No title - PDF Free Download

WATER RESEARCH A Journal of the International Water Association Editor-in-Chief Mogens Henze Institute of Environment &...

Author: Mark van Loosdrecht

294 downloads 3304 Views 14MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

WATER RESEARCH A Journal of the International Water Association

Editor-in-Chief Mogens Henze Institute of Environment & Resources Technical University of Denmark Bygningstorvet DK-2800 KGS Lyngby Denmark Tel: 45 4525 1477 Fax: 45 4593 2850 E-mail: [email protected]

Editors J. Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai The University of Tokyo Japan Gregory Korshin University of Washington USA Anna Ledin Technical University of Denmark Denmark Eberhard Morgenroth University of Illinois Urbana-Champaign USA W. Rauch University Innsbruck Austria Maria Reis Universidade Nova de Lisboa/FCT Portugal Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea Mark van Loosdrecht Delft University of Technology The Netherlands Thomas Ternes Bundesanstalt für Gewässerkunde Germany Stefan Wuertz Univ. of California, Davis USA Hanqing Yu University of Science & Technology of China China

Associate Editors Andrew Baker The University of Birmingham UK

Damien Batstone The University of Queensland Australia

S-L. Lo National Taiwan University Taiwan

G-H. Chen The Hong Kong University of Science & Technology Hong Kong China

Y. Matsui Hokkaido University Japan

Tom Curtis Univ. of Newcastle upon Tyne UK Ana Deletic Monash University USA Francis de los Reyes III North Carolina State University USA Rob Eldridge The University of Melbourne Australia Rosina Girones University of Barcelona Spain Stephen Gray Victoria University Australia Kate Grudpan Chiang Mai University Thailand Xiaodi Hao Beijing University of Civil Engineering and Architecture China E.E. Herricks University of Illinois - Urbana USA H-Y. Hu Tsinghua University China P.M. Huck University of Waterloo Canada Bruce Jefferson Cranﬁeld University UK Sergey Kalyuzhnyi Moscow State University Russian Federation

Max Maurer EAWAG Switzerland How Yong Ng National University of Singapore Singapore Satoshi Okabe Hokkaido University Japan S.L. Ong National University of Singapore Singapore Jong M. Park Pohang University of Science & Technology Korea Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Jean-Philippe Steyer LBE-INRA France M. Takahashi Hokkaido University Japan Kai Udert EAWAG Switzerland V.P. Venugopalan BARC Facilities India E. von Sperling Federal University of Minas Gerais Brazil A.I. Zouboulis Aristotle University Greece

Jaehong Kim Georgia Institute of Technology USA

Editorial Ofﬁce

Jes La Cour Jansen Lund Institute of Technology Sweden

Water Research Radarweg 29 1043 NX Amsterdam The Netherlands Tel: 31 20 485 2701 Fax: 31 20 485 2521; 31 20 485 3264 E-mail: [email protected]

G. Langergraber BOKU/Univ. of Natural Res. and Applied Life Scs. Austria

Mailing address

Publication information: Water Research (ISSN 0043-1354). For 2010, volume 44 is scheduled for publication. Subscription prices are available upon request from the publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/watres). Further information is available on this journal and other Elsevier products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon request. Claims for missing issues should be made within six months of the date of dispatch. Author Enquiries: Please visit this journal’s homepage at http://www.elsevier.com/locate/watres. You can track accepted articles at http://www.elsevier.com/ trackarticle and set up e-mail alerts to inform you of when an article’s status has changed. Also accessible from here is information on copyright, frequently asked questions and more. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. Orders, claims, and journal enquiries: Please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4537076 [outside the USA]; fax: (+1) (314) 5235153; e-mail: [email protected] Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865) 843970; e-mail: [email protected] Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3) 5561 5037; fax: (+81) (3) 5561 5047; e-mail: [email protected] Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510; e-mail: [email protected] Application for membership of International Water Association should be made to: Executive Director, IWA, Alliance House, 12 Caxton Street, London SW1H 0QS, U.K. (Tel.: +44 207 654 5500; Fax: +44 207 654 5555; e-mail: [email protected]; website: http://www.IWAhq.org.uk). Registered Charity (England) No. 289269. Individual membership is available from £30 upwards. For details contact IWA.

water research 44 (2010) 351

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Editorial

Editorial to special issue in Water Research Emerging contaminants in water The chemical pollution of natural waters is one of the big challenges of the 21st century. Based on the rapid evolution in analytical chemistry, with the possibility to detect more polar compounds a whole new suite of ‘‘emerging contaminants’’ such as pharmaceuticals, hormones and perfluorinated compounds has been identified in various compartments of the water cycle including both natural and technical aquatic systems. The discovery of these micropollutants in the aquatic environment has triggered research efforts to investigate sources and mitigation strategies with conventional and novel treatment processes and assess their importance with regard to ecotoxicology and human health. In this special issue on emerging contaminants, we have compiled 20 research articles and 2 reviews, which deal with these timely issues. As mentioned above, quantification of micropollutants in aquatic systems is a key requirement to assess their fate. Several studies in this issue address the determination of pharmaceuticals in archived biosolids (Halden et al.), in wastewater treatment (Lindberg et al.) and lagoon treatment (Wong et al.). Another study uses analytical data to evaluate the fraction of pharmaceutical residues in wastewater originating from hospitals (Ort et al.). Finally, a review covers the occurrence and fate of phytoestrogens in the environment (Liu et al.). To further elucidate the relevance of the micropollutants detected in various aquatic compartments, their (eco)toxicological potential has to be assessed. Several papers address related issues for individual compounds (lipid regulators, Fernandez-Pinas et al.) or classes of compounds (ionic liquids, Yun et al.) and for pharmaceuticals in advanced wastewater treatment systems such as powdered activated carbon and ozonation with in vivo and in vitro tests (Escher et al., Stalter et al.). Furthermore, one study is focused on toxicity nanotube suspensions (Tarabara et al.), which have been on the radar of emerging contaminants recently. Finally, one paper focuses on the toxicological relevance of emerging contaminants for drinking water (Schriks et al.). In recent years, municipal wastewater has been recognized as an important source of micropollutants to the

receiving water bodies. Therefore, mitigation strategies for the minimization of the discharge of these compounds play an increasingly important role in the urban water management. In this special issue, the removal of benzotriazoles (Reemtsma et al.) and pharmaceuticals, caffeine and DEET (Yu et al.) and other emerging contaminants (Rosal et al.) has been investigated for conventional wastewater treatment. Numerous papers address the oxidative removal of micropollutants from wastewater with chlorine, chlorine dioxide, ferrate, ozone, advanced oxidation processes, such as UV/ H2O2, (solar) photo-Fenton and non-thermal plasma (Lee et al., Malato et al., Reungoat et al., Gerrity et al., MendezArriaga). Other options for removal of micropollutants include membrane processes such as reverse osmosis (Hu et al.) and nanofiltration (Yangali-Quintanilla et al.) as well as sorption on sludge (Carrere et al.). The challenges caused by harmful algae producing toxins for desalination operations were reviewed by David Caron and co-workers. Finally, mitigation of micropollutants may also occur during managed aquifer recharge (Drewes et al.) or in biological Fenton-like processes (Vicent et al.). In short, we are very happy to provide you this Theme Issue on Emerging Contaminants. We appreciate the contributions by the authors, reviewers and editorial staff of Water Research to this project.

Thomas Ternes* Federal Institute of Hydrology (BFG), Am Mainzer Tor 1, 56068 Koblenz, Germany *Corresponding author. Tel.: þ49 261 1306 5560; fax: þ49 261 1306 5363. Urs von Gunten EAWAG, Ueberlandstrasse 133, Duebendorf CH-8600, Switzerland 0043-1354/$ – see front matter ª 2010 Published by Elsevier Ltd. doi:10.1016/j.watres.2010.01.015

water research 44 (2010) 352–372

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Review

Environmental fate and toxicity of ionic liquids: A review Thi Phuong Thuy Pham a, Chul-Woong Cho a, Yeoung-Sang Yun a,b,* a

Department of Bioprocess Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756, Republic of Korea Division of Semiconductor and Chemical Engineering and Research Institute of Industrial Technology, Chonbuk National University, Chonbuk 561-756, Republic of Korea b

article info

abstract

Article history:

Ionic liquids (ILs) are organic salts with low melting point that are being considered as

Received 31 May 2009

green replacements for industrial volatile organic compounds. The reputation of these

Received in revised form

solvents as ‘‘environmental friendly’’ chemicals is based primarily on their negligible vapor

27 August 2009

pressure. Nonetheless, the solubility of ILs in water and a number of literature

Accepted 12 September 2009

documenting toxicity of ILs to aquatic organisms highlight a real cause for concern. The

Available online 24 September 2009

knowledge of ILs behavior in the terrestrial environment, which includes microbial degradation, sorption and desorption, is equally important since both soil and aquatic

Keywords:

milieu are possible recipients of IL contamination. This article reviews the achievements

Ionic liquids

and current status of environmental risk assessment of ILs, and hopefully provides

Toxicity

insights into this research frontier.

Degradation

ª 2009 Elsevier Ltd. All rights reserved.

Biodegradation Environmental fate Sorption

Contents 1. 2.

3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toxicological aspect of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Effects of ILs in an enzyme level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Antibacterial activity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Toxicity of ILs to algae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Cytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Phytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Toxicity of ILs to invertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Inhibitory effects of ILs on vertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Environmental fate of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Chemical degradation of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Biodegradability of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Sorption of ILs in environmental systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

353 354 354 356 361 361 363 363 364 364 364 365 367

* Corresponding author. Division of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756, Republic of Korea. Tel.: þ82 63 270 2308; fax: þ82 63 270 2306. E-mail address: [email protected] (Y.-S. Yun). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.030

water research 44 (2010) 352–372

4.

1.

353

Concluding remarks and future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369

Introduction

Most of volatile organic compounds (VOCs) commonly used in industrial applications cause a major concern in the current chemical processing industry. The main problems are the toxicity of the organic solvents to both the process operators and the environment as well as the volatile and flammable nature of these solvents which make them a potential explosion hazard (Schmid et al., 1998). Recently, the deleterious effects of many solvents combined with serious environmental issues, such as atmospheric emissions and contamination of aqueous effluents are making their use prohibitive. Thus, many researchers have focused on the development of ‘‘green engineering’’ which represents research aimed at finding environmentally benign alternatives to harmful chemicals. Among the neoteric solvents applicable in ‘‘green technologies’’ ionic liquids (ILs) have garnered increasing attention over the others such as supercritical CO2 (Blanchard et al., 1999; Blanchard and Brennecke, 2001; Kazarian et al., 2000) and aqueous biphasic systems (Myasoedov et al., 1995; Rogers et al., 1995; Willauer et al., 1999). Ionic liquids, formerly known as molten salts, constitute one of the hottest areas in chemistry these days. Basically, they have melting points below 100 C, which can be achieved by incorporating a bulky asymmetric cation into the structure together with a weakly-coordinating anion (Ranke et al., 2004). The unique, highly solvating, yet non-coordinating environment of ILs provides an attractive medium for various types of chemical processes. Also, the physical properties of ILs can be tailored by a judicious variation in the length and branching of the alkyl chain and the anionic precursor (Fuler et al., 1997; Huddleston et al., 2001). In this way, ILs can be made taskspecific for a certain application. The almost limitless structural possibilities of ILs, as opposed to limited structural variations within molecular solvents, make them ‘‘designer

solvents’’ (Marsh et al., 2004; McFarlane et al., 2005; Sheldon, 2005). Some independent reports (Hagiwara and Ito, 2000; Olivier, 1999; Welton, 1999) and many reviews (Earle and Seddon, 2000; Rooney and Seddon, 2001) have highlighted ILs as representing a state-of-the-art, innovative approach to sustainable chemistry, with the argument that their vapor pressure is immeasurably low and they are not flammable. Recently, the application of these liquids as reaction media for organic synthesis, catalysis, or biocatalysis has been well documented (Earle and Seddon, 2000; Wasserscheid and Keim, 2000; Welton, 1999) (Fig. 1). Gordon (2001) pointed out that there is an obvious advantage in performing many reactions in ILs due to the improvement in process economics, reaction activity, selectivity and yield. Although ILs can lessen the risk of air pollution due to their insignificant vapor pressure, they do have significant solubility in water (Anthony et al., 2001; McFarlane et al., 2005; Wong et al., 2002). As a result, this is the most likely medium through which ILs will be released into the environment. Ionic liquids currently are not widely used in industrial applications; nonetheless, continued development and further use of these solvents may lead to accidental discharge and contamination. The properties that make them be the target of industrial interest (i.e. high chemical, thermal stability and non-volatility) suggest potential problems with degradation or persistence in the environment. In general, the deficiency of information and uncertainty surrounding the environmental impact of ILs is a major barrier to the utilization of these compounds by industry. Initial efforts have been made to overcome this drawback and offer a preliminary insight into the behavior of ILs in the aqueous environments. These studies provided extensive data sets, e.g. on (eco)toxicity, biodegradability, bioaccumulation and distribution of ILs in different environmental compartments. Therefore, it is necessary to consolidate all the available data in a single

Fig. 1 – Applications of ionic liquids.

354

water research 44 (2010) 352–372

review to lay the groundwork for more comprehensive community and ecosystem investigations. The overall objective of this review is to systematically gather and interpret existing information about the fate, removal options and (eco)toxicological assessment strategies of ILs.

2.

Toxicological aspect of ILs

The current literature represents a number of studies addressing the biological effects of ILs evaluated on the basis of toxicological test systems. The ILs toxicities towards these systems of different levels of biological complexity as well as several environmental compartments (Fig. 2) are successively discussed in the following subsections. All the structures of IL compounds discussed in this review were listed in Table 1. The acronyms used for these substances were adapted from Ranke et al. (2007a). In this way, the cation head groups were abbreviated as ‘‘IM’’ for imidazolium, ‘‘Py’’ for pyridinium, ‘‘Pyr’’ for pyrrolidinium, ‘‘Mor’’ for morpholinium, ‘‘Pip’’ for piperidinium, ‘‘Quin’’ for quinolinium, ‘‘N’’ for quaternary ammonium and ‘‘P’’ for quaternary phosphonium. The alkyl chains attached to the head group were given as numbers corresponding to the number of carbon in the alkyl residues. For example, the 1-butyl-3-methylimidazolium moiety was denoted as IM14. In case the carbon chain length equals or exceeds 10, the numbers were separated by a hyphen (e.g. IM1-10 indicated 1-decyl-3-methylimidazolium). Particularly, for pyridinium

entities, the carbon-bound alkyl chains were appended to the head group at different positions and the abbreviation was made by noting the position of attachment and a symbol for the attached group (e.g. Py4-2Me for 1-butyl-2-methylpyridinium). The anionic components were shortened as they are in the periodic table for the halides. For tetrafluoroborate, hexafluorophosphate, bis(trifluoromethylsulfonyl)imide, dicyanamide and hydrogen sulfate the abbreviations were BF4, PF6, (CF3SO2)2N, CN(N)2 and HSO4 in respective to their structural formula.

2.1.

Effects of ILs in an enzyme level

Enzyme inhibition data by ILs include those of the acetylcholinesterase from electric eel (Electrophorus electricus) (Arning et al., 2008; Jastorff et al., 2005; Matzke et al., 2007; Ranke et al., 2007b; Stasiewicz et al., 2008; Stock et al., 2004; Torrecilla et al., 2009; Zhang and Malhotra, 2005), the AMP deaminase (Sk1adanowski et al., 2005) and the antioxidant enzyme system of mouse liver (Yu et al., 2009a). The enzyme acetylcholinesterase plays the most important role in nerve response and function. Also, acetylcholinesterase catalyzes the hydrolysis of acetylcholinesters with a relative specificity for acetylcholine, which is a neurotransmitter common to many synapses throughout mammalian nervous systems (Fulton and Key, 2001; Massoulie´ et al., 1993). Thus, an inhibition of acetylcholinesterase leads to various adverse effects in neuronal processes, such as heart diseases or myasthenia

Fig. 2 – The flexible (eco)toxicological test battery considering aquatic and terrestrial compartments as well as different trophic levels including enzymes, luminescent marine bacteria, freshwater green algae, mammalian cells, duckweed, freshwater crustacean and zebrafish (Adapted from Matzke et al. (2007) by permission of the Royal Society of Chemistry).

355

water research 44 (2010) 352–372

Table 1 – Selection of cationic and anionic structures of commonly used ionic liquids. Head group

R2 N

N

+

Side chain R1 ¼ -C2H5, -C3H7, -C4H9, -C5H11, -C6H13, -C7H15, -C8H17, -C9H19, -C10H21, -C14H29, -C16H33, -C18H37, -C19H39 R2 ¼ -CH3, -C2H5

R1

Imidazolium (IM)

CH3 CH3

R1

+ N

R1 ¼ -C2H5, -C3H7, -C4H9, -C5H11, -C6H13, -C8H17

CH3

Pyridinium (Py)

CH3 N

+

R1 ¼ -C4H9, -C6H13, -C8H17

R1 Pyrrolidinium (Pyr)

O

N

CH3

+

R1 Morpholinium (Mor)

R1 ¼ -C4H9

Cation

R1

+

N

CH3

R1 ¼ -C4H9

Piperidinium (Pip)

+

N

R1 ¼ -C4H9, -C6H13, -C8H17

R1 Quinolinium (Quin)

R1

R2 +

N R4

R1-4 ¼ -CH3, -C2H5, -C3H7, -C4H9, -C6H13

R3

Quaternary ammonium (N)

R1

R2 +

P

R1-4 ¼ -C4H9, -C6H13, -C14H29

R3 (

R4

Quaternary phosphonium (P

(continued on next page)

356

water research 44 (2010) 352–372

Table 1 (continued) Head group Anion

Side chain Cl Br

Chloride Bromide

F Tetrafluoroborate [BF4]

B

F

-

F

F F

F

F

-

P

Hexafluorophosphate [PF6]

F

F F

F3 C

S

Bis(trifluoromethylsulfonyl)imide [(CF3SO2)2N]

O N

Dicyanamide [(CN)2N]

in humans (Chemnitius et al., 1999; Pope et al., 2005). Ranke et al. (2007b) published a comprehensive collection of acetylcholinesterase inhibition values for 292 compounds covering a large variety of ILs and closely related salts. Among these data, only those of the commonly tested ILs are summarized in Table 2 for the ease of comparison of ILs toxicity from molecular up to organism levels of biological complexity. It was found that all observed inhibitory effects on the enzyme could be exclusively accounted for the cationic moiety (Arning et al., 2008). In particular, the IL with pyridinium as cationic core structure inhibited the enzyme slightly stronger than the imidazolium analogue whereas the compounds based on phosphonium was less inhibitory. All anion species exerted no effect on the enzyme activity with only exception of the fluoride anion and the fluoride containing [SbF6] and [PF6] species. Both species are known to readily undergo hydrolysis in contact with moisture and thus the fluoride seems to be the active compound. The non-inhibiting effects of anion might be explained by their limited interactions with the active site of this enzyme (Matzke et al., 2007). In addition, a correlation between an increasing chain length of the side chains connected to the cationic head groups and an enhanced inhibitory potential of the ILs was found. It is believed that the mechanism involves the similarity of the positively charged imidazolium or pyridinium to the choline part that binds to the anionic site of the enzyme, such that the longer alkyl chain results in an improved fit (Stock et al., 2004). Sk1adanowski et al. (2005) discussed the usefulness of in vitro AMP deaminase inhibition assay as a potential molecular method in prospective risk analysis of imidazolium-based ILs. The results revealed that IM14 salts associated with [PF6], [BF4], p-tosylate and [Cl] demonstrated a dose-dependent inhibition of AMP deaminase activity. The IC50 values (concentration of ILs inhibiting 50% of enzyme activity) for those containing a fluorine compartment [PF6] and [BF4] are lower (5 mM) than those for [Cl] and p-tosylate (10 mM), which indicated the adverse effect of these fluoride-containing anions. The other study on enzyme inhibition assay dealt with

O

O -

CF3 S O

N -

N

N

the effects of acute exposure of intraperitoneal injection of aqueous IM18 Br on the antioxidant enzymes of the treated mouse liver (Yu et al., 2009a). The antioxidant enzymes tested included superoxide dismutase, catalase, glutathione peroxidase and glutathione-S-transferase. The results showed that administration of IM18 Br modified activities of these defense enzymes in mouse liver, and caused damage to livers of treated mice at median lethal dose (LD50) of 35.7 mg/kg. Though data published by these authors did not cover a large variety of ILs, the enzyme inhibition assays suggest the trend in which cationic moiety is the dominating factor influencing the toxicity of ILs, especially when substituted with a long alkyl side chain. Regarding the anion types, perfluoronated ions are of toxicological interest due to hydrolysis resulting in HF formation, while the others cause less prominent effect.

2.2.

Antibacterial activity of ILs

Bacteria serve as an ideal starting point for ILs toxicity estimations as they have short generation times. Preliminary toxicological investigations have shown quaternary ammonium and pyridinium compounds have critical inhibitory effects on a variety of bacteria and fungi (Babalola, 1998; Kelman et al., 2001; Li et al., 1998). In the studies of Pernak’s group (Cieniecka-Ros1onkiewicz et al., 2005; Pernak et al., 2001a; Pernak et al., 2001b; Pernak and Chwa1a, 2003; Pernak et al., 2003; Pernak et al., 2004a), they observed a trend of increasing toxicity with an increase in the alkyl chain length substituent in the pyridinium, imidazolium and quaternary ammonium salts to various bacteria including rods, cocci and fungi. As a measure of microbial activity of imidazolium and pyridinium ILs with varying alkyl chain lengths, Docherty and Kulpa (2005) also used a group of microorganisms possessing a variety of physiological and respiratory activities. It was found that imidazolium and pyridinium bromides incorporated hexyl- and octyl-chain had considerable antimicrobial effect to pure cultures of Escherichia coli,

Table 2 – Toxicity of ILs to different levels of biological complexity including enzyme, bacteria, algae, rat cell line, human cell lines, duckweed and invertebrate. Log10EC50 (mM)a

Compound Acetylcholin esterase IM12 Cl

2.0614

IM12 IM12 IM12 IM13 IM13 IM13 IM14

2.0514 2.0514 2.0314 2.2714 2.28 0.0318 2.2214 1.91 0.0411

BF4 PF6 (CF3SO2)2N Cl BF4 PF6 Cl

1.90 0.0218

IM14 BF4

1.98 0.01811

IM14 PF6

2.15 0.0518

IM14 (CF3SO2)2N IM14 (CN)2N

1.96 0.02111 1.95 0.0718

IM15 IM15 IM15 IM16

1.9614 1.8614 1.8514 1.9214

Cl BF4 PF6 Cl

IM16 Br

N.A.

IM16 IM16 IM16 IM17 IM17 IM17 IM18

1.8814 1.8814 2.1514 2.0714 2.1214 1.9114 1.6014

BF4 PF6 (CF3SO2)2N Cl BF4 PF6 Cl

IM18 Br

N.A.

4.5510 4.33 0.1119 N.A. N.A. N.A. N.A. 3.94 0.0613 N.A. 3.71 0.144 2.955 3.34 0.136 3.47 0.0419 4.01 0.054 3.07 0.0313 3.355 3.27 0.096 3.55 0.0413 3.10 0.176 3.12 0.3516 3.07 0.296 3.39 0.084 3.67 0.104 2.995 N.A. 3.14 0.0213 N.A. 1.9410 2.32 0.166 2.91 0.0913 1.42 0.124 0.815 3.18 0.0313 2.17 0.066 N.A. N.A. 2.44 0.0613 N.A. 1.19 0.116 1.01 0.0619 0.63 0.074 0.075

MCF7b

Lemna minor

N.A.

N.A.

N.A.

3.4414 3.9214 N.A. >4.3014 3.4714 >3.0014 3.5514

4.00 0.0420 N.A. 3.26 0.0420 N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. 2.8211

N.A. N.A. N.A. N.A. N.A. N.A. 1.934 1.93 0.061

N.A.

3.4314

3.44 0.1120

N.A.

N.A.

1.574 1.56 0.071 1.85 0.0622

N.A.

2.1111

3.1214

3.72 0.0517 3.66 0.0820

N.A.

2.497

1.684 1.67 0.111

4.15 0.069

2.20 0.0421

N.A.

3.1014

4.14 0.2217

N.A.

N.A.

2.55 0.159 N.A.

1.80 0.0712 N.A.

1.81 0.1511 N.A.

2.6814 3.1514

3.07 0.0820 N.A.

N.A. N.A.

2.45 0.0811 N.A.

1.854 1.85 0.101 N.A. N.A.

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. 1.92 0.0921

N.A. N.A. N.A. 0.0819

>3.0014 >3.0014 >3.0014 2.8514

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A.

N.A.

2.57 0.152

N.A.

N.A.

N.A.

N.A.

N.A.

N.A. 3.25 0.679 2.53 0.159 N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. 1.4621

N.A. N.A. N.A. N.A. N.A. N.A. 2.67 0.3719

2.9814 2.9114 2.2414 2.5314 2.5814 2.3014 2.0114

N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. 2.818 N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A.

0.784 1.06 0.0422 N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A.

1.65 0.252

N.A.

2.48 0.0420

N.A.

N.A.

Escherichia coli

Pseudo kirchneriella subcapitata

Scenedesmus vacuolatus

IPC-81

N.A.

N.A.

2.78 0.0619

N.A.

5.25 0.069 N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. 2.34 0.0121

N.A. N.A. N.A. N.A. N.A. N.A. 2.26 0.0811

N.A.

3.46 0.0623

4.60 0.029

N.A.

HeLa

Daphnia magnac N.A.

water research 44 (2010) 352–372

IM14 Br

Vibrio fischeri

1.334 0.54 0.1222

357

(continued on next page)

358

Table 2 (continued) Log10EC50 (mM)a

Compound Acetylcholin esterase 1.53 0.02511 2.0314 2.0314 N.A. 1.0914

IM1-10 BF4 IM1-10 PF6 IM1-14 Cl IM1-16 Cl IM1-18 Cl IM1-19 Cl IM1-19 BF4 IM1-19 PF6 IM22 Br IM23 Br IM24 BF4 IM25 BF4 IM26 Br IM26 BF4 IM2-10 Br Py Cl Py2 Cl Py3 Br Py3 (CF3SO2)2N Py4 Cl

1.10 0.0418 1.6814 0.5414 0.6814 0.9614 1.3614 1.4314 1.6214 2.0814 2.2114 2.03 0.0118 N.A. 1.7714 1.8414 0.9214 >3.0014 2.1014 2.2214 2.2114 1.7014

Py4 Br

1.7714

Py4 BF4 Py4 PF6 Py4 (CN)2N

1.8014 1.8414 N.A.

Py5 Br Py5 (CF3SO2)2N Py6 Cl Py6 Br Py6 PF6 Py6 (CF3SO2)2N Py8 Cl Py8 (CF3SO2)2N Py4-2Me Cl Py4-2Me BF4

1.5214 1.5514 1.7214 N.A. 1.7614 1.8514 1.6014 1.4014 0.7014 0.8214

Escherichia coli

Pseudo kirchneriella subcapitata

Scenedesmus vacuolatus

IPC-81

HeLa

MCF7b

Lemna minor

1.41 0.0713 0.95 0.126 N.A. 0.72 0.0413 0.50 0.0713 0.23 0.0619 0.18 0.0613 N.A. 0.15 0.0719 0.23 0.0819 1.45 0.0519 N.A. N.A. N.A. N.A. N.A. 2.8 0.0413 3.1413 N.A. 2.15 0.0513 N.A. N.A. N.A. N.A. N.A. 3.41 0.084 2.645 3.18 0.0619 3.40 0.014 2.735 N.A. N.A. 3.31 0.104 2.615 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. 2.64 0.159 N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A.

2.3011 N.A. N.A. N.A. 3.57 0.0619

1.5914 1.9614 1.6414 N.A. 1.3414

2.48 0.0220 N.A. 2.28 0.0220 N.A. N.A.

2.848 N.A. N.A. N.A. N.A.

0.907 N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 2.57 0.0621

N.A. N.A. 2.48 0.219 >2.0019 >2.0019 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 2.59 0.1119

0.7714 1.5014 0.4214 0.1914 0.0114 1.4014 1.6514 1.8514 >3.0014 >3.3014 3.2614 N.A. 2.0114 2.2614 0.5314 N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 4.36 0.0917 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 2.32 0.1819

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A.

N.A.

N.A.

3.9014

3.50 0.0720

N.A.

N.A.

N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

3.1814 N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. 1.2714 N.A. N.A. 3.2514

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. 1.074 N.A. N.A. N.A. N.A. N.A. N.A.

Daphnia magnac

water research 44 (2010) 352–372

IM18 BF4 IM18 PF6 IM18 (CF3SO2)2N IM19 BF4 IM1-10 Cl

Vibrio fischeri

1.1514 N.A.

Py4-3Me BF4 Py4-3Me PF6 Py4-3Me (CN)2N

1.53 0.0218 1.45 0.0218 1.2214

Py6-3Me Cl Py6-3Me Br

1.0614 N.A.

Py6-4Me Py6-4Me Py8-3Me Py8-3Me

Cl BF4 Cl Br

1.4414 1.4814 0.6414 N.A.

Py8-4Me Cl Py8-4Me BF4 Pyr14 Cl Pyr14 Br

N.A. 3.46 0.0623

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. 1.764

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

3.3014 N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. N.A.

N.A. 0.594

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A.

N.A. 2.1714 N.A. N.A.

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. 1.008

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. 0.404

N.A. N.A. N.A. N.A.

N.A. N.A. N.A. 3.67 0.283

N.A. N.A. 3.37 0.1019 N.A.

1.6314 1.4914 >4.3014 3.7714

N.A. N.A. N.A. N.A.

N.A. N.A. 2.16 0.2519 N.A.

N.A. N.A. N.A. N.A.

Pyr14 BF4 Pyr14 (CF3SO2)2N Pyr14 (CN)2N Pyr16 Cl Pyr16 (CF3SO2)2N Pyr18 Cl Pyr18 BF4 Pyr66 Mor14 Cl Mor14 Br Mor14 (CF3SO2)2N Pip14 Br

1.9114 2.1314 1.9814 2.4814 2.6014 2.3614 2.0214 2.0814 N.A. 2.7114 2.7814 1.8314

N.A. N.A. N.A. 2.9910 N.A. N.A. N.A. N.A. >4.3019 N.A. N.A. 4.27 0.0919

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. >2.3812 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. 2.5319 N.A. N.A. N.A. N.A. N.A. N.A. N.A. >4.0019 2.0019 3.27 0.1219

2.9014 3.0114 4.2314 2.9114 N.A. 2.5914 1.8214 1.2314 N.A. >4.3014 3.4314 4.0314

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. 2.98 0.3219 N.A. N.A. N.A. N.A. N.A. N.A. N.A. 3.11 0.1319 3.15 0.1319 0.4719

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

Pip14 (CF3SO2)2N Quin4 Br Quin4 BF4 Quin6 BF4 Quin8 Br Quin8 BF4 N1111 Br N1114 (CF3SO2)2N N1123 (CF3SO2)2N N1124 Cl N1124 (CF3SO2)2N N2222 Cl N2222 Br N2226 Br N4444 Br P4444 Br P666-14 Br

1.7814 0.7914 0.6214 0.4814 N.A 0.3014 N.A. 2.6014 2.3414 2.0614 2.0314 2.8014 N.A. N.A. 2.3014 2.6114 2.8514

N.A. N.A. N.A. N.A. N.A. N.A. >5.004 N.A. N.A. N.A. N.A. N.A. >5.004 2.46 0.164 3.27 0.074 2.714 3.41 0.024

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

2.0819 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. >4.0019 1.78 0.1719 N.A. N.A. N.A. N.A. N.A. N.A.

3.4114 2.3214 2.1614 1.0714 0.0314 0.1714 N.A. 3.6114 N.A. >4.3014 3.4313 >3.4814 N.A. N.A. 2.2514 1.6614 N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 4.26 0.0420 N.A. N.A. N.A. N.A.

N.A. N.A. N.A. 4.64 0.0215 4.258 N.A. 3.148 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 4.15 0.0415 4.158 2.938 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.

2.85 0.0719 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 0.83 0.6719 N.A. N.A. N.A. N.A. N.A. N.A. N.A.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 1.474 0.954 N.A.

(continued on next page)

359

N.A. N.A.

1.1114 1.2214 1.9214 1.9314

N.A. 2.75 0.134 2.125 N.A. N.A. 2.66 0.054 1.995 1.4410 2.06 0.164 1.485 N.A. N.A. N.A. 0.79 0.054 0.255 N.A. N.A. >4.3019 N.A.

water research 44 (2010) 352–372

Py4-3Me Cl Py4-3Me Br

360

References: 1Bernot et al. (2005a); 2Cho et al. (2007); 3Cho et al. (2008a,b); 4Couling et al. (2006); 5Docherty and Kulpa (2005); 6Garcia et al. (2005); 7Jastorff et al. (2005); 8Kumar et al. (2009); 9Lee et al. (2005); 10 Luis et al. (2007); 11Matzke et al. (2007); 12Pretti et al. (2008); 13Ranke et al. (2004); 14Ranke et al. (2007b); 15Salminen et al. (2007); 16Samorı` et al. (2007); 17Stepnowski et al. (2004); 18Stock et al. (2004); 19 Stolte et al. (2007a); 20Wang et al. (2007); 21Wells and Coombe (2006); 22Yu et al. (2009). a N.A. means not available (not determined). b Toxicity of ILs is expressed as log10IC50 (mM) in case of MCF7 cell line. c Toxicity of ILs is expressed as log10LC50 (mM) in case of D. magna.

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 1.90 0.0520 N.A. 0.4814 N.A. 0.2414 N.A. N.A. N.A. N.A. N.A.

Scenedesmus vacuolatus Pseudo kirchneriella subcapitata

N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 3.47 0.0818 >3.3017 >3.4814 3.40 0.218 BF4 PF6 (CF3SO2)2N (CN)2N P666-14 P666-14 P666-14 P666-14

Compound

Table 2 (continued)

Acetylcholin esterase

Vibrio fischeri

Escherichia coli

Log10EC50 (mM)a

IPC-81

HeLa

MCF7b

Lemna minor

Daphnia magnac

water research 44 (2010) 352–372

Staphylococcus aureus, Bacillus subtilis, Pseudomonas fluorescens and Saccharomyces cerevisiae. The anion performed nearly no effect on antimicrobial activity in the case of imidazolium analogues (Docherty and Kulpa, 2005; Garcia et al., 2005; Lee et al., 2005; Pernak et al., 2003; Pernak et al., 2004a) whereas this was not the case for phosphonium salts. Within the group of alkyltrihexylphosphonium ILs in the study of CienieckaRos1onkiewicz et al. (2005), both cation structure and the type of anion had effects on the biological activity. The antibacterial activity of ILs not only involves in hampering the growth rate of microbes but also interferes with their productivity. Matsumoto et al. (2004a) tested the toxicity of imidazolium-based ILs to lactic acid producing bacterium Lactobacillus rhamnosus to examine whether these compounds can replace conventional organic solvents in the extractive fermentation of lactate. The results showed that the bacterium L. rhamnosus grew, consumed glucose, and produced lactate in the presence of imidazolium-based ILs. A change of alkyl length in the imidazolium cation had little difference on the survival of the cells. In a similar study (Matsumoto et al., 2004b), they focused on hiochii bacteria, Lactobacillus homochiochii and Lactobacillus fructivorans and also found that the bacteria could produce lactic acid in the presence of ILs. Nonetheless, the lactic acid producing activities of these bacteria generally decreased with the extension of alkyl chain length in the imidazolium cation moiety. Water miscible ILs had various effects on the physiology of Clostridium sporogenes when tested as additives in culture media or reaction media for reduction of nitrobenzene (Dipeolu et al., 2008). In their study, 2-hydroxyethyltrimethylammonium dimethylphosphate and N,N-dimethylethanolammonium acetate increased the growth rate of C. sporogenes; by contrast, IM14 BF4 and IM12 EtSO4 inhibited growth. Although IM12 EtSO4 inhibited growth, it was sufficiently non-toxic to allow efficient reduction of nitrobenzene using harvested cells. Thus, it is recommended that both noninhibitory and partially inhibitory ILs should be screened for use in biotransformation. Nonetheless, Ganske and Bornscheuer (2006) referred that ILs could have substantial inhibitory effects on the growth of microorganisms when they explored the effects of the two most commonly used ILs IM14 BF4 and IM14 PF6 on the growth of E. coli, Pichia pastoris and Bacillus cereus. Regarding inhibition assays used in assessment of environmental potential risk of a compound in aquatic milieu, the bioluminescence assay using Vibrio fischeri (formerly known as Photobacterium phosphoreum) is one of the most applied (Kaiser and Palabrica, 1991; Steinberg et al., 1995). This is a rapid, costeffective, and well-established method for toxicity determination focusing on environmental issues, and also a standard ecotoxicological bioassay in Europe (DIN EN ISO 11348). The published data on ILs toxicity towards V. fischeri were listed on Table 2 and were comprehensively interpreted in the study of Peraccini et al. (2007). Although it has been claimed that modifications of the anion lead to changes in chemical and physical properties of ILs (Sheldon, 2001), no clear increase in toxicity caused by the anion could be observed, and toxicity seemed to be determined mainly by the cationic component (Ranke et al., 2004). This is likely explained by the fact that lipophilic part of the molecules can be intercalated into the

water research 44 (2010) 352–372

membrane, whereas their ionic head group is at least partially solvated in the aqueous solution, as suggested by Austin et al. (1998). The ILs toxicity was also observed to correlate directly with the length of the n-alkyl residues in the methylimidazolium cation (Romero et al., 2008). Interestingly, Ranke et al. (2004) noted a slight hormetic effect at concentrations below inhibitory concentrations. Concerning the anionic influence, compounds with [PF6] were found to be slightly more toxic than compounds with other anions in their study (Ranke et al., 2004). The anion [(CF3SO2)2N] showed no intrinsic toxicity to V. fischeri in the report of Matzke et al. (2007); in contrast, an increased in toxicity was found for all tested compounds combined with [(CF3SO2)2N] for V. fischeri (Stolte et al., 2007a). Couling et al. (2006) extended the bioluminescence inhibition assay to pyridinium derivatives and it was noted that the quaternary ammonium compounds seemed to be less toxic to V. fischeri than the pyridinium and imidazolium analogues. Also, the quantitative structureproperty relationship (QSPR) modeling suggested that imidazolium cations, with two nitrogen atoms, are predicted to be more toxic than pyridinium moieties, which only have one nitrogen atom in the structure. In addition, the QSPR correlation predicted that quaternary ammonium cations are less toxic than those with cations containing nitrogen-bearing rings, which was in agreement with the experimental results (Couling et al., 2006). However, in contrast to the cases of aromatic ILs and ammonium compounds, the authors were unsuccessful in modeling the behavior of phosphonium salts using the developed correlation.

2.3.

Toxicity of ILs to algae

As algae are primary producers, either directly or indirectly, of organic matter required by animals in freshwater food chains, their ecology is crucial in providing the energy for sustaining other higher trophic levels. The ubiquity of algae makes these organisms ideal for toxicological studies and, because they have a short life cycle they can respond quickly to environmental change (Blaise, 1993; Lewis, 1995). To date, several groups have focused their attention on the use of algal primary producers to assess the effects of ILs to aquatic environments (Cho et al., 2007; Cho et al., 2008a,b,c; Grabinska-Sota and Kalka, 2006; Kulacki and Lamberti, 2008; Lata1a et al., 2005; Matzke et al., 2007; Matzke et al., 2008; Pham et al., 2008a,b; Pretti et al., 2009; Stolte et al., 2007a; Wells and Coombe, 2006). Cho and co-workers used Pseudokirchneriella subcapitata (formerly known as Selenastrum capricornutum) to study the effect of different head groups, side chains and anions of ILs on algal growth rate and photosynthetic activity. The data revealed that the toxic influence of ILs on growth rates were more significant than those of photosynthetic performance (Pham et al., 2008b). Once again, the trend of increasing toxicity with increasing alkyl chain length was observed in their reports (Cho et al., 2007; Pham et al., 2008b). Regarding the anionic effects, P. subcapitata was sensitive to the anion moieties in the order: [SbF6] > [PF6] > [BF4] > [CF3SO3] > [C8H17OSO3] > [Br] z [Cl]. In particularly, it was found that with respect to IL incorporating perfluorinated anion (i.e. IM14 BF4), EC50 values (concentrations which lead to a 50% reduction of the exposed organisms

361

relative to control) of the previously prepared stock solution (6 months prior to experiment) were significantly lower compared to those of the freshly made one (Pham et al., 2008a). This might be due to hydrolytic effects of IM14 BF4 leading to fluoride formation, as confirmed by ion chromatography analysis. This implies that after ILs are released into the aqueous system; they can become more hazardous than expected by laboratory data with fresh ILs. In a detailed study on hydrolysis of fluoride-containing anions, Cho et al. (2008a) showed that IM14 SbF6 generated a greater amount of fluoride compared to IM14 BF4, but no fluoride formation occurred with the hexafluorophosphate. When only small amounts of fluoride ions were formed from IM14 SbF6 and IM14 BF4 within 96 h, the formed fluoride ion did not affect the algal growth rate. Nevertheless, the fluoride ion formation from IM14 BF4 increased with incubating time of the stock solution; thus, the toxicity might significantly increase according to the further formed fluoride ions. In view of cationic effect, Pyr14 Br was found to be the least toxic of all the ILs tested to P. subcapitata (Cho et al., 2008b). For the limnic green alga Scenedesmus vacuolatus, a severe toxicity was found for 1-butyl-4-(dimethylamino)pyridinium, whereas the quaternary ammonium and morpholinium compounds exhibited no toxicity (Stolte et al., 2007a). Despite the extensive studies on the toxicological impact of ILs towards freshwater phytoplankton, inhibition mechanism of both the growth rate and photosynthetic activity by ILs has not been described by the authors. Lata1a et al. (2005), who selected two marine algae Oocystis submarina (green algae) and Cyclotella meneghiniana (diatom) as testing organisms, found that the two species differed dramatically in their ability to recover from IL exposure. Additionally, it was discovered that IL toxicity declined with increasing salinity. The lower toxicity of IL in this case is probably due to the reduced permeability of IL cations through the algal cell walls. High amounts of chloride provide a good ion-pairing environment for imidazolium cations, which consequently compete with hydroxyl or silanol functional groups in the cell-wall structure of green alga and diatom, respectively. Though no information on EC50 values was described, the facts emerged from this work provide useful information in the further fate assessment of ILs in marine environments.

2.4.

Cytotoxicity of ILs

As a cellular test system, promyelotic leukemia rat cell line IPC-81 has been frequently used in cytotoxicity assays of ILs, with the reduction of the WST-1 dye as an indicator of cell viability (Matzke et al., 2007; Ranke et al., 2004; Ranke et al., 2007a; Stasiewicz et al., 2008; Stolte et al., 2006; Stolte et al., 2007b; Torrecilla et al., 2009). It was observed that ILs with polar ether, hydroxyl and nitrile functional groups within the side chains exhibited low cytotoxicity compared to those incorporated with ‘‘simple’’ alkyl side chains (Kumar et al., 2009; Stasiewicz et al., 2008; Stolte et al., 2007b). Those functional groups were thought to impede cellular uptake by membrane diffusion and reduce lipophilicity based interactions with the cell membrane (Stolte et al., 2007b). Taking a closer look at the effects of sub-structural elements of ILs, [(CF3SO2)2N] anion and 4-(dimethylamino)pyridinium cation

362

water research 44 (2010) 352–372

were described to have intrinsic effects of anion and head group on cytotoxicity, respectively (Stolte et al., 2007b). The well known side chain length effect (decrease in EC50 values with elongation of the alkyl side chain) could also be confirmed in these studies. To date many studies have analyzed the toxicity of ILs on human cell lines (Frade et al., 2007; Garcı´a-Lorenzo et al., 2008; Hassoun et al., 2002; Kumar et al., 2009; Salminen et al., 2007; Stepnowski et al., 2004; Wang et al., 2007). These in vitro systems have been extremely beneficial in studying the molecular basis of chemical’s biological activity, including its toxic mode of action (Blaauboer et al., 1998) and could facilitate extrapolation of in vitro data with regard to possible effects on humans (Malich et al., 1997). Most of studies dealt with HeLa cells exemplifying prototypical cells of the human epithelium which is normally the site of first contact of an organism with toxicants. According to Stepnowski et al. (2004), the cytotoxicity data implied that effects of IM14 cation coupled with chloride, tetrafluoroborate or hexafluorophosphate were probably dependent on the anionic moieties. The lowest effect concentrations for tetrafluoroborate species were found to be 0.63 mM, whereas hexafluorophosphate and chloride inhibited HeLa cell growth at comparably high concentrations of >10 mM. Surprisingly, when the anion effect was compared, the strongest inhibition was found for [PF6]. This might be due to hydrolysis affecting fluoride formation, thus causing serious toxicological consequences through the decomposition product. A similar phenomenon was observed by Ranke et al. (2004) in IPC-81 leukemia cells, where the lower toxicity of 1-n-butyl-3methylimidazolium hexafluorophosphate in comparison to the hexafluorophosphate anion alone was explained by reduced anion uptake due to the formation of an ion pair. The anion in this ion pair can, however, also be partially decomposed. This was shown in recent work by Swatloski et al. (2003), who identified traces of 1-n-butyl-3-methylimidazolium fluoride hydrate as a decomposition product formed during the purification of the 1-n-butyl-3-methylimidazolium hexafluorophosphate. As shown by Wang et al. (2007) the phosphonium bis(trifluoromethylsulfonyl)imide salts performed the highest inhibitory to HeLa cells, followed by alkylimidazolium, alkylpyridinium, alkyltriethylammonium and N-alkyl-N,Ndimethyl-N-(2-hydroxylethyl)ammonium salts, in decreasing order. For each cation class the toxicity increased with increasing chain length of the alkyl substituent for a given anion: 1-ethyl-3-methylimidazolium bromide yielded an EC50 of 8.4 mM, substituting the ethyl moiety for a butyl group led to an EC50 of 2.8 mM, and for an octyl moiety an EC50 of 0.3 mM. This result was consistent with what has been observed in other studies. Salts containing the tetrafluoroborate anion showed the highest EC50, followed closely by bromide and chloride. Bis(trifluoromethylsulfonyl)imide salts were significantly more toxic than their halide counterparts. However, the effect of changing the anion was smaller than that of changing the alkyl substituent, e.g. while 1-ethyl-3-methylimidazolium tetrafluoroborate was observed to have an EC50 of 9.9 mM, the corresponding bromide and bis(trifluoromethylsulfonyl)imide salts had EC50 of 8.4 and 1.8 mM, respectively – these all considerably less toxic than 1-octyl-3-methylimidazolium bromide.

The CaCo-2 cells were used in the study of Garcı´a-Lorenzo et al. (2008) with the aim of a convenient screening method for obtaining first rough estimates for the toxic potential of ILs. The obtained data showed that in general, ILs with longer alkyl chains were more lipophilic than those with shorter alkyl chains. The former can be presumed to have a tendency to be incorporated into the phospholipid bilayers of biological membranes. In this respect, some authors have indicated that the increased toxicity of longer ILs can be accounted for enhanced membrane permeability altering the physical properties of the lipid bilayer (Lata1a et al., 2005; Ranke et al., 2004; Stepnowski et al., 2004). Additionally, it has been proposed that the mode of toxic action for ILs takes place through membrane disruption because of the structural similarity of imidazolium-based ILs to detergent, pesticides and antibiotics able to cause membrane-bound protein disturbance (Docherty and Kulpa, 2005). Recently, Ranke et al. (2007a,b) have demonstrated that lipophilicity of ILs dominates their in vitro cytotoxicity over a wide range of structural variations. The contribution of the anionic part of the ILs to the observed biological effect was evaluated by comparing the EC50 values obtained for the cations IM16 and IM18, combined with two different anions [Cl] and [PF6]. For both cations, a stronger toxic effect was found for chloride derivatives, but not for fluoride containing hexafluorophosphate. A similar result was reported by Stock et al. (2004) where the inhibitory effects of IM14 Cl and IM14 PF6 on the acetylcholinesterase activity were compared. In addition, slightly higher cytotoxicity for the chloride derivative has also been observed when the cytotoxicity of IM14 Cl and IM14 PF6 on HeLa cells was tested (Stepnowski et al., 2004). This implies the effect of perfluorinated ions is not drastic to all but vary according to species of organisms tested. Several authors have pointed out that altering the anion has only minimal effects on the toxicity of several imidazolium compounds (Bernot et al., 2005a; Garcia et al., 2005; Ranke et al., 2004). This indicates that ILs toxicity seems to be related to the alkyl chain branching and to the hydrophobicity of the imidazolium cation but not to the various anions. In this respect, a recent study using the IPC-81 rat leukemia cell line with a large pool of anions demonstrated that most of the commercially available anions showed no or only marginal cytotoxic effects. However, anionic compartments with lipophilic and hydrolysable structural elements are likely to be of considerable relevance with respect to the toxicity of ILs (Stolte et al., 2006). In a recent study (Frade et al., 2007), the human cell lines such as HT-29 and CaCo-2 cells were utilized to estimate the inhibitory effect of ILs with several types of cations and anions. In both cells, IM14, IM12OH (1-(2-hydroxyethyl)-3methylimidazolium), IM12O2O1 (1-(2-(2-methoxyethoxy)ethyl)-3-methylimidazolium) and cholines were the least toxic cations independently of the anion. Within the studied combinations, it can be noted that IM14 PF6, IM14 acesulfame, IM12OH BF4/PF6, IM12O2O1 BF4/PF6, IM12OH acesulfame and IM12OH saccharine are not toxic and present good alternatives to organic solvents. Meanwhile, increasing the length of the substituent chain may contribute to a significant increasing of imidazolium toxicity. It was also noted that [(CF3SO2)2N] anion decreased the toxicity to a large extent,

water research 44 (2010) 352–372

independently of the cation and for both cell types, which was in accordance with Salminen et al. (2007).

2.5.

Phytotoxicity of ILs

The studies on phytotoxic activity of ILs were conducted mostly on the duckweed, Lemna minor, a common aquatic vascular plant (Jastorff et al., 2005; Larson et al., 2008; Matzke et al., 2007; Stolte et al., 2007a). In general, 1-alkyl-3methylimidazolium compounds with longer alkyl chains were more toxic to L. minor than those with short alkyl chain lengths. Imidazolium and pyridinium cations with butyl groups had similar EC50s (the concentrations that produced a 50% reduction in root growth) (39.07 and 32.54 mM, respectively); while the equivalent ammonium cation had a much higher EC50 (101.48 mM; i.e., less toxic) (Larson et al., 2008). In consideration of anionic effect, [(CF3SO2)2N] was found to cause moderate toxicity to this duckweed (EC50 ¼ 6300 mM) (Matzke et al., 2007). On the other hand, this anion had no or even a positive influence on the observed effects on L. minor (Stolte et al., 2007a). Focusing on the terrestrial environment, Matzke et al. (2009a) investigated the influence of differently composed soils, with varying contents of the clay minerals smectite and kaolinite, on the toxicity of different anion species of imidazolium-based ILs towards the wheat Triticum aestivum. The data showed that IM14 (CF3SO2)2N appeared the most toxic, independently of the type and concentration of added clay. This is totally in contrast to the findings of Stolte et al. (2007a), who reported that [(CF3SO2)2N] caused no harm to L. minor, indicating the toxic effect of this anion is different between certain plants. The toxicity of 1-butyl-3-methylimidazolium incorporated chloride, tetrafluoroborate and hydrogen sulfate was mainly controlled by the cationic moiety. The observed effects varied according to the added clay type and clay concentration. An increase of clay content resulted in less inhibitory effects of these substances. On the contrary, for IM14 combined with bis(trifluoromethylsulfonyl)imide the addition of clay minerals led to higher toxicity compared to the reference soil. Since results are contradictious further study is necessary to unravel the underlying mechanism. Moreover, a detailed study on the effect of IM14 BF4 on the wheat T. aestivum seedlings (Wang et al., 2009) showed that IM14 BF4 was hazardous to the early development of wheat and had varying effects on different organs. At low concentrations, IM14 BF4 did not inhibit, and even promoted, wheat seedling growth. Nonetheless, at high concentrations, this IL inhibited wheat seedling growth significantly and decreased chlorophyll content, thereby reducing photosynthesis and plant growth. Therefore, the authors suggested that dilution could decrease the toxicity of IM14 BF4 to plants and would be a good method for remediating IL-polluted environments. In another research, the phytotoxicity tests of chiral ILs containing (-)-nopyl derivatives were carried out in a plant house using spring barley (Hordeum vulgare) which is a monocotyledonous plant, and a common radish (Raphanus sativus L. subvar. radicula Pers.) which is a dicotyledonous plant (Ba1czewski et al., 2007). According to the data obtained, increasing the concentration of ILs resulted in a systematic decrease in the crop fresh weight of total sprouts and the crop fresh

363

weight per plant, both for spring barley and for common radish. It could also be noted that common barley was a more resistant plant which fairly well tolerates test IL concentrations up to 200 mg kg1 of soil; whereas, for radish, the growth and development inhibiting concentration is 100 mg kg1 of soil. Using the same target plant (H. vulgare), Pernak et al. (2004b) reported that the 1,3-dialkoxymethylimidazolium tetrafluoroborate salts introduced to the soil at concentration of 1,000 mg kg1, or 100 mg kg1 dry mass of soil, were found to exert a phytotoxic effect on monocotyledonous plants. On the other hand, at a concentration of 10 mg kg1 no such effect on the growth of the roots was notified. Concerning phytotoxicity of ILs to garden cress (Lepidium sativum L.) in soil environment, Studzin´ska and Buszewski (2009) have proved that hazardous effects of imidazolium ILs are closely connected with organic matter content in soil. Soil with more organic carbon was observed to sorb IL cations more extensively than soil with little or no organic matter; hence, the more fertile in soil, the lower probability of hazardous effect of ILs to plants. On the other hand, the hazardous character of analyzed ILs was strongly connected with their hydrophobicity, indicating that the more hydrophobic IL, the higher decrease of seed germination. Although intensive work has not been conducted on phytotoxic influence of ILs, the available data offer initial hints for environmental scientists dealing with the potential impact of ILs towards aqueous and terrestrial plants.

2.6.

Toxicity of ILs to invertebrates

Ecotoxicological literature of ILs to invertebrates mainly focus on the use of Daphnia magna as a test organism (Bernot et al., 2005a; Couling et al., 2006; Garcia et al., 2005; Grabinska-Sota and Kalka, 2006; Luo et al., 2008; Nockemann et al., 2007; Pretti et al., 2009; Samorı` et al., 2007; Wells and Coombe, 2006; Yu et al., 2009b). Daphnia is an important link between microbial and higher trophic levels (McQueen et al., 1986), and has been the subject of hundreds of intensive ecological studies. The results of all studies again observed the well-established link between toxicity and alkyl chain length of the tested ILs containing imidazolium, pyridinium or quaternary ammonium as counter cations. The most toxic compound towards D. magna was found to be IM18 Br whereas the least toxic one was IM14 Cl with log10EC50 values of 1.33 and 1.93, respectively (Table 2). Also, the nature of the anion was suggested to have smaller effects compared to those of the cation. In a recent study, Luo et al. (2008) investigated the developmental toxicity of IM18 Br on D. magna. It was found that this compound exhibited toxicity on the development of three generation of D. magna with the decrease of number of offspring and average brood size correlated to increasing IM18 Br concentrations. This indicated that IM18 Br could cause deleterious effect to the population of Daphnia and indirectly disturb freshwater food webs. Couling et al. (2006) used experimental data to determine which part of the IL molecule is responsible for the observed toxic effects through a quantitative structure-property relationship (QSPR) modeling. In this respect, correlative and predictive equations were generated and proved that there was a distinct influence of the length of alkyl residues attached to the aromatic nitrogen atoms

364

water research 44 (2010) 352–372

towards D. magna. Moreover, the models predicted that the toxicity increased slightly with increasing number of aromatic nitrogen atoms in cation ring. This implies that ammonium salts are less toxic than pyridinium salts, which in turn are less toxic than imidazolium moieties. Interestingly, it was noted that methylating the aromatic carbons could be effective in reducing toxicity to D. magna, indicating that 1-nbutylpyridinium bromide can be more toxic than 1-n-butyl-3methylpyridinium bromide, which is more toxic than 1-nbutyl-3,5-dimethylpyridinium analogue. The QSPR, though is still in its infancy, has contributed initial guidelines for a rationale design of a new category of ILs with an acceptable environmental profile. Other studies include data on the snail Physa acuta (Bernot et al., 2005b), the spring tail Folsomia candida (a soil invertebrate) (Matzke et al., 2007), Caenorhabditis elegans (a soil roundworm) (Swatloski et al., 2004) and Dreissena polymorpha (zebra mussel) (Costello et al., 2009). It was also demonstrated a positive relationship between alkyl chain length and toxicity in these reports. In the research of Bernot et al. (2005b), the estimated LC50 (median lethal concentration) ranged from 3.50 to 1799.8 mM (0.54 to 3.26 in the logarithmic form), which implied that P. acuta are less sensitive to ILs than are D. magna (log10LC50 (mM) ranging from 1.33 to 1.93 (Table 2)). Also, the authors observed that at low concentrations, the IL may suppress snail movement, but concentrations above this threshold level trigger an escape response, causing the organism to move faster. Grazing patterns, nonetheless, showed that snails grazed less at higher IL concentrations. Physa spp. are key components of freshwater food webs, because they graze algae and are themselves important prey for fish and invertebrate predators (Bernot and Turner, 2001; Osenberg and Mittelbach, 1989). Thus, nonlethal IL concentrations affected P. acuta behaviors, potentially influencing individual fitness and good web interactions.

2.7.

being 42.4, 43.4 and 85.1 mg/L, respectively, indicating that the developmental toxicity of IM18 Br in the frog was stagesensitive. The number of dead embryos was also found to increase with the increasing concentrations of the IL IM18 Br. The developmental impact of IM18 Br was claimed not only in this finding but also in the work of Luo et al. (2008), who investigated on D. magna. Other work in the literature has focused on the acute toxicity of ILs on rats and mice (Bailey et al., 2008; Cheng et al., 2009; Landry et al., 2005; Pernak and Czepukowicz, 2001; Sipes et al., 2008). The values of acute toxicity of 3-hexyloxymethyl1-methylimidazolium tetrafluoroborate were found to be LC50 ¼ 1400 and 1370 mg kg1 for female and male Wistar rats, respectively (Pernak and Czepukowicz, 2001). Bailey et al. (2008) studied the effects of prenatal exposure of mice to IM14 Cl due to the potential for human exposure as a result of water or soil contamination from industrial effluent or accidental spills. As shown in the experimental data, after being contacted to the IL, fetal weight was considerably reduced at the two highest concentrations (169 and 225 mg kg1 d1). Malformations were also somewhat more numerous at the highest dosage, suggesting that IM14 Cl may be teratogenic. Maternal toxicity was also present, indicating that IM14 Cl appeared to be developmentally toxic at maternally toxic dosages. Also, IM14 Cl has been shown to cause thermal irritation when applied topically to rats, but produced only minimal contact sensitization when evaluated in the mouse local lymph node assay (Landry et al., 2005). Additionally, in this report, it is worth noting that the transdermal toxicity of IM14 Cl was influenced by the vehicle of administration. Use of the organic solvent, dimethylformamide, accentuated the acute toxicity. Very high concentrations of IM14 Cl (up to 95% IM14 Cl in water) applied to the rat skin were markedly less acutely toxic. This result may have a practical guideline that to reduce the acute toxicity, ILs can be handled in pure form with water as a co-solvent.

Inhibitory effects of ILs on vertebrates

Zebrafish (Danio rerio) plays an important role in ecotoxicology as a prominent model vertebrate. Concerning toxicity of ILs to the zebrafish, Pretti et al. (2006) revealed that ILs may cause a completely different effect on fish according to their chemical structures. As imidazolium, pyridinium and pyrrolidinium showed a LC50 (lethal effect) >100 mg L1, they could be regarded as non-highly lethal towards zebrafish. On the other hand, the ammonium salts showed LC50 remarkably lower than that reported for organic solvents and tertiary amines. In general, these data referred that fish are less sensitive to ILs toxicity compared to other species belonging to lower trophic levels. In a recent report, Li et al. (2009) used the frog Rana nigromaculata as an amphibian model for toxicity testing. Amphibians are often the main vertebrate group prone to contaminant exposure in aquatic systems mostly because their larvae live in water (Lahr, 1997; Mann and Bidwell, 2000). In their study, they evaluated the toxic effects of IM18 Br on the early embryonic development of the frog R. nigromaculata. The results demonstrated that the highest embryonic mortality occurred in the neural plate stage, followed by the early gastrula and early cleavage stages with the LC50 values

3.

Environmental fate of ILs

3.1.

Chemical degradation of ILs

Ionic liquids possess excellent chemical and thermal stability, which gives, unfortunately, a negative aspect for their treatment after usage prior to disposal. To assess the persistence of ILs in the environment as well as verify possibilities of their cleanup by chemical methods, several groups have focused their attention on oxidative and thermal degradation of ILs in aqueous media (Awad et al., 2004; Baranyai et al., 2004; Berthon et al., 2006; Itakura et al., 2008; Li et al., 2007; Morawski et al., 2005; Siedlecka and Stepnowski, 2009; Siedlecka et al., 2008a,b; Stepnowski and Zaleska, 2005). Pioneering work in the field of oxidative degradation was done by Stepnowski and Zaleska (2005) and Morawski et al. (2005) who showed that the greatest degradation efficiency for imidazolium ILs was achieved with a combination of UV light and a catalytic oxidant such as hydrogen peroxide or titanium dioxide. Subsequently, Li et al. (2007) studied the oxidative degradation of 1,3-dialkylimidazolium ILs in hydrogen peroxide/acetic acid medium assisted by ultrasonic chemical irradiation. It was

water research 44 (2010) 352–372

observed that 99% of tested compounds was degraded after 72 h. In addition, advanced oxidative degradation in the presence of reactive peroxides generated by Fenton reagent has been applied for the removal of ILs from water (Siedlecka and Stepnowski, 2009; Siedlecka et al., 2008a,b). According to the results, in a Fenton system with 1 mM of Fe(III) and 100 mM of H2O2, more than 97% of IM14 Cl was observed to degrade after 90 min. For Pyr14 Cl, IM16 Cl and IM18 Cl, the levels of degradation were 92%, 88% and 68%, respectively. Investigations of the degradation mechanisms indicated IM18 Cl was more resistant to oxidation by OH radicals cleaved from H2O2, suggesting that the oxidation rates of imidazolium ILs by OH are structure-dependent (Siedlecka and Stepnowski, 2009). The level of degradation was dependent on the alkyl chain length, consistent with Stepnowski and Zaleska (2005), who indicated that lengthening the alkyl chain lowered the rate of IL degradation. On contrast, the different length of the side chains and the type of anions did not affect the degradation process (Li et al., 2007). Regarding the thermal degradation studies of alkylimidazolium salts (Awad et al., 2004), extension of the alkyl chain enhanced the thermo-oxidative degradation of imidazolium salts. Interestingly, methyl substitution in the 2-position (i.e. between the two N atoms) was observed to decrease the oxidative decomposition of imidazolium ILs. The longer alkyl chain was also observed to induce an enhancement in photocatalytic decomposition of ILs (Morawski et al., 2005), which was not in the case of oxidative degradation (Stepnowski and Zaleska, 2005; Siedlecka and Stepnowski, 2009). Nonetheless, detailed account on the degradation of ILs by photocatalysis is required to verify this phenomenon.

3.2.

Biodegradability of ILs

In contrast to chemical degradation, which requires the assistance of a certain oxidant for catalysis, biodegradation is the microbial breakdown of chemical compounds. Biodegradation seems to be more environmentally friendly compared to chemical decomposition process. The initial attempt to examine the degradation potential of different IM14 cations combined with [Br], [BF4], [PF6], [N(CN)2], [(CF3SO2)2N] and octylsulfate as the counter ion was done using the Sturm and Closed-Bottle test protocols by the group of Scammells (Garcia et al., 2005; Gathergood and Scammells, 2002; Gathergood et al., 2004; Gathergood et al., 2006). Nonetheless, no compound showed significant degree of biodegradation with the exception of the octylsulfate-containing IL. The next step study on the biodegradation of ILs involved the design of ILs containing biodegradable side chains (Gathergood and Scammells, 2002). The design was done according to the principles of Boethling (Boethling, 1994, 1996; Howard et al., 1991) who identified three important parameters including the potential sites of enzymatic hydrolysis (for example, esters and amides) and oxygen in form of hydroxyl, aldehyde or carboxylic acid groups as well as unsubstituted linear alkyl chains (especially 4 carbons) and phenyl rings, which represent possible sites for attack by oxygenases. However, for a balance between chemical properties and biodegradability, not all of these factors were suitable for ILs. The

365

addition of oxygen containing functional groups such as alcohols, aldehyde and carboxylic acids was reported to restrict the ILs performance as reaction media whereas the incorporation of phenyl rings was known to increase the melting points of IL solvents (McGuinness and Cavell, 2000). Therefore, ester or amide group was selected to be coupled in alkyl side chain of ILs. The introduction of ester groups derived from a C2 acid and C4 or higher alcohol in the 3-N-substitutent was demonstrated to increase the biodegradation of imidazolium-based ILs (Gathergood et al., 2004). This can be explained by the fact that introduction of ester moiety probably provides a site susceptible to enzymatic attack (Gathergood et al., 2004; Gathergood et al., 2006) and hence, improves the biodegradation level. Though the addition of amide group is informed to improve the biodegradation of organic compounds (Boethling, 1994, 1996; Howard et al., 1991), no critical enhancement of biological degradation was noted when this group was appended into the imidazoliumbased ILs (Gathergood et al., 2004). However, no compound could be classified as ‘‘readily biodegradable’’ corresponding to Organization for Economic Cooperation and Development (OECD) standards (U.S. EPA, 1998), for which 60–70% or greater biodegradation by activated sludge microbial inoculate is required within a 10-day window in a 28-day period. Finally, the combination of the octylsulfate anion and imidazolium cation containing ester side chains resulted in readily biodegradable IL (Gathergood et al., 2006). Recently, Stolte et al. (2008) also paid their attention on investigation of functional groups incorporating alkyl chain ILs. Nonetheless, the introductions of terminal hydroxyl, carboxyl, ether and nitrile groups did not improve the biological degradation as expected. Kumar et al. (2006) investigated the fate of IM14 BF4 when in contact with soil-microorganisms, wastewater microorganisms, Pseudomonas putida and E. coli. Although IM14 BF4 was indicated to be recalcitrant in Sturm and Closed-Bottle test assays as mentioned above, it was observed in this study that P. putida was able to break down IM14 BF4 after 15 days of incubation. The breakdown products were monitored using GC-MS and identified to be 1-H-methylimidazole and 1-H-butylimidazole, which were in consistent with the theoretical metabolism scheme proposed by Jastorff et al. (2003). In case of bacteria from soil and wastewater, the metabolic intermediates appeared on the 12th day. It was also noted that different intermediate peaks were observed at different retention time with different microbes, indicating that the degradation mechanism of IM14 BF4 may vary in correspondence to certain microbes and metabolic pathways. In another study, the biodegradation pathway of IM18 moiety (Fig. 3) was proposed based on intermediate products via HPLC-MS analysis after 24-day period of incubation with activated sludge (Stolte et al., 2008). The metabolism of IM18 cation appeared to undergo oxidation reactions catalyzed probably by mono-oxygenases, e.g. the cytochrome P450 system on the terminal methyl group (u-oxidation). The alcohol formed was subsequently oxidized and converted into aldehydes, and then into carboxylic acids by dehydrogenases. The resulting carboxylic acids then might undergo b-oxidation and finally generated two carbon fragments that can enter the tricarboxylic acid cycle as acetyl Co-A (Fig. 3).

366

water research 44 (2010) 352–372

retention time in min.

m/z+

intensity

8.9

195

4*105

13.5 / 14.4

211

3*105 / 2*105

m/z+

N

N

+

N

+

N

OH

OH

12.2 / 12.7

209

1*106 / 2*106

N

+

N

N

+

N

211

O

10.0 – 12.5

225

2*104 – 6*104

N

H

+

N

O

O

19.5

183

1*105

OH

N

N

+

N

+

N

209

OH

16.2

197

3*10

5

N

+

N

OH

26.5

155

0.5*105

OH

N

N

+

N OH

24.2

169

4*106

O +

O

N

N

+

225

+

N

OH O

26.7

141

1*105

N

N

+

O OH

Fig. 3 – Biodegradation pathways of 1-octyl-3-methylimidazolium by activated sludge microbial community (Reproduced from Stolte et al. (2008) by permission of the Royal Society of Chemistry).

The proposed pathways provide basic information to both environmental scientists and chemical engineers; however, no studies have sought to examine the toxicity of metabolic products after degradation of ILs. This issue is of paramount importance since metabolism might not always end in less toxic products. Subsequently, Wells and Coombe (2006) extended the microbial degradation study with ammonium, imidazolium, phosphonium and pyridinium compounds by measuring the biological oxygen demand. The authors observed no biodegradability of cations incorporated short chains (C 4) within this test series, which was in agreement with Docherty et al. (2007) and Stolte et al. (2008). For longer alkyl chains (C12, C16 and C18) containing ILs, a strong inhibitory effect of these compounds on the inoculum used was found, indicating the active microbial consortium was significantly impacted by ILs toxicity. In recent studies (Docherty et al., 2007; GrabinskaSota and Kalka, 2004; Harjani et al., 2008; Stasiewicz et al., 2008), pyridinium-based ILs were reported to be fully catabolized by microbial community in activated sludge. This can be inferred from the fact that degradation pathways for pyridine –

the precursor of pyridinium-based compounds – under aerobic and anaerobic conditions were intensively investigated in the work of Kaiser et al. (1996). With respect to the common 1,3-dialkylpyridinium ILs, Pham et al. (2009) reported that after 21 days of incubation, microorganisms from activated sludge were able to break down Py4-3Me Br. Analyses of HPLC and MS/MS demonstrated that this biodegradation led to the formation of 1-hydroxybutyl-3-methylpyridinium, 1-(2-hydroxybutal)-3-methylpyridinium, 1-(2-hydroxyethyl)3-methylpyridinium and methylpyridine. Based on these intermediate products, biodegradation pathways were also suggested (Fig. 4), thereby providing the basic information which might be useful for assessing the factors related to the environmental fate and behavior of this commonly used pyridinium IL. Although this is the first report on biodegradation intermediates and pathway of pyridinium ILs, the authors have failed to systematically screen a single microorganism or a microbial consortium responsible for biodegradation of ILs. Therefore, it is needed to further investigate which type of microorganism is adaptable to ILs and which is responsible for degradation process. Also, the broken

367

water research 44 (2010) 352–372

II H3C

+

N

H3C

CH3

+

N

CH3 OH

I H3C

OH

+

N

H3C

H3C

H3 C

OH

+

N

+

N

+

H

+

OH

+

O

+

N

H3 C

OH

H3C

+

N

+

N

CH3 H3C

H

+

H3C OH

H3C O

H3C OH

Fig. 4 – Biodegradation pathways of 1-butyl-3-methylpyridinium entity by microorganisms from activated sludge (Reprinted with permission from Pham et al. (2009). Copyright 2009 American Chemical Society).

structures much further than methylpyridinium were not measured in the study. In particular, the possibility of cleavage of heterocyclic ring in ILs molecular structure (both pyridinium in this work and imidazolium in the study of Stolte et al. (2008)) has not been indentified. This issue should be clarified through further studies. Interestingly, Py4-3Me Br was not found to be metabolized by the activated sludge community (Docherty et al., 2007). This was attributed likely to the high IL concentration used in the study of Docherty’s group, which consequently inhibited the microbial consortium. Nonetheless, it was demonstrated that the structural manipulation of the pyridinium skeleton may lead to ILs with greater biodegradable extent compared to imidazolium-based compouds (Harjani et al., 2008). Concerning the anionic effect, ILs with halide counter ions were postulated to be more stable to degradation than perfluoronated ions (Awad et al., 2004; Gathergood and Scammells, 2002). In a preliminary study, Gathergood and Scammells (2002) confirmed this assumption and showed that the biodegradation efficiency decreased in the order [PF6] > [BF4] > [Br] with 60%, 59% and 48% of CO2 evolution values, respectively. In a later study (Gathergood et al., 2006), it was found that the octylsulfate anion is considerably more biodegradable than the other commonly used anions. The alkyl chain with C4, C6 or C8 was found to increase the rate of degradation (Docherty et al., 2007; Stolte et al., 2008); nonetheless, further increasing the chain length to C12, C16 or C18 was noted to cause toxic effect towards inoculum (Wells and Coombe, 2006). However, it was stated that the long octyl side chain was not a compulsory factor for biodegradation, but more important is a certain overall lipophilicity of the compound (Stolte et al., 2008).

3.3.

Sorption of ILs in environmental systems

Because the aquatic and terrestrial environments are possible recipients for contaminants, the distribution and behavior of ILs in soil are also extremely important. The retention and mobility of ILs in soils and sediments are strongly influenced by its tendency to be sorbed onto the various components of the soil matrix (Stepnowski, 2005). Since imidazolium-based ILs possess high electron acceptor potential of delocalized aromatic systems and hydrophobic components (e.g. the alkyl chain) (Stepnowski, 2005), they could be sorbed onto soils and sediments via several mechanisms. In a preliminary study, Stepnowski (2005) proved that electrostatic interactions contributed to the sorption of the imidazolium cations. Moreover, totally contrast trends were also observed demonstrating an extremely strong and practically irreversible sorption onto fine-textured marine sediments and a relatively weakly and reversibly binding to peaty soil (with the highest organic carbon content) (Stepnowski, 2005). This indicates the importance of the mineral component of the soil (sediment) in the sorption mechanism of ILs. Also, in this work, the author pointed out that compounds with longer alkyl chains were irreversibly bound to the soil component, which was in agreement with Stepnowski et al. (2007) and Matzke et al. (2009b). Interestingly, Beaulieu et al. (2008) did not find a positive effect of alkyl chain length on the sorption of alkylimidazolium-based ILs to aquatic sediments and suggested that hydrophobic interactions were not the most important sorption mechanism. The contrast results between these groups imply that sorption mechanisms of ILs may vary according to properties and composition of the environmental systems. The studies suggest that ILs may be retained by aquatic sediments;

368

water research 44 (2010) 352–372

nonetheless, the toxic action of these sorbed materials towards aquatic organisms has not been addressed. Also, further efforts should be continued to elucidate the reversibility of IL sorption on these sediments to give a better understanding of the ILs fate after being released into aqueous environment. Matzke et al. (2009b) investigated the influences of the two different clay minerals kaolinite and smectite as well as of organic matter on the cation sorption and desorption behaviors of three imidazolium based ILs including IM14 BF4, IM18 BF4 and IM14 (CF3SO2)2N in soil. The addition of organic matter and clays was observed to increase the sorption/decrease the desorption of all ILs tested, and in particular smectite had striking effects on the sorption efficiency of all substances. It is worth noting that not only the cationic moiety with different alkyl side chain lengths but the anionic compartments can also play an importance role in the sorption/ desorption processes. Imidazolium compounds with BF 4 as a counter anion showed higher sorption capacity compared to that of [(CF3SO2)2N], indicating the high potential of this type of IL to form ionic pairs in the soil matrix (Matzke et al., 2009b). However, further work with a variety of ILs incorporated different cationic and anionic moieties should be carried out to verify these phenomena. Gorman-Lewis and Fein (2004) examined the sorption behavior of IM14 Cl onto a range of surfaces which are commonly found in the near-surface environment. The results suggested that IM14 Cl could be minimally retarded by noninterlayer clay system and might lead to unimpeded transport through subsurface groundwater. Also, the adsorption capacity of this IL onto bacterial surfaces was not high, which might be due to the low hydrophobicity of IM14 Cl. Additionally, investigations of the adsorption of IM14 Cl towards different media carried out by our group (Vijayaraghavan et al., 2009) have shown that retardation of this compound was possible only by an ion-exchange resin and activated carbon, which was in consistency with the work of Anthony et al. (2001). However, no significant adsorption of IM14 Cl in the media of a fermentation waste (Corynebacterium glutamicum) and dried activated sludge was observed in our study. Conclusively, the data currently available demonstrated that ILs incorporated imidazolium cation can be sorbed to organic matter, whether found in aquatic sediments or terrestrial soils, and the presence of clays significantly enhanced the sorption capacity.

4.

Concluding remarks and future directions

Ionic liquids, of which the most often cited attribute is their negligible vapor pressure, have been suggested as a green alternative to traditional organic solvents with the desire to minimize diffusion to the atmosphere. Low volatility, however, does not completely eliminate potential environmental hazards and might pose serious threats to aquatic and terrestrial ecosystems. The studies of environmental fate and toxicity of ILs have shown that the ILs commonly used to date are toxic in nature and their toxicities vary considerably across organisms and trophic levels. In general, the effect of anionic moieties is not drastic as the alkyl length effect except for the case of [(CF3SO2)2N], which shows a clear (eco)toxicological

hazard potential. The other perfluorinated anions have been also proved to be hazardous due to hydrolytically unstable properties. In addition, the introduction of functional polar groups to the alkyl chain has been shown to reduce the toxicity of ILs and increase the biodegradation efficiency to some extent. This indicates the possibility of tailoring ILs by coupling suitable functional groups to their structure, which in turn leads to a more environmental friendly compound. The side chain length effect has been found to be consistent in all levels of biological complexity as well as different environmental compartments. Also, an increase in alkyl-chain length, or lipophilicity, was observed to be related to an increase in the rate of degradation as well as an increase in toxicity. This indicates a conflict of aims between minimizing the toxicity and maximizing the biodegradability of these neoteric solvents. Regarding the cationic compartment, pyridinium has been found to be more environmental friendly than imidazolium from both viewpoints of toxicology and microbial degradation. It can therefore be suggested that the structural manipulation of the pyridinium skeleton should be considered in design of a sustainable IL. From the currently available data, it is clear that some commonly used ILs are very far away from the image of green chemicals that are often cited in the literature. The uncertainties in their sustainable development hinder the applications of ILs under real conditions. Although some attempts have been made to give important hints in the prospective design and synthesis of inherently safer ILs, comprehensive studies dealing with the behaviors of ILs in aqueous media still await to be conducted. The important features required for the thorough insight into environmental fate of ILs include, but are not limited to: Providing more fundamental understanding into the mechanism for IL-induced toxicity to different levels of biological complexity. The underlying mechanisms of IL toxicity have rarely been studied. - Assessing the biodegradability of cationic and anionic compartments and toxicity of their degradation intermediates. This may provide useful information in consciously designing safer chemicals. - Investigating the aerobic and anaerobic biodegradation of ILs, which would suggest initial guidelines for the treatment of ILs waste by using the existing aerobic and anaerobic wastewater treatment facilities. Especially, anaerobic degradation awaits to be investigated. - Defining which organisms or enzymes may promote degradation pathways and determining specific microbial consortium or cultivatable communities capable of biotransformation of ILs. - Performing the ecotoxicity and biodegradation tests in real environmental conditions instead of controlled conditions of laboratory experiments, which would be advantageous in understanding the fate and behavior of ILs under real conditions. For this, the potential toxicological effects at population level and community level should be addressed. It must be encouraged to use tools such as experimental mesocosms to study the effects of ILs at higher levels of organization. - Creating database of environmentally benign structure moieties of ILs based upon their toxicological and -

water research 44 (2010) 352–372

biodegradation information, which would be practically useful as a reference for manufacturers and regulators to properly develop and regulate the use of ILs.

Acknowledgements This work was supported by NRF Grant funded by the Korean Government (KRF-2007-521-D00106, NRL 2009-0083194, and in part WCU R31-2008-000-20029-0).

references

Anthony, J.L., Maginn, E.J., Brennecke, J.F., 2001. Solution thermodynamics of imidazolium-based ionic liquids and water. J. Phys. Chem. B 105, 10942–10949. Arning, J., Stolte, S., Bo¨schen, A., Stock, F., Pitner, W.-R., WelzBiermann, U., Jastorff, B., Ranke, J., 2008. Qualitative and quantitative structure activity relationships for the inhibitory effects of cationic head groups, functionalized side chains and anions of ionic liquids on acetylcholinesterase. Green Chem. 10, 47–58. Austin, R.P., Barton, P., Davis, A.M., Manners, C.N., Stansfield, M.C., 1998. The effect of ionic strength on liposome-buffer and 1-octanol-buffer distribution coefficients. J. Pharm. Sci. 87, 599–607. Awad, W.H., Gilman, J.W., Nyden, M., Harris, R.H., Sutto, T.E., Callahan, J., Trulove, P.C., DeLong, H.C., Fox, D.M., 2004. Thermal degradation studies of alkyl-imidazolium salts and their application in nanocomposites. Thermochim. Acta 409, 3–11. Babalola, G.O., 1998. Anti-bacterial activity of synthetic N-heterocyclic oxidizing compounds. Lett. Appl. Microbiol. 26, 43–46. Bailey, M.M., Townsend, M.B., Jernigan, P.L., Sturdivant, J., HoughTroutman, W.L., Rasco, J.F., Swatloski, R.P., Rogers, R.D., Hood, R.D., 2008. Developmental toxicity assessment of the ionic liquid 1-butyl-3-methylimidazolium chloride in CD-1 mice. Green Chem. 10, 1213–1217. Ba1czewski, P., Bachowska, B., Bia1as, T., Biczak, R., Wieczorek, W. M., Balin´ska, A., 2007. Synthesis and phytotoxicity of new ionic liquids incorporating chiral cations and/or chiral anions. J. Agric. Food Chem. 55, 1881–1892. Baranyai, K.J., Deacon, G.B., MacFarlane, D.R., Pringle, J.M., Scott, J.L. , 2004. Thermal degradation of ionic liquids at elevated temperatures. Aust. J. Chem. 57, 145–147. Beaulieu, J.J., Tank, J.L., Kopacz, M., 2008. Sorption of imidazoliumbased ionic liquids to aquatic sediments. Chemosphere 70, 1320–1328. Bernot, R.J., Turner, A.M., 2001. Predator identity and traitmediated indirect effects in a littoral food web. Oecologia 129, 139–146. Bernot, R.J., Brueseke, M.A., Evans-White, M.A., Lamberti, G.A., 2005a. Acute and chronic toxicity of imidazolium-based ionic liquids on Daphnia magna. Environ. Toxicol. Chem. 24, 87–92. Bernot, R.J., Kennedy, E.E., Lamberti, G.A., 2005b. Effects of ionic liquids on the survival, movement, and feeding behavior of the freshwater snail, Physa acuta. Environ. Toxicol. Chem. 24, 1759–1765. Berthon, L., Nikitenko, S.I., Bisel, I., Berthon, C., Faucon, M., Saucerotte, B., Zorz, N., Moisy, Ph, 2006. Influence of gamma

369

irradiation on hydrophobic room-temperature ionic liquids [BuMeIm] PF6 and [BuMeIm] (CF3SO2)2N. Dalton Trans, 2526–2534. Blaauboer, B.J., Balls, M., Barratt, M., Casati, S., Coecke, S., Mohamed, M.K., Moore, J., Rall, D., Smith, K.R., Tennant, R., Schwetz, B.A., Stokes, W.S., Younes, M., 1998. Alternative testing methodologies and conceptual issues. Environ. Health Perspect 106, 413–418. Blaise, C.R., 1993. In: Richardson, M.L. (Ed.), Ecotoxicology Monitoring. VCH, Weinheim, pp. 83–108. Blanchard, L.A., Hancu, D., Beckman, E.J., Brennecke, J.F., 1999. Green processing using ionic liquids and CO2. Nature 399, 28–29. Blanchard, L.A., Brennecke, J.F., 2001. Recovery of organic products from ionic liquids using supercritical carbon dioxide. Ind. Eng. Chem. Res. 40, 287–292. Boethling, R.S., 1994. Cationic Surfactants, Surfactant Science Series, vol. 53. Marcel Dekker, New York, pp. 95–135. Boethling, R.S., 1996. Designing safer chemicals. ACS Symp. Ser. 640, 156. Chemnitius, J.-M., Sadowski, R., Winkel, H., Zech, R., 1999. Organophosphate inhibition of human heart muscle cholinesterase isoenzymes. Chem. Biol. Interact 120, 183–192. Cheng, Y., Wright, S.H., Hooth, M.J., Sipes, I.G., 2009. Characterization of the disposition and toxicokinetics of N-butylpyridinium chloride in male F-344 rats and female B6C3F1 mice and its transport by organic cation transporter 2. Drug Metab. Dispos. 37, 909–916. Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Vijayaraghavan, K., Choe, W.-S., Yun, Y.-S., 2007. Toxicity of imidazolium salt with anion bromide to a phytoplankton Selenastrum capricornutum: effect of alkylchain length. Chemosphere 69, 1003–1007. Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Yun, Y.-S., 2008a. Influence of anions on the toxic effects of ionic liquids to a phytoplankton Selenastrum capricornutum. Green Chem. 10, 67–72. Cho, C.-W., Jeon, Y.-C., Pham, T.P.T., Vijayaraghavan, K., Yun, Y.-S., 2008b. The ecotoxicity of ionic liquids and traditional organic solvents on microalga Selenastrum capricornutum. Ecotoxicol. Environ. Saf. 71, 166–171. Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Min, J., Jung, H.Y., Lee, D.S., Yun, Y.-S., 2008c. Microalgal photosynthetic activity measurement system for rapid toxicity assessment. Ecotoxicology 17, 455–463. Cieniecka-Ros1onkiewicz, A., Pernak, J., Kubis-Feder, J., Ramani, A., Robertson, A.J., Seddon, K.R., 2005. Synthesis, anti-microbial activities and anti-electrostatic properties of phosphoniumbased ionic liquids. Green Chem. 7, 855–862. Costello, D.M., Brown, L.M., Lamberti, G.A., 2009. Acute toxic effects of ionic liquids on zebra mussel (Dreissena polymorpha) surviving and feeding. Green Chem. 11, 548–553. Couling, D.J., Bernot, R.J., Docherty, K.M., Dixon, J.K., Maginn, E.J., 2006. Assessing the factors responsible for ionic liquid toxicity to aquatic organisms via quantitative structure-property relationship modeling. Green Chem. 8, 82–90. Dipeolu, O., Green, E., Stephens, G., 2008. Effects of watermiscible ionic liquids on cell growth and nitro reduction using Clostridium sporogenes. Green Chem. 11, 397–401. Docherty, K.M., Kulpa, C.F., 2005. Toxicity and antimicrobial activity of imidazolium and pyridinium ionic liquids. Green Chem. 7, 185–189. Docherty, K.M., Dixon, J.K., Kulpa, C.F., 2007. Biodegradability of imidazolium and pyridinium ionic liquids by an activated sludge microbial community. Biodegradation 18, 481–493. Earle, M.J., Seddon, K.R., 2000. Ionic liquids. Green solvents for the future. Pure Appl. Chem. 72, 1391–1398. Frade, R.F.M., Matias, A., Branco, L.C., Afonso, C.A.M., Duarte, C. M.M., 2007. Effect of ionic liquids on human colon carcinoma HT-29 and CaCo-2 cell lines. Green Chem. 9, 873–877. Fuler, J., Carlin, R.T., Osteryoung, R.A., 1997. The room temperature ionic liquids 1-ethyl-3-methylimidazolium tetrafluoroborate:

370

water research 44 (2010) 352–372

electrochemical couples and physical properties. J. Electrochem. Soc. 144, 3881–3886. Fulton, M.H., Key, P.B., 2001. Acetylcholinesterase inhibition in estuarine fish and invertebrates as an indicator of organophosphorus insecticide exposure and effects. Environ. Toxicol. Chem. 20, 37–45. Ganske, F., Bornscheuer, U.T., 2006. Growth of Escherichia coli, Pichia pastoris and Bacillus cereus in the presence of the ionic liquids [BMIM] [BF4] and [BMIM] [PF6] and organic solvents. Biotechnol. Lett. 28, 465–469. Garcia, M.T., Gathergood, N., Scammells, P.J., 2005. Biodegradable ionic liquids. part II: effect of the anion and toxicology. Green Chem. 7, 9–14. Garcı´a-Lorenzo, A., Tojo, E., Tojo, J., Teijeira, M., Rodrı´guezBerrocal, F.J., Gonza´lez, M.P., Martı´nez-Zorzano, V.S., 2008. Cytotoxicity of selected imidazolium-derived ionic liquids in the human Caco-2 cell line. Sub-structural toxicological interpretation through a QSAR study. Green Chem. 10, 508–516. Gathergood, N., Scammells, P.J., 2002. Design and preparation of room-temperature ionic liquids containing biodegradable side chains. Aust. J. Chem. 55, 557–560. Gathergood, N., Garcia, M.T., Scammells, P.J., 2004. Biodegradable ionic liquids: part I. Concept, preliminary targets and evaluation. Green Chem. 6, 166–175. Gathergood, N., Scammells, P.J., Garcia, M.T., 2006. Biodegradable ionic liquids. Part III: the first readily biodegradable ionic liquids. Green Chem. 8, 156–160. Gordon, C.M., 2001. New developments in catalysis using ionic liquids. Appl. Catal. A 222, 101–117. Gorman-Lewis, D.J., Fein, J.B., 2004. Experimental study of the adsorption of an ionic liquid onto bacterial and mineral surfaces. Environ. Sci. Technol 38, 2491–2495. Grabinska-Sota, E., Kalka, J., 2004. An assessment of the toxicity of pyridinium chlorides and their biodegradation intermediates. Environ. Int 28, 687–690. Grabinska-Sota, E., Kalka, J., 2006. Toxicity of imidazolium chlorides to aquatic organisms. Polish J. Environ. Stud. 15, 405–409. Hagiwara, R., Ito, Y., 2000. Room temperature ionic liquids of alkylimidazolium cations and fluoroanions. J. Fluorine Chem. 105, 221–227. Harjani, J.R., Singer, R.D., Garcia, M.T., Scammells, P.J., 2008. The design and synthesis of biodegradable pyridinium ionic liquids. Green Chem. 10, 436–438. Hassoun, E.A., Abraham, M., Kini, V., Al-Ghafri, M., Abushaban, A., 2002. Cytotoxicity of the ionic liquid, 1- N-butyl-3methylimidazolium chloride. Res. Commun, Pharmacol. Toxicol 7, 23–31. Howard, P.H., Boethling, R.S., Stiteler, W., Meylan, W., Beauman, J., 1991. Development of a predictive model for biodegradability based on BIODEG, the evaluated biodegradation data base. Sci. Total Environ. 110, 635–641. Huddleston, J.G., Visser, A.E., Reichert, W.M., Willauer, H.D., Broker, G.A., Rogers, R.D., 2001. Characterization and comparison of hydrophilic and hydrophobic room temperature ionic liquids incorporating the imidazolium cation. Green Chem. 3, 156–164. Itakura, T., Hirata, K., Aoki, M., Sasai, R., Yoshida, H., Itoh, H., 2008. Decompostition and removal of ionic liquid in aqueous solution by hydrothermal and photocatalytic treatment. Environ. Chem. Lett. doi:10.1007/s10311-008-0177-7. Jastorff, B., Sto¨rmann, R., Ranke, J., Mo¨lter, K., Stock, F., Oberheitmann, B., Hoffmann, W., Hoffmann, J., Nu¨chter, M., Ondruschka, B., Filser, J., 2003. How hazardous are ionic liquids? Structure-activity relationships and biological testing as important elements for sustainability evaluation. Green Chem. 5, 136–142. Jastorff, B., Mo¨lter, K., Behrend, P., Bottin-Weber, U., Filser, J., Heimers, A., Ondruschka, B., Ranke, J., Schaefer, M.,

Schro¨der, H., Stark, A., Stepnowski, P., Stock, F., Sto¨rmann, R., Stolte, S., Welz-Biermann, U., Ziegert, S., Tho¨ming, J., 2005. Progress in evaluation of risk potential of ionic liquids-basis for an eco-design of sustainable products. Green Chem. 7, 362–372. Kaiser, K.L.E., Palabrica, V.S., 1991. Photobacterium phosphoreum, toxicity data index. Water Poll. Res. J. Can. 26, 361–431. Kaiser, J.P., Feng, Y.C., Bollag, J.M., 1996. Microbial metabolism of pyridine, quinoline, acridine, and their derivatives under aerobic and anaerobic conditions. Microbiol. Rev. 60, 483–498. Kazarian, S.G., Briscoe, B.J., Welton, T., 2000. Combining ionic liquids and supercritical fluids: in situ ATR-IR study of CO2 dissolved in two ionic liquids at high pressures. Chem. Commun., 2047–2048. Kelman, D., Kashman, Y., Rosenberg, E., Ilan, M., Ifrach, I., Loya, Y., 2001. Antimicrobial activity of the reef sponge Amphimedon viridis from the Red Sea: evidence for selective toxicity. Aquat. Microb. Ecol 24, 9–16. Kulacki, K.J., Lamberti, G.A., 2008. Toxicity of imidazolium ionic liquids to freshwater algae. Green Chem. 10, 104–110. Kumar, R.A., Papaı¨conomou, N., Lee, J.-M., Salminen, J., Clark, D.S., Prausnitz, J.M., 2009. In vitro cytotoxicities of ionic liquids: effect of cation rings, functional groups, and anions. Environ. Toxicol. 24, 388–395. Kumar, S., Ruth, W., Sprenger, B., Kragl, U., 2006. On the biodegradation of ionic liquid 1-butyl-3-methylimidazolium tetrafluoroborate. Chim. Oggi 24, 24–26. Lahr, J., 1997. Ecotoxicology of organisms adapted to life in temporary freshwater ponds in arid and semi-arid regions. Arch. Environ. Contam. Toxicol 32, 50–57. Landry, T.D., Brooks, K., Poche, D., Woolhiser, M., 2005. Acute toxicity profile of 1-butyl-3-methylimidazolium chloride. Bull. Environ. Contam. Toxicol 74, 559–565. Larson, J.H., Frost, P.C., Lamberti, G.A., 2008. Variable toxicity of ionic liquid-forming chemicals to Lemna minor and the influence of dissolved organic matter. Environ. Toxicol. Chem. 27, 676–681. Lata1a, A., Stepnowski, P., Nedzi, ˛ M., Mrozik, W., 2005. Marine toxicity assessment of imidazolium ionic liquids: acute effects on the Baltic algae Oocystis submarina and Cyclotella meneghiniana. Aqua. Toxicol 73, 91–98. Lee, S.-M., Chang, W.-J., Choi, A.-R., Koo, Y.-M., 2005. Influence of ionic liquids on the growth of Escherichia coli. Korean. J. Chem. Eng 22, 687–690. Lewis, M.A., 1995. In: Rand, G.M. (Ed.), Fundamentals of Aquatic Toxicology: Effects, Environment Fate, and Risk Assessment, second ed. Taylor and Francis, Washington, DC, USA, pp. 135–170. Li, G., Shen, J., Zhu, Y., 1998. Study of pyridinium-type functional polymers. II. Antibacterial activity of soluble pyridinium-type polymers. J. Appl. Polym. Sci. 67, 1761–1768. Li, X., Zhao, J., Li, Q., Wang, L., Tsang, S.C., 2007. Ultrasonic chemical oxidative degradations of 1,3-dialkylimidazolium ionic liquids and their mechanistic elucidations. Dalton Trans, 1875–1880. Li, X.-Y., Zhou, J., Yu, M., Wang, J.-J., Pei, Y.C., 2009. Toxic effects of 1-methyl-3-octylimidazolium bromide on the early embryonic development of the frog Rana nigromaculata. Ecotox. Environ. Saf. 72, 552–556. Luis, P., Ortiz, I., Alcado, R., Irabien, A., 2007. A novel group contribution method in the development of a QSAR for predicting the toxicity (Vibrio fischeri EC50) of ionic liquids. Ecotoxicol. Environ. Saf 67, 423–429. Luo, Y.-R., Li, X.-Y., Chen, X.-X., Zhang, B.-J., Sun, Z.-J., Wang, J.-J., 2008. The developmental toxicity of 1-butyl-3-octylimidazolium bromide on Daphnia magna. Environ. Toxicol 23, 736–744. Malich, G., Markovic, B., Winder, C., 1997. The sensitivity and specificity of the MTS tetrazolium assay for detecting the in

water research 44 (2010) 352–372

vitro cytotoxicity of 20 chemicals using human cell lines. Toxicology 124, 179–192. Mann, R.M., Bidwell, J.R., 2000. Application of the FETAX protocol to assess the development toxicity of nonylphenol ethoxylate to Xenopus laevis and two Australian frogs. Aquat. Toxicol 51, 19–29. Marsh, K.N., Boxall, J.A., Lichtenthaler, R., 2004. Room temperature ionic liquids and their mixtures–a review. Fluid Phase Equilibr 219, 93–98. Massoulie´, J., Pezzementi, L., Bon, S., Krejci, E., Vallette, F.-M., 1993. Molecular and cellular biology of cholinesterases. Prog. Neurobiol. 41, 31–91. Matsumoto, M., Mochiduki, K., Fukunishi, K., Kondo, K., 2004a. Extraction of organic acids using imidazolium-based ionic liquids and their toxicity to Lactobacillus rhamnosus. Sep.Purif. Technol. 40, 97–101. Matsumoto, M., Mochiduki, K., Kondo, K., 2004b. Toxicity of ionic liquids and organic solvents to lactic acid-producing bacteria. J. Biosci. Bioeng 98, 344–347. Matzke, M., Stolte, S., Thiele, K., Juffernholz, T., Arning, J., Ranke, J., Welz-Biermann, U., Jastorff, B., 2007. The influence of anion species on the toxicity of 1-alkyl-3-methylimidazolium ionic liquids observed in an (eco)toxicological test battery. Green Chem. 9, 1198–1207. Matzke, M., Stolte, S., Bo¨schen, A., Filser, J., 2008. Mixture effects and predictability of combination effects of imidazolium based ionic liquids as well as imidazolium based ionic liquids and cadmium on terrestrial plants (Triticum aestivum) and limnic green algae (Scenedesmus vacuolatus). Green Chem. 10, 784–792. Matzke, M., Stolte, S., Arning, J., Uebers, U., Filser, J., 2009a. Ionic liquids in soils: effects of different anion species of imidazolium based ionic liquids on wheat (Triticum aestivum) as affected by different clay minerals and clay concentrations. Ecotoxicology 18, 197–203. Matzke, M., Thiele, K., Mu¨ller, A., Filser, J., 2009b. Sorption and desorption of imidazolium based ILs in different soil types. Chemosphere 74, 568–574. McFarlane, J., Ridenour, W.B., Luo, H., Hunt, R.D., Depaoli, D.W., Ren, R.X., 2005. Room temperature ionic liquids for separating organics from produced water. Sep. Sci. Technol 40, 1245–1265. McGuinness, D.S., Cavell, K.J., 2000. Donor-functionalized heterocyclic carbine complexes of palladium(II): efficient catalysts for C-C coupling reactions. Organometallics 19, 741–748. McQueen, D.J., Post, J.R., Mills, E.L., 1986. Trophic relationships in freshwater pelagic ecosystems. Can. J. Fish. Aqua. Sci. 43, 1571–1581. Morawski, A.W., Janus, M., Goc-Maciejewska, I., Syguda, A., Pernak, J., 2005. Decomposition of ionic liquids by photocatalysis. Pol. J. Chem. 79, 1929–1935. Myasoedov, B.F., Molochnikova, N.P., Shkinev, V.M., Spivakov, B.Y., 1995. The Behavior of Actinides in Two-Phase Aqueous Systems Based on Polyethylene Glycol. Plenum, New York, pp. 91. Nockemann, P., Thijs, B., Driesen, K., Janssen, C.R., Hecke, K.V., Meervelt, L.V., Kossmann, S., Kirchner, B., Binnemans, K., 2007. Choline saccharinate and choline acesulfamate: ionic liquids with low toxicities. J. Phys. Chem. B 111, 5254–5263. Olivier, H., 1999. Recent developments in the use of non-aqueous ionic liquids for two-phase catalysis. J. Mol. Cat. A: Chem. 146, 285–289. Osenberg, C.W., Mittelbach, G.G., 1989. Effects of body size on the predator–prey interaction between pumpkinseed sunfish and gastropods. Ecol. Monogr. 59, 405–432. Pernak, J., Czepukowicz, A., 2001. New ionic liquids and their antielectrostatic properties. Ind. Eng. Chem. Res. 40, 2379– 2383.

371

_ J., Mirska, I., 2001a. Synthesis and Pernak, J., Rogoza, antimicrobial activities of new pyridinium and benzimidazolium chlorides. Eur. J. Med. Chem. 36, 313–320. Pernak, J., Kalewska, J., Ksycin´ska, H., Cybulski, J., 2001b. Synthesis and anti-microbial activities of some pyridinium salts with alkoxymethyl hydrophobic group. Eur. J. Med. Chem. 36, 899–907. Pernak, J., Chwa1a, P., 2003. Synthesis and anti-microbial activities of choline-like quaternary ammonium chlorides. Eur. J. Med. Chem. 38, 1035–1042. Pernak, J., Sobaszkiewicz, K., Mirska, I., 2003. Anti-microbial activities of ionic liquids. Green Chem. 5, 52–56. Pernak, J., Goc, I., Mirska, I., 2004a. Anti-microbial activities of protic ionic liquids with lactate anion. Green Chem. 6, 323–329. Pernak, J., Sobaszkiewicz, K., Foksowicz-Flaczyk, J., 2004b. Ionic liquids with symmetrical dialkoxymethyl-substituted imidazolium cations. Chem. Eur. J 10, 3479–3485. Pham, T.P.T., Cho, C.-W., Vijayaraghavan, K., Min, J., Yun, Y.-S., 2008a. Effect of imidazolium-based ionic liquids on the photosynthetic activity and growth rate of Selenastrum capricornutum. Environ. Toxicol. Chem. 27, 1583–1589. Pham, T.P.T., Cho, C.-W., Min, J., Yun, Y.-S., 2008b. Alkyl-chain length effects of imidazolium and pyridinium ionic liquids on photosynthetic response of Pseudokirchneriella subcapitata. J. Biosci. Bioeng 105, 425–428. Pham, T.P.T., Cho, C.-W., Jeon, C.-O., Chung, Y.-J., Lee, M.-W., Yun, Y.-S., 2009. Identification of metabolites involved in the biodegradation of the ionic liquid 1-butyl-3-methylpyridinium bromide by activated sludge microorganisms. Environ. Sci. Technol 43, 516–521. Peraccini, D., Chiappe, C., Intorre, L., Pretti, C., 2007. In: Letcher, T. M. (Ed.), Thermodynamics, Solubility and Environmental Issues. Elsevier, UK, pp. 259–278. Pope, C., Karanth, S., Liu, J., 2005. Pharmacology and toxicology of cholinesterase inhibitors: uses and misuses of a common mechanism of action. Environ. Toxicol. Pharmacol 19, 433–446. Pretti, C., Chiappe, C., Pieraccini, D., Gregori, M., Abramo, F., Monni, G., Intorre, L., 2006. Acute toxicity of ionic liquids to the zebrafish (Danio rerio). Green Chem. 8, 238–240. Pretti, C., Chiappe, C., Baldetti, I., Brunini, S., Monni, G., Intorre, L., 2009. Acute toxicity of ionic liquids for three freshwater organisms: Pseudokirchneriella subcapitata, Daphnia magna and Danio rerio. Ecotoxicol. Environ. Saf. 72, 1170–1176. Ranke, J., Mo¨lter, K., Stock, F., Bottin-Weber, U., Poczobutt, J., Hoffmann, J., Ondruschka, B., Filser, J., Jastorff, B., 2004. Biological effects of imidazolium ionic liquids with varying chain length in acute Vibrio fischeri and WST-1 cell viability assays. Ecotox. Environ. Saf 58, 396–404. Ranke, J., Mu¨ller, A., Bottin-Weber, U., Stock, F., Stolte, S., Arning, J., Sto¨rmann, R., Jastorff, B., 2007a. Lipophilicity parameters for ionic liquid cations and their correlation to in vitro cytotoxicity. Ecotox. Environ. Saf 67, 430–438. Ranke, J., Stolte, S., Sto¨rmann, R., Arning, J., Jastorff, B., 2007b. Design of sustainable chemical products – the example of ionic liquids. Chem. Rev. 107, 2183–2206. Rogers, R.D., Bond, A.H., Bauer, C.B., Zhang, J., Rein, S.D., Chomko, R.R., Roden, D.M., 1995. Partitioning behavior of 99 Tc and, 129 I from simulated Handford tank wastes using polyethylene-glycol based aqueous biphasic systems. Solvent Extr. Ion Exch. 13, 689–713. Romero, A., Santos, A., Tojo, J., Rodrı´guez, A., 2008. Toxicity and biodegradability of imidazolium ionic liquids. J. Hazard. Mater 151, 268–273. Rooney, D.W., Seddon, K.R., 2001. In: Wypych, G. (Ed.), Handbook of Solvents. ChemTech Publishing, Toronto, pp. 1459–1484. Salminen, J., Papaiconomou, N., Kumar, R.A., Lee, J.-M., Kerr, J., Newman, J., Prausnitz, J.M., 2007. Physicochemical properties and toxicities of hydrophobic piperidinium and pyrrolidinium ionic liquids. Fluid Phase Equilibr 261, 421–426.

372

water research 44 (2010) 352–372

Samorı`, C., Pasteris, A., Galletti, P., Tagliavini, E., 2007. Acute toxicity of oxygenated and nonoxygenated imidazoliumbased ionic liquids to Daphnia magna and Vibrio fischeri. Environ. Toxicol. Chem. 26, 2379–2382. Schmid, A., Kollmer, A., Mathys, R.G., Withot, B., 1998. Development toward large-scale bacterial bioprocesses in the presence of bulk amounts of organic solvents. Extremophiles 2, 249–256. Sheldon, R., 2001. Catalytic reactions in ionic liquids. Chem. Commun, 2399–2407. Sheldon, R.A., 2005. Green solvents for sustainable organic synthesis: state of the art. Green Chem. 7, 267–278. Siedlecka, E.M., Stepnowski, P., 2009. The effect of alkyl chain length on the degradation of alkylimidazolium- and pyridinium-type ionic liquids in a Fenton-like system. Environ. Sci. Pollut. Res. 16, 453–458. Siedlecka, E.M., Mrozik, W., Kaczyn´ski, Z., Stepnowski, P., 2008a. Degradation of 1-butyl-3-methylimidazolium chloride ionic liquid in a Fenton-like system. J. Hazard. Mater 154, 893–900. Siedlecka, E.M., Go1ebiowski, ˛ M., Kumirska, J., Stepnowski, P., 2008b. Identification of 1-butyl-3-methylimidazolium chloride degradation products formed in Fe(III)/H2O2 oxidation system. Chem. Anal. (Warsaw) 53, 943–951. Sipes, I.G., Knudsen, G.A., Kuester, R.K., 2008. The effects of dose and route on the toxicokinetics and disposition of 1-butyl3-methylimidazolium chloride in male F-344 rats and female B6C3F1 mice. Drug Metab. Dispos. 36, 284–293. Sk1adanowski, A.C., Stepnowski, P., Kleszczyn´ski, K., Dmochowska, B., 2005. AMP deaminase in vitro inhibition by xenobiotics. A potential molecular method for risk assessment of synthetic nitro- and polycyclic musks, imidazolium ionic liquids and N-glucopyranosyl ammonium salts. Environ. Toxicol. Phar. 19, 291–296. Stasiewicz, M., Mulkiewicz, E., Tomczak-Wandzel, R., Kurmirska, J., Siedlecka, E.M., Go1ebiowski, M., Gajdus, J., Czerwicka, M., Stepnowski, P., 2008. Assessing toxicity and biodegradation of novel, environmentally benign ionic liquids (1-alkoxymethyl-3-hydroxypyridinium chloride, saccharine and acesulfamates) on cellular and molecular level. Ecotox. Environ. Saf 71, 157–165. Steinberg, S.M., Poziomek, E.J., Engelmann, W.H., Rogers, K.R., 1995. A review of environmental applications of bioluminescence measurements. Chemosphere 30, 2155–2197. Stepnowski, P., Sk1adanowski, A.C., Ludwiczak, A., qaczyn´ska, E., 2004. Evaluating the cytotoxicity of ionic liquids using human cell line HeLa. Hum. Exp. Toxicol 23, 513–517. Stepnowski, P., 2005. Preliminary assessment of the sorption of some alkyl imidazolium cations as used in ionic liquids to soils and sediments. Aust. J. Chem. 58, 170–173. Stepnowski, P., Zaleska, A., 2005. Comparison of different advanced oxidation processes for the degradation of room temperature ionic liquids. J. Photoch. Photobio. A 170, 45–50. Stepnowski, P., Mrozik, W., Nichthauser, J., 2007. Adsorption of alkylimidazolium and alkylpyridinium ionic liquids onto natural soils. Environ. Sci. Technol 41, 511–516. Stock, F., Hoffmann, J., Ranke, J., Sto¨rmann, R., Ondruschka, B., Jastorff, B., 2004. Effects of ionic liquids on the acetycholinesterase–a structure-activity relationship consideration. Green Chem. 6, 286–290. Stolte, S., Arning, J., Bottin-Weber, U., Matzke, M., Stock, F., Thiele, K., Uerdingen, M., Welz-Biermann, U., Jastorff, B., Ranke, J., 2006. Anion effects on the cytotoxicity of ionic liquids. Green Chem. 8, 621–629. Stolte, S., Matzke, M., Arning, J., Bo¨schen, A., Pitner, W.-R., Welz-Biermann, U., Jastorff, B., Ranke, J., 2007a. Effects of

different head groups and functionalized side chains on the aquatic toxicity of ionic liquids. Green Chem. 9, 1170–1179. Stolte, S., Arning, J., Bottin-Weber, U., Mu¨ller, A., Pitner, W.-R., Welz-Biermann, U., Jastorff, B., Ranke, J., 2007b. Effects of different head groups and functionalized side chains on the cytotoxicity of ionic liquids. Green Chem. 9, 760–767. Stolte, S., Abdulkarim, S., Arning, J., Blomeyer-Nienstedt, A.-K., Bottin-Weber, U., Matzke, M., Ranke, J., Jastorff, B., Tho¨ming, J., 2008. Primary biodegradation of ionic liquid cations, identification of degradation products of 1-methyl-3-octylimidazolium chloride and electrochemical wastewater treatment of poorly biodegradable compounds. Green Chem. 10, 214–224. ´ ska, S., Buszewski, B., 2009. Study of toxicity of imidazolium Studzin ionic liquids to watercress (Lepidium sativum L.). Anal. Bioanal. Chem. 393, 983–990. Swatloski, R.P., Holbrey, J.D., Rogers, R.D., 2003. Ionic liquids are not always green: hydrolysis of 1-butyl-3-methylimidazolium hexafluorophosphate. Green Chem. 5, 361–363. Swatloski, R.P., Holbrey, J.D., Memon, S.B., Caldwell, G.A., Caldwell, K.A., Rogers, R.D., 2004. Using Caenorhabditis elegans to probe toxicity of 1-alkyl-3-methylimidazolium chloride based ionic liquids. Chem. Commun, 668–669. Torrecilla, J.S., Garcı´a, J., Rojo, E., Rodrı´guez, R., 2009. Estimation of toxicity of ionic liquids in leukemia rat cell line and acetylcholinesterase enzyme by principal component analysis, neural networks and multiple linear regressions. J. Hazard. Mater 164, 182–194. U.S. Environmental Protection Agency 1998. Fate, transport, and transformation test guidelines. OPPTS 835.3110. Readily Biodegradability. Vijayaraghavan, K., Pham, T.P.T., Cho, C.-W., Won, S.W., Choi, S. B., Mao, J., Kim, S., Kim, Y.-R., Chung, B.W., Yun, Y.-S., 2009. An assessment on the interaction of a hydrophilic ionic liquid with different sorbents. Ind. Eng. Chem. Res. 48, 7283–7288. Wang, L.-S., Wang, L., Wang, L., Wang, G., Li, Z.-H., Wang, J.-J., 2009. Effect of 1-butyl-3-methylimidazolium tetrafluoroborate on the wheat (Triticum aestivum L.) seedlings. Environ. Toxicol 24, 296–303. Wang, X., Ohlin, C.A., Lu, Q., Fei, Z., Hu, J., Dyson, P.J., 2007. Cytotoxicity of ionic liquids and precursor compounds towards human cell line HeLa. Green Chem. 9, 1191–1197. Wasserscheid, P., Keim, W., 2000. Ionic liquids–New ‘‘solutions’’ for transition metal catalysis. Angew. Chem. Int. Ed. 39, 3772–3789. Wells, A.S., Coombe, V.T., 2006. On the freshwater ecotoxicity and biodegradation properties of some common ionic liquids. Org. Pro. Res. Dev. 10, 794–798. Welton, T., 1999. Room-temperature ionic liquids, solvents for synthesis and catalysis. Chem. Rev. 99, 2071–2083. Willauer, H.D., Huddleston, J.G., Griffin, S.T., Rogers, R.D., 1999. Partitioning of aromatic molecules in aqueous biphasic systems. Sep. Sci. Technol 34, 1069–1090. Wong, D.S.H., Chen, J.P., Chang, J.M., Chou, C.H., 2002. Phase equilibria of water and ionic liquids [emim][PF6] and [bmim][PF6]. Fluid Phase Equilibr 194–197, 1089–1095. Yu, M., Li, S.-M., Li, X.-Y., Zhang, B.-J., Wang, J.-J., 2009a. Acute effects of 1-octyl-3-methylimidazolium bromide ionic liquid on the antioxidant enzyme system of mouse liver. Ecotox. Environ. Saf. 71, 903–906. Yu, M., Wang, S.-H., Luo, Y.-R., Han, Y.-W., Li, X.-Y., Zhang, B.-J., Wang, J.-J., 2009b. Effects of the 1-alkyl-3-methylimidazolium bromide ionic liquids on the antioxidant defense system of Daphnia magna. Ecotox. Environ. Saf. 72, 1798–1804. Zhang, C., Malhotra, S.V., 2005. Increased paraoxon detection by acetylcholinesterase inactivation with ionic liquid additives. Tatanta 67, 560–563.

water research 44 (2010) 373–384

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

A QSAR model for predicting rejection of emerging contaminants (pharmaceuticals, endocrine disruptors) by nanofiltration membranes Victor Yangali-Quintanillaa,b,*, Anwar Sadmania, Megan McConvillea,b, Maria Kennedya, Gary Amya,b a

UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands Delft University of Technology, Stevinweg 1, 2628CN Delft, The Netherlands

b

article info

abstract

Article history:

A quantitative structure activity relationship (QSAR) model has been produced for pre-

Received 25 February 2009

dicting rejection of emerging contaminants (pharmaceuticals, endocrine disruptors,

Received in revised form

pesticides and other organic compounds) by polyamide nanofiltration (NF) membranes.

4 June 2009

Principal component analysis, partial least square regression and multiple linear regres-

Accepted 26 June 2009

sions were used to find a general QSAR equation that combines interactions between

Available online 3 July 2009

membrane characteristics, filtration operating conditions and compound properties for predicting rejection. Membrane characteristics related to hydrophobicity (contact angle),

Keywords:

salt rejection, and surface charge (zeta potential); compound properties describing

Pharmaceuticals

hydrophobicity (log Kow, log D), polarity (dipole moment), and size (molar volume, molec-

Endocrine disruptors

ular length, molecular depth, equivalent width, molecular weight); and operating condi-

Nanofiltration

tions namely flux, pressure, cross flow velocity, back diffusion mass transfer coefficient,

Modelling

hydrodynamic ratio (Jo/k), and recovery were identified as candidate variables for rejection

QSAR

prediction. An experimental database produced by the authors that accounts for 106 rejection cases of emerging contaminants by NF membranes as result of eight experiments with clean and fouled membranes (NF-90, NF-200) was used to produce the QSAR model. Subsequently, using the QSAR model, rejection predictions were made for external experimental databases. Actual rejections were compared against predicted rejections and acceptable R2 correlation coefficients were found (0.75 and 0.84) for the best models. Additionally, leave-one-out cross-validation of the models achieved a Q2 of 0.72 for internal validation. In conclusion, a unified general QSAR equation was able to predict rejections of emerging contaminants during nanofiltration; moreover the present approach is a basis to continue investigation using multivariate analysis techniques for understanding membrane rejection of organic compounds. ª 2009 Elsevier Ltd. All rights reserved.

* Corresponding author: UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands. Tel.: þ31 15 215 1745; fax: þ31 15 215 2921. E-mail addresses: [email protected], [email protected] (V. Yangali-Quintanilla). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.06.054

374

1.

water research 44 (2010) 373–384

Introduction

Nanofiltration (NF) and reverse osmosis (RO) are technologies that provide medium to high rejections of organic compounds present as emerging contaminants (micropollutants) in water, namely endocrine disrupting compounds (EDCs), pharmaceutically active compounds (PhACs), and pesticides (Kiso et al., 2001; Scha¨fer et al., 2003; Kimura et al., 2003, 2004; Nghiem et al., 2004). The presence of micropollutants has been identified in surface water bodies, sewage treatment plant effluents, and stages of drinking water treatment plants and even at trace-levels in finished drinking water (Kolpin et al., 2002; Heberer, 2002; Castiglioni et al., 2006). The possible effects on aquatic organisms and human health, associated with the consumption of water containing low concentrations of single compounds, have been presented in toxicology studies (Pomati et al., 2006; Escher et al., 2005; Vosges et al., 2008). The studies demonstrate that researchers do not yet understand the exact risks from decades of persistent exposure to random combinations of low levels of pharmaceuticals, EDCs, and other organic compounds; hence, the long-term effects of consumption of water containing low concentrations of micropollutants will remain as an unanswered question for the foreseeable future. Meanwhile, water treatment facilities are implementing monitoring programs, research organizations dealing with water reuse have published reports, and studies have addressed the topic (Drewes et al., 2006; Verliefde et al., 2007). An important aspect to deal with the problem has been the identification of compound physicochemical properties and membrane characteristics to explain transport, adsorption and removal of micropollutants by different mechanisms, explicitly size/ steric exclusion, hydrophobic adsorption and partitioning, and electrostatic repulsion (Kiso et al., 2001; Scha¨fer et al., 2003; Kimura et al., 2003; Nghiem et al., 2004, Ozaki and Li, 2002; Van der Bruggen and Vandecasteele, 2002; Bellona and Drewes, 2005; Xu et al., 2005). A number of articles have proposed a mechanistic understanding of the interaction between membranes and organic compounds; others have tried to apply fitting parameter models to model rejection (Cornelissen et al., 2005; Kim et al., 2007; Verliefde et al., 2008). However, there have been few models to ‘‘predict’’ the rejection of compounds. To overcome that status, our objective was to create a general QSAR model to predict rejections based on an integral approach that considers membrane characteristics, filtration operating conditions and physicochemical compound properties. A quantitative structure activity relationship (QSAR) is a method that relates an activity of a set of compounds quantitatively to chemicals descriptors (structure or property) of those compounds (Sawyer et al., 2003). QSAR has the objective of prediction but maintaining a relationship to mechanistic interpretation. Applications of QSAR for the development of models to find relationships between membranes and organic compounds have been presented in journals related to drug discovery and medicinal chemistry for analysis of permeability of membranes to organic compounds (Ren et al., 1996; Fujikawa et al., 2007). The study of reverse

osmosis membranes has also experienced the application of QSAR principles. Campbell et al. (1999) performed a QSAR analysis of surfactants influencing attachment of a mycobacterium to cellulose and aromatic polyamide reverse osmosis membranes; their objective was to understand the relationship between surfactant molecular properties and activity on the membrane surface to inhibit bacterial attachment to the membrane in order to reduce biofilm formation and to increase permeate production. More recently, Libotean et al. (2008) developed an artificial neural network model based on quantitative structure-property relations, the model claim to predict organic solute passage through reverse osmosis membranes considering simultaneous correlation of organic solutes (molecular descriptors) and membrane properties. Alike, our study uses the concept of QSAR analysis to quantify an activity, compound rejection by a membrane, in terms of organic compound physicochemical properties, membrane characteristics (salt rejection, pure water permeability, molecular weight cut-off, charge, hydrophobicity) and operating conditions (pressure, flux, cross flow velocity, back diffusion mass transfer coefficient, recovery). In this work a QSAR model was constructed with internal experimental data used for training. The model was internally validated using measures of goodness of fit and prediction. Subsequently, after identification of a relationship in form of an equation, estimation of rejections for an external dataset for different compounds and membranes were used to externally validate the model. Similarly, rejections of more emerging organic contaminants can be predicted in advance, before nanofiltration or reverse osmosis applications. Nevertheless, the QSAR model is applicable in the range of boundary experimental conditions that will be defined in the experimental section of this publication (Section 3).

2.

Theory

2.1.

Principal component analysis

Principal component analysis (PCA) is a method that allows simplification of many variables into a group of a few variables that might be measuring the same principles of a system. It may occur that a system considers an abundance of variables to explain a process; in this case principal component analysis reduces the redundancy of information. The general objectives of PCA are data reduction and interpretation. Although p variables (components) are required to reproduce the total system variability, often much of this variability can be accounted for by a small number k of principal components. The k is the number of components (reduced) that represent the initial p variables. In general, PCA is concerned with whether the covariances or correlations between a set of observed variables x1, x2, ., xp can be explained in terms of a smaller number of unobservable components, c1, c2, ., ck, where k < p. Thus, there is as much information in the k components as there is in the original p variables. Comprehensive details about the theory of PCA can be found elsewhere (Jolliffe, 2002; Johnson and Wichern, 2007; Everitt and Dunn, 2001).

water research 44 (2010) 373–384

2.2.

Multiple linear regression

Multiple linear regression is a method of analysis for assessing the strength of the relationship between a set of explanatory variables known as independent variables, and a single response or dependent variable. Applying multiple regression analysis to a set of data results in what are known as regression coefficients, one for each explanatory variable. The multiple regression model for a response variable, y, with observed values, y1, y2, ., yn (where n is the sample size) and q explanatory variables, x1, x2, ., xq with observed values, x1i, x2i, ., xqi for i ¼ 1, ., n, is yi ¼ b0 þ b1 x1i þ b2 x2i þ / þ bq xqi þ 3i

(1)

The regression coefficients, b0, b1, ., bq, are generally estimated by least squares. The term 3i is the residual or error for individual i and represents the deviation of the observed value of the response for this individual from that expected by the model. These error terms are assumed to have a normal distribution with variance s2. The fit of a multiple regression model can be judged with calculation of the multiple correlation coefficient, R, defined as the correlation between the observed values of the response variable and the values predicted by the model. The squared value of R (R2) gives the proportion of the variability of the response variable accounted for by the explanatory variables. Analysis of variance (ANOVA) will provide an F-test of the null hypothesis that each of b0, b1, ., bq, is equal to zero, or in other words that R2 is zero (Landau and Everitt, 2004).

2.3. Principal component regression and partial least squares regression Principal component regression (PCR) is a method in which the components from the principal component method are used for regression. Hence, the principal components of the matrix X are used as regressors of a dependent Y. The orthogonality of the principal components eliminates the multicollinearity problem. But, the problem of choosing an optimum subset of predictors remains. A possible strategy is to keep only a few of the first components. But they are chosen to explain X rather than Y, and therefore, nothing guarantees that the principal components, which ‘‘explain’’ X, are relevant for Y. Problems may arise, however, if there is a lot of variation in X. PCR finds, somewhat uncritically, those latent variables that describe as much as possible of the variation in X. But sometimes the variable itself gives rise to only small variations in X, and if the interferences vary a lot, then the latent variables found by PCR may not be particularly good at describing Y. In the worst case important information may be hidden in directions in the X-space that PCR interprets as disturbance, and therefore leaves out. Partial least squares regression (PLS) is able to cope better with this problem, by forming variables that are relevant for describing Y. By contrast, PLS regression finds components from X that are also relevant for Y. Specifically, PLS regression searches for a set of components (called latent vectors) that performs a simultaneous decomposition of X and Y with the constraint that these components explain as much as possible of the

375

covariance between X and Y. This step generalizes PCA. The goal of PLS regression is to predict Y from X and to describe their common structure (Abdi, 2003; Jørgensen and Goegebeur, 2006).

3.

Experimental

3.1. Chemicals, membranes, materials and experimental conditions The list of organic compounds representing emerging contaminants is presented in Table 1. The compounds were selected considering: their occurrence in surface water and drinking water, their identification as priority emerging contaminants, the availability of analytical methods, their quantitative and qualitative representation of physicochemical properties. The pharmaceutical compounds (caffeine, sulfamethoxazole, acetaminophen, phenacetin, phenazone, carbamazepine, naproxen, ibuprofen, metronidazole), endocrine disrupting compounds (17b-estradiol, estrone, bisphenol A, nonylphenol, atrazine) and sodium alginate were purchased from Sigma–Aldrich (Sigma–Aldrich, Schnelldorf, Germany). Potassium chloride, sodium hydroxide, hydrochloric acid and magnesium sulphate anhydrous were purchased from J.T.Baker (J.T.Baker, Deventer, Netherlands). Two thin film composite NF membranes were selected for this study (NF-200 and NF-90, Dow-Filmtec, Dow Chemical Co., Midland, MI). The experimental setup consisted of two filtration SEPA CF II (GE Osmonics, Minnetonka, MN) cells and cell holders in parallel, in order to increase permeate production and achieve hydrodynamic conditions, two hydraulic pumps (Power Team, Bega Int. BV, Netherlands), a 60 litres stainless steel tank (Tummers, Netherlands), a positive displacement pump (Hydra-Cell pump, Wanner Eng. Inc., Minneapolis, MN), a frequency converter (VLT microdrive, Danfoss, SA), a chiller/heater (Julabo, Germany), control needle valves, pressure gauges, flow meters, a proportional pressure relief valve and stainless steel tubings (Swagelok BV, Netherlands), a digital balance (Sartorius, Germany) and, a computer for flow rate data acquisition. A piece of membrane was compacted with deionised water for 6 h at a pressure of 276 kPa before performing an experiment with the membrane. The experiments were conducted in a recycle mode in which permeate and concentrate were recirculated into the feed tank for the first 72 h (a pre-equilibration period); then, permeate was collected within the next 24 h. The feed solution of all the experiments contained a cocktail of 14 compounds (concentration ranging from 6.5 to 65 mg/L). The main reason for conducting experiments at concentrations of mg/L was to accelerate steady state (after membrane adsorption) conditions in a limited time (3 days). At very low concentrations more time would be needed to achieve steady state conditions and at short time tests low concentrations may lead to over-estimation of rejections. A specific flux decline of 15% of initial flux was targeted to foul the membranes using a feed solution containing w10 mg/L DOC of sodium alginate. The study of Lee et al. (2004) concluded that polysaccharides were important membrane foulants. Sodium alginate was used as

376

water research 44 (2010) 373–384

Table 1 – List of compounds and physicochemical properties. Name

Acetaminophen Phenacetine Caffeine Metronidazole Phenazone Sulfamethoxazole Naproxen Ibuprofen Carbamazepine Atrazine 17b-estradiol Estrone Nonylphenol Bisphenol A a b c d e f

Molecular Acid Log Kowb Log Da (pH 7) weight pKa 20 Ca (g/mol) 151 179 194 171 188 253 230 206 236 216 272 270 220 228

10.2 N/A N/A N/A N/A 5.7 4.3 4.3 N/A N/A 10.3 10.3 10.3 10.3

0.46 1.58 0.07 0.02 0.38 0.89 3.18 3.97 2.45 2.61 4.01 3.13 5.71 3.32

0.23 1.68 0.45 0.27 0.54 0.45 0.34 0.77 2.58 2.52 3.94 3.46 5.88 3.86

Dipole moment (debye)c

Molar volumed (cm3/ mol)

Molec. length (nm)e

Molec. width (nm)e

Molec. depth (nm)e

4.55 4.05 3.71 6.30 4.44 7.34 2.55 4.95 3.66 3.43 1.56 3.45 1.02 2.13

120.90 163.00 133.30 117.80 162.70 173.10 192.20 200.30 186.50 169.80 232.60 232.10 236.20 199.50

1.14 1.35 0.98 0.93 1.17 1.33 1.37 1.39 1.20 1.26 1.39 1.39 1.79 1.25

0.68 0.69 0.87 0.90 0.78 0.71 0.78 0.73 0.92 1.00 0.85 0.85 0.75 0.83

0.42 0.42 0.56 0.48 0.56 0.58 0.75 0.55 0.58 0.55 0.65 0.67 0.59 0.75

Equiv. Classificationf width (nm)e 0.53 0.54 0.70 0.66 0.66 0.64 0.76 0.64 0.73 0.74 0.74 0.76 0.66 0.79

HL-neutral HL-neutral HL-neutral HL-neutral HL-neutral HL-ionic HB-ionic HB-ionic HB-neutral HB-neutral HB-neutral HB-neutral HB-neutral HB-neutral

ADME/Tox Web Software. Experimental database: SRC PhysProp Database. Chem3D Ultra 7.0. ACD/ChemSketch Properties Batch. Molecular Modeling Pro. HL ¼ hydrophilic, HB ¼ hydrophobic.

surrogate of polysaccharides. Alginate is frequently used as a model for organic matter of algae origin (Henderson et al., 2008). Experiments were carried out for both clean and fouled NF-90 and NF-200 membranes. The selection of membranes was based on a qualitative rejection assessment of emerging contaminants with molecular weight of more than 150 Da by membranes with a MWCO between 200 and 300 Da. A total of eight experiments were performed; at hydrodynamic ratio of pure water permeation flux to back diffusion mass transfer coefficient (J0/k) of w1 and recovery 3% (NF-90 clean, NF-200 clean), at J0/k of w1 and recovery 8% (NF-200 clean), at J0/k of w2 and recovery 3 (NF-90 clean), and at J0/k of w2 and recovery 8 (NF-90 clean and fouled, NF-200 clean and fouled). The calculation method of k (back diffusion mass transfer coefficient) was presented in a previous publication (YangaliQuintanilla et al., 2009). All the experiments were carried out at a controlled temperature of 20 C, an ionic strength of 10 mM as KCl and a pH of 7. The transmembrane pressures were in the range of 276–483 kPa. The fluxes were between 4.3 and 30.2 L/m2 per day; cross flow velocities were between 0.5 and 4.5 cm/s. The experiments produced a total internal dataset of 106 rejection cases; the dataset can be accessed as supplementary data. The boundary experimental conditions of the internal dataset are presented in Table 2. The internal dataset was used to develop the model. An external dataset that gathered three different datasets was used for validation of the model. The external dataset is presented as supplementary data. Experimental conditions for the first part of the external dataset can be obtained from Kim et al. (2007) and Yangali-Quintanilla et al. (2008). Experimental conditions for the second and third parts can be found in Verliefde et al. (2008); the data correspond to filtration experiments using synthetic water solutions.

3.2. Analytical equipment, analyses of compounds and membranes The pharmaceuticals and endocrine disruptors (with the exception of atrazine) were analyzed by Technologiezentrum Wasser (TZW, Karlsruhe, Germany). The detection limit was 10 ng/L per compound. The uncertainty of estimates was of 15% according to a validation method of the analysis protocol and due to high concentrations of the samples (mg/L). The analyses of pharmaceuticals were performed according to

Table 2 – Data range of membrane characteristics, operating conditions and rejections. Variable Molecular weight cut-off (MWCO) Pure water permeability (PWP) Salt rejection (SR)a Zeta potential (ZP) Contact angle (CA) Pressure (P) Cross flow velocity (v) Back diffusion mass transfer coefficient (k) Flux (J ) Hydrodynamic ratio (J0/k) Recovery (recov) Rejection (rejection)

Unit

Min. value Max. value

Da

200

300

L/m2 per day/kPa – MV

0.86

2.23

kPa cm/s cm/s

0.96 48.04 39.3 276 0.5 2.70E-04

0.98 10.78 58.0 483 4.5 5.99E-04

L/m2 per day –

4.3 1

30.2 2

% %

3 17.7

8 99.0

a 2000 mg/L MgSO4, 25 C, recovery 15%, pressure 1034 kPa, pH 8.

water research 44 (2010) 373–384

the protocols described by Sacher et al. (2001, 2008). Analyses of 17b-estradiol, estrone, nonylphenol and bisphenol A were done by gas chromatography/mass spectrometry (GC/MS) after automated solid-phase extraction onto a polymeric material and subsequent silylation of the analytes. First, 10 ml of a 50 ng/ml solution of 4-n-nonylphenol in acetone which was used as internal standard for the overall procedure were added to an aliquot of the water sample (1000 ml). Automated solid-phase extraction (Tekmar AutoTrace, Germany) was done on plastic cartridges filled with 200 mg of bondelut material (Fa. Varian, Darmstadt, Germany). After the enrichment step the solid-phase material was dried in a gentle stream of nitrogen. Elution was done with 4 ml of acetone. The acetone was evaporated to 100 ml in a stream of nitrogen and to dryness in a drying oven at 80 C. The dry residue was reconstituted with 100 ml of a silylation reagent mixture (N-methyl-N-trimethylsilyltrifluoro acetamide (MSTFA)/2% trimethyliodo silane). After a reaction time of 20 min at 80 C (drying oven), determination of the derivatives was done by GC/MS using a 6890 GC/MS system from Agilent Technologies (Waldbronn, Germany). Concentrations of atrazine were determined using microplate enzyme-linked immunoabsorbent assay (ELISA) kits (Abraxis LLC, Warminster, PA). Atrazine was determined with a detection limit of 0.04 mg/L, and uncertainty of 15%. To determine the hydrophobicity of membranes, contact angles of clean and fouled membrane surfaces were measured with CAM200 optical contact angle and surface tension meter (KSV Instruments, Finland) at Delft University of Technology; to measure contact angle, the sessile drop method was used. Surface charge, in terms of zeta potential, of clean and fouled membranes was quantified using ELS-8000 zeta potential analyzer (Otsuka Electronics, Japan). The zeta potential analyses were determined using a Milli-Q water solution at pH 7 and ionic strength of 10 mM KCl. The zeta potential was determined using the electrophoresis method using a cell consisting of membrane and quartz cells. The zeta potential was calculated from the electrophoretic mobility using the Smoluchowski formula, a detailed explanation of calculation was provided in a previous publication (Shim et al., 2002). The pH of the solutions was measured using a calibrated Metrohm 691 pH-meter (Metrohm AG, Herisau, Switzerland); the electrical conductivity and temperature were measured with a WTW Cond 330i (WTW GmbH, Weilheim, Germany) portable conductivity meter. Clean and fouled membranes were characterized to determine magnesium sulphate salt rejection at standard conditions specified by manufacturers, a pure water solution containing 2000 mg/L of magnesium sulphate at 25 C and pH 8 was filtrated at pressure of 1034 kPa and recovery of 15%.

3.3.

Characterization and classification of compounds

The acid dissociation constant as log Ka (pKa) was used to determine the speciation of the organic compound in ionic species at pH 7. For hydrophobicity determination, log Kow and log D were used; log Kow is the octanol–water partition coefficient and log D is the ratio of the equilibrium concentrations of all species (unionized and ionized) of a molecule in octanol to the same species in the water phase. Values of pKa and log D

377

were calculated by ADME/Tox web software. Solubility and log Kow values were obtained from SRC Physprop experimental database. The value of the molecular dipole moment was equal to the vector sum of the individual bond dipole moments. Dipole moment was calculated by Chem3D Ultra 7.0, Cambridgesoft. Size descriptors included molar volume (MV), molecular length, molecular width, molecular depth and equivalent molecular width. The molecular length is defined as the distance between the two most distant atoms. The molecular width and molecular depth (width > depth) are measured by projecting the molecule on the plane perpendicular to the length axis and the equivalent molecular width is defined as the geometric mean of width and depth (Santos et al., 2006). Molar volumes were calculated using the program ACD/ChemSketch Properties Batch, ACD/Labs; and Molecular Modeling Pro, ChemSW, was used to compute size descriptors after optimization geometry of a molecule from the interaction of conformational analysis and energy minimization with a semi-empiric method MOPAC-PM3. Based on pKa and log Kow values, the compounds were classified as hydrophilic neutral, hydrophilic ionic, hydrophobic ionic and hydrophobic neutral (see Table 1). Compounds with log Kow 2 were referred to as hydrophobic; therefore those with log Kow < 2 were hydrophilic. The classification was based on an early reference (Connell, 1990). Although the value of 2 may seem low to consider hydrophobicity, the classification was not used in constructing the models and therefore the magnitude of log Kow or log D became more important. Table 1 shows the calculated values of molecular weight, pKa, log Kow, log D, dipole moment, molar volume, molecular length, molecular width, molecular depth and equivalent width.

4.

Results and discussion

4.1.

QSAR methodology

The procedure to find a general QSAR equation to describe rejection was performed in four phases. The first phase was the organization of data from the experimental part. The data comprised of 106 rejection cases. The database showing the rejection cases is presented as supplementary data. A total of 21 initial variables were used. The variables considered as compound descriptors were molecular weight (MW), solubility, log Kow, log D, dipole moment, molar volume, molecular length, molecular width, molecular depth and equivalent width; variables describing membrane characteristics were molecular weight cut-off (MWCO), pure water permeability (PWP), magnesium sulphate salt rejection (SR), charge of the membrane as zeta potential (ZP), and hydrophobicity as contact angle (CA); variables describing operating conditions were operating pressure (P), cross flow velocity (v), back diffusion mass transfer coefficient (k), flux (J ), ratio of pure water permeation flux J0 and back diffusion mass transfer coefficient (J0/k) and recovery. The range of values for membrane characteristics, operating conditions and rejections is presented in Table 2. The second phase was dedicated to the process of variables reduction using a correlation matrix and factor analysis with principal component analysis. The third phase corresponded to the regression analysis. In

378

water research 44 (2010) 373–384

the third phase three methodologies were implemented; the first was principal component analysis (PCA), with sequential application of multiple linear regressions (MLR). The second method was the use of partial least squares (PLS) regression and MLR; and the third method was the use of MLR only. The last phase was the validation process. The model was internally validated using measures of goodness of fit (regression coefficients) and prediction (leave-one-out cross-validation); Section 4.5 gives details about the validation process. External validation of the general QSAR model was implemented by predicting rejections for an external dataset of experiments performed with different compounds and membranes, and with comparable operating conditions. PCA and PLS were performed using the research and statistical package SPSS Statistics 16.0. Leave-one-out cross-validations of the models were performed with MobyDigs (Talete, Milano, Italy).

4.2. Variables reduction with principal component analysis and QSAR model The correlation matrix of the initial 21 variables was scrutinized in order to obtain a not positive definite matrix as a requisite of PCA; the matrix is accompanied as supplementary data. A matrix is called not positive definite when there are both positive and negative eigenvalues. In the case of symmetric matrices, such as a correlation matrix, positive definiteness will only hold if the matrix and every principal submatrix have a positive determinant. A non-positive definite input matrix may signal a perfect linear dependency of one variable on another, known as collinearity. This was the case for MWCO and salt rejection (SR) that were perfectly linearly correlated. Therefore application of PCA considering independently one variable or the other will give the same results of variables reduction and number of components. In other words, MWCO will not be excluded with PCA, the variable will be separated in advance and the results obtained for SR may be replaced by the variable MWCO, or vice versa. Once an appropriate matrix was defined, the variables were analyzed in terms of how significant their correlations with rejection were; those correlations are also shown as an additional row and column of the 21 21 variables matrix. Rejection is only a reference variable to evaluate correlation with the rest of variables. After a sequential implementation of PCA, three components were extracted; they defined the initial database of 21 variables with 11 variables describing three relations namely membrane/operating-conditions (comp. 1: flux, pure water permeability, salt rejection, zeta potential, mass transfer coefficient, cross flow velocity), hydrophobicity/size (comp. 2: length, log Kow, log D) and size (comp. 3: equivalent width, depth). The final three components accounted for 89.3% explanation of total variance. It is important to mention that these results were produced for the experimental dataset. The next step was the implementation of multiple linear regression (MLR) using the new set of variables. The use of MLR after PCA presents the advantage of a more simplified modelling approach. Moreover, the analysis of data before MLR may help to identify variables that are similar in response, which was the case of SR and MWCO. The dependent variable for all regression analyses was rejection. Two

methods of linear regression were used, the first method is called enter (forced) method; which performs a regression with the contribution of all variables entered to model the dependent variable. The second method is stepwise regression; which is a more sophisticated method. Each variable is entered in sequence and its contribution is assessed according to an F-test. In the present study an F-test with a statistical significance >0.10 implied removal of the variable, and F-test with a significance <0.05 implied entry of the variable in the model. If adding the variable contributes to the model then it is retained, but all other variables in the model are then retested to see if they are still contributing to the success of the model, otherwise an elimination process is carried out to remove variables that are no longer judged to improve the model. Therefore, the method should ensure that the model contains a set of appropriate predictor variables. Considering the previous explanation, the variables log Kow, log D, length, depth, eqwidth, PWP, SR, ZP, v, k, J and the 106 rejection cases were used to model rejection. The final regression resulted in variables SR, eqwidth, log D, length and depth. The model presented a correlation coefficient R2 of 0.75 with an F-test of 60.2, with a statistical significance of w0%. Besides, all coefficients of the model variables showed very acceptable significances (<0.001). Therefore the QSAR linear equation model for rejection can be written as rejection ¼

265:150 eqwidth 117:356 depth þ 81:662 length 5:229 log D þ 1358:090 SR 1447:817 (2)

Where, the units of the variables were specified in Tables 1 and 2. Eq. (2) can be mechanistically interpreted; rejection will increase by the effect of size, which is explained by the positive coefficients of length and equivalent width. The mechanism of steric hindrance due to size exclusion has been recognized as a main cause of rejection in many studies (Kiso et al., 2001; Van der Bruggen et al., 1999; Ozaki and Li, 2002; Nghiem et al., 2005). Contrarily, the negative coefficient of log D will decrease rejection, which clearly states that the effect of hydrophobicity lessens rejection due to adsorption and subsequent partitioning mechanisms. Indeed, partitioning is a combined effect that is not only compound property dependent (size, hydrophobicity) but also is related to membrane characteristics. It is important to mention that log D is assuming the role of hydrophobicity for neutral and ionic compounds; compounds with high log D values will adsorb to the membrane and may partition after saturation. Contrarily, ionic compounds will present log D values depending on the pH of the solution, thus may not adsorb when log D is very low or negative, and may adsorb at greater values of log D. Hydrophobicity influences rejection after adsorptive interactions with the membrane; this fact has been documented in some studies (Kiso et al., 2001; Kimura et al., 2004). The role of depth in the equation will compensate size exclusion contributions of length and equivalent width in a final rejection. The equation also shows that salt rejection (SR) is a parameter incorporating steric/size hindrance and electrostatic repulsion effects related to charge of the membrane and operating conditions. This effect may be related to cake-enhanced concentration polarization affecting the salt rejection of clean and fouled membranes (Hoek and

water research 44 (2010) 373–384

Elimelech, 2003). Thus, SR is ultimately serving as a comparison parameter between membranes of the same type (aromatic polyamide) but with narrow differences in pore size, and possibly with differences in charge. The QSAR equation merges information about interaction of membrane characteristics, filtration operating conditions and organic compounds properties to predict rejections during nanofiltration. According to Eq. (2), contact angle and zeta potential as measurements of hydrophilicity and membrane surface charge, respectively, were not part of the equation and therefore did not contribute quantitatively to model rejection. However, size/steric hindrance effects related to salt rejection, and hydrophobicity of the solutes were part of the model equation. In conclusion, rejection increased by size/steric hindrance effects; but hydrophobicity decrease rejection due to adsorption and partitioning mechanisms. The results of PCA and the model cannot be generalized to experimental conditions outside of the boundary experimental conditions defined in Section 3. For instance, excessive changes of pH affect the ionic speciation of charged compounds, obviously pKa values of solutes and pH of feed waters will determine boundary conditions for applicability of the model. Changes of membrane properties such as charge and pore size due to swelling will also influence the model. Other considerations for application of the model are the type of membrane used (aromatic polyamide), fluxes, pressures and cross flow velocities. Cellulose acetate or even different membrane chemistry than NF-90 or NF-200 will influence the PCA and the model. Nonetheless, the approach is valid and can be generalized under certain conditions in upscale NF applications.

4.3. QSAR model after partial least squares regression and MLR The following variables: MW, solubility, MV, MWCO, ZP, CA, P, v, J, Jo/k, recovery and width, were removed after partial least squares (PLS) regression. Therefore, the final PLS model is defined by variables named log Kow, log D, dipole, length, depth, eqwidth, PWP, SR and k. The main advantage of PLS is the ability to handle collinearity among the independent variables. In contrast, principal component analysis can not control collinearity. Another advantage of PLS was that the calculation process was simpler. However, PLS is used more as a predictive technique and not as an interpretive technique such as MLR. Therefore, in this exploratory analysis, PLS regression serves as a variable selection process and as prelude for implementation of MLR. Once again as occurred with PCA, a reduced number of variables simplified the implementation of MLR. After applying stepwise regression to the reduced number of variables, the model result obtained was also Eq. (2).

4.4.

QSAR model after multiple linear regression

To finalize the model development under various statistical application scenarios, the implementation of direct multiple linear regression was performed. The R2 is calculated for all possible subset models. Using this technique, the model with the largest R2 is declared the best linear model. However, this

379

technique has some disadvantages. First, the R2 increases with each variable included in the model. Therefore, this approach encourages including all variables in the best model although some variables may not significantly contribute to the model. This approach also contradicts the principal of parsimony that encourages as few parameters in a model as possible. Thus, the application of MLR without previous data analysis is a possibility when the number of variables is limited. Another disadvantage during the present study was that MLR was not able to distinguish collinearity between variables. This was the case for variables MWCO and SR; MWCO can replace the role of SR in Eq. (2). The new equation presented an R2 of 0.75; the F-test was 52.5 with a significance of w0%. The equation was the following rejection ¼

265:150 eqwidth 117:356 depth þ 81:662 length 5:229 log D 0:272 MWCO 62:565 (3)

It is important to mention that Eq. (3) may have been equally defined in Section 4.2 if the variable selected before implementing PCA was MWCO and not SR as explained before; therefore it is not surprising that Eqs. (2) and (3) only show differences in two coefficients. However, considering practical or operational facts, it may be more difficult to determine changes of MWCO when fouling occurs; on the other hand, salt rejections tests are part of monitoring practices. In addition, the variable of MWCO may be difficult to define when a range of MWCO exists for a membrane rather than an approximate single MWCO value if compared to salt rejection, although that fact is not desirable for a membrane because will greatly influence rejection of solutes with sizes close to or in the range of MWCO. It may be argued that a simple stepwise MLR will produce the same results of a PCA or PLS followed by MLR, however, the application of MLR without PCA or PLS has the disadvantage of removing or adding appropriate variables during the iterative process of stepwise MLR. The stepwise MLR is only dependent on the fulfillment of a statistical condition; it even may happen that different combination of variables defines a good equation. However, the advantage of PCA or PLS is that only important variables are considered, and only those will be part of the final MLR implementation.

4.5.

Internal and external validation of the QSAR model

Actual (measured) rejection values (106 rejection cases) versus modelled (fitted) rejections of the data used to generate the model are shown in Fig. 1, the dataset is provided as supplementary data; a 95% confidence interval shows that very few modelled rejections were out of that interval. Besides of a good fit of a model, it is necessary an assessment of the predictive power of the model, i.e. appropriate robustness (Eriksson et al., 2003). The R2 is the most widely used measure of the ability of a QSAR model to reproduce the internal data in the training (goodness of fit), but does not explain its robustness and prediction power. One technique to evaluate prediction is the leave-one-out cross-validation technique, in which one case at a time is iteratively held-out from the training set and the rest is used for model development and the excluded case is predicted by the developed model (Gramatica, 2007).

380

water research 44 (2010) 373–384

b HB-ion HB-neu HL-ion HL-neu

95% confidence interval

length, eqwidth, depth, log D, SR R² = 0.75

modeled rejection (%)

measured rejection (%)

measured rejection (%)

a

HB-ion HB-neu HL-ion HL-neu

95% confidence interval

length, eqwidth, depth, log D, MWCO R² = 0.75

modeled rejection (%)

Fig. 1 – QSAR model of experimental database: (a) Eq. (2); and (b) Eq. (3).

According to Gramatica, the predictive power of a model may be estimated by the goodness of prediction parameter Q2 leave-one-out (1-PRESS/TSS, where PRESS is the predictive error sum of squares and TSS is the total sum of squares). In general, a Q2 > 0.5 is regarded as good and Q2 > 0.9 as excellent (Eriksson et al., 2003). For the developed QSAR models, the model with SR (Eq. (2)) presented a Q2 leave-one-out of 0.72, and the model with MWCO (Eq. (3)) presented a Q2 leave-oneout of 0.72. After internal cross-validation it was demonstrated that Eqs. (2) and (3) were valid to model rejection, however, an adjustment must be made to the equation before using it to compare measured vs. predicted rejections for external databases. This adjustment was necessary to overcome the mathematical structure of the equation. Using a physical interpretation, it was evident that size parameters referring to variables length and equivalent width may be large enough to cause rejection predictions over 100%, which can be explained after observing positive coefficients for equivalent width and length. This situation may also be detrimental for rejection predictions of ionic compounds of medium to large size (0.6–1.2 nm as equivalent width) that mostly are rejected due to electrostatic repulsion and less steric hindrance. Therefore Eqs. (2) and (3) can be transformed to the following conditional equation rejection ¼

100 if QSAR model 100 QSAR model

(4)

An external dataset (that gathered three different datasets) was selected for external validation of the QSAR model. The external dataset is presented as supplementary data. The first part of the external dataset corresponds to membrane Filmtec NF-90. The second part corresponds to NF membrane Trisep TS-80 and the third part corresponds to NF membrane Desal HL. Desal HL membrane has a main difference with NF-90 and Trisep TS-80 membranes, viz. the MWCO of Desal HL is in the range of 150–300 Da, while NF-90 and Trisep TS-80 were reported to have a MWCO of 200 Da. Therefore an average MWCO of 225 Da was assumed for Desal HL during the application of Eq. (3). It is worthwhile to mention that the

second and third parts of the external dataset were generated using spiral wound membrane elements instead of flat sheet membranes. Fig. 2a show plotted results of measured rejections vs. predicted rejections after calculations with Eq. (4), for QSAR model with Eq. (2) (SR). According to Fig. 2a an R2 of 0.75 was obtained after considering all compounds of the external dataset. However, after observing the rejection cases by NF-90 for bromoform (BF) and trichloroethene (TCE), it appeared that BF and TCE may be considered atypical results because their rejections did not correspond to their size and hydrophobicity when compared to other compounds with comparable molecular descriptors. According to Table 3 BF has approximately the same hydrophobicity and polarity as CF, but BF is bigger than CF; therefore, measured rejection for BF was expected to be higher than rejection of CF (0%) due to size exclusion. The measured rejection of TCE (3%) was not comparable to the rejection of perchloroethene (39%) although they have the same length but a very small difference in equivalent width. In an experiment conducted by Kim et al. (2007) rejections of BF and TCE were of 50 and 33%, respectively, for a low pressure reverse osmosis membrane. It was also observed that the measured rejection (53%) of linuron (LNU) may be considered as atypical observation because a higher rejection was expected. According to Table 3, monolinuron is close to LNU in size and polarity; thus measured rejection of monolinuron was of 77%, higher than 53%. Also according to Table 3, carbamazepine is close to LNU in size and hydrophobicity; thus measured rejection of carbamazepine was of 94%, higher than 53%. A similar explanation was given for the low rejection (70%) of N-acetylL-tyrosine (NAT) by Desal HL compared to 94% rejection by TS-80 as can be seen in Table 3. Other non-expected rejection was that of 2-(1H)-quinoline (QNL), with a rejection of 22% when compared to 2-methoxyethanol and perchloroethene with rejections of 32 and 39%, respectively; even though the length of QNL is greater than the length of 2-methoxyethanol and perchloroethene, a lower rejection of QNL was observed. However, rejections of methacetin (MTC) and NDPA may be influenced by their small equivalent width as can be seen in

381

water research 44 (2010) 373–384

b 100

100

80

80

measured rejection (%)

measured rejection (%)

a

NAT

MTBE

60 LNU NDPA

40 ETH

MTC QNL

20

60

40

20

NDPA

R2 = 0.84

2

BF

0

0

20

MTC

TCE

R = 0.75 0

40

60

80

100

0

20

predicted rejection (%)

NF-90

TS-80

HL

40

60

80

100

predicted rejection (%)

Linear ( )

NF-90

TS-80

HL

Linear ( )

Fig. 2 – Predicted rejections for external dataset using magnesium sulphate SR: (a) all external data; and (b) selected external data.

Table 3. The response of the model for rejection of those particular compounds (with 2*equivalent width < or w length) was of over prediction, but the model presented better response for rejections of metribuzin, atrazin and N-acetyl-Ltyrosine (all with length >2*equivalent width) as observed in Table 3. Ethanol (ETH) and MTBE with rejections of 38 and 60%, respectively, were expected to be lower for Desal HL membrane because it was observed that rejection of ETH was of 9% for TS-80; and MTBE was expected to have a rejection compared to that of 2-methoxyethanol (32%) or perchloroethene (39%) due to proximities in size. After selection

and justified separation of the mentioned rejection cases, Fig. 2b presents the predictions of the external dataset with an R2 of 0.84. In a similar explanatory scenario, Figs. 3a,b show measured rejections vs. predicted rejections after calculations with Eq. (4), for QSAR model with equation 3 (MWCO). The main difference between Figs. 2b and 3b was that the model with MWCO (Fig. 3b) showed a lower R2 (0.80) than the model with SR (R2 ¼ 0.84), meaning that the latter had a better goodness of fit for external prediction response. Moreover, the characterization of magnesium sulphate salt rejection for a membrane may be preferred instead of MWCO, particularly

Table 3 – Partial list of rejections and compound properties of external dataset. Compound Chloroform Ethanol Ethanol Carbontetrachloride Bromoform MTBE Trichloroethene Perchloroethene 2-methoxyethanol 2-(1H)-quinoline NDPA NDPA Metribuzin Carbamazepine Linuron Monolinuron Atrazin Methacetin Methacetin N-acetyl-L-tyrosine N-acetyl-L-tyrosine

Abb.

ETH BF MTBE TCE

QNL NDPA NDPA

LNU

MTC MTC NAT NAT

Length

Eqwidth

Depth

Log D

Dipole

Measured rejection

Predicted rejection

0.53 0.64 0.64 0.64 0.69 0.77 0.78 0.78 0.87 1.00 1.16 1.16 1.17 1.20 1.21 1.22 1.26 1.28 1.28 1.33 1.33

0.42 0.52 0.52 0.6 0.56 0.63 0.49 0.59 0.52 0.52 0.60 0.60 0.74 0.73 0.69 0.69 0.74 0.52 0.52 0.71 0.71

0.35 0.51 0.51 0.57 0.48 0.59 0.36 0.45 0.51 0.36 0.53 0.53 0.64 0.58 0.53 0.65 0.55 0.42 0.42 0.60 0.60

1.97 0.31 0.31 2.83 2.40 0.94 2.29 3.40 0.77 1.26 1.36 1.36 0.47 2.45 3.20 2.30 2.61 1.03 1.03 2.18 2.18

1.12 1.55 1.55 0.30 1.00 1.37 0.95 0.11 0.25 3.38 3.40 3.40 0.52 3.66 2.11 2.02 3.43 2.20 2.20 3.45 3.45

0 9 38 35 0 60 3 39 32 22 45 19 97 88 53 79 91 38 5 94 70

0 14 0 26 33 25 36 46 34 53 69 55 99 94 84 77 100 72 58 100 100

Membrane NF-90 TS-80 Desal HL NF-90 NF-90 Desal HL NF-90 NF-90 TS-80 TS-80 TS-80 Desal HL TS-80 TS-80 TS-80 TS-80 TS-80 TS-80 Desal HL TS-80 Desal HL

382

water research 44 (2010) 373–384

a

b 100

80 NAT

MTBE

60 LNU NDPA ETH

40

MTC QNL NDPA

20

measured rejection (%)

measured rejection (%)

100

80

60

40

20

R2 = 0.80

2

MTC BF

0 0

R = 0.74

TCE

20

0 40

60

80

100

0

20

predicted rejection (%)

NF-90

TS-80

HL

40

60

80

100

predicted rejection (%)

Linear ( )

NF-90

TS-80

HL

Linear ( )

Fig. 3 – Predicted rejections for external dataset using MWCO: (a) all external data; and (b) selected external data.

for nanofiltration and low pressure reverse osmosis membranes; besides, the effect of fouling in membranes can also be quantified by salt rejection experiments. We can state that the QSAR model with SR demonstrated to be acceptable for the external dataset of NF-90, Trisep TS-80 and Desal HL with an R2 of 0.75 and 0.84 for the total external dataset and justified selected external dataset, respectively. Although the model can be valid with limitations related to boundary experimental conditions mentioned in Section 3, its applicability and approach can be of value for the construction of a model with combined datasets organized in training and testing groups.

5.

Acknowledgements The authors acknowledge Delft Cluster and EU Techneau Project for funding this project. The authors also acknowledge Tae-Uk Kim and Arne Verliefde for providing data and details of their research. The authors thank Filmtec (Dow Chemical Co.) for donating the membranes, Dr. Jaeweon Cho of GIST (Korea), Dr. Frank Sacher of TZW (Germany) and Steven Mookhoek of TU Delft (Netherlands) for contributing with analytical results and facilities.

Conclusions – A general QSAR model equation was developed to merge information about interaction of membrane characteristics, filtration operating conditions and solute properties to predict rejections of emerging contaminants during nanofiltration. – The QSAR model identified that the most important variables that influence rejection of organic solutes were log D, salt rejection, equivalent width, depth and length. – Rejection increased by size/steric hindrance effects, solute hydrophobicity decreased rejection due to adsorption and partitioning mechanisms. – Salt rejection incorporated steric hindrance and electrostatic repulsion effects that were related to the membrane structure and operating conditions. – The use of MWCO was acceptable for modelling purposes; however NF membranes with a broad range of MWCO (pore size and distribution) may difficult estimation of rejection of contaminants, thus magnesium sulphate salt rejection may be more appropriate.

Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2009.06.054

references

Abdi, H., 2003. Partial least squares (PLS) regression. In: LewisBeck, M., Bryman, A., Futing, T. (Eds.), Encyclopaedia of Social Sciences Research Methods. Sage, Thousand Oaks, CA. Bellona, C., Drewes, J.E., 2005. The role of membrane surface charge and solute physico-chemical properties in the rejection of organic acids by NF. Journal of Membrane Science 249 (1–2), 227–234. Campbell, P., Srinivasan, R., Knoell, T., Phipps, D., Ishida, K., Safarik, J., Cormack, T., Ridgway, H., 1999. Quantitative structure-activity relationship (QSAR) analysis of surfactants influencing attachment of a Mycobacterium sp. to cellulose

water research 44 (2010) 373–384

acetate and aromatic polyamide reverse osmosis membranes. Biotechnology and Bioengineering 64 (5), 527–544. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environmental Science & Technology 40 (1), 357–363. Connell, D.W., 1990. Bioaccumulation of Xenobiotic Compounds. CRC Press, Boca Raton, FL, pp. 75-110. Cornelissen, E.R., Verdouw, J., Gijsbertsen-Abrahamse, A.J., Hofman, J.A.M.H., 2005. A nanofiltration retention model for trace contaminants in drinking water sources. Desalination 178 (1–3), 179–192. Drewes, J.E., Amy, G., Kim, T.-U., Xu, P., Bellona, C., Oedekoven, M., Macalady, D., 2006. Rejection of WastewaterDerived Micropollutants in High-Pressure Membrane Applications Leading to Indirect Potable Reuse Effect of Membranes and Micropollutant Properties. WateReuse, Alexandria, VA. Eriksson, L., Jaworska, J., Worth, A.P., Cronin, M.T.D., McDowell, R.M., Gramatica, P., 2003. Methods for reliability and uncertainty assessment and for applicability evaluations of classification and regression-based QSARs. Environmental Health Perspectives 111 (10), 1361–1375. Escher, B.I., Bramaz, N., Eggen, R.I.L., Richter, M., 2005. In vitro assessment of modes of toxic action of pharmaceuticals in aquatic life. Environmental Science & Technology 39 (9), 3090–3100. Everitt, B.S., Dunn, G., 2001. Applied Multivariate Data Analysis. Arnold, London. Fujikawa, M., Nakao, K., Shimizub, R., Akamatsu, M., 2007. QSAR study on permeability of hydrophobic compounds with artificial membranes. Bioorganic & Medicinal Chemistry 15 (11), 3753–3767. Gramatica, P., 2007. Principles of QSAR models validation: internal and external. QSAR & Combinatorial Science 26 (5), 694–701. Heberer, T., 2002. Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicology Letters 131 (1–2), 5–17. Henderson, R.K., Baker, A., Parsons, S.A., Jefferson, B., 2008. Characterisation of algogenic organic matter extracted from cyanobacteria, green algae and diatoms. Water Research 42 (13), 3435–3445. Hoek, E.M.V., Elimelech, M., 2003. Cake-enhanced concentration polarization: a new fouling mechanism for salt-rejecting membranes. Environmental Science & Technology 37 (24), 5581–5588. Johnson, R.A., Wichern, D.W., 2007. Applied Multivariate Statistical Analysis. Pearson Prentice Hall, Upper Saddle River, NJ. Jolliffe, I.T., 2002. Principal Component Analysis. Springer-Verlag, New York. Jørgensen, B., Goegebeur, Y., 2006. Multivariate Data Analysis and Chemometrics. Department of Statistics, University of Southern Denmark. Kim, T.-U., Drewes, J.E., Summer, R.S., Amy, G., 2007. Solute transport model for trace organic neutral and charged compounds through nanofiltration and reverse osmosis membranes. Water Research 41 (17), 3977–3988. Kimura, K., Amy, G., Drewes, J., Heberer, T., Kim, T., Watanabe, Y., 2003. Rejection of organic micropollutants (disinfection byproducts, endocrine disrupting compounds, and pharmaceutically active compounds) by NF/RO membranes. Journal of Membrane Science 227 (1–2), 113–121. Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of neutral endocrine disrupting compounds (EDCs) and pharmaceutical active compounds (PhACs) by RO membranes. Journal of Membrane Science 245 (1–2), 71–78.

383

Kiso, Y., Sugiura, T., Kitao, T., Nishimura, K., 2001. Effects of hydrophobicity and molecular size on rejection of aromatic pesticides with nanofiltration membranes. Journal of Membrane Science 192 (1–2), 1–10. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in U.S. Streams, 1999–2000: a National Reconnaissance. Environmental Science & Technology 36 (6), 1202–1211. Landau, S., Everitt, B., 2004. A Handbook of Statistical Analysis using SPSS. Chapman & Hall/CRC, Boca Raton, FL. Lee, N., Amy, G., Croue, J.-P., Buisson, H., 2004. Identification and understanding of fouling in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM). Water Research 38 (20), 4511–4523. Libotean, D., Giralt, J., Rallo, R., Cohen, Y., Giralt, F., Ridgway, H., Rodriguez, G., Phipps, D., 2008. Organic compounds passage through RO membranes. Journal of Membrane Science 313 (1-2), 23–43. Nghiem, L.D., Scha¨fer, A.I., Elimelech, M., 2004. Removal of natural hormones by nanofiltration membranes: measurement, modelling, and mechanisms. Environmental Science & Technology 38 (6), 1888–1896. Nghiem, L.D., Scha¨fer, A.I., Elimelech, M., 2005. Pharmaceutical retention mechanisms by nanofiltration membranes. Environmental Science & Technology 39 (19), 7698–7705. Ozaki, H., Li, H., 2002. Rejection of organic compounds by ultralow pressure reverse osmosis membrane. Water Research 36 (1), 123–130. Pomati, F., Castiglioni, S., Zuccato, E., Fanelli, R., Vigetti, D., Rossetti, C., Calamari, C., 2006. Effects of a complex mixture of therapeutic drugs at environmental levels on human embryonic cells. Environmental Science & Technology 40 (7), 2442–2447. Ren, S., Das, A., Lien, E.J., 1996. QSAR analysis of membrane permeability to organic compounds. Journal of Drug Targeting 4 (2), 103–107. Sacher, F., Lange, F.T., Brauch, H.-J., Blankenhorn, I., 2001. Pharmaceuticals in groundwaters: analytical methods and results of a monitoring program in Baden-Wu¨rttemberg, Germany. Journal of Chromatography A 938 (1–2), 199–210. Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008. Pharmaceutical residues in the river Rhinedresults of a onedecade monitoring programme. Journal of Environmental Monitoring 10 (5), 664–670. Santos, J.L.C., de Beukelaar, P., Vankelecom, I.F.J., Velizarov, S., Crespo, J.G., 2006. Effect of solute geometry and orientation on the rejection of uncharged compounds by nanofiltration. Separation and Purification Technology 50 (1), 122–131. Sawyer, C.N., McCarty, P.L., Parkin, G.F., 2003. Chemistry for Environmental Engineering and Science. McGraw-Hill, New York, pp. 479-520. Scha¨fer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the natural hormone estrone from aqueous solutions using nanofiltration and reverse osmosis. Environmental Science & Technology 37 (1), 182–188. Shim, Y., Lee, H.-J., Lee, S., Moon, S.-H., Cho, J., 2002. Effects of natural organic matter and ionic species on membrane surface charge. Environmental Science & Technology 36 (17), 3864–3871. Van der Bruggen, B., Vandecasteele, C., 2002. Modeling of the retention of uncharged molecules with nanofiltration. Water Research 36 (5), 1360–1368. Van der Bruggen, B., Schaep, J., Wilms, D., Vandecasteele, C., 1999. Influence of molecular size, polarity and charge on the retention of organic molecules by nanofiltration. Journal of Membrane Science 156 (1), 29–41. Verliefde, A., Cornelissen, E., Amy, G., Van der Bruggen, B., van Dijk, H., 2007. Priority organic micropollutants in water

384

water research 44 (2010) 373–384

sources in Flanders and the Netherlands and assessment of removal possibilities with nanofiltration. Environmental Pollution 146 (1), 281–289. Verliefde, A.R.D., Cornelissen, E.R., Heijman, S.G.J., Verberk, J.Q.J.C., Amy, G.L., Van der Bruggen, B., van Dijk, J.C., 2008. The role of electrostatic interactions on the rejection of organic solutes in aqueous solutions with nanofiltration. Journal of Membrane Science 322 (1), 52–66. Vosges, M., Braguer, J.-C., Combarnous, Y., 2008. Long-term exposure of male rats to low-dose ethinylestradiol (EE2) in drinking water: effects on ponderal growth and on litter size of their progeny. Reproductive Toxicology 25 (2), 161–168.

Xu, P., Drewes, J.E., Bellona, C., Amy, G., Kim, T.-U., Adam, M., Heberer, T., 2005. Rejection of emerging organic micropollutants in nanofiltration-reverse osmosis membrane applications. Water Environment Research 77 (1), 40–48. Yangali-Quintanilla, V., Kim, T.-U., Kennedy, M., Amy, G., 2008. Modeling of RO/NF membrane rejections of PhACs and organic compounds: a statistical analysis. Drinking Water Engineering and Science 1 (1), 7–15. Yangali-Quintanilla, V., Sadmani, A., McConville, M., Kennedy, M., Amy, G., 2009. Rejection of pharmaceutically active compounds and endocrine disrupting compounds by clean and fouled nanofiltration membranes. Water Research 43 (9), 2349–2362.

water research 44 (2010) 385–416

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Harmful algae and their potential impacts on desalination operations off southern California David A. Carona,*, Marie-E`ve Garneaua, Erica Seuberta, Meredith D.A. Howarda,b, Lindsay Darjanya, Astrid Schnetzera, Ivona Cetinic´a, Gerry Filteauc, Phil Laurid, Burton Jonesa, Shane Trusselle a

Department of Biological Sciences, University of Southern California, 3616 Trousdale Parkway, Los Angeles, CA 90089-0371, USA Southern California Coastal Water Research Project, 3535 Harbor Blvd., Suite 110, Costa Mesa, CA 92626, USA c Separation Processes, Inc., 3156 Lionshead Avenue, Suite 2, Carlsbad, CA 92010, USA d West Basin Municipal Water District, 17140 Avalon Blvd., Suite 210, Carson, CA 90746, USA e Trussell Technologies, Inc., 6540 Lusk Boulevard, Suite C175, San Diego, CA 92121, USA b

article info

abstract

Article history:

Seawater desalination by reverse osmosis (RO) is a reliable method for augmenting

Received 2 April 2009

drinking water supplies. In recent years, the number and size of these water projects have

Received in revised form

increased dramatically. As freshwater resources become limited due to global climate

12 June 2009

change, rising demand, and exhausted local water supplies, seawater desalination will play

Accepted 23 June 2009

an important role in the world’s future water supply, reaching far beyond its deep roots in

Available online 30 June 2009

the Middle East. Emerging contaminants have been widely discussed with respect to wastewater and freshwater sources, but also must be considered for seawater desalination

Keywords:

facilities to ensure the long-term safety and suitability of this emerging water supply.

Harmful algal blooms

Harmful algal blooms, frequently referred to as ‘red tides’ due to their vibrant colors, are

Desalination

a concern for desalination plants due to the high biomass of microalgae present in ocean

Red tides

waters during these events, and a variety of substances that some of these algae produce.

Phytoplankton

These compounds range from noxious substances to powerful neurotoxins that constitute

Phytotoxins

significant public health risks if they are not effectively and completely removed by the RO membranes. Algal blooms can cause significant operational issues that result in increased chemical consumption, increased membrane fouling rates, and in extreme cases, a plant to be taken off-line. Early algal bloom detection by desalination facilities is essential so that operational adjustments can be made to ensure that production capacity remains unaffected. This review identifies the toxic substances, their known producers, and our present state of knowledge regarding the causes of toxic episodes, with a special focus on the Southern California Bight. ª 2009 Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: þ1 213 740 0203; fax: þ1 213 740 8123. E-mail address: [email protected] (D.A. Caron). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.06.051

386

1.

water research 44 (2010) 385–416

Introduction

1.1. General overview of harmful algal blooms: a growing global concern Microscopic algae constitute an essential component of all aquatic food webs. Photosynthetic production of organic material by this diverse group of species comprises the primary source of nutrition for all heterotrophic forms of life in much of the world’s ocean and freshwater ecosystems. Microalgae can reach high abundances in the plankton during periods of optimal growth and reduced grazing pressure by herbivores. Such localized mass proliferations are known as algal (or phytoplankton) blooms. In addition, a small proportion of microalgal species are capable of producing a number of noxious or toxic compounds that cause a variety of adverse effects on ecosystem structure and function. These substances pose the potential for ecosystem damage, food web disruption and marine animal mortality, and present a significant human health risk through the consumption of contaminated seafood and, in at least one case, direct exposure to water or aerosols containing these toxic compounds. Additionally, the algal biomass and the associated organic load cause significant desalination operational issues, impacting the pretreatment system and possibly forcing the treatment plant to be taken off-line (Petry et al., 2007). Countless human deaths resulting from the consumption of seafood contaminated with algal toxins have been avoided through rigorous monitoring programs, but sea life has not been so fortunate. Approximately one half of all unusual marine mammal mortality incidents are now attributable to the ingestion of food or prey contaminated by harmful algal blooms (Ramsdell et al., 2005). Losses in revenue due to the direct contamination of seafood products and indirect effects on tourism and other uses of coastal areas have been estimated in the tens of millions of dollars annually in the U.S. states along the Pacific coast (Trainer et al., 2002). There is now convincing evidence that harmful algal bloom (HAB) events are increasing at local, regional and global scales worldwide (Smayda, 1990; Hallegraeff, 1993, 2003; Anderson et al., 2002; Glibert et al., 2005a) and along the North American west coast in particular (Horner et al., 1997; Trainer et al., 2003). This increased occurrence may be due in part to better detection of HAB episodes in recent years or the global dispersal of toxic algal species via the transport of resting spores in ships’ ballast waters (Hallegraeff and Bolch, 1992; Burkholder et al., 2007), but another very likely cause is the increasing impact of anthropogenic activities on coastal ecosystems (Smayda, 1990; Anderson et al., 2002; Glibert et al., 2005b, 2006; Howard et al., 2007; Cochlan et al., 2008; Kudela et al., 2008a). Recent reports reveal extensive and, in some cases, newly emerging occurrences of HABs along the coasts of the U.S. (Fig. 1). These incidents engender a variety of noxious impacts on ecosystems and public health, including direct effects on organisms due to the production of acutely toxic substances, and indirect effects such as reduced availability of dissolved oxygen in the water column resulting from the decomposition of the extensive amounts of organic substances usually produced during such blooms. The

Fig. 1 – Distribution of some well-known regional HAB issues along U.S. shores, including (a) Alaska and (b) Hawaii. Causes and impacts of these poisoning events are defined in Tables 1–3. Summarized from information presented on the Harmful Algae webpage (http://www. whoi.edu/redtide/).

dramatic increases in biomass and organic load that accompany these events pose a significant threat to seawater desalination facilities (Gaid and Treal, 2007).

1.2.

Regional HAB issues along U.S. coastlines

Harmful algae are present throughout U.S. coastal waters, but not all species are of equal concern in all regions (Fig. 1). For example, toxic species of the dinoflagellate genus Alexandrium are common over vast stretches of the U.S. coastline, but coastal regions of the northeastern and northwestern U.S. appear to experience particularly high rates of occurrence of toxic ‘red tides’ caused by these species. The neurotoxins produced by Alexandrium, called saxitoxins, cause paralytic shellfish poisoning (PSP) in humans when ingested through contaminated seafood (particularly filter-feeding shellfish). Similarly, several toxic species of the diatom genus Pseudonitzschia occur along the entire U.S. coastline but significant concentrations of the neurotoxin, domoic acid, produced by these species have historically constituted a health threat primarily in the northeastern and northwestern U.S. (Bates et al., 1989) where it has been documented as the cause of amnesic shellfish poisoning (ASP) in humans. However, high concentrations of domoic acid in the plankton and in diverse planktivorous organisms have been recently documented along the entire Pacific coast of the U.S. (Scholin et al., 2000; Trainer et al., 2002; Schnetzer et al., 2007), as well as in the Gulf of Mexico (Pan et al., 1998). Domoic acid has been attributed to numerous marine animal mortalities along the U.S. west coast. In the Gulf of Mexico, primarily along the west coast of Florida, extensive and recurrent blooms of the dinoflagellate Karenia brevis produce a suite of toxins, known as brevetoxins, that can be aerosolized by breaking waves and

water research 44 (2010) 385–416

induce neurotoxic shellfish poisoning (NSP) in people inhaling the aerosols (see review of Kirkpatrick et al., 2004). The Tampa Bay seawater desalination facility is the only operating seawater desalination treatment plant of significant size in the United States. It is located along the west coast of Florida and is likely to encounter algal blooms that contain brevetoxin. Less toxic blooms also take place with regional specificity. The pelagophyte Aureococcus anophagefferens causes ‘brown tides’ in coastal waters of Rhode Island, near Long Island (NY) and southward along the mid-Atlantic coast of the U.S. since 1985. No specific toxins have been identified from A. anophagefferens, and no human fatalities have been directly attributed to these blooms. Nevertheless, this species appears to be unpalatable or inhibitory to many filter-feeding mollusks and has caused substantial mortality among these populations, including commercially valuable species (Bricelj and Lonsdale, 1997). Other microalgal species can disrupt food webs or cause reductions in water quality without producing acutely toxic conditions. Among these are the ‘colorful’ red tides of the dinoflagellate Lingulodinium polyedrum, a yessotoxin producer, that have occurred periodically throughout several decades along the south and central Californian coasts (Horner et al., 1997; Gregorio and Pieper, 2000). These blooms have so far been found to be relatively innocuous in these waters but massive accumulations of these cells could have significant impact on desalination plants because of increased turbidity, high suspended solids and organic loading of influent water. Furthermore, accumulations of cells in protected harbors can cause fish mortality by depleting oxygen dissolved in the water, further challenging influent screening and pretreatment systems at desalination plants. Other taxa, such as species of the prymnesiophyte genus Phaeocystis, produce substances that can lead to enormous buildups of sea foam along coasts (Armonies, 1989).

1.3.

Desalination, plankton and water quality issues

Large research programs have developed within different geographic areas throughout the U.S. to address regional HAB issues. These programs are designed to study the species, toxins and environmental causes of HAB outbreaks. These efforts, as well as local, county, state and federal monitoring programs provide basic information for marine resource use and have focused almost exclusively on threats to human health via the consumption of contaminated seafood. Unfortunately, few if any of these programs provide sufficient information on appropriate temporal and spatial resolution for thoroughly assessing the potential impact of HAB events on reverse osmosis desalination operations. Moreover, toxin analyses have primarily examined the presence of these substances in particulate material (plankton or animal tissue, particularly shellfish and finfish), and therefore may be poor predictors for the amount of toxins that might occur in seawater in the dissolved state during algal blooms, which would be most likely to be loaded onto reverse osmosis membranes during desalination. There are two potential impacts that HABs may have on seawater desalination facilities: (1) algal toxins in ocean water

387

pose a significant treatment challenge for the reverse osmosis system to ensure that these molecules are effectively removed and (2) increased turbidity, total suspended solids and total organic content resulting from algal biomass and growth challenge the entire desalination facility’s treatment train. The significance of these issues will depend on the specific algae forming a bloom and the toxin(s) or other substances that they produce, the magnitude and duration of the bloom, and the specific desalination process conducted. For example, multi-effect distillation and multi-flash distillation might be susceptible to (2) but would be much less affected by toxins in the water (1). Desalination using reverse osmosis presumably would be vulnerable to both issues. Therefore, for the latter desalination approach, a thorough understanding of HAB episodes in terms of incidence and seasonality, vertical and horizontal spatial distribution, as well as biological aspects such as algal composition within a geographical region could help optimize the design and operational efficiency of desalination plants employing reverse osmosis. This paper provides an overview of HABs occurring along the continental U.S. coastline with special emphasis on the southwestern U.S., and provides some insight on the potential impacts that these events may have on the seawater desalination process. In recent years, this geographical area has become a focal point of discussions regarding desalination (Cooley et al., 2006) because of its sizable population and the particularly tenuous nature of the water supply to this region. Although numerous issues involving the desalination process are now being examined (Separation Processes Inc., 2005; Gaid and Treal, 2007; Pankratz, 2008, 2009), very limited information exists on the risks that algal blooms pose to seawater desalination facilities. A review of the major species producing harmful blooms, the substances they produce, and information on the spatial and temporal distributions of blooms are presented along with some conclusions on their potential impacts. This paper also provides some general guidelines on how early detection may help prevent or minimize the impact of HABs on a desalination facility’s production capacity or its water quality.

2. Toxin producers and toxin concentrations of the west coast A variety of toxins including several powerful neurotoxins are produced by microalgae, and a number of these toxins and potentially toxic algal species have been detected on the U.S. west coast (Table 1). The ability to rapidly detect and quantify toxic algae in natural water samples is problematic at this time. Many of these species are difficult to identify using light microscopy. For this reason, new genetic and immunological methods for species identification and enumeration have been appearing rapidly in the literature (Miller and Scholin, 1998; Bowers et al., 2000, 2006; Coyne et al., 2001; Caron et al., 2003; Galluzzi et al., 2004; Anderson et al., 2005; Mikulski et al., 2005, 2008; Ahn et al., 2006; Handy et al., 2006; Moorthi et al., 2006; Iwataki et al., 2007, 2008; Demir et al., 2008; Matsuoka et al., 2008). Moreover, many toxin-producing algal species exhibit variable toxin production in response to environmental conditions, and among different strains of the same species

388

water research 44 (2010) 385–416

Table 1 – Planktonic species occurring along the west coast of the U.S. that are potential concerns for reverse osmosis operations. Microalgae

Toxin(s)

Poisoning Event

References

Amnesic Shellfish Poisoning (ASP)

Subba Rao et al. (1988), Bates et al. (1989), Martin et al. (1990), Buck et al. (1992), Garrison et al. (1992), Rhodes et al. (1996), Horner et al. (1997), Lundholm et al. (1997), Rhodes et al. (1998), Trainer et al. (2000, 2001), Baugh et al. (2006)

Diatoms Pseudo-nitzschia spp. P. australisb P. cuspidatab P. delicatissimab P. fraudulentab P. multiseriesb P. pungensb P. pseudodelicatissimab P. seriataa

Domoic acid (DA)

Dinoflagellates Alexandrium spp. A. acatenellaa A. catenellab A. fundyensea A. hiranoia A. ostenfeldiia A. tamarensea

Saxitoxins (STXs)

Dinoflagellates Lingulodinium polyedrumb Gonyaulax spiniferaa Protoceratium reticulatuma,c

Yessotoxins (YTXs)

Human and ecosystem effects None reported

Holmes et al. (1967), Satake et al. (1997, 1999), Draisci et al. (1999a), Paz et al. (2004, 2007), Armstrong and Kudela (2006), Rhodes et al. (2006), Howard et al. (2007)

Dinoflagellates Dinophysis spp. D. acuminataa D. acutaa D. caudate D. fortiia D. norvegicaa D. rotundataa D. triposa

Okadaic acid (OA) Dinophysistoxins (DTXs) Pectenotoxins (PTXs)

Diarrhetic Shellfish Poisoning (DSP)

Holmes et al. (1967), Yasumoto et al. (1980), Murata et al. (1982), Yasumoto et al., (1985), Cembella (1989), Lee et al. (1989), Horner et al. (1997), Cembella (2003), Miles et al. (2004), Shipe et al. (2008), Sutherland (2008)

Human effects Gastro-intestinal symptoms Neurologic symptoms Death Ecosystem effects Marine mammal mortalities Bird mortalities

Paralytic Shellfish Poisoning (PSP) Human effects Gastro-intestinal symptoms Paralysis Death Ecosystem effects Marine mammal mortalities

Human effects Gastro-intestinal symptoms Ecosystem effects None reported

Sommer and Meyer (1937), Gaines and Taylor (1985), Steidinger (1993), Scholin et al. (1994), Taylor and Horner (1994), Jester (2008)

Prorocentrum spp. P. micans P. minimuma,d Raphidophytes Chattonella marinaa Fibrocapsa japonicaa Heterosigma akashiwoa

a b c d

Brevetoxins (PbTxs)

Neurotoxic Shellfish Poisonining (NSP) Human effects Gastroenteritis Neurologic symptoms Respiratory irritation and/or failure Ecosystem effects Marine mammal mortalities Fish mortality events

Loeblich and Fine (1977), Hershberger et al. (1997), Gregorio and Connell (2000), Hard et al. (2000), Tyrell et al. (2002), O’Halloran et al. (2006)

Reported to produce toxin. Reported to produce toxin on the west coast of the United States. Conflicting reports on toxicity of P. reticulatum cultures isolated from California, Washington and Florida. Reported to be present on the west coast of Mexico.

even when isolated from the same geographic region (Smith et al., 2001; Trainer et al., 2001; Kudela et al., 2004). Laboratory experiments have revealed a wide range of physico-chemical factors that increase or decrease toxin

production by harmful species of algae, and which appear to be species-specific (see review of Grane´li and Flynn, 2006). Reports of factors inducing toxin production have sometimes been conflicting, presumably indicating that multiple factors, or

water research 44 (2010) 385–416

perhaps generally stressful conditions, may stimulate toxin production. Factors affecting toxin production include: (1) temperature (Ono et al., 2000); (2) light intensity (Ono et al., 2000); (3) salinity (Haque and Onoue, 2002a,b); (4) trace metal availability, especially iron (Ladizinsky and Smith, 2000; Rue and Bruland, 2001; Maldonado et al., 2002; Wells et al., 2005; Sunda, 2006) but also copper (Maldonado et al., 2002) and selenium (Mitrovic et al., 2004, 2005); (5) macronutrient availability including silicate (Pan et al., 1996b; Fehling et al., 2004; Kudela et al., 2004), phosphate (Pan et al., 1996a, 1998; Fehling et al., 2004), nitrogen (Bates et al., 1991; Pan et al., 1998; Kudela et al., 2004) and combinations of nutrient limitation (Anderson et al., 1990; Flynn et al., 1994; John and Flynn, 2000); (6) cellular elemental ratios of nutrients and physiological stress (Grane´li and Flynn, 2006; Schnetzer et al., 2007); (7) growth phase (Anderson et al., 1990; Bates et al., 1991; Flynn et al., 1994; Johansson et al., 1996; Maldonado et al., 2002; Mitrovic et al., 2004). The precise combination(s) of environmental factors that select for population growth of particular algal species within diverse natural assemblages, and the specific conditions that induce toxin production, are poorly understood for most harmful algae. This present state of knowledge makes it difficult to predict the timing, duration or spatial extent of the vast majority of HAB events and the toxic events resulting from them. Our ability to thoroughly characterize HABs is also complicated by the complex array of toxins produced by algae. Marine algal species produce a suite of toxic components (Yasumoto and Murata, 1993), and unidentified toxins undoubtedly remain to be described. Additionally, most toxins are actually composed of families of closely related compounds. Slightly different forms of a toxin can exhibit very different levels of toxicity, or may be characterized differently by some detection methods and analytical approaches (Garthwaite et al., 2001; Lefebvre et al., 2008). Such complexity and variability can sometimes yield vague or contradictory conclusions regarding the exact source of toxicity in a natural sample (Bates et al., 1978; Paz et al., 2004, 2007). Finally, characterization of HAB events is complicated by inherent difficulties associated with linking specific toxins measured in natural water samples to a specific algal species in a complex, natural phytoplankton assemblage and, as noted above, the presence of toxic species in a water sample does not necessarily indicate the presence of toxins. Despite these shortcomings, there is considerable knowledge of many of the major algal toxins and their producers in U.S. coastal waters that constitute the most important potential concerns for desalination activities because they are the most likely to be encountered in ocean water intakes.

2.1.

Domoic acid

2.1.1.

Toxin description and activity

Domoic acid (Fig. 2; Table 3) is an amino acid derivative belonging to the kainoid class of compounds containing three carboxyl groups and one secondary amino group (Wright et al., 1990; Jeffery et al., 2004). All four groups are charged at neutral pH, and the carboxyl groups become successively protonated as pH decreases, yielding five possible protonated forms of domoic acid (Quilliam, 2003; Jeffery et al., 2004). There

389

are currently ten known isomers of domoic acid, including the isodomoic acids A through H and the domoic acid 5’ diestereomer (Jeffery et al., 2004). Domoic acid and other members of the kainoid class are glutamate analogues that interfere with neurochemical pathways by binding to glutamate receptors of brain neurons (Wright et al., 1990; Quilliam, 2003). The resulting effect of these neuroexcitants, or excitotoxins, is a continuous stimulation of the neurons, which can lead to rupture and/or eventual formation of lesions (Wright et al., 1990). Depolarized neurons result in short-term memory loss (Clayden et al., 2005), which has led to the common name for the illness related to the consumption of seafood contaminated with domoic acid: amnesic shellfish poisoning (ASP). Symptoms of ASP include gastroenteritis (vomiting, diarrhea, abdominal cramps) that can be experienced in humans within 24 h after ingestion, and neurological symptoms of confusion, memory loss, disorientation, seizures, coma and/or cranial nerve palsies that are typically experienced within 48 h (Perl et al., 1990; Wright et al., 1990). The number of human illnesses resulting from domoic acid poisoning has been few (Horner et al., 1997), likely due to active monitoring of fisheries. However, cultured blue mussels (Mytilus edilus) contaminated with domoic acid poisoned 107 people and killed three during the first major documented ASP outbreak in 1987 on Prince Edward Island, Canada (Perl et al., 1990). ASP poses a serious threat to marine wildlife, and the deaths of thousands of marine mammals and sea birds have been attributed to domoic acid intoxication (Bates et al., 1989; Scholin et al., 2000; Gulland et al., 2002; Caron et al., unpublished data). The first documented poisoning episode of marine animals related to domoic acid on the U.S. west coast was attributed to Pseudo-nitzschia australis and occurred in September 1991 off central California (Table 2; Buck et al., 1992; Fritz et al., 1992). High concentrations of domoic acid were also detected in Washington and Oregon in the 1990s (Wekell et al., 1994; Adams et al., 2000; Trainer et al., 2002), and a decade later in coastal waters off southern California (Schnetzer et al., 2007). The frequency and severity of these toxic events appears to be increasing (Trainer et al., 2007).

2.1.2.

Producers

The production of domoic acid and its isomers is confined to approximately a dozen chain-forming marine pennate diatom species within the genus Pseudo-nitzschia (Bates and Trainer, 2006), a genus containing species that form long chains of cells attached at their ends (Fig. 3a and b). The main toxin producing species that have been documented on the U.S. west coast include: P. australis, P. delicatissima, P. fraudulenta, P. multiseries, P. pungens, P. pseudodelicatissima, P. seriata and P. cuspidata (Tables 1 and 2). These species are distinguished based on fine morphological features of their silica frustules (Fig. 3a and b). These distinctions are subtle and require careful electron microscopical analysis and elaborate taxonomic training. As a consequence, historical misidentifications are not unusual and debates regarding some species descriptions are still unresolved. It is surprising that the first reports of ASP on the west coast of the U.S. were not recorded until the 1990s, even though Pseudo-nitzschia species have been recorded in surveys of

390

water research 44 (2010) 385–416

Fig. 2 – Chemical structures of commonly encountered toxins produced by microalgae in U.S. coastal waters.

phytoplankton species in the Southern California Bight since 1917 (Allen, 1922, 1924, 1928, 1936, 1940, 1941; Reid et al., 1970, 1985; Lange et al., 1994; Fryxell et al., 1997; Thomas et al., 2001). Given that these species generally comprise a significant portion of the total diatom assemblage in these waters, it can be surmised that either toxin production has increased in these west coast species, or that poisoning events prior to the 1990s have occurred but have not been attributed to these diatoms. Historical accounts of ‘unusual animal mortality events’ along the U.S. west coast tend to support the latter hypothesis. There have been increasing numbers of toxic events recorded along the U.S. west coast (Table 2), notably in Puget Sound (Trainer et al., 2003, 2007), Monterey Bay (Vigilant and Silver, 2007; R. Kudela, unpubl. data), Santa Barbara Channel (Trainer et al., 2000; Anderson et al., 2006; Mengelt, 2006), San Pedro Channel (Busse et al., 2006; Schnetzer et al., 2007), Newport Beach (Busse et al., 2006) and San Diego (Lange et al., 1994; Busse et al., 2006). Most recently, toxic blooms of Pseudonitzschia in the Long Beach-Los Angeles Harbor and San Pedro Channel have been particularly toxic, with some of the highest domoic acid concentrations recorded for the U.S. west coast (Caron et al., unpublished data). The increased incidence and severity of these toxic episodes off the western U.S. coast parallels the increase in frequency and intensity of harmful

algal blooms observed globally (Smayda, 1990; Hallegraeff, 1993, 2003; Anderson et al., 2002; Glibert et al., 2005b).

2.2.

Saxitoxins

2.2.1.

Toxin description and activity

Saxitoxin is a complex guanidine-based alkaloid that exists as more than 30 identified analogues in nature (Llewellyn, 2006). It is the most powerful marine toxin currently known and among the most dangerous poisons on Earth, except for some venoms and bacterial toxins (Schantz et al., 1957). Due to its acute toxicity, saxitoxin is currently listed as a chemical weapon in Schedule 1 of the Chemical Weapons Convention (Llewellyn, 2006). Saxitoxins display a rigid tricyclic core (Fig. 2; Table 3) and are very stable in biological and physiological solutions (Rogers and Rapoport, 1980). This nitrogenrich molecule and its chemical relatives are polar and have a positive charge at pH 7.7 (Shimizu et al., 1981). Consequently, they are soluble in water and alcohols, and insoluble in organic solvents (Schantz et al., 1957). Saxitoxins are known to disrupt the flow of ions through voltage gate sodium channels (Catterall, 1992; Cestele and Catterall, 2000). It has also been recently discovered that they have the ability to bind to calcium (Su et al., 2004) and

Table 2 – Distribution and concentrations of marine toxins in plankton of confirmed toxin producers in U.S. west coast waters. Toxin(s) Domoic acid

Location and year

Causative species

Particulate mg L1 (nmol L1)

Cellular pg cell1 b.d.–4.6

Washington coast and Juan de Fuca Eddy, WA (1997, 1998)

P. pseudodelicatissima Pseudo-nitzschia spp.

b.d.–2.7 3.6–8.7

Penn Cove, WA (1997)

P. pungens P. multiseries P. australis P. pseudodelicatissima

b.d.–0.8

Washington coast, WA (2001)

P. australis

b.d.–0.03

Pseudo-nitzschia spp.

(0.4–15)

Puget Sound, WA (2005)

P. pseudodelicatissima Pseudo-nitzschia spp.

b.d.–14

Central Oregon coast, OR (1998)

P. australis

Pt. An˜o Nuevo, San Francisco, CA (1998)

References Adams et al. (2000), Trainer et al. (2001, 2002) Trainer et al. (1998)

Marchetti et al. (2004) 4

2 10 –0.3 0.1–94.4

a

(b.d.–4.3) (1–5)

a

Baugh et al. (2006)

Trainer et al. (2007)

0.5

35

Trainer et al. (2001)

P. pungens P. multiseries

0.1–0.7

0.3–6.3

Trainer et al. (2000)

Bolinas Bay, San Francisco, CA (2003)

P. australis

0.15–9.4

Monterey Bay, CA (1991, 1998)

P. australis

b.d.–12.3 0.1–6.7

3–37 7.2–75

Monterey Bay, CA (1998)

P. pseudodelicatissima P. multiseries

0.1–0.4 0.67

0.8–1.2 6

Monterey Bay, CA (2000)

Pseudo-nitzschia spp. P. australis

Monterey Bay, CA (2002–2003)

Pseudo-nitzschia spp.

Morro Bay, CA (1998)

P. australis

1.3–7.4

37–78

Trainer et al. (2000, 2001)

San Luis Obispo, CA (2003–2005)

P. australis P. multiseries

1.5–7.6

9–38

Mengelt (2006)

Point Conception, CA (1998)

P. australis

2.2–6.3

15–22

Trainer et al. (2000)

Santa Barbara, CA (1998)

P. australis P. pungens P. pseudodelicatissima

0.5–1.2

0.1–0.9

Trainer et al. (2000)

Howard et al. (2007)

b.d.–24

24

Buck et al. (1992), Garrison et al. (1992), Walz et al. (1994), Scholin et al. (2000) Trainer et al. (2000, 2001) b.d.–8491

water research 44 (2010) 385–416

Washington coast, WA (2003)

Dissolved pg mL1 (nmol L1)

Bargu et al. (2002, 2008)

Vigilant and Silver (2007)

391

(continued on next page)

392

Table 2 (continued ) Toxin(s)

Saxitoxins

Location and year

Causative species

Particulate mg L1 (nmol L1)

Cellular pg cell1

0.03–1.7

0.14–2.1

Anderson et al. (2006)

6–12

b.d.–80

Mengelt (2006)

b.d.–117

Schnetzer et al. (2007)

P. australis

Santa Barbara (Santa Rosa Island and north San Miguel) (2004)

P. australis P. multiseries

Southern California Bight, CA (2003, 2004)

Pseudo-nitzschia spp. P. australis P. cuspidata

5.6–12.7

San Diego and Orange counties, CA (2004)

P. australis P. multiseries

b.d.–2.33

Sequim Bay, WA (2004–2007)

Alexandrium spp.

0.02–0.5

Oregon coast, OR (2004)

Alexandrium spp.

0.004–0.028

Humboldt Bay, CA (2004)

A. catenella

San Mateo County coast, CA (2004) Monterey Bay, CA (2004)

A. catenella

Monterey Bay, CA (2003–2005)

A. catenella

Morro Bay, CA (2004)

A. catenella

Yessotoxin

La Jolla, CA (1993)

Lingulodinium polyedrum

Brevetoxins

Indian Inlet, Bald Eagle Creek and Torque Canal, DE (2000)

Chattonella cf. verruculosa

b.d.: Below detection limit. a Toxin concentration from cells in culture.

Busse et al. (2006)

150–800

Jester (2008)

2.1–62.6

a

Jester (2008)

0.6–31.3

a

Jester (2008)

1.4–16.6

a

b.d.–0.962

0.008–<0.2

Lefebvre et al. (2008) Jester et al. (unpublished data)

1.6–19a

A. catenella

References

Jester et al. (2009b) Jester (2008)

0.002–0.02a 0–0.05a

Armstrong and Kudela (2006) Howard et al. (2008)

6

Bourdelais et al. (2002)

water research 44 (2010) 385–416

Santa Barbara Channel, CA (2003)

Dissolved pg mL1 (nmol L1)

water research 44 (2010) 385–416

potassium channels (Wang et al., 2003) and to be a weak inhibitor of neuronal nitric oxide synthase (reviewed in Llewellyn, 2006). These activities directly affect the nervous system, and the consumption of seafood containing saxitoxin can result in serious human illness and death, commonly referred to as paralytic shellfish poisoning (PSP). Minor symptoms of PSP, such as burning or tingling sensation of the lips and face, dizziness, headache, salivation, intense thirst and perspiration, vomiting, diarrhea and stomach cramps, can be experienced within 30 min after the consumption of contaminated seafood (Llewellyn, 2006). The consumption of a lethal dose can result in death within hours due to muscular paralysis and respiratory difficulty followed by complete respiratory arrest. PSP outbreaks result in more than 2000 illnesses worldwide each year, with a 5–10% mortality rate (Hallegraeff, 2003). PSP toxins also have adverse effects on marine wildlife that can cause mortalities among fish, marine mammal and seabird populations (Geraci et al., 1989; Montoya et al., 1996; Shumway et al., 2003). There are presently no records of unusual animal mortality events along the Californian coast that are attributable to saxitoxin poisoning (Jester et al., 2009b), but occurrence of the toxins in species consumed by humans is sufficient to warrant year-round monitoring.

2.2.2.

Producers

Saxitoxins are biosynthesized by dinoflagellates in marine ecosystems, most notably species within the genus Alexandrium (Fig. 3c), as well as Gymnodinium catenatum, Pyrodinium bahamense var. copressum, and by some cyanobacteria in freshwater ecosystems (Hallegraeff, 2003). Blooms of toxic and noxious dinoflagellates are often referred to as ‘red tides’ because of the red discoloration of water created by the accessory pigments of the cells. However, toxic levels of saxitoxins can be attained at dinoflagellate abundances that do not significantly discolor the water because of the exceptionally high potency of saxitoxins (Burkholder et al., 2006). This situation exists for Alexandrium in that it does not typically reach ‘bloom’ abundances on the U.S. west coast, and constant toxin monitoring is necessary to identify toxic conditions (Langlois, 2007; IOC HAB Programme, 2008; Jester et al., 2009a). Alexandrium species and measurable saxitoxin concentrations are common along the U.S. west coast (Table 1), although concentrations reported for this toxin typically have not been as high as noted along the U.S. east coast (Table 2). Thus, few west coast studies have contributed to our understanding the dynamics of Alexandrium abundances and saxitoxin production while ongoing research in the Gulf of Maine represents the most comprehensive regional study of this dinoflagellate (Anderson et al., 2005; McGillicuddy et al., 2005). Combined field observations, laboratory studies and modeling efforts have led to a scenario for toxic events along the northeastern coast of the U.S. that involve an interplay between river runoff, resuspension of dinoflagellate cysts from coastal sediments, favorable offshore growth conditions, and winds that generate onshore flow into coastal shellfish areas. A monitoring study of PSP in Puget Sound (WA) from 1993 to 2007 underscores that the timing and location of PSP outbreaks and high Alexandrium abundances are highly variable and not easily predicted from local or large-scale climate data (Moore et al., 2009). However, the study points out that

393

periods of warm air and low stream flow may favor saxitoxin accumulation in sentinel mussels (Moore et al., 2009).

2.3.

Brevetoxins

2.3.1.

Toxin description and activity

Brevetoxins are polyether, non-polar compounds that depolarize cell membranes by opening voltage gate sodium ion channels and induce enhanced inward flux of ions into cells (Lin et al., 1981; Baden, 1983, 1989; Purkerson et al., 1999). Brevetoxins exist as two structural types and multiple analogs possessing various levels of potency (Baden, 1989; Cembella, 2003; Kirkpatrick et al., 2004). The types differ in their ladderframe polycyclic ether structural backbones and are designated type A and type B (Fig. 2). The brevetoxin derivatives found in the marine environment (PbTx-2, PbTx-3 and PbTx-9; Table 3) are produced most commonly by dinoflagellate and raphidophyte algae and are of the structural type B (Baden, 1989; Baden et al., 2005). Brevetoxins bind to site 5 of the voltage-sensitive sodium channel in neurons, causing these channels to remain open and fire repeatedly (Catterall, 1992; Cestele and Catterall, 2000). Brevetoxin poisoning in humans is referred to as neurotoxic shellfish poisoning (NSP), and includes gastrointestinal symptoms of nausea, diarrhea and abdominal pain, neurologic symptoms of paresthesia, and respiratory irritation and/or failure (Kirkpatrick et al., 2004). The effects of brevetoxins on human health are well documented along the western coast of Florida where severe, nearly annual red tides caused by the dinoflagellate Karenia brevis release large amounts of brevetoxins into the air when the fragile cells are broken in breaking waves at the water’s edge (Kusek et al., 1999; Kirkpatrick et al., 2004). The aerosolized toxins constitute a significant health risk when they are inhaled and, as a result, K. brevis blooms are one of the most intensively studied and best-understood regional HABs. Red tides caused by K. brevis have been implicated in marine mammal fatalities, fish kills and human illnesses. Brevetoxins have not been reported from the U.S. west coast, and therefore no known human fatalities or health issues have yet been attributed to brevetoxins from that region.

2.3.2.

Producers

Several dinoflagellate species and a few raphidophyte species produce a suite of brevetoxins (Baden, 1989). K. brevis, the most notorious brevetoxin producer within the Gulf of Mexico, has not been observed on the west coast of the U.S., but several species of raphidophytes that are potential brevetoxin producers have been documented (Tables 1 and 2). Heterosigma akashiwo (Fig. 3e), Chattonella marina (Fig. 3f) and Fibrocapsa japonica have been isolated and cultured from coastal waters off southern California. In general there are few reports of significant blooms of these species on the west coast, although blooms of raphidophytes in San Francisco Bay and Delaware Inland Bays have been observed with cell abundances in excess of 108 cells L1 (Herndon et al., 2003; Coyne et al., 2005). In part, this lack of information is a consequence of the fact that raphidophyte species are notoriously difficult to identify using traditional microscopical techniques because they preserve poorly (Hallegraeff and

394

water research 44 (2010) 385–416

water research 44 (2010) 385–416

Hara, 1995; Throndsen, 1997). Recently developed genetic approaches for the identification and quantification of some raphidophytes are beginning to provide much-needed tools for studying the distributions and ecology of these HAB species (Handy et al., 2006; Demir et al., 2008). Despite the difficulties of characterizing these blooms, fish kills have been attributed to raphidophyte blooms on the west coast of the U.S. although these studies have not quantified brevetoxins (Hershberger et al., 1997; Hard et al., 2000).

2.4.

Diarrhetic shellfish toxins

2.4.1.

Toxin description and activity

Toxins that cause diarrhetic shellfish poisoning (DSP) include okadaic acid, dinophysistoxins and pectenotoxins (Ramsdell et al., 2005). Okadaic acid is a monocarboxylic acid named for the marine sponge Halichondria okadai from which it was first isolated (Tachibana et al., 1981). Okadaic acid can also be found in natural water samples in polar and non-polar esteric forms (Prassopoulou et al., 2009). The first dinophysistoxin described was isolated from the mussel M. edilus and was found to be a methyl form of okadaic acid (Murata et al., 1982). Okadaic acid and dinophysistoxins are linear polyethers (Fig. 2) and the mode of action is the inhibition of protein phosphatases (Takai et al., 1987; Bialojan and Takai, 1988; Haystead et al., 1989), enzymes that play a key role in dephosphorylation in many biological processes including cell cycle regulation. The pectenotoxins are lipid soluble and differ structurally from other diarrhetic toxins in that they possess a lactone ring (Fig. 2), and not considered to be protein phosphatase inhibitors, but have a high actin-depolarizing action (Hori et al., 1999). There is speculation that pectenotoxins may not produce diarrhetic effects (Cembella, 2003). DSP toxins (Table 3) were named for the human symptoms resulting from the ingestion of contaminated shellfish, including inflammation of the intestinal tract, diarrhea, abdominal cramps, vomiting and nausea beginning 30 min to a few hours after ingestion (Hallegraeff, 2003). In addition to the symptoms listed above, okadaic acid is known to be a strong tumor promoter (Suganuma et al., 1988), although the potential health implications of this activity due to the ingestion of contaminated seafood is unknown. There are presently no documented cases of DSP resulting from okadaic acid, dinophysistoxins or pectenotoxins on the U.S. west coast, and thus these toxins are not regularly monitored in the marine environment. DSP toxins have been detected in mussels and water samples from California (Sutherland, 2008), so it is possible that DSP has occurred on the U.S. west coast but has been attributed to other sources of contamination.

2.4.2.

395

Producers

Okadaic acid and dinophysistoxins are produced by a few species of the dinoflagellate genus Prorocentrum (Cembella, 2003) and most commonly by species of the genus Dinophysis. Dinophysis species present on the western U.S coast include D. acuminata, D. acuta, D. caudata, D. fortii, D. norvegica, D. rotundata, and D. tripos (Table 2). D. acuminata and D. fortii have been documented in Californian waters for many years (Bigelow and Leslie, 1930). D. acuminata produces okadaic acid (Yasumoto et al., 1985), D. fortii produces okadaic acid, dinophysistoxins and pectenotoxins (Yasumoto et al., 1980; Murata et al., 1982) while D. rotundata and D. tripos produce dinophysistoxin-1 (Lee et al., 1989). Dinophysis species are technically not phytoplankton, but heterotrophic protists that retain chloroplasts acquired from their prey. Dinophysis acquires its chloroplasts by preying on ciliates, which in turn prey on cryptophyte algae. This complex trophic relationship has made the culture of these species unsuccessful until recently (Park et al., 2006), and therefore no information on the environmental conditions that influence toxin production by these species presently exists. However, genus members are easily distinguished by light microscopy because they possess a pronounced ‘keel’ and other unique morphological aspects (Fig. 3d). These species are often encountered in plankton samples. Abundances of 103 cells L1 are commonly encountered (Nishitani et al., 2005), and occasionally they may reach abundances in excess of 105 cells L1 (Carpenter et al., 1995).

2.5.

Yessotoxin

2.5.1.

Toxin Description

Yessotoxin (Fig. 2; Table 3) is a chiral molecule of high polarity due to the presence of two sulfate groups. The molecule consists of fused polyether rings organized into a ladder shaped skeleton (Yasumoto and Murata, 1993; Wright and Cembella, 1998), a structure similar to other ladder-like polyether toxins such as the ciguatera toxin complex (ciguatoxins and maitotoxin), gambieric acids and brevetoxins (Yasumoto and Murata, 1993; Wright and Cembella, 1998). There are nearly 100 analogs of yessotoxin that have been identified to date (Satake et al., 1997, 1999; Ciminiello et al., 1998, 2000, 2001; Daiguji et al., 1998; Miles et al., 2004, 2005a,b, 2006; Paz et al., 2006). The yessotoxin class was named for the species of scallop, Patinopecten yessoensis, in which it was initially detected (Murata et al., 1987). Yessotoxin was originally classified in the DSP-toxin class because it was detected with other DSP toxins, but it appears that it does not induce

Fig. 3 – Major algal toxin producers occurring along the U.S. west coast. (a) Pseudo-nitzschia australis, a producer of domoic acid; (b) Scanning electron micrograph of Pseudo-nitzschia australis; (c) Alexandrium catenella, a producer of saxitoxin; (d) Dinophysis sp., a producer of okadaic acid; (e) Heterosigma akashiwo, a producer of brevetoxins; (f) Chattonella marina, a producer of brevetoxins; (g) Cochlodinium sp.; (h) Lingulodinium polyedrum, a producer of yessotoxin; (i) Phaeocystis globosa colony; (j) foam produced by the prymnesiophyte, Phaeocystis accumulating along the shore; (k) Unconcentrated seawater from King Harbor, City of Redondo Beach, with significant discoloration due to an algal bloom; (l) Prorocentrum sp., the dominant organism in (k); and (m) higher magnification of Prorocentrum sp. from (l). Scale bars [ 10 mm. Photo (b) courtesy of Peter Miller, (c) courtesy of Carmelo Tomas, (j) courtesy of Cindi Heslin.

396

Table 3 – Summary of toxins that can be present in Southern California waters. MW: molecular weight. Toxin

Properties

Formula

MW

Mode of action

References

Hydrosoluble At pH 7: DA3

C15H21NO6

311.14

Binds to glutamate receptors in the brain disrupting normal neurochemical transmission

Wright et al. (1990), Quilliam (2003)

Saxitoxins (STXs)

Hydrosoluble pH 7: Stable

C10H17N7O4

299.3

Bind to site 1 of voltage-sensitive sodium channels and block sodium conductance; bind to calcium and potassium channels

Wong et al. (1971), Wang et al. (2003), Su et al. (2004), Catterall (1992), Cestele and Catterall (2000)

Brevetoxins (PbTxs) Brevetoxin 2 (PbTx 2) Brevetoxin 3 (PbTx 3) Brevetoxin 9 (PbTx 9) Diarrhetic shellfish toxins Okadaic acid (OA)

Liposoluble C50H70O14 C50H72O15 C50H74O14

895.1 897.1 899.1

Bind to site 5 of voltage-sensitive sodium channels, shifting activation to more negative membrane potentials and block channel activation

Lin et al. (1981), Baden (1983, 1989), Purkerson et al. (1999)

C44H68O13

805

Inhibits protein phosphatases, inhibits dephosphorylation of proteins

Tachibana et al. (1981), Murata et al. (1982), Yasumoto et al. (1985), Takai et al. (1987), Bialojan and Takai (1988), Haystead et al. (1989), Hori et al. (1999)

Dinophysistoxins (DTXs) Pectenotoxins (PTXs) Yessotoxins (YTXs)

Liposoluble Hydrosoluble

High actin-depolarizaing action C55H80O21S2Na2

1187.3

Activation of phosphodiesterase in the presence of external Ca2þ; Disruption of the E-cadherin–catenin system in epithelial cells and potentially disrupting its tumour suppressive functions

Murata et al. (1987), Takahashi et al. (1996), Alfonso et al. (2003), Ronzitti et al. (2004)

water research 44 (2010) 385–416

Domoic acid (DA)

water research 44 (2010) 385–416

diarrhetic effects (Ogino et al., 1997). Accordingly, the regulation of the European Commission on marine biotoxins now considers yessotoxins separately from DSP toxins (European Commission, 2002). The great number of yessotoxin analogs complicates toxicity studies, and may explain the sometimes contradictory reports of its mode of action. Studies have shown that lysosomes, the immune system and the thymus (with tumorigenic implications) are the biological targets of yessotoxin (Franchini et al., 2004; Malagoli et al., 2006), while other reports have indicated cardiotoxic effects (Terao et al., 1990; Ogino et al., 1997; Aune et al., 2002). The cardiotoxicity of yessotoxin might be attributed to phosphodiesterase activation in the presence of external calcium ions (Alfonso et al., 2003). Unlike the other marine toxins mentioned above, there have been no reported human health issues or marine mammal deaths associated with yessotoxins.

2.5.2.

Producers

There are three known yessotoxin-producing dinoflagellates, Protoceratium reticulatum, Lingulodinium polyedrum, and Gonyaulax spinifera, and they have all been observed in coastal waters off the western U.S. (Table 1). According to phylogenetic analyses of available rRNA gene sequences, the capacity for yessotoxin production appears to be restricted to the order Gonyaulacales (Howard et al., 2009). However, toxin production among strains within each species appears to be highly variable. Yessotoxin has been detected in L. polyedrum isolates cultured from around the globe (Tubaro et al., 1998; Draisci et al., 1999a; Strom et al., 2003; Paz et al., 2004), including isolates from Californian coastal waters (Armstrong and Kudela, 2006). The reported cellular concentrations in the latter cells ranged from below detection to 1.5 pg cell1, indicating that L. polyedrum is significantly less toxic than P. reticulatum or G. spinifera. Yessotoxin has been recorded in blue mussels at low concentrations along the U.S. west coast (Table 2) during red tides caused by L. polyedrum, as well as during non-bloom conditions, but yessotoxin production by this dinoflagellate appears to be less than toxin levels produced by isolates from other geographical regions (see Table 2 in Howard et al., 2008). Expansive and dense blooms of L. polyedrum (Fig. 3 h) have been reported in California since 1901, but there have only been anecdotal reports of health problems associated with the red tides caused by L. polyedrum (Kudela and Cochlan, 2000). Yessotoxin production by isolates of P. reticulatum has been confirmed (Satake et al., 1997; Boni et al., 2002; Miles et al., 2002; Riobo et al., 2002; Stobo et al., 2003; Samdal et al., 2004; Eiki et al., 2005), but concentrations ranging from below detection to 79 pg cell1 have been reported for isolates from Washington, California and Florida (Paz et al., 2004, 2007). Isolates of G. spinifera appear to be the most prolific yessotoxin producers. Concentrations in New Zealand isolates ranged from below detection to 200 pg cell1 (Rhodes et al., 2006), more than 20-fold higher than the per-cell toxicity of P. reticulatum and 600-fold higher than L. polyedrum. G. spinifera does not generally bloom in high densities on the U.S. west coast, but it has been frequently observed at low abundances (Howard et al., 2008; M. Silver, pers. comm.) and has reached bloom concentrations in

397

Tomales Bay, north of San Francisco (G. Langlois pers. comm.). Yessotoxins are monitored in New Zealand, Europe and Japan but they are not routinely measured on the U.S. west coast. Howard et al. (2008) was the first study to confirm yessotoxins in California and Washington coastal waters, albeit at very low concentrations.

2.6.

Toxin detection and quantification

A wide variety of methodologies and technologies exist to detect, characterize and quantify the major toxins produced by microalgae. These approaches can be broadly divided into those used to characterize biological activity (toxicity assays) and those used to identify specific chemical structure(s) (immunological, various analytical techniques). Because of the highly variable approaches employed, and in most cases the highly diverse set of compounds comprising a toxin class, the methods provide somewhat different estimates of absolute toxin concentrations or presumed toxicity. It is also noteworthy that the majority of the protocols used to measure algal toxins have focused on the analysis of tissue samples that may be a source of human contamination (e.g. shellfish or finfish tissue) or plankton material filtered from seawater samples. Relatively few studies have examined the concentrations of toxins dissolved in seawater (Table 2). For this reason, limits of detection for toxins in seawater are poorly known for most approaches. However, based on analytical approaches presently available, a practical limit of detection for a few of the major concerns (domoic acid and saxitoxins) is in the range of 0.1 mg per liter of seawater for immunological approaches, but a detection limit of approximately 0.01 mg per liter for dissolved domoic acid in seawater using highperformance liquid chromatography has been reported (Pocklington et al., 1990). Detection limits for toxins in particulate material in the water can be significantly lower because particles can be concentrated by filtration prior to extraction. Knowledge of the concentrations of dissolved toxins would be preferable from the perspective of reverse osmosis desalination operations because dissolved toxins (rather than cell-bound toxins) would most likely impact these membranes. Domoic acid has been detected and quantified in seawater, plankton, shellfish extract and homogenate, as well as sea bird and mammalian body fluid (e.g. blood, urine, amniotic fluid, cerebral spinal fluid). Analytical approaches for these measurements include commercially available immunological techniques (enzyme-linked immunosorbent assay; ELISA) (Garthwaite et al., 1998, 2001), high pressure liquid chromatography (HPLC) with ultraviolet (UV) diode array detection (DAD) (Quilliam et al., 1989; Quilliam, 2003), receptor binding assay (RBA) (Van Dolah et al., 1994), mouse bioassay (MBA) or liquid chromatography–mass spectrometry (LC–MS). The results obtained by these various approaches are not yet completely compatible or comparable, and therefore comparisons across studies using different analytical methods can be problematic. In general, the choice of an approach is a compromise between cost of analysis (or access to costly equipment), sample throughput, sensitivity and analytical goal (e.g. thorough chemical characterization versus overall

398

water research 44 (2010) 385–416

toxicity, or human health risk versus scientific understanding of bloom dynamics). Saxitoxins can be rapidly detected and quantified with commercial ELISA kits, but the high specificity of these tests precludes the recognition of certain members of the saxitoxin family, especially the neo-saxitoxin (Garthwaite et al., 2001). Saxitoxin detection and quantification is often accomplished by HPLC, RBA, LC–MS, and MBA. Regulatory programs for seafood consumption are still based on ‘lethal mouse dosage’. Similarly, detection and quantification of brevetoxin and its derivatives in seawater, shellfish, and mammalian body fluids can be accomplished using commercially available ELISA kits (Naar et al., 2002), by HPLC, RBA (Van Dolah et al., 1994), LC–MS and MBA. A comparative study also quantified brevetoxins by radioimmunoassay (RIA) and a neuroblastoma (N2A) cytotoxicity assay (Twiner et al., 2007). Quantification of the DSP toxin suite can be underestimated by ELISA because commercial ELISA assays are usually optimized to detect okadaic acid and not the dinophysistoxins (Garthwaite et al., 2001). HPLC with fluorimetric detection (HPLC-FLD) has been commonly used to detect and quantify okadaic acid, its polar and non-polar esters, as well as dinophysistoxins (Lee et al., 1987). The MBA method has also been routinely used for the detection of DSP toxins. The detection and quantification of yessotoxin is problematic because of the extensive suite of derivatives that may exist. HPLC-FLD analysis (Yasumoto and Takizawa, 1997), MBA, LC-MS (Draisci et al., 1999b; Paz et al., 2006, 2008) and ELISA (Samdal et al., 2004, 2005) have been employed.

2.7. Other potentially toxic, noxious and nuisance organisms Reports of newly occurring HAB species, or recognition of extant issues that have gone previously undocumented, are

increasing our awareness of other potentially harmful bloomforming algae along the U.S. west coast. For example, blooms of an emerging potentially toxic organism, Cochlodinium sp. (Fig. 3g) off central California have recently been reported (Curtiss et al., 2008; Iwataki et al., 2008; Kudela et al., 2008b). This species is difficult to identify using light microscopy, and therefore researchers have begun to use gene sequences to provide accurate identification (Iwataki et al., 2007, 2008; Matsuoka et al., 2008). While this organism has only recently reached sufficient abundances to discolor Californian waters, there has already been one reported abalone loss in central California that appears to be associated with a bloom of Cochlodinium (R. Kudela pers. comm). Species of the dinoflagellate Prorocentrum (Fig. 3k–m) occasionally bloom in Californian coastal waters where they can attain very high abundances periodically, causing discolorations of the water and nuisance accumulations of algae (Holmes et al., 1967; Shipe et al., 2008). The Prorocentrum species known to occur in Californian waters have not yet demonstrated DSP toxin production, but species from elsewhere in the world are known to produce theses toxins (Table 1). Similarly, massive blooms of the dinoflagellate Akashiwo sanguinea are common in coastal waters of southern and central California and have recently been the cause of seabird mortality due to surfactant-like proteins (Jessup et al., 2009). Although they may not be overtly toxic, these blooms can cause animal mortalities, deplete oxygen, and result in an increased organic and biomass loading to a seawater desalination facility. The prymnesiophyte Phaeocystis globosa infrequently attains high abundances off the Californian coast (Armonies, 1989). This species produces single cells that are <10 mm in size, but it also forms fluid-filled colonies several millimeters in diameter in which individual cells are embedded in the polysaccharide skin of the colony (Fig. 3i). Single cells of

Fig. 4 – Time series of chlorophyll a fluorescence, temperature and dissolved oxygen in King Harbor, City of Redondo Beach, CA. Note the short-term temporal fluctuations in these parameters that are a result of tide, wind and biological interactions. These measurements were collected using autonomously recording sensors that provide high resolution observations of chemical and physical properties that might indicate an algal bloom, or environmental factors that might stimulate a bloom.

water research 44 (2010) 385–416

Phaeocystis are consumed by many zooplankton species but the colonies are typically a poor food source. Selective feeding on single cells appears to favor colony formation and the accumulation of colonies in the water column (Netjstgaard et al., 2007). When released via colony destruction or algal life cycle events, the colony matrix material is easily worked into a ‘sea foam’ that can form layers many centimeters (even meters) thick at the ocean surface or along the coastline over fairly extensive regions (Fig. 3j).

3. Spatial and temporal patterns of harmful algae A fundamental aspect of the biology of harmful algal blooms, and of vital importance for desalination operations, is the tendency for rapid and dramatic changes in the spatial and temporal distributions of these species. These changes occur rapidly across a wide range of scales, and pose challenges for documenting and predicting these distributions. Numerous approaches and instruments have been developed to characterize the dynamics of phytoplankton communities. These approaches presently have significant limitations on their abilities to identify species composition of a bloom, but they provide crucial information on the emergence and longevity of bloom events as well as their vertical and geographical extent. This information can help seawater desalination facilities adjust operations to ensure reliable production for the duration of the bloom.

3.1.

Temporal variability

Significant temporal variations in the abundances of phytoplankton take place on time scales ranging from hours to decades. Short-term temporal variability (hours to a few weeks) can be a consequence of rapid population growth or consumption by herbivores, sinking of senescent populations, diel vertical migration, tidal movement, and aggregation or dispersal by physical processes such as water mass convergences or divergences. Diel vertical migration of several dinoflagellates has been attributed to geotaxis, phototaxis and nutrient status (Eppley et al., 1968; Blasco, 1978; Cullen and Horrigan, 1981; Levandowsky and Kaneta, 1987). Classic responses involve nighttime sinking out of surface waters to deeper water where nutrient concentrations are greater, and rising into surface waters for photosynthesis during daytime. Shifts in nutrient cell quotas that accompany these migrations may have significant implications for toxin production because cell toxicity can be related to nutrient status of the cells (Anderson et al., 1990; Flynn et al., 1994; John and Flynn, 2000; Flynn, 2002). Diel-to-weekly variations in phytoplankton abundance can be characterized using self-contained or wirelessly networked sensor packages. Data collected in King Harbor of the City of Redondo Beach, CA (Fig. 4) demonstrate the efficacy of these instruments for providing high temporal resolution of chlorophyll a fluorescence (which approximates phytoplankton biomass) and pertinent environmental factors (e.g. dissolved oxygen and temperature) that provide insight into the factors

399

controlling the pattern. These data reveal a 4-fold variation in phytoplankton standing stock over a two-day period. In addition, changes in water quality criteria were easily and rapidly identified (e.g. decrease in dissolved oxygen concentration near noontime on September 13). Daily variations in the latter parameter can be extreme at night during algal blooms. High resolution, short-term monitoring approaches allow rapid detection of sentinel parameters, and in turn provide information for the development of predictive models of HAB events. Chlorophyll a fluorescence sensors provide valuable information on the short-term temporal patterns of total algal biomass, but these instruments cannot identify noxious algal species within a phytoplankton assemblage. More sophisticated instruments now coming online offer that possibility. For example, the Environmental Sample Processor (http:// www.mbari.org/ESP) is an in situ instrument capable of performing real-time identifications of HAB and other microbial species, as well as toxin analyses such as domoic acid using on-board molecular analyses (Greenfield et al., 2006). Similarly, handheld devices now exist for the detection of some HAB species and toxins in the field (Casper et al., 2007). These rapid and highly specific analyses are becoming valuable tools for quick determinations of toxin presence resulting from algal blooms. These more costly instruments can be used to improve the information available on bloom composition once the cheaper and more readily available sensors identify an emerging bloom event. Seasonal variability of HAB species and their toxins along the Californian coast can be gleaned from the records of the Marine Biotoxin Monitoring Program (MBMP) of the California Department of Public Health (CDPH), a program started after a major domoic acid outbreak in the fall of 1991. At present, the annual effort involves the analysis of approximately 300 shellfish samples for domoic acid and >1000 samples for PSP toxins from all fifteen Californian coastal counties (CDPH, 2007). Shellfish toxin information provides a reasonable representation of toxins in the upper water column over seasonal and annual scales such as demonstrated in studies on toxic dinoflagellates (Montojo et al., 2006; Jester et al., 2009a; Moore et al., 2009). Detailed information on the sampling effort is provided in the MBMP annual reports (Langlois, 2007). Distributions of domoic acid and PSP toxins along the Californian coast during the period 2002–2007 based on the MBMP data reveal both seasonal and geographical trends (Figs. 5 and 6). Monthly averages for the 6-year period (histograms) and maximal concentrations (triangles and lines) showed detectable concentrations of PSP toxins and domoic acid in shellfish in nearly all months for all Californian coastal counties (Fig. 5). Domoic acid concentrations showed pronounced seasonality, with very high peaks during spring and much smaller peaks during fall. This temporal trend is concordant with previous observations along the southern Californian coast (Lange et al., 1994; Walz et al., 1994; Schnetzer et al., 2007; Shipe et al., 2008). Fall domoic acid peaks correspond to minor blooms of toxic Pseudo-nitzschia occasionally noted on the west coast of the U.S. (Bolin and Abbott, 1960; Buck et al., 1992; Walz et al., 1994; Trainer et al., 2002; Schnetzer et al., 2007).

400

water research 44 (2010) 385–416

Domoic acid (µg/100g)

70000 60000 50000 40000 30000 20000

Mean of monthly averages Absolute maximum

15000

10000

5000

0 1600 1400 1200 1000 800 600 400

PSP toxins (µg/100g)

300

200

100

0 Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Fig. 5 – Seasonal variability of domoic acid (upper panel) and PSP toxin (lower panel) concentrations along the Californian coast. Monthly averages of data collected from 2002 and 2007 summarized for fifteen Californian coastal counties are shown as histograms. Also shown are the maximal values recorded during each month over the entire study period (triangles and solid lines). Toxin concentrations were derived from shellfish tissue. Data from CDPH (Langlois, 2007).

Seasonal peaks in domoic acid along the U.S. west coast differ along a latitudinal gradient. Highest domoic acid concentrations in Washington have been observed during the fall (Trainer, 2002; Office of Shellfish and Water Protection, 2008; Moore et al., 2009) compared to spring blooms that are common in southern California (Schnetzer et al., 2007). This north-south trend in seasonality is also evident on a smaller latitudinal scale. Blooms along the Californian coast have been more frequently observed later in the year in northern counties (Humboldt versus Santa Barbara counties in Fig. 6). The time lag between bloom periods observed from south to north may be related to the timing of the California Current System (CCS) upwelling maximum, which brings nutrients into surface waters and promotes phytoplankton growth. The CCS upwelling occurs in early spring in southern California, in June off Washington,

and throughout summer in northern California and Oregon (Reid et al., 1956; Landry, 1989). Monthly averages as well as maximal PSP toxin concentrations showed less pronounced seasonal variability than domoic acid during the period 2002–2007, although the highest concentrations were recorded from July to September (Fig. 5). This seasonal pattern of maximal PSP is in agreement with the last 25 years of monitoring results (Langlois, 2007). It is also consistent with the 1927–1989 observations on the Californian coast indicating that most significant concentrations of the toxin take place between May and October (Price et al., 1991). The highest PSP toxin concentrations in shellfish have also been observed in the summer and the fall periods off the Washington and Oregon coasts (Trainer et al., 2002; Determan, 2003).

401

water research 44 (2010) 385–416

Domoic acid

PSP toxins Marin County 140

12000

120

10000

100

PSP toxins (µg/100 g)

North

Domoic acid (µg/100 g)

Humboldt County 14000

8000 6000 4000 2000

****

*

** **

**

2002 2003 2004 2005 2006 2007 * not detected

80 60 40 20

**

0

0 April

May

June

July

140

12000

120

10000

100

8000 6000 4000 2000

**

September

San Luis Obispo County

14000

PSP toxins (µg/100 g)

South

Domoic acid (µg/100 g)

Santa Barbara County

August

* * **

80 60 40 20

**

**

**

* *

**

*

0

0 April

May

June

July

August

September

Fig. 6 – Interannual variation in toxin concentration in four Californian coastal counties from 2002 to 2007 during three months exhibiting high concentrations of domoic acid (April, May, June) and saxitoxin (July, August, September). Humboldt and Marin counties are located north of Santa Barbara and San Luis Obispo counties. Toxin concentrations were measured from shellfish tissues. Data from CDPH (Langlois, 2007).

Diel to seasonal temporal patterns in ASP and PSP outbreaks off the U.S. west coast are augmented by large interannual variability in the intensity and frequency of HABs. Interannual variations in HABs presumably are related, at least in part, to changes in atmospheric and hydrographic features modulated by the 3–7 year cycles of El Nin˜o-Southern Oscillations (ENSOs) (Price et al., 1991; Horner et al., 1997), but the exact relationship between HABs and these climatic events is not clear because there are still relatively few observations spanning these temporal regimes. Warming off California during El Nin˜o episodes reduces seasonal upwelling, enhances physical stratification in the CCS and lowers the nutricline in the water column (the depth at which nutrient concentrations increase rapidly). It is clear that these changes in water stability and nutrient availability have significant impacts on plankton productivity and community structure, but the specific responses of the phytoplankton communities vis-a`-vis HAB events are not yet predictable (Barber and Chavez, 1983). ASP outbreaks occurred during El Nin˜o episodes of 1991 and 1997–98 along the coasts of Oregon, Washington and California (Table 2; Moore et al., 2008), but the specific factors contributing to toxic Pseudo-nitzschia blooms during these events could not be clearly identified (Horner and Postel, 1993; Trainer et al., 2000). Interannual variability in ASP and PSP concentrations in coastal waters along the Californian coast was evident and substantial in the MBMP dataset during the period 2002–2007 (Fig. 6). ASP and PSP outbreaks were frequent, but

they varied in intensity and the timing of peak concentrations between the years within a single geographical location, and between southern and northern Californian counties in the same year and season. Notably, the magnitude of this variability is on the same order of magnitude as the variability observed on short-term (daily) or seasonal time scales. Little is known regarding the longer time scale fluctuations in HABs along the U.S. west coast. Multi-decadal fluctuations in ocean temperature are known to provoke shifts in the biological regime (Chavez et al., 2003), and it is anticipated that climatic shifts would affect the timing, intensity or frequency of phytoplankton blooms. Long-term regime shifts may affect the occurrence of blooms of L. polyedrum, a producer of yessotoxin, along the coast of southern California (Tables 1 and 2). The Pacific Ocean was cooler in the years preceding 1976, and red tides dominated by L. polyedrum commonly developed along the Southern California Bight during fall (Gregorio and Pieper, 2000). During a recent warm regime (1976-mid 1990s), red tides occurred during winter and spring and persisted until summer in the region of the Los Angeles River mouth (Gregorio and Pieper, 2000). Recent massive blooms of L. polyedrum during fall may indicate a return to the pre-1976 conditions (Moorthi et al., 2006). The generality surmised from data depicting short- to longterm temporal variability in phytotoxin dynamics is that variability can be high at all scales. Given our present state of understanding regarding the specific combination of forcing factors that give rise to this high variability, it is difficult to

402

water research 44 (2010) 385–416

Fig. 7 – Two-dimensional presentations of chemical and physical data collected by an autonomous vehicle (Webb Slocum glider, Teledyne Webb Research, East Falmouth, MA) along a nearshore–offshore transect at Newport Beach, CA shown in (f) (indicated on map by the blue line). Contour plots are shown for (a) temperature, (b) chlorophyll a fluorescence, (c) salinity, (d) backscattered light and (e) water density. The Webb Slocum glider is an autonomous vehicle commonly employed in coastal ecosystems. This buoyancy-driven underwater vehicle generates horizontal motion by ascending and descending with pitched wings (Schofield et al., 2007). A rudder directs heading while buoyancy is controlled by pumping seawater into and out of the nose of the vehicle. This long-lived, low-power glider achieves horizontal velocities of approximately 25– 30 cm sL1 with vertical velocities of 10–15 cm sL1.

accurately predict the timing and magnitude of toxic blooms. For these reasons, monitoring at multiple temporal scales is necessary to adequately characterize plankton dynamics.

3.2.

Spatial variability

Spatial variability of HABs is considerable at multiple scales, analogous (and strongly related) to the temporal variability described above. Blooms can be highly localized (10s of meters) or expansive (100s of kilometers), and distributions vertically within the water column are heterogeneous over scales of centimeters to meters. The geographical extent and heterogeneous nature of the U.S. west coast results from differences in local hydrography that are manifested in smalland large-scale differences in spatial patterns of toxic blooms. Regional-scale variations in HAB distributions are illustrated by MBMP data during 2002–2007 for ASP and PSP concentrations observed in counties along northern and central California (Fig. 6). ASP events (frequency and level of

toxicity) were generally lower in northerly Humboldt County during this period relative to Santa Barbara County nearly 1000 km to the south. More recently, high domoic acid concentrations have been observed within the Southern California Bight, presumably indicating a continuing southward movement of the Pseudo-nitzschia spring blooms (Langlois, 2007; Schnetzer et al., 2007). On the other hand, Marin County (north of San Francisco) exhibited higher monthly PSP values during most months than San Luis Obispo County located nearly 500 km to the south. This general latitudinal trend in PSP events is consistent with findings that the three southernmost counties (Los Angeles, Orange and San Diego) generally experience low concentrations of PSP relative to northern California (Price et al., 1991; Langlois, 2007). Regional-scale and geographical differences in ASP and PSP events have also been reported along the coastline of Oregon (Trainer et al., 2002). Small-scale spatial variability (horizontally and vertically) can also be dramatic. HAB events can be highly restricted

water research 44 (2010) 385–416

geographically (e.g. relegated to a protected embayment). Even within spatially extensive blooms, phytoplankton biomass is often highly discontinuous over very small spatial scales because of differences in local circulation, and wind or wave forcing. Considerable spatial variability in the abundance of P. australis within Monterey Bay has been observed, a phenomenon that has been attributed to advective forces (Buck et al., 1992). Discontinuities within a water column, such as thermoclines, haloclines, nutriclines and light absorption often leads to the establishment of subsurface microlayers where phytoplankton biomass can be many-fold elevated relative to algal standing stocks only centimeters above or below (Dekshenieks et al., 2001; Rines et al., 2002). Characterization of phytoplankton spatial distributions include approaches ranging from shore-based sampling, to the use of oceanographic ships, to remote sensing of largescale patterns using satellite imagery. Vertical profiling of phytoplankton assemblages can be accomplished using over-the-side, ship-based instrument packages and more recently autonomous vehicles equipped with a variety of sensor packages. The use of autonomous vehicles to provide synoptic measurements of phytoplankton biomass from chlorophyll a fluorescence and pertinent chemical/physical parameters is state-of-the-art for obtaining two-dimensional cross-sections or three-dimensional patterns in the water column (Fig. 7). Autonomous vehicles provide time- and depth-stamped measurements of a variety of parameters that can be optimized for a specific mission. Such an instrument deployed off Newport Beach, CA (blue line in Fig. 7f) during May–June 2008 yielded detailed patterns of chemical/biological parameters that provided information on the extent and vertical distribution of phytoplankton biomass (Fig. 7a–e). Evidence of upwelling in this

33°45’N

33°40’N

33°35’N

118°10’W

118°00’W

0.01-0.50 0.51-1.00 1.01-1.50 1.51-2.00 2.01-2.50 2.51-3.00 3.01-3.50 3.51-4.00 4.01-4.50 4.51-5.00

Particulate domoic acid (µg L-1)

Fig. 8 – Spatial variability in domoic acid concentrations contained in the total particulate material (algal cells, other microbes and detritus) in surface waters within the San Pedro Bay area including the Long Beach-Los Angeles harbor area. Data were collected at 20 sampling stations (locations indicated by filled circles).

403

spring deployment was apparent from the upward-pointing isotherms (temperature) and isopycnals (density) on the shoreward end of the transect (left sides of Fig. 7a, e). This was particularly evident in the temperature plot 5–10 km from shore between the surface and 20 m (Fig. 7a, red circle). Nutrients (not shown) were generally depleted in surface waters, and increased with decreasing temperature. This physico-chemical structure is concordant with a significant subsurface maximum in chlorophyll a concentration, indicative of a response of the phytoplankton assemblage to elevated nutrient concentrations at this depth (red circle on left in Fig. 7b), as previously observed (Jones et al., 2002). The cross-sectional picture provided by the autonomous vehicle indicated that the phytoplankton community had a patchy structure on the scale of meters (vertically) and kilometers (horizontally). Two major horizontal patches were observed at 6–12 km and at 15–21 km from shore (red circles in Fig. 7b). Measurements in addition to chlorophyll a fluorescence can add information on the observed general patterns such as particle size distributions derived from the optical backscatter spectrum. The backscatter profile obtained at a wavelength of 532 nm indicated a patch of particles that were not of algal nature (Fig. 7d, red circle). The wavelength-dependent slope of the backscatter, which is dependent on the particle size distribution, can also be mapped to indicate the size-class of phytoplankton particles that dominate the chlorophyll a maxima. This information is particularly important for a seawater desalination facility, where the incoming particle size distribution is known to impact the source water filterability. Autonomous vehicles allow nearly synoptic measurements of the spatial distribution of phytoplankton and ancillary parameters. Many of these instruments operate for significant periods of time (weeks) and thereby supply temporal as well as spatial coverage. Knowledge of the temporal evolution and spatial organization of coastal marine systems enables a better understanding of the linkages between physical processes and the biological responses that contribute to the formation of algal blooms. Moreover, data from these instruments can be telemetered to the laboratory in near-real time and used to direct costly efforts such as shipboard sampling, or plan operations for land-based activities and measurements. Broad-scale, horizontal distributions can also be acquired via shipboard sampling programs (Fig. 8). Shipboard work is time and labor intensive, but onboard sample processing enables more sophisticated analyses than autonomous vehicles are presently capable of providing. Moreover, ships and other manned platforms permit time series studies at a single study site. Shipboard sampling conducted in April 2008 in the Long Beach-Los Angeles harbor area and the adjacent San Pedro Channel demonstrated considerable spatial variability in the distribution of HAB species and their toxins within this relatively small area (approximately 500 km2). Results indicated a patchy distribution of domoic acid in particulate material (i.e. within phytoplankton cells) collected near the surface with highest concentrations in the vicinity of the harbor breakwater and at several offshore locations (Fig. 8). Intermediate regions exhibited toxin concentrations that were more than an order of magnitude less than these maxima.

404

Environmental driving factors

45

7

40

Relative fluorescence TMP

6

35

5

30

4

25

3

20 15

2

10

1

5

Transmembrane pressure (TMP, psi)

Characterizing the factors that lead to the stimulation of harmful algae and the production of toxins by these algae has been an area of very active research for decades. These studies involve field observations to document the spatiotemporal extent of blooms and toxin concentrations in plankton and marine life, and laboratory experiments aimed at understanding the key environmental factors leading to HAB events and toxin production. The overall results gleaned from many years of work group into three basic categories: (1) factors and conditions leading to phytoplankton blooms in general, (2) factors leading specifically to the growth of HAB species, and (3) factors leading to toxin production. Numerous factors have been implicated as contributors to the observed global expansion of HABs (Smayda, 1990; Hallegraeff, 1993, 2003; Anderson et al., 2002; Glibert et al., 2005b). Phytoplankton blooms occur naturally as a consequence of the vertical mixing of deep, nutrient-rich waters into lighted surface waters. This process occurs seasonally in temperate environments due to winter storm events, and due to coastal upwelling events caused by appropriate regional wind conditions. There is no a priori reason why these ‘natural’ sources of nutrients cannot lead to HAB events, but the global increase in frequency and severity of HABs implies that human activities may be an underlying reason for this escalation. Eutrophication of coastal ecosystems is a growing global concern that has clear consequences for blooms of nearshore algal populations (Anderson et al., 2008; Heisler et al., 2008; Howarth, 2008). Nutrient enrichment has been implicated in harmful blooms occurring in some protected bays, but the linkage between nutrient discharges mediated by human activities and many HAB events is still unconfirmed. For example, field studies have shown that coastal upwelling of nitrate-rich waters can be a driving factor leading to toxigenic Pseudo-nitzschia blooms along the U.S. west coast (Horner et al., 2000; Scholin et al., 2000; Anderson et al., 2006) but the specific role of river/coastal runoff in domoic acid production is unclear (Scholin et al., 2000; Schnetzer et al., 2007). The importance of nutrient discharge into coastal waters is, of course, dependent on the amount of nutrients available for phytoplankton growth from natural sources but the latter term is poorly defined in most situations. Constructing nutrient budgets for coastal ecosystems is an area ripe for future work. In the meantime, it has been speculated that anthropogenic nutrient sources, such as elevated nutrient concentrations in river discharge, coastal runoff from agricultural land, and sewage discharge may significantly increase the total amounts of nutrients available for the growth of coastal phytoplankton (Scholin et al., 2000; Glibert et al., 2005a,b, 2006; Howard et al., 2007; Kudela et al., 2008a). There now exists a basic understanding of the general conditions that favor the growth of phytoplankton per se (Allen et al., 2008). Despite this basic understanding, there is still only limited information on the specific conditions that selectively stimulate the growth of harmful algal bloomforming species of phytoplankton. As a result, mathematical models that attempt to predict HAB events tend to be more correlative than deterministic (i.e., they identify the

conditions that may promote a HAB, rather than the conditions that will promote a bloom). One generality is that rarely can one identify a ‘silver bullet’, a single parameter or set of circumstances that provide an accurate prediction of the occurrence of a particular HAB species. The environmental circumstances leading to the dominance of a HAB population over all other species of algae in a given locale are composed of a complex set of physical, chemical and biological conditions with poorly known variances, and these conditions appear to be species-specific. Biological factors contributing to the success or demise of individual HAB taxa include allelopathy among competing phytoplankton, mixotrophy by HAB species, and the deterrence of potential consumers via the production of noxious or toxic compounds (Strom et al., 2003; Burkholder et al., 2008; Buskey, 2008; Flynn, 2008; Smayda, 2008). These biological interactions presuppose a period of stable environmental conditions in order that the scenarios of allelopathy, grazer deterrence or phagotrophic activity of HAB species can play themselves out. This requirement may explain why the formation of a stable water mass appears to play a role in the development of some HAB events (Scholin et al., 2000). Models predicting the population growth of potentially toxic algae are necessary for understanding bloom dynamics, but these models must also integrate information on toxin induction. Many toxins do not appear to be constitutively produced by algae, but are induced by a variety of specific environmental conditions that are not completely understood. Silica, phosphorus, nitrogen and trace metal limitations, and nutrient or elemental ratios (in addition to or instead of absolute concentrations) have all been implicated in toxin induction (Pan et al., 1996a, 1998; Rue and Bruland, 2001; Fehling et al., 2004; Wells et al., 2005; Grane´li and Flynn, 2006; Schnetzer et al., 2007). Again, species-specific differences (and perhaps strain-specific differences) may exist in the factors promoting toxin production. Physical aspects such as temperature and light intensity may stimulate toxin production by some harmful algae (Ono et al., 2000).

Relative fluorescence

3.3.

water research 44 (2010) 385–416

0

0 0

5

10

15

20

25

30

35

40

45

Days Fig. 9 – Correlation between chlorophyll a fluorescence (open squares, in relative fluorescence units) and transmembrane pressure (filled circles) in a pilot-scale reverse osmosis desalination system. Increased loading of phytoplankton biomass resulted in greater transmembrane pressure.

405

water research 44 (2010) 385–416

Pseudo-nitzschia spp. (cells L-1)

2.0x105

1.5x105

1.0x105

5.0x104

0

Particulate domoic acid (µg L-1)

4.0 3.0 2.0 1.0 0.5

0.0 3.0

Dissolved domoic acid (µg L-1)

2.5 2.0 1.5 1.0 0.5

6 ov -0 01 -N

6 01 -Se p-0

01 -Ju l-0 6

ay -06 01 -M

06 ar01 -M

n-0 6 01 -Ja

01 -N ov -05

5 01 -Se p-0

01 -Ju l-0 5

ay -05 01 -M

01 -M

ar-

05

0.0

Fig. 10 – Time series of abundances of Pseudo-nitzschia spp. cells (top), domoic acid concentrations in particulate material (middle) and dissolved in seawater (bottom) at a coastal monitoring site in El Segundo, CA.

4.

Desalination operations and HAB events

A thorough understanding of the specific factors and conditions giving rise to harmful algal blooms and toxin production in coastal waters will require a great deal of additional research before accurate models for predicting these toxic events will be readily available. Until then, appropriate monitoring strategies to detect imminent bloom events and the ability to track the evolution of an active bloom, coupled with an understanding of the potential toxins being produced,

the toxin chemistry, and their rejection by seawater reverse osmosis membranes, provide a seawater desalination facility with the best strategy for making operational adjustments to ensure that the treatment plant capacity or product water quality remains unaffected.

4.1. Concern with harmful algal blooms and their toxin production As seawater desalination has continued to become more costeffective and less energy intensive (Al-Sahlawi, 1999), many

406

water research 44 (2010) 385–416

communities are planning or implementing seawater desalination facilities (Al-Sahlawi, 1999; Burbano et al., 2007). The selected pretreatment procedures and the process engineering that determines the ultimate facility design is entirely dependent upon the source water quality (e.g. suspended solids, turbidity, organic material content, algal cell content, etc.) and its variations, particularly for facilities incorporating open intakes (Bonnelye et al., 2004b; Gaid and Treal, 2007). When the seawater desalination process is performed by reverse osmosis membranes, the selection and proper operation of a pretreatment system is paramount to the success of the downstream desalination process (Tenzer et al., 1999; Bonnelye et al., 2004b; Separation Processes Inc., 2005; Gaid and Treal, 2007; Petry et al., 2007). Algal blooms are known to have significant negative impacts on reverse osmosis desalination facilities. A variety of pretreatment trains have been considered to address the difficult source water quality associated with algal blooms, where the organic and biomass load increase dramatically (Adin and Klein-Banay, 1986; Al Arrayedhy, 1987; Hasan AlSheikh, 1997; Watson, 1997; Abdul Azis et al., 2000; Bonnelye et al., 2004a,b; Burbano et al., 2007; Gaid and Treal, 2007; Petry et al., 2007; Peleka and Matis, 2008). Recently, microfiltration and ultrafiltration membrane pretreatment has been identified as a component of a preferred pretreatment train due to the consistent, high quality water produced by membrane filtration, especially when compared to conventional processes (Wilf and Schierach, 2001). However, one significant drawback to implementation of these modern pretreatment technologies is that they are as susceptible, or possibly more so, to significant algal blooms (Bonnelye et al., 2004a). An early warning system can provide information to a seawater desalination facility so that functional changes can be made to efficiently maintain operations even as source water quality deteriorates. Turbidity sensors offer a rapid measurement of the total amount of suspended particles in the intake water for making these decisions. On in situ instruments such as the Slocum glider, backscatter measurements yield this type of information (Fig. 7d). Measurements of chlorophyll a fluorescence can augment this surveillance approach by estimating the degree to which algal bimoass contributes to the total load of suspended particles. Responses to deteriorating water quality may include chemical additions, use of additional pretreatment equipment, or additional staff preparations (e.g. maintenance activities, guaranteeing all membranes and filters are clean in preparation for an event) to continuously deliver a high quality feedwater to the reverse osmosis system that will produce the desalinated drinking water.

4.2. Addressing spatiotemporal variability in HAB abundance and early detection As detailed above, it is essential that a desalination facility incorporate a means of rapid algal bloom detection so that, when necessary, proper process changes can be made to maintain the production capacity. Sensors for detecting an eminent algal bloom can be located at the desalination facility to inform personnel regarding changes in water quality that are directly observed on the source water. Fig. 9 presents the

transmembrane pressure (TMP) of a microfiltration system that serves as pretreatment to a pilot-scale reverse osmosis desalination system along with the levels of chlorophyll a fluorescence observed in the feedwater. It is clear from this figure that increased membrane fouling rates (e.g. faster daily rise in the TMP) were associated with increasing chlorophyll a fluorescence (i.e. increased algal biomass) in the source water. It is well known that higher concentrations of algae cause increased membrane fouling rates in microfiltration systems that are frequently incorporated, or considered, in today’s desalination facilities (Gijsbertsen-Abrahamse et al., 2006; Lee and Walker, 2006; Reiss et al., 2006). A more complete approach might include a monitoring system located offshore that measures some of the primary factors influencing algal blooms, such as nutrient monitoring in near-real time using new in situ sensor technology (Glibert et al., 2008). Such information would be useful to both the desalination facility and HAB researchers who are continually improving their understanding of the causative factors that produce HABs and their associated toxins. Using the information provided by offshore sensors, the desalination facility personnel could note trends and shifts in driving factors that generate algal blooms and make any chemical orders or perform maintenance procedures that have significant lead times. The same offshore sensor might also incorporate real-time monitors of sentinel parameters for changes in algal biomass, such as turbidity and chlorophyll a, allowing the facility to prepare for changes in chemical additions and redundant equipment service. Monitoring of basic chemical parameters of seawater (e.g. chlorophyll a concentrations) will provide valuable information for facility operations, but this activity is not sufficient to fully assess potentially toxic conditions that might arise from algae that do not require high standing stocks to constitute a significant toxic threat, such as Alexandrium species. Species-specific approaches, such as automated in situ instruments or laboratory-based methods, as well as chemical/immunological analyses that identify and quantify specific algal toxins are necessary to more thoroughly characterize the potential hazards posed by HAB species. The consistent removal of these potentially toxic substances through the reverse osmosis process is both a function of size (e.g. molecular weight) and charge (e.g. zeta potential) (Amy et al., 2005). Depending on the size and charge of the contaminant of concern, the rejection, or removal, by the reverse osmosis process will differ. It is important that we continue to broaden our knowledge on potentially toxic substances excreted by algal stock and their associated blooms. The approach for obtaining this information would be best complemented with knowledge of the species that are present regionally, the potential problems they pose (e.g. specific toxins and the amounts of soluble microbial products and extracellular polymeric substances excreted), the spatial extent of HAB episodes, and their seasonality. The seasonality of Pseudo-nitzschia spp. and domoic acid near the intake of a pilot desalination plant in El Segundo, CA, exemplifies the usefulness of routine monitoring for identifying potentially toxic conditions in coastal waters adjacent to a plant (Fig. 10). Abundances of Pseudo-nitzschia spp. and concentrations of domoic acid contained in the

water research 44 (2010) 385–416

algal cells or dissolved in the intake water exhibited a springtime peak. Knowledge of the seasonality of this toxic bloom-forming species allows intensive sampling of coastal waters during spring when toxic events are more common, improving the overall effectiveness of the monitoring effort and making it more cost effective. Historical and real-time information on the spatial distribution of HABs can provide information vital for optimizing design and performance of desalination operations. Local/ regional hydrography, and resultant algal blooms can differ dramatically. When constructing a new intake pipeline, the selection of its location (e.g. depth and distance from shore) can be greatly enhanced through the use of offshore monitoring devices and efforts to take into account the presence of any local accumulations of algal biomass due to currents, water mass convergences/divergences or internal waves, and also subsurface maxima in algal abundance. Properly locating offshore monitoring can provide significant information that will allow optimal location of a new intake pipeline or identification of issues that might affect an existing one, thereby significantly reducing the organic and suspended solid loads present in the feedwater during algal bloom events. These considerations will ease pretreatment operations, reduce the cost of water production, and help improve the facility’s longevity.

5.

Conclusions

5.1. Potential impacts, unresolved issues and research prospectus The presence of harmful algae in coastal waters that might be employed in reverse osmosis desalination pose potential problems for these operations that have been known to even cause desalination facilities to temporarily cease production (Tenzer et al., 1999; Pankratz, 2008). As the number of seawater desalination facilities continues to grow with lower costs and increasing demand, it is essential that these operating facilities develop the tools necessary to allow process changes and ensure capacity objectives continue to be met. Regardless of the pretreatment configuration, changes in source water quality require adjustments and these changes need to carefully coordinate to ensure that the reverse osmosis membranes are not irreversibly fouled or damaged in the process. Benchmark work is required to establish the effectiveness of the seawater reverse osmosis process in dealing with HAB toxins and other phytoplankton-derived substances. Even if advanced pretreatment technologies such as microfiltration are implemented upstream of the reverse osmosis process, passage of transparent extracellular material produced by the algal bloom (Alldredge et al., 1993; Hong et al., 1997) may affect reverse osmosis membrane performance. Additionally, the physical durability of phytoplankton varies greatly and the pretreatment process might disrupt cells and create significantly higher concentrations of dissolved organic substances, including toxins, than were originally present in the source water. For example, dissolved domoic acid has been observed in the seawater passing through the prefiltration process (Fig. 10, bottom panel) but it is unclear if these values are

407

higher due to cell breakage as a result of the prefiltration process. Therefore, it is important that the international desalination community carefully characterize these potential contaminants and their removal to improve treatment approaches in seawater desalination. To our knowledge, there are no published reports on the effectiveness of reverse osmosis for removing dissolved algal toxins from seawater. Some of these toxin molecules (e.g. domoic acid) are near the theoretical molecular size of molecules rejected by reverse osmosis membranes, but experimental studies are required to validate the effective of this process on toxin removal. In addition, more information will be needed to understand the potential impact of discharged brine and pretreatment backwash water resulting from the reverse osmosis desalination process on the ecology of coastal ecosystems. The use of ferric sulfate or ferric chloride as a pretreatment coagulant would concentrate toxic algae and their associated toxins if they are present in the intake water. Similarly, the discharge of brine resulting from the reverse osmosis process would contain elevated concentrations of dissolved algal toxins relative to unfiltered seawater. The degree of concentration of these toxins would not be expected to be large, but the significance of these processes will depend on the starting concentrations in the raw intake or prefiltered water and the degree of concentration due to treatment. There is presently no information on algal toxins in these discharges. HABs on the U.S. west coast exhibit significant generalities across geographical and temporal scales (e.g. many of the same species occur throughout the region), but the details of bloom dynamics differ with geographic location, depth and season (and perhaps on interannual and decadal scales). The high degree of variability associated with these events makes constant monitoring of HABs in intake water for desalination a vital issue. Regional HAB programs and regulatory agencies along the U.S. west coast presently provide useful information for some known potential problems (e.g. ASP and PSP toxins) for end users that need information on coastal water quality. Awareness (and augmentation) of this information could improve planning and safe operation of desalination facilities. Monitoring of newly emerging HAB concerns (e.g. Cochlodinium spp.), or HABs and toxins that are presently poorly characterized (e.g. NSP, DSP, and yessotoxin poisoning) should also be implemented in the future to allow evaluation of their potential impacts on desalination processes. New technologies for toxin detection and quantification, and in situ monitoring of biological and chemical parameters are rapidly improving our ability to monitor coastal ecosystems and identify potentially problematic situations involving HABs. Advances in in situ observing technologies (sensor networks, autonomous sensor-equipped vehicles) provide the capability for obtaining unprecedented resolution in the spatial and temporal distributions of chemical and physical parameters, and some biologically important features (Sukhatme et al., 2007). New approaches and instruments for toxin detection help identify contaminated seafood products, and constitute sentinels for the threats of HABs to marine animal populations. Future uses of coastal waters for desalination will also benefit from, and contribute to, these activities.

408

water research 44 (2010) 385–416

Acknowlegements The authors are grateful to Mr. Mark Donovan at Separation Processes, Inc. for providing the data for Fig. 9. The preparation of this manuscript was supported in part by funding from a contract between the West Basin Municipal Water District, Department of Water Resources and the University of Southern California, National Oceanic and Atmospheric Administration grants NA05NOS4781228 and NA07OAR4170008, Sea Grant NA07OAR4170008, National Science Foundation grants CCR0120778 (Center for Embedded Networked Sensing; CENS), DDDAS-0540420, MCB-0703159, and a NASA Earth and Space Science Fellowship Grant NNX06AF88H. M.-E`. Garneau was supported by a fellowship from the Fonds que´be´cois de recherche sur la nature et les technologies (FQRNT).

references

Abdul Azis, P.K., Al-Tisan, I., Al-Daili, M., Green, T.N., Dalvi, A.G.I., Javeed, M.A., 2000. Effects of environment on source water for desalination plants on the eastern coast of Saudi Arabia. Desalination 132 (1–3), 29–40. Adams, N.A., Lesoing, M., Trainer, L., 2000. Environmental influences on domoic acid accumulation in razor clams on the Washington coast. Journal of Shellfish Research 19 (2), 1007–1015. Adin, A., Klein-Banay, C., 1986. Pretreatment of seawater by flocculation and settling for particulates removal. Desalination 58 (3), 227–242. Ahn, S., Kulis, D.M., Erdner, D.L., Anderson, D.M., Walt, D.R., 2006. Fiber-optic microarray for simultaneous detection of multiple harmful algal bloom species. Applied and Environmental Microbiology 72 (9), 5742–5749. Al Arrayedhy, M., 1987. Pre- and post-treatment at the RO plant at RA’s Abu Jarjur, Bahrain. Desalination 63 (C), 81–94. Al-Sahlawi, M.A., 1999. Seawater desalination in Saudi Arabia: economic review and demand projections. Desalination 123 (2–3), 143–147. Alfonso, A., de la Rosa, L., Vieytes, M.R., Yasumoto, T., Botana, L.M., 2003. Yessotoxin, a novel phycotoxin, activates phosphodiesterase activity: effect of yessotoxin on cAMP levels in human lymphocytes. Biochemical Pharmacology 65 (2), 193–208. Alldredge, A.L., Passow, U., Logan, B.E., 1993. The abundance and significance of a class of large, transparent organic particles in the ocean. Deep-Sea Research Part I Oceanographic Research Papers 40 (6), 1131–1140. Allen, W.E., 1922. Observations on surface distribution of marine diatoms between San Diego and Seattle. Ecology 3 (2), 140–145. Allen, W.E., 1924. Surface catches of marine diatoms and dinoflagellates made by U.S.S. Pioneer between San Diego and Seattle in 1923. In: University of California Publications in Zoology, vol. 26, pp. 243–248. Allen, W.E., 1928. Review of five years of studies on phytoplankton at Southern California piers, 1920–1924, inclusive. Bulletin of the Scripps Institution of Oceanography Technical Services 1, 357–401. Allen, W.E., 1936. Occurrence of marine plankton diatoms in a tenyear series of daily catches in Southern California. American Journal of Botany 23 (2), 60–63. Allen, W.E., 1940. Summary of results of twenty years of researches on marine phytoplankton. Proceedings of the 6th Pacific Scientific Congress 3, 576–583.

Allen, W.E., 1941. Twenty years’ statistical studies of marine plankton dinoflagellates of Southern California. American Midland Naturalist 26, 603–635. Allen, J.I., Smyth, T.J., Siddom, J.R., Holt, M., 2008. How well can we forecast high biomass algal biomass events in a eutrophic coastal sea? Harmful Algae 8 (1), 70–76. Amy, G., Kim, T.U., Yoon, J., Bellona, C., Drewes, J., Pellegrino, J., Heberer, T., 2005. Removal of micropollutants by NF/RO membranes. Water Science and Technology: Water Supply 5 (5), 25–33. Anderson, C.R., Brzezinski, M.A., Washburn, L., Kudela, R., 2006. Circulation and environmental conditions during a toxigenic Pseudo-nitzschia australis bloom in the Santa Barbara Channel, California. Marine Ecology Progress Series 327, 119–133. Anderson, D.M., Kulis, D.M., Sullivan, J.J., Hall, S., Lee, C., 1990. Dynamics and physiology of saxitoxin production by the dinoflagellates Alexandrium spp. Marine Biology 104 (3), 511–524. Anderson, D.M., Glibert, P.M., Burkholder, J.M., 2002. Harmful algal blooms and eutrophication: nutrient sources, composition, and consequences. Estuaries 25 (4B), 704–726. Anderson, D.M., Kulis, D.M., Keafer, B.A., Gribble, K.E., Marin, R., Scholin, C.A., 2005. Identification and enumeration of Alexandrium spp. from the Gulf of Maine using molecular probes. Deep-Sea Research Part II 52 (19–21), 2467–2490. Anderson, D.M., Burkholder, J.M., Cochlan, W.P., Glibert, P.M., Gobler, C.J., Heil, C.A., Kudela, R.M., Parsons, M.L., Rensel, J.E.J., Townsend, D.W., Trainer, V.L., Vargo, G.A., 2008. Harmful algal blooms and eutrophication: examining linkages from selected coastal regions of the United States. Harmful Algae 8 (1), 39–53. Armonies, W., 1989. Occurrence of meiofauna in Phaeocystis seafoam. Marine Ecology Progress Series 53, 305–309. Armstrong, M., Kudela, R., 2006. Evaluation of California isolates of Lingulodinium polyedrum for the production of yessotoxin. African Journal of Marine Science 28 (2), 399–401. Aune, T., Sørby, R., Yasumoto, T., Ramstad, H., Landsverk, T., 2002. Comparison of oral and intraperitoneal toxicity of yessotoxin towards mice. Toxicon 40 (1), 77–82. Baden, D.G., 1983. Marine food-borne dinoflagellate toxins. International Review of Cytology 83, 99–150. Baden, D.G., 1989. Brevetoxins: unique polyether dinoflagellate toxins. The FASEB Journal 3 (7), 1807–1817. Baden, D.G., Bourdelais, A.J., Jacocks, H., Michelliza, S., Naar, J., 2005. Natural and derivative brevetoxins: historical background, multiplicity and effects. Environmental Health Perspectives 113 (5), 621–625. Barber, R.T., Chavez, F.P., 1983. Biological consequences of El Nin˜o. Science 222 (4629), 1203–1210. Bargu, S., Powell, C.L., Coale, S.L., Busman, M., Doucette, G.J., Silver, M.W., 2002. Krill: a potential vector for domoic acid in marine food webs. Marine Ecology Progres Series 237, 209–216. Bargu, S., Powell, C.L., Wang, Z., Doucette, G.J., Silver, M.W., 2008. Note on the occurrence of Pseudo-nitzschia australis and domoic acid in squid from Monterey Bay, CA (USA). Harmful Algae 7 (1), 45–51. Bates, H.A., Kostriken, R., Rapoport, A., 1978. The occurrence of saxitoxin and other toxins in various dinoflagellates. Toxicon 16 (6), 595–601. Bates, S.S., Trainer, V.L., 2006. The ecology of harmful diatoms. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 81–93. Bates, S.S., Bird, C.J., Defreitas, A.S.W., Foxall, R., Gilgan, M., Hanic, L.A., Johnson, G.R., McCulloch, A.W., Dodense, P., Pocklington, R., Quilliam, M.A., Sim, P.G., Smith, J.C., Rao, D.V. S., Todd, C.D., Walter, J.A., Wrigh, J.L.C., 1989. Pennate diatom Nitzschia pungens as the primary source of domoic acid, a toxin in shellfish from eastern Prince Edwards Island, Canada. Canadian Journal of Fisheries and Aquatic Sciences 46 (7), 1203–1215.

water research 44 (2010) 385–416

Bates, S.S., DeFreitas, A.S.W., Milley, J.E., Pocklington, R., Quilliam, M.A., Smith, J.C., Worms, J., 1991. Controls on domoic acid production by the diatom Nitzschia pungens f. multiseries in cultures: nutrients and irradiance. Canadian Journal of Fisheries and Aquatic Sciences 48 (7), 1136–1144. Baugh, K.A., Bush, J.M., Bill, B.D., Lefebvre, K.A., Trainer, V.L., 2006. Estimates of specific toxicity in several Pseudo-nitzschia species from the Washington coast, based on culture and field studies. African Journal of Marine Science 28 (2), 403–407. Bialojan, C., Takai, A., 1988. Inhibitory effect of a marine-sponge toxin, okadaic acid, on protein phosphatases. Specificity and kinetics. Biochemical Journal 256 (1), 283–290. Bigelow, H.B., Leslie, M., 1930. Reconnaissance of the waters and plankton of Monterey Bay July 1928. Bulletin of the Museum of Comparative Zoology 70 (5), 430–581. Blasco, D., 1978. Observations on the diel migration of marine dinoflagellates off the Baja Californian coast. Marine Biology 46 (1), 41–47. Bolin, R.L., Abbott, D.P., 1960. Studies on the marine climate and phytoplankton of the central coastal area of California, 1954–1960. California Cooperative Oceanic Fisheries Investigations 9, 23–45. Boni, L., Ceredi, A., Guerrini, F., Milandri, A., Pistocchi, R., Poletti, R., Pompei, M., 2002. Toxic Gonyaulax Grindley Reinecke in the North-Western Adriatic Sea (Italy). In: Tenth International Conference on Harmful Algal Blooms, St. Petersburg, FL (abstract). Bonnelye, V., Snaz, M.A., Comte, C., Colas, F., Plasse, L., Gueguen, F., 2004a. Ultrafiltration versus conventional pre-treatment upstream RO: advantages and limits, Seoul, Korea. Bonnelye, V., Sanz, M.A., Durand, J.P., Plasse, L., Gueguen, F., Mazounie, P., 2004b. Reverse osmosis on open intake seawater: pre-treatment strategy. Desalination 167 (1–3), 191–200. Bourdelais, A.J., Tomas, C.R., Naar, J., Kubanek, J., Baden, D.G., 2002. New fish-killing alga in coastal Delaware produces neurotoxins. Environmental Health Perspectives 110 (5), 465–470. Bowers, H.A., Tengs, T., Glasgow Jr., H.B., Burkholder, J.M., Rublee, P.A., Oldach, D.W., 2000. Development of real-time PCR assays for rapid detection of Pfiesteria piscicida and related dinoflagellates. Applied and Environmental Microbiology 66 (11), 4641–4648. Bowers, H.A., Tomas, C., Tengs, T., Kempton, J.W., Lewitus, A.J., Oldach, D.W., 2006. Raphidophyceae (Chadefaud ex Silva) systematics and rapid identification: sequence analyses and real-time PCR assays. Journal of Phycology 42 (6), 1333–1348. Bricelj, V.M., Lonsdale, D.J., 1997. Aureococcus anophagefferens: causes and ecological consequences of brown tides in U.S. mid-Atlantic coastal waters. Limnology and Oceanography 42 (5), 1023–1038. Buck, K.R., Uttal-Cooke, L., Pilskaln, C.H., Roelke, D.L., Villac, M.C., Fryxell, G.A., Cifuentes, L., Chavez, F.P., 1992. Autecology of the diatom Pseudo-nitzschia australis, a domoic acid producer, from Monterey Bay, California. Marine Ecology Progress Series 84 (3), 293–302. Burbano, A.A., Adham, S.S., Pearce, W.R., 2007. The state of fullscale RO/NF desalination - Results from a worldwide survey. Journal of American Water Works Association 99 (4), 116–127. Burkholder, J.M., Azanza, R.V., Sako, Y., 2006. The ecology of harmful dinoflagellates. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 53–66. Burkholder, J.M., Hallegraeff, G.M., Melia, G., Cohen, A., Bowers, H.A., Oldach, D.W., Parrow, M.W., Sullivan, M.J., Zimba, P.V., Allen, E.H., Kinder, C.A., Mallin, M.A., 2007. Phytoplankton and bacterial assemblages in ballast water of U.S. military ships as a function of port of origin, voyage time, and ocean exchange practices. Harmful Algae 6 (4), 486–518.

409

Burkholder, J.M., Glibert, P.M., Skelton, H.M., 2008. Mixotrophy, a major mode of nutrition for harmful algal species in eutrophic waters. Harmful Algae 8 (1), 77–93. Buskey, E.J., 2008. How does eutrophication affect the role of grazers in harmful algal bloom dynamics? Harmful Algae 8 (1), 152–157. Busse, L.B., Venrick, E.L., Antrobus, R., Miller, P.E., Vigilant, V., Silver, M.W., Mengelt, C., Mydlarz, L., Prezelin, B.B., 2006. Domoic acid in phytoplankton and fish in San Diego, CA, USA. Harmful Algae 5 (1), 91–101. Caron, D.A., Dennett, M.R., Moran, D.M., Schaffner, R.A., Lonsdale, D. J., Gobler, C.J., Nuzzi, R., McLean, T.I., 2003. Development and application of a monoclonal-antibody technique for counting Aureococcus anophagefferens, an alga causing recurrent brown tides in the Mid-Atlantic United States. Applied and Environmental Microbiology 69 (9), 5492–5502. Carpenter, E.J., Janson, S., Boje, R., Pollehne, F., Chang, J., 1995. The dinoflagellate Dinophysis norvegica: biological and ecological observations in the Baltic Sea. European Journal of Phycology 109. Casper, E.T., Patterson, S.S., Bhanushali, P., Farmer, A., Smith, M., Fries, D.P., Paul, J.H., 2007. A handheld NASBA analyzer for the field detection and quantification of Karenia brevis. Harmful Algae 6 (1), 112–118. Catterall, W.A., 1992. Cellular and molecular biology of voltagegated sodium channels. Physiological Reviews 72 (Suppl. 4.), S15–S48. CDPH, 2007. Preharvest Shellfish Protection and Marine Biotoxin Monitoring Program. California Department of Public Health. Cembella, A.D., 1989. Occurrence of okadaic acid, a major diarrheic shellfish toxin, in natural populations of Dinophysis spp. from the eastern coast of North America. Journal of Applied Phycology 1 (4), 307–310. Cembella, A.D., 2003. Chemical ecology of eukaryotic microalgae in marine ecosystems. Phycologica 42 (4), 420–447. Cestele, S., Catterall, W.A., 2000. Molecular mechanisms of neurotoxin action on voltage-gated sodium channels. Biochimie 82 (9–10), 883–892. Chavez, F.P., Ryan, J., Lluch-Cota, S.E., Niquen, M., 2003. From anchovies to sardines and back: multidecadal change in the Pacific ocean. Science 299 (5604), 217–221. Ciminiello, P., Fattorusso, E., Forino, M., Poletti, R., Viviani, R., 2000. Structure determination of carboxyhomoyessotoxin, a new yessotoxin analogue isolated from Adriatic mussels. Chemistry Research in Toxicology 13 (8), 770–774. Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R., 2001. 42,43,44,45,46,47,55-Heptanor-41-oxohomoyessotoxin, a new biotoxin from mussels of the Northern Adriatic Sea. Chemical Research in Toxicology 14 (5), 596–599. Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R., Viviani, R., 1998. Isolation of adriatoxin a new analogue of yessotoxin from mussels of the Adriatic Sea. Tetrahedron Letters 39 (48), 8897–8900. Clayden, J., Read, B., Hebditch, K., 2005. Chemistry of domoic acid, isodomoic acids, and their analogues. Tetrahedron 61 (24), 5713–5724. Cochlan, W.P., Herndon, J., Kudela, R.M., 2008. Inorganic and organic nitrogen uptake by the toxigenic diatom Pseudo-nitzschia australis (Bacillariophyceae). Harmful Algae 8 (1), 111–118. Cooley, H., Gleick, P.H., Wolff, G., 2006. Desalination, with a grain of salt: a California perspective. In: Hart, I. (Ed.). Pacific Institute for Studies in Development, Environment, and Security, Hayward, CA, p. 88. Coyne, K.J., Hutchins, D.A., Hare, C.E., Cary, S.C., 2001. Assessing temporal and spatial variability in Pfiesteria piscicida distributions using molecular probing techniques. Aquatic Microbial Ecology 24 (3), 275–285.

410

water research 44 (2010) 385–416

Coyne, K.J., Handy, S.M., Demir, E., Whereat, E.B., Hutchins, D.A., Portune, K.J., Doblin, M.A., Cary, S.C., 2005. Improved quantitative real-time PCR assays for enumeration of harmful algal species in field samples using an exogenous DNA reference standard. Limnology and Oceanography Methods 3, 381–391. Cullen, J.J., Horrigan, S.G., 1981. Effects of nitrate on the diurnal vertical migration, carbon to nitrogen ratio, and the photosynthetic capacity of the dinoflagellate Gymnodinium splendens. Marine Biology 62 (2–3), 81–89. Curtiss, C.C., Langlois, G.W., Busse, L.B., Mazzillo, F., Silve, M.W., 2008. The emergence of Cochlodinium along the California Coast (USA). Harmful Algae 7 (3), 337–346. Daiguji, M., Satake, M., Ramstad, H., Aune, T., Naoki, H., Yasumoto, T., 1998. Structure and fluorometric HPLC determination of 1-desulfoyessotoxin, a new yessotoxin analog isolated from mussels from Norway. Natural Toxins 6 (6), 235–239. Dekshenieks, M.M., Donaghay, P.L., Sullivan, J.M., Rines, J.E.B., Osborn, T.R., Twardowski, M.S., 2001. Temporal and spatial occurrence of thin phytoplankton layers in relation to physical processes. Marine Ecology Progress Series 223, 61–71. Demir, E., Coyne, K.J., Doblin, M.A., Handy, S.M., Hutchins, D.A., 2008. Assessment of microzooplankton grazing on Heterosigma akashiwo using a species-specific approach combining quantitative real-time PCR (QPCR) and dilution methods. Microbial Ecology 55 (4), 583–594. Determan, T., 2003. Paralytic shellfish poisoning (PSP) patterns in Puget Sound shellfish in 2001-A report for the Puget Sound Ambient Monitoring Program Office of Food Safety and Shellfish Programs. Washington State Department of Health, Olympia. Draisci, R., Ferretti, E., Palleschi, L., Marchiafava, C., Poletti, R., Milandri, A., Ceredi, A., Pompei, M., 1999a. High levels of yessotoxin in mussels and presence of yessotoxin and homoyessotoxin in dinoflagellates of the Adriatic Sea. Toxicon 37, 1187–1193. Draisci, R., Palleschi, L., Giannetti, L., Lucentini, L., James, K.J., Bishop, A.G., Satake, M., Yasumoto, T., 1999b. New approach to the direct detection of known and new diarrhetic shellfish toxins in mussels and phytoplankton by liquid chromatography-mass spectrometry. Journal of Chromatrography 847 (1–2), 213–221. Eiki, K., Satake, M., Koike, K., Ogata, T., Mitsuya, T., Oshima, Y., 2005. Confirmation of yessotoxin production by the dinoflagellate Protoceratium reticulatum in Mutsu Bay. Fisheries Science 71 (3), 633–638. Eppley, R.W., Holm-Harisen, O., Strickland, J.D.H., 1968. Some observations on the vertical migration of dinoflagellates. Journal of Phycology 4 (4), 333–340. European Commission, 2002. Directive Commission Decision, No. C2002 1001. Official Journal of the European Communities March 15, 2002, 1.75/63, 16.3. Fehling, J., Davidson, K., Bolch, C.J., Bates, S.S., 2004. Growth and domoic acid production by Pseudo-nitzschia seriata (Bacillariophyceae) under phosphate and silicate limitation. Journal of Phycology 40 (4), 674–683. Flynn, K.J., 2002. Toxin production in migrating dinoflagellates: a modelling study of PSP producing Alexandrium. Harmful Algae 1 (2), 147–155. Flynn, K.J., 2008. Attack is not the best form of defense: lessons from harmful algal bloom dynamics. Harmful Algae 8 (1), 129–139. Flynn, K., Franco, J.M., Fernandez, P., Reguera, B., Zapata, M., Wood, G., Flynn, K.J., 1994. Changes in toxin content, biomass and pigments of the dinoflagellate Alexandrium minutum during nitrogen refeeding and growth into nitrogen or phosphorus stress. Marine Ecology Progres Series 111, 99–109. Franchini, A., Marchesini, E., Poletti, R., Ottaviani, E., 2004. Lethal and sub-lethal yessotoxin dose-induced morpho-functional

alterations in intraperitoneal injected Swiss CD1 mice. Toxicon 44 (1), 83–90. Fritz, L., Quilliam, M.A., Wright, J.L.C., Beale, A.M., Work, T.M., 1992. An outbreak of domoic acid poisoning attributed to the pennate diatom Pseudo-nitzschia australis. Journal of Phycology 28 (4), 439–442. Fryxell, G.A., Villac, M.C., Shapiro, L.P., 1997. The occurrence of the toxic diatom genus Pseudo-nitzschia (Bacillariophyceae) on the West Coast of the USA, 1920–1996: a review. Phycologia 36 (6), 419–437. Gaid, K., Treal, Y., 2007. Le dessalement des eaux par osmose inverse: l’expE`rience de VE`olia Water. Desalination 203 (1–3), 1–14. Gaines, G., Taylor, F.J.R., 1985. An exploratory analysis of PSP patterns in British Columbia: 1942–1984. In: Anderson, D.M., White, A.W., Baden, D.G. (Eds.), Toxic dinoflagellates. Elsevier Science, New York, pp. 439–444. Galluzzi, L., Penna, A., Bertozzini, E., Vila, M., Garces, E., Magnani, M., 2004. Development of a real-time PCR assay for rapid detection and quantification of Alexandrium minutum (a dinoflagellate). Applied and Environmental Microbiology 70 (2), 1199–1206. Garrison, D.L., Conrad, S.M., Eilers, P.P., Waldron, E.M., 1992. Confirmation of domoic acid production by Pseudo-nitzschia australis (Bacillariophyceae) cultures. Journal of Phycology 28 (5), 604–607. Garthwaite, I., Ross, K.M., Miles, C.O., Hansen, R.P., Foster, D., Wilkins, A.L., Towers, N.R., 1998. Polyclonal antibodies to domoic acid, and their use in immunoassays for domoic acid in sea water and shellfish. Natural Toxins 6 (3–4), 93–104. Garthwaite, I., Ross, K.M., Miles, C.O., Briggs, L.R., Towers, N.R., Borrell, T., Busby, P., 2001. Integrated Enzyme-Linked Immunosorbent Assay screening system for amnesic, neurotoxic, diarrhetic, and paralytic shellfish poisoning toxins found in New Zealand. Journal of AOAC International 84 (5), 1643–1648. Geraci, J.R., Anderson, D.M., Timperi, R.J., St. Aubin, D.J., Early, G.A., Prescott, J.H., Mayo, C., 1989. Humpback whales (Megaptera novaeangliae) fatally poisoned by dinoflagellate toxin. Canadian Journal of Fisheries and Aquatic Sciences 46 (11), 1895–1898. Gijsbertsen-Abrahamse, A.J., Schmidt, W., Chorus, I., Heijman, S.G.J., 2006. Removal of cyanotoxins by ultrafiltration and nanofiltration. Journal of Membrane Science 276 (1–2), 252–259. Glibert, P.M., Anderson, D.M., Gentien, P., Grane´li, E., Sellner, K.G., 2005a. The global, complex phenomena of harmful algal blooms. Oceanography 18 (2), 137–147. Glibert, P.M., Seitzinger, S., Heil, C.A., Burkholder, J.M., Parrow, M.W., Codispoti, L.A., Kelly, V., 2005b. The role of eutrophication in the global proliferation of harmful algal blooms. Oceanography 18 (2), 198–209. Glibert, P.M., Harrison, J., Heil, C., Seitzinger, S., 2006. Escalating worldwide use of urea – a global change contributing to coastal eutrophication. Biogeochemistry 77 (3), 441–463. Glibert, P.M., Kelly, V., Alexander, J., Cosdispoti, L.A., Boicourt, W.C., Trice, T.M., Michael, B., 2008. In situ nutrient monitoring: a tool for capturing nutrient variability and the antecedent conditions that support algal blooms. Harmful Algae 8 (1), 175–181. Grane´li, E., Flynn, K., 2006. Chemical and physical factors influencing toxin content. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 229–241. Greenfield, D.I., Marin, R., Jensen, S., Massion, E., Roman, B., Feldman, J., Scholin, C., 2006. Application of the Environmental Sample Processor (ESP) methodology for quantifying Pseudonitzschia australis using ribosomal RNA-targeted probes in sandwich and fluorescent in situ hybridization. Limnology and Oceanography Methods 4, 426–435.

water research 44 (2010) 385–416

Gregorio, D.E., Connell, L., 2000. Range of Heterosigma akashiwo (Raphidophyceae) expanded to include California, USA. In: Ninth International Conference on Harmful Algal Blooms, Tasmania, Australia (abstract). Gregorio, D.E., Pieper, R.E., 2000. Investigations of red tides along the southern California. Bulletin of Southern California Academy of Sciences 99 (8), 147–160. Gulland, F.M.D., Haulena, M., Fauquier, D., Langlois, G., Lander, M.E., Zabka, T., Duerr, R., 2002. Domoic acid toxicity in California sea lions (Zalophus californianus): clinical signs, treatment and survival. Veterinary Record 150 (15), 475–480. Hallegraeff, G., Hara, Y., 1995. Taxonomy of harmful marine raphidophytes. In: Hallegraef, G., Anderson, D.M., Cembella, A.D. (Eds.), Manual on Harmful Marine Microalgae. IOC UNCESCO, Paris, pp. 365–371. Hallegraeff, G.M., 1993. A review of harmful algal blooms and their apparent global increase. Phycologia 32 (2), 79–99. Hallegraeff, G.M., 2003. Harmful algal blooms: a global overview. In: Hallegraeff, G.M., Anderson, D.M., Cembella, A.D. (Eds.), Manual on Harmful Marine Microalgae. Monographs on Oceanographic Methodology. UNESCO Publishing, Paris pp. 25–49. Hallegraeff, G.M., Bolch, C.J., 1992. Transport of diatom and dinoflagellate resting spores in ships’ ballast water: implications for plankton biogeography and aquaculture. Journal of Plankton Research 14 (8), 1067–1084. Handy, S.M., Hutchins, D.A., Cary, S.C., Coyne, K.J., 2006. Simultaneous enumeration of multiple raphidophyte species by multiprobing and mutiplexing: capabilities and limitations. Limnology and Oceanography: Methods 4, 193–204. Haque, S.M., Onoue, Y., 2002a. Effects of salinity on growth and toxin production of a noxious phytoflagellate, Heterosigma akashiwo (Raphidophyceae). Botanica Marina 45 (4), 356–363. Haque, S.M., Onoue, Y., 2002b. Variation in toxin compositions of two harmful raphidophytes, Chattonella antiqua and Chattonella marina, at different salinities. Environmental Toxicology 17 (2), 113–118. Hard, J.J., Connell, L., Hershberger, W.K., Harrell, L.W., 2000. Genetic variation in mortality of chinook salmon during a bloom of the marine alga Heterosigma akashiwo. Journal of Fish Biology 56 (6), 1387–1397. Hasan Al-Sheikh, A.H., 1997. Seawater reverse osmosis pretreatment with an emphasis on the Jeddah Plant operation experience. Desalination 110 (1–2), 183–192. Haystead, T.A.J., Sim, A.T.R., Carling, D., Honnor, R.C., Tsukitani, Y., Cohen, P., Hardie, D.G., 1989. Effects of the tumour promoter okadaic acid on intracellular protein phosphorylation and metabolism. Nature 337, 78–81. Heisler, J., Glibert, P.M., Burkholder, J.M., Anderson, D.M., Cochlan, W., Dennison, W.C., Dortch, Q., Gobler, C.J., Heil, C.A., Humphries, E., Lewitus, A., Magnien, R., Marshall, H.G., Sellner, K., Stockwell, D.A., Stoecker, D.K., Suddleson, M., 2008. Eutrophication and harmful algal blooms: a scientific consensus. Harmful Algae 8 (1), 3–13. Herndon, J., Cochlan, W.P., Horner, R., 2003. Heterosigma akashiwo blooms in San Francisco Bay. Interagency Ecological Program for the San Francisco Estuary Newsletter 16, 46–48. Hershberger, P.K., Rensel, J.E., Postel, J.R., Taub, F.B., 1997. Heterosigma bloom and associated fish kill. Harmful Algae News 16 (1), 4. Holmes, R.W., Williams, P.M., Eppley, R.W., 1967. Red water in La Jolla, 1964–1966. Limnology and Oceanography 12 (3), 503–512. Hong, Y., Smith Jr., W.O., White, A.M., 1997. Studies on transparent exopolymer particles (TEP) produced in the Ross Sea (Antarctica) and by Phaeocystis antarctica (Prymnesiophyceae). Journal Of Phycology 33 (3), 368–376.

411

Hori, M., Matsuura, Y., Yoshimoto, R., Ozaki, H., Yasumoto, T., Karaki, H., 1999. Actin depolymerizing action by marine toxin, pectenotoxin-2. Folia Pharmacologica Japonica 114 (Suppl. 1), 225–229. Horner, R.A., Postel, J.R., 1993. Toxic diatoms in western Washington waters (U.S. west coast). Hydrobiologia 269/270 (1), 197–205. Horner, R.A., Garrison, D.L., Plumley, F.G., 1997. Harmful algal blooms and red tide problems on the U.S. west coast. Limnology and Oceanography 42 (2), 1076–1088. Horner, R.A., Hickey, B.M., Postel, J.R., 2000. Pseudo-nitzschia blooms and physical oceanography off Washington state. South African Journal of Marine Science 22 (1), 299–308. Howard, M.D.A., Cochlan, W.P., Ladizinsky, N., Kudela, R.M., 2007. Nitrogenous preference of toxigenic Pseudo-nitzschia australis (Bacillariophyceae) from field and laboratory experiments. Harmful Algae 6 (2), 206–217. Howard, M.D.A., Silvera, M., Kudela, R.M., 2008. Yessotoxin detected in mussel (Mytilus californicus) and phytoplankton samples from the U.S. west coast. Harmful Algae 7 (5), 646–652. Howard, M.D., Smith, G.J., Kudela, R.M., 2009. Phylogenetic relationships of yessotoxin-producing dinoflagellates, based on the large subunit and internal transcribed spacer ribosomal DNA domains. Applied and Environmental Microbiology 75 (1), 54–63. Howarth, R.W., 2008. Coastal nitrogen pollution: a review of sources and trends globally and regionally. Harmful Algae 8 (1), 14–20. IOC HAB Programme, 2008. IOC Harmful Algal Bloom Website. Intergovernmental Oceanic Commission of UNESCO. Iwataki, M., Kawami, H., Matsuoka, K., 2007. Cochlodinium fulvescens sp nov (Gymnodiniales, Dinophyceae), a new chainforming unarmored dinoflagellate from Asian coasts. Phycological Research 55 (3), 231–239. Iwataki, M., Kawami, H., Mizushima, K., Mikulski, C.M., Doucette, G.J., Relox, J.R., Anton, A., Fukuyo, Y., Matsuoka, K., 2008. Phylogenetic relationships in the harmful dinoflagellate Cochlodinium polykrikoides (Gymnodiniales, Dinophyceae) inferred from LSU rDNA sequences. Harmful Algae 7 (3), 271–277. Jeffery, B., Barlow, T., Moizer, K., Paul, S., Boyle, C., 2004. Amnesic shellfish poison. Food and Chemical Toxicology 42 (4), 545–557. Jessup, D.A., Miller, M.A., Ryan, J.P., Nevins, H.M., Kerkering, H.A., Mekebri, A., Crane, D.B., Johnson, T.A., Kudela, R.M., 2009. Mass stranding of marine birds caused by a surfactantproducing red tide. PLOS One 4, E4550. Jester, R., 2008. An investigation into the prevalence of Alexandrium derived toxins in marine food webs. Ph.D., University of California, Santa Cruz, Santa Cruz, pp. 116. Jester, R., Lefebvre, K.A., Langlois, G., Vigilant, V., Baugh, K.A., Silver, M.W., 2009a. A shift in the dominant toxin-producing algal species in central California alters phycotoxins in food webs. Harmful Algae 8 (2), 291–298. Jester, R.J., Baugh, K.A., Lefebvre, K.A., 2009b. Presence of Alexandrium catenella and paralytic shellfish toxins in finfish, shellfish and rock crabs in Monterey Bay, California, USA. Marine Biology 156 (3), 493–504. Johansson, N., Grane´li, E., Yasumoto, T., Carlsson, P., Legrand, C., 1996. Toxin production by Dinophysis acuminata and D. acuta cells grown under nutrient sufficient and deficient conditions. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. Intergovernmental Oceanographic Commission of UNESCO, Paris, pp. 277–279. John, E.H., Flynn, K.J., 2000. Growth dynamics and toxicity of Alexandrium fundyense (Dinophyceae): the effect of changing N:

412

water research 44 (2010) 385–416

P supply ratios on internal toxin and nutrient levels. European Journal of Phycology 35 (1), 11–23. Jones, B.H., Noble, M.A., Dickey, T.D., 2002. Hydrographic and particle distributions over the Palos Verdes continental shelf: spatial, seasonal and daily variability. Continental Shelf Research 22 (6–7), 945–965. Kirkpatrick, B., Fleming, L.E., Squicciarini, D., Backer, L.C., Clark, R., Abraham, W., Benson, J., Cheng, Y.S., Johnson, D., Pierce, R., Zaias, J., Bossart, G.D., Baden, D.G., 2004. Literature review of Florida red tide: implications for human health effects. Harmful Algae 3 (2), 99–115. Kudela, R.M., Cochlan, W.P., 2000. Nitrogen and carbon uptake kinetics and the influence of irradiance for a red tide bloom off southern California. Aquatic Microbial Ecology 21 (1), 31–47. Kudela, R.M., Roberts, A., Armstrong, M., 2004. Laboratory analyses of nutrient stress and toxin production in Pseudonitzschia spp. from Monterey Bay, California. In: Steidinger, K., Landsberg, J., Tomas, C., Vargo, G. (Eds.). Florida Environmental Research Institute and UNESCO, St. Petersberg, pp. 136–138. Kudela, R.M., Lane, J.Q., Cochlan, W.P., 2008a. The potential role of anthropogenically derived nitrogen in the growth of harmful algae in California, USA. Harmful Algae 8 (1), 103–110. Kudela, R.M., Ryan, J.P., Blakely, M.D., Lane, J.Q., Peterson, T.D., 2008b. Linking the physiology and ecology of Cochlodinium to better understand harmful algal bloom events: a comparative approach. Harmful Algae 7 (3), 278–292. Kusek, K.M., Vargo, G., Steidinger, K., 1999. Gymnodinium breve in the field in the lab and in the newspaper - scientific and journalistic analysis of Florida red tides. Contributions in Marine Science 34, 1–229. Ladizinsky, N., Smith, G.J., 2000. Accumulation of domoic acid by the coastal diatom Pseudo-nitzschia multiseries: a possible copper complexation strategy. Journal of Phycology 36, 41. Landry, M.R., 1989. Broad-scale distributional patterns of hydrographic variables on the Washington/Oregon shelf. In: Landry, M.R., Hickey, B.M. (Eds.), Coastal Oceanography of Washington and Oregon. Elsevier, New York, pp. 1–40. Lange, C.B., Reid, F.M.H., Vernet, M., 1994. Temporal distribution of the potentially toxic diatom Pseudo-nitzschia australis at a coastal site in Southern California. Marine Ecology Progress Series 104 (3), 309–312. Langlois, G., 2007. Marine Biotoxin Monitoring Program Annual Report. California Department of Public Health for California Department of Fish and Game, Sacramento. Lee, J., Walker, H.W., 2006. Effect of process variables and natural organic matter on removal of microcystin-LR by PAC-UF. Environmental Science and Technology 40 (23), 7336–7342. Lee, J.S., Yanagi, T., Kenma, R., Yasumoto, T., 1987. Fluorometric determination of diarrhetic shellfish toxins by highperformance liquid chromatography. Agricultural and Biological Chemistry 51 (3), 877–881. Lee, J.S., Igarashi, T., Fraga, S., Dah, E., Hovgaard, P., Yasumoto, T., 1989. Determination of diarrhetic toxins in various dinoflagellate species. Journal of Applied Phycology 1 (2), 147–152. Lefebvre, K.A., Bill, B.D., Erickson, A., Baugh, K.A., O’Rourke, L., Costa, P.R., Nance, S., Trainer, V.L., 2008. Characterization of intracellular and extracellular saxitoxin levels in both field and cultured Alexandruim spp. samples from Sequim Bay, Washington. Marine Drugs 6, 103–116. Levandowsky, M., Kaneta, P.J., 1987. Behaviour in dinoflagellates. In: Taylor, F.J.R. (Ed.), The Biology of Dinoflagellates. Blackwell, Oxford, pp. 360–397. Lin, Y.Y., Risk, M., Ray, S.M., Van Engen, D., Clardy, J., Golik, J., James, J.C., Nakanishi, K., 1981. Isolation and structure of brevetoxin B from the ‘red tide’ dinoflagellate Ptychodiscus

brevis (Gymnodinium breve). Journal of American Chemical Society 103 (22), 6773–6775. Llewellyn, L.E., 2006. Saxitoxin, a toxic marine natural product that targets a multitude of receptors. Natural Product Reports 23 (2), 200–222. Loeblich III, A.R., Fine, K.E., 1977. Marine chloromonads: More widely distributed in neritic environments than previously thought. Proceedings of the Biological Society of Washington 90 (2), 388–399. ˜ ., 1997. Studies Lundholm, N., Skov, J., Pocklington, R., Moestrup, A on the marine planktonic diatom Pseudo-nitzschia. 2. Autecology of P. pseudodelicatissima based on isolates from Danish coastal waters. Phycologia 36, 381–388. Malagoli, D., Marchesini, E., Ottaviani, E., 2006. Lysosomes as the target of yessotoxin invertebrate and vertebrate cell lines. Toxicology Letters 167 (1), 75–83. Maldonado, M.T., Hughes, M.P., Rue, E.L., Wells, M.L., 2002. The effect of Fe and Cu on growth and domoic acid production by Pseudo-nitzschia multiseries and Pseudo-nitzschia australis. Limnology and Oceanography 47 (2), 515–526. Marchetti, A., Trainer, V.L., Harrison, P.J., 2004. Environmental conditions and phytoplankton dynamics associated with Pseudo-nitzschia abundance and domoic acid in the Juan de Fuca eddy. Marine Ecology Progress Series 281, 1–12. Martin, J.L., Haya, K., Burridge, L.E., Wildish, D.J., 1990. Nitzschia pseudodelicatissima - a source of domoic acid in the Bay of Fundy, eastern Canada. Marine Ecology Progress Series 67 (2), 177–182. Matsuoka, K., Iwataki, M., Kawami, H., 2008. Morphology and taxonomy of chain-forming species of the genus Cochlodinium (Dinophyceae). Harmful Algae 7 (3), 261–270. McGillicuddy Jr., D.J., Signell, R.P., Stock, C.A., Keafer, B.A., Keller, M.D., Ehtland, R.D., Anderson, D.M., 2005. A mechanisms for offshore initiations of harmful algal blooms in the coastal Gulf of Maine. Journal of Plankton Research 25, 1131–1138. Mengelt, C., 2006. Ultraviolet Photoecology, Dark Survival, and Seasonal Abundance of Pseudo-nitzschia australis and P. multiseries in Coastal Waters of Central California. Dissertation, University of California, Santa Barbara. Mikulski, C.M., Morton, S.L., Doucette, G.J., 2005. Development and application of LSU rRNA probes for Karenia brevis in the Gulf of Mexico, USA. Harmful Algae 4 (1), 49–60. Mikulski, C.M., Park, Y.T., Jones, K.L., Lee, C.K., Lim, W.A., Lee, Y., Scholin, C.A., Doucette, G.J., 2008. Development and field application of rRNA-targeted probes for the detection of Cochlodinium polykrikoides Margalef in Korean coastal waters using whole cell and sandwich hybridization formats. Harmful Algae 7 (3), 347–359. Miles, C.O., Aasen, J., Dahl, E., Samdal, I., Aune, T., Briggs, L.R., Olseng, C.D., Edvardsen, B., Naustvoll, L.J., 2002. Yessotoxin in Norwegian Blue Mussels caused by Protoceratium reticulatum. In: Tenth International Conference on Harmful Algal Blooms, St. Petersburg, FL (abstract). Miles, C.O., Wilkins, A.L., Samdal, I.A., Sandvik, M., Petersen, D., Quilliam, M.A., Nausvoll, L.J., Rundberget, T., Torgersen, T., Hovgaard, P., Jensen, D.J., Cooney, J.M., 2004. A novel pectenotoxin, PTX-12, in Dinophysis spp. and shellfish from Norway. Chemical Research and Toxicology 17 (11), 1423–1433. Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J., Munday, R., Cooney, J.M., Beuzenberg, V., 2005a. Polyhydroxylated amide analogs of yessotoxin from Protoceratium reticulatum. Toxicon 45 (1), 61–71. Miles, C.O., Samdal, I.A., Aasen, J.A.G., Jensen, D.J., Quilliam, M.A., Petersen, D., Briggs, L.M., Wilkins, A.L., Rise, F., Cooney, J.M., MacKenzie, A.L., 2005b. Evidence for numerous analogs of

water research 44 (2010) 385–416

yessotoxin in Protoceratium reticulatum. Harmful Algae 4 (6), 1075–1091. Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J., Cooney, J.M., Beuzenberg, V., MacKenzie, L., 2006. Identification of 45-hydroxy-46,47-dinoryessotoxin, 44-oxo45,46,47-trinoryessotoxin, and 9-methyl-42,43,44,45,46,47,55hepta-nor-38-en-41-oxoyessotoxin, and partial characterization of some minor yessotoxins, from Protoceratium reticulatum. Toxicon 47 (2), 229–240. Miller, P.E., Scholin, C.A., 1998. Identification and enumeration of cultured and wild Pseudo-nitzschia (Bacillariophyceae) using species-specific LSU rRNA-targeted fluorescent probes and filter-based whole cell hybridization. Journal of Phycology 34 (2), 371–382. Mitrovic, S.M., Ferna´ndez Amandi, M., MacKenzie, L., Furey, A., James, K.J., 2004. Effects of selenium, iron and cobalt addition to growth and yessotoxin production of the toxic marine dinoflagellate Protoceratium reticulatum in culture. Journal of Experimental Marine Biology and Ecology 313 (2), 337–351. Mitrovic, S.M., Hamilton, B., MacKenzie, L., Furey, A., James, K.J., 2005. Persistence of yessotoxin under light and dark conditions. Marine Environmental Research 60 (3), 397–401. Montojo, U.M., Sakamoto, S., Cayme, M.F., Gatdula, N.C., Furio, E. F., Relox, J.R.J., Sato, S., Fukuyo, Y., Kodam, M., 2006. Remarkable difference in accumulation of paralytic shellfish poisoning toxins among bivalve species exposed to Pyrodinium bahamense var. compressum bloom in Masinloc bay. Philippines Toxicon 48 (1), 85–92. Montoya, N.G., Akselman, R., Franco, J.M., Carreto, J.I., 1996. Paralytic shellfish toxins and mackerel (Scomber japonicus) mortality in the Argentine sea. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. Intergovernmental Oceanographic Community, UNESCO, Paris, pp. 417–420. Moore, S.K., Mantua, N.J., Kellogg, J.P., Newton, J.A., 2008. Local and large-scale climate forcing of Puget Sound oceanographic properties on seasonal to interdecadal timescales. Limnology and Oceanography 53 (5), 1746–1758. Moore, S.K., Mantua, N.J., Hickey, B.M., Trainer, V.L., 2009. Recent trends in paralytic shellfish toxins in Puget Sound, relationships to climate, and capacity for prediction of toxic events. Harmful Algae 8 (3), 463–477. Moorthi, S.D., Countway, P.D., Stauffer, B.A., Caron, D.A., 2006. Use of quantitative real-time PCR to investigate the dynamics of the red tide dinoflagellate Lingulodinium polyedrum. Microbial Ecology 52 (1), 136–150. Murata, M., Shimatani, M., Sugitani, Y., Oshima, Y., Yasumoto, T., 1982. Isolation and structural elucidation of the causative toxin of the diarrhetic shellfish poisoning. Bulletin of the Japanese Society of Scientific Fisheries 48 (4), 549–552. Murata, M., Kumagai, M., Lee, J.S., Yasumoto, T., 1987. Isolation and structure of Yessotoxin, a novel polyether compound implicated in diarrhetic shellfish poisoning. Tetrahedron Letters 28 (47), 5869–5872. Naar, J., Bourdelais, A., Tomas, C., Kubanek, J., Whitney, P.L., Flewelling, L., Steidinger, K., Lancaster, J., Baden, D.G., 2002. A competitive ELISA to detect brevetoxins from Karenia brevis (formerly Gymnodinium breve) in seawater, shellfish, and mammalian body fluid. Environmental Health Perspectives 110 (2), 179–185. Netjstgaard, J.C., Tang, K.W., Steinke, M., Dutz, J., Koski, M., Antajan, E., Long, J.D., 2007. Zooplankton grazing on Phaeocystis: a quantitative review and future challenges. Biogeochemistry 83 (1–3), 147–172. Nishitani, G., Yamaguchi, M., Ishikawa, A., Yanagiya, S., Mitsuya, T., Imai, M., 2005. Relationships between occurrences

413

of toxic Dinophysis species (Dinophyceae) and small phytoplanktons in Japanese coastal waters. Harmful Algae 4 (4), 755–762. O’Halloran, C., Silver, M., Holman, T., Scholin, C., 2006. Heterosigma akashiwo in central California waters. Harmful Algae 5 (2), 124–132. Office of Shellfish and Water Protection, 2008. 2007 Annual Report: Commercial and Recreational Shellfish Areas in Washington State Olympia. Office of Shellfish and Water Protection. Ogino, H., Kumagai, M., Yasumoto, T., 1997. Toxicologic evaluation of Yessotoxin. Natural Toxins 5 (6), 255–259. Ono, K., Khan, S., Onoue, Y., 2000. Effects of temperature and light intensity on the growth and toxicity of Heterosigma akashiwo (Raphidophyceae). Aquaculture Research 31 (5), 427–433. Pan, Y., Subba Rao, D.V., Mann, K.H., 1996a. Changes in domoic acid production and cellular chemical composition of the toxigenic diatom Pseudo-nitzschia muliseries under phosphate limitation. Journal of Phycology 32 (3), 371–381. Pan, Y., Subba Rao, D.V., Mann, K.H., Li, K.W., Harrison, W.G., 1996b. Effects of silicate limitation on production of domoic acid, a neurotoxin, by the diatom Pseudo-nitzschia multiseries. I. Batch culture studies. Marine Ecology Progress Series 131 (1–3), 225–233. Pan, Y., Bates, S.S., Cembella, A.D., 1998. Environmental stress and domoic acid production by Pseudo-nitzschia: a physiological perspective. Natural Toxins 6 (3–4), 127–135. Pankratz, T., 2008. Red tides close desal plants. Water Desalination Report 44, 1. Pankratz, T., 2009. Speaking of red tides. Water Desalination Report 45, 2. Park, M.G., Kim, S., Kim, H.S., Myung, G., Kang, Y.G., Yih, W., 2006. First successful culture of the marine dinoflagellate Dinophysis. Aquatic Microbial Ecology 45 (2), 101–106. Paz, B., Riobo, P., Fernandez, A.L., Fraga, S., Franco, J.M., 2004. Production and release of yessotoxins by the dinoflagellates Protoceratium reticulatum and Lingulodinium polyedrum in culture. Toxicon 44 (3), 251–258. Paz, B., Riobo, P., Souto, M.L., Gil, L.V., Norte, M., Fernandez, J.J., Franco, J.M., 2006. Detection and identification of glycoyessotoxin A in a culture of the dinoflagellate Protoceratium reticulatum. Toxicon 48 (6), 611–619. Paz, B., Daranas, A.H., Cruz, P.G., Franco, J.M., Pizarro, G., Souto, M.L., Norte, M., Ferna´ndez, J.J., 2007. Characterization of okadaic acid related toxins by liquid chromatography coupled with mass spectrometry. Toxicon 50 (2), 225–235. Paz, B., Daranas, A.H., Norte, M., Riobo, P., Franco, J.M., Ferna´ndez, J.J., 2008. Yessotoxins, a group of marine polyether toxins: an overview. Marine Drugs 6 (6), 73–102. Peleka, E.N., Matis, K.A., 2008. Application of flotation as a pretreatment process during desalination. Desalination 222 (1–3), 1–8. Perl, T.M., Bedard, L., Kosatsky, T., Hockin, J.C., Todd, E.C.D., McNutt, L.A., Remis, R.S., 1990. Amnesic shellfish poisoning: a new clinical syndrome due to domoic acid. Canadian Diseases Weekly Report 16 (Suppl 1E), 7–8. Petry, M., Sanz, M.A., Langlais, C., Bonnelye, V., Durand, J.P., Guevara, D., Nardes, W.M., Saemi, C.H., 2007. The El Coloso (Chile) reverse osmosis plant. Desalination 203 (1–3), 141–152. Pocklington, R., Milley, J.E., Bates, S.S., Bird, C.J., deFreitas, A.S.W., Quilliam, M.A., 1990. Trace determination of domoic acid and phytoplankton by high-performance liquid chromatography of the fluorenylmethoxycarbonyl (FMOC) derivative. International Journal of Environmental Analytical Chemistry 38, 351–358. Prassopoulou, E., Katikou, P., Georgantelis, D., Kyritsakis, A., 2009. Detection of okadaic acid and related esters in mussels

414

water research 44 (2010) 385–416

during diarrhetic shellfish poisoning (DSP) episodes in Greece using the mouse bioassay, the PP2A inhibition assay and HPLC with fluorimetric detection. Toxicon 53 (2), 214–227. Price, D.W., Kizer, K.W., Hansgen, K.H., 1991. California’s paralytic shellfish poisoning prevention program, 1927–89. Journal of Shellfish Research 10 (11), 119–145. Purkerson, S.L., Baden, D.G., Fieber, L.A., 1999. Brevetoxin modulates neuronal sodium channels in two cell lines derived from rat brain. Neurotoxicology 20 (6), 909–920. Quilliam, M., 2003. Chemical methods for domoic acid, the amnesic shellfish poisoning (ASP) toxin. In: Hallegraeff, G., Anderson, D., Cembella, A. (Eds.), Manual on Harmful Marine Microalgae, Monographs on Oceanographic Methodology. Intergovernmental Oceanographic Commission (UNESCO), Paris, pp. 247–266. Quilliam, M.A., Sim, P.G., McCulloch, A.W., McInnes, A.G., 1989. High-performance liquid chromatography of domoic acid, a marine neurotoxin, with application to shellfish and plankton. International Journal of Environmental Analytical Chemistry 36 (3), 139–154. Ramsdell, J.S., Anderson, D.M., Glibert, P.M., 2005. HARRNESS, Harmful Algal Research and Response: a National Environmental Science Strategy 2005–2015. Ecological Society of America, Washington, DC, 96 pp. Reid, J.L., Roden, G.I., Wyllie, J.G., 1958. Studies of the California Current system. California Cooperative Oceanic Fisheries Investigations Progress Report, 1 July 1956–1 January 1958, pp. 27–56. Reid, F.M.H., Fuglister, E., Jordan, J.B., 1970. The ecology of the plankton off La Jolla, California, in the period April through September, 1967. Part 5: phytoplankton taxonomy and standing crop. Bulletin of the Scripps Institute of Oceanography Techical Series 17, 51–66. Reid, F.M.H., Lange, C.B., White, M.M., 1985. Microplankton species assemblages at Scripps Pier from March to November 1983 during the 1982–1983 El Nin˜o Event. Botanica Marina 28 (10), 443–452. Reiss, C.R., Robert, C., Owen, C., Taylor, J.S., 2006. Control of MIB, geosmin and TON by membrane systems. Journal of Water Supply: Research and Technology – AQUA 55 (2), 95–108. Rhodes, L., White, D., Syhre, M., Atkinson, M., 1996. Pseudonitzschia species isolated from New Zealand coastal waters: domoic acid production in vitro and links with shellfish toxicity. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. IOC of UNESCO, Paris pp. 155–158. Rhodes, L., Scholin, C., Garthwaite, I., Haywood, A., Thomas, A., 1998. Domoic acid producing Pseudo-nitzschia species deduced with whole cell DNA probe-based and immunochemical assays. In: Reguera, B., Blanco, J., Ferna´ndez, M.L., Wyatt, T. (Eds.), Harmful Algae. Xunta de Galicia and Intergovernmental Oceanographie Commission of UNESCO, pp. 274–277. Rhodes, L., McNabb, P., de Salas, M., Briggs, L., Beuzenberg, V., Gladstone, M., 2006. Yessotoxin production by Gonyaulax spinifera. Harmful Algae 5 (2), 148–155. Rines, J.E.B., Donaghay, P.L., Dekshenieks, M.M., Sullivan, J.M., Twardowski, M.S., 2002. Thin layers and camouflage: hidden Pseudo-nitzschia spp. (Bacillariophyceae) populations in a fjord in the San Juan Island, Washington, USA. Marine Ecology Progress Series 225, 123–137. Riobo, P., Paz, B., Fernandez, M., Fraga, S., Franco, J.M., 2002. Lipophylic toxins of different strains of Ostreopsidaceae and Gonyaulacaceae. In: Tenth International Conference on Harmful Algal Blooms, St. Petersburg, FL (abstract). Rogers, R.S., Rapoport, H., 1980. The pKas of saxitoxin. Journal of American Chemical Society 102 (24), 7335–7339.

Ronzitti, G., Callegari, F., Malaguti, C., Rossini, G.P., 2004. Selective disruption of the E-cadherin–catenin system by an algal toxin. British Journal of Cancer 90 (5), 1100–1107. Rue, E.L., Bruland, K.W., 2001. Domoic acid binds iron and copper: a possible role for the toxin produced by the marine diatom Pseudo-nitzschia. Marine Chemistry 76 (1–2), 127–134. Samdal, I.A., Naustvoll, L.J., Olseng, C.D., Briggs, L.R., Miles, C.O., 2004. Use of ELISA to identify Protoceratium reticulatum as a source of yessotoxin in Norway. Toxicon 44 (1), 75–82. Samdal, I.A., Aesen, J.A.G., Briggs, L.R., Dahl, E., Miles, C.O., 2005. Comparison of ELISA and LC–MS analyses for yessotoxins in blue mussels (Mytilus edulis). Toxicon 46, 7–15. Satake, M., Mackenzie, L., Yasumoto, T., 1997. Identification of Protoceratium reticulatum as the biogenetic origin of yessotoxin. Natural Toxins 5 (4), 164–167. Satake, M., Ichimura, T., Sekiguchi, K., Yoshimatsu, S., Oshima, Y., 1999. Confirmation of yessotoxin and 45, 46, 47trinoryessotoxin production by Protoceratium reticulatum collected in Japan. Natural Toxins 7 (4), 147–150. Schantz, E.J., Mold, J.D., Stanger, D.W., Shavel, J., Riel, F.J., Bowden, J.P., Lynch, J.M., Wyler, R.S., Riegel, B., Sommer, H., 1957. Paralytic shellfish poison. VI. A Procedure for the isolation and purification of the poison from toxic clam and mussel tissues. Journal of American Chemical Society 79 (19), 5230–5235. Schnetzer, A., Miller, P.E., Schaffner, R.A., Stauffer, B.A., Jones, B.H., Weisberg, S.B., DiGiacomo, P.M., Berelson, W.M., Caron, D.A., 2007. Blooms of Pseudo-nitzschia and domoic acid in the San Pedro Channel and Los Angeles harbor areas of the Southern California Bight, 2003-2004. Harmful Algae 6 (3), 372–387. Schofield, O., Kohut, J., Aragon, D., Creed, L., Graver, J., Haldeman, C., Kerfoot, J., Roarty, H., Jones, C., Webb, D., Glenn, S., 2007. Slocum gliders: robust and ready. Journal of Field Robotics 24 (6), 473–485. Scholin, C.A., Herzog, M., Sogin, M., Anderson, D.M., 1994. Identification of group-and strain-specific genetic markers for globally distributed Alexandrium (Dinophyceae). Part II. Sequence analysis of a fragment of the LSU rRNA gene. Journal of Phycology 30 (6), 999–1011. Scholin, C.A., Gulland, F., Doucette, G.J., Benson, S., Busman, M., Chavez, F.P., Cordaro, J., DeLong, R., De Vogelaere, A., Harvey, J., Haulena, M., Lefebvre, K.A., Lipscomb, T., Loscutoff, S., Lowenstine, L.J., Marin 3rd, R., Miller, P.E., McLellan, W.A., Moeller, P.D., Powell, C.L., Rowles, T., Silvagni, P., Silver, M., Spraker, T., Trainer, V.L., Van Dolah, F.M., 2000. Mortality of sea lions along the central California coast linked to a toxic diatom bloom. Nature 403 (6765), 80–84. Separation Processes Inc., 2005. Investigation of Microfiltration and Reverse Osmosis Seawater Desalination. National Water Research Institute. Shimizu, Y., Hsu, C.P., Genenah, A.A., 1981. Structure of saxitoxin in solutions and stereochemistry of dihydrosaxitoxins. Journal of American Chemical Society 103, 605–609. Shipe, R.F., Leinweber, A., Gruber, N., 2008. Abiotic controls of potentially harmful algal blooms in Santa Monica Bay, California. Continental Shelf Research 28 (18), 2584–2593. Shumway, S.E., Allen, S.M., Boersma, P.D., 2003. Marine birds and harmful algal blooms: sporadic victims or under-reported events? Harmful Algae 2 (1), 1–17. Smayda, T.J., 1990. Novel and nuisance phytoplankton blooms in the sea: evidence for a global epidemic. In: Grane´li, E., Gundstro¨m, B., Edler, L., Anderson, D.M. (Eds.), Toxic Marine Phytoplankton. Elsevier, New York, pp. 29–40. Smayda, T.J., 2008. Complexity in the eutrophication–harmful algal bloom relationship, with comment on the importance of grazing. Harmful Algae 8 (1), 140–151.

water research 44 (2010) 385–416

Smith, G., Ladizinsky, N., Miller, P., 2001. Amino acid profiles in species and strains of Pseudo-nitzschia from Monterey Bay, California: insights into the metabolic role(s) of domoic acid. In: Hallegraef, G., Blackburn, S., Bolch, C., Lewis, R. (Eds.). IOC, UNESCO, Hobart, Tasmania, pp. 324–327. Sommer, H., Meyer, K.F., 1937. Paralytic shellfish poisoning. Archives of Pathology 24, 560–598. Steidinger, K.A., 1993. Some taxonomic and biologic aspects of toxic dinoflagellates. In: Falconer, I. (Ed.), Algal Toxins in Seafood and Drinking Water. Academic Press, London, pp. 1–28. Stobo, L.A., Lewis, J., Quilliam, M.A., Hardstaff, W.R., Gallacher, S., Webster, L., Smith, E.A., McKenzie, M., 2003. Detection of yessotoxin in UK and Canadian isolates of phytoplankton and optimization and validation of LC–MS methods. Canadian Technical Report of Fisheries and Aquatic Sciences 2498, 8–14. Strom, S., Wolfe, G.V., Slajer, A., Lambert, S., Clough, J., 2003. Chemical defenses in the microplankton II: inhibition of protist feeding by B-dimethylsulfoniopropionate (DMSP). Limnology and Oceanography 48 (1), 230–237. Su, Z., Sheets, M., Ishida, H., Li, F., Barry, W.H., 2004. Saxitoxin blocks L-type ICa. Journal of Pharmacology and Experimental Therapeutics 308 (1), 324–329. Subba Rao, D.V., Quilliam, M.A., Pocklington, R., 1988. Domoic acid – a neurotoxic amino acid produced by the marine diatom Nitzschia pungens in culture. Canadian Journal of Fisheries and Aquatic Sciences 45 (12), 2076–2079. Suganuma, M., Fujiki, H., Suguri, H., Yoshizawa, S., Hirota, M., Nakayasu, M., Ojika, M., Wakamatsu, K., Yamada, K., Sugimura, T., 1988. Okadaic acid: an additional non-phorbol12-tetradecanoate-13-acetate-type tumor promoter. Proceedings of the National Academy of Sciences, United States of America 85 (6), 1768–1771. Sukhatme, G.A.D., Zhang, B., Oberg, C., Stauffer, B.A., Caron, D.A., 2007. The design and development of a wireless robotic networked aquatic microbial observing system. Environmental Engineering Science 24, 205–215. Sunda, W., 2006. Trace metals and harmful algal blooms. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 203–214. Sutherland, C.M., 2008. Diarrhetic Shellfish Toxins Linked to Local Dinophysis Populations in the California Coastal Waters of Monterey Bay. Dissertation, University of California, Santa Cruz, Santa Cruz. Tachibana, K., Scheuer, P.J., Tsukitani, Y., Kikuchi, H., Engen, D.V., Clardy, J., Gopichand, Y., Schmitz, F.J., 1981. Okadaic acid, a cytotoxic polyether from two marine sponges of the genus Halichondria. Journal of American Chemical Society 103 (9), 2469–2471. Takahashi, H., Kusumi, T., Kan, Y., Satake, M., Yasumoto, T., 1996. Determination of the absolute configuration of yessotoxin, a polyether compound implicated in diarrhetic shellfish poisoning, by NMR spectroscopic method using a chiral anisotropic reagent, methoxy-(2-naphthyl) acetic acid. Tetrahedron Letters 37 (39), 7087–7090. Takai, A., Bialojan, C., Troschka, M., Ru¨egg, J.C., 1987. Smooth muscle myosin phosphatase inhibition and force enhancement by black sponge toxin. FEBS Letters 217 (1), 81–84. Taylor, F.J.R., Horner, R.A., 1994. Red tides and other problems with harmful algal blooms in Pacific Northwest coastal waters. In: Review of the Marine Environment and Biota of Strait of Georgia, Puget Sound and Juan de Fuca Strait. Canadian Fisheries and Aquatic Science Technical Report 1948, pp. 175–186. Tenzer, B., Adin, A., Priel, M., 1999. Seawater filtration for fouling prevention under stormy conditions. Desalination 125 (1–3), 77–88.

415

Terao, K., Ito, E., Oarada, M., Murata, M., Yasumoto, T., 1990. Histopathological studies on experimental marine toxin poisoning: the effects in mice of yessotoxin isolated from Patinopecten yessoensis and of a desulfated derivative. Toxicon 28 (9), 1095–1104. Thomas, W.H., Zavoico, T., Hewes, C., 2001. Historical Phytoplankton Species Time-Series Data (1917–1939) from the North American Pacific Coast. Scripps Institution of Oceanography. Ref. no. 01-12. Throndsen, J., 1997. The plankton marine flagellates. In: Tomas, C.R. (Ed.), Identifying marine phytoplankton. Academic Press, San Diego, pp. 591–729. Trainer, V.L., 2002. Harmful algal blooms on the U.S. west coast. In: Taylor, F.J., Trainer, V.L. (Eds.), Harmful Algal Blooms in the PICES Region of the North Pacific, pp. 89–118. PICES Scientific Report 23. Trainer, V.L., Adams, N.G., Bill, B.D., Anulacion, F., Wekell, J.C., 1998. Concentration and dispersal of a Pseudo-nitzschia bloom in Penn Cove, Washington, USA. Natural Toxins 6 (3–4), 113–126. Trainer, V.L., Adams, N.G., Bill, B.D., Stehr, C.M., Wekell, J.C., Moeller, P., Busman, M., Woodruff, D., 2000. Domoic acid production near California coastal upwelling zones, June 1998. Limnology and Oceanography 45 (8), 1818–1833. Trainer, V.L., Adams, N.G., Wekell, J.C., 2001. Domoic acid producing Pseudo-nitzschia species off the U.S. west coast associated with toxification events. In: Hallegraeff, G.M., Blackburn, S.I., Bolch, C.J., Lewis, R.J. (Eds.), Harmful Algal Blooms 2000. Intergovernmental Oceanographic Commission of UNESCO, Paris, pp. 46–49. Trainer, V.L., Hickey, B.M., Horner, R.A., 2002. Biological and physical dynamics of domoic acid production off the Washington Coast. Limnology and Oceanography 47 (5), 1438–1446. Trainer, V.L., Eberhart, B.T.L., Wekell, J.C., Adams, N.G., Hanson, L., Cox, F., Dowell, J., 2003. Paralytic shellfish toxins in Puget Sound, Washington State. Journal of Shellfish Research 22 (1), 213–223. Trainer, V.L., Cochlan, W.P., Erickson, A., Bill, B.D., Cox, F.H., Borchert, J.A., Lefebvre, K., 2007. Recent domoic acid closures of shellfish harvest areas in Washington State inland waterways. Harmful Algae 6 (3), 449–459. Tubaro, A., Sidari, L., Della Loggia, R., Yasumoto, T., 1998. Occurrence of yessotoxin in phytoplankton and mussels from Northern Adriatic Sea. In: Proceedings of the VIII International Conference on Harmful Algae, Vigo, Spain (abstract). Twiner, M.J., Dechraoui, M.Y.B., Wang, Z., Mikulski, C.M., Henry, M. S., Pierce, R.H., Doucette, G.J., 2007. Extraction and analysis of lipophilic brevetoxins from the red tide dinoflagellate Karenia brevis. Analytical Biochemistry 369 (1), 128–135. Tyrell, J.V., Connell, L.B., Scholin, C.A., 2002. Monitoring for Heterosigma akashiwo using a sandwich hybridization assay. Harmful Algae 1 (2), 205–214. Van Dolah, F.M., Finley, E.L., Haynes, B.L., Doucette, G.J., Moeller, P.D., Ramsdell, J.S., 1994. Development of rapid and sensitive high throughput pharmacologic assays for marine phycotoxins. Natural Toxins 2 (4), 189–196. Vigilant, V.L., Silver, M., 2007. Domoic acid in benthic flatfish on the continental shelf of Monterey Bay, California, USA. Marine Biology 151 (6), 2053–2062. Walz, P.M., Garrison, D.L., Graham, W.M., Cattey, M.A., Tjeerdema, R.S., Silver, M.W., 1994. Domoic acid-producing diatom blooms in Monterey Bay, California: 1991–1993. Natural Toxins 2 (5), 271–279. Wang, J., Salata, J.J., Bennett, P.B., 2003. Saxitoxin is a gating modifier of hERG Kþ channels. Journal of General Physiology 121 (6), 583–598.

416

water research 44 (2010) 385–416

Watson, M., 1997. Control of Membrane Fouling Due to Algae. Applied Research Department. Technology Transfer Note. No.12. Wekell, J.C., Gauglitz, E.J.J., Barnett, H.J., Hatfield, C.L., Simons, D., Ayres, D., 1994. Occurrence of domoic acid in Washington state razor clams (Siliqua patula) during 1991–1993. Natural Toxins 2 (4), 197–205. Wells, M.L., Trick, C.G., Cochlan, W.P., Hughes, M.P., Trainer, V.L., 2005. Domoic acid: the synergy of iron, copper, and the toxicity of diatoms. Limnology and Oceanography 50 (6), 1908–1917. Wilf, M., Schierach, M.K., 2001. Improved performance and cost reduction of RO seawater systems using UF pretreatment. Desalination 135 (1–3), 61–68. Wong, J.L., Oesterlin, R., Rapoport, H., 1971. Struture of saxitoxin. Journal of American Chemical Society 93 (26), 7344–7345. Wright, J.L.C., Cembella, A.D., 1998. Ecophysiology and biosynthesis of polyether marine biotoxins. In: Anderson, D.M., Cembella, A.D., Hallegraeff, G.M. (Eds.), The Physiological

Ecology of Harmful Algal Blooms. Springer-Verlag, Heidelberg, pp. 427–451. Wright, J.L.C., Bird, C., de Freitas, A., Hampson, D., McDonald, J., Quilliam, M., 1990. Chemistry, biology, and toxicology of domoic acid and its isomers. Canadian Diseases Weekly Report 16 (Suppl. 1E), 21–26. Yasumoto, T., Murata, M., 1993. Marine toxins. Chemical Reviews 93, 1867–1909. Yasumoto, T., Takizawa, A., 1997. Fluorometric measurement of yessotoxins in shellfish by high-pressure liquid chromatography. Bioscience, Biotechnology and Biochemistry 61 (10), 1775–1777. Yasumoto, T., Oshima, Y., Sugawara, W., Fukuyo, Y., Oguri, H., Igarashi, T., Fujita, N., 1980. Identification of Dinophysis fortii as the causative organism of diarrhetic shellfish poisoning. Bulletin of the Japanese Society of Scientific Fisheries 46 (11), 1405–1411. Yasumoto, T., Murata, M., Oshima, Y., Sano, M., Matsumoto, G.K., Clardy, J., 1985. Diarrhetic shellfish toxins. Tetrahedron 41 (6), 1019–1025.

water research 44 (2010) 417–426

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Occurrence and removal of pharmaceuticals, caffeine and DEET in wastewater treatment plants of Beijing, China Qian Sui, Jun Huang, Shubo Deng, Gang Yu*, Qing Fan POPs Research Centre, Department of Environmental Science & Engineering, Tsinghua University, Beijing 10084, China

article info

abstract

Article history:

The occurrence and removal of 13 pharmaceuticals and 2 consumer products, including

Received 14 January 2009

antibiotic, antilipidemic, anti-inflammatory, anti-hypertensive, anticonvulsant, stimulant,

Received in revised form

insect repellent and antipsychotic, were investigated in four wastewater treatment plants

5 July 2009

(WWTPs) of Beijing, China. The compounds were extracted from wastewater samples by

Accepted 8 July 2009

solid-phase extraction (SPE) and analyzed by ultra-performance liquid chromatography

Available online 15 July 2009

coupled with tandem mass spectrometry (UPLC–MS/MS). Most of the target compounds were detected, with the concentrations of 4.4 ng L1–6.6 mg L1 and 2.2–320 ng L1 in the influents

Keywords:

and secondary effluents, respectively. These concentrations were consistent with their

Pharmaceuticals

consumptions in China, and much lower than those reported in the USA and Europe. Most

Wastewater

compounds were hardly removed in the primary treatment, while their removal rates ranging

Removal efficiency

from 12% to 100% were achieved during the secondary treatment. In the tertiary treatment,

Advanced treatment

different processes showed discrepant performances. The target compounds could not be

Risk assessment

eliminated by sand filtration, but the ozonation and microfiltration/reverse osmosis (MF/RO)

China

processes employed in two WWTPs were very effective to remove them, showing their main contributions to the removal of such micro-pollutants in wastewater treatment. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

With the progress of sensitive analytical techniques, the frequent detection of various pharmaceuticals in the aquatic environment has received global concerns of both the academic community and the public (Daughton and Ternes, 1999; Jones et al., 2005). After intake by humans or animals, the pharmaceuticals will be partially converted to metabolites, however, partially excreted unchanged or as conjugates, and finally delivered to the wastewater treatment plants (WWTPs). As there is no unit specifically designed to remove these compounds, the elimination by most WWTPs seems to be inefficient (Ternes, 1998; Castiglioni et al., 2006; Lishman et al., 2006; Nakada et al., 2006; Santos et al., 2007; Vieno et al., 2007b; Xu et al., 2007; Gulkowska et al., 2008; Paxeus, 2004). Together with treated wastewater, these compounds are

released to the aquatic environment, and consequently found to contaminate the receiving water bodies (Lindqvist et al., 2005; Kasprzyk-Hordern et al., 2009), or even raw water sources of drinking water treatment plant (Ternes et al., 2002; Vieno et al., 2007a; Radjenovic et al., 2008). Meanwhile, results of toxicology studies have revealed that some pharmaceuticals are suspected to have direct toxicity to certain aquatic organisms (Ferrari et al., 2003; Jjemba, 2006; Grung et al., 2008; Quinn et al., 2008). Besides, their continual but undetectable effects could accumulate slowly, and finally lead to irreversible change on wildlife and human beings (Daughton and Ternes, 1999). Therefore, the occurrence and behavior of pharmaceuticals in the WWTPs, which are both the sink and source of the compounds, should be focused on. So far, concentrations of pharmaceuticals from various therapeutic classes in the WWTPs have been well documented in the

* Corresponding author. Tel.: þ86 10 62787137; fax: þ86 10 62794006. E-mail address: [email protected] (G. Yu). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.010

418

water research 44 (2010) 417–426

North America (Thomas and Foster, 2005; Lishman et al., 2006), Japan (Nakada et al., 2006) and some European countries (Ternes, 1998; Castiglioni et al., 2006; Santos et al., 2007; Vieno et al., 2007b; Jones et al., 2007; Paxeus, 2004). Reported species and concentrations of pharmaceuticals varied from country to country, and plant to plant, owing to the different usage patterns. Meanwhile, the removal efficiencies of pharmaceuticals also varied much (Nakada et al., 2006; Gulkowska et al., 2008), indicating that the removal could be affected by both the compound-specific properties, and the factors concerning specific WWTPs, such as types of treatment processes, solids retention time (SRT), hydraulic retention time (HRT), temperature, etc. In recent years, very few studies about the situation in China have been reported. Only one specific therapeutic class, antibiotics, has been investigated by limited previous studies (Xu et al., 2007; Gulkowska et al., 2008; Chen et al., 2008). Therefore, it is necessary and important to investigate the occurrence and removal of pharmaceuticals from different therapeutic classes in the WWTPs of China. Due to the low efficiency of conventional wastewater treatment processes, some advanced treatment technologies have been evaluated. Ozonation was found to be effective to remove pharmaceuticals in real municipal WWTPs of Japan (Nakada et al., 2007; Okuda et al., 2008) and Germany (Ternes et al., 2003). Nanofiltration (NF) and reverse osmosis (RO) membrane filtration, the well-proven technologies to remove pharmaceuticals from different kinds of waters, have also been applied at bench, pilot and full scale (Khan et al., 2004; Nghiem et al., 2005; Drewes et al., 2005; Al-Rifai et al., 2007; Watkinson et al., 2007; Comerton et al., 2008; Radjenovic et al., 2008). Retention behavior of pharmaceuticals during the processes associated with physicochemical properties of pharmaceuticals, membranes as well as the solution chemistry, and mechanisms of pharmaceutical rejection have been discussed in Kimura et al. (2004), Nghiem et al. (2005), Nghiem and Coleman (2008) and Comerton et al. (2008). Recently, considering the requirement of reclaimed water, several advanced treatment facilities have been installed in the WWTPs of Beijing. However, the removal efficiency of micro-pollutants, such as pharmaceuticals, has not been evaluated yet. In the present study, we investigated the contamination levels of 13 pharmaceuticals and 2 consumer products from 8 classes (i.e. antibiotic, antilipidemic, anti-inflammatory, antihypertensive, anticonvulsant, stimulant, insect repellent and antipsychotic) in four WWTPs of Beijing, China, which have different advanced treatment units, and evaluated the elimination efficiencies of the target pharmaceuticals. To the best of our knowledge, this is the first report on the occurrence and removal of pharmaceuticals and consumer products from multiple classes in the WWTPs of China, especially for the situation during the advanced treatment processes.

2.

Materials and methods

2.1.

Chemicals

All the standards including chloramphenicol (CP), nalidixic acid (NA), trimethoprim (TP), bezafibrate (BF), clofibric acid (CA), gemfibrozil (GF), diclofenac (DF), indometacin (IM),

ketoprofen (KP), mefenamic acid (MA), metoprolol (MTP), carbamazepine (CBZ), caffeine (CF), N,N-diethyl-meta-toluamide (DEET) and sulpiride (SP) (Appendix) were of analytical grade (>90%), and purchased from Sigma–Aldrich (Steinheim, Germany). Isotopically labeled compounds, used as internal standards, were 13C-phenacetin obtained from Sigma– Aldrich, and 3D-mecoprop from Dr. Ehrenstorfer (Augsburg, Germany). HPLC grade methanol, acetone, dichloromethane, hexane, as well as formic acid were provided by Dikma (USA), and ultra-pure water was produced by a Milli-Q unit (Millipore, USA). Stock solutions of individual compound were prepared in methanol and mixture standards with different concentrations were prepared by diluting the stock solutions before each analytical run. All the solutions were stored at 4 C in the dark.

2.2.

Sample collection

Four full-scale municipal WWTPs, referred as A, B, C and D, were selected in our study. These WWTPs employ similar conventional treatment processes: primary treatment to remove particles coupled with secondary biological treatment. For the secondary biological treatment processes, WWTPs A and D employ anaerobic/anoxic/oxic (A2/O) activated sludge process, anoxic/oxic (A/O) activated sludge process is adopted in WWTP B, and WWTP C employs oxidation ditch (OD). Other detailed information on each WWTP, such as inhabitants served, daily flow, HRT and SRT are shown in Table 1. Part of the secondary effluents was further treated in WWTPs A, B and D, by the processes of ultrafiltration (UF)/ozone, sand filtration (SF) and microfiltration/reverse osmosis (MF/RO), respectively. In WWTP A, a dead-end ultrafiltration system (Zenon GE) is used. The whole system has 6 trains of Zee-Weed 1000 membrane. Each train contains 9 cassettes of 57–60 modules per cassette. The membrane, with the pore size of 0.02 mm, is made by PVDF. The module is operated in an outside/in configuration at a constant flow of 23 L (m2 h)1 and the total treatment capacity reaches 80,000 m3 d1. The membrane is hydraulically backwashed at a constant flow rate of 34 (m2 h)1, and 29 times per day. The backwash phase lasts for 1 min. Maintenance cleaning is conducted once per day. Membranes are soaked in the sodium hypochlorite solution (50 mg L1) for 25 min. For the ozonation process, gaseous ozone is generated from an ozone generator (Mitsubishi Electric). The ozone dosage and contact time in the reaction tank is 5 mg L1 and 15 min, respectively. The pH of the wastewater before ozonation ranges 6.5–8.0 and shows no significant change after ozonation. As the heart of the advanced treatment in WWTP D, a spiral-wound crossflow module is employed for the reverse osmosis (RO) membrane filtration. The RO membrane (Filmtec, DOW) is made from a thin-film composite polyamide material. Each module is designed to operate at a water flux of 1.3 m3 h1, and a product water recovery of 75–80%. The trans-membrane pressure is between 0.04 and 0.06 MPa, and the salt rejection remained at the level of 99%. Every 3–6 months, normally when the transmembrane pressure reaches above 0.06 MPa, the membrane is cleaned with 0.1%(w) sodium hydroxide solution (for organic foulants), 2%(w) citric acid (for inorganic foulants) and 0.5%(w) formaldehyde (as biocide). Schematic diagram of treatment processes in the four WWTPs is shown in Fig. 1.

419

water research 44 (2010) 417–426

Table 1 – Information of the WWTPs investigated. WWTP A B C D

Inhabitants served 103

Daily flow (103 m3)

HRT (h)

SRT (d)

Secondary treatment

Tertiary treatment

814 2400 480 2415

400 1000 200 600

11 11 15 10.67

12–15 20 12–16 15

A2/O A/O OD A2/O

UF/ozone SF – MF/RO

The samples were collected once from the four WWTPs during June and July 2008, with no compensation for HRT. All of them were collected as grab samples in duplicate (500 mL for influents and 1000 mL for the others) in prewashed amber glass bottles, kept in the cooler and transported to the laboratory. Immediately after delivery to the laboratory, they were filtered through prebaked (400 C, >4 h) glass microfiber filters (GF/F, Whatman) to remove particles and stored at 4 C before extraction.

2.3.

conditioned, wastewater samples, added with internal standards and adjusted to pH ¼ 7, were introduced to the cartridge via a PTFE tube, at a flow rate of 5–10 mL min1. After washing by 5 mL of 5.0% methanol solution, the cartridge was dried under vacuum for 2 h and eluted with 5 mL of methanol. The extract was then concentrated to 0.4 mL under a gentle nitrogen stream and stored at 4 C for analysis. Concentrations of the target compounds were analyzed using ultraperformance liquid chromatography coupled with tandem mass spectrometry (UPLC–MS/MS). Analytes were separated using Waters Acquity UPLC system (Waters Corporation, USA) equipped with Acquity UPLC BEH C18 column (50 2.1 mm, particle size of 1.7 mm), and detected by Quattro Premier XE tandem quadrupole mass spectrometry (Waters Corp., USA) equipped with an electrospray ionization source. The analysis was carried out in multiple reaction monitoring (MRM) mode,

Sample extraction and analysis

The method for the extraction and analysis of pharmaceuticals and consumer products is presented elsewhere (Sui et al., in press) and briefly described here. After the solid-phase extraction (SPE) cartridges (Oasis, HLB, 200 mg, 6 mL) were

a

Ozonation

Ultrafiltration

WWTP A

Tertiary Effluent Influent Grit Removal

A/O treatment

Secondary Clarifier

Secondary Effluent

Screen

b

Sand Filtration

WWTP B

Tertiary Effluent Influent Grit Removal

A2/O treatment

Primary Clarifier

Secondary Clarifier Secondary Effluent

Screen

c

WWTP C

Influent

Grit Removal

Oxidation Ditch

Screen

Secondary Clarifier

Secondary Effluent

MF

d

WWTP D

RO

Tertiary Effluent

`

Influent Grit Removal

Primary Clarifier

A2/O treatment

Secondary Clarifier

Secondary Effluent

Screen

Fig. 1 – Schematic diagram of the treatment processes in the four WWTPs and sampling site location (C).

420

water research 44 (2010) 417–426

and in general, two precursor ion/product ion transitions were monitored for one compound with the purpose of quantification and confirmation.

concentrations of some target compounds (i.e. caffeine, DEET, carbamazepine) resulted in slightly higher deviations.

2.4.

3.

Result and discussion

3.1.

Influents

Quality control

For each sampling, 500 mL Milli-Q water in an amber glass bottle as a field blank was brought to the WWTPs, exposed to the environment where the samples were taken from, and then delivered back to the laboratory with samples. For each set of samples (normally 10 samples), at least one procedural blank was prepared from ultra-pure water in the laboratory. Both the field blanks and procedural blanks were run identically to the wastewater samples, and the concentrations of target compounds were below the limit of quantification (LOQ). The absolute recoveries, calculated by comparing the concentrations of target compounds in spiked and unspiked wastewaters, were proved to be 73–102% and 50–95% in the effluent and influent for most compounds, respectively. While for several compounds (i.e. sulpiride, gemfibrozil, mefenamic acid), the absolute recoveries were not satisfactory. However, 13 C-phenacetin and 3D-mecoprop, the surrogate standards used for positive and negative ion mode respectively were able to compensate for the loss of most analytes, and relative recoveries were 67–130% for all the analytes in the effluent and 79–140% in the influents except mefenamic acid (251%) and nalidixic acid (178%). Therefore, the concentrations of these two compounds in the wastewater influents were not quantitatively determined and reported. The LOQs were 0.3–5.5 ng L1 and 0.7–20 ng L1 in the effluent and influent, respectively. Detailed information about the calibration, recoveries, LOQ, matrix effects, etc. were described in Sui et al. (in press), and briefly listed in Table 2. As duplicate samples were collected at each sampling site, mean concentrations were adopted. In most cases, deviations of duplicate samples were less than 20%. For some tertiary effluent samples, low

As shown in Fig. 2, 12 target compounds were detected in all the influent samples from the four WWTPs, while ketoprofen was below LOQ in all wastewater samples. The most abundant compounds detected were the consumer products, caffeine (3.4–6.6 mg L1) and N,N-diethyl-meta-toluamide (0.6–1.2 mg L1), probably due to the large consumption of drinks containing caffeine (i.e. coffee, tea, etc.) and wide application of insect repellent during the summer time when we sampled. Diclofenac, trimethoprim, sulpiride, carbamazepine, indometacin and metoprolol showed relatively high concentrations (Fig. 2). A similar composition distribution was observed among all the influents of the four WWTPs. The concentrations of target pharmaceuticals except diclofenac and trimethoprim, were much lower than those reported in the European and North American countries (Thomas and Foster, 2005; Lishman et al., 2006; Vanderford and Snyder, 2006; Santos et al., 2007; Gomez et al., 2007; Vieno et al., 2007b; Huerta-Fontela et al., 2008). For instance, the concentrations of ketoprofen in the wastewater influents were recorded to be 2.0 0.6 mg L1 in Finland (Lindqvist et al., 2005), 200 ng L1 in Australia (Al-Rifai et al., 2007), and 300–1360 ng L1 in Spain (Santos et al., 2007), while in the influents of four WWTPs in Beijing, it could not be detected. Concerning gemfibrozil, which is used to lower cholesterol and triglyceride levels in the blood, the contamination level found in the present study was 24–140 ng L1, even 1 or 2 order of magnitude lower than those in the USA (4770 ng L1, Vanderford and Snyder, 2006) and Canada (418 ng L1,

Table 2 – Instrumental quantification limit (IQL), limit of quantification (LOQ), absolute recovery (AR), relative recovery (RR) and matrix suppression of target compounds. Compounds

BF CA CBZ CF CP DEET DF GF IM KP MA MTP NA SP TP

IQL (pg)

0.5 10 2.5 2.5 2.5 2.5 10 10 2.5 10 10 2.5 2.5 0.5 2.5

LOQ (ng L1) Effluent

Influent

0.3 5.4 1.0 1.7 1.0 1.3 4.7 4.2 1.3 5.5 5.5 1.1 1.1 0.4 1.0

0.7 16 2.8 3.3 2.3 3.2 9.4 20 2.8 18 5.2 3.3 2.1 1.0 2.7

AR (n ¼ 6, %) Effluent

a Value in the brackets refers to the deviation of the recovery.

74 74 100 60 98 76 86 95 80 69 73 88 91 51 102

(3) (4) (4) (3) (7) (3) (12) (12) (12) (9) (11) (3) (9) (5) (7)

a

RR (n ¼ 6, %) a

Influent 58 50 71 61 86 63 85 40 73 44 154 60 95 42 74

(5) (4) (12) (37) (9) (6) (19) (11) (13) (3) (24) (3) (9) (3) (16)

a

Effluent 94 94 130 77 124 99 109 120 101 87 92 115 118 67 132

(4) (6) (6) (4) (10) (5) (16) (16) (16) (12) (15) (5) (12) (7) (9)

Matrix effect (%) a

Influent 94 82 133 114 140 118 139 65 119 72 251 113 178 79 139

(10) (8) (23) (70) (17) (14) (32) (18) (22) (7) (41) (9) (20) (7) (31)

27 30 6 49 39 24 4 28 1 46 13 15 3 63 5

421

water research 44 (2010) 417–426

a

10000

Concentration (ng L-1)

Influents

1000

100

10

Cpred ¼ SP

CBZ

MTP

IM

DF

GF

CA

BF

TP

CP

CF

DEET

1

10000

Concentration (ng L-1)

Pharmaceuticals & Consumer Products

b

1000

Lishman et al., 2006). The low levels of target pharmaceuticals were probably due to the lower per capita consumption in China than in the countries with higher socioeconomic statuses, where medical care is more prevalent (Thomas and Foster, 2005). The per capita consumption rate of gemfibrozil in China is estimated to be 0.036 mg person1 d1 (Table 3), lower than those in Germany (0.2 mg person1 d1, Ternes, 1998) and Canada (0.2 mg person1 d1, Lishman et al., 2006). Since the levels of target pharmaceuticals were somewhat different from those of European and North American countries, we theoretically calculate the concentration of pharmaceuticals in the wastewater influent by the following equation (Lindqvist et al., 2005; Nakada et al., 2006)

Secondary effluents

100

10

SP

CBZ

MTP

MA

IM

DF

GF

CA

BF

TP

CF

CP

DEET

1

Pharmaceuticals & Consumer Products Fig. 2 – Concentrations of target pharmaceuticals in wastewater influents (a) and secondary effluents (b) of four WWTPs in Beijing.

T e% I 1012 365 P Q

(1)

where Cpred is the predicted concentration of the pharmaceutical in wastewater influent (ng L1); T is the total production of a pharmaceutical both for human and animal use in China per year (ton year1), P is the population of China, e% is the amount of the pharmaceutical excreted unchanged, I is the number of inhabitants served and Q is the influent flow (m3 d1). The predicted concentrations of gemfibrozil, diclofenac, indometacin, ketoprofen, carbamazepine, and sulpiride were comparable to those measured in the influents (Table 3). Much lower measured concentration than predicted concentration of chloramphenicol was probably because it had been forbidden for use in food and aquaculture in China since 2005, and the available data about the production of pharmaceuticals were based on the year of 2004. It should be noticed that since there is no available data on the total consumption of any pharmaceutical, we used figures for total production of individual pharmaceutical instead. Therefore, the differences between the amounts actually produced and applied as well as the amount used in human and veterinary medicine could not be distinguished, which might result in overestimation of the theoretical concentration. Nevertheless, the comparability between the predicted concentrations and measured concentrations illustrates the overall reasonability of the approach.

Table 3 – Outputs, per capita consumption, predicted concentrations (PECs) and measured concentrations (MECs) of some pharmaceuticals in the wastewater influents of WWTPs investigated. Compound

CP TP GF DF IM KP CBZ SP

Outputa (tons year1)

Per capita consumption (mg person1 d1)

Excreted unchangedb (%)

PEC (ng L1)

MEC (mean, ng L1)

1929 2352 17 328 277 92 395 98

4.051 4.939 0.036 0.689 0.582 0.193 0.830 0.206

5–10 45 76 15 10–20 2.7 2–3 15–70

844 6173 76 287 242 15 58 243

31 400 60 318 129 n.d.c 113 157

a From CMEIN (2005). b From Bolton and Null (1981), Ternes (1998), Khan and Ongerth (2004), Niwa et al. (2005), Nakada et al. (2006), Jjemba (2006). c n.d. ¼ Not detected.

422

water research 44 (2010) 417–426

Compound

MECeff (ng L1)

PNECa (ng L1)

MEC/PNEC

15 140 4.7 13 204 n.d. 79 108

182,000 16,000 6000 42,000 100 306,000 31,000 420

0.0001 0.0087 0.0007 0.0003 2.0440 – 0.0025 0.2574

CF TP BF CA DF KP MTP CBZ

a From Santos et al. (2007), Lindqvist et al. (2005), Grung et al. (2008), Ferrari et al. (2003), Huschek et al. (2004).

the dominant contributor in the wastewater effluent, the risk quotient was higher than 1, implicating a risk to the aquatic environment.

3.3.

Removal efficiency of conventional treatment

The removal efficiency during the primary treatment was low, indicating no significant adsorption of target compounds to the particles removed in this stage (Fig. 4). Most of the pharmaceuticals and consumer products have log Kow values of less than 3.0, so they are not expected to adsorb significantly to the particles. Other pharmaceuticals with higher Kow values, such as gemfibrozil, have much lower pKa values than the pH of wastewater. Therefore, they are dissociated and expected to be >98% in the aqueous phase (Thomas and Foster, 2005), and not bound to the particles. During the secondary treatment, the average removal rate for different compounds ranged from 12% to 100% (Fig. 4). Caffeine, bezafibrate, trimethoprim and DEET were effectively removed, with the average efficiency of 100 0%, 88 12%, 76 24% and 69 21%, respectively. These results were comparable with those found in the previous studies (Ternes, 1998; Okuda et al., 2008; Thomas and Foster, 2005; Castiglioni et al., 2006). Caffeine was proved to be readily biodegradable (Okuda et al., 2008; Thomas and Foster, 2005; Huerta-Fontela

120

Primary treatment Secondary treatment

100

A

40 20

-40 0

20

40

60

80

100

Pharmaceutical Compostion (%) Fig. 3 – Composition profiles of target pharmaceuticals in secondary effluent samples from four WWTPs in Beijing.

-60

Pharmaceutical Fig. 4 – Removal efficiencies of target pharmaceuticals during the conventional treatment.

SP

CBZ

GF

CA

-20

BF

0 TP

B

60

CF

C

80

DEET

Others GF MTP IM CBZ SP TP DF DEET

Removal efficiency (%)

WWTP

D

MTP

Similar to the influent samples, ketoprofen was below the LOQ in all the secondary effluent samples. Nalidixic acid and chloramphenicol were detected only in one WWTP, with the concentration of 8.1 and 19 ng L1, respectively. The mean concentrations of the other 12 compounds ranged from 5 to 200 ng L1 (Fig. 2). Diclofenac, N,N-diethyl-meta-toluamide, trimethoprim, sulpiride and indometacin showed high concentrations in the secondary effluents. Carbamazepine and metoprolol followed, with the concentrations ranging from 69 to 120 ng L1, and 60 to 108 ng L1, respectively. Other compounds, such as caffeine, gemfibrozil and mefenamic acid, occurred at the lowest levels. Despite of a wide variation of trimethoprim from different WWTPs, the composition profiles of target pharmaceuticals in secondary effluents from the four WWTPs were quite similar (Fig. 3). The concentration levels of most pharmaceuticals and consumer products detected in the secondary effluent were also lower than those reported in the Europe. They were over 100 ng L1, in some cases even up to 500 ng L1 in the wastewater effluents of the European countries (Santos et al., 2007; Gomez et al., 2007; Ternes, 1998; Vieno et al., 2007b). While in the present study, 10 out of 15 compounds were less than 100 ng L1, and none of them exceeded 400 ng L1 in any effluent samples (Fig. 2). Our results were in agreement with those in Japan (Nakada et al., 2006), Korea (Kim et al., 2007) and some other cities of China (Xu et al., 2007; Gulkowska et al., 2008; Chen et al., 2008). For instance, the concentrations of chloramphenicol in the effluents of 4 WWTPs in Guangzhou were
Table 4 – Measured concentrations in the effluent samples, predicted no-effect concentrations and risk quotients (MEC/PNEC) of target compounds.

IM

Secondary effluent

DF

3.2.

423

water research 44 (2010) 417–426

et al., 2008; Gomez et al., 2007). The removal rate of bezafibrate was found to be 87% in six Italian WWTPs in summer time (Castiglioni et al., 2006), very similar to that observed in our study. The concentrations of DEET were decreased by more than 80% during the biological treatment in the WWTP of Japan (Okuda et al., 2008), slightly better than our results. A second group of pharmaceuticals, including three antiinflammatory drugs, clofibric acid, gembrozil, metoprolol and sulpiride, had lower removal rates with large variation in different WWTPs studied. For instance, 28–53% of diclofenac, a representative of the anti-inflammatory drugs, was removed by secondary treatment in the WWTPs, which was between 26% in Finland (Lindqvist et al., 2005) and 69% in Germany (Ternes, 1998). The elimination of these compounds may be highly dependent on the configurations and operation conditions of individual WWTP as well as wastewater characteristics, and thus no definitive conclusion could be reached. Higher load of carbamazepine was found in the secondary effluent than in the primary effluent, indicating negative removal efficiency during the secondary treatment. Some carbamazepine was found to be excreted as the form of conjugates (Vieno et al., 2007b), which was biodegraded to carbamazepine by enzymatic processes during the secondary treatment, resulting in additional amounts of carbamazepine in the secondary effluent. However, as the calculations of all the removal efficiencies were based on grab samples that were not sampled with a hydraulic lag in the present study, some error might be brought in due to diurnal variation of the concentration. Therefore, the present study only provided a snapshot of the removal of pharmaceuticals and consumer products in the WWTPs of Beijing. To better illustrate that, 24-h composite samples that are lagged by HRT should be collected and analyzed in further studies. It has been reported that high HRT (>12 h) and SRT (>10 d) may contribute to an increased removal rate of pharmaceuticals (Jones et al., 2007; Vieno et al., 2007b). In the present study, the WWTP C, in which the HRT was higher than the others, was the best in removing these compounds, due to increased contact time of target compounds and the microorganisms. On the other hand, the different SRTs did not have significant effects on the removal efficiency, probably because the SRTs in all the four WWTPs were relatively high (>10 d), and without large differences. In addition, it is noteworthy that the WWTP C employed oxidation ditches, which showed better removal of natural estrogens and estrogenic activity than A/O (Hashimoto et al., 2007). It also could be the reason for the higher removal efficiencies in the WWTP C. Further investigation for different types of WWTPs is necessary to confirm the results mentioned above.

3.4.

Removal efficiency in advanced treatment processes

The removal efficiencies of the pharmaceuticals during the SF, UF/ozonation, as well as MF/RO treatment in three corresponding WWTPs are listed in Table 5. Generally, sand filtration was not effective for these compounds. Only trimethoprim, DEET and gemfibrozil were removed slightly during this treatment process. It should be noticed that these compounds were efficiently removed in the secondary treatment, indicating that the biodegradation on

Table 5 – Removal efficiencies (%) of target pharmaceuticals and consumer products by advanced treatment processes in studied WWTPs. Compound

DEET CF TP BF CA GF DF IM MA MTP CBZ SP

WWTP A

WWTP B

WWTP D

UF

Ozone

SF

MF/RO

0–50 <0 0–50 0–50 <0 0–50 0–50 0–50 0–50 0–50 <0 0–50

50–80 50–80 >90 0–50 50–80 80–90 >90 >90 80–90 80–90 >90 >90

0–50 <0 80–90 0–50 <0 50–80 <0 <0 <0 <0 0–50 0–50

>90 50–80 >90 >90 80–90 >90 >90 >90 0–50 >90 >90 >90

the biofilm present on the sand particle, rather than the removal with particles, may be the main reason for their elimination (Gobel et al., 2007). The results showed that ozonation is effective in removing most of the target compounds, probably due to the operation conditions employed in WWTP A (ozone dosage: 5 mg L1, contact time: 15 min). Carbamazepine, diclofenac, indomethacin, sulpiride and trimethoprim were significantly eliminated, with the removal rates of above 95%. The double bond in the azepine ring of carbamazepine and pyrrole ring of indomethacin, and the non-protonated amine of diclofenac and trimethoprim were susceptible to ozone attack (Vieno et al., 2007a; Nakada et al., 2007; Westerhoff et al., 2005). The removal efficiencies of DEET and metoprolol were modest. The amide group, which is not reactive with ozone, could be the reason for the modest removal of DEET (Nakada et al., 2007). Low removal efficiencies were found for bezafibrate, clofibric acid, as well as caffeine. Only 14% of bezafibrate disappeared in the ozone process, consistent with its low rate constants with ozone (590 50 M1 S1, Huber et al., 2003). The reaction site of bezafibrate is the R-oxysubstituent (–O–C(CH3)2COOH) on one of the aromatic rings. However, as the pKa of bezafibrate is 3.6, the R-oxysubstituent cannot be deprotonated and consequently the overall rate constant at pH > 4 is much lower (Huber et al., 2003). It should be noticed that during the ozonation, most of the pharmaceuticals were not mineralized but transformed to the oxidation products. For instance, three oxidation products containing quinazoline-based functional groups were identified during the ozonation of CBZ (Mcdowell et al., 2005). The good performance of ozonation in the present study was consistent with Ternes et al. (2003), Huber et al. (2005) and Okuda et al. (2008). When 5 mg L1 ozone was applied to the effluent of a municipal WWTP in Germany (contact time: 18 min), target compounds, such as trimethoprim, carbamazepine, indomethacin, clofibric acid, were removed by more than 50% (Ternes et al., 2003). Huber et al. (2005) conducted a pilot study on the oxidation of pharmaceuticals during ozonation of conventional activated sludge (CAS) and membrane bio-reactor (MBR) effluents with various ozone dosages, and found that macrolide and sulfonamide antibiotics, estrogens, and acidic pharmaceuticals diclofenac,

water research 44 (2010) 417–426

a

Tertiary treatment Conventional treatment

120 100 80 60 40 20

SP

CBZ

MTP

IM

DF

GF

CA

BF

TP

CP

-20

CF

0 DEET

Removal Contribution (%)

Pharmaceutical

b 180

Tertiary treatment Secondary treatment Primary treatment

160 140 120 100 80 60 40 20

-40

SP

CBZ

MTP

IM

DF

GF

CA

BF

TP

CF

0 -20

DEET

naproxen and indomethacin were oxidized by more than 90–99% for ozone doses 2 mg L1 in all effluents. The elimination by ultrafiltration in the WWTP A was low for all the investigated compounds. The molecular weight cutoff (MWCO) of UF membranes was much higher than 1000 Da, thus UF membranes showed poor retention of all the investigated pharmaceuticals, of which the molecular weight are less than 400 Da. The removal of individual target compound was less than 50%, and might be due to the adsorption onto the membrane. It has been also demonstrated that UF membrane typically had less than 40% retention of 27 PPCPs, and the mass balances calculated based on the concentration of each compound in feed, permeate and retentate showed the observed retention was significantly governed by adsorption (Yoon et al., 2006). In contrast, MF/RO employed in WWTP D was very effective. In the effluent of MF/RO, all the target compounds except caffeine were not detected. Generally, one or combination of three basic mechanisms could be involved during the rejection of solute by NF/RO membrane: steric effect, charge exclusion and adsorption (Radjenovic et al., 2008). For most pharmaceuticals, the rejections were considered to be dominated by steric interaction in ‘‘tight’’ NF or RO membrane filtration (Nghiem et al., 2005; Radjenovic et al., 2008). As most investigated compounds have molecular weights about 200–400 Da, smaller than MWCO of RO membrane applied, excellent rejection of most pharmaceuticals by RO membrane was observed in this study as well as in previous studies (Kimura et al., 2004; Al-Rifai et al., 2007; Radjenovic et al., 2008). Besides, membrane fouling and the presence of organic matter in the wastewater effluents likely contributed to higher rejections of pharmaceuticals, especially for some hydrophobic ionogenic compound (Nghiem and Coleman, 2008; Comerton et al., 2008). Nevertheless, the rejections of two compounds, caffeine and mefenamic acid were slightly lower (i.e. 50–80% and 0–50%, respectively). The concentration of mefenamic acid in feed wastewaters of MF/RO membrane process was very low, only a bit higher than its LOQ in the wastewater effluent, which could be the reason for the low rejection rate. The low retention of caffeine in the present study was inaccordance with Drewes et al. (2005). They found that in two full-scale RO facilities, target EDCs and PPCPs were efficiently rejected to below detection limit except for caffeine, still detected in the permeates. The physiochemical properties might explain the low rejection rate of caffeine. As a representative of hydrophilic and non-ionic compounds, the rejection driven by charge exclusion and adsorption is negligible, and steric exclusion is solely responsible for the retention of caffeine (Nghiem et al., 2005). However, the molecular weight of caffeine is 195 Da, smaller than other target compounds, and might result in the decreased removal efficiency during the RO membrane filtration process. Compared to the other two, the WWTPs employing ozone and RO membrane filtration as advanced treatment were more efficient in removing pharmaceuticals. For these WWTPs, the advanced treatment made a significant contribution to the total elimination of most pharmaceuticals (Fig. 5). Therefore, the utility of efficient advanced treatment could be considered as a tool to reduce pharmaceuticals in the

Contribution to removal efficiency (%)

424

-60

Pharmaceutical Fig. 5 – Contributions of primary, secondary (or conventional treatment) and tertiary treatment to the total elimination of selected pharmaceuticals in WWTP A (a) and WWTP D (b).

municipal wastewater treatment plants. However, the problems of membrane fouling and further treatment or disposal of retentate challenge the application of RO membrane filtration (Van der Bruggen et al., 2008). For ozonation, as most of the pharmaceuticals could not be mineralized, and oxidation products are formed from parent pharmaceutical compounds (Mcdowell et al., 2005), more research is required to identify the oxidation products and their potential toxicity during the partial oxidation process (Nakada et al., 2007). Besides, economic feasibility should be evaluated by estimating the energy consumption and investment and operation costs for both advanced treatment processes (Joss et al., 2008).

4.

Conclusion

13 out of 15 pharmaceuticals and consumer products from eight classes were detected at four WWTPs in Beijing, China. The concentrations of most compounds in the influent and secondary effluent were lower than those reported in the USA and Europe, but consistent with the production profile of the

water research 44 (2010) 417–426

pharmaceuticals in China. According to the result of risk assessment for the secondary effluent, only diclofenac might pose a risk to the aquatic environment. The removal efficiencies by the conventional treatment varied for different compounds, depending on their chemical structures, physiochemical properties, as well as the specific treatment processes utilized at each WWTP. Further removal could be achieved by adopting some advanced treatment processes, such as ozonation and MF/RO. However, others, such as sand filtration, showed low efficiency in removing these compounds from secondary effluent.

Acknowledgement This study was supported by the National Science Fund for Distinguished Young Scholars (No. 50625823).

Appendix. Supplementary data Supplementary information related to this article can be found at doi:10.1016/j.watres.2009.07.010.

references

Al-Rifai, J.H., Gabelish, C.L., Schafer, A.I., 2007. Occurrence of pharmaceutically active and non-steroidal estrogenic compounds in three different wastewater recycling schemes in Australia. Chemosphere 69 (5), 803–815. Bolton, S., Null, G., 1981. Caffeine: psychological effects, use and abuse. Orthomolecular Psychiatry 10 (3), 202–211. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environmental Science and Technology 40 (1), 357–363. Chen, H.C., Wang, P.L., Ding, W.H., 2008. Using liquid chromatography–ion trap mass spectrometry to determine pharmaceutical residues in Taiwanese rivers and wastewaters. Chemosphere 72 (6), 863–869. China Medicine Economic Information Net (CMEIN), 2005. Chinese Medical Statistical Yearbook. Products section. China Medicine Economic Information Net (CMEIN), Beijing, China. Comerton, A.M., Andrews, R.C., Bagley, D.M., Hao, C.Y., 2008. The rejection of endocrine disrupting and pharmaceutically active compounds by NF and RO membranes as a function of compound and water matrix properties. Journal of Membrane Science 313 (1–2), 323–335. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107 (S6), 907–938. Drewes, J.E., Bellona, C., Oekekoven, M., Xu, P., Kim, T.U., Amy, G., 2005. Rejection of wastewater-derived micropollutants in high-pressure membrane applications leading to indirect potable reuse. Environmental Progress 24 (4), 400–409. European Environment Agency, 1998. Environmental Risk Assessment – Approaches, Experiences and Information Source. European Environment Agency, London, pp. 68–86. Ferrari, B., Paxeus, N., Giudice, R.L., Pollio, A., Grric, J., 2003. Ecotoxicological impact of pharmaceuticals found in treated

425

wastewaters: study of carbamazepine, clofibric acid, and diclofenac. Ecotoxicology and Environmental Safety 55 (3), 359–370. Gobel, A., Mcardell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate of sulfonamides, macrolides, and trimethoprim in different wastewater treatment technologies. Science of the Total Environment 372 (2–3), 361–371. Gomez, M.J., Bueno, M.J.M., Lacorte, S., Fernandez-Alba, A.R., Aguera, A., 2007. Pilot survey monitoring pharmaceuticals and related compounds in a sewage treatment plant located on the Mediterranean coast. Chemosphere 66 (6), 993–1002. Grung, M., Kallqvist, T., Sakshaug, S., Skurtveit, S., Thomas, K.V., 2008. Environmental assessment of Norwegian priority pharmaceuticals based on the EMEA guideline. Ecotoxicology and Environmental Safety 71 (2), 328–340. Gulkowska, A., Leung, H.W., So, M.K., Taniyasu, S., Yamashita, N., Yeung, L.W.Y., Richardson, B.J., Lei, A.P., Giesy, J.P., Lam, P.K.S., 2008. Removal of antibiotics from wastewater by sewage treatment facilities in Hongkong and Shenzhen, China. Water Research 42 (1–2), 395–403. Hashimoto, T., Onda, K., Nakamura, Y., Tada, K., Miya, A., Murakami, T., 2007. Comparison of natural estrogen removal efficiency in the conventional activated sludge process and the oxidation ditch process. Water Research 41 (10), 2117–2126. Huber, M.M., Canonica, S., Park, G.Y., Gunten, U.V., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environmental Science and Technology 37 (5), 1016–1024. Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D., Mcardell, C.S., Ried, A., Siegrist, H., Ternes, T.A., Gunten, U.V., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science and Technology 39 (11), 4290–4299. Huerta-Fontela, M., Galceran, M.T., Martin-Alonso, J., Ventura, F., 2008. Occurrence of psychoactive stimulatory drugs in wastewaters in north-eastern Spain. Science of the Total Environment 397 (1–3), 31–40. Huschek, G., Hansen, P.D., Maurer, H.H., Krengel, D., Kayser, A., 2004. Environmental risk assessment of medicinal products for human use according to European Commission recommendations. Environmental Toxicology 19 (3), 226–240. Jjemba, P.K., 2006. Excretion and ecotoxicity of pharmaceutical and personal care products in the environment. Ecotoxicology and Environmental Safety 63 (1), 113–130. Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2005. Human pharmaceuticals in wastewater treatment processes. Critical Reviews in Environmental Science and Technology 35 (4), 401–427. Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2007. The occurrence and removal of selected pharmaceutical compounds in a sewage treatment works utilising activated sludge treatment. Environmental Pollution 145 (3), 738–744. Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade wastewater treatment for removing organic micropollutants. Water Science and Technology 57 (2), 251–255. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Research 43 (2), 363–380. Khan, S.J., Wintgens, T., Sherman, P., Zaricky, J., Schafer, A.I., 2004. Removal of hormones and pharmaceuticals in the advanced water recycling demonstration plant in Queensland, Australia. Water Science and Technology 50 (5), 15–22. Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical residues in Australian sewage by quantities of use and fugacity calculations. Chemosphere 54 (3), 355–367.

426

water research 44 (2010) 417–426

Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Research 41 (5), 1013–1021. Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of neutral endocrine disrupting compounds (EDCs) and pharmaceutical active compounds (PhACs) by RO membranes. Journal of Membrane Science 245 (1–2), 71–78. Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J., Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006. Occurrence and reductions of pharmaceuticals and personal care products and estrogens by municipal wastewater treatment plants in Ontario, Canada. Science of the Total Environment 367 (2–3), 544–558. Lindqvist, N., Tuhkanen, T., Kronberg, L., 2005. Occurrence of acidic pharmaceuticals in raw and treated sewages and in receiving waters. Water Research 39 (11), 2219–2228. Mcdowell, D.C., Huber, M.M., Wagner, M., Gunten, U.V., Ternes, T.A., 2005. Ozonation of carbamazepine in drinking water: identification and kinetic study of major oxidation products. Environmental Science and Technology 39 (20), 8014–8022. Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., Takada, H., 2006. Pharmaceutical chemicals and endocrine disrupters in municipal wastewater in Tokyo and their removal during activated sludge treatment. Water Research 40 (17), 3297–3303. Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S., Sato, N., Takada, H., 2007. Removal of selected pharmaceuticals and personal care products (PPCPs) and endocrine-disrupting chemicals (EDCs) during sand filtration and ozonation at a municipal sewage treatment plant. Water Research 41 (19), 4372–4382. Nghiem, L.D., Schafer, A.I., Elimelech, M., 2005. Pharmaceutical retention mechanisms by nanofiltration membrane. Environmental Science and Technology 39 (19), 7698–7705. Nghiem, L.D., Coleman, P.J., 2008. NF/RO filtration of the hydrophobic ionogenic compound triclosan: transport mechanisms and the influence of membrane fouling. Separation and Purification Technology 62 (3), 709–716. Niwa, T., Inoue, S., Shiraga, T., Takagi, A., 2005. No inhibition of cytochrome P450 activities in human liver microsomes by sulpiride, an antipsychotic drug. Biological and Pharmaceutical Bulletin 28 (1), 188–191. Okuda, T., Kobayashi, Y., Nagao, R., Yamashita, N., Tanaka, H., Tanaka, S., Fujii, S., Konishi, C., Houwa, I., 2008. Removal efficiency of 66 pharmaceuticals during wastewater treatment process in Japan. Water Science and Technology 57 (1), 65–71. Paxeus, N., 2004. Removal of selected non-steroidal antiinflammatory drugs (NSAIDs), gemfibrozil, carbamazepine, beta-blockers, trimethoprim and triclosan in conventional wastewater treatment plants in five EU countries and their discharge to the aquatic environment. Water Science and Technology 50 (5), 253–260. Quinn, B., Gagne, F., Blaise, C., 2008. An investigation into the acute and chronic toxicity of eleven pharmaceuticals (and their solvents) found in wastewater effluent on the cnidarian, Hydra attenuata. Science of the Total Environment 389 (2–3), 306–314. Radjenovic, J., Petrovic, M., Ventura, F., Barcelo, D., 2008. Rejection of pharmaceuticals in nanofiltration and reverse osmosis

membrane drinking water treatment. Water Research 42 (14), 3601–3610. Santos, J.L., Aparicio, I., Alonso, E., 2007. Occurrence and risk assessment of pharmaceutically active compounds in wastewater treatment plants. A case study: Seville city (Spain). Environment International 33 (4), 596–601. Sui, Q., Huang, J., Deng, S. B., Yu, G. Rapid determination of pharmaceuticals from multiple therapeutic classes in wastewater by solid-phase extraction and ultra-performance liquid chromatography tandem mass spectrometry. Chinese Science Bulletin, in press. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32 (11), 3245–3260. Ternes, T.A., Meisenheimer, M., Mcdowell, D., Sacher, F., Brauch, H.J., Haist-gulde, B., Preuss, G., Wilme, U., Zulei-seibert, N., 2002. Removal of pharmaceuticals during drinking water treatment. Environmental Science and Technology 36 (17), 3855–3863. Ternes, T.A., Stuber, J., Herrmann, N., Mcdowell, D., Ried, A., Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal of pharmaceuticals, contrast media and musk fragrances from wastewater? Water Research 37 (8), 1976–1982. Thomas, P.M., Foster, G.D., 2005. Tracking acidic pharmaceuticals, caffeine, and triclosan through the wastewater treatment process. Environmental Toxicology and Chemistry 24 (1), 25–30. Vanderford, B.J., Snyder, S.A., 2006. Analysis of pharmaceuticals in water by isotope dilution liquid chromatography/tandem mass spectrometry. Environmental Science and Technology 40 (23), 7312–7320. Van der Bruggen, B., Manttari, M., Nystrom, M., 2008. Drawbacks of applying nanofiltration and how to avoid them: a review. Separation and Purification Technology 63 (2), 251–263. Vieno, N.M., Harkki, H., Tuhkanen, T., Kronberg, L., 2007a. Occurrence of pharmaceuticals in river water and their elimination in a pilot-scale drinking water treatment plant. Environmental Science and Technology 41 (14), 5077–5084. Vieno, N., Tuhkanen, T., Kronberg, L., 2007b. Elimination of pharmaceuticals in sewage treatment plants in Finland. Water Research 41 (5), 1001–1012. Watkinson, A.J., Murby, E.J., Costanzo, S.D., 2007. Removal of antibiotics in conventional and advanced wastewater treatment: implications for environmental discharge and wastewater recycling. Water Research 41 (18), 4164–4176. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science and Technology 39 (17), 6649–6663. Xu, W.H., Zhang, G., Li, X.D., Zou, S.C., Li, P., Hu, Z.H., Li, J., 2007. Occurrence and elimination of antibiotics at four sewage treatment plants in the Pearl River Delta, South China. Water Research 41 (19), 4526–4534. Yoon, Y., Westerhoff, P., Snyder, S.A., Wert, E.C., 2006. Nanofiltration and ultrafiltration of endocrine disrupting compounds, pharmaceuticals and personal care products. Journal of Membrane Science 270 (1–2), 88–100.

water research 44 (2010) 427–438

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Application of the combination index (CI)-isobologram equation to study the toxicological interactions of lipid regulators in two aquatic bioluminescent organisms Ismael Rodea-Palomaresa,1, Alice L. Petreb,1, Karina Boltesb, Francisco Legane´sa, Jose´ Antonio Perdigo´n-Melo´nb, Roberto Rosalb, Francisca Ferna´ndez-Pin˜asa,* a

Departamento de Biologı´a, Facultad de Ciencias, Universidad Auto´noma de Madrid, 2 Darwin Street, Cantoblanco, 28049 Madrid, Spain Departamento de Ingenierı´a Quı´mica, Universidad de Alcala´, Alcala´ de Henares, E-28871 Madrid, Spain

b

article info

abstract

Article history:

Pharmaceuticals in the aquatic environment do not appear singly and usually occur as

Received 26 March 2009

complex mixtures, whose combined effect may exhibit toxicity to the aquatic biota. We

Received in revised form

report an environmental application of the combination index (CI)-isobologram equation,

13 July 2009

a method widely used in pharmacology to study drug interactions, to determine the nature

Accepted 18 July 2009

of toxicological interactions of three fibrates toward two aquatic bioluminescent organ-

Available online 25 July 2009

isms, Vibrio fischeri and the self-luminescent cyanobacterial recombinant strain Anabaena CPB4337. The combination index-isobologram equation method allows computerized

Keywords:

quantitation of synergism, additive effect and antagonism. In the Vibrio test, the fibrate

Antagonism

combinations showed antagonism at low effect levels that turned into an additive effect or

Combination index-isobologram

synergism at higher effect levels; by contrast, in the Anabaena test, the fibrate combinations

equation

showed a strong synergism at the lowest effect levels and a very strong antagonism at high

Cyanobacterium

effect levels. We also evaluated the nature of the interactions of the three fibrates with a real

Fibrates

wastewater sample in the cyanobacterial test. We propose that the combination index-

Synergism

isobologram equation method can serve as a useful tool in ecotoxicological assessment. ª 2009 Elsevier Ltd. All rights reserved.

Vibrio fischeri

1.

Introduction

Fibrates and statins (HMG-CoA reductase inhibitors) are the main lipid-lowering drugs prescribed either alone or in combination therapy in order to decrease plasma cholesterol levels and reduce the incidence of coronary heart disease. Although partially displaced by statins, the total number of fibrate prescriptions is in constant increase in the United States (Holoshitz et al., 2008). Fibric acids are the active forms

of fibrates and belong to the nuclear receptor superfamily of ligand-activated transcription factors. Gemfibrozil and fenofibrate are the fibrates currently marketed in the US, whereas bezafibrate is also available in Europe and other developed countries (Lambropoulou et al., 2008). Fenofibric acid, 2-[4-(4chlorobenzoyl)phenoxy]-2-methylpropanoic acid, is the active metabolite of fenofibrate, the inactive prodrug marketed and dispensed. Gemfibrozil, 5-(2,5-dimethylphenoxy)-2,2-dimethylpentanoic acid and bezafibrate, p-[4-[chlorobenzoylamino-

* Corresponding author. Tel.: þ34 9 1497 8176; fax: þ34 9 1497 8344. E-mail address: [email protected] (F. Ferna´ndez-Pin˜as). 1 Both authors contributed equally to this work. 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.026

428

water research 44 (2010) 427–438

ethyl]-phenoxy]-b-methylpropionic acid, are also fibric acid derivatives with similar pharmacokinetic behaviour (Miller and Spence, 1998). The occurrence of lipid regulators in the discharge of treated urban and municipal wastewater has been relatively well documented. Bezafibrate has been detected in effluents of two British STP with averages up to 230 ng/L (KasprzykHordern et al., 2009). Metcalfe et al. (2003) found around 1 mg/L of gemfibrozil in effluents of Canadian STP, whereas fenofibrate has been reported in concentrations up to 0.5 mg/L in the influent of several Brazilian STP (Stumpf et al., 1999). Andreozzi et al. (2003) found lipid regulators in the effluent of several European STP at concentrations up to 4.76 mg/L (gemfibrozil), 1.07 mg/L (bezafibrate) and 0.16 mg/L (fenofibrate). Rosal et al. (2008), reported the occurrence of bezafibrate and gemfibrozil at levels of 139 and 608 ng/L respectively in the effluent of a Spanish STP. In the same plant Rodrı´guez et al. (2008) found 165 ng/L of fenofibric acid, 61 ng/L of bezafibrate and 143 ng/L of gemfibrozil. It is also significant that removal efficiencies observed in current STP are not always high. Fent et al. (2006) reported maximum removal rates of 50–75% for fenofibric acid and gemfibrozil and somewhat higher for bezafibrate, although for the later, efficiencies below 15% have also been reported. Stumpf et al. (1999) reported a 45% removal of fenofibric acid by an activated sludge conventional treatment. KasprzykHordern et al. (2009) encountered an average degradation of bezafibrate not higher than 67%. On the other hand, Castiglioni et al. (2006) reported that the removal efficiency of bezafibrate during an activated sludge treatment greatly varied from 15% in winter to 87% in summer. At measured environmental concentrations as those reported above (mostly in the ng/L and mg/L range), many studies have shown that the risk of acute toxicity is unlikely (Fent et al., 2006; Han et al., 2006; Borgmann et al., 2007); however, there is a lack of data on chronic toxicity effects. Moreover, pharmaceuticals in the aquatic environments occur as complex mixtures from different classes, not as single contaminants (Gros et al., 2007); thus, although the concentration of individual pharmaceuticals is low, their mixture could prove ecotoxicologically significant (Brain et al., 2004). Current methods of risk assessment usually focus on the assessment of single chemicals, which may underestimate the risk associated with toxic action of mixtures; probably for this reason, in the last years there is an increasing number of studies dealing with complex mixtures of pharmaceuticals (Cleuvers, 2003, 2004; Crane et al., 2006; Han et al., 2006; Borgmann et al., 2007; Christensen et al., 2007; Pomati et al., 2008; Quinn et al., 2009). However, assessment of combined toxicities is not an easy issue. Basically, two different models are in use for the prediction of mixture toxicity, i.e., concentration addition, when pharmaceuticals have a similar mode of toxic action, and response addition or independent action, when pharmaceuticals have different modes of toxic action (Cleuvers, 2003; Teuschler, 2007). However, toxicological interactions, synergisms or antagonisms, between the pharmaceuticals and their effects can occur independently of mode of action; moreover, in most cases, the pharmacological mechanisms of action is known but the toxic mode of action may remain unknown (Cleuvers, 2003; Chou, 2006). In an effort

to overcome this limitation, we report an environmental application of a method widely used in pharmacology to interpret drug interactions; this method, termed as the median-effect/combination index (CI)-isobologram equation (Chou, 2006) allows quantitative determinations of chemical interactions where CI <1, ¼1 and >1 indicate synergism, additive effect and antagonism, respectively. One important property of the method is that previous knowledge of the mechanisms of action of each chemical is not required. Besides, the method takes into account both the potency and the shapes of the dose-effect curve of each chemical. The method has been computerized allowing an automated simulation of synergism and antagonism at different concentrations and at different effect levels of the chemicals in a mixture. The aim of our study was to assess the nature of the toxicological interactions of three fibrates, gemfibrozil, bezafibrate and fenofibric acid, by the method of combination index (CI)-isobologram equation. The three pharmaceuticals were used singly or in two- and three-drug combinations. As toxicity endpoint we have chosen the bioluminescent response of two prokaryotes, the naturally luminescent Vibrio fischeri and the recombinant bioluminescent cyanobacterium Anabaena sp. PCC 7120 CPB4337 (hereinafter, Anabaena CPB4337), both bioluminescent organisms have proved very useful in evaluating the toxicity of individual fibrates in a previous study (Rosal et al., 2009). For Anabaena CPB4337, we also evaluated the nature of the interactions of the three fibrates with a wastewater sample from a local STP, which already proved very toxic to the cyanobacterium (Rosal et al., 2009).

2.

Materials and methods

2.1.

Materials

Gemfibrozil (þ99%) and bezafibrate (þ98%) were purchased from Sigma–Aldrich. Fenofibric acid was produced from fenofibrate (Sigma–Aldrich, þ99% purity) by hydrolysis. A suspension of fenofibrate in isopropanol (30 wt.%, 400 mL) was refluxed during 4 h with an aqueous sodium hydroxide solution (2.0 M, 200 mL). After cooling to less than 70 C, a solution of hydrochloric acid (1.0 M, 325 mL) was slowly added while keeping the temperature over 60 C. The product crystallized after cooling and keeping at room temperature during 4 or more h. The product was filtered and rinsed with water and dried overnight at 60 C under nitrogen. The purity of the product was over 97% checked by HPLC. Solubility of acidic drugs in water is strongly pH dependent with few data considering this variable. Comerton et al. (2007) reported a solubility of 10.9 mg/L of gemfibrozil in water, but we could solve over 125 mg/L in 2 mM MOPS (3-[N-morpholino] propanesulfonic acid) at pH 6 and higher quantities for the pH at which V. fischeri bioassays were performed. In all cases, we avoided the use of solvents and the upper limit for the concentrations of the studied compounds was their solubility in pure water or wastewater at the pH of the bioassay. Wastewater samples were collected from the secondary clarifier of a STP located in Alcala´ de Henares (Madrid) that

429

water research 44 (2010) 427–438

receives domestic wastewater with a minor contribution of industrial effluents from facilities located near the city. This STP used a conventional activated sludge treatment and has been designed for a total capacity of 375,000 equivalent inhabitants with a maximum flow rate of 3000 m3/h. In a recent previous study (Rosal et al., 2009), we found that this wastewater was very toxic to Anabaena cells with a wastewater dilution as low as 0.11 causing 50% luminescence inhibition (wastewater EC50).

2.2.

Toxicity tests

Bioassays with the photo-luminescent bacteria Vibrio fischeri were carried out according to ISO 11348-3 standard protocol (ISO, 2007). This bioassay measures, during the prescribed incubation period, the decrease in bioluminescence induced in the cell metabolism due to the presence of a toxic substance. The bacterial assay used the commercially available Biofix Lumi test (Macherey-Nagel, Germany). The bacterial reagent is supplied freeze-dried (Vibrio fischeri NRRL-B 11177) and was reconstituted and incubated at 3 C for 5 min before use. The desired pH was set by using NaOH or HCl. The analysis media was 0.34 M NaCl (2% w/v) and tests were performed at 15 C and the measurements of light were made using a luminometer (Optocomp I). The effect of toxicants or toxicant mixtures (i.e., fibrates or fibrate combinations) was measured as percent inhibition with respect to the light emitted under test conditions in the absence of any toxic influence. Toxicity values were routinely obtained after 30 min exposure. Phenol and ZnSO4 7 H2O have been used as toxicity standards and all tests have been replicated to ensure reproducibility. The bioassays using the recombinant bioluminescent cyanobacterium Anabaena CPB4337 were based on the inhibition of constitutive luminescence caused by the presence of any toxic substance (Rodea-Palomares et al., 2009; Rosal et al., 2009). Anabaena CPB4337 was routinely grown at 28 C in the light, ca. 65 mmol photons m2 s1 on a rotary shaker in 50 mL AA/8 (Allen and Arnon, 1955) supplemented with nitrate (5 mM) in 125 ml Erlenmeyer flasks and 10 mg/mL of neomycin sulphate (Nm). Luminescence inhibition-based toxicity assays were performed as follows: 160 mL from five to seven serial dilutions of each tested toxicant or toxicant mixture (i.e.; fibrates or fibrate combinations) plus a control (ddH2O buffered with MOPS at pH 5.8) were disposed in an opaque white 96-well microtiter plates. 40 mL cells, grown as described, were washed twice and resuspended in ddH2O buffered with MOPS at pH 5.8 and were added to the microtiter plate wells to reach a final cell density at OD750 nm of 0.5. The luminescence of each sample was recorded every 5 min for up to 1 h in the Centro LB 960 luminometer. Three independent experiments with duplicate samples were carried out for all Anabaena toxicity assays. CuSO4 has been used as toxicity standard and all tests have been replicated to ensure reproducibility.

2.3.

Experimental design of fibrate combinations

Solutions of gemfibrozil (Gm), bezafibrate (Bz) and fenofibric acid (Fn) prepared as described above were used singly and in two (Bz þ Gm; Fn þ Gm; Fn þ Bz) and three (Fn þ Gm þ Bz)

combinations. Anabaena and Vibrio fischeri cells were treated with serial dilutions of each fibrate individually and with a fixed constant ratio (1:1), based on the individual EC50 values, in their binary and ternary combinations. Five dilutions (serial dilution factor ¼ 2) of each fibrate and combination plus a control were tested in three independent experiments with replicate samples. For evaluating the nature of the interaction of fibrates with wastewater, binary combinations of each fibrate plus wastewater (Fn þ WW; Gm þ WW; Bz þ WW) and a quaternary combination of the three fibrates plus wastewater (Fn þ Gm þ Bz þ WW) were also prepared and tested for Anabaena CPB4337. Anabaena cells were treated with serial dilutions of each fibrate and wastewater individually and with a fixed constant ratio (1:1), based on the individual EC50 values, in their binary and quaternary combinations. Five dilutions (serial dilution factor ¼ 2) of each fibrate and wastewater and their combinations plus a control were tested in three independent experiments with replicate samples. The experimental design is shown in Table 1. All individual fibrate, wastewater and their combination assays were carried out at the same time as recommended by Chou (2006) to maximize computational analysis of data.

2.4. Median-effect and combination index (CI)isobologram equations for determining combined fibrate interactions The results were analyzed using the median-effect/combination index (CI)-isobologram equation by Chou (2006) and Chou and Talalay (1984) which is based on the median-effect principle (mass-action law) (Chou, 1976) that demonstrates that there is an univocal relationship between dose and effect independently of the number of substrates or products and of the mechanism of action or inhibition. This method involved plotting the dose-effect curves for each compound and their combinations in multiple diluted concentrations by using the median-effect equation: fa ¼ fu

m

D Dm

(1)

Where D is the dose, Dm is the dose for 50% effect (e.g., 50% inhibition of bioluminescence or EC50), fa is the fraction affected by dose D (e.g., 0.75 if cell bioluminescence is inhibited by 75%), fu is the unaffected fraction (therefore, fa ¼ 1 fu), and m is the coefficient of the sigmoidicity of the dose-effect curve: m ¼ 1, m > 1, and m < 1 indicate hyperbolic, sigmoidal, and negative sigmoidal dose-effect curve, respectively. Therefore, the method takes into account both the potency (Dm) and shape (m) parameters. If Eq. (1) is rearranged, then: D ¼ Dm½fa=ð1 faÞ

1=m

(2)

The Dm and m values for each fibrate are easily determined by the median-effect plot: x ¼ log (D) versus y ¼ log ( fa/fu) which is based on the logarithmic form of Eq. (1). In the medianeffect plot, m is the slope and log (Dm) is the x-intercept. The conformity of the data to the median-effect principle can be ready manifested by the linear correlation coefficient (r) of the data to the logarithmic form of Eq. (1) (Chou, 2006).

Pure fibrate experiments

Fibrates plus wastewater experiments

Vibrio fischeri Dilutions

1

⁄4 (EC50) 1 ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)

Gm

(D)2 (D)1 0.4 8.75 0.8 17.5 1.6 35 3.2 70 6.4 140 Two toxicant combo (D)1 þ (D)2 (1.6:35)

Single toxicant Bz

Fn

(D)3 37.5 75 150 300 600

(D)1 (D)2 2.5 2.5 5 5 10 10 20 20 40 40 Two toxicant combo (D)1 þ (D)2 (1:1)

⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)

0.4 8.75 0.8 17.5 1.6 35 3.2 70 6.4 140 (D)1 þ (D)3 (1.6:150)

1

⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)

0.4 0.8 1.6 3.2 6.4 (D)2 þ (D)3 (35:150)

37.5 75 150 300 600

2.5 5 10 20 30* (D)2 þ (D)3 (1:5)

1

8.75 17.5 35 70 140 Three toxicant combo (D)1 þ (D)2 þ (D)3 (1.6:35:150)

37.5 75 150 300 600

0.4 0.8 1.6 3.2 6.4

37.5 75 150 300 600

1

1

⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50) 1

1

⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50) 1

8.75 17.5 35 70 140

Anabaena CPB4337

2.5 5 10 20 25* (D)1 þ (D)3 (1:5)

Gm

Single toxicant Bz (D)3 12.5 25 50 100 200

Fn

1

⁄4 (EC50) 1 ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)

1

2.5 5 10 20 25*

Gm

(D)1 (D)2 2.5 2.5 5 5 10 10 20 20 40 40 Two toxicant combo (D)1 þ (D)4 (1:0.01)

Bz

WW**

(D)3 12.5 25 50 100 200

(D)4 0.025 0.05 0.1 0.2 0.4

⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)

2.5 5 10 20 40 (D)2 þ (D)4 (1:0.01)

0.025 0.05 0.1 0.2 0.4

12.5 25 50 100 150*

1

2.5 5 10 20 40 (D)3 þ (D)4 (1:0.002)

0.025 0.05 0.1 0.2 0.4

2.5 5 10 20 40 Three toxicant combo (D)1 þ (D)2 þ (D)3 (1:1:5)

12.5 25 50 100 200

1

2.5 5 10 20 40

12.5 25 50 100 200

2.5 5 10 20 40

1

⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)

1

⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)

12.5 25 50 100 200

1

0.025 0.05 0.1 0.2 0.4

Four toxicant combo (D)1 þ (D)2 þ (D)3 þ (D)4 (1:1:5:0.01) 1/8 (EC50) ⁄4 (EC50) 1 ⁄2 (EC50) 1 (EC50) 2 (EC50) 1

1.25 2.5 5 10 20

1.25 2.5 5 10 20

6.25 12.5 25 50 100

0.0125 0.025 0.05 0.1 0.2

For the Anabaena test, the design for the experiment with the wastewater [WW (D4)] sample is also included. The experimental design is based on EC50 ratios as proposed by Chou and Talalay (1984). EC50 is the effective concentration of a toxicant which caused a 50% bioluminescence inhibition. The combination ratio was approximately equal to the EC50 ratio of the combination components (i.e., close to their equipotency ratio). *Upper maximal possible dose due to the solubility limit of fibrates in pure water. **EC50 for wastewater is the dilution which caused 50% luminescence inhibition. (D)1, (D)2 and (D)3 in mg/L, (D)4 is the dilution of wastewater in ddH2O.

water research 44 (2010) 427–438

1

Anabaena CPB4337

Single toxicant Fn

430

Table 1 – Experimental design for determining toxicological interactions of fenofibric acid [Fn (D)1], gemfibrozil [Gm (D)2], bezafibrate [Bz (D)3] and their binary and ternary combinations for Vibrio fischeri and Anabaena CPB4337 bioluminescence tests.

431

water research 44 (2010) 427–438

These parameters were then used to calculate doses of the fibrates and their combinations required to produce various effect levels according to Eq. (1); for each effect level, combination index (CI) values were then calculated according to the general combination index equation for n chemical combination at x% inhibition (Chou, 2006):

n

ðCIÞx ¼

n X

ðDÞj

j¼1

ðDx Þj

¼

P ðDx Þ1n ½Dj = n1 ½D

n X j¼1

ðDm Þj

(3)

n h io1=mj fax j = 1 fax j

where n(CI)x is the combination index for n chemicals (e.g., fibrates) at x% inhibition (e.g., bioluminescence inhibition); (Dx)1n is the sum of the dose of n chemicals that exerts x% P inhibition in combination, {[Dj]/ n1 ½D} is the proportionality of the dose of each of n chemicals that exerts x% inhibition in combination; and (Dm)j {( fax)j/[1 ( fax)j]}1/mj is the dose of each drug alone that exerts x% inhibition. From Eq. (3), CI <1, ¼1 and >1 indicates synergism, additive effect and antagonism, respectively.

2.5.

versus fa, the fraction affected by a particular dose; see Eq. (1)) and polygonograms (a polygonal graphic representation depicting synergism, additive effect and antagonism for three or more drug combinations). Linear regression analyses were computed using MINITAB Release 14 for Windows (Minitab Inc; USA).

Analysis of results

Computer program CompuSyn (Chou and Martin, 2005, Compusyn Inc, USA) was used for calculation of dose-effect curve parameters, CI values, fa-CI plot (plot representing CI

3.

Results

3.1. Toxicological interactions of fibrate combinations in Vibrio fischeri and Anabaena CPB4337 bioluminescence tests Applying the combination index-isobologram method, we evaluated the nature of gemfibrozil (Gm), fenofibric acid (Fn) and bezafibrate (Bz) interactions both in Vibrio fischeri and Anabaena CPB4337 bioluminescence tests. Table 2 shows the dose-effect curve parameters (Dm, m and r) of the three fibrates singly and their binary and ternary combinations, as well as mean combination index (CI) values of fibrate combinations. Dm was the dose required to produce the medianeffect (analogous to the EC50); Dm values for Fn were the lowest both, in Vibrio and Anabaena tests, Dm values for Gm were in the same range for both Vibrio and Anabaena while Bz

Table 2 – Dose-effect relationship parameters and mean combination index (CI) values (as a function of fractional inhibition of luminescence) of gemfibrozil (Gm), fenofibric acid (Fn), and bezafibrate (Bz) individually and of their binary and ternary combinations on Vibrio fischeri and Anabaena CPB4337 bioluminescence tests. Drug combo

Vibrio fischeri Dose-effect parameters Dm

Fn Gm Bz Gm þ Bz Fn þ Bz Fn þ Gm Fn þ Gm þ Bz

mg/L

(mM)

1.45 20.58 252.07 78.20 153.79 9.84 55.69

(4.01) (82.11) (696.46) (234.20) (424.93) (38.74) (166.69)

CI values

m

r

0.78 1.53 1.15 1.54 1.09 1.15 1.23

0.989 0.966 0.975 0.991 0.981 0.973 0.993

EC10

– – – 1.13 0.13 2.98 0.15 0.99 0.17 1.46 0.06

EC50

Add Ant Add Ant

– – – 0.97 0.04 1.71 0.03 0.75 0.05 1.01 0.02

EC90

Add Ant Syn Add

– – – 0.86 0.05 1.17 0.06 0.86 0.08 0.99 0.03

Syn Ant Syn Add

Anabaena CPB4337 Dose-effect parameters Dm mg/L Fn Gm Bz Gm þ Bz Fn þ Bz Fn þ Gm Fn þ Gm þ Bz

8.53 10.69 12.56 19.17 13.92 12.26 6.62

CI values

m

r

0.96 0.81 1.08 0.84 0.76 0.46 0.53

0.971 0.959 0.990 0.972 0.965 0.955 0.960

EC10

EC50

EC90

(mM) (23.62) (42.67) (34.70) (56.88) (38.49) (41.45) (19.45)

– – – 1.06 0.15 0.55 0.06 0.13 0.02 0.09 0.01

Add Syn Syn Syn

– – – 1.57 0.06 1.19 0.04 1.29 0.05 0.57 0.02

Ant Ant Ant Syn

– – – 2.5 0.22 2.59 0.14 12.9 2.33 3.92 0.19

Ant Ant Ant Ant

The parameters m, Dm and r are the antilog of x-intercept, the slope and the linear correlation coefficient of the median-effect plot, which signifies the shape of the dose-effect curve, the potency (EC50), and conformity of the data to the mass-action law, respectively (Chou, 1976; Chou and Talalay, 1984; Chou, 2006). Dm and m values are used for calculating the CI values (Eq. (3)); CI <1, ¼1, and >1 indicate synergism (Syn), additive effect (Add), and antagonism (Ant), respectively. EC10, EC50 and EC90, are the doses required to inhibit bioluminescence 10, 50 and 90%, respectively. Computer software CompuSyn was used for automated calculation and simulation.

432

water research 44 (2010) 427–438

Dm values were an order of magnitude higher in the Vibrio test (Rosal et al., 2009); m was the Hill coefficient used to determine the shape of the dose-response curve, hyperbolic (m ¼ 1), sigmoidal (m > 1) or negative sigmoidal (m < 1); also shown in the table, linear regression correlation coefficients (r-values) of the median-effect plots were >0.95 in all cases, indicating the conformity of the data to the median-effect principle which qualifies for further studies using this method. The Dm and m values for single fibrates and for their combination mixtures were used for calculating synergism or antagonism based on the CI Eq. (3) (Chou, 2006). Fig. 1 shows the fa-CI plot of fibrate interactions both for Vibrio (Fig. 1a) and Anabaena tests (Fig. 1b); the fa-CI plot depicts the CI value

a

3.0

Combination index, CI

2.5

2.0

1.5

1.0

0.5

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Fraction affected, fa

b

3.0

Combination index, CI

2.5

2.0

1.5

versus fa (effect level or fraction of luminescence inhibited by a fibrate singly or in combination with respect to the control) for two (Fn þ Bz; Fn þ Gm and Bz þ Gm) and three fibrate (Fn þ Gm þ Bz) combinations. The fa-CI plot is an effectoriented plot that shows the evolution of the kind of interaction (synergism, antagonism, additive effect) as a function of the level of the effect ( fa) of a particular toxicant on the reference organism ( fa, where ECa ¼ fa 100; i.e., EC10 ¼ f10 100). In the Vibrio test (Fig. 1a), the Bz þ Gm and Fn þ Gm binary combination showed a slight antagonism at very low fa values and slight synergism (Fn þ Gm) or nearly additive effects (Bz þ Gm) at the highest fa values, the Fn þ Bz combination showed a strong antagonism at low effect levels but the antagonism decreased and approached an additive kind of interaction at the highest fa levels; the ternary combination (Fn þ Gm þ Bz) showed a moderate antagonism at low fa values that also turned into a nearly additive effect at fa values above 0.4. Correlation analyses were made between CI values of the fibrate ternary combination and CI values of each of the fibrate binary combinations to determine which binary combination interaction was predominant in the ternary mixture (Table 3); the highest correlation coefficient was found for the Fn þ Bz combination (r ¼ 0.91), suggesting that this combination interaction predominated in the three fibrate mixture. The fa-CI plot of the Anabaena test (Fig. 1b) showed the opposite pattern of interactions as the three binary and the ternary combinations showed from slight to strong synergism at the lowest fa values that turned into a very strong antagonism at fa values over 0.5; the ternary combination (Fn þ Gm þ Bz) closely followed the interaction pattern of the binary Fn þ Gm combination, this is confirmed by the highest correlation coefficient found between the CI values of the ternary combination and the CI values of the Fn þ Gm combination (r ¼ 0.996) which suggests that in the Anabaena test, this particular combination seemed to be the predominant in the ternary toxicological interaction. Selected average CI values for both Vibrio fischeri and Anabaena CPB4337 tests at three representative dose levels (EC10, EC50 and EC90) and the combined effects are summarized in Table 2.

1.0

3.2. Toxicological interactions of wastewater and fibrate combinations in the Anabaena CPB4337 bioluminescence test

0.5

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Fraction affected, fa

Fig. 1 – Combination index plot ( fa-CI plot) for a set of three fibrate combinations: Fn D Bz (–6–), Bz D Gm (–B–), Fn D Gm (–,–) and Fn D Gm D Bz (–;–) for Vibrio fischeri test (a) and Anabaena CPB4337 test (b). CI values are plotted as a function of the fractional inhibition of bioluminescence ( fa) by computer simulation (CompuSyn) from fa [ 0.10 to 0.95. CI <1, [1 and >1 indicates synergism, additive effect and antagonism, respectively. At least three independent experiments with two replicates were used. The vertical bars indicate 95% confidence intervals for CI values based on sequential deletion analysis (SDA) (Chou and Martin, 2005). Fn [ fenofibric acid, Bz [ bezafibrate and Gm [ gemfibrozil.

In a recent previous study (Rosal et al., 2009), we found that a wastewater sample collected from a local STP was very toxic to Anabaena cells with a wastewater dilution of 0.11 causing 50% luminescence inhibition (wastewater EC50). The observed toxicity was attributed to the combined toxicities of over thirty micropollutants, which included fibrates as well as other pharmaceuticals (Rosal et al., 2008). We sought to investigate the nature of the interaction between the wastewater (WW) and the three fibrates in binary (Fn þ WW; Bz þ WW and Gm þ WW) and quaternary (Fn þ Gm þ Bz þ WW) combinations; for these experiments, the wastewater itself was regarded as a toxicant; the experimental design was analogous to the one for the three fibrate interactions and is also shown in Table 1. The r-values of the median-effect plots were >0.95 in all cases, indicating that the data conformed to the medianeffect principle (not shown). Fig. 2 shows the fa-CI plot for each

433

water research 44 (2010) 427–438

Table 3 – Correlation analyses between CI values of fibrate ternary and fibrate D wastewater quaternary combinations ( y) and their binary combinations (x) for Vibrio fischeri and Anabaena CPB4337 tests. Test organism

Combinations

V. fischeri

Fn þ Gm þ Bz

versus

Anabaena CPB4337

Fn þ Gm þ Bz

versus

WW þ Fn þ Gm þ Bz

versus

Regression parameters

Gm þ Bz Fn þ Bz Fn þ Gm Gm þ Bz Fn þ Bz Fn þ Gm Gm þ Bz Fn þ Bz Fn þ Gm Fn þ WW Gm þ WW Bz þ WW

xo

m

r

0.614 0.594 0.067 5.876 2.716 0.282 0.079 0.199 0.464 0.253 2.131 0.003

1.77 0.281 1.40 4.39 3.00 0.247 0.372 0.246 0.017 1.31 2.41 0.865

0.83 0.91 0.81 0.91 0.941 0.996 0.999 0.998 0.897 0.999 0.89 0.999

Fn ¼ fenofibric acid, Bz ¼ bezafibrate, Gm ¼ gemfibrozil, WW ¼ wastewater. The parameters of linear regression equations: x0 (value of y when x ¼ 0); m (slope) and r (correlation coefficient) with all p-values of 0.001. Analyses were computed using MINITAB Release 14 for Windows.

of the binary fibrate-wastewater combination and the quaternary combination; as can be observed, in a broad range of fa values, the binary combinations showed a strong synergism; however, at fa values above 0.8, the binary Fn þ WW and Bz þ WW combinations approached an additive effect and at fa values above 0.95, these two combinations yielded antagonism; by contrast, the Gm þ WW combination became even more synergistic. The quaternary combination interaction showed a strong synergism through a broad range of fa values but also turned into slight antagonism at fa values above 0.95,

Combination index, CI

2.0

1.5

1.0

0.5

0.0 0.0

0.2

0.4

0.6

0.8

1.0

Fractions affected, fa

Fig. 2 – Combination index plot ( fa-CI plot) for a set of three fibrates and toxic wastewater sample in their binary and quaternary combinations: Gn D WW (–6–), Fn D WW (–B–), Bz D WW (–,–) and Fn D Gm D Bz D WW (–;–) for the Anabaena CPB4337 test. CI values are plotted as a function of the fractional inhibition of bioluminescence ( fa) by computer simulation (CompuSyn) from fa [ 0.10 to 0.95. CI <1, [1 and >1 indicates synergism, additive effect and antagonism, respectively. At least three independent experiments with two replicates were used. The vertical bars indicate 95% confidence intervals for CI values based on sequential deletion analysis (SDA) (Chou and Martin, 2005). Fn [ fenofibric acid, Bz [ bezafibrate, Gm [ gemfibrozil, WW [ wastewater.

closely resembling the pattern of the Fn þ WW and Bz þ WW interactions which is confirmed by the highest r value (r ¼ 0.999) in the correlation analyses (Table 3), which suggests a predominant effect of Fn and Bz in the quaternary interaction. The computer software CompuSyn (Chou and Martin, 2005) displays a type of graphic termed polygonogram, which is a semiquantitative method of representing interactions between three or more compounds at a determined fa value. This graphic allows a simplified visual presentation of the overall results. Fig. 3 shows the polygonogram for the three fibrates and the wastewater at four fa values; synergism is indicated by solid lines and antagonism by broken ones; the thickness of the lines indicates the strength of the interaction. The polygonogram clearly shows the synergistic interaction of wastewater in combination with each of the three fibrates at low fa values and the antagonistic interaction that appeared at the highest fa value, 0.99, for the Fn þ WW and the Bz þ WW combinations. The same wastewater sample collected from a local STP was proved as responsible of stimulation of the bioluminescence activity of Vibrio fischeri to 110–120% of that of the control. Moreover, the EC50 values for the fibrates in the wastewater were higher than those for fibrates in pure water (Rosal et al., 2009). The same trend was observed comparing the dose-effect curve parameters (Dm, m and r) for the ternary combination (Fn þ Gm þ Bz) of fibrates in ddH2O and wastewater. The dose required to produce the median-effect (Dm) in Vibrio fischeri test when (Fn þ Gm þ Bz) were solved in wastewater was 131.936 compared to 55.6951 mg/L required when ddH2O was employed. CI values could not be calculated for Vibrio fischeri due to the fact that the wastewater itself was not toxic to this bacterium; synergism or antagonism could not be properly estimated (Chou, 2006).

4.

Discussion

The three fibrates that we have used in our study are lipid modifying agents that are effective in lowering elevated

434

water research 44 (2010) 427–438

Fig. 3 – Polygonograms showing the toxicological interactions of three fibrates and a toxic wastewater sample in their binary combinations (Fn D Bz, Bz D Gm, Fn D Gm, Gm D WW, Fn D WW, Bz D WW) as calculated by CompuSyn (Chou and Martin, 2005) for Anabaena CPB4337 test at four effect levels: fa [ 0.1 (a), fa [ 0.5 (b), fa [ 0.9 (c) and fa [ 0.99 (d). Solid lines indicate synergism, broken lines indicate antagonism. The thickness of the line represents the strength of synergism or antagonism. Figure generated by CompuSyn (Chou and Martin, 2005).

plasma triglycerides and cholesterol in humans (Staels et al., 1998). These pharmaceuticals are highly used, ubiquitous and persistent (Daughton and Ternes, 1999), they are found at ng/L to mg/L levels in many STP effluents, surface waters, estuaries of rivers and even in sea water (for a review, see Hernando et al., 2007). Although non-target organisms; the continuous release of these substances into the environment may cause acute or chronic toxicity to the aquatic biota. Regarding fibrates, in the recent literature there are many reports dealing with individual toxicity of different fibrates in a range of aquatic organisms from primary producers to consumers; a great variability has been found in the sensitivity of the different test organisms toward these pharmaceuticals (Hernando et al., 2007). However, pharmaceuticals such as fibrates do not occur singly in a polluted environment and are usually found as mixtures, therefore, for risk assessment strategies it is important to know the combined effects of pharmaceuticals in non-target organisms (Teuschler, 2007). There are two concepts widely used for the prediction of mixture toxicity: concentration addition (CA) and independent action (IA) (Backhaus et al., 2003; Vighi et al., 2003; Backhaus et al., 2004; Junghans et al., 2006). CA is used for mixtures whose components act in a similar mode of action while IA is based on the idea of dissimilar action, meaning that the compounds have different mechanisms of action; however, as discussed by Cleuvers (2003) the terms similar/

dissimilar action may be misleading. Pharmaceuticals such as fibrates may have the same pharmacological mechanism of action [i.e., interaction with the binding peroxisome proliferator-activated receptor a (PPARa)] in their target organism, humans; however, if fibrates released in the aquatic environments prove toxic to different non-target organisms, the exact mechanism of toxicity (probably different to the pharmacological mode of action) should be investigated in depth before choosing which approach, CA or IA, to use. In fact, only if toxicity is regarded as non-specific at all, the concept of CA may be used although it may also have limitations. Cleuvers (2003) found that two totally different pharmaceuticals, a fibrate and an anti-epileptic drug, followed the concept of CA in the Daphnia toxicity test and the concept of IA in an algal test; both pharmaceuticals apparently shared the same nonspecific toxic mode of action for both organisms; so it appeared that the concept of CA or IA did not depend on a similar/dissimilar mode of action but on the tested organism. The author also discussed that by definition, when using CA, substances applied below their individual noneffect concentration (NOEC) will contribute to the total effect of the mixture while when using IA, substances applied below their NOEC will not contribute to the total effect of the mixture, meaning that any combination effect will probably be higher if the substances follow the concept of CA and this may be misleading when considering the terms synergism or

water research 44 (2010) 427–438

antagonism because as also discussed by Chou (2006), synergism or antagonism may occur independently of a similar or dissimilar mode of action. In this context, Fent et al. (2006) tested mixtures of different kinds of pharmaceuticals (including fibrates) that might have estrogenic activity in a yeast reporter system; they applied the CA model and found that it had severe limitations when the dose-response curves of the individual pharmaceuticals were not identical or at low effect concentrations. As pharmaceuticals released in the environment may have such diverse dose-effect relationships, the lack of appropriate prediction suggests limitation of the CA mixtures concept. To study the nature of the combined fibrate interactions (synergism, additive effect, antagonism) for the Vibrio fischeri and Anabaena CPB4337 bioluminescence tests, we have followed the combination index (CI)-isobologram equation method of Chou (2006) and Chou and Talalay (1984); a method widely used to study drug interactions in pharmacology. This method may be considered a fractional analysis technique for drug interactions (Berenbaum, 1981; Bovill, 1998) that is independent of the mode of action and considers both the potency (EC50, Dm) and the shape (m) of the dose-effect curve for each drug. The method allows prediction of synergism/ antagonism at all effect levels ( fa) for a combination of n drugs; in contrast with the classical graphical isobologram method (Berenbaum, 1981; Bovill, 1998) that cannot be used for more than three compounds and have also graphical limitations to show all effect levels. By using this method, we have been able to determine the nature of interactions for a wide range of effect levels of three fibrates in binary and ternary combinations in two different bioluminescent organisms. However, the nature of these interactions was not uniform along the fa levels range in any of the two organisms. In Vibrio fischeri, antagonism predominated at low and intermediate fa levels but at the highest effect levels, interactions became additive or slightly synergistic. In Anabaena, a dual synergistic/antagonistic behaviour was observed with synergism predominating at fa levels below 0.4–0.5 and strong antagonism above these fa values. It is difficult to give an explanation to this phenomenon because the combination index method only allows quantitative determination of synergism or antagonism and the elucidation of the mechanism by which synergism or antagonism occurs is a separate issue that needs a different kind of approach. However, tentatively, antagonism, which could be considered the predominant interaction in Vibrio fischeri and Anabaena, might be explained by the structural similarity of fibrates which are related pharmaceuticals that share a common structural motif, a cyclic head and a hydrophobic tail (Rosal et al., 2009); at the fa levels where antagonism is found in both organisms, fibrates may compete with one another for the same target/ receptor sites. The slight synergism found at very high levels in Vibrio fischeri could perhaps be explained by the fact that at very high concentrations, fibrates may somehow combine to increase toxicity by an unspecific way of action that is probably not related to their pharmacological mechanism. Perhaps, the most puzzling interaction is the observed high synergism at very low fa levels in Anabaena; the mechanism of such synergistic interaction is not readily apparent. One could speculate that these fibrates at very low concentrations could

435

involve what Jia et al. (2009) in their extensive review of mechanisms of drug combinations call ‘‘facilitating actions’’ that means that secondary actions of one drug enhances the activity or level of another drug in the mixture or alternatively ‘‘complementary actions’’ when drugs act at the same target at different sites, at overlapping sites or at different targets of the same pathway. However, in the literature there are very few reports on possible targets of fibrates on the prokaryotic cell; English et al. (1994) reported that peroxisome proliferators such as fibrates have been shown to induce cytochrome P450BM-3 which catalyzes the hydroxylation of fatty acids, in Bacillus megaterium. Garbe (2004) reported that fibrates induced methyltransferase Rv0560c with a function in the biosynthesis of isoprenoid compounds in Mycobacterium tuberculosis; Garbe (2004) suggested that both effects may act on the plasma membrane, modulating its properties. In mitochondria, which have significant features that resemble those of prokaryotes, fibrates have been found to inhibit respiratory complex I (NDH-1 complex) and to interfere with mitochondrial fatty acid oxidation (Scatena et al., 2007). Whether fibrates may exert similar effects in Vibrio fischeri and Anabaena to those observed in Bacillus or mitochondria needs further research. In this context, we have found that, as the faCI plots show, fibrate interactions do not follow the same pattern in both bacteria, this may be due to the different origin and position in the food web of Vibrio fischeri, a heterotrophic marine prokaryote and Anabaena CPB4337, a recombinant strain of an obligate phototrophic freshwater prokaryote; in fact, Anabaena presents intracellular photosynthetic membranes called thylakoids where several functionally distinct NDH-1 complexes have been found with roles both in respiration and photosynthesis (Battchikova and Aro, 2007). If fibrates are also affecting NDH-1 complexes in Anabaena, their effects might be very different to those in Vibrio fischeri; so, although we have measured the same toxicity endpoint in both bacteria, i.e., luminescence inhibition, the combined effects of fibrates seem to depend on the test organism. Ince et al. (1999) assessed toxic interactions of heavy metals in binary mixtures on Vibrio fischeri and the freshwater aquatic plant Lemna minor and found that most binary metal mixtures exhibited only antagonistic interactions in the plant opposed to fewer antagonistic and some synergistic interactions in the heterotrophic bacterium. These authors also found that in the bacterium, the nature of the interaction (synergism or antagonism) also changed with the effect level of the binary metal combinations, although the authors did not provide a mechanistic explanation for this variability. Cheng and Lu (2002) made a comparison of joint interactions of organic toxicants in binary mixtures in Escherichia coli and Vibrio fischeri and found that toxicants with the same mechanisms of toxicity displayed mostly additive or antagonistic interactions in E. coli and Vibrio fischeri; however a synergistic interaction was found between glutardialdehyde and butyraldehyde in Vibrio. Synergistic effects in both bacteria were mostly associated with toxicants with different mechanisms of toxicity, although antagonism clearly predominated. They also found that for a total of 44 organic binary mixtures, only six mixtures resulted in identical type of interaction in both bacteria. From our results and those of other authors’ (Ince et al., 1999; Cheng and Lu, 2002; Cleuvers, 2003) one may

436

water research 44 (2010) 427–438

conclude that previous knowledge of the mechanism of toxic action of a compound is not useful enough to predict which kind of interactions it will display when combined with other toxicants with the same or different toxic mechanism; also, as we have shown, the nature of the interaction may depend on the effect level of the mixture. In addition, different types of organisms will show completely different responses to mixtures of potential toxicants. We previously found that a local wastewater was very toxic for the Anabaena CPB4337 test but non-toxic at all for the Vibrio fischeri or Daphnia magna tests. This wastewater is a mixture of over thirty micropollutants, mostly pharmaceuticals of different therapeutics groups that, besides the fibrates used in this study, included antibiotics, analgesics/anti-inflammatories, b-blockers, antidepressants, anti-epileptics/psychiatrics, ulcer healing compounds, diuretics and bronchodilators; personal care products and some priority organic pollutants are also present (Rosal et al., 2008). The method of Chou allows to combine one drug mixture with another drug mixture and determine their interactions; therefore, we studied the nature of the interaction of fibrates and wastewater in the Anabaena bioluminescence test; interestingly, we found that in a wide range of effect levels, the interaction of wastewater and the three fibrate combination was synergistic; particularly, at very low fa values which means that fibrates are at low concentrations and the wastewater is diluted several-fold, the method predicted a strong synergism; this may be due, as discussed above, to the observed synergistic interactions of fibrates with one another as well as interactions with some of the detected micropollutants when present at very low concentrations. This observed synergism may be environmentally relevant since most pharmaceuticals such as fibrates do not usually show acute toxicity on non-target organisms when tested at real environmental concentrations (Hernando et al., 2007) but in a mixture, if they act synergistically, they could prove toxic for a test organism even at low concentrations; these results agree with those found by Hernando et al. (2004) who reported synergistic toxic effects for Daphnia magna test when wastewater was spiked with environmental concentrations of several pharmaceuticals including fibrates. By contrast, our results show that at high fa values ( fa > 0.8), the combined interaction of the quaternary fibrates þ wastewater combination, the binary Fn þ WW and Bz þ WW combinations approached an additive effect and eventually became antagonistic; in our previous study, the wastewater itself decreased Anabaena bioluminescence by 84% with a lower confidence limit of 76% and an upper confidence limit of 91%; when the wastewater was spiked with increasing concentrations of each fibrate we found that, with the exception of gemfibrozil, the EC50 values for the fibrates in the wastewater were higher than those for fibrates in pure water; this was attributed either to reduced bioavailability or to antagonistic effects of fibrates with other chemicals present in the wastewater; although we did not use the method of Chou, we obtained similar results to the ones we report in this study; that is, at high effect levels (>84% luminescence inhibition) the interaction of fibrates with wastewater, except the Gm þ WW combination, showed antagonism. Based on our results, we propose that the combination index (CI)-isobologram equation, a method widely used in

pharmacology both for in vitro and in vivo bioassays, may also be applied in environmental toxicology as a general method to define interactions of potential toxicants in mixtures in any test organism and/or toxicological endpoint of interest and could be especially useful for risk assessment strategies that take into account the toxicological interactions of substances in a mixture.

5.

Conclusions

We report an environmental application of the combination index (CI)-isobologram equation to study the nature of the interactions of fibrate combinations in two bioluminescent aquatic organisms. The method allowed calculating synergism or antagonism of binary and ternary fibrate combinations at all effect levels simultaneously; we could also test the method with a real wastewater sample in binary and quaternary combination with the fibrates, finding that at very low effect levels, the fibrates acted synergistically with the wastewater in the Anabaena test. The proposed method may be used with other test organisms and/or toxicological endpoints and could be particularly useful for risk assessment approaches to toxicity of complex mixtures.

Acknowledgements The research was funded by the Spanish Ministry of Education through grants CTM2005-03080/TECNO and CSD2006-00044 and the Comunidad de Madrid, grants 0505/AMB-0395 and 0505/MB/0321.

references

Allen, M.B., Arnon, D.I., 1955. Studies on nitrogen-fixing blue green algae. I Growth and nitrogen fixation by Anabaena cylindrica Lemm. Plant Physiol 30 (4), 366–372. Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in STP effluents and their solar photodegradation in aquatic environment. Chemosphere 50 (10), 1319–1330. ˚ ., Blanck, H., Faust, M., Backhaus, T., Altenburger, R., Arrhenius, A Finizio, A., Gramatica, P., Grote, M., Junghans, M., Meyer, W., Pavan, M., Porsbring, T., Scholze, M., Todeschini, R., Vighi, M., Walter, H., Horst Grimme, L., 2003. The BEAM-project: prediction and assessment of mixture toxicities in the aquatic environment. Continental Shelf Res. 23 (17–19), 1757–1769. Backhaus, T., Arrhenius, A., Blanck, H., 2004. Toxicity of a mixture of dissimilarly acting substances to natural algal communities: predictive power and limitations of independent action and concentration addition. Environ. Sci. Technol 38 (23), 6363–6370. Battchikova, N., Aro, E.-M., 2007. Cyanobacterial NDH-1 complexes: multiplicity in function and subunit composition. Physiol. Plant 131 (1), 22–32. Berenbaum, M.C., 1981. Criteria for analyzing interactions between biologically active agents. Adv. Cancer Res. 35, 269–335. Borgmann, U., Bennie, D.T., Ball, A.L., Palabrica, V., 2007. Effect of a mixture of seven pharmaceuticals on Hyalella azteca over multiple generations. Chemosphere 66 (7), 1278–1283.

water research 44 (2010) 427–438

Bovill, J.G., 1998. Analysis of drug interactions. Baillie`re’s Clin. Anaesthesiol 12 (2), 135–168. Brain, R.A., Johnson, D.J., Richards, S.M., Hanson, M.L., Sanderson, H., Lam, M.W., Young, C., Mabury, S.A., Sibley, P.K., Solomon, K.R., 2004. Microcosm evaluation of the effects of an eight pharmaceutical mixture to the aquatic macrophytes Lemna gibba and Myriophyllum sibiricum. Aquat. Toxicol 70 (1), 23–40. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environ. Sci. Technol 40 (1), 357–363. Chen, C.-Y., Lu, C.-L., 2002. An analysis of the combined effects of organic toxicants. Sci. Total Environ 289 (1–3), 123–132. Chou, T.C., 1976. Derivation and properties of Michaelis–Menten type and Hill type equations for reference ligands. J. Theor. Biol. 59 (2), 253–276. Chou, T.C., 2006. Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacol. Rev. 58 (3), 621–681. Chou, T.C., Martin, N., 2005. CompuSyn for Drug Combinations: PC Software and User’s Guide: A Computer Program for Quantification of Synergism and Antagonism in Drug Combinations and the Determination of IC50 and ED50 and LD50 Values. ComboSyn, Inc., Paramus, NJ. Chou, T.C., Talalay, P., 1984. Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv. Enzyme Regul 22, 27–55. Christensen, A.M., Faaborg-Andersen, S., Ingerslev, F., Baun, A., 2007. Mixture and single-substance toxicity of selective serotonin reuptake inhibitors toward algae and crustaceans. Environ. Toxicol. Chem. 26 (1), 85–91. Cleuvers, M., 2003. Aquatic ecotoxicity of pharmaceuticals including the assessment of combination effects. Toxicol. Lett. 142 (3), 185–194. Cleuvers, M., 2004. Mixture toxicity of the anti-inflammatory drugs diclofenac, ibuprofen, naproxen, and acetylsalicylic acid. Ecotoxicol. Environ. Saf 59 (3), 309–315. Comerton, A.M., Andrews, R.C., Bagley, D.M., Yang, P., 2007. Membrane adsorption of endocrine disrupting compounds and pharmaceutically active compounds. J. Memb. Sci. 303 (1–2), 267–277. Crane, M., Watts, C., Boucard, T., 2006. Chronic aquatic environmental risks from exposure to human pharmaceuticals. Sci. Total Environ 367 (1), 23–41. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health Perspect 107 (6), 907–938. English, N., Hughes, V., Wolf, C.R., 1994. Common pathways of cytochrome P450 gene regulation by peroxisome proliferators and barbiturates in Bacillus megaterium ATCC14581. J. Biol. Chem. 269 (43), 26836–26841. Fent, K., Escher, C., Caminada, D., 2006. Estrogenic activity of pharmaceuticals and pharmaceutical mixtures in a yeast reporter gene system. Reprod. Toxicol 22 (2), 175–185. Garbe, T.R., 2004. Co-induction of methyltransferase Rv0560c by naphthoquinones and fibric acids suggests attenuation of isoprenoid quinone action in Mycobacterium tuberculosis. Can. J. Microbiol. 50 (10), 771–778. Gros, M., Petrovic, M., Barcelo, D., 2007. Wastewater treatment plants as a pathway for aquatic contamination by pharmaceuticals in the Ebro river basin (northeast Spain). Environ. Toxicol. Chem. 26 (8), 1553–1562. Han, G.H., Hur, H.G., Kim, S.D., 2006. Ecotoxicological risk of pharmaceuticals from wastewater treatment plants in Korea: occurrence and toxicity to Daphnia magna. Environ. Toxicol. Chem. 25 (1), 265–271. Hernando, M.D., Petrovic, M., Ferna´ndez-Alba, A.R., Barcelo´, D., 2004. Analysis by liquid chromatography-electrospray

437

ionization tandem mass spectrometry and acute toxicity evaluation for beta-blockers and lipid-regulating agents in wastewater samples. J. Chromatogr. A 1046 (1–2), 133–140. Hernando, M.D., Agu¨era, A., Ferna´ndez-Alba, A.R., 2007. LC-MS analysis and environmental risk of lipid regulators. Anal. Bioanal. Chem. 387 (4), 1269–1285. Holoshitz, N., Alsheikh-Ali, A.A., Karas, R.H., 2008. Relative safety of gemfibrozil and fenofibrate in the absence of concomitant cerivastatin use. Am. J. Cardiol 101 (1), 95–97. ¨ stu¨n, B., 1999. Ince, N.H., Dirilgen, N., Apikyan, I.G., Tezcanli, G., U Assessment of toxic interactions of heavy metals in binary mixtures: a statistical approach. Arch. Environ. Contam. Toxicol 36, 365–372. International Organization for Standardization, 2007. Water quality - Determination of the inhibitory effect of water samples on the light emission of Vibrio fischeri (Luminescent bacteria test), ISO 11348-3 Revised version, Geneva, Switzerland. Jia, J., Zhu, F., Ma, X., Cao, Z.W., Li, Y.X., Chen, Y.Z., 2009. Mechanisms of drug combinations: interaction and network perspectives. Nat. Rev. Drug Discov 8 (2), 111–128. Junghans, M., Backhaus, T., Faust, M., Scholze, M., Grimme, L.H., 2006. Application and validation of approaches for the predictive hazard assessment of realistic pesticide mixtures. Aquat. Toxicol 76 (2), 93–110. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Res. 43 (2), 363–380. Lambropoulou, D.A., Hernando, M.D., Konstantinou, I.K., Thurman, E.M., Ferrer, I., Albanis, T.A., Fernandez-Alba, A.R., 2008. Identification of photocatalytic degradation products of bezafibrate in TiO2 aqueous suspensions by liquid and gas chromatography. J. Chromatogr. A 1183 (1–2), 38–48. Metcalfe, C.D., Miao, X.S., Koenig, B.G., Struger, J., 2003. Distribution of acidic and neutral drugs in surface waters near sewage treatment plants in the lower Great Lakes, Canada. Environ. Toxicol. Chem. 22 (12), 2881–2889. Miller, D.B., Spence, J.D., 1998. Clinical pharmacokinetics of fibric acid derivatives (fibrates). Clin. Pharmacokinet 34 (2), 155–162. Pomati, F., Orlandi, C., Clerici, M., Luciani, F., Zuccato, E., 2008. Effects and interactions in an environmentally relevant mixture of pharmaceuticals. Toxicol. Sci. 102 (1), 129–137. Quinn, B., Gagne, F., Blaise, C., 2009. Evaluation of the acute, chronic and teratogenic effects of a mixture of eleven pharmaceuticals on the cnidarian, Hydra attenuata. Sci. Total Environ 407 (3), 1072–1079. Rodea-Palomares, I., Gonza´lez-Garcı´a, C., Legane´s, F., Ferna´ndezPin˜as, F., 2009. Effect of pH, EDTA, and anions on heavy metal toxicity toward a bioluminescent cyanobacterial bioreporter. Arch. Environ. Contam. Toxicol. doi:10.1007/ s00244-008-9280-9. Rodrı´guez, A., Rosal, R., Perdigo´n, J.A., Mezcua, M., Agu¨era, A., Hernando, M.D., Leto´n, P., Ferna´ndez-Alba, A.R., Garcı´aCalvo, E., 2008. In: Barcelo´, D., Petrovic, M. (Eds.), Emerging Contaminants from Industrial and Municipal Waste. SpringerVerlag, Berlin, pp. 127–175. Rosal, R., Rodrı´guez, A., Perdigo´n-Melo´n, J.A., Mezcua, M., Hernando, M.D., Leto´n, P., Garcı´a-Calvo, E., Agu¨era, A., Ferna´ndez-Alba, A.R., 2008. Removal of pharmaceuticals and kinetics of mineralization by O3/H2O2 in a biotreated municipal wastewater. Water Res. 42 (14), 3719–3728. Rosal, R., Rodea-Palomares, I., Boltes, K., Ferna´ndez-Pin˜as, F., Legane´s, F., Gonzalo, S., Petre, A., 2009. Ecotoxicity assessment of lipid regulators in water and biologically treated wastewater using three aquatic organism. Environ. Sci. Pollut. Res. doi:10.1007/s11356-009-0137-1.

438

water research 44 (2010) 427–438

Scatena, R., Bottoni, P., Botta, G., Martorana, G.E., Giardina, B., 2007. The role of mitochondria in pharmacotoxicology: a reevaluation of an old, newly emerging topic. Am. J. Physiol. Cell Physiol 293 (1), C12–C21. Staels, B., Dallongeville, J., Auwerx, J., Schoonjans, K., Leitersdorf, E., Fruchart, J.C., 1998. Mechanism of action of fibrates on lipid and lipoprotein metabolism. Circulation 98 (19), 2088–2093. Stumpf, M., Ternes, T.A., Wilken, R.D., Rodrigues, S.V., Baumann, W., 1999. Polar drug residues in sewage and natural waters in the state of Rio de Janeiro. Brazil. Sci. Total Environ 225 (1–2), 135–141.

Teuschler, L.K., 2007. Deciding which chemical mixtures risk assessment methods work best for what mixtures. Toxicol. Appl. Pharmacol 223 (2), 139–147. Vighi, M., Altenburger, R., Arrhenius, A., Backhaus, T., Bodeker, W., Blanck, H., Consolaro, F., Faust, M., Finizio, A., Froehner, K., Gramatica, P., Grimme, L.H., Gronvall, F., Hamer, V., Scholze, M., Walter, H., 2003. Water quality objectives for mixtures of toxic chemicals: problems and perspectives. Ecotoxicol. Environ. Saf 54 (2), 139–150.

water research 44 (2010) 439–448

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Toxication or detoxication? In vivo toxicity assessment of ozonation as advanced wastewater treatment with the rainbow trout Daniel Staltera,*, Axel Magdeburga, Mirco Weilb, Thomas Knackerb, Jo¨rg Oehlmanna a

Goethe University Frankfurt am Main, Biological Sciences Division, Department of Aquatic Ecotoxicology, Siesmayerstrasse 70, 60054 Frankfurt, Hessen, Germany b ECT Oekotoxikologie GmbH, Bo¨ttgerstrasse 2–4, 65439 Flo¨rsheim, Germany

article info

abstract

Article history:

Ozonation as advanced wastewater treatment method is an effective technique for

Received 23 March 2009

micropollutant removal. However, the application of this method carries the inherent

Received in revised form

danger to produce toxic oxidation byproducts. For an ecotoxicological assessment

16 July 2009

conventionally treated wastewater, wastewater after ozonation and ozonated wastewater

Accepted 18 July 2009

after sand filtration were evaluated in parallel at an operating treatment plant via the fish

Available online 25 July 2009

early life stage toxicity test (FELST) using rainbow trout (Oncorhynchus mykiss).

Keywords:

exposed to ozonated WW. This was accompanied by a significant decrease in body weight

Sewage treatment

and length compared to reference water, to the conventionally treated WW and to the

Pharmaceuticals

ozonated water after sand filtration. Hence sand filtration obviously prevents from adverse

Oxidation byproducts

ecotoxicological effects of ozonation.

The FELST revealed a considerable developmental retardation of test organisms

Vitellogenin

An additional test with yolk-sac larvae resulted in a significant reduction of vitellogenin

Emerging contaminants

levels in fish exposed to ozonated wastewater compared to fish reared in conventionally

Advanced oxidation process

treated wastewater. This demonstrates the effective removal of estrogenic activity by

Rainbow trout

ozonation.

Fish early life stage toxicity test

Adverse ozonation effects may have been a result of the conversion of chemicals into more toxic metabolites. However, sand filtration reduced toxication effects indicating that these oxidation byproducts are readily degradable or adsorbable. The results indicate that in any case ozonation should not be applied without subsequent post treatment appropriate for oxidation byproducts removal (e.g. sand filtration). ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

Wastewater (WW) is one of the major sources of micropollutants introduced into the aquatic environment (Schwarzenbach et al., 2006). The large spectrum of pharmaceuticals and personal care products (PPCPs) occurring in

surface waters (Daughton and Ternes, 1999), particularly with regard to mixture toxicity, may pose a potential threat to aquatic wildlife. But also single substances may have the capability to endanger the ecosystem as for example diclofenac can lead to an impairment of the general health condition of rainbow trout at environmentally relevant concentrations

* Corresponding author. Tel.: þ49 69 7982 4882; fax: þ49 69 7982 4748. E-mail address: [email protected] (D. Stalter). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.025

440

water research 44 (2010) 439–448

Nomenclature AOC C COD DOC EE2 ELISA EQS FELST FS GJIC MS222 NH4–N

assimilable organic carbon control chemical oxygen demand dissolved organic carbon 17a-ethinylestradiol enzyme linked immunosorbent assay environmental quality standard fish early life stage toxicity test wastewater after final sedimentation gap junctional intercellular communication tricaine methanesulfonate ammonium nitrogen

(Schwaiger et al., 2004; Triebskorn et al., 2004). 17a-ethinylestradiol (EE2) is considered to be the most potent estrogenic active compound for fish with the lowest observed effect concentration of 0.1 ng/L for vitellogenesis in rainbow trout (Purdom et al., 1994). Moreover it leads to a reduced egg fertilisation success and skewed sex ratio toward females of fat head minnows at a LOEC of 0.32 ng/L (Parrott and Blunt, 2005). Furthermore, industrial surfactants like alkylphenolic ethoxylates are likely to be responsible for feminization effects in rainbow trout (Jobling et al., 1996) and complex mixtures of estrogenic chemicals present in the environment have been shown to act additively (Brian et al., 2005). The concentrations of such compounds in WW often exceed effect concentrations (Ternes et al., 1999; Ying et al., 2002; Tixier et al., 2003) and thus even treated WW contains high amounts of estrogenic substances that impair the endocrine system of exposed fish (Larsson et al., 1999; Rodgers-Gray et al., 2001) resulting in feminization effects of males and consequently reduced fertility of field populations (Jobling et al., 2002). Besides, in the European Union the reduction of the contamination of surface waters with hazardous substances is defined by the Water Framework Directive requiring a ‘‘good status’’ for all coastal and inland waters until 2015 (European Commission, 2000). One tool is the implementation of environmental quality standards (EQSs) for mono substance pollutants exhibiting a significant risk to the aquatic ecosystem. The discharge of such compounds, classified as priority substances, is envisaged to be progressively reduced by 2025 or 5 years after inclusion in the list for priority substances, respectively. However, till now so called ‘‘emerging contaminants’’ are only marginally addressed and PPCPs are not included in this list. But the latter has to be reviewed at least every 4 years and provisions have been made to add several hazardous emerging contaminants. Amongst others the inclusion of diclofenac, EE2 and carbamazepine is discussed and additionally recommended by the European Parliament since 2007 (Jensen, 2007). Moreover EQSs meanwhile have been proposed for several PPCPs based on reliable ecotoxicity data (e.g. diclofenac: 0.1 mg/L, EE2: 0.03 ng/L, carbamazepine: 0.5 mg/L; Jahnel et al., 2006; Moltmann et al., 2007). Unfortunately, some PPCPs exceed the EQSs in surface waters of some urbanised

NO3–N O OECD OS P PAH PAC PPCPs rcf SPM VTG WW WWTP

nitrate nitrogen wastewater after ozonation Organisation for Economic Cooperation and Development wastewater after ozonation and sand filtration phosphate polycyclic aromatic hydrocarbons powdered activated carbon pharmaceuticals and personal care products relative centrifugal force suspended particulate matter vitellogenin wastewater wastewater treatment plant

regions. That is, inter alia, a result of the fact that conventional WW treatment plants are not designed for appropriate removal of PPCPs (Daughton and Ternes, 1999). Nevertheless in many countries surface waters serve as raw water resources for drinking water supply. Due to the public’s demand for safe drinking water and ecosystem health advanced WW treatment methods are required to allow for sufficient removal of micropollutants from sewage treatment effluents. Therefore, end of pipe techniques like activated carbon filtration and ozonation of WW might be crucial to achieve regulative requirements in a medium-term perspective. Huber et al. (2005) demonstrated that ozonation of secondary effluent is an effective tool for the removal of a wide range of pharmaceuticals among them diclofenac, carbamazepine and the estrogens estradiol, estrone and EE2, with degradation rates of more than 90% for ozone doses of 2 mg/L. Merely iodinated X-ray contrast media and a few acidic pharmaceuticals were oxidized only partially. Nakada et al. (2007) confirmed the effective elimination of 22 pharmaceuticals during ozonation in an operating treatment plant. A further advantage of the ozonation is the sanitizing property of the method (Tyrrell et al., 1995). The effective removal of micropollutants and pathogens indicates that ozonation is a suitable technique for advanced WW treatment to reduce the contamination of the aquatic environment. Actually a reduced toxicity of a mixture of six pharmaceuticals treated by ozonation towards algae and rotifer was demonstrated by Andreozzi et al. (2004). Reduced estrogenic activity of an EE2-containing WW after ozonation was shown by Huber et al. (2004) and ozone treatment of WW significantly decreased the toxic potency of urban pollutants to the freshwater mussel Elliptio complanata (Gagne et al., 2007). However, ozone treatment typically transforms chemical compounds but does not mineralize them entirely. Consequently, not only the concentrations of mother compounds but also the inclusion of transformation products and their mixtures should be assessed for toxicity, too. For a risk-benefit analysis an extensive ecotoxicological evaluation of ozonated WW is essential. In vitro bioassays covering different modes of toxic action (e.g. estrogenicity, acetylcholine esterase activity or non-specific baseline toxicity) and performed with enriched water samples are

water research 44 (2010) 439–448

2.2.

Table 1 – Wastewater quality summary (mg/L). SPM

COD

P

NH4–N

NO3–N

pH

0.1 0.04 0.27

5.4 10.1

8.3 8.12 8.4

Median 10% percentile 90% percentile

Final sedimentation 4.8 17 0.19 3.6 15 0.17 8.96 23.8 0.22

n

37

Median 10% percentile 90% percentile

Ozonation and 2 15 1.6 12.2 2.4 19.8

sand filtration 0.17 0.04 0.14 0.04 0.19 0.15

9.8 7.28 12.48

n

37

37

37

23

23

37

37

37

2

23

8.3 8.02 8.4 23

suitable tools for toxicity characterization of WW as shown by Escher et al. (2008). These methods were shown to be highly sensitive even for identification and characterization of low toxic water samples. Nevertheless, as typical screening tools they are not designed to replace chronic in vivo tests of whole effluent samples. To allow for a more comprehensive and integrative assessment of the potentially hazardous impact of WW ozonation we applied the fish early life stage toxicity test (FELST) with rainbow trout (Oncorhynchus mykiss).

2.

Materials and methods

Ozonated and conventionally treated WW were tested in parallel on site at the wastewater treatment plant (WWTP) Wu¨eri (Regensdorf, Switzerland) with the fish early life stage toxicity test (FELST) in a flow-through system.

2.1.

Characterization of the wastewater treatment plant

The municipal WWTP Wu¨eri operates experimentally with a full scale ozonation reactor after final sedimentation and with a sand filtration step after the ozone reactor. Table 1 shows the WW quality parameters. The water parameters after the final sedimentation and after the ozone reactor are on the same level and therefore exemplarily shown for final sedimentation water. Low ammonium and phosphate concentrations indicate that the treatment plant is already working well. The dissolved organic carbon (DOC) ranged from 5.4 to 5.9 mg/L and the pH was nearly constant in all test waters. The WWTP serves for a population equivalent of 25,000. The median discharge in the experimental period was 6190 m3/day (10th percentile: 4430 m3/day, 90th percentile: 10,500, n ¼ 109) and the applied ozone concentration was in a range between 0.4 and 1 mg O3/mg DOC.

441

Experimental setup

WW from three different sampling points of serial treatment steps was tested (Fig. 1): after the final sedimentation (FS), after the ozone reactor (O) and after additional sand filtration (OS). Test waters were passively transported through high-grade steel pipes to aerated high-grade steel reservoir tanks. The retention time in the pipes was adjusted to be at least 45 min to avoid ozone residuals reaching the exposure vessels. Indeed no ozone was detected by indigo blue method (Bader and Hoigne, 1981) in the reservoir tank during maximum required flow-through and during maximum applied ozone concentration (1 mg O3/mg DOC). From reservoir tanks test waters were transported through polytetrafluoroethylene tubes via a peristaltic pump (IPC24, Ismatec, Wertheim– Mondfeld, Germany) to the exposure vessels each equipped with a passive discharge device and tempered using a temperature-controlled water bath. Constant temperature conditions in the water bath were achieved with a flowthrough cooling system (Van der Heijden, Do¨rentrup, Germany). The flow-through rates in the exposure vessels ranged from 11 mL per minute up to 44 mL per minute (2–8 fold water exchange per day in the exposure vessels) depending on the fish size to match the loading rate criteria (OECD, 1992b). Reconstituted water according to OECD guideline 203 (1992a) was used as control water (C). The FELST was performed with the rainbow trout (Oncorhynchus mykiss) according to OECD guideline 210 (1992b) with a constant water temperature of 10 2 C as well as darkness for embryo development and 12 2 C with 12/12 h light/dark photoperiod post hatch. Sixty newly fertilized eggs per replicate were exposed to the test waters for 65 days in high-grade steel 10 L tanks. One test series was performed with unfiltered WW and a second with membrane filtered WW (pore size: 0.4 mm, Kubota Corp., Osaka, Japan) to minimize microbial impacts. Macromolecules and organic compounds were not retained. However, it has to be considered that membrane filtration additionally removes suspended particulate matter and consequently all particle bound pollutants. The filter membranes were placed in the reservoir vessels. A third test was performed with yolk-sac fry (5 days post hatch, 30 larvae per exposure vessel) and non-filtered test waters because we postulated a reduced sensitivity to pathogen contamination of the larvae compared to the egg stage. The test duration was 64 days. All tests were performed with undiluted WW to increase the probability to detect differences between the treatment groups. With the beginning of swim up (the swim up process marks a developmental transition from larval stage to juvenile fish stage and is characterized with the

Fig. 1 – Sampling points at the wastewater treatment plant. Abbreviations: FS, after final sedimentation; O, after the ozone reactor; OS, after sand filtration.

442

water research 44 (2010) 439–448

beginning of exogenous ingestion) the fish were fed four times per day (trout starter, 4% body weight per day). The tests were run with four (C) and three (FS, O, OS) replicates in the first FELST with unfiltered WW and two (C, OS) and three replicates (FS, O) per test water in the remaining two tests. The latter were performed in parallel and therefore it was not possible to use consistently three replicates per WW-group due to limited space capacities. In each test replicates were placed randomized in the water bath. Observations on egg coagulation, hatching, mortality, swim up, malformation and abnormal behaviour were recorded daily. Fish were humanely killed by MS222 (tricaine methanesulfonate, Sigma– Aldrich, St. Louis, USA) overdose. Individual fish were blotted dry, weight and body length were determined. Afterwards fish were frozen in liquid nitrogen and stored at 80 C until vitellogenin detection in whole body homogenates.

2.3.

Vitellogenin detection

Whole body homogenates of 11 fish (from the third test with yolk-sac fry) per replicate were prepared as described by Holbech et al. (2006) with slight modifications. Aliquots of 0.3 g frozen fish, excised between head and pectoral fin, were mixed with 10 fold of the body weight of homogenisation buffer (50 mM Tris–HCl pH 7.4; 1% protease inhibitor cocktail (P 8340, Sigma–Aldrich, St. Louis, USA)) and homogenated with a dispersing apparatus (T18 basic Ultra–Turrax, IKA, Staufen, Germany). The homogenate was centrifuged for 30 min at 20,000 rcf and the supernatant was used for vitellogenin analysis. Vitellogenin (VTG) was detected with a rainbow trout vitellogenin ELISA test kit (Biosense, Bergen, Norway) using a 1:20 dilution.

2.4.

Statistical analysis

Complete statistical analyses (Kruskal–Wallis with Dunn’s post test, Fisher’s exact test, non-linear regression with variable slope) were performed using GraphPad Prism version 5.0 for windows (GraphPad software, San Diego, CA, USA). All given error values indicate the standard error (SE) and in all figures error bars display the SE. Kruskal–Wallis with Dunn’s post test was chosen to test on significant differences because data were not normally distributed in all cases. For quantal data Fisher’s exact test was applied (mortality, swim up, coagulation). When toxicity endpoints could be analysed individually for each test animal (biomass, body length, vitellogenin), the collected data are presented and statistically evaluated on a per specimen basis.

3.

Results

3.1.

FELST with unfiltered wastewater

Unfiltered WW caused an increased coagulation rate of the exposed eggs in the FELST (Fig. 2). The eggs exposed to WW after final sedimentation (FS) were completely coagulated after 18 days with a 50% coagulation time of 12.1 days. The coagulation of the eggs exposed to WW after the ozone reactor (O) was considerably delayed compared to FS. However the

Fig. 2 – Cumulative coagulation of Oncorhynchus mykiss eggs (mean values ± SE) exposed to differently treated wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 40 days all treatment groups differ significantly from each other (Fisher’s exact test: p < 0.001; n [ 240 (C), 180 (FS, O, OS)).

coagulation after 40 days achieves still an average rate of 87.2 4.8% with a 50% coagulation time of 17.6 days. The lowest coagulation rate in the WW treatments occurred after sand filtration (OS; 64.4 4.0% after 40 days; 50% coagulation time: 25.8 days). After 10–15 days exposure, fungus mycelia (first appearing in the FS vessels) were observed in all WW exposure vessels. Mycelia were found on and between eggs as well as vorticellas on the eggs whereas the reference water (C) remained observably free from mycelia and vorticellas. Hatching success in the control group (53.4 5.2%) did not meet validity criteria (>66%; OECD guideline 210).

3.2.

FELST with membrane filtered wastewater

To exclude microbial impairment the second FELST was performed with membrane filtered WW and eggs were obtained from another fish hatchery. All test vessels remained free from microbial contamination throughout the test duration. The egg coagulation rates in the WW treatment groups of this experimental series were reduced compared to the first FELST with a maximum of 25% (O, OS) and only 20.1 1.1% after FS (Fig. 3A). Egg coagulation was significantly increased and hatching success significantly decreased in the WW treatment groups compared to the control ( p < 0.05, Fisher’s exact). The coagulation rates were slightly but not significantly increased in O and OS compared to FS (Fig. 3A). The hatching progress was slightly delayed in O compared to FS and OS but hatching success achieved at least 75% in all treatments (Fig. 3B). The control group met the validity criteria according to OECD guideline 210 (egg coagulation: 10.0 1.7%, hatching success: 90.0 1.7%).

water research 44 (2010) 439–448

443

Fig. 3 – Cumulative coagulation of eggs (A) and cumulative hatching of larvae (B) from Oncorhynchus mykiss (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 40 and 36 days, respectively all wastewater treatment groups are significantly different from the control (Fisher’s exact test: p < 0.01–0.05; n [ 120 (C, OS), 180 (FS, O)).

Fig. 4 shows the cumulative swim up of hatched fish beginning after 45 days of exposure. The swim up is considerably delayed in all WW treatment groups compared to the control. This effect is most notable in O, even if compared to

FS and OS. At the end of the experiment only 83.3 1.6% of the fish swam up in O while 100% swam up in C and FS and 97.6 2.4% in OS. Hereby the swim up success after 64 days is significantly decreased in O compared to the other treatment groups ( p < 0.01, Fisher’s exact). The 50% swim up time in the control is 47.4 1.0 days, which is only slightly increased in FS and OS (50.2 1.0; 49.3 1.0) but obviously increased in O (57.5 1.04). The biomass as well as the body length of fish is significantly decreased ( p < 0.001, Kruskal–Wallis with Dunn’s post test) in all WW treatments compared to the control (Fig. 5). Both endpoints in fish exposed to O are furthermore significantly decreased compared to the FS group ( p < 0.001) and significantly decreased compared to OS ( p < 0.05). Generally the mortality is comparatively low in all WW treatment groups (Fig. 6), as they are fulfilling the validity criteria for controls according to OECD guideline 210 (survival after hatch 70%). Nevertheless the mortality in the WW treatment groups is slightly increased compared to C. The highest and significantly increased mortality rate was detected in the ozonated water (24.0 4.0%, p < 0.05, Fisher’s exact).

3.3. Fig. 4 – Cumulative swim up of Oncorhynchus mykiss larvae (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 64 days the O group differs significantly from the other treatment groups (Fisher’s exact test: p < 0.001–0.01; n [ 95 (C), 116 (FS), 103 (O), 75 (OS)).

Fish test starting with yolk-sac fry & vitellogenesis

The fish test starting with the yolk-sac stage revealed no statistically significant differences in development between WW treatments and the control. The biomass exhibited no major deviations between exposure groups (<6% deviation from the control). Mortality was very low with the highest value in the control of 3.3 0.2% and a mortality of 0% in O (FS: 2.5 2.5%, OS: 1.7 1.7%). The vitellogenin content was significantly increased in specimens of the FS group (67.3 26.9 ng/mL; p < 0.05,

444

water research 44 (2010) 439–448

Fig. 5 – Biomass (A) and body length (B) in percentage relative to the control (0.28 ± 0.01 g and 34.1 ± 0.2 mm, respectively) of Oncorhynchus mykiss (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Significant differences to the control are indicated with white asterisks, between treatments with black asterisks (Kruskal–Wallis with Dunn’s post test: +, p < 0.05; +++, p < 0.001; n [ 73–114).

Kruskal–Wallis with Dunn’s post test) compared to fish maintained in reference water (11.0 6.6 ng/mL) whereas it was significantly decreased in the O and OS groups (4.6 1.9 and 4.8 1.9 ng/mL, respectively; p 0.01) compared to FS (Fig. 7).

4.

Discussion

4.1.

FELST with unfiltered wastewater

presumably are a result of microbial contamination as the reference water (C) remained free from mycelia and vorticellas whereas in the WW treatment groups fungi mycelia and vorticellas were found after 10–15 days exposure. However, the hatching success in the control (53.4 5.2%) did not comply with the validity criteria (>66%) according to OECD guideline (1992b) indicating that egg quality was not sufficient to run a valid test. The disinfectant effect of ozonation (Tyrrell et al., 1995) is likely to be responsible for the delayed egg

Unfiltered WW led to an increased coagulation rate of the exposed eggs in the FELST (Fig. 2). High coagulation rates

Fig. 6 – Mortality of Oncorhynchus mykiss post hatch (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Significant differences to control are indicated with white asterisks (Fisher’s exact test: +, p < 0.05; n [ 108 (C), 143 (FS), 135 (O), 90 (OS)).

Fig. 7 – Whole body vitellogenin concentration of Oncorhynchus mykiss (mean values ± SE) after 60 days exposure to differently treated wastewaters starting with the yolk-sac stage. Abbreviations: FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; C, control water; SE, standard error. Significant differences to control are indicated with white asterisks, between treatments with black asterisks (Kruskal–Wallis with Dunn’s post test: +, p < 0.05; ++, p < 0.01; n [ 22 (C, OS) – 33 (FS, O)).

water research 44 (2010) 439–448

coagulation in the O group compared to FS. Ozonation is also known to produce high amounts of assimilable organic carbons (AOC) which may have allowed a fast regrowth of microorganisms (Huang et al., 2005), resulting in still high coagulation rates in the O group. Sand filtration reduces the amount of suspended particulate matter (SPM) and of AOC (Wang and Summers, 1996). This may have reduced microbial development and resulting coagulation rates.

4.2.

FELST with membrane filtered wastewater

Exposure to membrane filtered WW revealed a distinct delay of the swim up process in the O group compared to FS and OS (Fig. 4) accompanied by a significant decrease in body weight and body length (Fig. 5). The reduced biomass and body length after ozonation is most likely a result of the delayed development because fish start exogenous ingestion at the onset of the swim up process. Consequently trout that swim up later may suffer from developmental disadvantages. Oxidative byproducts in ozonated WW may have impeded embryonic and/or larval development of fish. Possibly because sand filtration is able to remove ozonation metabolites (e.g. aldehydes, glyoxal, AOC; Wang and Summers, 1996) the effect in OS was reduced. However, no single compound could be clearly identified for the retardation effect. Possibly the sum of aldehydes (e.g. formaldehyde, glyoxal, acetaldehyde), carboxylic acids (e.g. formate), ketones and brominated organic compounds formed due to ozonation (Huang et al., 2005; Wert et al., 2007) caused these effects. Unfortunately, no chronic toxicity data for juvenile rainbow trout are available in literature for these compounds. Nevertheless, Wang and Summers (1996) were able to demonstrate that sand filtration efficiently removes aldehydes produced during ozone application, supporting this assumption. Besides, the formation of more complex organic metabolites evoked by ozonation may result in an increased toxicity compared to chemical precursors as it has already been documented for polycyclic aromatic hydrocarbons (PAHs). In the experiments of Luster-Teasley et al. (2002, 2005) chrysene and pyrene and byproducts as a result of ozonation were examined for their ability to disrupt gap junctional intercellular communication (GJIC), an indicator for tumor promoting properties. Among the transformation products, aldehydic compounds exhibited an increased toxicity compared to the precursor substance while their carboxylic structure analogue showed no GJIC disrupting activity. For the antiepileptic drug carbamazepine McDowell et al. (2005) identified three new oxidation products of unknown toxicity after ozone treatment. Furthermore, a possible increase in mutagenicity after ozonation of WW, as observed by Monarca et al. (2000) with the Ames test, verifies the potential of ozonation to produce toxic oxidation byproducts. In this context conversion of the widely used fungicide tolylfluanide into the carcinogen N-nitrosodimethylamine during ozonation was discovered by Schmidt and Brauch (2008). Based on these results it is likely that ozonation of WW leads to an increased number of unknown and potentially toxic metabolites depending on the composition of WW at the outset and the post treatment (e.g. sand filtration). Petala et al. (2006) demonstrated with the Vibrio fischeri bioluminescence test,

445

that toxicity originating from ozonated WW decreases with increasing storage time, indicating a rapid decomposition of toxic metabolites. In spite of the developmental retardation and the reduced biomass no significant differences in mortality rates were observed between the WW treatment groups (Fig. 6). Mortality was highest directly after ozonation (O, 24.0 4.0%). Considering the developmental retardation and the decreased biomass of fish exposed to ozonated water the increased mortality of O group specimens compared to FS and OS is noticeable and, although not significant, potentially a result of a reduced fitness of the fish in the ozonated water. It might be assumed that the retarded development is a consequence of an unspecific and general impairment of the fish’s health condition. This may increase their sensitivity towards environmental and anthropogenic stressors, thus leading to an increased mortality. Furthermore, under field conditions the delayed development possibly increases the risk for the fish to fall prey to predators since before swim up the larvae rest on the bottom, not capable to abscond effectively. Based on the assumption that the compounds causing adverse effects after ozonation are readily degradable it is possible that the detoxication after the ozone reactor will occur in the river as well. However, the discharge of ozonated WW without sand filtration would bear the risk to endanger fish populations in a considerable range of the receiving river, depending on WW load and flow velocity. Therefore, ozonation should always be followed by a sand filtration and further studies should focus on whether sand filtration is capable to remove oxidation byproducts sufficiently.

4.3.

Fish test starting with yolk-sac fry & vitellogenesis

The fish test with non-filtered WW starting with the yolk-sac stage resulted in negligible and insignificant deviations of development and biomass between treatment groups. Ozonation had no effect on these toxicity endpoints. These results indicate that yolk-sac fry 5 days post hatch are less sensitive to ozonation metabolites, compared to fish embryos and newly hatched fish, respectively. Nevertheless, the amount of suspended particulate matter in the exposure vessels and in the Teflon tubes was accompanied by an increased biofilm formation. This may have contributed to an increased detoxication of the ozonated WW due to (bio-) degradation of oxidation byproducts. The significant increase of vitellogenin content in fish exposed to FS compared to the reference water indicates an environmentally relevant contamination of the WW with estrogenic active compounds (Fig. 7). After ozonation the VTG content decreased even below the control level. The formation of antiestrogenic compounds is not likely because the VTG decrease compared to control level is not significant and phenols, as an important functional group interacting with the estrogen receptor (Nishihara et al., 2000), are particularly susceptible to ozone attack (von Gunten, 2003). These results confirm the high efficiency of ozonation to eliminate estrogenic contamination in WW as already shown by Huber et al. (2004) with an in vitro test system. Based on these results it can be assumed that ozonation is well suited to reduce the estrogenic burden of WW below environmental relevance.

446

water research 44 (2010) 439–448

Regarding ecosystem health it should be considered that estrogen and xenoestrogen concentrations detected in the environment have been shown to impair the sustainability of wild fish populations (Kidd et al., 2007), arguing for the establishment of an end of pipe technique for estrogen removal to protect fish populations in surface waters receiving high amounts of WW.

4.4.

Ozonation vs. alternatives

Currently there is a controversy regarding the appropriate advanced WW treatment techniques to reduce contamination of the aquatic ecosystem with micropollutants. Alternative processes other than ozonation also deliver promising results with regard to the removal of contaminants. Wastewater treatment with powdered activated carbon (PAC) achieves reduction rates of 75–90% for pharmaceuticals (including X-ray contrast media) with PAC dosages of 10–20 mg/L (Pu¨ttmann et al., 2008). The main advantage of PAC treatment is that a reduced chemical concentration is equivalent to removal of chemicals whereas ozonation leads basically to a transformation of the compounds. According to cost estimations PAC treatment is expected to be about 30% more expensive than ozonation (Joss et al., 2008). However, the main disadvantage of PAC treatment might be the need for the disposal of the used and contaminated carbon. Membrane filtration is suitable for micropollutants retention but as a result of considerably higher requirement for energy and technical equipment economically not competitive with ozonation or activated carbon treatment (Joss et al., 2008). Overall, end of pipe techniques are presumably a suitable solution to reduce toxicity of hazardous WW in a mediumterm perspective. Nevertheless in the long term source control strategies such as wastewater separation (e.g. urine separation), ecologically correct disposal of drugs by end users, reuse or recycling by the pharmaceutical industry or alternative medical treatments to drug therapies (Daughton, 2003; Joss et al., 2006) could offer environmentally friendly and affordable options.

5.

Conclusions

The sanitizing effect of ozonation was confirmed. Coagulation rates of eggs exposed to ozonated wastewater were on a lower level compared to those exposed to conventionally treated wastewater. Disinfection is presumably only efficient when linked to sand filtration because of rapid recovery of microorganisms due to increased assimilable organic carbon formation as a result of ozonation. Membrane filtered wastewater (for removal of microorganisms) reveals developmental retardation directly after ozonation. After sand filtration this adverse effect disappears. The reduced biomass and body length in fish exposed to ozonated wastewater is most probably a result of the formation of toxic oxidation byproducts. The mortality in the ozonated wastewater was significantly increased as a result of retarded development. Impairment of the fish’s health condition may increase the sensitivity towards environmental and anthropogenic stressors.

Developmental retardation might increase the risk for the fish to fall prey to predators because swim up stage is delayed. Ozonation should not be applied without appropriate barrier for oxidation byproducts. Effectiveness of sand filtration for removal of oxidation byproducts should be further evaluated. Reduction of vitellogenin content in fish exposed to ozonated wastewater on control level confirms the suitability of this technique to reduce estrogenic activity, possibly below environmental relevance.

Acknowledgements The authors should like to express their gratitude to the staff from WWTP Wu¨eri for their technical cooperation and Adriano Joss from the EAWAG for helpful suggestions and technical support. Ulrike Schulte-Oehlmann is acknowledged for critical comments and helpful suggestions on the manuscript. This study was part of the EU project Neptune (contract no 036845, SUSTDEV-2005-3.II.3.2) within the Energy, Global Change and Ecosystems Programme of the Sixth Framework (FP6-2005-Global-4) and co-funded by Bundesamt fu¨r Umwelt (BAFU), Bern (CH) within the Strategy MicroPoll Programme (contract no 05.0013.PJ/F471-0916).

references

Andreozzi, R., Campanella, L., Fraysse, B., Garric, J., Gonnella, A., Lo Giudice, R., Marotta, R., Pinto, G., Pollio, A., 2004. Effects of advanced oxidation processes (AOPs) on the toxicity of a mixture of pharmaceuticals. Water Science and Technology 50 (5), 23–28. Bader, H., Hoigne, J., 1981. Determination of ozone in water by the indigo method. Water Research 15 (4), 449–456. Brian, J.V., Harris, C.A., Scholze, M., Backhaus, T., Booy, P., Lamoree, M., Pojana, G., Jonkers, N., Runnalls, T., Bonfa, A., Marcomini, A., Sumpter, J.P., 2005. Accurate prediction of the response of freshwater fish to a mixture of estrogenic chemicals. Environmental Health Perspectives 113 (6), 721–728. Daughton, C.G., 2003. Cradle-to-cradle stewardship of drugs for minimizing their environmental disposition while promoting human health. II. Drug disposal, waste reduction, and future directions. Environmental Health Perspectives 111 (5), 775–785. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107, 907–938. Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S., Vermeirssen, E.L.M., 2008. Monitoring of the ecotoxicological hazard potential by polar organic micropollutants in sewage treatment plants and surface waters using a mode-of-action based test battery. Journal of Environmental Monitoring 10 (5), 622–631. European Commission, 2000. Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy. Official Journal of the European Communities L327. Retrieved from: http://eur-lex.europa.eu/

water research 44 (2010) 439–448

LexUriServ/LexUriServ.do?uri¼OJ: L:2000:327:0001:0072:EN: PDF [12.02.2009]. Gagne, F., Andre, C., Cejka, P., Gagnon, C., Blaise, C., 2007. Toxicological effects of primary-treated urban wastewaters, before and after ozone treatment, on freshwater mussels (Elliptio complanata). Comparative Biochemistry and Physiology C – Toxicology & Pharmacology 145 (4), 542–552. Holbech, H., Kinnberg, K., Petersen, G.I., Jackson, P., Hylland, K., Norrgren, L., Bjerregaard, P., 2006. Detection of endocrine disrupters: evaluation of a fish sexual development test (FSDT). Comparative Biochemistry and Physiology C – Toxicology & Pharmacology 144 (1), 57–66. Huang, W.J., Fang, G.C., Wang, C.C., 2005. The determination and fate of disinfection by-products from ozonation of polluted raw water. Science of the Total Environment 345 (1-3), 261–272. Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of estrogenic activity and formation of oxidation products during ozonation of 17a-ethinylestradiol. Environmental Science & Technology 38 (19), 5177–5186. Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D., McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science & Technology 39 (11), 4290–4299. Jahnel, J., Neamtu, M., Schudoma, D., Frimmel, F.H., 2006. Scientific risk assessment of considered water relevant substances (Bestimmung von Umweltqualitaetsnormen fuer potenziell gewaesserrelevante Stoffe). Acta Hydrochimica et Hydrobiologica 34 (4), 389–397. Jensen, G.K., 2007. European Parliament increases number of chemicals earmarked for priority action by European Commission. The Lancet Oncology 8 (5), 382–383. Jobling, S., Sheahan, D., Osborne, J.A., Matthiessen, P., Sumpter, J.P., 1996. Inhibition of testicular growth in rainbow trout (Oncorhynchus mykiss) exposed to estrogenic alkylphenolic chemicals. Environmental Toxicology and Chemistry 15 (2), 194–202. Jobling, S., Coey, S., Whitmore, J.G., Kime, D.E., Van Look, K.J.W., McAllister, B.G., Beresford, N., Henshaw, A.C., Brighty, G., Tyler, C.R., Sumpter, J.P., 2002. Wild intersex roach (Rutilus rutilus) have reduced fertility. Biology of Reproduction 67 (2), 515–524. Joss, A., Klaschka, U., Knacker, T., Liebig, M., Lienert, J., Ternes, T.A., Wennmalm, A., 2006. Source control, source separation. In: Ternes, T.A., Joss, A. (Eds.), Human Pharmaceuticals, Hormones and Fragrances. The Challenge of Micropollutants in Urban Water Management. ISBN 1843390930, pp. 353–384. Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade wastewater treatment for removing organic micropollutants? Water Science and Technology 57 (2), 251–255. Kidd, K.A., Blanchfield, P.J., Mills, K.H., Palace, V.P., Evans, R.E., Lazorchak, J.M., Flick, R.W. (2007). Collapse of a fish population after exposure to a synthetic estrogen. Proceedings of the National Academy of Sciences of the United States of America 104(21), 8897–8901. Larsson, D.G.J., Adolfsson-Erici, M., Parkkonen, J., Pettersson, M., Berg, A.H., Olsson, P.E., Forlin, L., 1999. Ethinylestradiol – an undesired fish contraceptive? Aquatic Toxicology 45 (2-3), 91–97. Luster-Teasley, S.L., Yao, J.J., Herner, H.H., Trosko, J.E., Masten, S.J., 2002. Ozonation of chrysene: evaluation of byproduct mixtures and identification of toxic constituent. Environmental Science & Technology 36 (5), 869–876. Luster-Teasley, S.L., Ganey, P.E., DiOrio, M., Ward, J.S., Maleczka, R.E., Trosko, J.E., Masten, S.J., 2005. Effect of byproducts from the ozonation of pyrene: biphenyl-2,20 ,6,60 tetracarbaldehyde and biphenyl-2,20 ,6,60 -tetracarboxylic acid

447

on gap junction intercellular communication and neutrophil function. Environmental Toxicology and Chemistry 24 (3), 733–740. McDowell, D.C., Huber, M.M., Wagner, M., Von Gunten, U., Ternes, T.A., 2005. Ozonation of carbamazepine in drinking water: identification and kinetic study of major oxidation products. Environmental Science & Technology 39 (20), 8014–8022. Moltmann, J.F., Liebig, M., Knacker, T., Keller, M., Scheurer, M., Ternes, T., 2007. Relevance of Endocrine Disrupting Substances and Pharmaceuticals in Surface Waters (Gewa¨sserelevanz endokriner Stoffe und Arzneimittel). Final report. Federal Environment Agency. Funding code 20524205. Monarca, S., Feretti, D., Collivignarelli, C., Guzzella, L., Zerbini, I., Bertanza, G., Pedrazzani, R., 2000. The influence of different disinfectants on mutagenicity and toxicity of urban wastewater. Water Research 34 (17), 4261–4269. Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S., Sato, N., Takada, H., 2007. Removal of selected pharmaceuticals and personal care products (PPCPs) and endocrine-disrupting chemicals (EDCs) during sand filtration and ozonation at a municipal sewage treatment plant. Water Research 41 (19), 4373–4382. Nishihara, T., Nishikawa, J., Kanayama, T., Dakeyama, F., Saito, K., Imagawa, M., Takatori, S., Kitagawa, Y., Hori, S., Utsumi, H., 2000. Estrogenic activities of 517 chemicals by yeast two-hybrid assay. Journal of Health Science 46 (4), 282–298. OECD, 1992a. Fish Acute Toxicity Test. OECD Guidelines for the Testing of Chemicals. No. 203. Organisation for Economic Cooperation and Development, Paris. OECD, 1992b. Fish Early Life Stage Toxicity Test. OECD Guidelines for the Testing of Chemicals. No. 210. Organisation for Economic Cooperation and Development, Paris. Parrott, J.L., Blunt, B.R., 2005. Life-cycle exposure of fathead minnows (Pimephales promelas) to an ethinylestradiol concentration below 1 ng/L reduces egg fertilization success and demasculinizes males. Environmental Toxicology 20 (2), 131–141. Petala, M., Samaras, P., Zouboulis, A., Kungolos, A., Salkellaropoulos, G., 2006. Ecotoxicological properties of wastewater treated using tertiary methods. Environmental Toxicology 21 (4), 417–424. Pu¨ttmann, W., Keil, F., Oehlmann, J., Schulte-Oehlmann, U., 2008. Strategy to reduce pharmaceuticals in drinking water – technical approach (Wassertechnische Strategien zur Reduzierung der Trinkwasserbelastung durch Arzneimittelwirkstoffe). Umweltwissenschaften und Schadstoff-Forschung 20 (3), 209–226. Purdom, C.E., Hardiman, P.A., Bye, V.V.J., Eno, N.C., Tyler, C.R., Sumpter, J.P., 1994. Estrogenic effects of effluents from sewage treatment works. Chemical Ecology 8 (4), 275–285. Rodgers-Gray, T.P., Jobling, S., Kelly, C., Morris, S., Brighty, G., Waldock, M.J., Sumpter, J.P., Tyler, C.R., 2001. Exposure of juvenile roach (Rutilus rutilus) to treated sewage effluent induces dose-dependent and persistent disruption in gonadal duct development. Environmental Science & Technology 35 (3), 462–470. Schmidt, C.K., Brauch, H.J., 2008. N, N-dimethosulfamide as precursor for N-nitrosodimethylamine (NDMA) formation upon ozonation and its fate during drinking water treatment. Environmental Science & Technology 42 (17), 6340–6346. Schwaiger, J., Ferling, H., Mallow, U., Wintermayr, H., Negele, R.D., 2004. Toxic effects of the non-steroidal antiinflammatory drug diclofenac Part 1: histopathological alterations and bioaccumulation in rainbow trout. Aquatic Toxicology 68 (2), 141–150. Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge

448

water research 44 (2010) 439–448

of micropollutants in aquatic systems. Science 313 (5790), 1072–1077. Ternes, T.A., Stumpf, M., Mu¨ller, J., Haberer, K., Wilken, R.D., Servos, M., 1999. Behavior and occurrence of estrogens in municipal sewage treatment plants – I. Investigations in Germany, Canada and Brazil. Science of the Total Environment 225 (1-2), 81–90. Tixier, C., Singer, H.P., Oellers, S., Mu¨ller, S.R., 2003. Occurrence and fate of carbamazepine, clofibric acid, diclofenac, ibuprofen, ketoprofen, and naproxen in surface waters. Environmental Science & Technology 37 (6), 1061–1068. Triebskorn, R., Casper, H., Heyd, A., Eikemper, R., Ko¨hler, H.R., Schwaiger, J., 2004. Toxic effects of the non-steroidal antiinflammatory drug diclofenac Part II. Cytological effects in liver, kidney, gills and intestine of rainbow trout (Oncorhynchus mykiss). Aquatic Toxicology 68 (2), 151–166.

Tyrrell, S.A., Rippey, S.R., Watkins, W.D., 1995. Inactivation of bacterial and viral indicators in secondary sewage effluents, using chlorine and ozone. Water Research 29 (11), 2483–2490. von Gunten, U., 2003. Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Research 37 (7), 1443–1467. Wang, J.Z., Summers, R.S., 1996. Biodegradation behavior of ozonated natural organic matter in sand filters. Revue des Sciences de l’Eau 9 (1), 3–16. Wert, E.C., Rosario-Ortiz, F.L., Drury, D.D., Snyder, S.A., 2007. Formation of oxidation byproducts from ozonation of wastewater. Water Research 41 (7), 1481–1490. Ying, G.G., Williams, B., Kookana, R., 2002. Environmental fate of alkylphenols and alkylphenol ethoxylates – a review. Environment International 28 (3), 215–226.

water research 44 (2010) 449–460

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

The role of organic matter in the removal of emerging trace organic chemicals during managed aquifer recharge T. Rauch-Williams, C. Hoppe-Jones, J.E. Drewes* Advanced Water Technology Center (AQWATEC), Colorado School of Mines, Environmental Science and Engineering Division, Golden, CO 80401-1887, United States

article info

abstract

Article history:

This study explored the effect of different bulk organic carbon matrices on the fate of trace

Received 20 April 2009

organic chemicals (TOrC) during managed aquifer recharge (MAR). Infiltration through

Received in revised form

porous media was simulated in biologically active column experiments under aerobic and

11 August 2009

anoxic recharge conditions. Wastewater effluent derived organic carbon types, differing in

Accepted 20 August 2009

hydrophobicity and biodegradability (i. e., hydrophobic acids, hydrophilic carbon, organic

Available online 27 August 2009

colloids), were used as feed substrates in the column experiments. These carbon substrates while fed at the same concentration differed in their ability to support soil biomass growth

Keywords:

during porous media infiltration. Removal of degradable TOrC (with the exception of

Trace organic chemicals (TOrC)

diclofenac and propyphenazone) was equal or better under aerobic versus anoxic porous

Groundwater recharge

media infiltration conditions. During the initial phase of infiltration, the presence of

Effluent organic matter

biodegradable organic carbon (BDOC) enhanced the decay of degradable TOrC by

Managed aquifer recharge

promoting soil biomass growth, suggesting that BDOC served as a co-substrate in a co-

Riverbank filtration

metabolic transformation of these contaminants. However, unexpected high removal

Biotransformation

efficiencies were observed for all degradable TOrC in the presence of low BDOC concen-

Co-metabolism

trations under well adopted oligotrophic conditions. It is hypothesized that removal under

Primary substrate

these conditions is caused by a specialized microbial community growing on refractory carbon substrates such as hydrophobic acids. Findings of this study reveal that the concentration and character of bulk organic carbon present in effluents affect the degradation efficiency for TOrC during recharge operation. Specifically aerobic, oligotrophic microbiological soil environments present favorable conditions for the transformation of TOrC, including rather recalcitrant compounds such as chlorinated flame retardants. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

Managed aquifer recharge (MAR) systems, such as riverbank filtration (RBF) and soil aquifer treatment (SAT), are widely used natural processes for drinking water augmentation projects using source water that might be impaired by wastewater discharge. Previous studies have demonstrated that MAR systems are effective in dampening and reducing

the concentrations of dissolved organic carbon (DOC) as well as various trace organic contaminants (TOrC) that might be present in impaired source waters (Drewes and Fox, 1999; Brauch et al., 2000; Gru¨nheid et al., 2005). The presence of TOrC has become a key concern for drinking water augmentation projects during the past decade (Kolpin et al., 2002; Heberer, 2002; Focazio et al., 2008). Although adverse human health effects caused by these compounds at concentrations

* Corresponding author. E-mail address: [email protected] (J.E. Drewes). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.08.027

450

water research 44 (2010) 449–460

commonly observed in impaired water resources are very unlikely (Schwab et al., 2005), minimizing exposure of wastewater derived contaminants in these projects is desired. Previous research on the fate of TOrC during MAR has primarily focused on collecting anecdotal and site specific information on their occurrence and removal (Drewes et al., 2002; Montgomery-Brown et al., 2003; Gru¨nheid et al., 2005; Dillon et al., 2008). Studies delineating the mechanisms and boundary conditions for the transformation of wastewater derived TOrC during MAR are lacking. Previous studies demonstrated that the type and bioavailability of effluent organic matter (EfOM) controls the extent of soil biomass growth in MAR systems (Rauch and Drewes, 2004; Rauch and Drewes, 2005). EfOM may consequently effect the degradation of TOrC by serving as a co-substrate in microbiologically facilitated transformations (Stratton et al., 1983). The diversity and expression of the soil microbial community also depends on composition and concentration of the organic carbon substrate controlling trophic cycles in the subsurface (Preuß and Nehrkorn, 1996; Szewzyk et al., 1998). The composition of EfOM (i.e. in terms of its bioavailability) is primarily determined by the degree of wastewater treatment employed (Drewes and Fox, 1999), which can vary widely from primary to conventional to advanced wastewater treatment. As a result, effluent qualities fed to MAR systems can vary in biodegradable dissolved organic carbon (BDOC) concentrations from less than 1 up to 15 mg/L or more. As a consequence, soil microbial communities growing on different levels of BDOC can differ widely in total biomass and diversity. The objectives of this research were to investigate the role of 1) abiotic vs. biotic conditions, 2) BDOC and 3) the type of organic carbon matrices on the removal of select TOrC, such as pharmaceutical residues, personal care products, and household chemicals, during MAR.

2.

Methodology

2.1.

Target organic contaminants

Compounds selected for this study represent small molecular weight organic chemicals (180 to 360 Dalton) that are hydrophilic at neutral pH regimes as indicated by an octanol/water partition coefficient at pH 7 (log DpH¼7 of less than 2.6). TOrC with these properties have a high potential to migrate into groundwater and are not expected to be adsorbed onto porous media. The molecular structures and physicochemical properties of the target compounds chosen for this study are presented in Table 1. These compounds cover a wide range of biodegradability as previously reported for soil/water systems. The anticonvulsants carbamazepine and primidone have been classified as recalcitrant during wastewater treatment and MAR in earlier studies (Heberer, 2002; Drewes et al., 2003; Clara et al., 2004). The chlorinated phosphate esters tris (1-chloroisopropyl)-phosphate (TCPP) and tris(2-chloroethyl)phosphate (TCEP) are two widely used flame retardants and are persistent in the aquatic environment (Heberer et al., 2001; Fries and Pu¨ttmann, 2003). During bank filtration in Germany, however, TCPP and TCEP exhibited a significant reduction, which was attributed to biotransformation in the aquifer

(Heberer et al., 2003). Amy and Drewes (2006) also reported removal of TCEP to concentrations below the detection limit after two years of subsurface travel in an MAR facility supporting that chlorinated flame retardants can be biotransformed under anoxic conditions. Propyphenazone is a poorly biodegradable analgesic that persists during RBF (Heberer et al., 2003) and SAT (Drewes et al., 2003). For diclofenac, a popular analgesic drug, low removal due to biodegradation or adsorption was reported (Buser et al., 1998; Mo¨hle et al., 1999) unless soils contain a high organic carbon content (Drillia et al., 2003). Several studies report a faster degradation of diclofenac under anoxic conditions as compared to aerobic conditions (Zwiener and Frimmel, 2003; Hua et al., 2003). Opposing results were reported by Schmidt et al. (2004) in that diclofenac was almost completely removed during aerobic bank filtration but recalcitrant during anaerobic recharge. Ibuprofen, ketoprofen, and naproxen are common analgesics that are well degradable during wastewater treatment (Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004) and during soil infiltration (Sedlak and Pinkston, 2001; Drewes et al., 2003). Gemfibrozil is a commonly prescribed blood lipid regulator found in unconfined shallow aquifers impacted by wastewater infiltration in the low to moderate ng/L-concentration range (Heberer and Stan, 1997; Drewes and Shore, 2001; Heberer, 2002). Gemfibrozil was removed below the limit of detection within a few weeks during groundwater recharge using SAT (Drewes et al., 2003).

2.2.

Analytical methods

2.2.1.

GC/MS analysis

The following TOrC were analyzed by gas chromatography coupled with mass spectrometry (GC/MS) using a HP 6890 gas chromatograph and a HP 5973 quadrupole mass spectrometer from Agilent Technologies (Waldbronn, Germany) adopting a method published by Reddersen and Heberer (2003): gemfibrozil, primidone, diclofenac sodium salt, carbamazepine, ketoprofen, naproxen, phenacetine, tris (2-chloroethyl)phosphate (TCEP) (Sigma Aldrich Chemicals), tris(chloropropyl)phosphate (TCPP), and propyphenazone (Pfaltz & Bauer, Inc.). Stock solutions were prepared by dissolving the compounds in milli-Q water adjusted to a pH of 10, during sonification and in the dark. A volume of 500 mL of sample was collected and filtered (0.45 mm, Whatman) prior to solid phase extraction. 10,11-dihydrocarbamazepine (Sigma Aldrich Chemicals) and 2-(m-chlorophenoxy) propionic acid (Sigma Aldrich Chemicals) were used as surrogate standards. The limits of detection (LofD) and limit of quantification (LoQ) for the target compounds ranged from 5 to 10 ng/L and from 10 to 50 ng/L, respectively. Only parent target compounds were investigated in this study, metabolites and conjugates were not analyzed.

2.2.2.

HPLC analysis

A high performance liquid chromatography (HPLC) HP 1100 (Agilent Technologies) combined with a UV diode array detector (DAD) was used to quantify concentrations of primidone, phenacetine, carbamazepine, naproxen, diclofenac, and ibuprofen in the lower mg/L range during adsorption breakthrough tests. The samples were directly injected without filtration or other

451

water research 44 (2010) 449–460

Table 1 – Physical–chemical properties of target organic compounds. Compound

Carbamazepine (anticonvulsant)

Structure

pKa

log P (log DpH7)

Expected degradability

References

13.9

2.67

Persistent

Mersmann et al., 2002; Schmidt et al., 2004

Diclofenac sodium (analgesic/antiinflammatory)

4.0

4.06 (1.06)

Degradable (anoxic)

Buser et al., 1998; Mo¨hle et al., 1999; Drillia et al., 2003; Zwiener and Frimmel, 2003, Hua et al., 2003; Schmidt et al., 2004

Gemfibrozil (blood lipid regulator)

4.8

4.39 (2.19)

Relatively persistent

Heberer and Stan, 1997; Drewes and Shore, 2001; Heberer, 2002; Drewes et al., 2003

3.72 (1.12)

Degradable (better aerobic)

Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003

Degradable

Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003

Degradable

Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003

Relatively persistent

Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003

Ibuprofen (analgesic/antiinflammatory)

Ketoprofen (analgesic/antiinflammatory)

Naproxen (analgesic/antiinflammatory)

Phenacetine (antipyretic)

4.4

4.2

4.0

n/a

2.81 (0.01)

3.00 (0.00)

1.63 (1.63)

Primidone (anticonvulsant)

12.3

0.4 (0.4)

Persistent

Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003

Propyphenazone (analgesic)

n/a

1.74 (1.74)

Poor/ persistent

Heberer et al., 2003; Drewes et al., 2003

(continued on next page)

452

water research 44 (2010) 449–460

Table 1 (continued ) Compound

Structure

Tris(2chloroethyl)phosphate (TCEP) (flame retardant)

pKa

n/a

log P (log DpH7)

Expected degradability

0.48 (0.48)

Relatively persistent

References

Heberer et al., 2001; Heberer et al., 2003, Amy and Drewes, 2006; Fries and Pu¨ttmann, 2003

n/a not applicable. pKa and log P values calculated by software ACD. log DpH¼7 values calculated using equations proposed by Scherrer and Howard, (1977).

sample preparation. The mobile phase consisted of two solvents: 1) type 1 water (adjusted to pH 2.5 with o-phosphoric acid (HPLC grade, Fisher Scientific) and buffered with 25 mmol potassium phosphate monobasic (Fisher Scientific)), and 2) acetonitrile (HPLC grade, Mallinckrodt ChromAR HPLC) with the same concentration of o-phosphoric acid as solvent 1). Sample injection (100 mL) occurred in triplicates for each sample at a flow rate of 2 mL/min. A mobile phase gradient was applied during the run for solvent 2 (acetonitrile) of 30% at t ¼ 0 min., 80% at t ¼ 12 min., and 30% at t ¼ 15 min. with a post-run time of 7 min. for column cleaning (30% of solvent 2). The detection wavelengths (band width 10 nm) were set to 205 nm at t ¼ 0 min. for quantification of primidone, phenacetine, and carbamazepine, 230 nm at t ¼ 6.8 min. (naproxen), and 205 nm at 7.8 min. for detection of diclofenac and ibuprofen with a reference wavelength of 330 nm (band width 60 nm) for all compounds. Standards were run in the range of 5–500 mg/L. The detection limits for each compound were 5 mg/L.

2.2.3.

DOC/UV absorbance/nitrate/ammonium

A Sievers 800 Total Organic Carbon Analyzer (GE, Boulder) was used for DOC quantification after microfiltration (0.45 mm, Whatman) (Standard Method 5310C). UV absorbance (UVA) measurements were conducted at 254 nm using a Beckman Coulter DU 800 Spectrophotometer after 0.45 mm filtration (Standard method 5910B). The specific UV absorbance (SUVA) was calculated as the ratio of UVA and DOC. Ammonium and nitrate concentrations were measured using the Nessler and the chromotropic acid method (HACH) with a detection range of 0.02–2.0 mg/l and 0.2–30 mg/L, respectively.

2.2.4.

Biomass

Soil biomass was determined as total viable biomass (i.e. viable, not necessarily active bacteria) using phospholipid extraction (PLE) as described in Rauch and Drewes (2005). Analyses were conducted in triplicates from the top-soil (0–2 cm, infiltration zone) of the columns.

2.2.5.

Wastewater effluent

Secondary treated wastewater effluent served as the feed to column systems and as source for subsequent organic carbon fractionation. The secondary treated effluent was collected from a local wastewater treatment plant employing nitrification and partial denitrification. The average DOC concentration of the effluent was 8.74 1.44 mg/L.

2.2.6.

Organic carbon fractionation

Organic matter (less than 1 mm in size as defined for this study) of the secondary effluent sample was isolated into three organic fractions: colloidal organic carbon, hydrophobic acids (HPO-A), and hydrophilic carbon (HPI) following the procedure described in Rauch and Drewes (2005) with some modifications. The sample was concentrated by vacuum rotary evaporation at 45 C (concentration factor: 40). The concentrate was separated by dialysis (Spectra/Por, Spectrum, 6000– 8000 Dalton) at pH 4–5 into organic colloids and DOC. The permeate of the dialysis (containing DOC) was collected and further separated into HPI and HPO-A by XAD-8 fractionation according to Leenheer et al. (2000) using a capacity factor of k0 ¼ 4. A carbon mass balance was performed for each carbon fractionation based on UVA and DOC measurements for quality control. In average, the secondary treated effluent contained 16 percent colloidal organic carbon, 38 percent HPO-A, 37 percent HPI, and 8 percent hydrophobic neutrals (HPO-N) (Rauch and Drewes, 2004). Organic carbon isolates were diluted to 3 mg/L DOC using milli-Q water, adjusted for ion strength, macro nutrient (nitrogen, phosphate) and micro nutrient concentrations (Rauch and Drewes, 2004) and promptly utilized as column feed waters.

2.3.

Experimental set-up

2.3.1.

Anoxic column system

The columns (PC system) were operated to study TOrC removal during simulated MAR under anoxic conditions and up to 3–4 weeks of retention time in the subsurface. The anoxic columns consisted of four 1-m plexiglass columns (15 cm i.d.) in series filled with aquifer material (d50 ¼ 0.8 mm, foc ¼ 0.003%) (Fig. SI-1, Supplemental Information). The column system was operated in flow-through mode at a loading rate of 0.065 m/d under saturated, anoxic (denitrifying) flow conditions (Table 2). The hydraulic retention time of the system using four columns in series was previously determined as 25 days. The columns had been continuously fed with secondary or tertiary treated effluents for over 6 years and for 5 months with the secondary treated effluent quality employed in this study prior to spiking of TOrC. The column influent was regularly purged with nitrogen gas to keep dissolved oxygen concentrations below 0.5 mg/L. Samples were collected once to twice a week from column influent and the four column effluents, respectively, and analyzed for DOC, UV absorbance, pH, conductivity and TOrC concentrations.

453

water research 44 (2010) 449–460

Table 2 – Operational parameters and water quality changes of columns used in this study.a Label

PC C1 C2 C3 C4 Abiotic

Feed water matrix

Length (m)

Anoxic columns (EfOM) Hydrophobic acids (HPO-A) Hydrophilic carbon (HPI) Effluent organic matter (EfOM) Organic colloids (Org. Colloids) Abiotic column, Type I water

41m¼ 4m 0.3

Organic Carbon

Feed water (mg/L) 6.9

Effluent (mg/L)

BOC (mg/L)

Predominant Redox Condition

Total Viable Soil Biomassb (nmol PO3 4 /g d.w soil)

Anoxic

17.3 1.3

3.1 0.9

See Fig. 1 for DOC column profile 2.7 0.02 0.4 0.9

Aerobic

3.3 0.4

0.3

2.8 0.4

2.6 0.1

0.3 0.4

Aerobic

10.5 5.3

0.3

3.0 0.4

2.7 0.4

0.4 0.4

Aerobic

20.6 2.5

0.3

3.1 0.5

2.2 0.3

0.8 0.5

Aerobic

27.2 1.8

0.3

<0.1

<0.1

0

Abiotic

0.5 0.3

a All experiments conducted at feed water conductivity of 940–1500 mS/cm and pH 7.0–7.9. b Biomass samples were collected from the infiltration zone of all columns at 0–2 cm depth.

2.3.2.

Organic carbon fraction columns

Four columns (C1–C4) were operated to examine the effect of different organic carbon matrices on soil biomass growth and TOrC removal during aerobic conditions and short retention times (less than 1 day) in the subsurface. This system consisted of four parallel plexiglass columns (L ¼ 30 cm, i.d. ¼ 5 cm) filled with silica sand (d50 ¼ 0.65 mm, foc ¼ 0.004%). The columns were operated under saturated, aerobic flow conditions at a loading rate of 0.9 m/d representing a retention time of about 19 h in the porous media. The system had been acclimated for over two years to the infiltration of the four organic carbon fractions derived from secondary effluent (HPO-A, HPI, organic colloids, EfOM), respectively. Approximately six weeks prior to spiking TOrC to the column systems, the feed water concentrations of all four fractions was adjusted to 3 mg/L as DOC. Samples for DOC, UVA and pH were taken once to twice a week from the four column influents and effluents. Three consecutive TOrC spiking events were conducted. For each event, hydraulically corresponding samples of column influents and effluents were collected, respectively, after a complete breakthrough of the spike solution was observed using conductivity as a conservative tracer (typically after 24 h). After the 1st spiking event (day 0), the 2nd and 3rd spikes occurred on day 11 and 22, respectively. Set-up and operation conditions of the different columns employed in this study are summarized in Table 2. All small column systems received feed waters that were adjusted to a pH of 8.0 0.2 to minimize adsorption effects of acidic TOrC onto porous media.

2.3.3.

Adsorption test

An additional column was operated under abiotic conditions to assess the role of physical adsorption of selected TOrC onto porous media. The column set-up and operation was the same as described for the biologically active columns C1–C4. The adsorption column was filled with aquifer material from an RBF site (d50 ¼ 0.55 mm, bulk density ¼ 1.81 g/cm3, effective porosity ¼ 0.36, fOC ¼ 0.066%). Six TOrC were spiked into secondary effluent in a concentration range of 175–226 mg/L adjusted to a pH of 7.0 (feed water). To minimize biotransformation during this experiment, the column was kept

abiotic by adding sodium azide at a concentration of 2 mM. Sodium bromide was added as a conservative tracer and bromide was measured by ion chromatography (Dionex, Sunnyvale, CA) in the column effluent. The column was operated in single flow-through mode under saturated flow conditions at a hydraulic loading rate of 0.33 m/d. Effluent samples were collected using a fraction collector and analyzed by HPLC analysis. Retardation factors (Rd) were fitted to the step input breakthrough curves for each compound using the software program CXTFIT (Toride et al., 1995). Adsorption coefficients (Kd) were calculated according to the following relationship Rd ¼

rKd þ1 3

(1)

where Rd is the retardation factor, r is the sand bulk density, and 3 is the sand porosity.

3.

Results and discussion

3.1.

Adsorption behavior of target TOrC

In order to assess whether physical retardation would be a contributing factor in the removal of select TOrC in the column experiments, an abiotic column experiment was conducted to study sorption of the compounds onto the virgin porous media used in the columns. None of the six TOrC included in this experiment exhibited significant retardation during flow through porous media as compared to the conservative tracer bromide (indicated by Rd values very close to 1, Table 3). These findings reveal that sorption did not play an important role in the attenuation of these compounds onto the porous media used in the columns.

3.2.

Organic carbon removal

During the course of the study, the anoxic column system exhibited a steady removal of DOC (55%). The DOC removal observed in the column system followed a first-order kinetic

454

water research 44 (2010) 449–460

Table 3 – Retardation and sorption coefficients of select TOrC derived during abiotic column study. TOrC Naproxen Ibuprofen Phenacetine Diclofenac Primidone Carbamazepine

Rd

Kd (mL/g)

Mass recovery (%)

1.30 1.10 1.02 1.41 1.00 1.90

0.08 0.03 n.a. 0.01 <0.001 0.24

94.2 92.7 n.a. 93.6 102 94.0

degradation profile (Fig. 1). SUVA increased from 1.7 to 2.1 within 25 days of hydraulic residence time indicating a preferred removal of saturated organic compounds. Average DOC concentrations observed in the influent and effluent of column systems C1, C2, C3, and C4 fed with different organic carbon fractions are summarized in Table 2. Average influent DOC concentrations in all columns ranged from 2.8 to 3.1 mg/L. Columns C1–C4 exhibited average DOC removal by 14 (31) percent, 9 (15) percent, 12 (15) percent, and 25 (14) percent, respectively (Table 2). Standard deviations of BDOC concentrations in all columns were determined by propagation of error, and were for all fractions, with the exception of organic colloids, too high to compare statistically significant carbon removal among the different carbon feed fractions. In previous work, using the same column set-up but higher organic carbon influent concentrations (5 mg/L instead of 3 mg/L), the authors were able to show that the BDOC content in the organic carbon fractions increased relatively in the order of HPO-A, HPI, EfOM, and organic colloids (Rauch and Drewes, 2004).

3.3.

Prevailing redox conditions in the columns

Dissolved oxygen concentrations in the water fed to the anoxic column system were consistently below 0.3 mg/L. Denitrification resulted in a removal of 96 percent nitrate in the anoxic columns from an average influent concentration of 9.3 mg/L NO 3 -N. The pH increased slightly from 7.6 to 7.9 on average after a retention time of 25 days (Fig. 1). The columns C1–C4 were operated under fully aerobic conditions. Dissolved oxygen concentrations in the effluents of all four aerobic columns remained above 3 mg/L. Nitrification was observed in

Fig. 1 – DOC, SUVA, and soil biomass profile in anoxic columns (error bars [ Stdev). (The pH in the anoxic columns ranged between 7.6 and 7.9).

the columns (data not presented), but denitrification did not occur.

3.4.

Soil biomass

In the anoxic column system, soil biomass measured as PLE decreased rapidly with depth (Fig. 1). Due to little bioavailable organic carbon in the HPO-A feed water, column C1 sustained comparatively low amounts of soil microbial biomass (Table 2). However, in comparison to the virgin sand (prior to exposure to any aqueous organic carbon) biomass growth in presence of HPO-A was quantifiable even in such carbon poor (oligotrophic) conditions. Soil biomass in the infiltration zone (top 0–2 cm) of the HPI (C2), EfOM (C3) and organic colloid column (C4) increased in correlation with BDOC concentrations in the respective feed water matrices. This trend is not apparent from the BDOC data collected during this study, but is supported by previous work conducted when the same column systems were operated for a long period of time at slightly higher influent DOC concentrations of 5 mg/L (Rauch and Drewes, 2004). The variability of the DOC analysis in column influent and effluent samples likely resulted in a propagation of error for calculating the BDOC concentration that prevented showing the same trend for the lower feed DOC concentration used in this study.

3.5.

Removal of TOrC under anoxic conditions

The removal of TOrC under anoxic flow conditions was studied at environmentally relevant concentrations (100–900 ng/L) (Table 4). Compounds exhibiting concentrations of less than 100 ng/L in the secondary effluent (i.e. primidone, propyphenazone, diclofenac, and ketoprofen) were artificially amended prior to the experiments, in order to be able to derive meaningful removal efficiencies in the columns. TOrC influent and effluent concentrations for all column studies summarized in Table 4 are presented for synoptic samples taking into account the hydraulic residence time of the respective column systems. During the column studies, the concentrations of TOrC in the anoxic feed water remained stable for at least 4 weeks (with the exception of ibuprofen which exhibited a rapid loss after 8 days, data not shown). The observed removal of TOrC in the anoxic column system is presented in Table 4 and Fig. 2. Naproxen exhibited an almost linear removal during the first 2 m of infiltration (or 12.5 days residence time) and remained at a concentration of 10–25 ng/L during subsequent travel. Similarly, ketoprofen and diclofenac concentrations were reduced within 1–2 m of soil infiltration and subsequently remained relatively stable at concentrations below the limit of quantification for ketoprofen and in the range of 75–135 ng/L for diclofenac. Low and relatively stable TOrC concentrations in the 3rd and 4th anoxic column may indicate that concentration thresholds for certain TOrC exist below which further degradation is inhibited, as previously reported by others (Van der Meer et al., 1987; McCarty et al., 1981; Margesin and Schinner, 2001; Rittmann and McCarty, 2001). In contrast to bulk organic carbon removal (first-order kinetic), the TOrC attenuation (Fig. 2) seem to follow a linear removal with depth indicating a zero-order kinetic. While

Table 4 – Summary of TOrC concentrations observed in column studies. Anoxic Column TOrC (ng/L)

Carbamazepine Diclofenac sodium

Feed water 279–296 (n ¼ 3) 307–416 (n ¼ 2)

1-m Column effluent n.a. 75

2-m Column effluent n.a. 134

HPO-A (C1) 3-m Column effluent n.a. 110

4-m column effluent 318 80–120 (n ¼ 2)

444 141 (n ¼ 5)

323–325 (n ¼ 2)

<25

<25

<25

Ibuprofen

60

40

<10

<10

<10

Ketoprofen

707 111 (n ¼ 6)

570–592 (n ¼ 2)

<10

<10

<10

Naproxen

412 103 (n ¼ 5)

303–350 (n ¼ 2)

21

17–20 (n¼ 2)

17–20 (n ¼ 2)

Phenacetine

n.a.

Primidone Propyphenazone

562–573 (n ¼ 2) 642 107 (n ¼ 7)

TCEP

TCPP

323 68 (n ¼ 5)

805 172 (n ¼ 7)

629 520–630 (n ¼ 2)

352–377 (n ¼ 3)

955–1010 (n ¼ 3)

n.a. n.a.

n.a.

n.a.

n.a. 36–87 (n ¼ 2)

314

n.a.

764 42–112 (n ¼ 2)

378

942–1075 (n ¼ 2)

Column effluent

Column influent

n.a. 604 538 73 373 770 381 626 948 356 772 605 180 889 711 176 534 539 289

538 190 72 26 <25 62 10 84 <10 38 22 37 18 21 38 <50 <50 <50 133 809 556 1104 908 596 1080 1274 557

Column influent

n.a. n.a. 70 23 544 507 302 496 528 499 406 706 935 438 581 721 795 506 508

n.a. 148 809 542 856 966 572 995 1503 958

Column effluent

EfOM (C3)

n.a. n.a. <10 <10 431 307 59 6 33 <10 318 145 <25 232 36 21 n.a. <50 <50

n.a. n.a. 328 578

318 73 770 547 381 638 369 356 640 589 180 722 37 176 607 141 289

710 132

682 139

721

255

599 691

229 17

670 820

34 421

924 817

428 176

846 392

215 <50

222

605 848 556

863 572

n.a. 1087 958

588

605 n.a.

636 1120 596

n.a. <50 121

<50 n.a.

n.a. 868 404

Column effluent

n.a. 294 72 336 108 62 104 102 <10 500 244 37 4 44 38 <50 <50 <50

588 733 542

n.a.

755 973

Column influent

n.a. 698 341

336 651

Column effluent

Org. colloids (C4)

1057 n.a.

712 1720 557

water research 44 (2010) 449–460

Gemfibrozil

Column influent

HPI (C2)

949 n.a.

n.a.: not available.

455

456

water research 44 (2010) 449–460

indicate a linear removal profile for ketoprofen, naproxen, gemfibrozil, and ibuprofen, the removal profile might also have been affected by a change in soil microbial community composition with column depth caused by changing organic carbon substrate composition and concentrations along the flow path. Thus, definite conclusions about the kinetic order are not possible without additional experiments. Due to the low ambient concentrations of trace organic contaminants (occurring in the ng/L range) it is unlikely that energy from TOrC metabolism is sufficient for biomass maintenance and growth. It was hypothesized that at trace levels, TOrC were transformed by co-metabolism (co-utilization). In this context the term co-metabolism describes a transformation of a trace compound, which is by itself unable to support cell replication, and requires the presence of another transformable organic compound (primary carbon source) that allows microorganisms to obtain energy for their metabolism and growth (Brandt et al., 2003; Stratton et al., 1983; Clara et al., 2005). Because the primary substrate (biodegradable bulk organic carbon) was present in the column experiment in concentrations two orders of magnitude above common concentrations of TOrC in the environment, co-metabolism offers a possible explanation for the observed removal of the target TOrC, that were fed at trace concentrations. This hypothesis was further tested by examining the effect of organic carbon subgroups present in EfOM and differing in biodegradability on soil biomass growth and TOrC removal.

Fig. 2 – Removal of TOrC in anoxic column system.

first-order degradation kinetics are indicative of organic carbon degradation at concentrations below the half-velocity coefficient Ks, or under mass transport limitations that keep biodegradation rates below the biological capacity. Zero-order reaction kinetics are indicative of degradation below the biological degradation potential, where either substrate concentrations of trace contaminants are significantly higher than the affinity for the specific substrate, or the limiting substrate is delivered at a constant rate (Simoni et al., 2001; Alexander, 1999). It should be noted that, while the column results

Gemfibrozil

Ketoprofen

Naproxen

ng/L removed in column

900 800 700 600 500 400 300 200 100 0 3.3

10.4

27.2

HPO-A

HPI

Org. Colloids

nmol org-PO43- g-1

Fig. 3 – Relationship between TOrC removal, soil biomass, and organic carbon substrate in columns C1, C2, and C4 (1st spike event).

3.6. Effect of aqueous organic carbon matrices on TOrC removal In order to further study the effect of soil biomass and aqueous organic carbon matrices on the removal of TOrC, aerobic column experiments were conducted using different organic carbon bulk fractions (EfOM, HPO-A, HPI, and organic colloids) isolated from secondary treated effluent. The different organic bulk substrates were fed at similar DOC concentrations to different column systems and were able to support different amounts of viable soil biomass (Table 2). During three consecutive spiking tests, hydraulically corresponding influent and effluent samples were collected from each column (Fig. 3) with the exception of the column fed with organic colloids for which only two spiking tests were conducted. The results of the first spiking test for the HPI and organic colloid columns are presented in Fig. 3. Findings suggest that the more viable soil biomass was present on the column media, the more complete was the removal for gemfibrozil, ketoprofen, and naproxen. In other terms the concentration of biodegradable effluent derived organic carbon substrate limited the transformation of TOrC. (Phenacetine was removed below the quantification limit in all columns and removal could, thus, not be related to biomass activity in the columns.) Organic colloids (BDOC content 0.86 0.48 mg/L) appeared to have promoted the degradation of TOrC by serving as the primary substrate and establishing a relatively higher microbial population that was able to co-metabolize TOrC better as compared to HPI organic carbon. TOrC removal in the HPI column was possibly lower because HPI substrate (BDOC content 0.25 0.41 mg/L) supported less biomass

457

water research 44 (2010) 449–460

1st spike

2nd spike (day 11 after 1st spike)

3rd spike (day 22 after 1st spike) 925

100 90

544

506

700

508

490 495 499

561

243 550

80 removal %

70 60

206

50

200

40 30

87

113

20 10 0 Ketoprofen

Naproxen

Phenacetine

Gemfibrozil

Ibuprofen

Fig. 4 – Microbial adaptation to TOrC in HPI column (values on top of bars represent removal in ng/L).

growth to metabolize TOrCs. During longer exposure times to the contaminants the performance of the HPI column for TOrC removal improved. In the 2nd and 3rd spiking event, ketoprofen, phenacetine, gemfibrozil, and naproxen exhibited increasingly better removal exceeding 95 percent 22 days after the first spike (Fig. 4) suggesting that the microbial community had adapted to metabolizing TOrC. In accordance with the performance of the anoxic column system (Fig. 2), ibuprofen was equally efficient removed (>85%) in the presence of all organic carbon fractions and in all three spiking events (Table 4 and Fig. 4) under both aerobic and anoxic conditions. This confirms previously reported findings that ibuprofen is very well degradable under various field conditions (Sedlak and Pinkston, 2001; Drewes et al., 2003). In contrast to the anoxic column system, propyphenazone was not eliminated in any of the four short columns (Table 4). Consistent with previous studies, propyphenazone was found to require longer retention times (>1 week) to exhibit partial removal under MAR conditions (Drewes et al., 2003; Heberer et al., 2003). The anticonvulsant drugs were not eliminated in any of the columns investigated (Table 4). Based on these study results, effluentderived BDOC does not seem to be a suitable substrate to stimulate co-metabolic decay of the selected anticonvulsant drugs (i.e. primidone, carbamazepine) under environmentally relevant bulk and TOrC concentrations. Whereas TCPP was not removed in the anoxic columns, this flame retardant seems to be removed in the aerobic HPO-A, HPI, and EfOM column after a lag time of two to three weeks after starting to feed this compound to the column, possibly indicating biological acclimatization leading to some degree of attenuation of this compound (Table 4). However, results are inconsistent when comparing influent and effluent concentrations in the different columns, therefore further studies are recommended to solidify this observation. Given that the first column of the anoxic system was operated at a longer residence time (about 6 days) compared to the aerobic columns (less than 1 day), it is noticeable that most degradable TOrC (i.e. gemfibrozil, ibuprofen, ketoprofen, naproxen) were significantly better transformed under aerobic conditions as compared to anoxic recharge

conditions. Diclofenac exhibited consistent removal during anoxic conditions but not during aerobic recharge, which is consistent with findings reported by Zwiener and Frimmel (2003) and Hua et al. (2003).

3.7.

Removal of TOrC in oligotrophic environments

Although the HPO-fraction provided little soil biomass growth 1 (<5 nmol org-PO3 4 g ) in the infiltration zone (Table 2), column C1 exhibited-unexpectedly-a significant and instantaneous transformation of target TOrC after initial exposure during the first spiking event. Phenacetine, ibuprofen, gemfibrozil, naproxen, and ketoprofen were all consistently removed by more than 90 percent in all three spiking events (Fig. 3 and Table 4). In comparison to all other feed water matrices, the HPO-A column exhibited an equal (regarding ibuprofen) or faster removal for all degradable target TOrC. These findings suggest that the type of effluent derived bulk organic carbon effects both, the quantity of soil biomass growth and the composition of microbial community and that both parameters are essential in understanding TOrC transformation during MAR. It is postulated, that HPO-A (humic-like) carbon promoted the development of an oligotrophic soil microbial community well capable of degrading TOrC to low residual concentrations. Given that EfOM usually is a mixture of all three carbon fractions (HPO-A, HPI, and organic colloids), EfOM may have stimulated the growth of a more diverse biomass in column C3 as compared to C1, C2, or C4 that were only fed with a subfraction of the organic substrate. This may explain why TOrC removal was relatively better in the column fed with EfOM, than would be expected based on soil biomass present in C3 (i.e. for gemfibrozil and naproxen) (Table 4). Daughton and Ternes (1999) acknowledged the role of oligotrophic metabolism for the decay of TOrC in the environment. Accordingly, the biotransformation of bulk organic carbon is governed by enzyme-saturating substrate concentrations (copiotrophic metabolism) during soil infiltration. TOrC are present at low concentrations, at below enzyme-saturating levels, which may necessitate oligotrophic metabolism.

458

water research 44 (2010) 449–460

Because oligotrophic organisms are more likely to be prevalent in native environments characterized by low-carbon fluxes than in wastewater treatment plants, the degradation of TOrC may be more efficient during soil percolation than in wastewater treatment plants (Daughton and Ternes, 1999). It is anticipated that oligotrophic organisms reside in MAR systems in soil depths below the immediate infiltration zone (which in contrast is likely dominated by microbial organisms growing primarily on easily degradable EfOM). Further investigations directed at the interrelation between effluent derived organic carbon and the succession of biocommunities in soil based on trophic carbon sequences seem key to the understanding of mechanisms and limitations to TOrC removal in MAR systems.

Acknowledgments

4.

Appendix. Supplementary data

Conclusions

Hydrophilic, small-molecular size TOrC can survive conventional and advanced wastewater treatment processes and are one of the key concerns in drinking water augmentation projects using impaired source waters. More research is needed to better understand the trophodynamic role of organic carbon on the removal of TOrC in MAR applications. In this study we conducted column experiments to illuminate key factors for the metabolic removal of twelve TOrC during soil infiltration. Experiments were designed to resemble natural environments (flow through systems, long biological acclimatization times, complex aqueous carbon compositions, environmentally relevant bulk and TOrC concentrations) in order to reveal interactions between bulk organic carbon and TOrC removal that are relevant for full-scale MAR conditions. Removal efficiencies varied widely among the target compounds but were generally higher during aerobic soil infiltration with the exception of diclofenac that was faster degraded during anoxic recharge. Propyphenazone required a longer residence time for removal than was provided in the aerobic columns employed in this study. The biological reactions for all other degradable TOrC were fast at ambient concentrations (ng/L level) and completed within the first 2 m of porous media infiltration. Threshold concentrations below which further degradation was negligible were observed for acidic TOrC under anoxic conditions in the lower ng/L-range. Intensive research has been conducted on stimulating the degradation of certain pollutants by artificially adding carbon substrates (Horvath, 1973, Semprini, 1997). Findings of this study demonstrated that naturally available organic matter in aqueous environments can promote the degradation of certain TOrC by serving as a secondary carbon substrate for their removal. The composition of effluent-derived bulk organic carbon had a strong effect on the biological removal of the target TOrC and is based on several mechanisms. BDOC, prevalent in form of colloidal and hydrophilic carbon, stimulates soil biomass growth and induces secondary substrate utilization of TOrC. Soil biomass in aerobic systems was able to adapt to most acidic drugs showing an improved removal over time. Soil biomass growth occurred even in response to the infiltration of recalcitrant hydrophobic acids. This organic carbon substrate induced an oligotrophic microbial community that was more effective in removing TOrC than during

copiotrophic metabolism in the presence of higher concentrations of BDOC. Consistent with previous observations, the anticonvulsant drugs carbamazepine and primidone exhibited no removal under any conditions examined in this study.

Partial funding for this study was provided by the Gwangju Institute of Technology in Korea. We are thankful for analytical support during this study provided by Dr. Thomas Heberer at the Technical University in Berlin, Germany, Matt Oedekoven, and Stephan Wagner.

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2009.08.027.

references

Amy, G., Drewes, J.E., 2006. Soil-aquifer treatment (SAT) as a natural and sustainable wastewater reclamation/reuse technology: fate of wastewater effluent organic matter (EfOM) and trace organic compounds. Environmental Monitoring and Assessment 129 (1–3), 19–26. Alexander, M., 1999. Biodegradation and Bioremediation. Academic Press, San Diego, California. Brandt, B.W., Van Leeuwen, I.M.M., Kooijman, S.A.L.M., 2003. A general model for multiple substrate biodegradation. Application to co-metabolism of structurally non-analogous compounds. Water Research 37, 4843–4854. Brauch, H.-J., Sacher, F., Denecke, E., Tacke, T., 2000. Effectiveness of bank filtration for the removal of polar organic trace contaminants. GWF Wasser Abwasser 141 (4) 2260–234 (in German). Buser, H.R., Poiger, T., Mu¨ller, M.D., 1998. Occurrence and fate of the pharmaceutical drug diclofenac in surface waters: rapid photodegradation in a lake. Environmental Science and Technology 32 (22), 3449–3456. Buser, H.R., Poiger, T., Mu¨ller, M.D., 1999. Occurrence and environmental behavior of the chiral pharmaceutical drug ibuprofen in surface waters and in wastewater. Environmental Science and Technology 33 (15), 2529–2535. Carballa, M., Omil, F., Lema, J.M., Lllompart, M., Jares, C.G., Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Research 38, 2918–2926. Clara, M., Strenn, B., Kreuzinger, N., 2004. Carbamazepine as a possible anthropogenic marker in the aquatic environment: investigations on the behaviour of carbamazepine in wastewater treatment and during groundwater infiltration. Water Research 38, 947–954. Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005. The solids retention time – a suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Research 39, 97–106. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107 (Suppl. 6) December 1999.

water research 44 (2010) 449–460

Dillon, P., Page, D., Vanderzalm, J., Pavelic, P., Toze, S., Bekele, E., Sidhu, J., Prommer, H., Higginson, S., Regel, R., RinckPfeiffer, S., Purdie, M., Pitman, C., Wintgens, T., 2008. A critical evaluation of combined engineered and aquifer treatment systems in water recycling. Water Science and Technology 57 (5), 753–762. Drewes, J.E., Fox, P., 1999. Fate of natural organic matter (NOM) during groundwater recharge using reclaimed water. Water Science and Technology 40 (9), 241–248. Drewes, J.E., Shore, L.S., 2001. Concerns about pharmaceuticals in water reuse, groundwater recharge, and animal waste. In: American Chemical Society (Ed.), Pharmaceuticals and personal care products in the environment. Symp. Ser, vol. 791, pp. 206–228. Drewes, J.E., Heberer, T., Reddersen, K., 2002. Fate of pharmaceuticals during indirect potable reuse. Water Science & Technology 46 (3), 73–80. Drewes, E.J., Heberer, T., Rauch, T., Reddersen, K., 2003. Fate of pharmaceuticals during groundwater recharge. Ground Water Monitoring and Remediation 23 (3), 64–72. Drillia, P., Stamatelatou, K., Lyberatos, G., 2003. Mobility of Pharmaceuticals in Soil, 8th Conference on Environmental Science and Technology, Lemnos, Greece. Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States–II untreated drinking water sources. Science of the Total Environment 402 (2–3), 201–216. Fries, E., Pu¨ttmann, W., 2003. Monitoring of the three organophosphate esters TBP, TCEP and TBEP in river water and ground water (Oder, Germany). Journal of Environmental Monitoring 5 (2), 346–352. Gru¨nheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved organic carbon (DOC) and trace organic compounds by bank filtration and artificial recharge. Water Research 39, 3219–3228. Heberer, T., Stan, H.-J., 1997. Determination of clofibric acid and N-(phenylsulphonyl)-sarcosine in sewage, river and drinking water. International Journal of Environmental Analytical Chemistry 67, 113–124. Heberer, T., Verstraeten, I.M., Meyer, M.T., Mechlinski, A., Reddersen, K., 2001. Occurrence and fate of pharmaceuticals during bank filtration–preliminary results from investigations in Germany and the United States. Water Resources Update 20, 4–17. Heberer, T., 2002. Occurrence, fate and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicology Letters 131, 5–17. Heberer, T., Mechlinski, A., Fanck, B., 2003. NASRI – Occurrence and Fate of Pharmaceuticals During Bank Filtration, Conference Wasser Berlin, KompetenzZentrum Wasser Berlin, pp. 41–49. Horvath, R.S., 1973. Enhancement of co-metabolism of chlorobenzoates by the co-substrate enrichment technique. Applied Microbiology 25 (6), 961–963. Hua, J., An, P., Winter, J., Gallert, C., 2003. Elimination of COD, microorganisms and pharmaceuticals from sewage by trickling through sandy soil below leaking sewers. Water Research 37, 4395–4404. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S. D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in U. S. streams, 1999–2000: a national reconnaissance. Environmental Science & Technology 36, 1202–1211. Leenheer, J.A., Croue, J.P., Benjamin, M., Korshin, G.V., Hwang, C.., Bruchet, A., Aiken, G.R., 2000. Comprehensive isolation of natural organic matter from water for spectral characterizations and reactivity testing. In: Barett, S.E., Krasner, S.W., Amy, G.L.

459

(Eds.), Natural Organic Matter and Disinfection By-Products, vol. 761. ACS Symposium Series, pp. 68–83. Margesin, R., Schinner, F., 2001. Bioremediation (natural attenuation and biostimulation) of diesel-oil-contaminated soil in an alpine glacier skiing area. Applied and Environmental Microbiology 67 (7), 2127–3133. McCarty, P.L., Reinhard, M., Rittmann, B.E., 1981. Trace organics in groundwater. Environmental Science and Technology 15 (1), 40–51. Mersmann, P., Scheytt, T., Heberer, T., 2002. Column experiments on the transport of pharmaceutically active compounds in the saturated zone. Acta Hydrochimica et Hydrobiologica 30 (5–6), 275–284 (in German). Mo¨hle, E., Kempter, C., Kern, A., Metzger, J.W., 1999. Examination of the degradation of drugs in municipal sewage plants using liquid chromatography-electrospray mass spectrometry. Acta Hydrochimica et Hydrobiologica 27 (6), 430–436 (in German). Montgomery-Brown, J., Drewes, J.E., Fox, P., Reinhard, M., 2003. Behavior of alkylphenol polyethoxlate metabolites during soil aquifer treatment. Water Research 37, 3672–3681. Preuß, G., Nehrkorn, A., 1996. Succession of microbial communities during bank filtration and artificial groundwater recharge Artificial Recharge of Groundwater. Proceedings of an International Symposium, A.-L. Kivima¨ki and T. Suokko, eds., Helsinki, Finland, June 3–5, 1996, Nordic Hydrological Programme, pp. 215–221. Rauch, T., Drewes, J.E., 2004. Assessing the removal potential of soil-aquifer treatment systems for bulk organic matter. Water Science & Technology 50 (2), 245–253. Rauch, T., Drewes, J.E., 2005. A novel approach for quantifying bulk organic carbon removal in groundwater recharge systems. Journal of Environmental Engineering 131 (6), 909–923. Reddersen, K., Heberer, T., 2003. Multi-compound methods for the detection of pharmaceutical residues in various waters applying solid phase extraction (SPE) and gas chromatography with mass spectrometric (GC-MS) detection. Journal of Separation Sciences 26, 1443–1450. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology. Principles and Applications. McGraw-Hill, Boston. Scherrer, R.A., Howard, S.M., 1977. Use of distribution coefficients in quantitative structure-activity relationships. Journal of Medical Chemistry 20 (1), 53–58. Schmidt, C.K., Lange, F.T., Brauch, H.-J., 2004. Assessing the Impact of Different Redox Conditions and Residence Times on the Fate of Organic Micropollutants during Riverbank Filtration, Fourth International Conference on Pharmaceuticals and Endocrine Disrupting Chemicals in Water, October 13–15, Minneapolis, Minnesota. Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff, R.D., D’aco, V.J., Anderson, P.D., 2005. Human pharmaceuticals in US surface waters: a human health risk assessment. Regulatory Toxicology and Pharmacology 42 (3), 296–312. Sedlak, D.L., Pinkston, K.E., 2001. Factors affecting the concentrations of pharmaceuticals released to the aquatic environment. Journal of Contemporary Water Research and Education 120, 56–64. September. Semprini, L., 1997. Strategies for the aerobic co-metabolism of chlorinated solvents. Current Opinion in Biotechnology 8 (3), 296–308. Simoni, S.F., Scha¨fer, A., Harms, H., Zehnder, A.J.B., 2001. Factors affecting mass transfer limited biodegradation in saturated porous media. Journal of Contaminant Hydrology 50, 99–120. Stratton, R.G., Namkung, E., Rittmann, B.E., 1983. Secondary utilization of trace organic by biofilms on porous media. Journal American Water Works Association 75, 463–469.

460

water research 44 (2010) 449–460

Szewzyk, U., Kalmbach, S., Manz, W., La˚ngmark, J., and Stenstro¨m, T.-A., 1998. Artificial groundwater as the future water supply of greater Stockholm IV: Phylogenetic diversity of the unsaturated zones of experimental sand filter columns. Artificial recharge of groundwater. Proceedings of the third international symposium on artificial recharge of groundwater–Tisar 98, J.H. Peters, ed., Amsterdam, Netherlands, 21–25 September 1998, Rotterdam, pp. 391–393. Toride, N., Leij, F.J., van Genuchten, M.T., 1995. The CXTFIT Code for Estimating Transport Parameters from Laboratory or Field

Tracer Experiments. Version 2.0. U.S. Salinity Laboratory Res. Rep. 137, U.S. Salinity, Riverside, CA. Van der Meer, J.R., Roelofsen, W., Schraa, G., Zehnder, A.J.B., 1987. Degradation of low concentrations of dichlorobezenes and 1,2,4thrichlorobenzene by Pseudomonas sp. strain P51 in nonsterile soil columns. FEMS Microbiology Letters 45 (6), 333–341. Zwiener, C., Frimmel, F.H., 2003. Short-term tests with a pilot sewage plant and biofilm reactors for the biological degradation of the pharmaceutical compounds clofibric acid, ibuprofen, and diclofenac. Science of the Total Environment 309, 201–211.

water research 44 (2010) 461–476

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Toxicological relevance of emerging contaminants for drinking water quality Merijn Schriks a,*, Minne B. Heringa a, Margaretha M.E. van der Kooi a, Pim de Voogt a,b, Annemarie P. van Wezel a a

KWR Watercycle Research Institute, P.O. Box 1072, 3430 BB Nieuwegein, The Netherlands Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV, Amsterdam, The Netherlands

b

article info

abstract

Article history:

The detection of many new compounds in surface water, groundwater and drinking water

Received 17 May 2009

raises considerable public concern, especially when human health based guideline values

Received in revised form

are not available it is questioned if detected concentrations affect human health. In an

14 August 2009

attempt to address this question, we derived provisional drinking water guideline values

Accepted 21 August 2009

for a selection of 50 emerging contaminants relevant for drinking water and the water

Available online 26 August 2009

cycle. For only 10 contaminants, statutory guideline values were available. Provisional drinking water guideline values were based upon toxicological literature data. The

Keywords:

maximum concentration levels reported in surface waters, groundwater and/or drinking

Risk assessment

water were compared to the (provisional) guideline values of the contaminants thus

Drinking water

obtained, and expressed as Benchmark Quotient (BQ) values. We focused on occurrence

(Provisional) drinking water

data in the downstream parts of the Rhine and Meuse river basins. The results show that

guideline value Threshold of Toxicological Concern (TTC)

for the majority of compounds a substantial margin of safety exists between the maximum concentration in surface water, groundwater and/or drinking water and the (provisional) guideline value. The present assessment therefore supports the conclusion that the majority of the compounds evaluated pose individually no appreciable concern to human health. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

Due to anthropogenic activities, freshwater systems worldwide are confronted with thousands of compounds. In the European Union, for example, there are more than 100 000 registered chemicals (EINECS), of which 30 000–70 000 are in daily use. About 300 million tons of synthetic compounds annually used in industrial and consumer products, partially find their way to natural waters (Schwarzenbach et al., 2006). A major contribution to chemical contamination originates from wastewater discharges that impact surface water quality

with incompletely removed organic contaminants (Kolpin et al., 2004; Snyder et al., 2001). Additional contamination comes from diffuse agricultural activities, in which over 140 million tons of fertilizers and several million tons of pesticides are applied each year, and from atmospheric deposition. Such contamination can become an increasing problem for drinking water supplies, especially since the European REACH legislation may drive producers to develop newly designed less lipophilic/bioaccumulative chemicals that will be inherently more difficult to remove by traditional drinking water treatment techniques.

* Corresponding author. Tel.: þ31 30 6069564. E-mail address: [email protected] (M. Schriks). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.08.023

462

water research 44 (2010) 461–476

Recently, Loos et al. (2009) presented an EU-wide monitoring study on 35 organic compounds in European river waters in concentrations up to 40 mg/L. In addition, we have shown the occurrence of emerging polar contaminants, such as benzotriazoles and metabolites of illicit drugs (e.g. benzoylecgonine, desalkylflurazepam and 9-carbonic acid-d9-THC) in groundwaters and surface waters in the Netherlands (Hogenboom et al., 2009; Van Leerdam et al., 2009). Many of these emerging contaminants raise considerable toxicological and public concern, especially when human health based guideline values are unavailable. At present, the World Health Organization (WHO) and the US Environmental Protection Agency (US EPA) have derived approximately 125 statutory guideline values for contaminants in drinking water (Cotruvo, 1988; US EPA, 2006; WHO, 2006). However, the potential health effects of many emerging contaminants present in the water cycle and the potential human health concern associated with direct water ingestion have not been evaluated and statutory standards are not available. Therefore, we proposed earlier to assess the potential human health concern of unknown non-genotoxic compounds (lacking structural alerts that raise concern for potential genotoxicity) by comparing the environmental or drinking water concentration to a TTC (Threshold of Toxicological Concern) derived target value (Mons et al., 2008). The TTC concept was developed in the context of food safety to obtain a first clue on risks of unregulated chemicals present at low levels. The TTC thus derived is based on the molecular structure of the chemical involved and its related mode of toxic action (Munro et al., 1996). Assuming a daily intake of 2 l/day of drinking water, and a maximum contribution of 10% from drinking water to the total exposure – both of which are standard assumptions for deriving drinking water-quality guidelines (WHO, 2006) – TTC based target values proposed for drinking water are 0.1 mg/L for non-genotoxic compounds and 0.01 mg/L for genotoxicants (Mons et al., 2008). This value is based on a TTC level of 1.5 mg/person/d (0.15 mg/person/d for substances containing structural alerts that raise concern for potential genotoxicity with an acceptable lifetime cancer risk of 106) for compounds in food (Kroes et al., 2004). Since the TTC based value is rather conservative, an over-estimation of the actual risk may be the result. For emerging contaminants, a more profound human health based assessment may therefore be very valuable. The objectives of the present study were twofold. The first objective was to collect existing drinking water guideline values for a selection of 50 emerging contaminants relevant for the water cycle. If existing guideline values were not available, provisional guideline values were derived with the aid of relevant toxicological literature data. The second aim was to compare the maximum concentration levels reported in surface water, groundwater and/or drinking water to the (provisional) guideline values of the contaminants thus obtained, and express this as a Benchmark Quotient (BQ) value (further abbreviated as ‘‘BQ value’’). The present study does not attempt to quantify mixture interactions, since for compounds with an unknown mode of action there is no accepted methodology for such an assessment.

2.

Materials and methods

The toxicological assessment of the compounds presented in this paper comprises of a tiered approach in five consecutive steps (Fig. 1). First, the compounds to be assessed were selected. Second, n-octanol–water partition coefficients (log Kow) were obtained and compounds with a log Kow > 3 were excluded from further assessment. This log Kow cut off value is applied as a default threshold; for compounds with a log Kow above 3 it is less likely that they pass drinking water treatment plants (Westerhoff et al., 2005). Third, if available, statutory drinking water guideline values were obtained from websites of competent authorities; else provisional guideline values were derived with the aid of toxicological data relevant for humans as reported in literature. Fourth, measured maximum surface water, groundwater and/or drinking water concentrations were obtained from various sources and compared to (provisional) guideline values. Finally, a BQ value was calculated from the maximum concentrations reported and the (provisional) drinking water guideline values obtained. These steps are described in more detail below.

2.1.

Selection of compounds for assessment

A priority list representing a broad range of chemical classes was formulated with more than 100 compounds of interest. The arguments for inclusion were (i) questions related to toxicity posed by Dutch drinking water companies, (ii) potential low removal efficiency during drinking water production, (iii) appearance in recent literature and (iv) occurrence in surface waters, groundwaters and drinking water as determined by ourselves and others in various screening studies.

2.2.

Collection of compound-specific data

2.2.1.

n-Octanol–water partition coefficients (log Kow)

All log Kow values were obtained with the aid of the estimation program KOWWIN (US EPA, v1.67). An exception was made for perfluoroctane sulfonate (PFOS) and perfluoroctanoic acid (PFOA), for which accurate log Kow values cannot be calculated with estimation software. For these compounds the log Kow values were obtained from a database (Krop and de Voogt, 2008).

2.2.2.

Toxicological data

As illustrated in Fig. 1 (step 3), the first step was to obtain existing statutory drinking water guideline values from e.g. the US EPA (URL1) and the WHO (URL2). If not available, the second step was to obtain an established (by an (inter)national organization) Tolerable Daily Intake (TDI), Acceptable Daily Intake (ADI) or Reference Dose (RfD) and subsequently a provisional drinking water guideline value was derived as further described in Section 2.3. If not available, in a third step toxicity data collection focused primarily on established (by an (inter)national organization) lowest/no observed (adverse) effect levels (LO/NO(A)ELs) and subsequently a TDI was calculated as further described in Section 2.3. Finally, in a fourth step, miscellaneous toxicological information was collected and a TDI was calculated accordingly. In the case of insufficient human relevant toxicological data the

water research 44 (2010) 461–476

463

Fig. 1 – Flow diagram of the assessment conducted in the present study. Abbreviations: log Kow, n-octanol–water partition coefficient; GLV, drinking water guideline value; ADI, acceptable daily intake; RfD, reference dose; TDI, tolerable daily intake; LO/NO(A)EL, lowest/no observed (adverse) effect level; numbers correspond to consecutive steps as described Section 2. Compound categories: see Section 2.2.2.

compound of interest was not further evaluated and removed from the list. To facilitate the interpretation for which compounds the toxicity database is strong and less strong, all compounds were categorized (Table 2); (A) representing compounds with a statutory drinking water guideline value, (B) representing compounds with an established TDI, ADI or RfD, (C) representing compounds for which the TDI was calculated with an established LO(A)EL or NO(A)EL and (D) representing compounds for which the TDI was calculated with miscellaneous toxicological information. TDIs, ADIs, RfDs and/or other chronic toxicity data were sourced from peer-reviewed scientific papers and from other sources such as ‘‘grey literature’’. In addition, a literature search was performed on the internet and/or toxicological relevant data were obtained from the US EPA IRIS database, the European Chemicals Bureau (ECB), the Organization for Economic Cooperation and Development (OECD), the Dutch National Institute for Health and the Environment (RIVM), the US Food and Drug Administration (US FDA), the Joint Meeting FAO/WHO Meetings on Pesticide Residues (JMPR), the Dutch Expert Committee for Occupational Standards (DECOS), the Dutch board for authorization of plant protection products and biocides (CTGB), the Scientific Committee on Occupational Exposure Limits (SCOEL), the US National Toxicology Program (NTP), the Joint FAO/WHO Committee on Food Additives (JECFA), the Food and Agriculture Organization (FAO), the Danish veterinary and food administration, the European Union (EU), the US National Research Council (NRC), the EFSA scientific panel on contaminants in the food chain (CONTAM) and the Hazardous Substance Data Bank (HSDB).

2.2.3.

Occurrence data

Collection of occurrence data focused primarily on maximum concentrations of compounds measured in the downstream parts of the Rhine and Meuse river basins during the past decade. If not available, maximum concentrations in other surface waters and/or groundwaters were sought. The primary source of occurrence data of compounds in surface waters were the annual reports of the Dutch Association of River Water Companies (RIWA) and the German Association of River Water Companies (ARW). Occurrence data of compounds in drinking water were obtained from the REWAB data set (restricted water-quality data from the Dutch water companies). If not available, alternative resources were searched such as ‘‘grey literature’’, unpublished data, or publicly available sources such as RIVM, the Dutch ministry of transport, public works and water management (Rijkswaterstaat) and the WHO. Additional data on the occurrence of compounds was obtained from peer-reviewed scientific papers and an internet based literature search.

2.3. Derivation of provisional drinking water guideline values A drinking water guideline value represents the concentration of a constituent that does not exceed tolerable risk to the health of the consumer over a lifetime (WHO, 2006). In some cases, an odour-threshold value may be much lower than the health based guideline value. To calculate a provisional health based guideline value, the general methodology was applied

464

water research 44 (2010) 461–476

Table 1 – List of compounds assessed in the present study. No.

Compound 2

CAS

log Kow

No.

Compound

1071-83-6 138261-41-3

4.0a 0.6a

Iohexol1

66108-95-0

3.1a

30

Iomeprol (iomeron)1

78649-41-9

1.4b

1.4b 2.3a 2.1a 2.0a 1.4a 2.5a 2.5a 1.5a

31 32 33 34 35 36 37 38

62883-00-5 73334-07-3 34123-59-6 1634-04-4 37350-58-6 3622-84-2 111991-09-4 62-75-9

2.4a 2.1a 2.9a 0.9a 1.9a 2.3b 1.2b 0.6a

1698-60-8 882-09-7

1.1a 2.6a

39 40

94-75-7

2.8a

41

84-66-2 134-62-3 109-89-7 111-96-6

2.4a 2.2a 0.6a 0.4a

42 43 44 45

Iopamidol1 Iopromide1 Isoproturon3 Methyl tert-butyl ether (MTBE)4 Metoprolol6 n-Butylbenzenesulphonamide2 Nicosulfuron3 n-Nitrosodimethylamine (NDMA)2 p,p0 -Sulfonyldiphenol2 Perfluoroctanesulfonate (PFOS) (potassium-salt)5 Perfluorooctanoic acid (PFOA)5 Phenazone6 Simazine3 Sulfamethoxazole6 Tolyltriazole2

67-43-6

4.2b

46

87674-68-8

2.2a

47

Dimethylamine (DMA)2

124-40-3

0.4a

48

24

Diuron3

330-54-1

2.7a

49

25

Ethyl tert-butyl ether (ETBE)4

637-92-3

1.9b

50

2 3

6 7 8 9 10 11 12 13

2,6-Dichlorobenzamide (BAM)3 4-Methylbenzenesulfonamide (p-toluenesulfonamide, 4-tolylsulfonamide)2 Acetylsalicylate (aspirin, acetyl salicylic acid)6 Alpha-amino-3-hydroxy5-methyl-4-isoxazole propionic acid (AMPA)3 Amidotrizoic acid (diatrizoic acid)1 Bentazone3 Benzene2 Benzothiazole2 Benzotriazole (1H-benzotriazole)2 Bis(2-chloroisopropyl)ether (BCIPE)2 Carbamazepine6 Carbendazim3

14 15

Chloridazon (pyrazon)3 Clofibric acid6

16

Dichlorophenoxyacetic acid (2,4-D)3

17 18 19 20

22

Diethyl phthalate2 Diethyl toluamide (DEET)3 Diethylamine (DEA)2 Diethylene glycol dimethyl ether (diglyme, bis(2-methoxy ethyl)ester))2 Diethylene triamine penta acetic acid (DTPA)2 Dimethenamid3

23

4 5

21

2008-58-4 70-55-3

0.8a 0.9b

27 28

Ethylenediamine tetra acetic acid (EDTA)2 Glyphosate3 Imidacloprid3

50-78-2

1.2a

29

77521-29-0

2.2b

117-96-4 25057-89-0 71-43-2 95-16-9 95-14-7 108-60-1 298-46-4 10605-21-7

Log Kow 3.9b

1,4-Dioxane

26

CAS 60-00-4

1

a

123-91-1

0.3

80-09-1 2795-39-3

1.7b 1.08c

335-67-1

2.8c

60-80-0 122-34-9 723-46-6 29385-43-1

0.4a 2.2a 0.9a NA

Trichloroethene2

79-01-6

2.4a

Triethylphosphate (ethylphosphate) (TEP)2 Triphenylphosphine oxide (TPPO)2 Tris(2-chloroethyl) phosphate (TCEP)2 Urotropine (methenamine, hexamine)2

78-40-0

0.8a

791-28-6

2.8a

115-96-8

1.4a

100-97-0

4.2b

NA not available. Chemical categories: 1Iodinated contrast media; 2Miscellaneous organic compounds; 3Miscellaneous pesticides; 4Oxygenated gasoline additives; 5Perfluorinated organic compounds; 6Pharmaceuticals. a log Kow values were derived from US EPA’s KOWWIN experimental database. b log Kow values were calculated with the aid of US EPA’s KOWWIN. c log Kow values were obtained from Krop and de Voogt (2008).

as described by Van Leeuwen (2000) and the WHO (2006). For compounds without a statutory drinking water guideline value, first the Tolerable Daily Intake (TDI) was determined. The point of departure (POD) for calculating the TDI was mostly a chronic LO(A)EL, NO(A)EL, benchmark dose level, maximum tolerated dose (MTD) or lowest effective safe dose. In case only inhalatory toxicity data could be found, a routeto-route extrapolation was carried out according to toxicological methods as described by Stokinger and Woodward (1958). An appropriate safety factor to extrapolate between species (inter-species differences), inter-individual differences (intraspecies differences), exposure route/duration and quality of the data was utilized as part of the TDI calculation (Van Leeuwen and Vermeire, 2007). Secondly, a drinking water equivalent level (DWEL) was calculated by multiplying the TDI

by a typical average body weight of 70 kg and division by a daily water consumption of 2 l. Finally, to account for the fraction of the TDI allocated to drinking water, the DWEL was multiplied by an allocation factor to give the provisional guideline value. In most cases, when there was insufficient exposure information to derive chemical-specific allocation factors, a default allocation factor of 10% was used.

2.4. Evaluation of water-quality data in the context of human health Of each compound the maximum concentration level reported in surface waters, groundwater and/or drinking water was compared to its (provisional) guideline value and was expressed as a BQ value (concentration in water divided by

Table 2 – Parameters used for derivation of (provisional) drinking water guideline values.

1,4-Dioxane

POD is an oral slope factor as derived by US EPA (1988a) of 0.011 per mg/kg bw/d from a 110-week study in rats (NCI, 1978). POD is a NOAEL of 4.5 mg/kg bw/d for decreased body weight in both sexes and increased liver weight in males as derived by the Danish EPA (2004) from a study in which dogs were fed BAM in the diet for 2 years (Wilson and Thorpe, 1971), with an uncertainty factor of 300 (100 for inter- and intraspecies variation and 3 for uncertainties in the dataset). POD is a NOAEL of 300 mg/kg bw/d for reproductive effects (decrease lactation index and litter weight at birth) as derived from a GLP compliant study in which rats were administered 4methylbenzenesulfonamide via oral gavage for 42 days (OECD/SIDS, 1994), with an uncertainty factor of 400 (100 for inter- and intraspecies variation and 4 for extrapolation to chronic exposure). POD is a LOEL of 10 mg/person from a human study as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month toxicity study in rats (study reference unknown). POD is the highest therapeutic dose of 50 mg/person/d (0.71 mg/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 9 mg/kg bw/d as derived by the WHO (1998) from a 2-year dietary toxicity study in rats (study reference unknown). POD is a risk estimate as derived by the WHO (2003a) from a 2-year gavage study in rats and mice (NTP, 1986). POD is a NOEL of 5.1 mg/kg bw/d as derived by the WHO/JECFA (2003) from a study in which rats were administered benzothiazole in the diet for 90 days (Morgareidge, 1971). Daily observations revealed no treatment related effects in histopathological/haematological parameters, body weight, food consumption and liver/kidney weights. An uncertainty factor of 200 was applied (100 for inter- and intraspecies variation and 2 for extrapolation to chronic exposure). POD is a LOAEL of 295 mg/kg bw/d for histological changes in the liver, decreased body weight gain and inflammation of the prostate/uterus as derived by the Dutch Expert Committee for Occupational Standards (DECOS, 2000) from a study in which rats were administered benzotriazole in the diet for 78 weeks (BIBRA Toxicology International, 1995), with an uncertainty factor of 1000 (100 for inter- and intraspecies variation and 10 for extrapolation of a LOAEL to a NOAEL). POD is a NOAEL of 35.8 mg/kg bw/d as derived by the US EPA (1989) from a 24-month chronic toxicity study in mice (Mitsumori et al., 1979). POD is a maximum tolerated dose (MTD) of 250 mg/kg bw/d as derived by Snyder et al. (2008) from a 2year study in rats showing evidence of carcinogenicity (Singh et al., 2005). POD is a NOAEL of 2.5 mg/kg bw/d as derived by the WHO/JECFA (1995) from a 2-year study in dogs (Sherman, 1972). POD is a NOAEL of 5.4 mg/kg bw/d for adverse histopathological/haematological changes, decreased food intake, lower body weight gain and higher organ weights (liver, kidney, thyroid gland) as derived from a study in which rats were orally administered chloridazon (method of administration unspecified) for 7 weeks (ECB, 2000b), with an uncertainty factor of 100 (inter- and intraspecies variation). POD is a LOEL of 1 mg/kg bw/d as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007) from an 8-week oral study in humans (Larsen et al., 1994).

2,6-Dichlorobenzamide (BAM)

4-Methylbenzenesulfonamide (p-toluenesulfonamide, 4-tolylsulfonamide) Acetylsalicylate (aspirin, acetyl salicylic acid) Alpha-amino-3-hydroxy-5-methyl4-isoxazole propionic acid (AMPA) Amidotrizoic acid (diatrizoic acid) Bentazone Benzene Benzothiazole

Benzotriazole (1H-benzotriazole)

Bis(chloroisopropyl)ether (BCIPE) Carbamazepine Carbendazim Chloridazon (pyrazon)

Clofibric acid

Categorya SUFb TDI, ADI or RfD (Provisional) (mg/kg bw/d) guideline value (mg/L) 30d

B

NA

NA

C

300

0.015

D

400

0.75

2600

B

20

0.007

25

A

100

0.3

900c

B

10

NA

250 000

A

100

0.1

300c

A

NA

NA

10c,d

C

200

0.026

90

C

1000

0.295

1000

B

1000

0.04

140

B

NA

0.00034

B

100

0.03

105

D

100

0.054

189

B

100

0.01

30

52.5

1

(continued on next page)

465

Point of departure (POD)

water research 44 (2010) 461–476

Compound

Compound

Dichlorophenoxyacetic acid (2,4-D) Diethyl phthalate Diethyl toluamide (DEET)

Diethylamine (DEA)

Diethylene triamine penta acetic acid (DTPA)

Dimethenamid Dimethylamine (DMA)

Diuron Ethyl tert-butyl ether (ETBE)

Ethylenediamine tetra acetic Acid (EDTA) Glyphosate Imidacloprid Iohexol

POD is a NOAEL of 1 mg/kg bw/d as derived by the WHO (2003b) from a 1-year study of toxicity in dogs and a 2-year study of toxicity and carcinogenicity in rats (study reference unknown). POD is a NOAEL of 750 mg/kg bw/d as derived by US EPA (1988b) from a 16-week toxicity study in rats (Brown et al., 1978). POD is a NOEL of 100 mg/kg/bw/d based on clinical signs, reduced haemoglobin/haematocrit levels and histological changes in liver, lymph nodes and uterus as derived by the California Environmental Protection Agency (California EPA, 2000) from a study in which beagle dogs were orally administered DEET (gelatin capsules) for 1 year (Goldenthal, 1994), with an uncertainty factor of 56 (14 for inter- and intraspecies variation and 4 for extrapolation to chronic exposure). POD is a LOAEL of 75 mg/m3 for reduced mean body weights and adverse histopathological effects (lesions of the nasal mucosa) as derived by the scientific committee on occupational exposure limits (SCOEL, 2002) from a study in which rats were exposed to DEA via the inhalatory route for 24 weeks (6.5 h/d, 5d/wk) (Lynch et al., 1986), with an uncertainty factor of 50 (5 for the absence of human data and a NOAEL and 10 for route-to-route extrapolation uncertainties). POD is a NOAEL 25 mg/kg bw/d for developmental effects (adversely affected implants per liter and decreased weight gain) as derived by the WHO (2002) from a study in which rabbits were administered diglyme via oral gavage for 13 days (NTP, 1987), with an uncertainty factor of 500 (100 for inter- and intraspecies variation and 5 for uncertainties in the dataset). POD is a NOAEL of 100 mg/kg bw/d for developmental effects (increased fetal deformations) as derived from a study according to OECD guideline 414 in which rats were administered DTPA (in its sodium form) via oral gavage during day 6–15 of pregnancy (ECB, 2000c), with an uncertainty factor of 1000 (100 for inter- and intraspecies variation and 10 for extrapolation to chronic exposure). POD is a NOAEL of 7 mg/kg bw/d as derived by the WHO/JMPR (2005) from a 24-month study in rats given diets containing racemic dimethenamid (study reference unknown). POD is a LOAEL of 19 mg/m3 for concentration-related lesions in the respiratory/olfactory mucosa as derived by the scientific committee on occupational exposure limits (SCOEL, 1991) from a study in which rats and mice were exposed to DMA via the inhalatory route for 2 years (6 h/d, 5d/wk) (CIIT, 1990), with an uncertainty factor of 50 (5 for the absence of human data and a NOAEL and 10 for routeto-route extrapolation uncertainties). POD is a NOEL of 0.625 mg/kg bw/d as derived by US EPA (1988c) from a 2-year feeding study in dogs (DuPont, 1964). POD is a NOAEL of 500 ppm (29.1 mg/kg bw/dh) for testes degeneration as derived from a study in which rats were exposed to ETBE via the inhalatory route for 13 weeks (Medinsky et al., 1999), with an uncertainty factor of 200 (100 for inter- and intraspecies variation and 2 for extrapolation to chronic exposure). POD is a NOAEL of 250 mg/kg bw/d (190 mg/kg bw/d as the free acid) as derived by the WHO/JECFA (1973) from a 2-year toxicity study in rats (study reference unknown). POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month study of toxicity in rats fed technical-grade glyphosate (study reference unknown). POD is a NOAEL of 5.7 mg/kg bw/d as derived by the WHO/JMPR (2001) from a 2-year study of toxicity and carcinogenicity in rats (study reference unknown). POD is a safe dose of 75 g/person/d (1.07 g/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007).

Categorya SUFb TDI, ADI or RfD (Provisional) (mg/kg bw/d) guideline value (mg/L) A

100

0.01

30c

B

1000

0.8

2800

C

56

1.8

6250

C

50

2.14f

750

C

500

0.05

175

D

1000

0.1

350

B

100

0.07

245

C

50

0.54g

190

B

300

0.002

7

D

200

0.15

525e

A

100

1.9

600b,c

A

100

0.3

900c

B

100

0.06

210

B

10

NA

375 000

water research 44 (2010) 461–476

Diethylene glycol dimethyl ether (diglyme, bis(2-methoxy ethyl)ester))

Point of departure (POD)

466

Table 2 (continued)

Iomeprol (iomeron)

Iopamidol Iopromide Isoproturon Methyl tert-butyl ether (MTBE)

Metoprolol n-Butylbenzenesulphonamide

p,p0 -Sulfonyldiphenol

Perfluoroctane sulfonate (PFOS) Perfluorooctanoic acid (PFOA)

Phenazone Simazine Sulfamethoxazole Tolyltriazole

D

2100

1.9

6700

B

10

NA

415 000

B

10

NA

250 000

A

1000

0.003

B

1000

0.3

B

100

0.014

50

D

600

0.083

292

B A

1000 NA

0.2 NA

D

600

0.017

B

200

0.00015

0.5

B

200

0.0015

5.3

B

100

0.036

125

A

1000

0.52

2c

B

200

0.13

440

D

600

0.25

875

9c 9400e

700 0.1c,d

60

467

(continued on next page)

water research 44 (2010) 461–476

Nicosulfuron n-Nitrosodimethylamine (NDMA)

POD is a NOEL of 2 g Iodine/kg bw/d (equal to 4 g iomeprol/kg bw/d) for adverse effects on liver and kidney (non-lipid cytoplasmic vacuolization of hepatocytes and renal tubular epithelium cells) as derived from a study in which dogs were intravenously exposed to iomeprol for 28 days (Morisetti et al., 1994), with an uncertainty factor of 2100 (35 for inter- and intraspecies, 10 for route-to-route extrapolation and 6 for extrapolation to chronic exposure). POD is a safe dose of 83 g/person/d (1.19 g/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a safe dose of 50 g/person/d (0.71 g/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 3 mg/kg bw/d as derived by the WHO (2003c) from a 90-day study in dogs and a 2-year feeding study in rats (study reference unknown). POD is a NOAEL of 300 mg/kg bw/d as derived by the Dutch National Institute for Health and the Environment (RIVM) (Swartjes et al., 2004) from a 90-day oral toxicity study in rats (Robinson et al., 1990). POD is a LOEL of 100 mg/person/d (1.42 mg/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 50 mg/kg bw/d for adverse treatment related macro- or microscopic effects (liver enlargement, hepatocyte hypertrophy, thymic atrophy and lymphocytolysis) as derived from a study according to OECD guideline 407 (GLP compliant) in which rats were administered nbutylbenzenesulphonamide via oral gavage for 28 days (Proviron Fine Chemicals, 2003), with an uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic exposure). POD is a NOAEL of 199 mg/kg bw/d as derived from a 2-year study in rats (URL3). POD is a tumor dose (TD05, dose level that causes 5% in increase in tumour incidence over the background) of 18 mg/kg bw/d as derived by the WHO (2008) from a detailed 2-year cancer dose– response study in rats (Peto et al., 1991a,b). POD is a NOEL of 10 mg/kg bw/d for histopathological effects (hyperplasia of the mucosal epithelium, hypertrophy of hepatocytes), decreased food consumption/body weight gain and increased liver weights as derived from a study according to OECD guideline 421 in which male and female rats were administered p,p0 -sulfonyldiphenol via oral gavage for respectively 45 days and from 14 days before mating to day 3 of lactation (Mitsubishi Chemical Safety Institute Ltd, date of study unknown), with an uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic exposure). POD is a NOAEL of 0.03 mg/kg bw/d as derived by the Scientific Panel on Contaminants in the Food Chain (CONTAM) (EFSA, 2008) from a 182-day study in Cynomolgus monkeys (Seacat et al., 2002). POD is a benchmark dose level (BMDL10) of 0.3 mg/kg bw/d as derived by the Scientific Panel on Contaminants in the Food Chain (CONTAM) (EFSA, 2008) from a number of studies in mice and male rats (study references unknown). POD is a LOEL of 250 mg/person/day (3.57 mg/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 0.52 mg/kg bw/d as derived by the WHO (1996) from a 2-year combined chronic toxicity/oncogenicity study in rats (Ciba-Geigy, 1988; unpublished study submitted to WHO). POD is a NOAEL of 25 mg/kg/d as derived by Schwab et al. (2005) from a 60-week study in rats (Swarm et al., 1973). POD is a NOAEL of 150 mg/kg bw/d for observed mild apathy as derived from a study in which rats were administered tolyltriazole via oral gavage for 29 days (Benzotriazoles Coalition, 2001; ECB, 2000a), with an uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic exposure).

Compound

Trichloroethene Triethylphosphate (ethylphosphate) (TEP)

Triphenylphosphine oxide (TPPO)

Tris(2-chloroethyl)phosphate (TCEP)

POD is a benchmark dose level (BMDL10) of 0.146 mg/kg bw/d as derived by the WHO (2005b)/Health Canada (2003) from a developmental toxicity study in rats (Dawson et al., 1993). POD is a NOEL of 335 mg/kg bw/d for fertility effects (effects on litter size) as derived from a study in which rats were administered TEP via the food for a unknown period (OECD/SIDS, 1998), with an uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic exposure). POD is a NOAEL of 8 mg/kg bw/d for salivation, vomiting, diarrhea, histopathological (liver damage and skeletal muscle atrophy)/haematological (elevated GPT, GOT and alkaline phosphatase activities, reduced haemoglobin/haematocrit levels) parameters as derived from a study in which dogs were administered TPPO via the food for 3 months (ECB, 2000d), with an uncertainty factor of 1000 (100 for inter- and intraspecies variation and 10 for extrapolation to chronic exposure). POD is a NOAEL of 22 mg/kg bw/d for increased relative liver and kidney weights as derived from a study in which rats were administered TCEP via oral gavage for 16 weeks (NTP, 1991), with an uncertainty factor of 1000 (10 for inter- and intraspecies variation and 10 for extrapolation to chronic exposure and uncertainty in genotoxic potential). POD is a NOAEL of 15 mg/kg bw/d as derived by the WHO/JECFA (1974) from a teratogenicity study (exposure from the fourth to fifty-sixth day after mating) in dogs (Hurni and Ohder, 1973).

Categorya SUFb TDI, ADI or RfD (Provisional) (mg/kg bw/d) guideline value (mg/L) 20c

A

100

0.0015

D

600

0.56

1950

D

1000

0.008

28

D

1000

0.022

77

B

100

0.15

500

NA not available. a Categories: A) statutory drinking water guideline value available; B) established TDI, ADI or RfD available; C) TDI calculated with a established LO(A)EL or NO(A)EL; D) TDI calculated with miscellaneous toxicological information. b Uncertainty factors. c WHO drinking water guidelines (WHO, 2006). d Based on a specific cancer risk level of 105. e The odour-threshold value for drinking water preparation is 15 mg/L for MTBE and w1 mg/L for ETBE (Swartjes et al., 2004; Van Wezel et al., 2009). f Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 150 mg/person/d (2.14 mg/kg bw/d) can be calculated from the 8-h threshold limit value (15 mg/m3) assuming 100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m3. g Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 40 mg/kg/person/d (0.54 mg/kg bw/d) can be calculated from the 8-hour threshold limit value (3.8 mg/m3) assuming 100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m3. h A TDI of 0.15 mg/kg bw/day can be calculated from the NOAEL (500 ppm ¼ 29.1 mg/kg/bw/d) assuming 100% oral/inhalatory absorption, a specific lung retention of 25% and a minute ventilation volume for rat of 45 mL.

water research 44 (2010) 461–476

Urotropine (methenamine, hexamine)

Point of departure (POD)

468

Table 2 (continued)

water research 44 (2010) 461–476

guideline value) in order (i) to provide perspective on what the occurrence of emerging contaminants might signify to human health and (ii) to help prioritize further investigations. A BQ value of 1 represents a (drinking) water concentration equal to the (provisional) guideline value. Compounds with a BQ value of 1 in drinking water may be of potential human health concern if the water were to be consumed over a lifetime period. Compounds with a BQ value 0.1 in drinking water were identified as those that may warrant further investigation; this is consistent with various US State and Federal practices (Toccalino, 2007). For compounds found in surface waters and groundwater the BQ value threshold to carry out an additional assessment was set at an arbitrary value of 0.2, since these source waters are purified in drinking water treatment plants which provides extra safety. Compounds in surface waters/groundwater or drinking water with a BQ value of 0.2 or 0.1 respectively, are presumed to present no appreciable concern to human health.

3.

Results

3.1.

Selection of compounds

For only 50 compounds out of the original list, statutory guideline values or useful toxicity and occurrence data could be found. These compounds constitute the final list, which includes compounds from various groups such as iodinated contrast media, pharmaceuticals, oxygenated gasoline additives, perfluorinated organic compounds, miscellaneous organic compounds and pesticides (Table 1). Natural and synthetic steroid hormones such as 17b-estradiol, 17a-ethynylestradiol and estrone were not included in this assessment, as they are removed relatively easily in drinking water purification processes (Nghiem et al., 2004).

3.2.

(Provisional) drinking water guideline values

For 10 compounds WHO statutory drinking water guideline values were available and these compounds were classified as category A. For the remaining 40 compounds a provisional guideline value was established with the aid of toxicological data. An established TDI, ADI or RfD was available for 22 compounds (category B). In 7 cases when there was no TDI, ADI or RfD available, an established NO(A)EL or LO(A)EL was used to calculate a TDI and subsequently a provisional drinking water guideline value (category C). For the remaining 11 compounds, miscellaneous toxicological data was used to calculate a TDI and subsequently a provisional drinking water guideline value (category D). As tabulated in Table 2, (provisional) guideline values ranged from 0.0001 mg/L for NDMA to 415 mg/L for the iodinated contrast medium iopamidol. All iodinated contrast media had relatively high provisional guideline values, ranging from 6.7 mg/ L (iomeprol) to 415 mg/L (iopamidol). In the cases of MTBE and ETBE, the human health based guideline values were at least one order of magnitude higher than the corresponding odour-threshold based guideline values of respectively 15 mg/ L and w1 mg/L (Swartjes et al., 2004; Van Wezel et al., 2009). For two compounds (DMA and DEA) the provisional guideline

469

values were established by route-to-route extrapolation of inhalatory LOAELs to oral LOAELs. Since benzene was evaluated to be genotoxic/carcinogenic (IARC group I) and NDMA (IARC group 2A) and 1,4-dioxane (IARC group 2B) are respectively suspected non-genotoxic and genotoxic carcinogens, their corresponding (provisional) guideline values are provided as an upper bound lifetime cancer risk to an individual of 105, or the odds that one case of cancer would result for every 100 000 persons subjected to continuous exposure over a 70-year lifetime.

3.3. Concentration of compounds in surface waters, groundwaters and drinking water The maximum concentrations of compounds reported in surface waters and/or groundwaters are summarized in Table 3. Measured maximum surface water concentrations were available for 37 of the 50 compounds in the annual reports of RIWA and ARW. For two compounds (MTBE and clofibric acid) maximum concentrations in Dutch groundwater are reported. For the remaining compounds, the maximum concentration reported in surface waters was taken from other sources (see Section 2.2.3). The six compounds with the highest reported maximum concentrations in surface waters were EDTA (29 mg/L), DTPA (12.2 mg/L), p,p0 -sulfonyldiphenol (10 mg/L), urotropine (10 mg/ L), 1,4-dioxane (10 mg/L) and AMPA (5 mg/L), whereas in groundwater a relatively high concentration was found for MTBE (27.3 mg/L) showing the environmental relevance of this compound. The highest maximum concentration of iodinated contrast media in surface waters was reported for iomeprol (0.97 mg/L). Table 3 also summarizes the maximum concentrations of compounds reported in drinking water. Data on the occurrence of compounds in drinking water were relatively scarce, and limited to 35 compounds. For 18 compounds, drinking water concentrations were obtained from the Dutch REWAB database and for 17 compounds drinking water concentrations were taken from reports by others. Drinking water concentrations for the remaining compounds could not be found. The highest maximum concentration reported was for EDTA (13.6 mg/L), followed by DTPA (9 mg/L), metoprolol (2.1 mg/L) and BCIPE (1.9 mg/L).

3.4. Comparison of compound concentrations to (provisional) guideline values (BQ value) For all compounds found in surface waters, groundwaters and drinking water the calculated BQ value was <1 (Table 3). The three compounds exhibiting the highest BQ values (i.e. posing the highest potential human health concern) in surface water are 1,4-dioxane, carbamazepine and PFOS (Fig. 2A). For MBTE and ETBE BQ values of respectively 1.8 and 1.2 can be calculated, when comparing their maximum concentrations reported in surface water to the odourthreshold based guideline values of 15 mg/L and w1 mg/L (not shown in Fig. 2). These BQ values do not indicate a concern for human health per se, but rather indicate that the maximum environmental concentration reported exceeds the odour-threshold. This implies that for MTBE and ETBE

Compound

Surface waters and groundwaters Max conc (mg/L) (number of measurements, year) 10 (NA, 1997) 0.05 (40, 2002-2006) 0.06 (20, 2005)

Ref

Drinking water BQ valuea

Max conc (mg/L) (number of measurements, year)

Source

Ref

BQ valuea

SW, NL SW, NL SW, NL

(4) (11) (13)

0.3 0.001 0.00002

0.5 (NA) 0.23 (14, 2002) NA

UDW, NL FDW, USA

(4) (10)

0.02 0.004

0.065 (NA, 2007)

SW, NL

(15)

0.003

0.12 (12, 2007)

FDW, NL

(15)

0.005

5 (499, 2005)

SW, NL

(11)

0.006

1.1 (6, 2001)

FDW, NL

(10)

0.001

SW, GER SW, NL SW, NL SW, NL SW, NL GW, NL SW, NL SW, BE SW, BE BFGW, NL SW, NL SW, NL SW, NL SW, NL SW, NL

(2) (11) (11) (6) (11) (6) (11) (11) (11) (15) (11) (11) (11) (11) (11)

0.000003 0.0003 0.07 0.0003 0.0005 0.02 0.2 0.01 0.002 0.003 0.007 0.0003 0.00001 0.0004 0.02

0.25 (6, 2006) 0.28 (11, 2006) 0.96 (12, 2005) 0.01 (10, 2007) 0.2 (10, 2007) 1.9 (9, 1982-1984) 0.03 (2, 2007) NA NA 0.14 (2, 2007) 0.11 (5, 2002) NA 0.03 (1, 2005) NA 0.15 (13, 2007)

FDW, NL FDW, NL TW, NL FDW, NL FDW, NL FDW, NL FDW, NL

(10) (10) (10) (14) (14) (6) (15)

0.000001 0.0009 0.1 0.0001 0.0002 0.01 0.03

FDW, NL TW, NL

(15) (10)

0.005 0.004

TW, NL

(10)

0.000005

UDW, NL

(1)

0.0009

SW, NL SW, NL SW, NL SW, BE SW, GER SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL GW, NL SW, NL SW, NL SW, NL SW, NL

(11) (9) (11) (11) (2) (11) (11) (11) (11) (11) (11) (11) (11) (7) (11) (6) (11) (11)

0.03 0.0005 0.002 0.1 0.002 (1.2b) 0.05 0.001 0.0003 0.000001 0.0001 0.000002 0.000002 0.03 0.003 (1.8b) 0.004 0.003 0.0002 0.07

9 (2, 2001) NA NA 0.08 (2, 2005) NA 13.6 (7, 2001) 0.46 (3, 2006) NA 0.06 (1, 2007) 0.01 (1, 2006) 0.1 (6, 2006) 0.04 (2, 2007) 0.02 (1, 2004) 1.25 (27, 2006) 2.1 (2, 2005) 0.05 (2, 2004) NA 0.002 (21, 2007)

FDW, NL

(10)

0.03

TW, NL

(10)

0.01

FDW, NL TW, NL

(10) (10)

0.02 0.0005

FDW, NL FDW, NL FDW, NL FDW, NL TW, NL FDW, NL FDW, NL TW, NL

(15) (10) (10) (15) (10) (10) (10) (13)

0.0000002 0.000001 0.0000002 0.0000002 0.002 0.0001 (0.08b) 0.04 0.0002

UDW, NL

(5)

0.02

0.63 0.1 0.74 0.03 0.54 2.9 0.227 1.5 0.3 0.091 0.2 0.9 0.06 0.29 3.64

(189, 2006) (126, 2007) (116, 2001) (3, 2008) (11, 2007) (15, 1984-1985) (263, 2003) (111, 2006) (68, 2002) (NA, 2007) (34, 2006) (8, 2005) (36, 2005) (38, 2007) (11, 2007)

12.2 (53, 2005) 0.12 (4, 2005) 0.34 (42, 2005) 0.68 (386, 2002) 1.2 (97, 2006) 29 (192, 2005) 1.2 (291, 2006) 0.06 (1, 2007) 0.5 (180, 2005) 0.97 (172, 2007) 0.714 (188, 2006) 0.56 (186, 2004) 0.31 (256, 2002) 27.3 (14, 2003-2005) 0.2 (114, 2006) 0.78 (300-400, 2004-2006) 0.17 (5, 2007) 0.0071 (38, 2006)

water research 44 (2010) 461–476

1,4-Dioxane 2,6-Dichlorobenzamide (BAM) 4-Methylbenzenesulfonamide (p-toluenesulfonamide, 4-tolylsulfonamide) Acetylsalicylate (aspirin, acetyl salicylic acid) Alpha-amino-3-hydroxy-5-methyl4-isoxazole propionic acid (AMPA) Amidotrizoic acid (diatrizoic acid) Bentazone Benzene Benzothiazole Benzotriazole (1H-benzotriazole) Bis(chloroisopropyl)ether (BCIPE) Carbamazepine Carbendazim Chloridazon (pyrazon) Clofibric acid Dichlorophenoxyacetic acid (2,4-D) Diethyl phthalate Diethyl toluamide (DEET) Diethylamine (DEA) Diethylene glycol dimethyl ether (diglyme, bis(2-methoxy ethyl)ester)) Diethylene triamine penta acetic acid (DTPA) Dimethenamid Dimethylamine (DMA) Diuron Ethyl tert-butyl ether (ETBE) Ethylenediamine tetra acetic Acid (EDTA) Glyphosate Imidacloprid Iohexol Iomeprol (iomeron) Iopamidol Iopromide Isoproturon Methyl tert-butyl ether (MTBE) Metoprolol n-Butylbenzenesulphonamide Nicosulfuron n-Nitrosodimethylamine (NDMA)

Source

470

Table 3 – Reported concentrations in surface waters, groundwaters and drinking water and comparison to (provisional) drinking water guideline values expressed as Benchmark Quotient (BQ) values.

Sources: bank-filtrated groundwater (BFGW); groundwater (GW); surface water (SW); finished drinking water (FDW); tap water (TW); unspecified drinking water (UDW). Locations: Belgium (BE); Germany (GER); the Netherlands (NL); United States of America (USA). References: (1) Anonymous data Dutch drinking water companies; (2) ARW database; URL4; (3) Brauch et al., 2000; (4) ECB, 2002; (5) Kleinneijenhuis and Puijker, 2008; (6) KWR internal data (KWR, 2009); (7) De Voogt et al., 2008; (8) Loos et al., 2009; (9) Mout et al., 2007; (10) REWAB database, 2009; (11) RIWA database; URL5 (12) Skutlarek et al., 2006; (13) van Beelen, E.S.E. (HWL, the Netherlands), personal communication; (14) Van Leerdam et al., 2009; (15) Versteegh et al., 2007. NA: Not available; an absence of measured concentrations above the detection limit. a BQ value: ratio of reported maximum concentration to the (provisional) guideline value (cf. Table 2). b Based on an odour-threshold value (cf. Table 2).

(6) FDW, NL

0.005

(10) FDW, NL

0.09

(12) (12) (15) (10) (15)

p,p0 -Sulfonyldiphenol Perfluoroctane sulfonate (PFOS) Perfluorooctanoic acid (PFOA) Phenazone Simazine Sulfamethoxazole Tolyltriazole Trichloroethene Triethylphosphate (ethylphosphate) (TEP) Triphenylphosphine oxide (TPPO) Tris(2-chloroethyl)phosphate (TCEP) Urotropine (methenamine, hexamine)

10 (300, 2005-2006) 0.11 (NA, 2008) 0.647 (NA, 2005/2006) 0.11 (26, 2005) 0.13 (124, 2006) 0.11 (170, 2005) 0.29 (11, 2007) 1.35 (206, 2006) 0.189 (25, 2007) 0.344 (88, 2003) 0.29 (8, 2006) 10 (NA, 1997-1998)

SW, BE SW, NL SW, GER SW, NL SW, BE SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL SW, GER

(6) (8) (13) (12) (11) (11) (11) (11) (11) (11) (11) (3)

0.1 0.2 0.1 0.0009 0.07 0.0003 0.0003 0.07 0.0001 0.01 0.004 0.02

NA 0.02 (NA, 2005/2006) 0.52 (NA, 2005/2006) 0.03 (8, 2005) 0.06 (4, 2004) 0.03 (4, 2007) NA 1.75 (8, 2005) NA 0.13 (6, 2007) NA NA

TW, GER TW, GER FDW, NL FDW, NL FDW, NL

0.04 0.1 0.0002 0.03 0.00007

water research 44 (2010) 461–476

471

there is a concern for drinking water production, because consumers will not accept odourous drinking water. The BQ values of the remaining compounds calculated for surface waters and groundwater ranged between 0.1 (p,p0 -sulphonylphenol, diuron) and 0.000001 (iohexol). For all iodinated contrast media the concentration in surface waters was at least three orders of magnitude less than the (provisional) guideline value. For drinking water, the BQ values of these compounds were even lower (Fig. 2B). For two compounds (benzene and PFOA) found in drinking water, the BQ value was equal to 0.1 indicating that additional assessments such as establishing trends may be warranted. However, for 15 compounds occurrence data in drinking water were not available and therefore the human health concern associated with drinking water consumption due to presence of any of these compounds remains unknown.

4.

Discussion

Rapid new developments in analytical chemistry lead to the detection and quantification of many emerging contaminants in drinking water and its environmental sources (surface water and groundwater). Since toxicological information is often absent, such compounds are a growing concern for drinking water companies and their customers. The present study attempts to address potential human health concern associated with water containing emerging contaminants. The 50 compounds included in this study represent a broad range of chemical classes for which maximum concentrations in surface waters, groundwater and/or drinking water were obtained in the downstream parts of the Rhine and Meuse basins. The results as presented in Fig. 2 indicate that a substantial margin exists between the (provisional) guideline value and the maximum concentrations of most compounds reported in surface waters, groundwaters and/or drinking water. The compounds evaluated with a relatively high BQ value (i.e. a high potential human health concern) and a known carcinogenic action are 1,4-dioxane, benzene and NDMA. The (provisional) guideline values for 1.4-dioxane (30 mg/L), benzene (10 mg/L) and NDMA (0.1 mg/L) used in the present study are based on a specific cancer risk level of 105. However, when applying a specific risk level of 106, as is common practice in the Netherlands, the provisional guidelines value would be 3 mg/L, 1 mg/L and 0.01 mg/L, respectively. This would result in BQ values much higher than the arbitrary thresholds for surface waters and drinking water employed in the present study. This indicates that very low concentrations of these compounds in drinking water could lead to a potential carcinogenic effect, and we conclude that for these compounds it is important to monitor trends in their (environmental) occurrence. Furthermore, Fig. 2 illustrates that the (provisional) guideline values of the majority of non-genotoxic compounds are at least two orders of magnitude above the Threshold of Toxicological Concern (TTC)-based drinking water target value for non-genotoxic compounds (0.1 mg/L). In addition, the (provisional) guideline values (expressed as a specific risk level of 106) of the three compounds with known carcinogenic action (1,4-dioxane, benzene and NDMA) are equal (NDMA) or much higher (1,4-

472

water research 44 (2010) 461–476

Fig. 2 – Comparison of compound concentrations in (A) surface/groundwaters and (B) drinking water to (provisional) guideline values. Benchmark Quotient (BQ) thresholds are indicated with dashed lines. Threshold of Toxicological Concern (TTC) based target value for non-genotoxic compounds (0.1 mg/L) is indicated with a dotted line. Numbers correspond to compounds as tabulated in Table 1.

dioxane and benzene) than the TTC derived target value of 0.01 mg/L for genotoxic compounds. This illustrates that the TTC based drinking water target value may be a conservative value ideally suited for exposure based waiving of compounds for which there is no sufficient toxicological information, which can be followed up by a more data-intensive evaluation. Two perfluorinated organic compounds were evaluated in the present study. For PFOA in drinking water a BQ value equal to the arbitrary threshold of 0.1 was calculated, whereas for PFOS in surface waters a BQ value of 0.2 was calculated. These persistent compounds are becoming a global problem, and PFOA and PFOS have already been detected in the ng/L range in, e.g. European and Japanese tapwaters (Ericson et al., 2007; Loos et al., 2007; Norimitsu et al., 2004). Recently, Skutlarek et al. (2006) observed at sampling site Neheim (river Ruhr catchment, a tributary of the river Rhine, Germany)

concentrations of PFOA of 0.65 mg/L in Lake Moehne, and 0.53 mg/L in corresponding drinking water, respectively. The authors concluded that water treatment steps may not effectively eliminate perfluorinated compounds to a sufficient extent, although approximately 50% of the waterworks at the Ruhr river are equipped with activated carbon filters. Hence, more research should be devoted to the behavior of perfluorinated organic compounds in drinking water treatment processes. Several structurally related iodinated contrast media (iopamidol, iohexol, iomeprol and iopromide) were evaluated in the present research. However, their calculated BQ values are much lower than the BQ threshold above which further investigations would be warranted. Iopromide, for example, is a relatively non-toxic compound with a reported safe dose (intravenous) of 50 g/person/d (Versteegh et al., 2007). Despite

water research 44 (2010) 461–476

the absence of human health effects, these compounds may deserve further attention from an environmental impact point of view. Since environmental sublethal effects of iodinated contrast media to organisms are largely unknown, taken together with high persistence and environmental presence at relatively high concentrations, additional environmental assessments may be necessary. For compounds with a low (provisional) guideline value as identified in the present study (e.g. carbamazapine), additional environmental monitoring may be warranted to characterize concentrations and to establish trends in their occurrence. As shown by Walraven and Laane (2009), river flow rates may influence contaminant concentrations seasonally, thus resulting in substantially varying BQ values. For example, it can be observed that the riverine concentration of the fuel oxygenate MTBE is highly dependent on the flow of the river Meuse. Similar patterns may occur for other compounds, resulting in (temporarily) exceedance of the BQ threshold. The evaluation as presented here supports the conclusion that the majority of the selected compounds as found in surface waters, groundwater and drinking water do not pose an appreciable concern to human health. This finding of no adverse effect to human health from exposure to trace quantities of compounds (e.g. pharmaceuticals) in surface waters and/or drinking water is supported by other results reported in the literature. Kingsbury et al. (2008) recently evaluated the potential health effects of 148 organic compounds in source water and finished water. The authors showed that the annual mean concentration of all compounds detected in finished water was less than the established human health benchmarks. Furthermore, Snyder et al. (2008) arrived at the same findings after evaluating human health effects associated with potential drinking water exposure of a suite of 62 indicator pharmaceuticals and potential endocrine disrupting compounds. Despite the absence of any concern to human health, drinking water remains a major point of consumer concern and some residual uncertainties need further exploration. For example, drinking water guideline values are developed using toxicity information for single compounds. Hence, the longterm cumulative dose-additive or synergistic effects of low concentrations of contaminants co-occurring as mixtures on human health and potentially sensitive sub-populations remain currently unknown. Understanding and implementing of such information is important for the development of future (enforceable) guideline values. Finally, the relatively large data gap on occurrence of compounds in drinking water should compel further research and assessment, especially for those compounds with a low (provisional) guideline value.

5.

Major conclusions

For most compounds evaluated in the present assessment, a substantial margin exists between the (provisional) guideline value and the maximum concentrations in surface waters, groundwaters and/or drinking water. The TTC based drinking water target values (0.1 mg/L and 0.01 mg/L for non-carcinogenic compounds, respectively) as proposed earlier are the conservative values they are meant

473

to be. They are optimally suited to provide exposure based waiving. The concentrations in drinking water of compounds such as MTBE, ETBE, 1,4-dioxane, NDMA and benzene should be monitored closely, since their guideline values are easily exceeded. Alkylated perfluorinated compounds such as PFOA and PFOS are environmentally persistent compounds and their increasing occurrence in (the sources of) drinking water should be monitored closely. For compounds with a very low (provisional) guideline value (e.g. mutagenic and carcinogenic compounds) it is important to better establish trends in their environmental occurrence. From a toxicological point of view iodinated contrast media as present in drinking water, such as amidotrizoic acid iopamidol, iohexol and iopromide, are not a direct concern for human health. However, further environmental assessment may be necessary, especially since the sublethal (ecological) effects of these compounds are largely unknown. Better understanding of the potential mixture effects of emerging compounds present in drinking water is important for the development of future guideline values.

Acknowledgements The authors wish to acknowledge the significant contribution to this work by Leo Puijker (KWR) and Ruud Jansen. Thanks also to Dr. Corine Houtman (HWL, the Netherlands) for her valuable comments on the manuscript. The research described was funded by the Joint Research Programme of the Dutch Water utilities (BTO).

references

Benzotriazoles Coalition, 2001. Benzotriazoles category justification and testing rationale. Synthetic Organic Chemical Manufacturers Organization. http://www.epa.gov/oppt/ chemrtk/pubs/summaries/benzo/c13456.pdf (accessed June 2009). BIBRA Toxicology International, 1995. Toxicity Profile of Benzotriazole. BIBRA Toxicology International, Carshalton (Surrey), UK. Brauch, H.J., Sacher, F., Denecke, E., Tacke, T., 2000. Efficiency of river-bank filtration for the removal of polar organic compounds (Werksamkeit der Uferfiltration fuer die Enfernung von polaren organischen Spurenstoffen). WasserAbwasser 141 (4), 226–234. Brown, D., Butterworth, K.R., Gaunt, I.F., Grasso, P., Gangolli, S.D., 1978. Short-term oral toxicity study of diethyl phthalate in the rat. Food Cosmet. Toxicol. 16, 415–422. California EPA (California Environmental Protection Agency), 2000. N,N-diethyl-meta-toluamide (DEET) risk characterization document. Department of pesticide regulation. Document number RCD 00-01, Sacramento, California, USA. CIIT (Chemical Industry Institute of Toxicology), 1990. Twenty four month final report inhalation toxicity of dimethylamine in F344 rats and B6C3F1 mice. CIIT Archives, Docket #11957. Ciba-Geigy, 1988. Combined chronic toxicity/oncogenicity study in rats. Basle, Ciba-Geigy, 1988 (unpublished study submitted to WHO).

474

water research 44 (2010) 461–476

Cotruvo, J.A., 1988. Drinking water standards and risk assessment. Regul. Toxicol. Pharmacol. 8 (3), 288–299. Danish EPA (Danish Environmental Protection Agency), 2004. Evaluation of Health Hazards by Exposure to BAM (2,6Dichlorobenzamide) and Risk Characterisation of Drinking Water Exposure. Environmental Project Nr. 943 2004. Dawson, B.V., Johnson, P.D., Goldberg, S.J., Ulreich, J.B., 1993. Cardiac teratogenesis of halogenated hydrocarbon-contaminated drinking water. J. Am. Coll. Cardiol. 21 (6), 1466–1472. De Voogt, P., Vink, C., Puijker, L.M., 2008. MTBE and ETBE in Dutch groundwater (MTBE en ETBE in Nederlands grondwater), KWR 08.043, KWR, Nieuwegein, The Netherlands. DuPont de Nemours and Company, Inc., 1964. Report MRID No. 00017763, 00080899. DECOS (Dutch Expert Committee for Occupational Standards), 2000. 1,2,3-benzotriazole. Health-based recommended occupational exposure limit. No. 2000/14OSH, The Hague, The Netherlands. ECB (European Chemicals Bureau), 2000a. IUCLID dataset on benzotriazoles. ECB (European Chemicals Bureau), 2000b. IUCLID dataset on Chloridazon. ECB (European Chemicals Bureau), 2000c. IUCLID dataset on Diethylene triamine penta acetic acid. ECB (European Chemicals Bureau), 2000d. IUCLID dataset on triphenylphosphine oxide. ECB (European Chemicals Bureau), 2002. European Risk Assessment Report 1,4-dioxane. EFSA (European Food Safety Authority), 2008. Opinion of the scientific panel on contaminants in the food chain on perfluorooctane sulfonate (PFOS), perfluorooctanoic acid (PFOA) and their salts. EFSA J. 53, 1–131. Ericson, I., Nadal, M., van Bavel, B., Lindstro¨m, G., Domingo, J.L., 2007. Levels of perfluorochemicals in water samples from Catalonia, Spain: is drinking water a significant contribution to human exposure? Environ. Sci. Pollut. Res. 15, 614–619. Goldenthal, E.I. (International Research and Development Corp.), 1994. Evaluation of DEET in a one-year chronic oral toxicity study in dogs. DEET Joint Venture/Chemical Specialties Manufacturers Assoc. DPR vol. 50191-164, Rec. No. 131265. Health Canada, 2003. Benchmark Dose for TCE in Drinking Water. Health Canada, Healthy Environments and Consumer Safety Branch, Biostatistics Unit, April, Ottawa, Ontario. Hogenboom, A.C., Van Leerdam, J.A., De Voogt, P., 2009. Accurate mass screening and identification of emerging contaminants in environmental samples by liquid chromatography–hybrid linear ion trap Orbitrap mass spectrometry. J. Chromatogr. A 1216 (3), 510–519. Hurni, H., Ohder, H., 1973. Reproduction study with formaldehyde and hexamethylenetetramine in beagle dogs. Food Cosmet. Toxicol. 11 (3), 459–462. Kingsbury, J.A., Delzer, C.D., Hopple, J.A., 2008. Anthropogenic organic compounds in source water of nine community water systems that withdraw from streams, 2002–2005, USGS, Reston, Virginia, USA, Report 2008–5208. Kleinneijenhuis, A.J., Puijker, L.M., 2008. Monitoring program for DMSA, DMST, N,N-dimethylsulfamide and its ozone reaction product NDMA in untreated and ozone-treated water. TNO report 7686, Delft, The Netherlands. Kolpin, D.W., Skopec, M., Meyer, M.T., Furlong, E.T., Zaugg, S.D., 2004. Urban contribution of pharmaceuticals and other organic wastewater contaminants to streams during differing flow conditions. Sci. Total Environ. 328 (1–3), 119–130. Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., Van Schothorst, F., Vos, J.G., Wurtzen, G., 2004. Structurebased thresholds of toxicological concern (TTC): guidance for

application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65–83. Krop, H. de Voogt, P., 2008. Physicochemical parameters and source markers of PFAS. Report IVAM ISO doc O/0818, Amsterdam, The Netherlands, pp. 1–147. KWR Watercycle Research Institute, 2009. KWR internal data, Nieuwegein, the Netherlands. Larsen, M.L., Illingworth, D.R., O’Malley, J.P., 1994. Comparative effects of gemfibrozil and clofibrate in type III hyperlipoproteinemia. Atherosclerosis 106, 235–240. Loos, R., Wollgast, J., Huber, T., Hanke, G., 2007. Polar herbicides, pharmaceutical products, PFOS, PFOA, and nonylpheniol and its carboxylates and ethoxylates in surface and tap waters around Lake Maggiore in Northern Italy. Anal. Bioanal. Chem. 387, 1469–1478. Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S., Bidoglio, G., 2009. EU-wide survey of polar organic persistent pollutants in European river waters. Environ. Pollut. 157 (2), 561–568. Lynch, D.W., Moorman, W.J., Stober, P., Lewis, T.R., Iverson, W.O., 1986. Subchronic inhalation of diethylamine vapour in Fisher344 rats: organ toxicity. Fund. Appl. Toxicol. 6 (3), 559–565. Medinsky, M.A., Wolf, D.C., Cattley, R.C., Wong, B., Janszen, D.B., Farris, G.M., Wright, G.A., Bond, J.A., 1999. Effects of a thirteen-week inhalation exposure to ethyl tertiary butyl ether on Fischer-344 rats and CD-1 mice. Toxicol. Sci. 51, 108–118. Mitsubishi Chemical Safety Institute Ltd. (date of study unknown). 4,4’-sulfonyldiphenol. Summary of repeated dose and developmental studies. Kashima Laboratory, Ibaraki, Japan. Mitsumori, K., Usui, T., Takahashi, K., Shirasu, Y., 1979. Twentyfour month chronic toxicity studies of dichlorodiisopropyl ether in mice. J. Pestic. Sci. 4 (3), 323–335. Mons, M., Van der Hoek, J.P., Stoks, P., Van der Kooij, D., 2008. Vision on and guideline values for xenobiotic substances in drinking water (Visie op en streefwaarden voor milieuvreemde stoffen in drinkwater). H2O 4, 1–4 (in Dutch). Morgareidge, K., 1971. 90-Day feeding study with benzothiazole in rats. Unpublished report from Food and Drug Research Laboratories, Inc., Maspeth, New York, USA. Submitted to WHO by Flavor and Extract Manufacturers’ Association of the United States. Morisetti, A., Tirone, P., Luzzani, F., de Haen, C., 1994. Toxicological safety assessment of iomeprol, a new X-ray contrast agent. Eur. J. Radiol. 1 (18 Suppl), S21–31. Mout, L., Jansen, L., van der Pijll, S., Nierop, S., 2007. Implementation program dimethenamid-p. Clean sources, now and in the future (uitvoeringsprogramma dimethenamid-p. Schone bronnen, nu en in de toekomst: tweede reeks knelpunten (http://www. schonebronnen.nl/). The Hague, The Netherlands. Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996. Correlation of structural class with no-observed effect levels: a proposal for establishing a threshold of concern. Food Chem. Toxicol 34, 829–867. NCI (National Cancer Institute), 1978. Bioassay of 1,4-dioxane for possible carcinogenicity, CAS No. 123-91-1. NCI Carcinogenesis Technical Report Series No. 80. DHEW Publication No. (NIH) PB-285-711, Bethesda, Maryland, USA. Nghiem, L.D., Schafer, A.I., Elimelech, M., 2004. Removal of natural hormones by nanofiltration membranes: measurement, modeling, and mechanisms. Environ. Sci. Technol. 38 (6), 1888–1896. Norimitsu, S., Kouji, H., Kayoko, I., Kazuaki, S., Takeo, Y., Akio, K., 2004. Perfluorooctanoate and perfluorooctane sulfonate concentrations in surface water in Japan. J. Occup. Health 46, 49–59. NTP (National Toxicology Program), 1986. Toxicology and carcinogenesis studies of benzene in F344/N rats and B6C3F1

water research 44 (2010) 461–476

mice (gavage studies). Research Triangle Park, NC, USA, US Department of Health and Human Services. (Technical report series No. 289). NTP (National Toxicology Program), 1987. Teratologic evaluation of diethylene glycol dimethyl ether (CAS No. 111-96-6) administered to New Zealand White rabbits on gestation days 6 through 19. Research Triangle Park, NC, USA, National Institute of Environmental Health Sciences, National Toxicology Program (NTP-87-108; PB 87–209532). NTP (National Toxicology Progam), 1991. Toxicology and carcinogenesis studies of tris(2-chloroethyl)phosphate in F344/N rats and B6C3F1 mice. Technical report series No. 391. NTP TR 391, Research Triangle Park, NC, USA. OECD/SIDS (Organization for Economic Development and Cooperation/Screening Information DataSet), 1994. pToluenesulfonamide. UNEP publications, Nairobi, Kenya. OECD/SIDS (Organization for Economic Development and Cooperation/Screening Information DataSet), 1998. Triethylphosphate. UNEP publications, Nairobi, Kenya. Peto, R., Gray, R., Brantom, P., Grasso, P., 1991a. Effects on 4080 rats of chronic ingestion of N-nitrosodiethylamine or Nnitrosodimethylamine: a detailed dose–response study. Cancer Res. 51, 6415–6451. Peto, R., Gray, R., Brantom, P., Grasso, P., 1991b. Dose and time relationships for tumor induction in the liver and esophagus of 4080 inbred rats by chronic ingestion of Nnitrosodiethylamine or N-nitrosodimethylamine. Cancer Res. 51, 6452–6469. Proviron Fine Chemicals, 2003. IUCLID dataset on Nbutylbenzenesulphonamide. REWAB Database, 2009. KWR Watercycle Research Institute, Nieuwegein, The Netherlands. Robinson, M., Bruner, R.H., Olson, G.R., 1990. Fourteen- and ninety-day oral toxicity studies of methyl-t-butyl ether in Sprague-Dawley rats. J. Am. Coll. Toxicol. 9, 525–540. Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., Von Gunten, U., Wehrli, B., 2006. The challenge of micropollutants in aquatic systems. Science 313 (5790), 1072–1077. SCOEL (Scientific Committee on Exposure Limits), 1991. No 11b. Recommendations of the SCOEL for DMA. SCOEL (Scientific Committee on Exposure Limits), 2002. No 91. Recommendations of the SCOEL for DEA. Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff, R.D., D’Aco, V.J., Anderson, P.D., 2005. Human pharmaceuticals in US surface waters: a human health risk assessment. Regul. Toxicol. Pharmacol. 42 (3), 296–312. Seacat, A.M., Thomford, P.J., Hansen, K.J., Olsen, G.W., Case, M.T., Butenhoff, J.L., 2002. Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in cynomolgus monkeys. Toxicol. Sci. 68 (1), 249–264. Sherman, H., 1972. Long-term feeding studies in rats and dogs with 2-benzimadazole carbamic acid, methyl ester (INE-965) (50% and 70% MBC wettable powder formulations). Unpublished report from E.I. DuPont de Nemours and Co., Inc., Haskell Laboratory, Newark, Delaware, USA. Singh, G., Driever, P.H., Sander, J.W., 2005. Cancer risk in people with epilepsy: the role of antiepileptic drugs. Brain 128 (Pt 1), 7–17. Skutlarek, D., Exner, M., Farber, H., 2006. Perfluorinated surfactants in surface and drinking waters. Environ. Sci. Pollut. Res. 13 (5), 299–307. Snyder, S.A., Villeneuve, D.L., Snyder, E.M., Giesy, J.P., 2001. Identification and quantification of estrogen receptor agonists in wastewater effluents. Environ. Sci. Technol. 35 (18), 3620–3625. Snyder, S.A., Trenholm, R.A., Snyder, E.M., Bruce, G.M.P., Hemming, J.D.C., 2008. Toxicological Relevance of EDCs and

475

Pharmaceuticals in Drinking Water. Awwa Research Foundation, Denver, USA, p. 484. Stokinger, H.E., Woodward, R.L., 1958. Toxicologic methods for establishing drinking water standards. J. Am. Water Works Ass. 50, 515–529. Swarm, R.L., Roberts, G.K., Levy, A.C., Hines, L.R., 1973. Observations on the thyroid gland in rats following the administration of sulfamethoxazole and trimethoprim. Toxicol. Appl. Pharm. 24 (3), 351–363. Swartjes, F.A., Baars, A.J., Fleuren, R.H.L.J., Otte, P.F., 2004. Risk limits for MTBE (methyl tertiair-butyl ether) in soil, sediment, groundwater, surface water, drinking water and for drinking water production (risicogrenzen voor MTBE (Methyl tertiairButyl Ether) in bodem, sediment, grondwater, oppervlaktewater, drinkwater en voor drinkwaterbereiding) RIVM report 711701039/2004, Bilthoven, The Netherlands. Toccalino, P.L., 2007. Development and application of healthbased screening levels for use in water-quality assessments. US Geological Survey Scientific Investigations Report 20075106, 12 p, Reston, Virginia, USA. URL1: http://www.epa.gov/waterscience/criteria/drinking/ dwstandards.pdf, accessed June 2009. URL2: http://www.who.int/water_sanitation_health/dwq/ guidelines/en/, accessed June 2009. URL3: http://www.ctb.agro.nl/ctb_files/11995_01.html, accessed May 2009 (In Dutch). URL4: http://www.iawr.org/index.php, publications, accessed January 2009. URL5: http://www.riwa-rijn.org/riwa_en.html, publications, accessed January 2009. US EPA (United States Environmental Protection Agency) IRIS database (1988a). 1,4-dioxane. US EPA (United States Environmental Protection Agency) IRIS database, 1988b. Diethyl phthalate. US EPA (United States Environmental Protection Agency) IRIS database, 1988c. Diuron. US EPA (United States Environmental Protection Agency) IRIS database, 1989. Bis(chloroisopropyl)ether. US EPA (United States Environmental Protection Agency), 2006. National Primary Drinking Water Standards. http://www.epa. gov/safewater/consumer/pdf/mcl.pdf (accessed August 2009). Van Leerdam, J.A., Hogenboom, A.C., Kooi van der, M.E., de Voogt, P., 2009. Determination of polar 1-H-benzotriazoles and benzothialzoles in water by solid-phase extraction and liquid chromatography LTQ Orbitrap mass spectrometry. Int. J. Mass Spectrom. 282 (3), 99–107. Van Leeuwen, F.X., 2000. Safe drinking water: the toxicologist’s approach. Food Chem. Toxicol. 38 (1 Suppl), S51–S58. Van Leeuwen, C.J., Vermeire, T., 2007. Risk Assessment of Chemicals, second ed., Springer, ISBN 978-4020-6101-1. Van Wezel, A.P., Puijker, L., Vink, C., Versteegh, A., de Voogt, P., 2009. Odour and flavour thresholds of gasoline additives (MTBE, ETBE and TAME) and their occurrence in Dutch drinking water collection areas. Chemosphere 76, 672–676. Versteegh, J., Van Der Aa, N., Dijkman, E., 2007. Pharmaceuticals in drinking water and sources of drinking water - results of monitoring campaign 2005/2006 (geneesmiddelen in drinkwater en drinkwaterbronnen - resultaten van het meetprogramma 2005/2006), 53, RIVM, Bilthoven, The Netherlands. Number: R(T)-1-67. Walraven, N., Laane, R.W., 2009. Assessing the discharge of pharmaceuticals along the Dutch coast of the North sea. Rev. Environ. Contam. Toxicol. 199, 1–18. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environ. Sci. Technol. 37 (17), 6649–6663. WHO (World Health Organization), 1996. Simazine in drinkingwater. Background document for development of WHO

476

water research 44 (2010) 461–476

Guidelines for drinking-water quality. WHO/SDE/WSH/03.04/ 42, Geneva, Switzerland. WHO (World Health Organization), 1998. Guidelines for drinking water quality. Second edition, volume 1 recommendations: addendum, ISBN 92 4 154514 3, Geneva, Switzerland. WHO (World Health Organization), 2002. Diethylene glycol dimethyl ether. Concise International Chemical Assessment Document 41, Geneva, Switzerland. WHO (World Health Organization), 2003a. Benzene in drinking water. WHO report WHO/SDE/WSH/03.04/24, Geneva, Switzerland. WHO (World Health Organization), 2003b. 2,4-D in drinking water. WHO report WHO/SDE/WSH/03.04/70, Geneva, Switzerland. WHO (World Health Organization), 2003c. Isoproturon in Drinking-water. Background document for development of WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/ 03.04/37, Geneva, Switzerland. WHO (World Health Organization), 2005a. Glyphosate and AMPA in drinking-water. Background document for development of WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/ 03.04/97, Geneva, Switzerland. WHO (World Health Organization), 2005b. Trichloroethene in Drinking-water, Background document for development of WHO Guidelines for Drinking-water Quality. WHO/SDE/WSH/ 05.08/22, Geneva, Switzerland. WHO (World Health Organization), 2006. Guidelines for drinking water quality, First Adendum to third ed., Recommendations, vol. 1, 2006, Geneva, Switzerland. WHO (World Health Organization), 2008. NNitrosodimethylamine in drinking-water. Background

document for preparation of WHO Guidelines for drinkingwater quality. Geneva, World Health Organization (WHO/HSE/ AMR/08.03/8), Geneva, Switzerland. WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 1973. Food Additives Series 32; Toxicological Evaluation of Certain Food additives and Contaminants. Prepared by the forty-first meeting of the joint FAO/WHO expert committee. WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 1974. Food Additives Series No. 5; Seventeenth Report of the Joint FAO/WHO Expert Committee on Food Additives, World Health Organization technical report series, 1974, No. 539; FAO Nutrition Meetings Report Series, 1974, No. 53. WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 1995. 892. Carbendazim (Pesticide residues in food: 1995 evaluations Part II Toxicological & environmental). WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 2003. WHO Food Additives Series No. 50; sulfur-containing heterocyclic compounds. WHO/JMPR (World Health Organization/Joint Meeting on Pesticide Residues), 2001. Toxicological evaluations imidacloprid. Pesticide residues in food. WHO/JMPR (World Health Organization/Joint Meeting on Pesticide Residues), 2005. Dimethenamid-p. pp. 189–239. Wilson, A.B., Thorpe, E., 1971. Toxicity studies on the ‘‘Prefix’’ residue 2,6-dichlorobenzamide: two year oral experiment with dogs. Tunstall Laboratory. Report number: Lab Project Number: T507531; Group research report: TLGR.0028.71, pp. 1–28.

water research 44 (2010) 477–492

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Monitoring the biological activity of micropollutants during advanced wastewater treatment with ozonation and activated carbon filtration M. Macova a, B.I. Escher a,*, J. Reungoat b, S. Carswell a,c, K. Lee Chue a, J. Keller b, J.F. Mueller a a

University of Queensland, National Research Centre for Environmental Toxicology (EnTox), 39 Kessels Road, Brisbane, QLD 4108, Australia University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia c Queensland Health Forensic and Scientific Services, Organics Laboratory, 39 Kessels Road, Brisbane, QLD 4108, Australia b

article info

abstract

Article history:

A bioanalytical test battery was used to monitor the removal efficiency of organic micro-

Received 13 May 2009

pollutants during advanced wastewater treatment in the South Caboolture Water Reclamation

Received in revised form

Plant, Queensland, Australia. This plant treats effluent from a conventional sewage treatment

28 August 2009

plant for industrial water reuse. The aqueous samples were enriched using solid-phase

Accepted 7 September 2009

extraction to separate some organic micropollutants of interest from metals, nutrients and

Available online 16 September 2009

matrix components. The bioassays were chosen to provide information on groups of chemicals with a common mode of toxic action. Therefore they can be considered as sum indicators

Keywords:

to detect certain relevant groups of chemicals, not as the most ecologically or human health

Bioassays

relevant endpoints. The baseline toxicity was quantified with the bioluminescence inhibition

Baseline toxicity

test using the marine bacterium Vibrio fischeri. The specific modes of toxic action that were

Phytotoxicity

targeted with five additional bioassays included aspects of estrogenicity, dioxin-like activity,

Estrogenicity

genotoxicity, neurotoxicity, and phytotoxicity. While the accompanying publication discusses

Genotoxicity

the treatment steps in more detail by drawing from the results of chemical analysis as well as

Toxic equivalency concept

the bioanalytical results, here we focus on the applicability and limitations of using bioassays for the purpose of determining the treatment efficacy of advanced water treatment and for water quality assessment in general. Results are reported in toxic equivalent concentrations (TEQ), that is, the concentration of a reference compound required to elicit the same response as the unknown and unidentified mixture of micropollutants actually present. TEQ proved to be useful and easily communicable despite some limitations and uncertainties in their derivation based on the mixture toxicity theory. The results obtained were reproducible, robust and sensitive. The TEQ in the influent ranged in the same order of magnitude as typically seen in effluents of conventional sewage treatment plants. In the initial steps of the treatment chain, no significant degradation of micropollutants was observed, and the high levels of dissolved organic carbon probably affected the outcome of the bioassays. The steps of coagulation/flocculation/dissolved air flotation/sand filtration and ozonation decreased the effectbased micropollutant burden significantly. ª 2009 Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: þ61 (0) 7 3274 9180; fax: þ61 (0) 7 3274 9003. E-mail address: [email protected] (B.I. Escher). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.025

478

1.

water research 44 (2010) 477–492

Introduction

Water quantity and water quality issues have led many countries to explore water reuse and water reclamation as alternatives to provide water for direct use in industrial applications as well as agricultural and gardening use, and for indirect use to supplement water reservoirs. Australia is amongst the countries most affected by uneven distribution of water resources and droughts. Therefore, various water treatment and desalination plants have been put into operation, among them the Water Reclamation Plant in South Caboolture, Queensland, Australia (van Leeuwen et al., 2003). Besides hygiene issues, organic micropollutants are of major concern because classical wastewater treatment with biological activated sludge only partially removes hazardous chemicals. Furthermore there is the potential that persistent and toxic micropollutants and transformation products could build up during the recycling process. Urban and agricultural wastewaters contain a high number of organic micropollutants, including pharmaceuticals and personal care products, hormones and industrial chemicals, biocides and pesticides (Schwarzenbach et al., 2006). A large number of studies have investigated the fate and removal of various individual and specific groups of micropollutants by treatment processes like biological activated sludge treatment (Ternes et al., 2004; Joss et al., 2005), ozonation and advanced oxidation processes (von Gunten, 2003; Huber et al., 2004; Huber et al., 2005; Lee et al., 2008) or activated carbon treatment (Sanchez-Polo et al., 2006; Snyder et al., 2007) but little is known how these treatment processes change the composition and biological activity of the micropollutants. Toxicity testing may provide complementary information to chemical analysis on the sum of micropollutants present in treated water. However, whole-effluent toxicity testing, which is a valuable tool for hazard assessment of industrial and municipal effluents (Chapman, 2000), is inappropriate to evaluate the fate of micropollutants during advanced water treatment because acute toxicity tests are not sensitive enough for purified water and long-term chronic tests are too expensive and time-consuming for routine monitoring. In addition, the physical treatment of wastewater alters the matrix and electrolyte composition and it becomes virtually impossible to differentiate the effect caused by micropollutants from that induced by the matrix, such as organic matter content, pH, ionic strength or nutrient content. Therefore, over the last years, we have developed a framework for bioanalytical quantification of organic micropollutants, which has three main cornerstones: Firstly, the aqueous samples are enriched using solid-phase extraction (SPE) to separate the organic micropollutants of interest from metals and matrix components. This SPE method was previously validated for recovery and extraction yield for selected chemicals including pesticides and pharmaceuticals (Bengtson Nash et al., 2005a; Escher et al., 2005; Escher et al., 2008a). Secondly, a battery of bioanalytical methods were selected that cover a non-specific bioassay, the bioluminescence inhibition of the marine bacterium Vibrio fischeri and

five bioassays for specific modes of toxic action that are indicative for groups of chemicals of particular relevance for human and environmental health, including aspects of estrogenicity, dioxin-like activity, genotoxicity, neurotoxicity, and phytotoxicity (Muller et al., 2007; Muller et al., 2008; Escher et al., 2008a; Escher et al., in press). Table 1 summarizes the key features of the bioassays and more details are given in the Supporting Information. Finally, we developed a coherent data evaluation method that is based on the toxic equivalency concept (Villeneuve et al., 2000; Escher et al., 2008b). Thus, the effects at various steps of the treatment chain can be easily compared and treatment efficiency can be compared between the different effect endpoints. Bioassays in combination with SPE have been successfully applied by several groups to evaluate various water treatment options (Escher et al., 2008a; Cao et al., 2009). In particular, the bioluminescence inhibition test with Vibrio fischeri and the Salmonella based test for genotoxicity have been widely used despite their recognized limitations (Johnson, 2005; Petala et al., 2006; Petala et al., 2008). The present work is more extensive covering several additional toxicity endpoints, allowing high-throughput monitoring applications while remaining cost-efficient. Here we present this framework for the first time to evaluate all steps of enhanced water treatment in the South Caboolture Water Reclamation Plant (van Leeuwen et al., 2003). This paper focuses on the advantages and limitations in the application of the bioanalytical test battery, and the interpretation and significance of the results. In the accompanying paper, the results obtained from the bioanalytical tools are compared with chemical analytical data for a variety of pharmaceuticals and pesticides and conclusions for advanced wastewater treatment are drawn (Reungoat et al., 2010).

2.

Materials and methods

2.1.

Samples and sites

Four consecutive sets of 24-hour composite samples were collected on 11-07-08, 22-07-08, 27-07-08 and 06-08-08 using refrigerated autosamplers at 10 sites of the South Caboolture Water Reclamation Plant (Table 2).

2.2.

Solid phase extraction

Immediately after sampling, concentrated HCl (36%) was added to a final concentration of 5 mM for preservation. It was demonstrated in earlier work that a pharmaceutical cocktail in a wastewater matrix had highest recoveries for HLB at pH 3 (Escher et al., 2005). Samples were extracted using 1 g OASIS HLB solid phase material in 20 mL cartridges (Waters, Australia) following centrifugation (only S1 sample) and filtration with a glass fibre filter. After conditioning the cartridges with 10 mL methanol and 20 mL of 5 mM HCl in MilliQ water, a known volume of sample was percolated under vacuum. The cartridges were dried overnight under vacuum and were eluted with 10 mL

479

water research 44 (2010) 477–492

Table 1 – Bioanalytical test battery (adapted from Muller (2008)). Assay

Measured by

Target mode of toxic action

Bioluminescence inhibition test

Reduction in luminescence of the naturally bioluminescent marine bacteria Vibrio fischeri

Effect on energy status through cytotoxicity and/or baseline toxicity

All chemicals

Acetylcholinesterase (AChE) Inhibition Assay

Product of the enzymatic reaction (colorimetric)

Organophosphates and carbamate insecticides

Imaging-PAM Assay

Increased Chl a fluorescence, inversely proportional to PS II photosynthetic yield Cellular proliferation in estrogenic dependent cell line

Inhibition of the enzyme, which hydrolyses the neurotransmitter acetylcholine PS II derived photosynthesis inhibition Estrogenicity

International Standard Organisation, 1998; Johnson, 2005; Farre´ et al., 2006; Escher et al., 2008a Ellman et al., 1961; Deutsche Norm, 1995; Hamers et al., 2000

Triazine and phenylurea herbicides

Schreiber et al., 2002; Schreiber et al., 2007

Estrogens, estrogenic industrial chemicals Polychlorinated dibenzodioxins/ furans (PCDD/F) and biphenyls (PCB) or polycyclic aromatic hydrocarbons (PAH) Chlorinated byproducts, aromatic amines, polycyclic aromatic hydrocarbons (PAH)

Soto et al., 1995; Ko¨rner et al., 1999

E-SCREEN

AhR-CAFLUX

Induction of green fluorescent protein under the control of Ah receptor; (fluorescence)

Dioxin-like activity; Aryl hydrocarbon (Ah) receptor activation

umuC assay

Induction of ß-galactosidase enzyme as an indicator of DNA damage; Product of the enzymatic reaction (colorimetric)

Genotoxicity; DNA damage (SOS-response)

methanol and 10 mL hexane:acetone (1:1). All eluates were evaporated to dryness and reconstituted with 0.5 mL of ethanol.

Chemicals that give response

dilution factorbioassay ¼

Literature reference

Nagy et al., 2002; Zhao and Denison, 2004

Oda et al., 1985; Reifferscheid et al., 1991; International Standard Organisation, 2000

volume of extract added to bioassay total volume of bioassay (2)

2.3.

Bioanalytical tools

The bioassays were performed as previously reported (Escher et al., 2005; Muller et al., 2007; Muller et al., 2008; Escher et al., 2008a; Escher et al., in press) using the methods cited in Table 1 and described in detail in the Supporting Information.

The final relative enrichment factor REF is the combination of the enrichment of the extraction and the dilution in the bioassay (Eq. 3) (Escher et al., 2006; Muller et al., 2007) and represents the enrichment or dilution of the original sample in each bioassay (Table 3). The REF is equivalent to concentration and is expressed in the units [Lwater sample/Lbioassay]. REF ¼ dilution factorbioassay enrichment factorSPE

2.4.

(3)

Relative enrichment factor of the samples REF

For each sample, the enrichment factor of the solid phase extraction was calculated using (Eq. 1) which represents the enrichment of the extract compared to the source water. enrichment factorSPE ¼

Vwater Vextract

(1)

Sample volume varied from 2.0 to 2.5 L. The final volume of each extract was 0.5 mL therefore the enrichment factorSPE was between 3500 and 5500 depending on the sample. An aliquot of the enriched sample extracts were then added to the microtiter plate of the respective bioassay and serially diluted by a test medium to obtain a concentrationeffect curve. A dilution factor of each bioassay was calculated using (Eq. 2). The dilution factors ranged from 0.05 to 0.007 for the highest dose depending on the bioassay.

2.5.

Evaluation of the concentration-effect curves

Due to the nature of the endpoints, the six bioanalytical tests of the battery could not be evaluated with a single data evaluation model (Table 3). In general, all concentration effect curves of the reference compounds and sample extracts followed a log-logistic function (Eq. 4), which was fitted using Prism 5.0 software (GraphPad, San Diego, CA, USA). Adjustable parameters are the minimum and maximum of the effect, min and max, the slope s, and the effect concentration inducing 50% of the maximum effect, EC50. If the concentration-effect curve dropped sharply at the range of higher doses indicating that the sample has a cytotoxic effect, the cytotoxic concentrations were excluded from the concentration-effect assessment.

480

water research 44 (2010) 477–492

Table 2 – Sample details. Samples were each taken after the described process. More details are given in Reungoat et al. (2010). TOC/DOC data represent the average of two samplings (22-07-08 and 06-08-08). Sample ID

Blank S1 S2 S3 S4 S5 S6 S7 C3 C4 C5

Site description

TOC (mg/L)

MilliQ water Influent (Effluent from WWTP) Denitrification Pre-Ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water

DOC (mg/L)

avg

sd

avg

sd

n.a. 20 24 21 11 9.9 6.6 7.1 8.5 4.0 3.6

n.a. 4 n.a. 7 1 0.9 0.6 0.5 1 0.4 0.07

n.a. 17 21 21 11 10 6.6 6.8 8.8 4.2 4.1

n.a. 4 n.a. 4 0.5 0.9 0.9 0.7 0.8 0.04 0.08

avg - average; sd - standard deviation; n.a. - not applicable.

effect ¼ min þ

maxmin 1þ10sðlog EC50 log ðconcentration of reference compound or REF of sampleÞÞ (4)

Three bioassays, the inhibition of bioluminescence of Vibrio fischeri, inhibition of AChE and inhibition of PSII photosynthetic yield, all produced effects between 0% and 100%, therefore the maximum effect in (Eq. 4) could be set to 100% (Table 3). Two bioassays, the AhR-CAFLUX and the E-SCREEN, yield information on the induction of a receptor or proliferation of cells and there is no absolute maximum effect. Dependent on the mode of action, the maximum effect can be higher or lower than the maximum effect of the reference compounds. The extent of difference of maximally attainable effect yields some additional information on the mode of action and the type of compounds that induce the effect. Details of each assay are discussed below. The curve fit is then performed in such a way that the maximum effect in equation 4 is an additional fitting parameter. Finally the umuC assay exhibited a linear concentrationeffect curve. At low effect levels the log-logistic concentration effect curve (Eq. 4) is congruent with a linear concentrationeffect curve. Unlike the other assays, where effects were compared to a reference compound, genotoxicity is expressed as the concentration that induces the threshold of induction defined by the EN ISO guideline (International Standard Organisation, 2000). This way of data evaluation needs further refinement in the future because it is not directly comparable to the results in the other bioassays and the effect concentration cannot be translated into TEQ. The detection limit of the bioassays was defined as three times the standard deviation of the response using the lowest concentration of the standard that induced an effect significantly different from the control.

concentrations (Villeneuve et al., 2000; Escher et al., 2008b). The equivalent concentration represents the concentration of the reference compound required to produce the same effect as the mixture of different compounds in the sample. Equivalent concentrations were calculated from concentration-effect curves of the reference compound and the samples. Both reference compounds and samples generally followed a sigmoidal log–concentration-effect curve (Eq. 4), which is equivalent to a linear function with respect to nonlogarithmic concentration at small effect levels. Sample extracts are normally composed of a mixture of unknown substances at unknown concentrations, so the concentration unit of the sample in the concentration effect curve is based on the Relative Enrichment Factor (REF) in units of [Lwater sample/Lbioassay]. Toxic equivalent concentrations (TEQ) were calculated as the ratio of EC50 values of the reference compound to the EC50 of the sample (Eq. 5). TEQ ¼

EC50 ðreference compoundÞ EC50 ðsampleÞ g 3 2 L g bioassay 5 4Lwater sample ¼ Lwater sample

ð5Þ

Lbioassay

Since the EC50(reference compound) is expressed in concentration units such as [g/Lbioassay] and the EC50(sample) in concentration units [Lwater sample/Lbioassay], representing the enrichment or dilution (REF) of the sample required to elicit the 50% effect, the toxic equivalent concentration is expressed in [g/Lwater sample]. In principle, any effect level can be used to derive TEQ, provided that the concentration-effect curves are parallel i.e. have the same slope, which is in fact one of the prerequisites for the validity of the toxic equivalency concept (Villeneuve et al., 2000). Therefore, if the effect of the sample in the bioassay does not reach 50%, toxic equivalency can be estimated using EC20 of the sample and EC20 of the corresponding reference compound.

2.6. Estimating equivalent concentrations in the samples

2.7.

For an easy-to-follow reporting of the effect, in all but one bioassay, the effects were reported in terms of toxic equivalent

To determine background levels of contamination associated with the extraction process as well as assessing any effect of

QA/QC

481

water research 44 (2010) 477–492

Table 3 – Evaluation of the concentration effect curves and expression of the bioanalytical results. Assay

Concentration-effect assessment

REF range

Reference compound; EC50 of the reference Positive control compound

Result expression

Detection limit

Baseline Toxicity – Top: 100 Bioluminescence Bottom: 0 inhibition in Slope: 1 Vibrio fischeri Neurotoxicity – AChE Phytotoxicity – Max-I-PAM

0.4–38

Virtual baseline toxicant; Phenola

12 mg/L; 130 mg/L

Baseline toxicity equivalent concentrations (baseline-TEQ)

20% inhibition of bioluminescence

0.04–115

Parathion

120 mg/l

0.3 mg/L

1.5–258

Diuron

16 mg/L

Estrogenicity – E-SCREEN

Top: w Bottom: w Slope: 1

0.0006–55 17ß-Estradiol (E2)

2 ng/L

Ah-Receptor – AhR-CAFLUX

Top: at least 100 and then w Bottom: w Slope: 1 Linear regression

0.004–55

2,3,7,8Tetrachlorodibenzodioxin (TCDD)

8 ng/L

Parathion equivalent concentrations (PTEQ) Diuron equivalent concentrations (DEQ) Estradiol equivalent concentrations (EEQ); Relative proliferative effect (RPE) TCDD equivalent concentrations (TCDDEQ)

2.2–110

(-S9) 2-amino anthracenea (þS9) 4-nitroquinolineN-oxidea

n.a.

Genotoxicity – Umu

0.01 mg/L

0.02 ng/L

0.04 ng/L

REF that elicits genotoxic Maximum REF effect with threshold achieved in the Induction Ratio of assayb 1.5

a Only used as positive control, not as reference compound. b Quantification limit; w variable; n.a. - not applicable.

the solvent, 2.0 L of MilliQ water were extracted identically to the samples and analysed in all bioassays as a procedural blank. For quality control and assurance purposes two replicates of randomly selected samples (2 sites per each sampling; for details see Supporting information Table S1–S6) were collected and analysed in all bioassays on a different day than the first replicate to evaluate repeatability of the SPE and the bioassays. Each sample extract was tested in the bioassays in duplicates or triplicates per run, depending on the assay. All individual data sets are summarised in the Supporting information Table S1–S6. Since the variability among the four sampling events was not higher than the variability between the sample replicates collected at the same time, we report in the main manuscript the average sd of four sample extracts at four different consecutive sampling events (n ¼ 4, whith exception of S2, where n ¼ 3). Analysis of variance (ANOVA) was used to test the differences among the average TEQ of samples S1-C5, including the blank, in comparison to S1.

3.

Results and discussion

3.1. Baseline toxicity – bioluminescence inhibition in vibrio fischeri The EC50 of phenol was 130 37 mg/L over 8 plates in 3 different repetitions of the assay, confirming the reproducibility and repeatability of the assay. While in all other assays, the reference compound served two purposes, as quality control and to define the reference for the TEQ, in the bioluminescence inhibition test, phenol was only used for quality control.

Toxicity of the sample was expressed in baseline toxicity equivalents (baseline-TEQ) derived from a baseline toxicity QSAR (quantitative structure-activity relationship) (Fig. 1A) using a virtual compound with log Kow of 3 and molecular weight of 300 g/mol as a reference, which equates to an EC50 of 12 mg/L (Escher et al., 2008b). The rationale behind this choice is the fact that there is not a singular positive control, as every chemical exhibits baseline toxicity. For specifically acting compounds the baseline effect is marginal and cannot be resolved if the single chemical is tested. However, in a mixture with a large number of chemicals with a variety of specific modes of action, as is the case in a wastewater sample, the baseline toxicity might actually dominate the overall mixture effect (Warne and Hawker, 1995). Thus, baseline toxicity will provide an integrative measure of the combination of chemicals that act together in the mixture and each chemical’s contribution is essentially weighted by its hydrophobicity only (Escher and Schwarzenbach, 2002). Fig. 1A depicts the relationship between hydrophobicity and effect concentration on a log-log plot and how the EC50 of a virtual baseline toxicant is derived (note that the hydrophobicity scale is based on liposome-water partitioning not octanol-water partitioning), therefore the virtual baseline toxicant with log Kow of 3 is situated slightly higher than 3 on the log-hydrophobicity scale. Samples were tested for baseline toxicity at 8 different concentrations after serial dilution 1:2, with the REF ranging from 0.4 to 38. All previously tested baseline toxicants exhibited a common slope of 1 (Escher et al., 2008b), and therefore the slope of the samples was fixed to 1, too. Effect concentrations (EC50) of the samples used to calculate the toxic equivalent concentrations were expressed in REFs,

482

water research 44 (2010) 477–492

Fig. 1 – Concentration-effect curves of the reference compounds and selected samples tested in bioluminescence inhibition assay (A, B), AChE inhibition assay (C, D) and I-PAM (E, F) assay. Bottom and top of each curve was fixed to 0 and 100, respectively. Quality of the fit expressed as R2 was >0.90, except for S7-effluent sample tested in AChE assay, where the R2 was 0.74. (A) Baseline toxicity QSAR for the bioluminescence inhibition test with Vibrio fischeri and derivation of the EC50 of the virtual baseline toxicant that is used as reference compound in this assay (data and QSAR from Escher et al. (2008b)). Error bars indicate SD.

representing the enrichment or dilution of the sample required to elicit the 50% effect, and are summarised in Table 4. The EC50 of the blank was 59, which means that if MilliQ water were enriched 59 times, the blank extract would elicit 50% bioluminescence inhibition. This EC50 was extrapolated from experimental data, where the highest measured REF was 38, which caused a 25% effect. It is not reasonable to enrich the samples to a higher REF because the enrichment of impurities of solvents and materials or contamination during the enrichment procedure would mask the measurable effect. The influent, denitrification and ozonation samples all exhibited EC50 in the range of 4–5.6, i.e. the samples had to be

enriched by approximately five times to see an effect. The concentration window for a visible effect is relatively narrow. A sample that is not enriched (REF ¼ 1) or even diluted would not give any visible response. The necessary enrichment is not the only reason for using SPE for sample preparation: a majority of matrix compounds are removed by SPE, such as salts and particulates. All organisms are very sensitive to variable electrolyte concentration. In this assay too little salt would decrease the bioluminescence of the marine bacterium (Escher et al., 2008a), and salting up is difficult because the salt content of water samples is unknown. In contrast to salts and particulate matter, we assume that dissolved organic carbon

483

water research 44 (2010) 477–492

Table 4 – Sample enrichment required to produce a 50% effect in each assay expressed as the average ± sd of four 24 h sampling events. ID

S1 S2 S3 S4 S5 S6 S7 C3 C4

C5

Site description

Blank Influent (Effluent from WWTP) Denitrification Pre-ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water

Bioluminescence inhibition

AChE

I-PAM

EC50 (REF)

EC50 (REF)

EC50 (REF)

avg

sd

59 5.6 4.4 4.0 9.8 14 28 29 11

0.81 1.6 0.94 0.98 3.7 3.2 7.8 11 2.6

36

14

BDL (<20%)

2381

1393

45

22

BDL (<20%)

2380

3349

(DOC), in particular the low molecular weight fraction, is partially retained by SPE. Baseline-TEQ increased slightly from influent (2.3 mg/L) to denitrification (2.8 mg/L) and after pre–ozonation (3.2 mg/L). However the stepwise increase was not statistically significant (ANOVA, p ¼ 0.05). Significant decrease by more than a factor of two in TEQ was observed after coagulation/flocculation/DAFF (1.4 mg/L). Baseline-TEQ after the main ozonation, activated carbon filtration and in the effluent ranged

avg

sd

avg

BDL (<20%) 58 19 53 12 42 12 80 14 92 17 BDL (<20%) 1093 930 129 13

sd

BDL (<10%) 429 272 133 70 146 93 1290 986 1187 758 2569 2116 1573 962 3205 4956

between 0.48 and 0.89 mg/L, levels not significantly different from the blank. Sand filtration as a polishing step directly after coagulation/flocculation/DAFF did not significantly alter the TEQ, while biological activated carbon filtration decreased the TEQs to the levels not significantly different from the blank (Table 5). Overall, DOC concentration correlated remarkably well with the baseline-TEQ (Fig. 2). The bioluminescence inhibition test cannot differentiate between micropollutants and organic

Table 5 – Summary of the bioanalytical results representing the average±sd of 4 samplings (expressed as equivalent concentrations except umuC assay where the data are expressed as REF of the sample required to elicit genotoxic effect). ID

Site description

Bioluminescence inhibition Baseline-TEQ (mg/L) avg

Blank S1 Influent (Effluent from WWTP) S2 Denitrification S3 Pre-ozonation S4 Coagulation/flocculation/ DAFF S5 Main ozonation S6 Activated carbon filtration S7 Effluent C3 Biological sand filter fed with S4 water C4 Biological activated carbon filter fed with S4 water C5 Biological activated carbon filter fed with S5 water

sd

AChE PTEQ (mg/L)

I-PAM

E-SCREEN

AhR-CAFLUX

umuC -S9

c

DEQ (mg/L) EEQ (ng/L) TCDDEQ (ng/L) 1/EC IR1.5 (1/REF)

avg sd avg

sd

avg

sd

avg

sd

avg

sd

0.21 2.3

0.01 0.7

<0.3 <0.01 3.2 1.0 0.12 0.1

<0.02 6.0 2.1

0.055 0.087b

0.01 0.03

<0.01 0.21

0.1

2.9 3.2a 1.4a

0.5 0.7 0.5

3.3 4.2 2.1

0.8 0.30 1.3 0.34 0.4 0.05

0.1 0.2 0.04

8.7 3.5 9.8a

0.10b 0.11b 0.11b

0.003 0.01 0.04

0.16 0.17 0.08a

0.04 0.06 0.06

0.91a,b 0.51a,b 0.52a,b 1.2a,b

0.2 0.2 0.2 0.3

1.9a 0.5 0.03 <0.3 0.02 0.36a 0.4 0.02 1.3a 0.1 0.08

0.03 0.02 0.01 0.06

0.43a 0.18 <0.02 <0.06 0.64a 0.53

0.091b 0.10b 0.10b 0.10b

0.02 0.06 0.01 0.01

0.02a <0.01 <0.01 0.075a

0.01

0.38a,b

0.1

<0.3

0.02

0.01

<0.02

0.070b

0.01

<0.01

0.34a,b

0.1

<0.3

0.04

0.03

<0.02

0.077b

0.02

<0.01

2.7 1.3 1.3

a Significantly different from influent, P < 0.05 tested against S1 with ANOVA, Bonferroni’s post test. b Not significantly different from the blank; ANOVA, Bonferroni’s post test. c TCDDEQ after acid silica gel clean-up.

0.08

484

water research 44 (2010) 477–492

Fig. 2 – Relationship between DOC concentration and baseline-TEQ. The black diamonds are the data from this study (from Tables 1 and 5). The empty squares refer to a previous study performed in Switzerland (all data refer to secondary effluent) (Escher et al., in press). Error bars indicate SD.

matter that is co-extracted with SPE. Assimilable organic carbon may be taken up by microorganisms and contribute to bioluminescence inhibition, while higher molecular-weight fractions of organic matter may bind micropollutants, potentially causing a decrease in their bioavailability and thus decrease the response. Note also that organic micropollutants contribute to DOC in DOC measurements. Therefore it is virtually impossible to differentiate between assimilable organic carbon and organic micropollutants. The TEQ of the influent of the reclamation plant, which is the effluent of a conventional wastewater treatment plant (WWTP), was higher than of the secondary effluent of a WWTP in Switzerland (Escher et al., 2008a; Escher et al., in press). It must also be noted that the DOC levels were significantly lower in the Swiss WWTP effluent than in the present study. In fact, the baseline-TEQ of the Swiss study (Escher et al., in press) fall on the lower left corner of Fig. 2, but they fit the correlation. This observation supports the conclusion that DOC contributed to the effects measured in this bioassay.

3.2. Neurotoxicity – acetylcholinesterase inhibition assay The concentration-effect curve of the reference compound, parathion, followed a log-logistic function with the slope 1 and the EC50 of 120 32 mg/L (Fig. 1C), using a commercially available acetylcholinesterase enzyme from Electrophorus electricus. The EC50 of parathion in our work was higher than data reported in literature: 27 mg/L (Escher et al., in press) and 26 mg/L (Escher et al., 2008a) with the use of bovine acetylcholinesterase. The difference in the sensitivity of the assay might be due to a different origin of acetylcholinesterase enzyme. The samples were tested at 12 different concentrations after serial dilution 1:2 with the REF range from 0.04 to 115. The best-fit slope of the samples’ concentration-effect curve was 1 (Fig. 1D) as it was for the reference compound. Effective concentrations required to elicit 50% inhibition are summarised in Table 4. Influent samples had to be enriched 52 times to elicit 50% inhibition, in contrast to effluent, which would

require an enrichment of 1100 times to elicit the same effect. As discussed above, we did not apply such high enrichment factors but had to extrapolate the EC50. The effect of the blank and samples treated with activated carbon was less than 20%, defined as the detection limit of the assay. Therefore the effective concentrations were not extrapolated from the concentration-effect curves. Results of the acetylcholinesterase inhibition assay were expressed as parathion equivalent concentrations (PTEQ) and are summarised in Table 5. PTEQ of the samples showed firstly a slight increase after pre-ozonation to 4.2 mg/L in comparison with the influent PTEQ of 3.1 mg/L, and then a significant decrease to 1.9 mg/L after the main ozonation. Activated carbon filtration further reduced PTEQ below the detection limit (<0.3 mg/L). The PTEQ in the effluent (0.36 mg/L) was only slightly above the detection limit of the assay. Sand filtration did not alter the PTEQ of the coagulation/flocculation/DAFF effluent, while treatment with the biological activated carbon filters reduced the PTEQ to below detection limit (<0.3 mg/L).

3.3.

Phytotoxicity – PS II inhibition I-PAM assay

The best-fit slope of the concentration-effect curve of the reference compound diuron was 1 and the EC50 was 16 6 mg/L (Fig. 1E). This EC50 represents an average of 12 concentrationeffect curves from 3 different runs indicating excellent repeatability of the assay. Diuron EC50 reported in this study was in the same order of magnitude as earlier work (6.7 mg/L) using Phaeodactylum tricornutum (Muller et al., 2007) and as the 2 hour endpoint in the combined algae test with Pseudokirchneriella subcapitata of 3 mg/L (Escher et al., 2008b; Escher et al., in press). The samples were enriched up to a REF of 258. The influent samples showed some cytotoxicity at these high enrichments so the phytotoxicity could not be evaluated. Despite the high enrichment of samples, the specific phytotoxic effects were still moderate, not exceeding 35% of the maximum effect of diuron. The best fit of the slope fitted to equation 5 would be higher than 1 (Fig. 1F, broken line) but given that only one data point forced this fit, the slope was fixed to the slope of the reference compound as usual (slope 1, Fig. 1F dotted line). We have previously observed that the yield of photosynthesis (Y)

Fig. 3 – Relationship between baseline-TEQ and PTEQ (triangles) or DEQ (diamonds) for the three samples influent, denitrification and pre-ozonation (S1 to S3). Error bars indicate SD.

water research 44 (2010) 477–492

485

Fig. 4 – Concentration-effect curves of selected samples and corresponding standards tested in E–SCREEN (A, B) and AhRCAFLUX (C, D) assay. The slope of the curves was fixed to 1. Bottom of the curves was an adjustable parameter in both assays. The top of the curve was adjustable in the E–SCREEN assay, while in AhR-CAFLUX, the top was set to the level of at least 100% of the TCDD maximal effect. Quality of the fit expressed as R2 ranged for selected samples and reference compounds from 0.96 to 0.99, except for Blank and S7-effluent samples tested in E-SCREEN assay, where the R2 was 0.3 and 0.4, respectively due to a very low response of both samples. Error bars indicate SD.

measurement is somewhat compromised, when several nonspecifically acting chemicals in the mixture mask the phytotoxic effect (Escher et al., 2008a). In fact, when single baseline toxicants (Escher et al., 2008b) were tested for the inhibition of photosynthesis yield they showed a response when measuring Y after 2 hours incubation, albeit at higher concentrations than for the corresponding growth inhibition endpoint, and the curve was steeper with a slope of 1.9. Since it is impossible to quantitatively differentiate between the contributions from phytotoxicity and cytotoxicity (unless cytotoxicity is so overwhelming that it makes the photosynthesis yield measurements impossible), we used the slope of the reference compound diuron to fit the samples even though the quality of fit was sub-optimal. The phytotoxic response of the samples after 2 hours of incubation expressed as diuron equivalent concentration DEQ is summarised in Table 5. Denitrification significantly increased the DEQ (0.29 mg/L) in comparison with the influent DEQ of 0.11 mg/L. The pre-ozonation step itself had no significant effect (0.34 mg/L). After coagulation/flocculation/DAFF, the DEQ decreased by 85% to 0.05 mg/L. DEQ decreased in small steps after the main ozonation and activated carbon filtration stages and in the effluent samples. No significant additional decrease in DEQ was observed after sand filter or biological activated carbon filters. All DEQ after coagulation/flocculation/DAFF are close to the detection limit and are based on

a single data point, therefore should be interpreted with some caution. The initial slight but not statistically significant increase in TEQ from sample S1 to S3 is unexpected at first sight. It is interesting to note, however, that this small increase was common for baseline toxicity and slightly less clear also for neurotoxicity (Figs. 2 and 3). The increase in baseline toxicity was explained by an increase of DOC, which is presumably partially co-extracted at low pH. The small increase in DEQ could also be caused by baseline toxicants interfering with the measurement of the photosynthesis yield. We have not observed such an effect in our previous work (Escher et al., 2008a; Escher et al., in press). However, we have never encountered DOC concentrations exceeding 10 mg/L in any of our previous work. It is unclear how much of the DOC is coextracted with the micropollutants by SPE. The effect of DOC needs to be investigated in future work but is beyond the scope of the present study. Possibly, extraction at a higher pH might reduce the fraction of co-extracted DOC due to the higher negative charge of DOC under these conditions. Previous work on comparing herbicides concentrations with DEQ often showed a relatively good correlation and a substantial fraction (> 50% of the response) could be explained by the measured herbicides (Bengtson Nash et al., 2005b; Bengtson Nash et al., 2006; Escher et al., 2006; Muller et al., 2008). There were some cases, such as samples collected

486

water research 44 (2010) 477–492

Table 6 – Sample enrichment required to produce a defined effect in each assay: 50% effect in E–SCREEN assay and 20% of maximal TCDD effect in AhR-CAFLUX assay. Results represent the average ± sd of 4 samplings. ID

Site description

E-SCREEN RPE

S1 S2 S3 S4 S5 S6 S7 C3 C4

C5

Blank Influent (Effluent from WWTP) Denitrification Pre-ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water

Mode of Action

avg

sd

0.05 0.85 0.76 0.76 1.0 0.92 0.19 0.27 0.65

– 0.2 0.1 0.2 0.2 0.2 0.1 0.1 0.2

Not estrogenic Full agonist Partial agonist Partial agonist Full agonist Full agonist Not estrogenic Not estrogenic Partial agonist

0.12

0.06

0.16

0.1

EC50 (REF) avg

After Clean-up EC20 TCDD (REF)

sd

avg

sd

BDL (<20%) 0.45 0.1 0.32 0.1 0.79 0.2 0.27 0.06 6.7 2 BDL (<20%) BQL (<50%) 3.1 1

13 1.6 1.4 1.7 1.6 3.1 4.4 4.4 1.7

1 0.6 0.4 0.3 0.1 0.6 2 0.9 0.3

28 19 14 14 16 17 20 15 16

0.4 6 2 1 6 3 12 2 0.9

Not estrogenic

BDL (<20%)

6.1

1

22

3

Not estrogenic

BDL (<20%)

8.2

2

21

5

Estrogenic activity: E-SCREEN assay

In the E-SCREEN, bottom and top of the concentration-effect curve (Eq. 4), both adjustable parameters, refer to the minimum or maximum stimulatory response, i.e., proliferation of the estrogen-dependent cells in E–SCREEN. The slopes of the concentration-effect curves of the reference compound, 17ß-estradiol, and the samples were fixed to 1 (Figs. 4A and B). The maximum of the curve yields information on the mode of action, thus being able to distinguish among full agonist,

sd

Before Clean-up EC20 TCDD (REF) avg

from the Thames River, where the biological DEQ were much higher than the chemically analysed DEQ (Bengtson Nash et al., 2006). The authors suggested this difference was due to the presence of additional herbicides that were not analysed with the analytical chemical methodology, but given the new evidence it is also possible that additional cytotoxicity by nonspecifically acting micropollutants and DOC could have contributed to the discrepancy.

3.4.

AhR-CAFLUX

partial agonist and no estrogenic effect, a parameter called relative proliferative effect (RPE). The reference compound 17ß-estradiol had an EC50 of 2.4 ng/L, which is in good agreement with previous work of 2.2 ng/L (Ko¨rner et al., 1999). EC50 of the estradiol in the E–SCREEN assay was comparable with the EC50 in the ER– CALUX assay of 2.3 ng/L (Legler et al., 2002), indicating the similar sensitivity of both assays. The detection limit of the E– SCREEN assay was defined as 20% of the maximal estradiol effect. The detection limit for the whole method was calculated based on the highest relative enrichment of the samples, providing the sample was not cytotoxic, resulting in a detection limit of 0.02 ng/L. Samples collected in the influent and after denitrification were cytotoxic at the highest concentrations tested, therefore the two highest concentrations were excluded and the detection limit of the whole method for those samples was 0.2 ng/L. The assay is still more sensitive than chemical analysis of the estrogenic compounds. Nine concentrations of the sample extract serially diluted 1:4 were tested in the E–SCREEN assay, resulting in REF from 0.0006

Fig. 5 – Concentration-effect assessment of selected samples tested in umuC assay without (A) or with (B) metabolic activation. Samples follow a linear function with respect to non-logarithmic concentration. Quality of the fit expressed as R2 ranged for selected samples from 0.80 to 0.99, except for Blank and S7-effluent samples. Error bars indicate SD.

water research 44 (2010) 477–492

Table 7 – Sample enrichment required to produce a genotoxic effect [ Induction Ratio of 1.5 (IR1.5) in umuC assay without and with metabolic activation. Results represent the average ± sd of 4 samplings. ID

Site description

umuC -S9 EC

IR1.5

avg

S1 S2 S3 S4 S5 S6 S7 C3 C4

C5

Blank Influent (Effluent from WWTP) Denitrification Pre-ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water

>78 5.2 6.9 6.3 15 78 >84 >69 24

umuC þ S9

(REF)

EC

sd

avg

2 2 2 7 20

13

IR1.5

>78 17 22 18 28 >87 >84 >73 34

>80

>80

>79

>79

(REF) sd

3 3 3 8

16

(1667 times dilution) to 55 (55 times enrichment of the original water sample). Sample REF required to produce a 50% effect (EC50) in the E–SCREEN assay ranged from 0.27 (equivalent to 4 times dilution of the original water sample) to 6.7 (equivalent to 6.7 times enrichment of the original water sample) (Table 3). The wide range of REF windows allows one to monitor a wide range of samples collected throughout the treatment chain. The estrogenicity of the samples was quantified with two parameters: the relative proliferative effect (RPE) and the estradiol equivalent concentration (EEQ). RPE of the samples, representing the ratio of the maximum proliferation induced by test samples compared to 17b-estradiol (also referred to as relative efficacy), are summarised in Table 6. The samples from the influent up to ozonation were fully or partially agonistic, while the remainder of the samples were not classified estrogenic and no EEQ could be derived for these samples. When a sample was classified as either a partial or full agonist, an estradiol equivalent concentration (EEQ) was calculated from the EC50. Influent to the reclamation plant (representing the effluent of the WWTP) showed full agonistic activity (RPE > 0.8) with the estradiol equivalent concentration EEQ of 6 ng/L (Table 5). This was comparable with the EEQs reported for WWTP effluent in the literature: <1–7.8 ng/L in Germany (E–SCREEN assay, (Ko¨rner et al., 2001)), <1 to >10 over a wide range of different WWTPs and 0.8–5.9 ng/L in one given WWTP at various samplings during three years in Switzerland (YES assay, (Rutishauser et al., 2004; Escher et al., in press)), <1–4.2 ng/L in Australia and New Zealand (estrogen receptor binding assay, (Leusch et al., 2006)), <1–16 ng/L in the Netherlands (ER–CALUX, (Murk et al., 2002)), 1–67 ng/L in Australia (E–SCREEN assay (Tan et al., 2007)). Samples collected after denitrification and pre-ozonation (S2 and S3, respectively) showed partial agonistic activity (RPE of 0.5–0.8), and the EEQs were not significantly different from the influent. Surprisingly, EEQ of the sample collected after coagulation/flocculation/DAFF treatment was significantly

487

higher (EEQ of 9.8 ng/L) than in influent and the sample again showed full agonistic activity. Full agonistic activity was also detected after the main ozonation (RPE > 0.8); however, the EEQ was significantly decreased to 0.43 ng/L. The remarkable removal efficiency of ozonation on estrogenicity is well established (Lee et al., 2008) and in a recent full-scale ozonation in a WWTP in Switzerland the ozonation step showed additional >95% removal efficiency (provided the influent of the ozonation reactor had EEQ > 2 ng/L) and the resulting effluent of the ozonation contained 0.6 ng/L EEQ (Escher et al., in press). Estrogenicity was further markedly altered after activated carbon filtration. Estrogenic effect of the sample treated with activated carbon (S6) was less than 20% of the estradiol effect (RPE < 0.2), which corresponds to the detection limit of the assay, therefore the EEQ was less than 0.02 ng/L. Estrogenicity (RPE) of the effluent sample was slightly increased, but did not reach 50% effect of estradiol, therefore the sample was classified as not estrogenic and the EEQ was not quantified. The estrogenic effect was below quantification limit. Taking into account the REF of the sample in this assay, the EEQ of the effluent was less than 0.06 ng/L. Besides the main treatment chain, additional treatment processes were assessed in this study: sand filtration of the coagulation/flocculation/DAFF effluent (C3) and biological activated carbon filtration of the coagulation/flocculation/ DAFF effluent (C4) and main ozonation effluent (C5). Sand filtration significantly decreased the EEQ of the coagulation/ flocculation/DAFF effluent from 9.8 ng/L to 0.63 ng/L. Biological activated carbon filtration decreased the estrogenic effect of both effluents to below detection limit (<0.02 ng/L). The same assay was also applied to detect estrogenicity in the raw sewage sample from Caboolture WWTP with high EEQ ranging from 68 to 91 ng/L in 3 different samples collected on different days (unpublished results). The levels were comparable with EEQ recently reported in the raw sewage in the Brisbane area using the same bioassay, with up to 74 ng/L EEQ (GWRC, 2008). The measured EEQ in the raw sewage was also similar to the levels reported in other studies, <4–185 ng/L in Australia and New Zealand (estrogen receptor binding assay, (Leusch et al., 2006)), up to 120 ng/L in the Netherlands (ER–CALUX, (Murk et al., 2002)) and 58–70 ng/L in Germany (E–SCREEN, (Ko¨rner et al., 2000)).

3.5.

Ah-receptor response: AhR-CAFLUX assay

In the AhR-CAFLUX assay, the bottom and top of the curve refer to the minimum and maximum induction of fluorescence under the control of AhR. Sample responses in AhRCAFLUX assay measured in relative fluorescence units (RFU) were converted to a percentage of maximal response of reference compound 2,3,7,8-tetrachlorodibenzo-p-dioxin (% TCDD max response). If the effect of the sample was higher than maximal TCDD effect, the top of the curve was adjustable. However, if the sample did not reach 100% TCDD effect, the top of the curve was fixed to 100. The slope of the concentration-effect curve of the reference compound TCDD and the samples was fixed to 1 (Figs. 4C and D). The EC50 of the reference compound TCDD ranged between 6.3 and 9.2 ng/L over 5 different runs, confirming the

488

water research 44 (2010) 477–492

reproducibility and repeatability of the assay and is in good agreement with the previous levels from the literature of 5.9 ng/L (Nagy et al., 2002) and 3.9 ng/L (Zhao and Denison, 2004). TCDD was chosen as the reference compound as it is one of the most potent agonists for this assay. However, high affinity ligands for the AhR also include dibenzofurans and biphenyls as well as a variety of polycyclic aromatic hydrocarbons, benzoflavones and other chemicals (Denison and Nagy, 2003). To assess the contribution of chemicals other than dioxins, furans and dioxin-like PCBs to the potency estimates, sulfuric acid silica gel clean-up was performed in this study, which removed most organic chemicals (e.g. PAHs) except persistent chemicals such as dioxins, furans and PCBs. Original sample extracts, as well as the extracts cleaned with sulfuric acid silica gel were tested with the AhR-CAFLUX assay at 5 different concentrations after serial dilution 1:10, with the REF ranging from 55 to 0.004. The response of many samples, especially after clean-up, did not reach 50% TCDD response. Therefore effective concentrations of the samples required to elicit 20% effect of TCDD (EC20 TCDD) were interpolated from concentration-effect curves (Table 6). The EC20 TCDD of the samples ranged from 1.4 to 8.2. The effective concentration of the same samples was significantly altered after clean up and ranged from 14 to 22, i.e. samples after acid silica gel clean up had to be enriched approximately 3–10 times more to elicit the same effect. The response in the AhR-CAFLUX assay was expressed as the 2,3,7,8-TCDD-equivalent concentration (TCDDEQ). TCDDEQ was significantly decreased by the main ozonation to 0.44 ng/L in comparison with the TCDDEQ of 0.83 ng/L in the influent. Sand filtration did not change the TCDDEQ of the coagulation/flocculation/DAFF effluent, while treatment with the biological activated carbon filters reduced the TCDD equivalent concentration to 0.33 ng/L a level not significantly different from the blank. The TCDDEQ of the samples after the sulfuric acid silica gel clean-up were reduced to levels not significantly different from the blank (Table 5). These results indicate that the observed effect in the AhR-CAFLUX assay prior to the sulfuric acid silica gel clean-up step may be attributed to chemicals other than dioxins, furans and PCBs.

3.6.

Genotoxicity-UmuC assay

The umuC genotoxicity assay can detect both cytotoxic and genotoxic effects. The extracts were tested both with (þS9) and without (S9) exogenous metabolic activation. Response on this assay is determined as an induction ratio (IR) (Fig. 5). An IR 1.5 is considered genotoxic, providing the sample is not cytotoxic (growth < 0.5). For samples that demonstrate significant genotoxic response, the effect concentrations ECIR 1.5 were derived from the linear concentration-effect curve as depicted in Fig. 5. Effect concentrations are in dimensionless units of REF and represent how many times the samples must be concentrated or diluted to elicit a threshold IR of 1.5 in the assay (Table 7). Results are expressed as 1/ECIR 1.5 therefore a higher number represents a higher genotoxic effect (Table 5). Unfortunately no TEQ could be derived for this endpoint but in theory this should be possible and in the future we will work on implementing the TEQ concept for this assay. This is somehow

problematic because of the frequently observed cytotoxicity of the samples, which can only be partially accounted for. Overall, the umuC test without metabolic activation gave higher responses than the test with metabolic activation indicating that some of the micropollutants that are responsible for genotoxicity are detoxified by the liver enzyme fraction S9. As in all other bioassays, the denitrification and preozonation did not affect the activity in the umuC test but coagulation/flocculation/DAFF decreased the genotoxicity without S9 by 58% and that with S9 by 35%. Subsequent main ozonation completely removed the effect for the test with prior S9 treatment and drastically reduced it in the absence of S9.

4.

Conclusion

A similar test battery has been previously used for various applications in wastewater treatment efficiency and water quality assessment. To apply the test battery to a treatment train with very subtle steps and for relatively pure water was a challenge. This challenge was met by the following measures: A comprehensive framework was developed that centres around the bioassays but links to sample preparation with SPE prior to testing and to the data evaluation scheme. Quality control was implemented and all effects were related to reference compounds, thereby correcting for slight day-to-day variability of the test results. The use of SPE allows a relatively non-specific extraction of water samples to concentrate a wide range of relevant pollutants to levels where they can exhibit a measurable response, while matrix components such as salts and metals are removed from the sample during SPE. The detection window of the bioassays is adjustable. For the application in the present study the concentration range of the sample in the bioassay ranged between 1700 times dilution and 250 times enrichment of the sample. The detection window depends on the constitution of the sample and the type of bioassay and is only limited by the effect of the blank at very high enrichment and nonspecific cytotoxicity of the sample for each of the assays. By dynamically changing the relative enrichment window of the sample throughout the study one can adapt the various bioassays to a wide variety of samples and results are robust and reproducible. Results are reported as toxic equivalent concentrations, a format that allows for comparison to results from chemical analysis. This is also a key feature of the bioanalytical toolsthey are no direct indicators for environmental health but they are indicators of the presence of mixtures of chemicals, accounting for the mixture toxicity of the unidentified chemicals in an environmental sample. To report bioanalytical results as toxic equivalent concentrations is better suited for hazard assessment and extrapolation than the commonly reported % effect of the water sample in a given bioassay because it is too difficult to extrapolate effects while it is possible to compare effect concentrations of chemicals across different level of biological organisation. So for example, if a sample elicits 0.5 mg/L DEQ one can compare this value with the environmental quality standard for

water research 44 (2010) 477–492

diuron to estimate what potential long-term effects could be expected from such an exposure. The study also highlighted some difficulties that remain and some lessons are to be learned. One of the weak points of the approach is the sample preparation with SPE. However, this is a common problem for chemical analysis and can only be circumvented if there are recovery standards for every single compound to be analysed. Such an approach is not possible for bioassays because spiked chemicals would add to the mixture toxicity. A shortcoming of the SPE method that was noticed here for the first time is the co-extraction of an unknown fraction of dissolved organic matter, an issue that requires further attention in the future. In addition, volatile compounds will evaporate during SPE but they would also cross contaminate in a 96 well plate, so it is a procedural advantage to have removed the volatile chemicals prior to toxicity testing. It must also be noted that oxidation by-products formed during ozonation are often more hydrophilic than their parent compound and might therefore not be extracted efficiently by SPE. In addition, one must be aware that mixture effects are assessed when applying bioanalytical tools to complex samples and there is always the problem that the subtle effect of newly formed by-products might be shielded by other micropollutants present in the mixture. Some of the oxidation byproducts stemming from ozonation are very reactive, for example aldehydes and other electrophiles formed from ß-blockers during ozonation (Benner and Ternes, 2009), others are very hydrophilic organics such as NDMA (N-nitrosodimethylamine) or inorganics such as bromate (Hollender et al., in press) none of these would be caught with the bioanalytical tools used here but require targeted chemical analysis for quantification. However, ozonation often only mildly oxidizes organic micropollutants and many resulting transformation products are likely to retain some of their toxic potential (Escher et al., 2008c) or lose their specific receptor-binding effect but retain non-specific baseline toxicity (Lee et al., 2008) and will thus contribute to the mixture toxicity in the bioanalytical tools applied in the present study. While the results are discussed in more detail and compared to chemical analysis in the accompanying paper (Reungoat et al., 2010), some general conclusions can be drawn: The influent of the water reclamation plant had similar levels of effects for the different endpoints as we have encountered in previous studies for wastewater after secondary treatment (Escher et al., 2008a; Escher et al., in press). The effluent of the water reclamation plant had strongly reduced effect levels compared to the influent, reducing the baseline-TEQ by 79%, the DEQ by 76%, the estrogenicity to below the detection limit of 0.02 ng/L EEQ, the PTEQ by 88% and the genotoxicity to below the detection limit. Only the AhR-CAFLUX assay did not yield any information on the removal efficiency of the treatment because the level of persistent inducers of the arylhydrocarbon receptor remained close to the detection limit. This was not too unexpected given that these are typically very hydrophobic compounds, which are usually adsorbed and not present in

489

the water column. In contrast, the non-persistent inducers of the AhR were clearly reduced by a factor of two during ozonation. The most efficient steps with respect to removal of the toxic responses in all selected assays proved to be the coagulation/flocculation/DAFF, the main ozonation and (biological) activated carbon filtration steps. This paper demonstrates the first comprehensive application of the bioanalytical tool framework on an advanced water treatment plant. One future goal would be to automate the assays to a degree that they can be used routinely by commercial laboratories or water companies themselves. At this stage, there remain some scientific questions to be resolved, including (1) the sample treatment and dosing, (2) extension of the battery and (3) data interpretation. (1) SPE and dosing by solvent spiking will always change the composition of an undefined mixture. A combination of passive sampling to a biomimetic polymer and passive dosing directly from the polymer into the bioassay might be a way to overcome this limitation. (2) A battery of bioassays will never be comprehensive because of the myriads of receptors and regulatory pathways in an organism. Nevertheless the battery would benefit from inclusion of additional endpoints, such as chemically induced immunosuppression or even other examples of mode-of-action categories already included, e.g. other endocrine endpoints. (3) At this stage we can say how much of a toxic equivalent concentrations is removed by a certain treatment step and we can give comparative results but we will never be able to deduce ecological or health consequences from that. This discussion must be lead independently. One solution would be to compare the TEQ of the mixture with the defined Environmental Water Quality Criteria (EQS) for single substances of the Water Framework Directive (European Commission, 2006) or guideline values for drinking water (e.g. NHMRC–NRMMC, 2004) but not for all reference compounds are such EQS and guideline values available.

Acknowledgements This work was co-funded by the Urban Water Security Research Alliance under the Enhanced Treatment Project, the CRC Water Quality and Treatment Project No. 2.0.2.4.1.1. Dissolved Organic Carbon Removal by Biological Treatment and by EnTox. The National Research Centre for Environmental Toxicology (EnTox) is a joint venture of the University of Queensland and Queensland Health Forensic and Scientific Services (QHFSS). The authors acknowledge the Moreton Bay Water for access to the South Caboolture Water Reclamation Plant; Ray McSweeny and Paul McDonnell (Moreton Bay Water) for their help during sampling and Chris Pipe-Martin (Ecowise) for providing information on the South Caboolture Water Reclamation Plant. The authors are grateful to Renee Muller (Gold Coast Water), Fred Leusch (Griffith University/EnTox), Karen Kennedy (EnTox) and Pam Quayle (EnTox) for constructive discussions and for help with the data analysis method, Marita Goodwin (EnTox) for running various assays and for helpful comments Christina Carswell (QHFSS) for extracting the

490

water research 44 (2010) 477–492

samples and Chris Paxman (EnTox) for technical assistance. We would like to thank also Michael Denison (University of California Davis, USA) for providing the H4G1.1c2 cells, Georg Reifferscheid (German Federal Institute of Hydrology, Germany) for providing the bacteria Salmonella typhimurium TA1535/pSK1002, and Ana Soto (Tufts University, USA) for providing the MCF7-BOS cells.

Appendix. Supporting information Supporting information related to this article can be found at doi:10.1016/j.watres.2009.09.025.

references

Bengtson Nash, S.M., Schreiber, U., Ralph, P.J., Mueller, J.F., 2005a. The combined SPE: ToxY-PAM phytotoxicity assay; application and appraisal of a novel biomonitoring tool for the aquatic environment. Biosensors and Bioelectronics 20 (7), 1443–1451. Bengtson Nash, S.M., McMahon, K., Eaglesham, G., Mueller, J.F., 2005b. Application of a novel phytotoxicity assay for the detection of herbicides in Hervey Bay and the Great Sandy Straits. Marine Pollution Bulletin 51, 351–360. Bengtson Nash, S.M., Goddard, J., Muller, J.F., 2006. Phytotoxicity of surface waters of the Thames and Brisbane river estuaries: a combined chemical analysis and bioassay approach for the comparison of two systems. Biosensors and Bioelectronics 21, 2086–2093. Benner, J., Ternes, T.A., 2009. Ozonation of propranolol: formation of oxidation products. Environmental Science & Technology 43, 5086–5093. Cao, N., Yang, M., Zhang, Y., Hu, J., Ike, M., Hirotsuji, J., Matsui, H., Inoue, D., Sei, K., 2009. Evaluation of wastewater reclamation technologies based on in vitro and in vivo bioassays. Ecotoxicology and Environmental Safety 407 (5), 1588–1597. Chapman, P.M., 2000. Whole effluent toxicity testing – usefulness, level of protection, and risk assessment. Environmental Toxicology and Chemistry 19 (1), 3–13. Denison, M.S., Nagy, S.R., 2003. Activation of the aryl hydrocarbon receptor by structurally diverse exogenous and endogenous chemicals. Annual Reviews in Pharmacology and Toxicology 43, 309–334. Deutsche Norm, 1995. In: DIN, N.W.N.i. (Ed.), Suborganismische Verfahren: Teil 1: Bestimmung von CholinesteraseHemmenden Organophosphat- und Carbamat-Pestiziden (Cholinesterase-Hemmtest) DIN 38415-1: 1995-02, Berlin, Germany. Ellman, G.L., Courtney, K.D., Andres, V., Featherstone, R.M., 1961. A new and rapid colorimetric determination of acetylcholinesterase activity. Biochemical Pharmacology 7, 88–95. Escher, B., Schwarzenbach, R.P., 2002. Mechanistic studies on baseline toxicity and uncoupling as a basis for modeling internal lethal concentrations in aquatic organisms. Aquatic Sciences 64, 20–35. Escher, B.I., Bramaz, N., Maurer, M., Richter, M., Sutter, D., von Ka¨nel, C., Zschokke, M., 2005. Screening test battery for pharmaceuticals in urine and wastewater. Environmental Toxicology and Chemistry 24 (3), 750–758. Escher, B.I., Quayle, P., Muller, R., Schreiber, U., Mueller, J., 2006. Passive sampling of herbicides combined with effect analysis in algae using a novel high-throughput phytotoxicity assay

(Maxi-Imaging- PAM). Journal Environmental Monitoring 8, 456–464. Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S., Vermeirssen, E., 2008a. Monitoring of the ecotoxicological hazard potential by polar organic micropollutants in sewage treatment plants and surface waters using a mode-of-action based test battery. Journal Environmental Monitoring 10, 622–631. Escher, B.I., Bramaz, N., Mueller, J.F., Quayle, P., Rutishauser, S., Vermeirssen, E., 2008b. Toxic equivalent concentrations (TEQs) for baseline toxicity and specific modes of action as a tool to improve evaluation of ecotoxicity tests on environmental samples. Journal Environmental Monitoring 10, 612–621. Escher, B.I., Baumgartner, R., Lienert, J., Fenner, K., 2008c. Predicting the ecotoxicological effects of transformation products. SpringerLink online. In: Boxall, A. (Ed.), Handbook of Environmental Chemistry – Degradation of Synthetic Chemicals in the Environment. Springer, Heidelberg, Germany. doi:10.1007/6998_2_015. http://www.springerlink. com/content/48r2t364q183q654/?p ¼ ec460152bfd 14906adc89378463e4997&pi ¼ 5 published Wednesday, April 30, 2008. Escher, B.I., Bramaz, N. and Ort, C., 2009. Monitoring the treatment efficiency of a full scale ozonation on a sewage treatment plant with a mode-of-action based test battery. Journal Environmental Monitoring, doi: 10.1039/ b907093a. European Commission, 2006. Proposal for a Directive of the European Parliament and the Council on Environmental Quality Standards in the Field of Water Policy and Amending Directive 2000/60/EC COM(2006) 398 Final. COM(2006) 398 Final. European Parliament, Committee on the Environment, Public Health and Food Safety, Brussels. Farre´, M., Martinez, E., Hernando, M., Fernandez-Alba, A., Fritz, J.I., Unruh, E., Mihail, O., Sakkas, V., Morbey, A., Albanis, T.A., Brito, F., Hansen, P.D., Barcelo, D., 2006. European ring exercise on water toxicity using different bioluminescence inhibition tests based on Vibrio fischeri, in support to the implementation of the water framework directive. Talanta, 323–333. GWRC, 2008. In: Leusch, F.D.L. (Ed.), Tools to Detect Estrogenic Activity in Environmental Waters. Global Water Research Coalition, London, UK, p. 84. Hamers, T., Molin, K.R.J., Koeman, J.H., Murk, A.J., 2000. A smallvolume bioassay for quantification of the esterase inhibiting potency of mixtures of organophosphates and carbamate insceticides in rainwater: development and optimization. Toxicological Sciences 58, 60–67. Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M., McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H., 2009. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale postozonation followed by sand filtration. Environmental Science & Technology, Articles ASAP (Article), in press. doi: 10.1021/ es9014629. Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of estrogenic activity and formation of oxidation products during ozonation of 17 alpha-ethinylestradiol. Environmental Science & Technology 38, 5177–5186. Huber, M.M., Go¨bel, A., Joss, A., Hermann, N., Lo¨ffler, D., McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science & Technology 39, 4290–4299. International Standard Organisation 1998 Water Quality– determination of the inhibitory effect of water samples on the light emission of Vibrio fischeri (luminescent bacteria test), EN ISO 11348-3, Geneva, Switzerland.

water research 44 (2010) 477–492

International Standard Organisation 2000 Water quality – Determination of the genotoxicity of water and waste water using the umu-test, EN ISO 13829(2000) and 38415-3 (1996), Geneva, Switzerland. Johnson, B.T., 2005. Microtox acute toxicity test. In: Blaise, C., Ferard, J.-F. (Eds.), Small-Scale Freshwater Toxicity Investigations. Toxicity Test Methods, vol. 1. Springer. Joss, A., Keller, E., Alder, A.C., Gobel, A., McArdell, C.S., Ternes, T., Siegrist, H., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Research 39 (14), 3139–3152. Ko¨rner, W., Hanf, V., Schuller, W., Kempter, C., Metzger, J., Hagenmaier, H., 1999. Development of a sensitive E-screen assay for quantitative analysis of estrogenic activity in municipal sewage plant effluents. Ecotoxicology and Environmental Safety 225 (1–2), 33–48. Ko¨rner, W., Bolz, U., Sussmuth, W., Hiller, G., Schuller, W., Hanf, V., Hagenmaier, H., 2000. Input/output balance of estrogenic active compounds in a major municipal sewage plant in Germany. Chemosphere 40 (9–11), 1131–1142. Ko¨rner, W., Spengler, P., Bolz, U., Schuller, W., Hanf, V., Metzger, J.W., 2001. Substances with estrogenic activity in effluents of sewage treatment plants in southwestern Germany. 2. Biological analysis. Environmental Toxicology and Chemistry 20 (10), 2142–2151. Lee, Y., Escher, B.I., von Gunten, U., 2008. Efficient removal of estrogenic activity during oxidative treatment of waters containing steroid estrogens. Environmental Science & Technology 42 (17), 6333–6339. Legler, J., Zeinstra, L.M., Schuitemaker, F., Lanser, P.H., Bogerd, J., Brouwer, A., Vethaak, A.D., De Voogt, P., Murk, A.J., Van der Burg, B., 2002. Comparison of in vivo and in vitro reporter gene assays for short-term screening of estrogenic activity. Environmental Science & Technology 36 (20), 4410–4415. Leusch, F.D., Chapman, H.F., van den Heuvel, M.R., Tan, B.L., Gooneratne, S.R., Tremblay, L.A., 2006. Bioassay-derived androgenic and estrogenic activity in municipal sewage in Australia and New Zealand. Ecotoxicology and Environmental Safety 65 (3), 403–411. Muller, R., Tang, J., Thier, R., Mueller, J., 2007. Combining passive sampling and toxicity testing for evaluation of mixtures of polar organic chemicals in sewage treatment plant effluents. Journal Environmental Monitoring 9, 104–109. Muller, R. 2008 Passive sampling and bioanalysis of organic chemicals in surface water. Ph.D. thesis. University of Queensland, St. Lucia, Qld, Australia, 201pp. Accessible at http://library.uq.edu.au/search~S7?/XRenee þ Muller&searchscope ¼ 7&SORT ¼ AZ/XRenee þ Muller&searchscope ¼ 7&SORT ¼ AZ&extended ¼ 0&SUBKEY ¼ Renee%20Muller/21%22C23%22C23%22CB/frameset&FF ¼ XRenee þ Muller&searchscope ¼ 27&SORT ¼ AZ&21%22C21%22C Muller, R., Schreiber, U., Escher, B.I., Quayle, P., Bengtson Nash, S., Mueller, J., 2008. Rapid exposure assessment of PSII herbicides in surface water using a novel chlorophyll a fluorescence imaging assay. Science Total Environment 401, 51–59. Murk, A.J., Legler, J., van Lipzig, M.M.H., Meerman, J.H.N., Belfroid, A.C., Spenkelink, A., van der Burg, B., Rijs, G.B.J., Vethaak, D., 2002. Detection of estrogenic potency in wastewater and surface water with three in vitro bioassays. Environmental Toxicology and Chemistry 21 (1), 16–23. Nagy, S., Sanborn, J., Hammock, B., Denison, M., 2002. Development of a green fluorescent protein-based cell bioassay for the rapid and inexpensive detection and characterization of Ah receptor agonists. Toxicological Sciences 65, 200–210. NHMRC–NRMMC, 2004. Australian Drinking Water Guidelines (ADWG). National Health and Medical Research Council

491

(NHMRC) in Collaboration With the Natural Resource Management Ministerial Council (NRMMC), Commonwealth of Australia. www.nhmrc.gov.au/publications/synopses/_ files/adwg_11_06.pdf. Oda, J., Nakamura, S.I., Oki, I., Kato, T., Shinagawa, H., 1985. Evaluation of the new system (umu-test) for the detection of environmental mutagens and carcinogens. Mutatation Research 147, 219–229. Petala, M., Samaras, P., Zouboulis, A., Kungolos, A., Salkellaropoulos, G., 2006. Ecotoxicological properties of wastewater treated using tertiary methods. Environmental Toxicology 21 (4), 417–424. Petala, M., Samaras, P., Zouboulis, A., Kungolos, A., Sakellaropoulos, G.P., 2008. Influence of ozonation on the in vitro mutagenic and toxic potential of secondary effluents. Water Research 42 (20), 4929–4940. Reifferscheid, G., Heil, J., Oda, Y., Zahn, R.K., 1991. A microplate version of the SOS/umu-test for rapid detection of genotoxins and genotoxic potentials of environmental samples. Mutation Research 253 (3), 215–222. Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F., Keller, J., 2010. Removal of micropollutants and reduction of biological adverse effects in a full scale reclamation plant using ozonation and activated carbon filtration. Water Research 44 (2), 625–637. Rutishauser, B.V., Pesonen, M., Escher, B.I., Ackermann, G.E., Aerni, H.-R., Suter, M.J.-F., Eggen, R.I.L., 2004. Comparative analysis of estrogenic activity in sewage treatment plant effluents involving three in vitro assays and chemical analysis of steroids. Environmental Toxicology and Chemistry 23, 857–864. Sanchez-Polo, M., Salhi, E., Rivera-Utrilla, J., von Gunten, U., 2006. Combination of ozone with activated carbon as an alternative to conventional advanced oxidation processes. Ozone-Science & Engineering 28 (4), 237–245. Schreiber, U., Mueller, J.F., Haugg, A., Gademann, R., 2002. New type of dual-channel PAM chlorophyll fluorometer for highly sensitive water toxicity biotests. Photosynthesis Research 74, 317–330. Schreiber, U., Quayle, P., Schmidt, S., Escher, B.I., Mueller, J., 2007. Methodology and evaluation of a highly sensitive algae toxicity test based on multiwell chlorophyll fluorescence imaging. Biosensors and Bioelectronics 22, 2554–2563. Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge of micropollutants in aquatic systems. Science 313 (5790), 1072–1077. Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J., Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of membranes and activated carbon in the removal of endocrine disruptors and pharmaceuticals. Desalination 202, 156–181. Soto, A.M., Sonnenschein, C., Chung, K.L., Fernandez, M.F., Olea, N. , Serrano, F.O., 1995. The E-SCREEN assay as a tool to identify estrogens: an update on estrogenic environmental pollutants. Environmental Health Perspectives 103 (Suppl. 7), 113–122. Tan, B.L.L., Hawker, D.W., Muller, J.F., Leusch, F.D.L., Tremblay, L.A., Chapman, H.F., 2007. Comprehensive study of endocrine disrupting compounds using grab and passive sampling at selected wastewater treatment plants in South East Queensland, Australia. Environment International 33 (5), 654–669. Ternes, T.A., Joss, A., Siegrist, H., 2004. Scrutinizing pharmaceuticals and personal care products in wastewater treatment. Environmental Science & Technology 38 (17), 393A–399A. van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water reclamation at South Caboolture, Queensland, Australia. Ozone: Science & Engineering 25, 107–120. Villeneuve, D.L., Blankenship, A.L., Giesy, J.P., 2000. Derivation and application of relative potency estimates based on in-vitro bioassays. Environmental Toxicology Chemistry 19, 2835–2843.

492

water research 44 (2010) 477–492

von Gunten, U., 2003. Ozonation of drinking water: part I. Oxidation kinetics and product formation. Water Research 37 (7), 1443–1467. Warne, M.S.J., Hawker, D.W., 1995. The number of components in a mixture determines whether synergistic and antagonistic or additive toxicity predominate – the funnel

hypothesis. Ecotoxicology and Environmental Safety 31 (1), 23–28. Zhao, B., Denison, M., 2004. Dvelopment and characterization of a green fluorescence protein-based rat cell bioassay system for detection of Ah receptor ligands. Organohalogen Compounds 66, 3332–3337.

water research 44 (2010) 493–504

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

An evaluation of a pilot-scale nonthermal plasma advanced oxidation process for trace organic compound degradation Daniel Gerrity*, Benjamin D. Stanford, Rebecca A. Trenholm, Shane A. Snyder Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, P.O. Box 99954, Las Vegas, NV 89193-9954, USA

article info

abstract

Article history:

This study evaluated a pilot-scale nonthermal plasma (NTP) advanced oxidation process

Received 22 May 2009

(AOP) for the degradation of trace organic compounds such as pharmaceuticals and

Received in revised form

potential endocrine disrupting compounds (EDCs). The degradation of seven indicator

3 September 2009

compounds was monitored in tertiary-treated wastewater and spiked surface water to

Accepted 10 September 2009

evaluate the effects of differing water qualities on process efficiency. The tests were also

Available online 17 September 2009

conducted in batch and single-pass modes to examine contaminant degradation rates and the remediation capabilities of the technology, respectively. Values for electrical energy per

Keywords:

order (EEO) of magnitude degradation ranged from <0.3 kWh/m3-log for easily degraded

Advanced oxidation process (AOP)

compounds (e.g., carbamazepine) in surface water to 14 kWh/m3-log for more recalcitrant

Nonthermal plasma (NTP)

compounds (e.g., meprobamate) in wastewater. Changes in the bulk organic matter based

Trace organic compound

on UV254 absorbance and excitation-emission matrices (EEM) were also monitored and correlated to contaminant degradation. These results indicate that NTP may be a viable

Pharmaceutical Endocrine

disrupting

compound

alternative to more common AOPs due to its comparable energy requirements for contaminant degradation and its ability to operate without any additional feed chemicals.

(EDC)

ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

Wastewater-derived contaminants, including pharmaceuticals and personal care products (PPCPs), endocrine disrupting compounds (EDCs), and other trace organic compounds, have been a significant aspect of environmental research efforts for decades (Snyder et al., 2003). However, adverse impacts on aquatic ecosystems and increased public awareness of trace organic compounds in water supplies have stimulated recent interest in this issue (Snyder et al., 2003). Due to rapid population growth and increasingly stressed water supplies, particularly in semi-arid regions, many cities are turning to alternative sources of drinking water such as indirect potable reuse. In some systems, wastewater effluent is having a greater impact on reservoir water quality due to drought and overdraft (Benotti et al., 2009a). Over time, these trends may

exacerbate the effects of trace organic compounds and force utilities to address the issue by augmenting their treatment trains. Although scientists are conflicted on the direct human health effects of PPCPs and EDCs at trace levels, the demonstrated effects on aquatic ecosystems have prompted studies of occurrence in wastewater, receiving water, source water, and drinking water (Ternes, 1998; Kolpin et al., 2002; Kim et al., 2007; Benotti et al., 2009c). Uncertainty in health effects coupled with the ubiquity of these compounds has necessitated subsequent studies on the efficacy of various treatment processes for their removal or transformation. Some processes (e.g., coagulation/flocculation/sedimentation, UV irradiation at disinfection doses (z40 mJ/cm2), and microfiltration/ultrafiltration) are ineffective for the treatment of trace organic compounds, whereas others (e.g., chlorine,

* Corresponding author. Tel.: þ1 (702) 856 3666; fax: þ1 (702) 856 3647. E-mail address: [email protected] (D. Gerrity). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.029

494

water research 44 (2010) 493–504

ozone, granular activated carbon (GAC), nanofiltration/reverse osmosis, and advanced oxidation) can achieve significant removal or transformation (Ternes et al., 2002; Westerhoff et al., 2005; Snyder et al., 2007). Advanced oxidation processes (AOPs) are often used to achieve further reductions in trace organics, target recalcitrant compounds, and comply with certain water reuse regulations (e.g., California Department of Public Health Title 22 requirements for NDMA and 1,4dioxane). The most common AOPs consist of UV/peroxide and ozone/peroxide, but emerging technologies such as nonthermal plasma (NTP) provide viable alternatives. This study focuses on a novel pilot-scale NTP reactor developed by Aquapure Technologies, Ltd. (Upper Galilee, Israel). The NTP uses high-voltage electrical pulses across fiber-like electrodes to form a corona discharge, which in turn generates UV light, ozone, and hydroxyl radicals (Locke et al., 2006). The light generated by the corona discharge spans wavelengths of approximately 250 nm to 1,000 nm with multiple peaks at wavelengths associated with radical formation (Locke et al., 2006). Although the light is lowintensity (Locke et al., 2006), it has been credited with a capacity for NDMA destruction (Even-Ezra et al., 2009), which is generally inefficient in ozone-based oxidation processes (Mitch et al., 2003). NTP is considered to be highly efficient because little energy is lost in heating the surrounding fluid, which allows the energy to be focused on the excitation of electrons (Pekarek, 2003). In contrast to the electron beam technology, which introduces electrons from an external source, the NTP used in this study employs a corona discharge that excites electrons in the ambient air directly above the target water matrix (Pekarek, 2003). Other NTP configurations are also available, including those that create a corona discharge in aerosols or within the target water matrix (Pekarek, 2003; Locke et al., 2006). The level of treatment can be adjusted by changing the frequency (e.g., from 500 to 1,000 Hz) or voltage (up to 40 kV) of the electrical pulses. The ionizing capability of the electrical pulses generates singlet oxygen atoms, which in turn form ozone and hydroxyl radicals. One of the most significant benefits of NTP is the fact that oxidants can be generated without the addition of costly chemicals or UV lamps, which require cleaning and are hindered by high turbidity and matrix absorbance. However, the NTP technology has limited commercial availability and has not been tested in full-scale drinking water and wastewater treatment applications so its long-term reliability, efficiency, and effectiveness are still unknown. In addition, the NTP technology is currently limited to low-flow applications due to its thin-film configuration. Although plasma-based technologies have been studied extensively for the degradation of volatile organic compounds (VOCs) (Masuda et al., 1995; Hwang and Jo, 2005; Ighigeanu et al., 2005; Thevenet et al., 2007; Kim et al., 2008), many of these studies have focused on air quality and/or bench-scale configurations. Due to limitations associated with conventional AOPsdprimarily the need for chemical additiond studies of novel AOPs for water and wastewater treatment applications are gaining popularity. A recent pilot-scale study utilizing the Aquapure technology evaluated the efficacy of NTP in degrading trichloroethylene (TCE), N-nitrosodimethylamine (NDMA), 1,4-dioxane, and methyl tert-butyl

ether (MTBE) in batch experiments and single-pass remediation (Even-Ezra et al., 2009). The current study uses the same pilot-scale system to evaluate the degradation of PPCPs and EDCs in water and wastewater. Despite the lack of PPCP/EDC data for plasma-based technologies, other AOPs, particularly UV/peroxide (Huber et al., 2003; Pereira et al., 2007; Snyder et al., 2007; Benotti et al., 2009b), ozone/peroxide (Acero and von Gunten, 2001; Snyder et al., 2007), and photocatalysis (Benotti et al., 2009b; Westerhoff et al., 2009), have been tested extensively with respect to trace organic degradation. In their review of NTP applications, Locke et al. (2006) discussed the need for comparisons of NTP with more common AOPs to determine whether there is a net benefit associated with employing NTP technologies. Among other research needs, Locke et al. (2006) specifically identified energy consumption and applicability to different waste matrices as useful points of comparison. To this end, the objective of this study is to expand the NTP knowledge base by evaluating its efficacy and energy efficiency in degrading seven PPCPs and EDCs in tertiary-treated wastewater and spiked surface water.

2.

Experimental section

2.1.

Selection of wastewater-derived contaminants

In order to narrow the scope of this research, a subset of the numerous compounds detected in previous occurrence studies was selected for evaluation. The indicator compounds were selected based on their magnitude and frequency of occurrence in water and wastewater (Snyder et al., 2007), varying physical/chemical characteristics and resulting susceptibility to treatment (Ternes et al., 2002; Westerhoff et al., 2005; Snyder et al., 2007), and ease of analytical methods (Trenholm et al., 2009). The seven compounds targeted in this study include meprobamate (anti-anxiety), dilantin (also known as phenytoin; anticonvulsant), primidone (anticonvulsant), carbamazepine (anticonvulsant), atenolol (betablocker), trimethoprim (antibiotic), and atrazine (herbicide and suspected EDC). Their structures and characteristics are described in Snyder et al. (2007). Although these compounds have generated considerable interest in the research, treatment, and regulatory arenas, only atrazine is currently regulated by the United States (U.S.) Environmental Protection Agency (EPA) at a maximum contaminant level (MCL) of 3 mg/L.

2.2.

Nonthermal plasma pilot unit

The pilot-scale unit depicted in Fig. 1 is an ‘‘electrode-to-plate’’ NTP prototype (Locke et al., 2006) developed by Aquapure Technologies, Ltd.; the pilot skid contains two reactors connected in series. Water flows in a thin film (z5 mm) along the stainless steel ground electrode (anode) where it is exposed to high-voltage electrical pulses. The electrical pulses from the generator have frequencies ranging from 500 to 1,000 Hz, maximum voltage of 8.0 kV, maximum current of 100 A, maximum energy of 1 J, and rise time of approximately 18 ns (Locke et al., 2006; Even-Ezra et al., 2009). Depending on the applied voltage and frequency, each reactor draws between 0.4 and 1.0 kW, and the external pumping and controls draw

water research 44 (2010) 493–504

495

Fig. 1 – Schematic of NTP pilot unit. The bottom figure illustrates a profile view of the electrodes within the reactor.

approximately 2.8 kW. The electrical pulses generate a corona discharge that stretches approximately 1 cm from the carbon fiber-like cathode to the water surface. The distance between the carbon electrode and water surface can be adjusted at multiple points along the reactor to level the system and optimize the process for a particular water matrix. The 1-cm distance was selected to achieve a consistent corona discharge and minimize sparking within the reactor, which would be destructive to the carbon fibers. Despite the harsh conditions experienced by the gas-phase electrode, no visible wear was detected on the carbon fibers over the duration of the experiments. With continuous, long-term operation of the NTP reactor, it is likely that the carbon fibers would eventually require replacement, but this life span is currently unknown. After traveling through the first reactor, the water is collected in a storage tank before being pumped into the ozone injector system or the second reactor. The ozone injector system combines a portion of the treated water from the first reactor with ozone-rich air (z2 g/m3) from the reactor headspace using a Venturi inductor. For the single-pass configuration, this mixture is collected in another storage tank for additional ozone contact time before being mixed and discharged with effluent from the first reactor. Excess air from the ozone contactor is then drawn out of the system by vacuum and quenched in an ozone destruct unit. The hydraulic residence times in the reactor, storage tank, and ozone contactor are approximately 0.5 min, 4–7.5 min, and 1.5 min, respectively. The hydraulic residence times in the reactor and storage tank depend on the process flow rate, whereas the residence time in the ozone contactor remains relatively constant because the injector flow rate (z40 L/min) is controlled by

a separate pump. The process is then repeated in the second reactor before being pumped out of the pilot skid. For the batch configuration, the effluent pump is replaced by a recirculation pump that allows the water to cycle continuously through a single reactor and ozone contactor. Additional details related to the NTP reactor used in this study are provided in Even-Ezra et al. (2009).

2.3.

Experimental design

The study was divided into three experiments: (1) a batch experiment with 150 L of tertiary-treated wastewater (i.e., grit removal, primary settling, activated sludge, secondary settling, and dual media filtration) at a recirculation rate of 8.0 L/min, (2) a single-pass experiment with tertiary-treated wastewater at a flow rate of 15.5 L/min, and (3) a single-pass experiment with spiked surface water at a flow rate of 11.4 L/min. For experiment (3), spiked surface water (Colorado River water from Lake Mead, NV) was pumped into the reactor from a 3,000-gallon polyethylene water tank (American Tank Company, Windsor, CA). The spiking solution was prepared by dissolving neat standards in water in order to avoid the introduction of radical scavenging solvents such as methanol. The spiking solution was added to the 3,000-gallon tank followed by recirculation for 24 h to facilitate homogenization. The two water types (wastewater and surface water) were selected to provide an assessment of NTP treatment efficiency for matrices with different organic loadings. The initial pH, total alkalinity, total organic carbon (TOC), UV254 absorbance, and trace organic compound concentrations for the three experiments are provided in Table 1. In all tables, ‘‘Batch’’ refers to the batch

496

water research 44 (2010) 493–504

4.8 kWh/m3 for the overall pilot skid. Samples were collected from the effluent line exiting the pilot skid. Residual oxidants in each sample were quenched with calcium thiosulfate. During the wastewater experiments, the dissolved ozone concentration in the NTP reactor effluent ranged from approximately 0.05 mg/L to 0.10 mg/L depending on the generator settings. The single-pass surface water experiment (3) incorporated nine different generator scenarios ranging from 0.6 to 2.6 kWh/m3 for the generators and 4.8 to 6.8 kWh/ m3 for the pilot skid. For the surface water, samples were taken from Reactor 1, Ozone Tank 1, Reactor 2, and the pilot skid effluent to evaluate the extent of contaminant degradation in each system component. During the surface water experiment, the dissolved ozone concentrations in the reactor effluent ranged from 0.25 to 0.50 mg/L depending on the generator settings.

Table 1 – Initial water quality conditions. Parameter pH Total Alkalinity TOC UV254 Abs. Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazined

Batch (1)

WW (2)a,c

SSW (3)b,c

7.0 126 mg/L as CaCO3 5.9 mg-C/L 0.119 cm1 275 ng/L 119 ng/L 124 ng/L 176 ng/L 378 ng/L 36 ng/L N/A

7.0 126 mg/L as CaCO3 5.6 mg-C/L N/A 256 25e ng/L 165 10 ng/L 146 6 ng/L 219 9 ng/L 413 8 ng/L 39 2 ng/L N/A

8.0 137 mg/L as CaCO3 2.6 mg-C/L 0.036 cm1 933 15 ng/L 1,005 74 ng/L 1,375 96 ng/L 1,600 115 ng/L 1,018 57 ng/L 1,200 82 ng/L 1,118 159 ng/L

a WW (2) ¼ Single-pass wastewater experiment. b SSW (3) ¼ Single-pass spiked surface water experiment. c Four influent samples were taken throughout the experiment to calculate a representative ambient concentration and evaluate temporal variability in target compound concentrations. d Atrazine was spiked in the SSW experiment but was not detected in the wastewater. e Errors represent 1 standard deviation.

2.4.

Analytical methods

Ozone gas concentrations in the reactor headspace were determined by online measurements with an OMC20 ozone monitor and IN-2000-L2-LC ozone analyzer (IN USA, Norwood, MA). Dissolved ozone residuals were measured with the indigo trisulfonate method (Yates and Stenstrom, 2000). TOC and UV254 absorbance were analyzed according to standard methods 5310B and 5910, respectively. Excitation-emission matrices (EEM) were also developed for the single-pass wastewater samples using a QuantaMaster UV-Vis QM4 Steady State Spectrofluorometer (Photon Technology International, Inc., Birmingham, NJ). Trace organic compounds were extracted and analyzed using on-line solid phase extraction and liquid chromatography with tandem mass spectrometry (SPE-LC-MS/MS) according to Trenholm et al. (2009). All standards and reagents were of the highest purity commercially available. All

wastewater experiment, ‘‘WW’’ refers to the single-pass wastewater experiment, and ‘‘SSW’’ refers to the single-pass spiked surface water experiment. The generator settings for the NTP pilot differed for each of the experiments, as described in Table 2. For the batch experiment (1), a single reactor/generator was operated at a frequency of 500 Hz and a voltage of 8.0 kV. To assess the degradation rate, ten samples were collected at generator (AOP-specific) energy consumption values ranging from 0 to 7.3 kWh/m3. For the single-pass wastewater experiment (2), four different generator scenarios provided treatment levels ranging from 0.7 to 1.8 kWh/m3 for the generators and 3.8 to

Table 2 – Experimental test conditions. Experiment

Generator 1 Frequency (Hz)

Voltage (kV)

Generator 2 Energy (kWh/m3)

Frequency (Hz)

Voltage (kV)

Energy (kWh/m3)

Total generator energy

Total skid energy

(kWh/m3)

(kWh/m3)

(1)

Batcha

Off

Off

0.00

500

8.0

N/A

N/A

N/A

(2)

WWa – 1 WW – 2 WW – 3 WW – 4

Off 800 600 800

Off 8.0 8.0 8.0

0.00 1.04 0.87 1.04

500 Off 500 500

8.0 Off 8.0 8.0

0.71 0.00 0.71 0.71

0.71 1.04 1.58 1.75

3.75 4.08 4.62 4.79

(3)

SSWb – 1 SSW – 2 SSW – 3 SSW – 4 SSW – 5 SSW – 6 SSW – 7 SSW – 8 SSW – 9

Off Off 500 Off 1000 500 500 1000 1000

Off Off 8.0 Off 8.0 8.0 8.0 8.0 8.0

0.00 0.00 0.97 0.00 1.53 0.97 0.97 1.53 1.53

600 800 Off 1000 Off 600 1000 600 1000

5.9 5.9 Off 5.9 Off 5.9 5.9 5.9 5.9

0.63 0.85 0.00 1.07 0.00 0.63 1.07 0.63 1.07

0.63 0.85 0.97 1.07 1.53 1.60 2.04 2.16 2.60

4.78 5.01 5.12 5.23 5.68 5.75 6.19 6.31 6.75

a The batch and WWTP experiments were conducted with ambient concentrations of trace organics in tertiary-treated wastewater. b The SSW experiments were conducted with raw surface water (Lake Mead) spiked with a suite of trace organics.

497

water research 44 (2010) 493–504

3.

1.0

0.8

0.6

C/C0

pharmaceuticals were obtained from Sigma-Aldrich (St. Louis, MO, USA). Meprobamate-d3 and trimethoprim-d9 were obtained from Toronto Research Chemicals (Ontario, Canada). Phenytoin-d10 and atenolol-d7 were obtained from C/D/N Isotopes (Pointe-Claire, Canada). Carbamazepine-d10 and primidone-d5 were obtained from Cambridge Isotope Laboratories (Andover, MA, USA). All solvents were trace analysis grade from Burdick and Jackson (Muskegon, MI). Reagent water was obtained using a Milli-Q Ultrapure Water Purification System (Millipore, Bedford, MA, USA). All concentrated stocks were prepared in methanol and stored at 20 C, while mixed spiking solutions were prepared in reagent water and stored at 4 C. On-line SPE was performed using a Symbiosis Pharma (Spark Holland, Emmen, the Netherlands) automated SPE in XLC mode. Briefly, a 10-mL volume of sample was measured in a volumetric flask at which time isotopically-labeled standards were spiked at 100 ng/L. This provided sufficient sample volume for duplicates, matrix spikes, and dilutions, if necessary. A 1.5-mL aliquot of each sample was then transferred to a 2-mL autosampler vial, although only 1.0 mL was used for extractions. Extractions were performed using Waters Oasis HLB Prospekt cartridges (30 mm, 2.5 mg, 10 1 mm, 96 tray) (Milford, MA). Prior to sample loading, each cartridge was sequentially conditioned with 1 mL of dichloromethane (DCM), MTBE, methanol, and reagent water (Milli-Q). Samples were loaded onto the SPE cartridge at 1 mL/min after which the cartridge was washed with 1 mL of reagent water. After sample loading, the analytes were eluted from the SPE cartridge to the LC column with 200 mL methanol, using the LC peak focusing mode. A 5-mM ammonium acetate solution and methanol gradient was used for LC mobile phases with a flow rate of 800 mL/min. Analytes were separated using a 150 4.6 mm Luna C18(2) with a 5-mm particle size (Phenomenex, Torrance, CA). Method reporting limits (MRLs) were chosen at 3 to 5 times the method detection limit (MDL). MRLs were 10 ng/L for all compounds, except for atenolol, which was 25 ng/L. All analytes were quantified using isotope dilution. Stringent QA/QC protocols (i.e., matrix spikes, duplicate samples, field blanks, and laboratory blanks) were followed throughout the duration of the experiment. Using the on-line SPE method, the concentrations of duplicate samples rarely varied by greater than 5%.

0.4

0.2

0.0 0

1

2

3

4

5

6

7

8

Generator Energy Consumption (kWh/m3) Meprobamate

Dilantin

Primidone

Carbamazepine

Atenolol

Trimethoprim

Fig. 2 – Contaminant degradation profiles from the batch experiment (1) with ambient concentrations of trace organic compounds in tertiary-treated wastewater. This graph illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption. The dashed lines represent samples that were below the MRL for those compounds.

The degradation of each detected contaminant was assumed to be pseudo first order based on a plot of ln(C/C0) versus generator energy consumption. Degradation rates, 95% confidence intervals, and R2 values are provided in Table 3. Due to the low initial trimethoprim concentration, the extent of degradation was limited, as indicated by the dashed line representing the MRL in Fig. 2. Despite the limited degradation, it is apparent that trimethoprim and carbamazepine are highly susceptible to NTP treatment. This result is consistent with previous studies assessing oxidation strategies for trace organic compounds. Westerhoff et al. (2005) reported nearly 100% degradation of trimethoprim and carbamazepine with standard chlorination and ozonation practices. Dilantin and atenolol degraded more slowly during the NTP treatment, which is also consistent with oxidation by chlorine or ozone (Westerhoff et al., 2005). Finally, meprobamate has been shown to be highly resistant to chlorination and ozonation in comparison to other trace organic compounds (Westerhoff et al., 2005). This conclusion is supported by the NTP results in that meprobamate required nearly double the energy to

Results and discussion

3.1. Experiment 1 – batch experiment with tertiarytreated wastewater Contaminant degradation profiles during the batch wastewater experiment are provided in Fig. 2. The initial ambient concentrations for the target contaminants ranged from 36 ng/L for trimethoprim to as high as 378 ng/L for atenolol (Table 1). Although included in the analytical method, atrazine was below the method reporting limit (MRL) for both wastewater experiments. As indicated in Table 2, the generator was operated at a frequency of 500 Hz, a voltage of 8.0 kV, and a recirculation rate of 8.0 L/min.

Table 3 – Pseudo first order rate constants for batch experiment (1) in tertiary-treated wastewater. Compound Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazinea

k (m3/kWh)

95% CIb

R2

N

0.36 0.52 0.40 1.03 0.61 0.71 N/A

0.03 0.15 0.15 0.35 0.22 N/A N/A

0.98 0.92 0.88 0.93 0.88 0.92 N/A

9 5 5 4 5 2 N/A

a Atrazine concentrations were below MRL in all samples. b 95% CI ¼ 95% confidence interval.

498

water research 44 (2010) 493–504

achieve a specified level of degradation. Primidone was not quite as resistant as meprobamate, but its degradation was slower and more variable than the other compounds. With respect to bulk organic parameters, there was no significant change in TOC over the duration of the experiment. The consistent TOC level was expected considering that mineralization is highly energy intensive in hydroxyl radicaldominated processes (Gerrity et al., 2009), and high CT values are required in ozone-dominated processes (Rosal et al., 2009). The limited energy consumption, low dissolved ozone concentrations, and limited ozone contact time were likely insufficient to induce measurable organic mineralization in the NTP reactor. Although there was no reduction in TOC, UV254 absorbance was reduced by more than 65%; this change was correlated to the degradation of target compounds in Fig. 3 as a potential surrogate measure of process efficacy. Quantifying trace organic compounds requires specialized equipment and training, and the analytical methods are very expensive and time-consuming. For this reason, it is generally impractical for water and wastewater utilities to employ trace contaminant monitoring programs for their systems. Implementation of this type of surrogate framework (i.e., reduction in UV254 absorbance in lieu of trace contaminant degradation) would allow utilities to assess the performance of certain treatment processes for emerging contaminants. For oxidation processes, UV254 absorbance is an easily measured parameter that appears to show consistent correlations with the degradation of trace organic compounds (Wert et al., 2009). Fig. 3 provides support for this surrogate framework because it demonstrates strong correlations between the degradation of the target compounds and reduction in UV254 absorbance. The slopes of the regression lines indicate that relative contaminant degradation (i.e., percent degradation) is actually more rapiddmore than twice as fast for nearly all compoundsdthan the reduction in UV254 absorbance. The slopes also show a general agreement with the degradation profiles

in Fig. 2. Although trimethoprim was expected to degrade rapidly, its slope of 6.25 is likely an outlier and artifact of its small sample size. With the exception of trimethoprim, the magnitude of the slopes and the negative intercepts are similar to those reported in the Wert et al. (2009) ozonation study. The negative intercepts indicate that there are initial reductions in UV254 absorbance that precede degradation of the target compounds. This may be attributable to initial oxidant/radical scavenging by the bulk organic matter. Values for electrical energy per order (EEO) of magnitude degradation (Bolton et al., 1996), which is often measured in kWh/m3-log, are also provided in Table 4. Since the batch experiment involved a continuous treatment, the EEO values were calculated based on first order degradation rates and linear regression. In other words, a linear regression was generated for ln(C/C0) against generator energy consumption, and the EEO was calculated as ln(0.1) divided by the slope of the linear regression. The EEO values ranged from 2.2 kWh/m3 for carbamazepine to 6.4 kWh/m3 for meprobamate. It is important to note that the batch EEO values may be overestimates because the batch system requires approximately 10–15 min to achieve its maximum ozone concentration after starting the generator.

3.2. Experiment 2 – single-pass experiment with tertiary-treated wastewater Contaminant degradation profiles for the single-pass wastewater experiment are illustrated in Fig. 4. The initial ambient concentrations for the target contaminants were similar to the batch experiment and ranged from 39 ng/L for trimethoprim to as high as 413 ng/L for atenolol (Table 1). As indicated in Table 2, this experiment involved a single-reactor configuration and both reactors operating in series at a flow rate of 15.5 L/min. All scenarios for this experiment involved voltages of 8.0 kV. The reactor consumed 0.7 kWh/m3 with Generator 2 operating at a frequency of 500 Hz during the

% Degradation of Trace Organic Compounds

100%

80%

Linear Regression Models: 60%

Contaminant

40%

20%

Slope Intercept R2

N

Meprobamate

1.56

-14%

0.97

8

Dilantin

2.09

-22%

0.99

4

Primidone

2.08

-28%

0.98

4

Carbamazepine 2.75

-30%

1.00

3

Atenolol

2.68

-44%

0.99

4

Trimethoprim

6.25

-90%

1.00

2

0% 0%

20%

40%

60%

80%

100%

% Reduction in UV254 Absorbance

Fig. 3 – Correlation between the degradation of target compounds and UV254 absorbance during the batch experiment (1). This analysis excludes samples that were below the MRL and the final five carbamazepine samples (>3 kWh/m3; see Fig. 2) due to high variability near the MRL. The slopes were all significant at the 0.03 level or better, and intercepts were all significant at the 0.07 level or better.

499

water research 44 (2010) 493–504

Table 4 – AOP-specific EEO values (kWh/m3-log). Contaminant

(1)

(2)

(3)

Batcha

Min. WWb

Med. WWb

Max. WWb

Min. SSWb

Med. SSWb

Max. SSWb

6.4 4.4 5.8 2.2 3.8 3.2 N/A

4.6 2.7 2.8 1.3 2.5 1.5 N/A

10 3.1 4.3 1.8 2.9 2.2 N/A

14 3.5 4.8 2.1 3.3 3.0 N/A

2.1 1.1 1.1 <0.3 0.4 <0.3 2.2

3.5 2.0 2.2 <0.7 1.0 <0.7 3.7

5.3 3.1 3.3 <1.2 1.7 <1.2 6.3

Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazine

UVc SSW

UV/H2O2c,d SSW

UV/TiO2c,e SSW

6.6 2.1 3.7 2.3 1.4 0.8 3.3

1.0 1.0 0.6 0.4 0.5 0.4 1.2

6.8 2.2 3.9 2.1 2.0 1.5 4.7

a Batch EEO values based on linear regression assuming first order kinetics. This analysis excludes samples that were below the MRL and the final five carbamazepine samples (>3 kWh/m3; see Fig. 2) due to high variability near the MRL. b WW and SSW were comprised of discrete sampling events so regression was not appropriate. c Adapted from Benotti et al. (2009b). The water matrix was identical to that of the SSW samples from this study (i.e., Lake Mead). d UV/H2O2 with 10 mg/L of H2O2. e UV/TiO2 (photocatalysis) with approximately 500 mg/L of titanium dioxide (TiO2).

WW-1 scenario, and the reactor consumed 1.0 kWh/m3 with Generator 1 operating at a frequency of 800 Hz during the WW-2 scenario. WW-3 and WW-4 involved both reactors operating in series at 600/500 Hz (1.6 kWh/m3) and 800/500 Hz (1.8 kWh/m3), respectively. The energy consumption values for the entire skid, which include energy related to the generators, pumps, and controls, are also provided for each scenario. The relative degradation rates were identical to those of the batch experiment. For example, carbamazepine and trimethoprim experienced the most rapid degradation, and primidone and meprobamate were the most recalcitrant compounds. At the highest level of energy consumption (1.8 kWh/m3), every compound except meprobamate experienced at least 70% degradation. The evaluation of

1.0

0.8

C/C0

0.6

0.4

0.2

0.0 0.0

0.5

1.0

1.5

2.0

Generator Energy Consumption (kWh/m3) Meprobamate

Dilantin

Primidone

Carbamazepine

Atenolol

Trimethoprim

Fig. 4 – Contaminant degradation profiles from the singlepass experiment (2) with tertiary-treated wastewater. The initial contaminant concentrations represent ambient conditions in the wastewater. This graph illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption. The dashed line represents samples that were below the MRL for trimethoprim.

trimethoprim was hindered by the low initial ambient concentration in the wastewater. EEO values for the recalcitrant compounds (meprobamate and primidone) decreased by 64% and 42%, respectively, as the energy consumption increased from 0.7 to 1.8 kWh/m3. EEO values for dilantin and atenolol decreased by approximately 20% over the same range of energy consumption. However, the EEO values for the most susceptible compounds (carbamazepine and trimethoprim) actually increased by 38% and 100%, respectively, due to their rapid degradation at low levels of energy consumption. Most compounds were characterized by EEO values ranging from 1 to 5 kWh/m3-log, but meprobamate required 13–14 kWh/m3-log with one reactor in operation. In general, the single-pass wastewater experiment was more efficient than the batch experiment with respect to EEO values due to the presence of ozone at the beginning of each discrete scenario, which provides a more appropriate representation of actual treatment conditions. As with the batch experiment, TOC concentrations did not change during the various treatment scenarios, but the transformation of bulk organic matter was evaluated with excitation-emission matrices (Fig. 5). Based on the regions described in Chen et al. (2003), a portion of the organic matter was characterized as soluble microbial byproduct-like (region IV), though a much larger portion of the organic matter was characterized as humic acid-like (region V). It is apparent that the NTP treatment quickly transformed the natural organic matter in both of these regions as observed by the dramatic decrease in fluorescence intensity. Although the organic matter was transformed into compounds with less ability to fluoresce, the constant TOC values indicate that the treatment was insufficient to induce organic mineralization. This highlights the importance of transformation products in AOP applications. Although the target compounds are degraded, the lack of mineralization indicates the formation of byproducts related to the bulk organic matter and possibly the trace organic compounds, which may or may not be harmful to humans and/or the environment. Due to the complexity of the target water matrices, identifying byproducts associated with the target compounds was beyond the

500

water research 44 (2010) 493–504

Fig. 5 – Excitation-emission matrices as a function of generator energy consumption for the single-pass experiment (2) with tertiary-treated wastewater. The regions are described as follows: (I) Aromatic protein, (II) aromatic protein II, (III) fulvic acid-like, (IV) soluble microbial byproduct-like, and (V) humic acid-like (Chen et al., 2003).

scope of this study, but future studies should address this issue in greater detail.

3.3. Experiment 3 – single-pass experiment with spiked surface water Contaminant degradation profiles, including that of atrazine, for the single-pass spiked surface water experiment are illustrated in Fig. 6. The initial spiked concentrations were much higher than the ambient wastewater concentrations as all of the compounds were spiked at greater than 900 ng/L (Table 1). Similar to the single-pass wastewater experiment, this experiment involved a single-reactor configuration and both reactors operating in series at a flow rate of 11.4 L/min (Table 2). For this experiment, Generator 1 was operated at a voltage of 8.0 kV, and Generator 2 was operated at a voltage of 5.9 kV. The generator and overall skid energy consumption values ranged from 0.6 to 2.6 kWh/m3 and 4.8 to 6.8 kWh/m3, respectively. The relative degradation rates were similar to those of the wastewater experiments. For example, meprobamate was the most difficult to degrade, while trimethoprim and carbamazepine rapidly dropped below the MRL in all treated samples. The lower organic loading of the surface water (i.e., lower initial TOC concentration and UV254 absorbance), which

resulted in a higher residual ozone concentration (0.25 to 0.50 mg/L), likely contributed to the improved degradation. In contrast to the wastewater experiments, atenolol degradation was more similar to trimethoprim and carbamazepine, while primidone degradation was more similar to that of dilantin. As expected, the degradation of atrazine, which is considered a recalcitrant compound (Westerhoff et al., 2005), was almost identical to that of meprobamate. For all of the compounds with concentrations above the MRL, the degradation peaked at about 70% for the most recalcitrant compounds and 85% for the moderately recalcitrant compounds after reaching an energy consumption of 1.6 kWh/m3. At the highest energy level, the effluent concentrations of the residual compounds ranged from 140 ng/L for dilantin to 390 ng/L for atrazine. During Experiment 3, there was also greater fluctuation for several time points. Since the samples represent discrete, rather than continuous (i.e., batch), treatment levels, one would anticipate greater variability, but there was still a relatively consistent decrease in contaminant concentration over the range of energy values tested. The EEO values for all of the compounds were dramatically lower in the surface water experiment than the wastewater experiments, presumably due to the lower scavenging capacity of the surface water. Meprobamate and atrazine had the highest EEOs with maximum values of 5.3 and 6.3 kWh/

501

water research 44 (2010) 493–504

1

0.8

C/C0

0.6

0.4

0.2

0 0.0

0.5 Meprobamate

1.0 1.5 2.0 Generator Energy Consumption (kWh/m3) Dilantin

Primidone

Carbamazepine

Atenolol

2.5 Trimethoprim

3.0 Atrazine

Fig. 6 – Contaminant degradation profiles from the single-pass experiment (3) with spiked surface water. This graph illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption. m3-log, respectively. The maximum values generally occurred during the experiments when one or both of the generators were operating at 1,000 Hz. Although operating both generators at 1,000 Hz achieved the highest level of degradation, increasing the pulse frequency generally yielded diminishing rates of return with respect to efficiency. Operating Generator 2 at a frequency of 600 Hz provided the most efficient treatment conditions for all of the target compounds. However, Generator 1 had been operated immediately prior to that scenario, so the residual ozone from Reactor 1 likely contributed to this increased efficiency. Although this overestimates the anticipated degree of degradation, it indicates that it may be beneficial to incorporate ozone contact tanks before and after each reactor to maximize degradation. Based on this

observation, the median values are likely the most appropriate EEO estimates since they always corresponded to both reactors operating simultaneously. For the compounds with reportable concentrations, the median EEO values ranged from 1.0 to 3.7 kWh/m3 for atenolol and atrazine, respectively. A subset of samples was also analyzed to assess the contribution of the system components (i.e., reactor vs. ozone contactor) to contaminant degradation. The results from one set of samples (SSW-8) are provided in Fig. 7; the other sample sets yielded similar trends and therefore are not shown. Based on these results, it appears that the ozone contactors provided the majority of the degradation for the more recalcitrant compounds, including meprobamate, dilantin, primidone,

1600

1400

Concentration (ng/L)

1200

1000

800

600

400

200

0 Influent

Reactor 1

Ozone 1

Reactor 2

Effluent

Sample Location Meprobamate

Dilantin

Primidone

Carbamazepine

Atenolol

Trimethoprim

Atrazine

Fig. 7 – Comparison of the pilot system components in degrading the target compounds during the single-pass experiment (3) with spiked surface water. This data refers to samples from SSW-8 with generator and skid energy consumption values of 2.2 and 6.3 kWh/m3, respectively. The small increase in concentrations from Ozone 1 to Reactor 2 is explained by mixing of Reactor 1 and Ozone 1 effluent prior to passage into Reactor 2 (see Fig. 1).

502

% Degradation of Trace Organic Compounds

water research 44 (2010) 493–504

100%

Linear Regression Models: 80%

60%

40%

Contaminant

Slope Intercept R2

N

Meprobamate

2.40

0

0.72

9

Dilantin

1.35

45%

0.84

9

Primidone

1.52

40%

0.81

9

Atrazine

2.27

0

0.74

9

20%

0% 0%

5%

10%

15%

20%

25%

30%

35%

% Reduction in UV254 Absorbance Meprobamate

Dilantin

Primidone

Atrazine

Fig. 8 – Correlation between the degradation of detected target compounds and UV254 absorbance during the single-pass experiment (3) with spiked surface water. The slopes were all significant at the 0.001 level, the intercepts for primidone and dilantin were significant at the 0.001 level, and the intercepts for meprobamate and atrazine were considered insignificant ( p > 0.09; regressions for meprobamate and atrazine were forced through the origin).

atenolol, and atrazine. However, carbamazepine and trimethoprim experienced rapid degradation in the reactor itself with the additional oxidation pathways (i.e., UV light, hydroxyl radicals, and ozone). The spiked surface water experiment was also evaluated based on general water quality parameters. For example, the initial bromide concentration in the surface water was approximately 62 mg/L, and additional analyses indicated that bromate increased only slightly (maximum of 2.4 mg/L) above the ambient surface water concentration of 1.7 mg/L. Similar to the wastewater experiments, there was no significant change in TOC over the duration of the experiment, but the reduction in UV254 absorbance was correlated to the degradation of the target contaminants in Fig. 8. In a manner similar to Fig. 3, Fig. 8 illustrates strong positive correlations between contaminant degradation and reduction in UV254 absorbance. Furthermore, the degradation of target contaminants was still more rapid than the reduction in UV254 absorbance. In contrast to the batch wastewater experiment and data provided in Wert et al. (2009), the regression lines for dilantin and primidone have positive intercepts, while the intercepts for meprobamate and atrazine were found to be insignificant ( p > 0.09). Wert et al. (2009) proposed that the correlations between contaminant degradation and UV254 absorbance were independent of wastewater quality, but it appears that the correlations differ when the matrices themselves are drastically different (i.e., wastewater vs. surface water). Specifically, the positive intercepts for dilantin and primidone in the surface water experiment indicate that these target compounds started to degrade (z40–45%) prior to any reduction in UV254 absorbance. In addition to contributing to their smaller slopes, this may indicate that there was less oxidant/radical scavenging and that the bulk organic matter was more recalcitrant in the surface water compared to the wastewater. It is also interesting to note that meprobamate has the smallest absolute intercept for the wastewater experiment and an insignificant

intercept for the surface water experiment, and atrazine also has an insignificant intercept for the surface water experiment. Therefore, the initial degradation of these highly resistant compounds is more similar to the initial degradation of the bulk organic matter. For the surface water experiments, this supports the theory that the aromatic organic matter in the surface water was more recalcitrant than that of the wastewater. Based on these observations, it appears that the UV254 surrogate framework is still valid, but there may be differences in the correlations for dissimilar water matrices. Table 4 provides EEO values for the three experiments in this study, and it also provides EEO values from Benotti et al. (2009b), which evaluated the efficacy of photolysis, UV/H2O2, and photocatalysis in degrading PPCPs and EDCs. Benotti et al. (2009b) provides an interesting point of comparison since the water matrix used for that study (spiked surface water from Lake Mead), target compounds, and analytical methods were identical. As indicated by the data, NTP was generally more efficient than photolysis and photocatalysis for each of the compounds, but was either comparable or slightly less efficient than UV/H2O2 at a peroxide dose of 10 mg/L. It is important to note that the NTP technology does not require UV lamps or costly peroxide feeds to generate the oxidative species. Therefore, NTP may be a viable alternative to the more common UV/H2O2 process despite the slightly higher energy requirements for some compounds. Although the NTP achieved comparable efficiencies to those of UV/H2O2 in these experiments, identification of the best alternative for a particular application also depends on capital costs and reliability. Since the NTP technology has not been fully commercialized, it is difficult to compare its capital costs and reliability with more established technologies. Future evaluations of large-scale NTP reactors, including long-term studies, will provide valuable information in determining how NTP compares to more established technologies on a life-cycle basis.

water research 44 (2010) 493–504

4.

Conclusions

With respect to treatment efficacy, the pilot-scale NTP system demonstrated rapid degradation of the target compounds. The relative degradation rates were consistent between the three phases of the study with carbamazepine and trimethoprim as the most susceptible compounds and meprobamate, primidone, and atrazine as the most recalcitrant compounds. Atenolol and dilantin were generally more resistant to oxidation than carbamazepine and trimethoprim. These similar relative degradation rates justify the use of indicator compounds in evaluations of PPCP and EDC treatment. The EEO values for the optimal scenarios were generally less than 5 kWh/m3-log for all of the target compounds, particularly in the spiked surface water with a lower organic loading. With process optimization (i.e., optimization of electrode height for each water matrix) and limited modifications to the system, including more effective use of ambient ozone, the EEO values could likely be reduced further. More common AOPs, including UV/H2O2 and ozone/H2O2, require significant chemical addition and residual H2O2 quenching, which represent a significant portion of their operational costs. The primary benefit of the NTP AOP is the ability to generate UV light, ozone, and hydroxyl radicals without chemical addition or the use of UV lamps. However, the true capital costs and reliability of large-scale NTP reactors for water and wastewater treatment are still relatively unclear. Also, since oxidation by molecular ozone may be the dominant mechanism for many compounds, future studies should provide direct comparisons of NTP with conventional ozone generators to compare system performance. These expanded studies are necessary before one can fully determine the applicability of the novel NTP technology for water and wastewater treatment. However, these preliminary results indicate that NTP may be a viable alternative to more common AOPs due to its chemical-free operation and comparable energy requirements for trace contaminant degradation.

Acknowledgments We would like to thank the SNWA Applied Research and Development Center staff, particularly Janie Zeigler, Shannon Ferguson, Christy Meza, and Elaine Go, for assisting with the various analyses. We would also like to thank Dvir Solnik, Itay Even-Ezra, Gal Bitan, and Shahar Nuriel of Aquapure Technologies for use of the nonthermal plasma pilot unit and for providing crucial training and guidance during the study.

references

Acero, J.L., von Gunten, U., 2001. Characterization of oxidation processes: ozonation and the AOP O3/H2O2. J. AWWA 93 (10), 90–100. Benotti, M.J., Stanford, B.D., Snyder, S.A., 2009a. Increased loading of pharmaceuticals and endocrine disrupting compounds to source drinking water due to climate change. J. Environ. Qual. Accepted.

503

Benotti, M.J., Stanford, B.D., Wert, E.C., Snyder, S.A., 2009b. Evaluation of a photocatalytic reactor membrane pilot system for the removal of pharmaceuticals and endocrine disrupting compounds from water. Water Res. 43 (6), 1513–1522. Benotti, M.J., Trenholm, R.A., Vanderford, B.J., Holaday, J.C., Stanford, B.D., Snyder, S.A., 2009c. Pharmaceuticals and endocrine disrupting compounds in U.S. drinking water. Environ. Sci. Technol. 43 (3), 597–603. Bolton, J.R., Bircher, K.G., Tumas, W., Tolman, C.A., 1996. Figuresof-merit for the technical development and application of advanced oxidation processes. J. Adv. Oxid. Technol. 1 (1), 13–17. Chen, W., Westerhoff, P., Leenheer, J.A., Booksh, K., 2003. Fluorescence excitation-emission matrix regional integration to quantify spectra for dissolved organic matter. Environ. Sci. Technol. 37 (24), 5701–5710. Even-Ezra, I., Mizrahi, A., Gerrity, D., Snyder, S., Salveson, A., Lahav, O., 2009. Application of a novel plasma-based advanced oxidation process for efficient and cost effective destruction of refractory organics in tertiary effluents and contaminated groundwater. Desalin. Water Treat. Accepted. Gerrity, D., Mayer, B., Ryu, H., Crittenden, J., Abbaszadegan, M., 2009. A comparison of pilot-scale photocatalysis and enhanced coagulation for disinfection byproduct mitigation. Water Res. 43, 1597–1610. Huber, M.M., Canonica, S., Park, G., von Gunten, U., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci. Technol. 37, 1016–1024. Hwang, Y.H., Jo, Y.M., 2005. Decomposition of odorous gases in a pilot-scale nonthermal plasma reactor. J. KOSAE 21 (2), 57–65. Ighigeanu, D., Martin, D., Zissulescu, E., Macarie, R., Oproiu, C., Cirstea, E., Iovu, H., Calinescu, I., Iacob, N., 2005. SO2 and NOx removal by electron beam and electrical discharge induced non-thermal plasmas. Vacuum 77, 493–500. Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Res. 41, 1013–1021. Kim, H., Han, J., Sakaguchi, Y., Minami, W., 2008. Simultaneous oxidation of NOx and SO2 by a new non-thermal plasma reactor enhanced by catalyst and additive. Plasma Sci. Technol. 10 (1), 53–56. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S. D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in U. S. streams, 1999–2000: A national reconnaissance. Environ. Sci. Technol. 36, 1202–1211. Locke, B.R., Sato, M., Sunka, P., Hoffmann, M.R., Chang, J.S., 2006. Electrohydraulic discharge and nonthermal plasma for water treatment. Ind. Eng. Chem. Res. 45, 882–905. Masuda, S., Hosokawa, S., Tu, X., Wang, Z., 1995. Novel plasma chemical technologiesdPPCP and SPCP for control of gaseous pollutants and air toxics. J. Electrostatics 34, 415–438. Mitch, W.A., Sharp, J.O., Trussell, R.R., Valentine, R.L., AlvarezCohen, L., Sedlak, D.L., 2003. N-nitrosodimethylamine (NDMA) as a drinking water contaminant: a review. Environ. Eng. Sci. 20 (5), 389–404. Pekarek, S., 2003. Non-thermal plasma ozone generation. Acta Polytech. 43 (6), 47–51. Pereira, V.J., Linden, K.G., Weinberg, H.S., 2007. Evaluation of UV irradiation for photolytic and oxidative degradation of pharmaceutical compounds in water. Water Res. 41 (19), 4413–4423. Rosal, R., Rodriguez, A., Perdigon-Melon, J.A., Petre, A., GarciaCalvo, E., 2009. Oxidation of dissolved organic matter in the effluent of a sewage treatment plant using ozone combined with hydrogen peroxide (O3/H2O2). Chem. Eng. J. 149, 311–318.

504

water research 44 (2010) 493–504

Snyder, S.A., Wert, E.C., Lei, H., Westerhoff, P., Yoon, Y., 2007. Removal of EDCs and Pharmaceuticals in Drinking and Reuse Treatment Processes. Water Research Foundation, Denver, CO. Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003. Pharmaceuticals, personal care products, and endocrine disruptors in water: implications for the water industry. Environ. Eng. Sci. 20 (5), 449–469. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Res. 32 (11), 3245–3260. Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F., Brauch, H., Haist-Gulde, B., Preuss, G., Wilme, U., ZuleiSeibert, N., 2002. Removal of pharmaceuticals during drinking water treatment. Environ. Sci. Technol. 36, 3855–3863. Thevenet, F., Guaitella, O., Puzenat, E., Herrmann, J.M., Rousseau, A., Guillard, C., 2007. Oxidation of acetylene by photocatalysis coupled with dielectric barrier discharge. Catal. Today 122, 186–194.

Trenholm, R.A., Vanderford, B.J., Snyder, S.A., 2009. On-line SPE LC-MS/MS analysis of pharmaceutical indicators in water: a green alternative to conventional methods. Talanta 79, 1425–1432. Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Using ultraviolet absorbance and color to assess pharmaceutical oxidation during ozonation of wastewater. Environ. Sci. Technol. 43, 4858–4863. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environ. Sci. Technol. 39, 6649–6663. Westerhoff, P., Moon, H., Minakata, D., Crittenden, J., 2009. Oxidation of organics in retentates from reverse osmosis wastewater reuse facilities. Water Res. 43 (16), 3992–3998. Yates, R.S., Stenstrom, M.K., 2000. Gravimetric sampling procedure for aqueous ozone concentrations. Water Res. 34 (4), 1413–1416.

water research 44 (2010) 505–520

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Single-walled carbon nanotubes dispersed in aqueous media via non-covalent functionalization: Effect of dispersant on the stability, cytotoxicity, and epigenetic toxicity of nanotube suspensions Alla L. Alpatova a, Wenqian Shan a, Pavel Babica b, Brad L. Upham b,c, Adam R. Rogensues a, Susan J. Masten a, Edward Drown d,1, Amar K. Mohanty d,2, Evangelyn C. Alocilja e, Volodymyr V. Tarabara a,* a

Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48824, USA Center for Integrative Toxicology, Michigan State University, East Lansing, MI 48824, USA c Department of Pediatrics and Human Development, Michigan State University, East Lansing, MI 48824, USA d School of Packaging, Michigan State University, East Lansing, MI 48824, USA e Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI 48824, USA b

article info

abstract

Article history:

As the range of applications for carbon nanotubes (CNTs) rapidly expands, understanding

Received 2 June 2009

the effect of CNTs on prokaryotic and eukaryotic cell systems has become an important

Received in revised form

research priority, especially in light of recent reports of the facile dispersion of CNTs in

14 September 2009

a variety of aqueous systems including natural water. In this study, single-walled carbon

Accepted 17 September 2009

nanotubes (SWCNTs) were dispersed in water using a range of natural (gum arabic,

Available online 30 September 2009

amylose, Suwannee River natural organic matter) and synthetic (polyvinyl pyrrolidone, Triton X-100) dispersing agents (dispersants) that attach to the CNT surface non-covalently

Keywords:

via different physiosorption mechanisms. The charge and the average effective hydrody-

Single-walled carbon nanotubes

namic diameter of suspended SWCNTs as well as the concentration of exfoliated SWCNTs

Dispersion

in the dispersion were found to remain relatively stable over a period of 4 weeks. The

Non-covalent functionalization

cytotoxicity of suspended SWCNTs was assessed as a function of dispersant type and

Cytotoxicity

exposure time (up to 48 h) using general viability bioassay with Escherichia coli and using

Epigenetic toxicity

neutral red dye uptake (NDU) bioassay with WB-F344 rat liver epithelia cells. In the E. coli viability bioassays, three types of growth media with different organic loadings and salt contents were evaluated. When the dispersant itself was non-toxic, no losses of E. coli and WB-F344 viability were observed. The cell viability was affected only by SWCNTs dispersed using Triton X-100, which was cytotoxic in SWCNT-free (control) solution. The epigenetic toxicity of dispersed CNTs was evaluated using gap junction intercellular communication (GJIC) bioassay applied to WB-F344 rat liver epithelial cells. With all SWCNT suspensions except those where SWCNTs were dispersed using Triton X-100 (wherein GJIC could not be measured because the sample was cytotoxic), no inhibition of GJIC in the presence of SWCNTs was observed. These results suggest a strong dependence of the toxicity of

* Corresponding author. Tel.: þ517 432 1755; fax: þ517 355 0250. E-mail address: [email protected] (V.V. Tarabara). 1 Present address: Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824, USA. 2 Present address: Department of Plant Agriculture, University of Guelph, 50 Stone Rd. E., Guelph, Ontario, N1 G 2W1 Canada. 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.042

506

water research 44 (2010) 505–520

SWCNT suspensions on the toxicity of the dispersant and point to the potential of noncovalent functionalization with non-toxic dispersants as a method for the preparation of stable aqueous suspensions of biocompatible CNTs. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

The discovery (Iijima, 1991) and subsequent extensive characterization of carbon nanotubes (CNTs) have revealed a class of materials with extraordinary electrical, mechanical, and thermal properties (Tasis et al., 2006). The wider application of CNTs in electronic, optical, sensing, and biomedical fields has been impeded by the low solubility of as-produced CNTs in polar liquids and by the strong tendency of CNTs to aggregate due to hydrophobic–hydrophobic interactions (Lin et al., 2004). Dispersing CNTs and ensuring long-term stability of CNTs suspended in a liquid medium have proved especially challenging for aqueous systems.

2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003; McDonald et al., 2006; Grossiord et al., 2007). These treatments were applied both in the absence (Salzmann et al., 2007) and in the presence of solubilizing agents: NOM (Hyung et al., 2007), Triton X-100 (Islam et al., 2003), Triton X-405 (Chappell et al., 2009), PVP-1300 (Didenko et al., 2005), GA (Bandyopadhyaya et al., 2002), and starch (Star et al., 2002). A three-step approach to solubilizing single-walled carbon nanotubes (SWCNTs) with amylose was suggested by Kim et al. (Kim et al., 2004): dispersion of SWCNT in water by sonication followed by treatment with amylose in dimethylsulfoxide (DMSO)–H2O mixture, followed by sonication allowing for molecularly controlled encapsulation of CNTs.

1.1.

1.2.

Dispersing CNTs in aqueous media

Recent efforts on the development of efficient and facile methods of dispersing CNTs in aqueous media have been focused on the hydrophilization of CNT with molecules that bind to the CNT surface non-covalently (O’Connell et al., 2001; Bandyopadhyaya et al., 2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003; Didenko et al., 2005; Wang et al., 2005; Bonnet et al., 2007; Grossiord et al., 2007; Hyung et al., 2007; Liu et al., 2007). Such non-covalent functionalization has great promise as the modification-induced changes in the electronic and mechanical properties of CNTs are minimized (Yang et al., 2007). Various surfactants (Islam et al., 2003; Moore et al., 2003; Grossiord et al., 2007), synthetic polymers (e.g., polyvinyl pyrrolidone (O’Connell et al., 2001; Didenko et al., 2005), poly(ethylene glycol) (Vaisman et al., 2006), polyphosphazene (Park et al., 2006)), natural organic matter (NOM) (Hyung et al., 2007; Liu et al., 2007; Saleh et al., 2009), biomolecules (e.g., proteins (Karajanagi et al., 2006; Zong et al., 2007), aminoacids (Georgakilas et al., 2002), DNA (Enyashin et al., 2007)), and carbohydrates (e.g., cyclodextrines (Dodziuk et al., 2003), amylose (Bonnet et al., 2007), starch (Star et al., 2002), and GA (Bandyopadhyaya et al., 2002)) have been evaluated as dispersants for CNTs. Dispersion via non-covalent functionalization is based on the direct contact between a CNT and a dispersant molecule (Liu et al., 1998; Grossiord et al., 2007). Such modification of the CNT surface facilitates the disaggregation (i.e. debundling) of CNT bundles into smaller diameter bundles (Liu et al., 1998) or even individual CNTs (O’Connell et al., 2002; Hyung et al., 2007) and leads to the stabilization of suspended CNTs via steric or electrostatic repulsion mechanisms or both. (See Supporting Documentation (SD), Section S.1.1, for a brief review of mechanisms of non-covalent functionalization of CNTs.) The dispersion of CNTs in water has been enhanced by mixing (Didenko et al., 2005; Hyung et al., 2007), sonication (Bandyopadhyaya et al., 2002; Liu et al., 2007; Salzmann et al., 2007), or mixing followed by sonication (O’Connell et al., 2001,

Stability of CNT suspensions in water

Previous studies of the long-term changes in suspensions of dispersed CNTs have focused on monitoring changes in the concentration of suspended CNTs (Jiang and Gao, 2003; Tseng et al., 2006; Lee et al., 2007; Marsh et al., 2007). By measuring UV–vis absorption at certain wavelength: 253 nm (Jiang and Gao, 2003), 300 nm (Sinani et al., 2005), 500 nm (Bahr et al., 2001; Lee et al., 2007), 530 nm (Marsh et al., 2007), and 800 nm (Hyung et al., 2007) the change in the concentration of suspended CNTs with time was determined. Aqueous suspensions of non-functionalized CNTs are known to be unstable. There are considerable quantitative differences, however, in the reported stability data for nonfunctionalized CNTs. The concentration of unmodified multiwalled carbon nanotubes (MWCNTs) suspended in deionized water was reported to decline 86 % over 2 h in one study (Marsh et al., 2007) and only 50% over 500 h in another study (Jiang and Gao, 2003). The suspensions of unmodified SWCNTs in deionized water were found to completely precipitate after only 4 h (Tseng et al., 2006). It was found that non-covalent modification of CNT surface drastically improved the stability of CNT suspensions (Jiang and Gao, 2003; Tseng et al., 2006; Hyung et al., 2007; Lee et al., 2007; Marsh et al., 2007). The stability of the suspension of non-covalently functionalized CNTs was found to be a function of the type (Hyung et al., 2007; Lee et al., 2007) and concentration (Chappell et al., 2009) of the dispersant, CNT length (Marsh et al., 2007) and the presence of inorganic salts in the solution (Saleh et al., 2009). SWCNTs stabilized with oligothiophene-terminated poly(ethylene glycol) produced suspensions that were significantly more stable than CNTs dispersed in water with the aid of sodium dodecyl sulfate (SDS) and Pluronic F127 (Lee et al., 2007). CNT suspensions in model Suwannee River NOM (SRNOM) solutions and in Suwannee River water were found to be considerably more stable than suspensions of CNTs dispersed in 1% aqueous solution of SDS (Hyung et al., 2007). Chappell et al. showed

water research 44 (2010) 505–520

that the stability of MWCNTs dispersed using two types of humic acids, Triton X-405, Brij 35, and SDS was enhanced in dose-dependent manner (Chappell et al., 2009). In a study of the aggregation kinetics of MWCNTs in aquatic systems, Suwannee River humic acid was shown to significantly enhance stability of MWCNTs suspensions in the presence of monovalent and divalent salts (Saleh et al., 2009). While studying the stability of MWCNTs in water, Marsh and coworkers (Marsh et al., 2007) observed that wrapping of the annealed CNTs with 1% SDS or charge doping increased their stability in the suspension; the authors demonstrated that shorter CNTs form more stable dispersions than longer CNTs of the same diameter. In summary, there is growing evidence that by the appropriate choice of a dispersant that modifies CNT surface non-covalently, highly stable CNT suspensions can be produced. However, to date there have been no systematic investigations that correlated long-term changes in size and charge of CNTs dispersed via non-covalent functionalization with the dispersion stability.

1.3.

Toxicity of CNTs

1.3.1.

Toxicity of CNTs towards eukaryotic cells: cytotoxicity

Most nanomaterial toxicity studies have been performed with mammalian cells, in particular with lung and skin cell cultures reflecting the understanding that the most likely routes of an organism’s exposure to nanomaterials are respiratory and dermal contact. With the development of methods of CNT dispersion in aqueous media, the assessment of the toxicity of such dispersed CNTs becomes very important in view of their increased mobility and potential to enter water supplies. While the toxicity of CNTs was studied with respect to the type of the CNT (MWCNT versus SWCNT) (Ding et al., 2005; Jia et al., 2005; Kang et al., 2008a), surface functionalization (Sayes et al., 2006; Yu et al., 2007; Meng et al., 2009; Yun et al., 2009), CNT length (Muller et al., 2005; Kang et al., 2008b; Simon-Deckers et al., 2008), exposure dose (Cherukuri et al., 2006; Flahaut et al., 2006; Sayes et al., 2006; Kang et al., 2007; Pulskamp et al., 2007; Yang et al., 2008, 2009; Ye et al., 2009), degree of purification (Fiorito et al., 2006; Flahaut et al., 2006; Elias et al., 2007; Simon-Deckers et al., 2008), and degree of dispersion (Wick et al., 2007), only very limited information exists on the biocompatibility of CNTs as a function of surface coating. Several hypotheses have been put forth to explain the likely pathways of CNT toxicity: (i) oxidative stress induced by the formation of reactive oxygen species (ROS) generated at the surface of CNTs (Manna et al., 2005; Yang et al., 2008; Ye et al., 2009); (ii) the presence of residual catalyst, which is used during the CNT manufacturing process (Kang et al., 2008a); (iii) physical contact between a CNT and a cell (Kang et al., 2007); or (iv) a combination of these factors (Kang et al., 2008a). The dispersant used for the stabilization of CNTs in suspension was suggested as another possible cause of the observed nanotube toxicity (Sayes et al., 2006) (See SD, Section S.1.2 for a brief review of possible mechanisms of CNT toxicity). In most studies of the toxicity of non-covalently functionalized CNTs the dispersants were surfactants. In vivo assessment of the toxicity of SWCNTs modified with Pluronic F108 administered intravenously to rabbits showed the

507

absence of the acute toxicity (Cherukuri et al., 2006). Pluronic F-127-coated MWCNTs did not affect cell viability, apoptosis, and ROS formation in mouse and human neuroblastoma cells (Vittorio et al., 2009), and mouse cerebral cortex (Bardi et al., 2009). In contrast, MWCNTs dispersed using Pluronic F68 caused cell death, changes in cell size and complexity, ROS production, interleukin-8 (IL-8) gene expression and nuclear factor (NF)-jB activation (Ye et al., 2009). CNTs dispersed using Tween 80 were toxic to mesothelioma cells (Wick et al., 2007), led to an inflammation of murine allergic airway with augmented humoral immunity (Inoue et al., 2009), and induced inflammatory and fibrotic responses when intracheally administrated to rats (Muller et al., 2005). More ROS were observed upon exposure of human lung epithelial or primary bronchial epithelial cells to SWCNTs dispersed using dipalmitoylphosphatidylcholine, a major component of lung surfactant (Herzog et al., 2009), as compared to dipalmitoylphosphatidylcholine-free samples. Low toxicity was observed in in vivo experiments when SWCNTs were dispersed using Tween 80 and intravenously injected into mice (Yang et al., 2008). In another study, SWCNTs dispersed in 1% SDS aqueous solution showed no cytotoxicity with respect to the human alveolar epithelial cells (Wo¨rle-Knirsch et al., 2006). CNTs dispersed using Pluronic PF-127 solution did not affect viability, apoptosis and ROS generation in the human neuroblastoma cells after 3 days of incubation; however, cell proliferation decreased as incubation time increased to 2 weeks (Vittorio et al., 2009). In the only paper that mentioned the potential toxicity effect of the dispersant, the authors suggested that excess surfactant was responsible for the observed increase in toxicity; in this work, the controlled exposure of cells to 1% Pluronic F108 produced a 10% decrease in the cells viability (Sayes et al., 2006). Very little is known about the effect of non-covalent wrapping with dispersants other than surfactants on the biocompatibility of CNTs. In vitro cytotoxicity of CNTs wrapped with poly(methyl vinylketone) decorated with a-N-acetylgalactosamine (Chen et al., 2006), nano-1 peptide (Chin et al., 2007) and cholesterolend-capped poly(2-methacryloyloxyethyl phosphorylcholine) (Xu et al., 2008) were examined after their contact with human lung epithelial-like cells, normal skin fibroblasts and human umbilical vein endothelial cell line. No impact on cell growth and proliferation was demonstrated in all three cases. Cytotoxicity of GA-stabilized MWCNTs upon exposure to A549 cells was observed by LDH, XTT and MTT bioassays in (SimonDeckers et al., 2008). Two possible reasons were hypothesized to be responsible for this effect: (i) increased availability of GAstabilized MWCNTs and (ii) different intercellular accumulation pathway of GA-stabilized MWCNTs as compared to when carbon nanotubes were exposed in bundles.

1.3.2. Toxicity of CNTs towards eukaryotic cells: genotoxicity and epigenetic toxicity Most toxicological studies of CNTs focused on the evaluation of cytotoxicity (Sayes et al., 2006; Elias et al., 2007; Gutierrez et al., 2007; Kang et al., 2007; Wick et al., 2007; Kang et al., 2008a,b; Yang et al., 2008; Bardi et al., 2009; Herzog et al., 2009; Inoue et al., 2009; Kang et al., 2009; Vittorio et al., 2009; Yang et al., 2009). However, genotoxicity (Kang et al., 2008a; Di Sotto et al., 2009; Lindberg et al., 2009; Wirnitzer et al., 2009; Yang et al., 2009) and

508

water research 44 (2010) 505–520

epigenetic toxicity (see SD, Section S.1.3) (Upham et al., 1994); (Trosko et al., 1998) are other possible causes of cell damage or other adverse effects. The genotoxicity of SWCNTs dispersed using fetal bovine serum at (5–10) mg/mL concentration, with respect to primary mouse embryo fibroblasts was demonstrated using Comet assay (Yang et al., 2009). In another recent study, CNTs (dispersed in BEGM cell culture medium and subjected to ultrasonication) induced a dose-dependent increase in DNA damage as indicated by Comet assay and caused a significant increase in micronucleated cells (micronucleus assay) in human bronchial epithelial cells (Lindberg et al., 2009). Kang et al. observed high levels of stress-related gene products in Escherichia coli upon its exposure to CNTs, with the quantity and magnitude of expression being much higher in the presence of SWCNTs (Kang et al., 2008a); in this study CNTs were either deposited on a PVDF membrane surface or dispersed in saline solution. No mutagenic effect was observed in Salmonella microcosme test with baytubes (high purity MWCNTs agglomerates, sonicated 10 min in deionized water) (Wirnitzer et al., 2009) and in bacterial reverse mutation assay (Ames test) with Salmonella typhimurium TA 98 and TA 100 strains and with E. coli WP2uvrA strain exposed to MWCNTs dispersed in DMSO (Di Sotto et al., 2009). There have been no reports to date on the epigenetic toxicity of carbon nanomaterials.

1.3.3.

2007). Lyon et al. showed that in media with high salt content, the size of nC60 aggregates tends to increase as compared to that observed in low salt media (Lyon et al., 2005). In another study, MWCNTs were found to stimulate growth of unicellular protozoan Tetrahymena pyriformis in growth medium, which contained proteose peptone, yeast extract, and glucose; the increase in growth was attributed to the formation of peptone-MWCNTs conjugates, which were taken up by the microorganism (Zhu et al., 2006). While the aforementioned studies indicate that salt and organic composition of the medium, in which exposure studies are performed, may influence the interaction of CNTs with bacteria, there have been no studies that comparatively evaluated CNT toxicity in growth media with different organic loadings and salt contents.

1.5.

Objectives of this study

This study addressed some of the knowledge gaps identified above. Aqueous suspensions of SWCNTs functionalized by a range of non-covalently bound dispersants of natural (NOM, GA, amylose) and synthetic (PVP, Triton X-100) origin, were prepared and evaluated in terms of their physicochemical and toxicity properties. The study pursued the following objectives:

Toxicity of CNTs in prokaryotic cells

Most toxicity studies have focused on the effect of CNTs on mammalian cell lines. Only limited information exists on the cytotoxic effects of CNTs towards bacterial cells. Recently, antimicrobial activity of SWCNTs (Kang et al., 2007, 2008a, 2009) and MWCNTs (Kang et al., 2008a,b, 2009) either suspended in aqueous solution (Kang et al., 2007, 2008a) or deposited on the surface of a PVDF microfilter (Kang et al., 2007, 2008a,b, 2009) towards gram-negative bacteria (E. coli and Pseudomonas aeruginosa (Kang et al., 2007, 2008a,b, 2009) and gram-positive (Staphylococcus epidermidis and Bacillus subtilis (Kang et al., 2009)) bacteria was reported. It was suggested that membrane damage to the cells was caused by the direct physical contact between the CNT and cell (Kang et al., 2007) or by a combination of direct physical contact and oxidative stress (Kang et al., 2008a). No effect on the percentage of E. coli inactivation was observed upon exposure of SRNOM-stabilized SWCNTs as compared of SWCNTs dispersed in the absence of SRNOM (Kang et al., 2009). Significant antimicrobial activity of CNTs composite films against Staphylococcus aureus and Staphylococcus warneri was reported in a separated study (Narayan et al., 2005). In contrast to the findings of above studies, no inhibition of E. coli growth and proliferation was reported in a study where microchannel-structured MWCNTs scaffolds were immersed into the culture medium with the cells (Gutierrez et al., 2007).

1.4. Effect of growth media on the toxicity of nanoparticles The existing literature highlights the importance of the growth media in cytotoxicity testing and links the apparent cytotoxic effect of the nanomaterial to the salt and organic content of the culture and related physicochemical characteristics of the nanomaterial (Lyon et al., 2005; Tong et al.,

(i) To evaluate the long-term stability and its physicochemical determinants for non-covalently functionalized SWCNTs as a function of dispersant type. The evolution of the concentration, size, and charge of suspended SWCNTs was studied over the period of 28 days. (ii) To assess time-dependent cytotoxicity of non-covalently functionalized SWCNTs with respect to bacteria and mammalian cells as a function of dispersant type and growth media. The effect of SWCNTs on E. coli was studied by measuring cell viability after 3 h, 24 h, and 48 h of exposure in three types of growth media with different organic loadings and salt contents. The cytotoxicity of dispersed SWCNTs towards mammalian cells was assessed in NDU bioassay with rat liver epithelial cells after 30 min and 24 h of incubation. (iii) To assess time-dependent epigenetic toxicity of non-covalently functionalized SWCNTs as a function of dispersant type. In this study, we evaluated epigenetic toxicity of the SWCNTs to rat liver epithelial cells as a function of dispersant type in GJIC bioassay after 30 min and 24 h of incubation.

2.

Approach

The set of chemically diverse dispersants was chosen based on their demonstrated effectiveness in solubilizing CNTs (Bandyopadhyaya et al., 2002; O’Connell et al., 2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003; Hyung et al., 2007; Liu et al., 2007) and potential adverse effects (Burnette, 1960; Chourasia and Jain, 2004; Dayeh et al., 2004; Schmitt et al., 2008). NOM, GA and PVP have LD50 doses of (54.8–58.5) mg (intravenous administration in mice) (EMEA, 1999), 2,000 mg (oral administration in rats) (Schmitt et al., 2008), and 100,000 mg (oral administration in rats) (Burnette, 1960) per kg

water research 44 (2010) 505–520

of body weight, respectively. Amylose has been reported to be used in a colon-specific drug delivery due to its low toxicity and high biodegradability (Chourasia and Jain, 2004). Triton X-100 was reported to be toxic to protozoa, fish, and mammalian cells (Dayeh et al., 2004). Three batches of dispersed SWCNTs were prepared. The first batch was used to comprehensively evaluate the longterm stability of SWCNT suspensions over a period of 4 weeks in terms of concentration, effective hydrodynamic diameter, and z-potential of stabilized SWCNTs. The second batch was used to evaluate the viability of E. coli cells after their contact with dispersed SWCNTs by the quantification of colony forming units. To elucidate the effects of ionic and organic composition of the growth medium, the cytotoxicity of the SWCNTs suspended in three growth media of different organic and salt compositions was evaluated. First, in order to assess the acute cytotoxicity of dispersed SWCNTs, we conducted experiments in 0.1 M NaCl. Then, in order to investigate the effect of SWCNTs on the ability of E. coli form colonies over time, we employed three types of growth media: (i) LB medium with higher organic chemical and salt content, (ii) MD medium with low salt content and organic load, and (iii) 0.1 M NaCl solution. The third batch was used to evaluate cyto- and epigenetic toxicity of the dispersed SWCNTs against rat liver epithelial cells by the (i) NDU and (ii) GJIC bioassays. The NDU assesses cell viability by measuring the accumulation of neutral red dye in lysosomes, which depends on the cell’s capacity to maintain pH gradients through the maintaining membrane integrity and production of adenosine triphosphate. The amount of dye incorporated by the cell is quantified spectrophotometrically (Borenfreund and Puerner, 1985). The NDU bioassay has been used to assess cytotoxicity of CNTs (Flahaut et al., 2006). The GJIC bioassay (Borenfreund and Puerner, 1985; Weis et al., 1998; Herner et al., 2001; Satoh et al., 2003) utilizes the ability of epigenetic tumor promoters to alter level of GJIC (Yamasaki, 1990; Trosko et al., 1991). The degree of GJIC was quantified by measuring the distance (area) the fluorescent dye (1200 Da) travels (occupies) between the cells after a given time. Both NDU and GJIC bioassays were carried out with WB-F344 rat liver epithelial cells exposed to dispersed SWCNTs for different periods of time. The WB-F344 cell line was chosen because these normal diploid rat liver epithelial cells have already used in numerous studies of cytotoxic or epigenetic effects (e.g., Herner et al., 2001) thus allowing us to have a comparative basis.

3.

Materials and methods

3.1.

CNTs

SWCNTs (purity > 90%), produced by catalytic chemical vapor deposition, were obtained from Cheap Tubes, Inc (Brattleboro, VT) and used as received. The SWCNTs were used as obtained, allowing us to mimic what might occur in the environment. As indicated by the manufacturer, the SWCNTs had an inside diameter in the range of 0.8 nm to 1.6 nm, an outer diameter in the range of 1 nm to 2 nm and were 5 mm to 30 mm in length.

509

3.2. Non-covalent functionalization: dispersants and dispersion procedures 3.2.1.

Dispersants

Gum arabic (approx. 250 kDa; a complex mixture of saccharides and glycoproteins obtained from the acacia tree), PVP (approx. 29 kDa), Triton X-100 (approx. 625 Da; polyethylene glycol p-(1,1,3,3-tetramethylbutyl)-phenyl ether) and amylose (molecular weight not specified; a polymeric form of glucose and a constituent of potato starch) were purchased from Sigma–Aldrich (Milwaukee, WI). SRNOM reverse osmosis isolation was obtained from International Humic Substances Society (St Paul, MN). The molecular structure of dispersants used in this study is presented in Fig. 1.

3.2.2.

Dispersion procedure

Aqueous solutions of PVP, Triton X-100 were prepared by dissolving PVP (40 mg) and Triton X-100 (0.4 mL) in 40 mL of water and adjusting the pH of both solutions to 7 with 0.1 M HCl. The aqueous solution of GA was prepared by mixing 1 g of GA in 50 mL of water and adjusting the pH of the mixture to 7 with 0.1 M HCl; this mixture was left to settle for 24 h and then 40 mL of supernatant was collected for use in the stability or toxicity studies. To prepare NOM solutions, two flasks of 40 mL of water, each containing 8 mg of NOM, were stirred for 24 h. Each solution was then filtered through a 0.22 mm filter under vacuum. The pH of one solution was kept at its original value (approx. 3.5) while the pH in another solution was adjusted to 7 with 0.1 M NaOH. SWCNTs were added to the above solutions to result in a 1 mg (SWCNT)/mL loading and were sonicated in Aquasonic 50 T water bath (VWR Scientific Products Corp, West Chester, PA) for 20 min. SWCNTs were dispersed with amylose based on modified version of the three-step approach reported by Kim et al. (Kim et al., 2004). SWCNTs (40 mg) were sonicated in 25 mL of water (pH 7) at approximately 75 W for 5 min using Sonicator 3000 (Misonix, Inc., Farmingdale, NY) equipped with a microprobe. 100 mg of amylose in 6.28 mL of DMSO were prepared and added to the sonicated suspension of SWCNT in water so that the DMSO/water ratio was 20% by volume. The resulting mixture was sonicated for another 5 min in order to remove the excess amylose and DMSO. The suspension was sonicated, centrifuged and refilled with water four times (See SD, Section S.2.3). Following the sonication, SWCNTs suspensions were divided into 12 mL aliquots and each aliquot was transferred into 15 mL centrifuge tube. After 24 h standing time, the top 4 mL of each suspension was withdrawn in order to be used in the stability or toxicity studies. Overall, three batches of dispersed SWCNT suspensions were prepared and were used, correspondingly, for 1) the study of the long-term stability of SWCNTs suspensions, 2) the general viability bioassay with E. coli, and 3) NDU and GJIC bioassays.

3.3. Physicochemical characterization of SWCNT suspensions In order to quantitatively assess the long-term stability SWCNT suspensions, a combination of characterization methods was employed.

510

water research 44 (2010) 505–520

(a-1)

a

GALP

1

1 3

ARAF

3

3

ARAF

ARAF

GALP

O OH

OH

O OH

OH

1

1

3

CH2OH

3

ARAF

1 6

(a-2)

GALP

CH3

OH

1

1

3

GALP

3

1

6

GALP

1

3 3

1

3

RHAP

GALP

RHAP

6

GALP

1

1

1

GA

GA

4

4

1

1

ARAF

ARAF

OH

(a-3)

GALP 6 1

OH

OH

6

6

(a-4) HOOC

O

O

OH H2COH

HO HO

OH OH

OH

GALP = D-GALACTOPYRANOSE ARAF = L-ARABOFURANOSE GA = D-GLUCURONIC ACID RHAP = L-RHAMNOPYRANOSE

b

OH

c HO

O

O

O

N

OH

O

O COOH

H2N

OH

HO

n

HOOC

d

COOH

OH

e O

COH

OH

O

n

C8 H17

O

OH

O OH

n

Fig. 1 – Molecular structure of solubilizers: (a) Gum arabic (a-1) galactose, (a-2) rhamnose, (a-3) arabinose, (a-4) glucuonic acid; (b) Poly(vinyl pyrrolidone); (c) A building block of humic acids, which has a compositional similarity to SRNOM; Dots represent chiral centers; (d) Triton X-100; (e) Amylose.

3.3.1. UV–vis spectrophotometry. Quantification of SWCNTs concentration in suspension The absorbance of SWCNT suspensions over the (200–800) nm wavelength range was measured over a period of 4 weeks (Multi-spec 1501, Shimadzu, Kyoto, Japan). For each type of SWCNT suspension prepared, the absorbance of SWCNT-free solution of the corresponding dispersant was used as a baseline. SWCNT suspensions used in our toxicity studies likely contained both individual and bundled SWCNTs. In fact, DLS measurements indirectly confirmed (see Section 4.2) the presence of dispersed SWCNT bundles and direct TEM imaging (see SD, Section S.3.3, Fig. S5) also showed the presence of SWCNTs bundles (note that TEM results need to be interpreted with caution as evaporation-induced aggregation could have contributed to bundling during TEM sample preparation). Incomplete exfoliation is the most likely and most environmentally relevant scenario; unfortunately, it also

entails difficulties with quantifying the total concentration of suspended nanotubes as larger CNT bundles tend to separate from the suspension. In this study, we used UV–vis spectrophotometry to estimate the concentration of exfoliated SWCNTs in suspensions. It is known that only fully exfoliated SWCNTs absorb in the (200–1200) nm wavelength range. Bundled NTs do not absorb significantly in this range due to the tunnel coupling between metallic and semiconductive SWNTs (Lauret et al., 2004; Grossiord et al., 2007). Therefore, the UV absorbance by a SWCNT suspension can be used to selectively measure the concentration of individually dispersed (i.e. fully exfoliated or debundled) SWCNTs. To determine the concentration of suspended exfoliated SWCNTs, we first prepared a separate set of suspensions with ten-fold reduced SWCNT loading with respect to the CNT loading used in long-term stability and toxicity studies. Four suspensions (with PVP, NOM pH 3.5, NOM pH 7, and GA used

water research 44 (2010) 505–520

as dispersants) with 10-fold reduced SWCNT content (0.1 g/L) were prepared using the same procedures as described in Section 3.2 except that they were subjected to rigorous, prolonged sonication for 1 h at power of (70–80) W using horn sonicator (Sonicator 3000, Misonix, Inc., Farmingdale, NY). No settling of SWCNTs was observed over the short term following the application of this treatment indicating complete dispersion of SWCNTs. UV–vis absorbance spectra were recorded at different dilutions and a calibration curve for absorption at 500 nm (Bahr et al., 2001; Huang et al., 2002; Sinani et al., 2005; Lee et al., 2007; Salzmann et al., 2007) as a function of concentration was constructed, and coefficient of molecular extinction, 3, was determined. This coefficient was used to estimate the concentration of exfoliated SWCNTs in suspensions used in long-term stability and toxicity experiments. Note that by subjecting suspended SWCNTs to a very intense sonication treatment we aimed at maximizing the extent of exfoliation; however, it was not possible to ascertain that the exfoliation was complete. Thus coefficients of molecular extinction and exfoliated SWCNT concentrations determined using the recorded calibration curves were estimated values. SWCNT characteristics measured upon the preparation of CNT suspensions are given in Table S1 in SD.

3.3.2.

Size and charge of dispersed SWCNTs

The effective hydrodynamic diameter and z-potential of suspended SWCNTs were determined after 1, 4, 7, 14, 21, and 28 days of settling. The size and charge were measured by dynamic light scattering (DLS) and phase analysis light scattering techniques, respectively (ZetaPALS, BI_MAS Option, Brookhaven Instrument Corp., Holtsville, NY). The Smoluchowski equation was applied to convert the measured electrophoretic mobility of dispersed SWCNT to z-potential. Transmission electron microscopy (TEM) imaging was used as an auxiliary method to aid in the interpretation of dynamic light scattering data (see SD, Sections S.2.5 and S.3.4).

3.4.

Toxicity assessment: E. coli viability assay

3.4.1.

Media preparation

Luria–Bertani (LB) growth medium was prepared according to the standard procedure (Atlas, 1993). Minimal Davis medium with 90% reduced potassium phosphate concentration (MD medium) was prepared (Fortner et al., 2005). 0.1 M NaCl solution was prepared by dissolving 9 g of NaCl in 1 L of water (pH 7) and autoclaving it for 15 min at 1 bar and 121 C. For the toxicity assessments, each component of LB, MD and NaCl media was prepared in 4-fold higher concentration as compared to original protocol and the aliquot part of the corresponding medium was added to SWCNTs suspension (as further described in ‘‘Quantification of cell viability’’ Section). Luria–Bertani Petri plates were prepared according to the published method (Atlas, 1993).

3.4.2.

Preparation of E. coli culture

E. coli K12 stock was prepared in glycerol and stored at 80 C. Prior to use, the stock was defrosted, and 30 mL of LB medium were inoculated with 5 mL of the stock. After overnight growth at 37 C, 5 mL of this suspension was spread onto LB agar plate and cultured at 37 C. Once distinct colonies were formed, the

511

agar plate was transferred to the refrigerator and kept at 4 C for up to one month. E. coli suspensions to be used in SWCNT cytotoxicity studies were prepared by scraping one colony from the surface of a Petri plate by aseptic loop and immersing the loop into 10 mL of LB or MD media in a 50 mL centrifuge tube. Tubes were placed on a shaker in an incubator 37 C for 12 h. When 0.1 M NaCl was used as an exposure medium in colony forming units bioassay, E. coli were first grown in the LB medium, centrifuged for 5 min at 2250 rpm and washed with 0.1 M NaCl as follows: the supernatant was decanted and replaced with an equal volume of 0.1 M NaCl, vortexed and resuspended by centrifugation. The washing procedure was repeated twice, presuming that after this treatment most of the remaining organic constituents of LB medium were removed from the E. coli suspension.

3.4.3.

Quantification of cell viability

The SWCNTs suspension (1.425 mL), growth medium (475 mL) and 100 mL of the stock E. coli suspension were transferred into a 15 mL centrifuge tube and incubated under gentle shaking at 37 C. Samples were taken after 3, 24, and 48 h and a series of dilutions (104–106) was prepared for each sample. Five samples of 10 mL and 20 mL from each dilution were placed onto an agar plate and incubated at 37 C until distinct colonies developed. Colony forming units (CFU)/1 mL were calculated for each sample. Each experiment was run in triplicates with negative (bacterial suspension with the corresponding amount of ultrapure water) and vehicle (solution of the corresponding dispersant) controls, herein called vehicle control I and vehicle control II, respectively. The results are reported as a fraction of control (FOC) standard deviation (STD), calculated as the ratio of the average number of colonies grown after E. coli exposure to SWCNTs suspensions or vehicle control II to the average number of colonies grown in vehicle control I plates.

3.5. Toxicity assessment: neutral red dye uptake bioassay For the NDU bioassay we adapted a published procedure (Borenfreund and Puerner, 1985; Weis et al., 1998; Satoh et al., 2003). A solution (0.015 w/v) of neutral red dye (3-amino-7(dimethylamino)-2-methylphenazine hydrochloride) in D-medium was incubated at 37 C for 2 h and filtered through a 0.22 mm syringe filter (Millipore Corp., New Bedford, MA) to remove undissolved dye and ensure sterile conditions. Confluent WB-F344 cells (see SD, Section S.2.6 for details on the preparation of the cells) were exposed to 500 mL of each of SWCNTs suspension and incubated at 37 C for 30 min and 24 h under gentle shaking in a humidified atmosphere containing 5% CO2. After cells were exposed to SWCNTs, the exposure medium was removed by aspiration and the cells were washed with 1 mL of phosphate saline buffer (PBS). Following washing, 2 mL of the neutral red dye solution per plate was added and the cells were incubated for 1 h at 37 C in the humidified atmosphere containing 5% CO2. Upon incubation, the cells were rinsed three times with PBS, and 2 mL of aqueous solution containing 1% acetic acid and 50% ethanol was added to each plate to lyse the cells. 1.5 mL of the lysate was transported into 2 mL microcentrifuge tube and the optical density was

512

water research 44 (2010) 505–520

recorded at 540 nm using a Beckman DU 7400 diode array detector (Beckman Instruments, Inc., Schaumburg, IL). The background absorbance was measured at 690 nm and then subtracted from the original absorbance. Each experiment was conducted in triplicates. The neutral red dye uptake was reported as the FOC (absorption of neutral red in the chemically treated sample divided by the absorption of neutral red in the nontreated control I sample). FOC values of 1.0 indicates noncytotoxic response while FOC values <0.8 indicate that less dye was absorbed by cells and that chemical is cytotoxic at that dose (El-Fouly et al., 1987; Satoh et al., 2003).

3.6. Toxicity assessment: gap junction intercellular communication (GJIC) bioassay The principles of GJIC bioassay were originally described in (Borenfreund and Puerner, 1985) and further developed in (Weis et al., 1998; Herner et al., 2001; Satoh et al., 2003). SWCNTs suspensions (500 mL) were introduced to the cell culture plates with confluent WB-F344 cells and incubated for 30 min and 24 h at 37 C in humidified atmosphere containing 5% CO2 under gentle shaking. Then, cells were washed with PBS solution and 1 mL of 0.05% solution of Lucifer yellow fluorescent dye in PBS buffer was added to each plate. Three different scrapes were made on the bottom of a cell culture plate using a surgical steel scalpel blade. The dye was absorbed through the monolayer of confluent cells. The transfer of dye through gap junctions lasted 3 min, followed by a thorough rinse with PBS solution to remove extracellular dye. Cells were fixed with 0.5 mL of 4% formalin solution in PBS. The migration of the dye in the cells was observed at 200 magnification using a Nikon epifluorescence microscope equipped with a Nikon Cool Snap EZ CCD camera and the images were digitally acquired using a Nikon NIS-Elements F2.2 imaging system. The fluorescence area of the dye migration from the scrape line was quantified using Image J 1.40 g (a public domain Java image processing program, http:// rsbweb.nih.gov/ij/index.html). Each experiment was performed in duplicate with same controls as in the NDU bioassay. Data are reported as the FOC, which is the ratio of the area of dye spread in the chemically treated sample to the area of dye spread in the vehicle control I samples. A FOC value of 1.0 corresponds to full communication between cells, FOC values of (0.5–0.9) represent partial GJIC inhibition, (0.3–0.5) values correspond to significant inhibition of the GJIC and values <0.3 denote complete inhibition of the GJIC (Weis et al., 1998). Student’s t-tests at 95% confidence interval were run for all three bioassays to determine whether the difference between the results obtained in the control I plates and the results in the tested plates were significant. The concentrations of SWCNTs upon addition to suspension of cells in the general E. coli viability bioassay, and in NDU and GJIC bioassays are reported in Table 1.

4.

Results and discussion

The knowledge of the evolution of physicochemical properties (e.g., charge and size) of dispersed CNTs can provide insights into the basis for both CNT stability and possible toxicity.

4.1.

Concentration of dispersed SWCNTs in suspensions

First, SWCNTs were suspended directly in ultrapure water without a dispersant; this suspension was used as a control. After sonication for 20 min, the mixture became a lightly colored suspension that completely separated into a dark subnatant layer and a clear supernatant after only 1 h of settling. In contrast, 20 min of sonication of the SWCNT–water mixture in the presence of a dispersant produced relatively stable colored suspensions (see SD, Fig. S1) that precipitated gradually with the precipitation rate varying as a function of the type of dispersant. The amount of the precipitate that formed at the bottom of a centrifuge tube visibly increased with settling time for all types of SWCNT suspensions. Immediately after sonication, all SWCNT suspensions prepared using the different dispersants appeared black. After 4 weeks of settling, in suspensions of NOM-stabilized and amylose-stabilized SWCNTs the color of the aqueous solution of the dispersant was revealed. After 7 weeks of settling, the color change could be detected in all samples with the suspension of amylose-stabilized SWCNTs (see SD, Fig. S1) undergoing the most dramatic change to an almost transparent supernatant phase (Figure S1 (b)). The UV–vis absorption measurements were performed to determine the concentration of suspended exfoliated SWCNTs as a function of settling time. The calibration curves for the intensity of the absorbance at 500 nm as a function of SWCNT concentration (see Section 2.3) were recorded for all suspensions (see SD, Fig. S2) except for amylose-stabilized SWCNTs. Because of the observed loss of SWCNTs during centrifugation/washing steps for amylose-dispersed SWCNTs (see SD, Section S.2.3), the concentration of amylosedispersed SWCNTs could not be measured. The calibration curves were linear (R2 0.99). The extinction coefficients for suspensions of SWCNTs dispersed using GA-, PVP-, Triton X-100-, NOM (pH 3.5)-, and NOM (pH 7) were determined to be 1.81 102, 1.97 102, 2.51 102, 2.54 102, and 2.70 102 (L mg1 cm1), respectively. The concentration of exfoliated SWCNTs in suspension as a function of settling time is given in Fig. 2a. The dispersion and exfoliation of SWCNTs bundles in the presence of GA and PVP was attributed to the wrapping of a long polymeric chain around the nanotube core (Bandyopadhyaya et al., 2002; Didenko et al., 2005), whereas supramolecular encapsulation was hypothesized as the mechanism of CNT modification by helical amylose (Kim et al., 2004).

Table 1 – Concentrations of the SWCNTs upon addition to cells in the general viability bioassay with E. coli, NRDU bioassay, and GJIC bioassay. Solubilizer

Concentration of SWCNTs in general E. coli viability bioassay, mg/L

Concentration of SWCNTs in NRDU and GJIC bioassays, mg/L

NOM pH 3.5 NOM pH 7.0 PVP GA Triton X-100

68.27 40.12 77.38 142.91 107.19

11.22 11.93 23.42 34.97 26.43

513

water research 44 (2010) 505–520

b

200

Effective diameter (nm)

Concentration (mg/L)

a

160 120 80 40 0

1000 800 600 400 200 0

0

7

14

21

28

0

7

Time (d) 0

d

-15

-potential (mV)

-potential (mV)

c

14

21

28

21

28

Time (d)

-30

-45

0

-15

-30

-45

-60

0

7

14

21

28

Time (d)

0

7

14

Time (d)

Fig. 2 – The time-dependent characterization of (a) estimated concentration of the exfoliated SWCNTs in the dispersion, (b) effective diameter and (c) z-potential of dispersed SWCNTs, and (d) z-potential of dispersants alone (-B- for GA/SWCNTs, -,- for PVP/SWCNTs, -6- for NOM3.5/SWCNTs, -:- for NOM7/SWCNTs, --- for Triton X-100/SWCNTs, -C- for Amylose/ SWCNTs). Notes: (1) The z-potential measurement was not conducted for amylose solution, because amylose is not watersoluble under room temperature. (2) Reported are results of only those measurements that were statistically significant, i.e. when sufficiently high photon count rates were recorded in dynamic light scattering measurements. (3) The absorption by suspensions of amylose-stabilized SWCNTs was not measured because the nanotube content in these suspensions could not be precisely determined; this was due to the incomplete separation at the centrifugation step of the suspension preparation process.

Triton X-100 stabilizes nanotubes by the formation of hemimicelles that cover nanotube surface with benzene rings providing p–p stacking between the surfactant molecule and nanotube core (Islam et al., 2003). The exfoliation (debundling) of SWCNTs in the presence of SRNOM was suggested to occur through the interaction of the aromatic moieties of natural organic matter and nanotube surface (Hyung et al., 2007; Liu et al., 2007; Hyung and Kim, 2008). Despite the differences in the physiosorption mechanisms responsible for the stabilization of CNTs in aqueous suspensions, all dispersants were effective, albeit to somewhat different extents. For all SWCNT suspensions, the calculated values of the concentration of exfoliated SWCNTs correlated well to the visual observations described above. As seen from Fig. 2a, the dispersion efficiency in terms of the concentration of dispersed SWCNTs was a function of the type of the dispersant, with GA producing suspensions having highest concentration of SWCNTs. The data were consistent among the three prepared batches of SWCNTs (see SD, Table S1). Didenko et al. (Didenko et al., 2005) suggested that after covering one single carbon nanotube with a long polymeric molecule, the remaining strands would react with other uncovered or partially covered SWCNTs thus bundling several nanotubes together. We speculate that this mechanism where

one polymer molecule links two or more nanotubes may be responsible for the higher dispersing efficiency of GA (by far the largest molecule among the target dispersants) towards SWCNTs compared to other dispersants. This hypothesis is supported by the measurements of the effective hydrodynamic diameter (Fig. 2b), and the estimation of the length of suspended particles with the size and the length of GA/ SWCNTs being higher than those of PVP, Triton X-100 and SRNOM (both pHs). As the measured diameter of GA-stabilized SWCNTs aggregates (which could also be expected to be rather porous) was still relatively small, gravity settling did not lead to a significant removal of GA-stabilized SWCNTs from the dispersion. When comparing the solubilizing ability of SRNOM at different pH, one could see that SRNOM is a more efficient dispersant at pH 3.5 than at pH 7. This is consistent with the findings by Huyng et al. (Hyung and Kim, 2008) who reported that adsorption of SRNOM to MWCNTs increased when pH decreased due to more compact and coiled conformation of NOM at acidic pHs. When pH increases, carboxylic and phenolic groups of SRNOM deprotonate resulting in higher electrostatic repulsion between a CNT and a SRNOM molecule and, as a consequence, in a lower amount of organic matter adsorbed on the CNT surface. The better dispersion of

514

water research 44 (2010) 505–520

a Fraction of control (FOC)

SWCNTs at pH 3.5 could also be attributed, in part, to the steric hindrance imposed by SRNOM when a higher surface density of NOM on the surface of CNT bundles results in stronger repulsion between CNTs. This observation is supported by the long-term z-potential measurements (Fig. 2c), where the surface charge of stabilized SRNOM/SWCNTs at pH 3.5 became less negative up to day 7, and then stabilized with increasing settling time. At the same time the surface charge of SWCNTs dispersed in SRNOM solution at pH 7 gradually increased after day 5. Indeed, had the electrostatic repulsion been the sole mechanism of SWCNTs stabilization in SRNOM solutions, the surface charge on the SWCNTs would have more rapidly become less negative at pH 3.5 than at pH 7. In summary, in terms of the effectiveness of SWCNT stabilization, the dispersants were ranked as follows: GA > Triton X-100 > PVP > NOM (pH 3.5) > NOM (pH 7).

0.8 0.6 0.4 0.2

NOM 3.5 NOM 7

GA

Amylose

PVP

Triton

1.2

Charge of dispersed SWCNTs

For all suspensions, the z-potential of dispersed SWCNTs was negative and nearly constant over the entire duration of the experiment (Fig. 2c and SD, Table S1). Each z-potential

Fraction of control (FOC)

Hydrodynamic size of dispersed SWCNTs 1 0.8 0.6 0.4 0.2 0 NOM 3.5

c Fraction of control (FOC)

The hydrodynamic size of CNTs in a suspension is an important characteristic that affects the stability and, possibly, toxicity of dispersed CNTs. It should be noted that the effective hydrodynamic diameter of suspended SWCNTs measured using DLS can only be used as a very rough estimate. While the DLS data are interpreted with the assumption that the primary scatterers are spherical with an aspect ratio of 1, dispersed SWCNTs are long and tubular, with a very high aspect ratio (see SD, Fig. S3). To obtain a better approximation of the size distribution of dispersed SWCNT, the multimodal size distribution model was used. TEM imaging was employed to obtain auxiliary information on the size and morphology of stabilized SWCNT. Even though the measurement of effective hydrodynamic diameter cannot be relied on to compare the sizes of suspended SWCNTs in different dispersion media, these measurements can be used to compare how the average particle size for a given dispersant changes with settling time. The values of effective diameter as a function of time for different SWCNT suspensions are given in Fig. 2b. Generally, the effective size of SWCNTs dispersed using Triton X-100, GA and NOM did not change significantly with increasing settling time (Fig. 2b). In the case of amylosedispersed SWCNTs, a slight decrease in the effective size was observed due to settling of larger and unstable SWCNT aggregates, which left behind more uniformly sized smaller amylose-stabilized CNT clusters. The concentration of SWCNTs in the suspension was negatively correlated with the effective size of SWCNT. The notable apparent exception was the suspension of SWCNTs dispersed using GA. However, it should be noted that the aqueous suspension of GA contained GA colloids of the size that was 1) comparable to the size of dispersed SWCNTs and 2) decreased during the 4 weeks of settling (see SD, Fig S5). Thus, the effective size measurements for SWCNTs dispersed using GA should be interpreted with caution.

4.3.

1

0

b 4.2.

1.2

NOM 7

GA

Amylose

PVP

Triton

1.2 1 0.8 0.6 0.4 0.2 0 NOM 3.5 NOM 7

GA

Amylose PVP

Triton

Fig. 3 – Viability of E. coli in (a) 0.1 M NaCl, (b) LB medium, and (c) MD medium. Exposure time: (a) 3 h, (b) 48 h, (c) 48 h. Open, hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose ultrapure water was used as control.

measurement for SWCNT suspensions was accompanied by a measurement of the z-potential of the dispersants in a SWCNT-free aqueous solution (Fig. 2d). After 1 d, only the GA solution scattered light sufficiently to enable a statistically meaningful z-potential measurement. After 4 d, solutions of NOM (pH 7), PVP, and Triton X-100 also scattered light

515

water research 44 (2010) 505–520

1.2

b Fraction of control (FOC)

Fraction of control (FOC)

a

1 0.8 0.6 0.4 0.2

1.2 1 0.8 0.6 0.4 0.2 0

0 NOM3.5 NOM7

GA

Amylose

PVP

Triton

NOM3.5

NOM7

GA Amylose

PVP Triton

Fig. 4 – Neutral red dye uptake by WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open, hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I and control II were ultrapure water.

sufficiently. After 7 d, all the dispersants except for NOM (pH 3.5) achieved acceptable count rates during z-potential measurements. The gradual change in the scattering ability of the dispersant solutions may correspond to the aggregation of the dispersant molecules. The dispersants were ranked in order from the most to the least negative z-potential of dispersed SWCNTs: NOM (pH 7) > NOM (pH 3.5) > GA z Triton X-100 > PVP z amylose. The highly negative charge of NOMstabilized SWCNTs was likely due to the charge of NOM. The low stability of amylose-dispersed SWCNTs could be attributed to the combination of low surface charge and steric repulsion between stabilized SWCNTs in the dispersion.

4.4. Assessment of cell toxicity in prokaryotic and eukaryotic systems 4.4.1.

General viability bioassay with E. coli (prokaryotic)

Immediately after E. coli cells were exposed to SWCNTs suspended in growth medium as well as after 3 h of exposure, no SWCNT aggregation was visually observed (see SD, Table S2) in experiments with all three types of the media – 0.1 M NaCl, MD (medium with lower salt organics content) and LB (medium with higher salt and organics content). After 24 h of incubation, limited precipitation was observed for all SWCNT

suspensions in LB and MD media. After 48 h of incubation more precipitation occurred in each type of media. No inhibition of E. coli colony forming ability was observed after 3 h of incubation with amylose-, NOM-, GA- and PVPstabilized SWCNTs in 0.1 M NaCl (Fig. 3a). The FOC values did not decline in any of these samples and no significant difference in the influence on CFU counts was found between dispersed SWCNTs and solutions of corresponding dispersants. When the bacteria suspension was brought in contact with Triton X-100-stabilized SWCNTs, approx. 25% loss of the cell viability was measured as compared to vehicle control I samples. However, there was no statistical difference between FOC of Triton-stabilized SWCNTs and the control (solution of Triton X-100 only). It remains unclear if the observed cytotoxicity of the these suspensions was due to i) SWCNTs stabilized by Triton X-100 or ii) the residual ‘‘free’’ (i.e. not associated with suspended SWCNTs) Triton X-100 potentially present in the solution or iii) the combined effect of both Triton X-100/SWCNT and dissolved Triton X-100. The complete separation of the Triton X-100/SWCNT and dissolved Triton X-100 could not be accomplished using centrifugation. Even for very long centrifugation times, the supernatant had grayish color indicating that some fraction of SWCNTs was not removed.

Fig. 5 – Representative phase contrast images of WB-F344 cells incubated with 500 mL of H2O (control I), 500 mL of Triton X-100 (control II) and 500 mL of Triton X-100-stabilized SWCNTs. Black dots observed in Control II and Triton X-100-solubilized SWCNTs samples correspond to dead WB-F344 cells. All images were taken at 2003 magnification. Scale bar is 50 mm.

516

water research 44 (2010) 505–520

Fig. 6 – Representative phase contrast (upper row) and UV epifluorescent (bottom row) images of WB-F344 cells incubated with 500 mL of H2O (control I), 500 mL of NOM pH 3.5 (control II) and 500 mL of NOM-stabilized SWCNTs (pH 4). All images were taken at 2003 magnification. Bright dots correspond to cells that absorb the dye; the absorption is indicative of cellular health. All images were taken at 2003 magnification. Scale bar is 50 mm.

1.2 1 0.8 0.6 0.4 0.2

plates. As the exposure time increased to 48 h, E. coli resumed its growth; with FOC of both Triton X-100 suspensions approaching values for vehicle control I sample (Fig. 3b). When dispersed SWCNTs were tested in MD medium, no losses in cell viability for amylose-, GA-, PVP- and NOM-stabilized SWCNTs were measured (Fig. 3c; also see SD, Fig. S7). In the case of Triton X-100 suspensions, the reduction in E. coli survival was observed after 3 h of exposure; comparable losses of the E. coli viability were also observed for both Triton X-100-stabilized SWCNTs and the Triton X-100 only solution. However, after 24 h of exposure, fewer E. coli colonies were

b

1.2

Fraction of control (FOC)

a Fraction of control (FOC)

The ability of E. coli to grow and to form colonies in the presence of amylose-, GA-, PVP-, and NOM-stabilized SWCNTs in LB medium mimicked both control samples regardless of the contact time (see SD, Fig. 3b and Fig. S6). On the contrary, 21 and 18 % mortality rates after 3 h of incubation were observed for E. coli in Triton X-100/SWCNTs suspension and Triton X-100 solution, respectively. After 24 h of contact, the number of colonies grown on the Petri plates decreased by 30 % when bacteria were in contact with Triton X-100-stabilized SWCNTs and by 27 % when bacteria were in Triton X-100 only solution (see SD, Fig. S6) as compared to vehicle control I

1 0.8 0.6 0.4 0.2 0

0 NOM 3.5

NOM 7

GA

Amylose

PVP

NOM 3.5

NOM 7

GA

Amylose

PVP

Fig. 7 – GJIC in WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open, hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I and control II were ultrapure water.

water research 44 (2010) 505–520

observed on Petri plates from suspensions of Triton X-100stabilized SWCNTs and Triton X-100 solution only (vehicle control II) (33 and 44 %, respectively) (see SD, Fig. S7). As the incubation time increased to 48 h, bacterial CFU counts in the presence of both Triton X-100 samples increased and approached that of vehicle control I samples (Fig. 3c). As was previously mentioned, the exposure of Triton X-100-stabilized SWCNTs to MD medium for 24 h resulted in formation of large flake-like aggregates, which settled out of the suspension. It is possible that this low salt medium induced specific interactions between bacterial cells and Triton X-100-coated SWCNTs, which were suppressed in the medium having a higher ionic strength. This led to a larger decrease in E. coli viability in these samples in comparison with the losses in Triton X-100 solution only. Damaged or dead cell attached to the SWCNTs aggregates and formed debris at the bottom of the testing tubes. The simultaneous effect of bacterial and nanoparticles settling out of the solution has been previously described (Tang et al., 2007) where the response of Shewanella oneidensis to C60–NH2 nanoparticles was studied. The fact that we observed inhibition of E. coli viability after 24 h of exposure in MD medium and did not observed this effect in LB medium highlights the importance of growth medium in cytotoxicity testing. Similarly, E. coli exposure to SWCNTs, stabilized with amylose, GA, PVP and NOM (both pH) did not affect CFU counts in any of the three media pointing to the importance of the dispersant in the assessments of biocompatibility of CNTs.

4.4.2.

NDU cytotoxicity bioassay on eukaryotic cells

The assessment of cytotoxicity was determined by the viable uptake of neutral red into F344-WB rat liver epithelial cells. Nonviable cells lack the ability to absorb this dye. Results (Fig. 4) indicated that SWCNTs dispersed using NOM (pH 3.5), NOM (pH 7), GA, amylose and PVP were not cytotoxic. However, Triton X-100 detergent induced a significant cytotoxic effect independent of SWCNTs. This latter observation can also be seen under phase contrast microscopy of cells treated with Triton X-100 for 30 min (Fig. 5). The morphology of the cells drastically changed from that of the normal control cells, in which the cells became clearer due to loss of most cytosolic contents, except for the cell nuclei. After a 24 h treatment with Triton X-100, the cells completely detached or solubilized from the culture plates (data not shown).

4.5. Assessment of epigenetic toxicity on eukaryotic cells: GJIC bioassay A representative fluorescent micrograph of the GJIC assay after treatment with (NOM þ SWCNTs) is shown in Fig. 6. The migration of the fluorescent dye through several cell layers is an indicator that the gap junction channels are open. After WB-F344 cells were incubated with dispersed SWCNTs suspensions for (1) 30 min and (2) 24 h, there was no significant affect on GJIC for all suspensions tested (Fig. 7). Due to Triton X-100-stabilized SWCNTs being cytotoxic, GJIC was not measured in cells exposed to this mixture. Also, when cells were examined under phase

517

contrast, no changes in size and shape of the individual cells were detected (Fig. 6). Hence, it can be concluded that under these experimental conditions WB-F344 cells exposed to dispersed SWCNTs retain normal intercellular communication irrespective of the applied treatment.

5.

Conclusions

The stability of aqueous suspensions of SWCNTs non-covalently functionalized by a range of natural (gum arabic, amylose, Suwannee River natural organic matter) and synthetic (polyvinyl pyrrolidone, Triton X-100) dispersants that bind to SWCNT surface via different physiosorption mechanisms was evaluated. Despite the differences, all SWCNT suspensions remained relatively stable over a period of 4 weeks as indicated by the evolution of the concentration of suspended exfoliated SWCNT and by the fact that both the average effective hydrodynamic diameter and z-potential of suspended SWCNTs were relatively constant. The epigenetic toxicity of dispersed SWCNTs was evaluated for the first time using a gap junctional intercellular communication assay with WB-F344 rat liver epithelial cells resulting in no inhibition by SWCNTs non-covalently functionalized using GA, PVP, amylose, and SRNOM. There were no cytotoxic effects of these SWCNTs suspensions on either the prokaryotic, bacterial (E. coli), or eukaryotic (WB-F344 rat liver epithelial) cell types. Only when the dispersant itself was toxic, were losses of cell viability observed. Bacterial CFU counts or neutral red uptake was affected only by SWCNTs dispersed using Triton X-100, which showed cytotoxicity in the SWCNT-free solution (vehicle control II). The results suggest a strong dependence of the toxicity of SWCNT suspensions on the toxicity of the solubilizing agent and point to the potential of non-covalent functionalization with non-toxic dispersants as a method for the preparation of aqueous suspensions of biocompatible CNTs.

Acknowledgements This work was supported by the National Science Foundation research grants CBET-0608320, CBET-056828 and OISE0530174, by the National Institute of Environmental Health Sciences grants R01 ES013268-01A2 and Grant-in-Aid for Science Research. We thank Michael Housinger for his assistance with the preparation of SWCNT suspensions. ARR acknowledges the support by Michigan AWWA Fellowship for Water Quality and Treatment Study. The contents of this work are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute of Environmental Health Sciences.

Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at doi: doi:10.1016/j.watres.2009.09.042.

518

water research 44 (2010) 505–520

references

Atlas, R.M., 1993. Handbook of Microbiological Media. CRC Press, Boka Raton. Bahr, J.L., Mickelson, E.D., Bronikovski, M.J., Smalley, R.E., Tour, J.M., 2001. Dissolution of small diameter single-wall carbon nanotubes in organic solvents. Chemical Communications, 193–194. Bandyopadhyaya, R., Nativ-Roth, E., Regev, O., YerushalmiRozen, R., 2002. Stabilization of individual carbon nanotubes in aqueous solutions. Nano Letters 2 (1), 25–28. Bardi, G., Tognini, P., Ciofani, G., Raffa, V., Costa, M., Pizzorusso, T., 2009. Pluronic-coated carbon nanotubes do not induce degeneration of cortical neurons in vivo and in vitro. Nanomedicine: Nanotechnology, Biology, and Medicine 5, 96–104. Bonnet, P., Albertini, D., Bizot, H., Bernard, A., Chauvet, O., 2007. Amylose/SWNT composites: from solution to film – synthesis, characterization and properties. Composite Science and Technology 67, 817–821. Borenfreund, E., Puerner, J.A., 1985. Toxicology determined in vitro by morphological alterations and neutral red absorption. Toxicology Letters 24, 119–124. Burnette, L.W., 1960. A review of the physiological properties of PVP. Proceedings of the Scientific Section of the Toilet Goods Association 38, 1–4. Chappell, M.A., George, A.J., Dontsova, K.M., Porter, B.E., Price, C.L., Zhou, P., Morikawa, E., Kennedy, A.J., Steevens, J.A., 2009. Surfactive stabilization of multi-walled carbon nanotube dispersions with dissolved humic substances. Environmental Pollution 157, 1081–1087. Chen, X., Tam, U.C., Czlapinski, J.L., Lee, G.S., Rabuka, D., Zettl, A., Bertozzi, C.R., 2006. Interfacing carbon nanotubes with living cells. Journal of American Chemical Society 128, 6292–6293. Cherukuri, P., Gannon, C.J., Leeuw, T.K., Schmidt, H.K., Smalley, R.E., Curley, S.A., Weisman, R.B., 2006. Mammalian pharmacokinetics of carbon nanotubes using intrinsic nearinfrared fluorescence. Proceedings of the National Academy of Sciences 103, 18882–18886. Chin, S.F., Baughman, R.H., Dalton, A.B., Diekmann, G.R., Draper, R.K., Mikoryak, C., Musselman, I.H., Poenitzsch, V.Z., Xie, H., Pantano, P., 2007. Amphiphilic helical peptide enhances the uptake of single-walled carbon nanotubes by living cells. Experimental Biological Medicine 232 (9), 1236–1244. Chourasia, M.K., Jain, S.K., 2004. Polysaccharides for colon targeted drug delivery. Drug Delivery 11 (2), 129–148. Dayeh, V.R., Chow, S.L., Schirmer, K., Lynn, D.H., Bols, N.C., 2004. Evaluating the toxicity of Triton X-100 to protozoan, fish, and mammalian cells using fluorescent dyes as indicators of cell viability. Ecotoxicology and Environmental Safety 57 (3), 375–382. Di Sotto, A., Chiaretti, M., Carru, G.A., Bellucci, S., Mazzantia, G., 2009. Multi-walled carbon nanotubes: lack of mutagenic activity in the bacterial reverse mutation assay. Toxicology Letters 184, 192–197. Didenko, V.V., Moore, V.C., Baskin, D.S., Smalley, R.E., 2005. Visualization of individual single-walled carbon nanotubes by fluorescent polymer wrapping. Nano Letters 5 (8), 1563–1567. Ding, L.H., Stilwell, J., Zhang, T.T., Elboudware, J.O., Jiang, H.J., Selegue, J.P., Cooke, P.A., Gray, J.W., Chen, F.F., 2005. Molecular characterization of the cytotoxic mechanism of multiwall carbon nanotubes and nanoonions on human skin fibroblast. Nano Letters 5, 2448–2464. Dodziuk, H., Ejchart, A., Anczewski, W., Ueda, H., Krinichnaya, E., Dolgonosa, G., Kutner, W., 2003. Water solubilization,

determination of the number of different types of single-wall carbon nanotubes and their partial separation with respect to diameters by complexation with h-cyclodextrin. Chemical Communications, 986–987. El-Fouly, M.H., Trosko, J.E., Chang, C.C., 1987. Scrape-loading and dye transfer: a rapid and simple technique to study gap junctional intercellular communications. Experimental Cell Research 168, 422–430. Elias, A.L., Carrero-Sanchez, J.C., Terrones, H., Endo, M., Laclette, H.P., Mauricio, M., 2007. Viability studies of pure carbon- and nitrogen-doped nanotubes with Entamoeba Histolytica: from amoebicidal to biocompatible structures. Small 3 (10), 1723–1729. Enyashin, A.N., Gemming, S., Seifert, G., 2007. DNA-wrapped carbon nanotubes. Nanotechnology 18, 245702. Fiorito, S., Serafino, A., Andreola, F., Bernier, P., 2006. Effects of fullerenes and single-wall carbone nanotubes on murine and human macrophages. Carbon 44, 1100–1105. Flahaut, E., Durrieu, M.C., Remy-Zolghadri, M., Bareille, R., Baquey, C., 2006. Investigation of the cytotoxicity of CCVD carbon nanotubes towards human umbilical vein endothelial cells. Carbon 44 (6), 1093–1099. Fortner, J.D., Lyon, D.Y., Sayes, C.M., Boyd, A.M., Falkner, J.C., Hotze, E.M., Alemany, L.B., Tao, Y.J., Guo, W., Ausman, K.D., Colvin, V.L., Hughes, J.B., 2005. C60 in water: nanocrystal formation and microbial response. Environmental Science and Technology 39 (11), 4307–4316. Georgakilas, V., Tagmatarchis, N., Pantarotto, D., Bianco, A., Briand, J.-P., Prato, M., 2002. Amino acid functionalisation of water soluble carbon nanotubes. Chemical Communications, 3050–3051. Grossiord, N., Schhoo, P., Meuldijk, J., Koning, C.E., 2007. Determination of the surface coverage of exfoliated carbon nanotubes by surfactant molecules in aqueous solution. Langmuir 23, 3646–3653. Gutierrez, M.C., Garcia-Carvajal, Z.Y., Hortiguela, M.J., Yuste, L., Rojo, F., Ferrer, M.L., Monte, F., 2007. Biocompatible MWCNT scaffolds for immobilization and proliferation of E. coli. Journal of Materials Chemistry 17, 2992–2995. Herner, H.A., Trosko, J.E., Masten, S.J., 2001. The epigenetic toxicity of pyrene and related ozonation byproducts containing an aldehyde functional group. Environmental Science and Technology 35 (17), 3576–3583. Herzog, E., Byrne, H.J., Davoren, M., Casey, A., Duschl, A., Oostingh, G.J., 2009. Dispersion medium modulates oxidative stress response of human lung epithelial cells upon exposure to carbon nanomaterial samples. Toxicology and Applied Pharmacology 236, 276–281. Huang, W.J., Taylor, S., Fu, K.F., Lin, Y., Zhang, D.H., Hanks, T.W., Rao, A.M., Sun, Y.P., 2002. Attaching proteins to carbon nanotubes via diimide-activated amidation. Nano Letters 2 (4), 311–314. Humic acids and their sodium salts. EMEA-The European agency for the evaluation of medicinal products. Available via. http:// www.emea.europa.eu/pdfs/vet/mrls/055499en.pdf, 1999. Hyung, H., Fortner, J.D., Hughes, J.B., Kim, J.H., 2007. Natural organic matter stabilizes carbon nanotubes in the aqueous phase. Environmental Science and Technology 41 (1), 179–184. Hyung, H., Kim, J.-H., 2008. Natural organic matter (NOM) adsorption to multi-walled carbon nanotubes: effect of NOM characteristics and water quality parameters. Environmental Science and Technology 42 (12), 4416–4421. Iijima, S., 1991. Helical microtubules of graphitic carbon. Nature 354 (6348), 56–58. Inoue, K.-i., Koike, E., Yanagisawa, R., Hirano, S., Nishikawa, M., Takano, H., 2009. Effects of multi-walled carbon nanotubes on a murine allergic airway inflammation model. Toxicology and Applied Pharmacology 237 (3), 306–316.

water research 44 (2010) 505–520

Islam, M.F., Rojas, E., Bergey, D.M., Johnson, A.T., Yodh, A.G., 2003. High weight fraction surfactant solubilization of single-wall carbon nanotubes in water. Nano Letters 3 (2), 269–273. Jia, G., Wang, H., Yan, L., Wang, X., Pei, R., Yan, T., Zhao, Y., Guo, X., 2005. Cytotoxicity of carbon nanomaterials: singlewall nanotube, multi-wall nanotube, and fullerene. Environmental Science and Technology 39, 1378–1383. Jiang, J., Gao, L., 2003. Production of aqueous colloidal dispersions of carbon nanotubes. Journal of Colloid and Interface Science 260, 89–94. Kang, S., Herzberg, M., Rodrigues, D.F., Elimelech, M., 2008a. Antibacterial effects of carbon nanotubes: size does matter! Langmuir 24 (13), 6409–6413. Kang, S., Mauter, M.S., Elimelech, M., 2008b. Physicochemical determinants of multiwalled carbon nanotube bacterial cytotoxicity. Environmental Science and Technology 42 (19), 7528–7534. Kang, S., Mauter, M.S., Elimelech, M., 2009. Microbial cytotoxicity of carbon-based nanomaterials: implications for river water and wastewater effluent. Environmental Science and Technology 43 (7), 2648–2653. Kang, S., Pinault, M., Pfefferle, L.D., Elimelech, M., 2007. Singlewalled carbon nanotubes exhibit strong antimicrobial activity. Langmuir 23 (17), 8670–8673. Karajanagi, S.S., Yang, H.C., Asuri, P., Sellitto, E., Dordick, J.S., Kane, R.S., 2006. Protein-assisted solubilization of singlewalled carbon nanotubes. Langmuir 22 (4), 1392–1395. Kim, O.-K., Je, J., Baldwin, J.W., Kooi, S., Pehrsson, P.E., Buckley, L.J., 2004. Solubilization of single-wall carbon nanotubes by supramolecular encapsulation of helical amylose. Journal of American Chemical Society 125, 4426–4427. Lauret, J.-S., Voisin, C., Cassabois, G., Roussignol, P., Delalande, C., Capes, L.E., Valentin, E., Jost., O., 2004. Photocreated carrier dynamics in isolated carbon nanotubes. Semiconductor Science and Technology 19, S486–S488. Lee, J.U., Huh, J., Kim, K.H., Park, C., Jo, W.H., 2007. Aqueous suspension of carbon nanotubes via non-covalent functionalization with oligothiophene-terminated poly(ethylene glycol). Carbon 45, 1051–1057. Lin, Y., Taylor, S., Li, H.P., Fernando, K.A.S., Qu, L.W., Wang, W., Gu, L.R., Zhou, B., Sun, Y.P., 2004. Advances toward bioapplications of carbon nanotubes. Journal of Materials Chemistry 14 (4), 527–541. Lindberg, H.K., Falck, G.C.-M., Suhonen, S., Vippola, M., Vanhalab, E., Catala´na, J., Savolainena, K., Norppa, H., 2009. Genotoxicity of nanomaterials: DNA damage and micronuclei induced by carbon nanotubes and graphite nanofibres in human bronchial epithelial cells in vitro. Toxicology Letters 186, 166–173. Liu, J., Rinzler, A.G., Dai, H.J., Hafner, J.H., Bradley, R.K., Boul, P.J., Lu, A., Iverson, T., Shelimov, K., Huffman, C.B., RodriguezMacias, F., Shon, Y.S., Lee, T.R., Colbert, D.T., Smalley, R.E., 1998. Fullerene pipes. Science 280 (5367), 1253–1256. Liu, Y.Q., Gao, L., Zheng, S., 2007. Debundling of single-walled carbon nanotubes by using natural polyelectrolytes. Nanotechnology 18 (36), 365702. Lyon, D.Y., Fortner, J.D., Sayes, C.M., Colvin, V.L., Hughes, J.B., 2005. Bacterial cell association and antimicrobial activity of a C60 water suspension. Environmental Toxicology and Chemistry 24, 2757–2762. Manna, S.K., Sarkar, S., Barr, J., Wise, K., Barrera, E.V., Jejelowo, O., Rice-Ficht, A., Ramesh, G.T., 2005. Single-walled carbon nanotube induces oxidative stress and activates nuclear transcription factor-kB in human keratinocytes. Nano Letters 5, 1676–1684. Marsh, D.H., Rance, G.A., Zaka, M.H., Whitby, R.J., Khlobystov, A.N., 2007. Comparison of the stability of multiwalled carbon

519

nanotube dispersions in water. Physical Chemistry Chemical Physics 9 (40), 5490–5496. McDonald, T.J., Engtrakul, C., Jones, M., Rumbles, G., Heben, M.J., 2006. Kinetics of PL quenching during single-walled carbon nanotube rebundling and diameter-dependent surfactant interactions. Journal of Physical Chemistry B 110, 25339–25346. Meng, J., Yanga, M., Song, L., Kong, H., Wang, C.Y., Wang, R., Wang, C., Xie, S.S., Xua, H.Y., 2009. Concentration control of carbon nanotubes in aqueous solution and its influence on the growth behavior of fibroblasts. Colloids and Surfaces B: Biointerfaces 71, 148–153. Moore, V.C., Strano, M.S., Haroz, E.H., Hauge, R.H., Smalley, R.E., Schmidt, J., Talmon, Y., 2003. Individually suspended singlewalled carbon nanotubes in various surfactants. Nano Letters 3 (10), 1379–1382. Muller, J., Huaux, F., Moreau, N., Misson, P., Heilier, J.-F., Delos, M., Arras, M., Fonseca, A., Nagy, J.B., Lison, D., 2005. Respiratory toxicity of multi-wall carbon nanotubes. Toxicology and Applied Pharmacology 207 (31), 221–231. Narayan, R.J., Berry, C.J., Brigmon, R.L., 2005. Structural and biological properties of carbon nanotube composite films. Materials Science and Engineering B 123, 123–129. O’Connell, M.J., Bachilo, S.M., Huffman, C.B., Moore, V.C., Strano, M.S., Haroz, E.H., Rialon, K.L., Boul, P.J., Noon, W.H., Kittrell, C., Ma, J., Hauge, R.H., Weisman, R.B., Smalley, R.E., 2002. Band gap fluorescence from individual single-walled carbon nanotubes. Science 297, 593–596. O’Connell, M.J., Boul, P.J., Ericson, L.M., Huffman, C., Wang, Y., Haroz, E., 2001. Reversible water-solubilization of singlewalled carbon nanotubes by polymer wrapping. Chemical Physics Letters 342, 265–271. Park, H.J., Heo, H.Y., Lee, S.C., Park, M., Lee, S.-S., Kim, J., Chang, J.Y., 2006. Dispersion of single-walled carbon nanotubes in water with polyphosphazene polyelectrolyte. Journal of Inorganic and Organometallic Polymers and Materials 16 (4) 359–364. Pulskamp, K., Worle-Knirsch, J.M., Hennrich, F., Kern, K., Krug, H.F., 2007. Human lung epithelial cells show biphasic oxidative burst after single-walled carbon nanotube contact. Carbon 45, 2241– 2249. Saleh, N.D., Pfefferle, L.D., Elimelech, M., 2009. Aggregation kinetics of multiwalled carbon nanotubes in aquatic systems: measurements and environmental implications. Environmental Science and Technology 42 (21), 7963–7969. Salzmann, C.G., Chu, B.T.T., Tobias, G., Llewellyn, S.A., Green, M.L.H., 2007. Quantitative assessment of carbon nanotube dispersions by Raman spectroscopy. Carbon 45 (5), 907–912. Satoh, A.Y., Trosco, J., Masten, S.J., 2003. Epigenetic toxicity of hydroxylated biphenyls and hydroxylated polychlorinated biphenyls on normal rat liver epithelial cells. Environmental Science and Technology 37, 2727–2733. Sayes, C.M., Liang, F., Hudsona, J.L., Mendeza, J., Guob, W., Beach, J.M., Moore, V.C., Doyle, C.D., West, J.L., Billups, W.E., Ausmanb, K.D., Colvin, V.L., 2006. Functionalization density dependence of single-walled carbon nanotubes cytotoxicity in vitro. Toxicology Letters 161, 135–142. Schmitt, D., Tran, N., Riefler, S., Jacoby, J., Merkel, D., Marone, P., Naouli, N., 2008. Toxicologic evaluation of modified gum acacia: mutagenicity, acute and subchronic toxicity. Food and Chemical Toxicology 46, 1048–1054. Simon-Deckers, A., Gouget, B., Mayne-L’Hermiteb, M., HerlinBoimeb, N., Reynaudb, C., Carrie`rea, M., 2008. In vitro investigation of oxide nanoparticle and carbon nanotube toxicity and intracellular accumulation in A549 human pneumocytes. Toxicology 253, 137–146. Sinani, V.A., Gheith, M.K., Yaroslavov, A.A., Rakhnyanskaya, A.A., Sun, K., Mamedov, A.A., Wicksted, J.P., Kotov, N.A., 2005.

520

water research 44 (2010) 505–520

Aqueous dispersions of single-wall and multiwall carbon nanotubes with designed amphiphilic polycations. Journal of American Chemical Society 127 (10), 3463–3472. Star, A., Steuerman, D.W., Heath, J.R., Stoddart, J.F., 2002. Starched carbon nanotubes. Angewandte Chemie International Edition 41 (14), 2508–2512. Tang, Y.J., Ashcroft, J.M., Chen, D., Min, G., Kim, C.-H., Murkhejee, B., Larabell, C., Keasling, J.D., Chen, F.F., 2007. Charge-associated effects of fullerene derivatives on microbial structural integrity and central metabolism. Nano Letters 7 (3), 754–760. Tasis, D., Tagmatarchis, N., Bianco, A. and Prato, M., 2006. Chemistry of carbon nanotubes. 106 (3), 1105–1136. Tong, Z., Bischoff, M., Nies, L., Applegate, B., Turco, R., 2007. Impact of fullerene (C60) on a soil microbial community. Environmental Science and Technology 41, 2985–2991. Trosko, J.E., Chang, B.V., Madhucar, J.E., Klauning, J.E., 1991. Chemical, oncogene, and growth factor inhibition of gap junctional intercellular communication: an alternative hypothesis to carcinogenesis. Pathobiology 58, 265–278. Trosko, J.E., Chang, C.-C., Upham, B., Wilson, M., 1998. Epigenetic toxicology as toxicant-induced changes in intracellular signalling leading to altered gap junctional intercellular communication. Toxicology Letters 102–103, 71–78. Tseng, C.-H., Wang, C.-C., Chen, C.-Y., 2006. Modification of multi-walled carbon nanotubes by plasma treatment and further use as templates for growth of CdS nanocrystals. Nanotechnology 17, 5602–5612. Upham, B.L., Masten, S.J., Lochwood, B.R., Trosko, J.E., 1994. Nongenotoxic effects of polycyclic aromatic hydrocarbons and their ozonation by-products on the intercellular communication of rat liver epithelial cells. Fundamental and Applied Toxicology 23, 470–475. Vaisman, L., Marom, G., Wagner, H.D., 2006. Dispersions of surface-modified carbon nanotubes in water-soluble and water-insoluble polymers. Advanced Functional Materials 16 (3), 357–363. Vittorio, O., Raffa, V. and Cuschieri, A., Influence of purity and surface oxidation on cytotoxicity of multi-wall carbon nanotubes with human neuroblastoma cells. Nanomedicine: Nanotechnology, Biology, and Medicine, in press, doi:10.1016/j.nano.2009.02.006. Wang, Y., Iqbal, Z., Malhotra, S.V., 2005. Functionalization of carbon nanotubes with amines and enzymes. Chemical Physics Letters 402, 96–101. Weis, L.M., Rummel, A.M., Masten, S.J., Trosko, J.E., Upham, B.L., 1998. Bay or baylike regions of polycyclic aromatic hydrocarbons were potent inhibitors of gap junctional

intercellular communication. Environmental Health Perspectives 106 (1), 17–22. Wick, P., Manser, P., Limbach, L.K., Dettlaff-Weglikowskab, U., Krumeich, F., Roth, S., Stark, W.J., Bruinink, A., 2007. The degree and kind of agglomeration affect carbon nanotube cytotoxicity. Toxicology Letters 168, 121–131. Wirnitzer, U., Herbold, B., Voetz, M., Ragot, J., 2009. Studies on the in vitro genotoxicity of baytubes, agglomerates of engineered multi-walled carbon-nanotubes (MWCNT). Toxicology Letters 186, 160–165. Wo¨rle-Knirsch, J.M., Pulskamp, K., Krug, H.F., 2006. Oops they did it again! Carbon nanotubes hoax scientists in viability assays. Nano Letters 6 (6), 1261–1268. Xu, F.-M., Xu, J.-P., Ji, J., Shen, J.-C., 2008. A novel biomimetic polymer as amphiphilic surfactant for soluble and biocompatible carbon nanotubes (CNTs). Colloids and Surfaces B: Biointerfaces 67, 67–72. Yamasaki, Y., 1990. Gap junctional intercellular communication and carcinogenesis. Carcinogenesis 11, 1051–1058. Yang, H., Liu, C., Yang, D., Zhanga, H., Xia, Z., 2009. Comparative study of cytotoxicity, oxidative stress and genotoxicity induced by four typical nanomaterials: the role of particle size, shape and composition. Journal of Applied Toxicology 29, 69–78. Yang, S.-T., Wang, X., Jia, G., Guc, Y., Wang, T., Nie, H., Ge, C., Wang, H., Liu, Y., 2008. Long-term accumulation and low toxicity of single-walled carbon nanotubes in intravenously exposed mice. Toxicology Letters 181, 182–189. Yang, Z., Chen, X.H., Chen, C.S., 2007. Noncovalent-wrapped sidewall multi-walled carbon nanotubes functionalization with polyimide. Polymer Composites 28 (1), 36–41. Ye, S.-F., Wu, Y.-H., Hou, Z.-Q., Zhang, Q.-Q., 2009. ROS and NF-jB are involved in upregulation of IL-8 in A549 cells exposed to multi-walled carbon nanotubes. Biochemical and Biophysical Research Communications 379, 643–648. Yu, B.Z., Yang, J.S., Li, W.X., 2007. In vitro capability of multi-walled carbon nanotubes modified with gonadotrophin releasing hormone on killing cancer cells. Carbon 45 (10), 1921–1927. Yun, Y., Dong, Z., Tan, Z., Schulz, M.J., Shanov, V., 2009. Fibroblast cell behavior on chemically functionalized carbon nanomaterials. Materials Science and Engineering C 29, 719–725. Zhu, Y., Ran, T., Li, Y., Guo, J., Li, W., 2006. Dependence of the cytotoxicity of multi-walled carbon nanotubes on the culture medium. Nanotechnology 16, 4668–4674. Zong, S., Cao, Y., Jua, H., 2007. Direct electron transfer of hemoglobin immobilized in multiwalled carbon nanotubes enhanced grafted collagen matrix for electrocatalytic detection of hydrogen peroxide. Electroanalysis 19 (7–8), 841–846.

water research 44 (2010) 521–532

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Oxidation of atenolol, propranolol, carbamazepine and clofibric acid by a biological Fenton-like system mediated by the white-rot fungus Trametes versicolor Ernest Marco-Urrea a, Jelena Radjenovic´ b, Gloria Caminal c, Mira Petrovic´ b,d, Teresa Vicent a,*, Damia` Barcelo´ b,e a

Departament d’Enginyeria Quı´mica and Institut de Cie`ncia i Tecnologia Ambiental, Universitat Auto`noma de Barcelona (UAB), 08193 Bellaterra, Spain b Department of Environmental Chemistry, IDAEA-CSIC, c/Jordi Girona 18–26, 08034 Barcelona, Spain c Unitat de Biocata`lisis Aplicada associada al IQAC (CSIC-UAB), Escola Te`cnica Superior d’Enginyeria, UAB, 08193 Bellaterra, Spain d Institucio´ Catalana de Recerca i Estudis Avanc¸ats (ICREA), Barcelona, Spain e Institut Catala` de Recerca de l’Aigua (ICRA), Parc Cientı´fic i Tecnolo´gic de la Universitat de Girona, Pic de Peguera, 15, 17003 Girona, Spain

article info

abstract

Article history:

Biological advanced oxidation of the pharmaceuticals clofibric acid (CA), carbamazepine

Received 31 March 2009

(CBZP), atenolol (ATL) and propranolol (PPL) is reported for the first time. Extracellular

Received in revised form

oxidizing species were produced through a quinone redox cycling mechanism catalyzed by

16 September 2009

an intracellular quinone reductase and any of the ligninolytic enzymes of Trametes versi-

Accepted 21 September 2009

color after addition of the lignin-derived quinone 2,6-dimethoxy-1,4-benzoquinone (DBQ)

Available online 22 September 2009

and Fe3þ-oxalate in the medium. Time-course experiments with approximately 10 mg L1 of initial pharmaceutical concentration resulted in percent degradations above 80% after

Keywords:

6 h of incubation. Oxidation of pharmaceuticals was only observed under DBQ redox

Trametes versicolor

cycling conditions. A similar degradation pattern was observed when CBZP was added at

Pharmaceuticals

the environmentally relevant concentration of 50 mg L1. Depletion of DBQ due to the attack

Hydroxyl radical

of oxidizing agents was assumed to be the main limiting factor of pharmaceutical degra-

Carbamazepine

dation. The main degradation products, that resulted to be pharmaceutical hydroxylated

Beta-blockers

derivatives, were structurally elucidated. The detected 4- and 7-hydroxycarbamazepine

Clofibric acid

intermediates of CBZP degradation were not reported to date. Total disappearance of intermediates was observed in all the experiments at the end of the incubation period. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

The presence of pharmaceuticals and personal care products (PPCPs) and their metabolites in waters has become an emerging environmental issue. To date, most attention has been focused on identification, fate and distribution of PPCPs

in municipal wastewater treatment plants (WWTPs), which are commonly found at very low concentrations (low ppb levels) (Radjenovic et al., 2007, 2009, Reemtsma et al., 2006). To avoid their potential adverse health effects on both humans and environment, research efforts are underway to develop efficient techniques for their removal. Among the alternatives

* Corresponding author. Departament d’Enginyeria Quı´mica and Institut de Cie`ncia i Tecnologia Ambiental. Universitat Auto`noma de Barcelona (UAB). 08193 Bellaterra, Spain. Tel.: þ34 935812142; fax: þ34 935812013. E-mail address: [email protected] (T. Vicent). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.049

522

water research 44 (2010) 521–532

to traditional water treatment processes, advanced oxidation processes (AOPs) have received great attention to degrade PPCPs (Klavarioti et al., 2009; Song et al., 2008). AOPs can be broadly described as aqueous phase oxidation methods based on the intemediacy of highly reactive species such as hydroxyl radicals (,OH). This radical acts as a non-selective, very strong oxidant agent with the ability to react with chemicals giving dehydrogenated or hydroxylated derivates, until achieving their total mineralization. To date, the production of ,OH radicals was based on chemical, photochemical, photocatalytical or electrochemical techniques, but little is known about the possibility to use biological systems to produce ,OH radicals for pollutant degradation (Klavarioti et al., 2009). This biological approach was proposed recently for several whiterot fungi (WRF) including Trametes versicolor and Pleurotus eringyi through a simple quinone redox cycling mechanism that is referred to as advanced biooxidation (Go´mez-Toribio et al., 2009a). WRF are basidiomycetes that are capable of extensive aerobic lignin depolymeration and mineralization due to the low substrate specificity and high reactivity of the enzymes they produce (principally laccase and peroxidases). Such combinations of enzymes with auxiliary oxidoreductive enzymes such as cytochrome P450 system are commonly involved in the degradation of several PPCPs as for example carbamazepine, clofibric acid, ibuprofen and several endocrine disrupting chemicals (Cabana et al., 2007; Marco-Urrea et al., 2009a). In this work, the anti-epileptic drug carbamazepine (CBZP), lipid regulator clofibric acid (CA), and b-blockers propranolol (PPL) and atenolol (ATL) were the selected pharmaceuticals due to their long-term use in Europe and North America and their subsequent occurrence in the aquatic environment (Alder et al., 2006). The metabolism of these pharmaceuticals in mammals has been well established. It is known that CBZP undergoes extensive hepatic metabolism by the cytochrome P450 system, with oxidation to 10,11-dihydro-10,11-epoxy carbamazepine, its further hydration to 11-dihydro-10,11epoxy carbamazepine and conjugation of the latter one with glucuronide being the main metabolic pathway of CBZP (Kerr et al., 1994). CA is the pharmacologically active derivative of clofibrate and several other fibrates, and it is metabolized to reactive acylating derivatives that have been shown to transacylate glutathione forming the corresponding glutathione conjugate (Grillo and Benet, 2001). The primary metabolic pathways of PPL are glucuronidation, side-chain oxidation and ring oxidation, with the main metabolites arising from Ndealkylation of the isopropanolamine side-chain, naphthalene ring hydroxylation, and side-chain O-glucuronidation (Luan et al., 2005). ATL is excreted eagerly via urine as an unchanged compound (90%) (Dollery, 1991), with a small percentage of atenolol-glucuronide (0.8–4.4%) and hydroxyatenolol (1.1– 4.4%, hydroxylation of the benzilic position). Limited removal of ATL and CA and no removal of CBZP have been frequently observed in municipal WWTPs and subsequently they deserve more attention due to the high risk of passing through later barriers in partly closed water cycles (Radjenovic et al., 2007; Reemtsma et al., 2006; Ternes, 1998). Joss et al. (2003) found that CBZP was not expected to produce acute toxic effects in the aquatic biota, but chronic and

synergistic effects with other chemicals could not be excluded. According to their results and regarding the present European legislation on the classification and labelling of chemicals (92/32/EEC), they classified CBZP as ‘‘R52/53 Harmful to aquatic organisms and may cause long-term adverse effects in the aquatic environment’’. On the other side, CA has been reported to be a non-hazardous, yet environmentally persistent compound (estimated persistence of 21 years in the environment) (Buser et al., 1998; Ferrari et al., 2003). Nevertheless, the possibility of endocrine disruption activity of CA through interference with cholesterol synthesis was reported (Pfluger and Dietrich, 2001), which is particularly important regarding its long-term effects. Although specific environmental effects of b-blockers are low, in a mixture with other bblockers the effect of concentration addition can occur, as shown in tests with Daphnia magna and phototoxicity assays with green algae (Escher et al., 2006). Considering these facts, removal techniques need to be developed to effectively degrade the selected pharmaceuticals. The objective of this study was to demonstrate for the first time the degradation of the aforementioned pharmaceuticals by biologically induced oxidizing species, produced by the WRF T. versicolor. The advanced biooxidation strategy consists of the incubation of fungi with a lignin-derived quinone (2,6,dimethoxy-1,4-benzoquione, DBQ) and chelated ferric ion (Fe3þ-oxalate). Under these conditions, fungi catalyze the conversion of the quinone into hydroquinone (DBQH2) by an intracellular quinone reductase, and subsequent oxidation of DBQH2 to semiquinone radicals (DBQ,) is performed in the extracellular medium by any of the lignin modifying enzymes of the WRF (laccase and peroxidases). Then, Fenton’s reagent is formed by DBQ, autoxidation catalyzed by Fe3þ, in which Fe2þ and superoxide anion radical (O, 2 ) are generated and Fe2þ(DBQ, þ Fe3þ-oxalate / DBQ þ Fe2þ-oxalate; , , 3þ oxalate þ O2 # Fe -oxalate þ O2 ), followed by O2 dis, þ , mutation (O, 2 þ HO2 þ H / O2 þ H2O2). Production of OH via quinone redox cycling described above was previously proposed by estimating production of 2-thiobarbituric acid reactive substances (TBARS) from 2-deoxyribose and hydroxylation of 4-hydroxybenzoic acid producing 3,4-dimethoxybenzoic acid, showing a high correlation between formation of Fenton’s reagent and TBARS production (Go´mez-Toribio et al., 2009a). The inhibitory effect of ,OH scavengers and catalase on TBARS production rate further strengthened ,OH generation by this mechanism (Go´mez-Toribio et al., 2009a; Guille´n et al., 2000). However, besides production of ,OH radicals, other oxidizing species such as ferryl ion might also be produced in typical Fenton reactions under certain conditions of pH and concentration of organic and inorganic ligands (Hug and Leupin, 2003).Therefore, more research is needed to ascertain the relative contribution of ,OH produced under quinone redox cycling conditions. This quinone redox cycle mechanism applied to WRF increases the range of environmental pollutants susceptible to be degraded by these microorganisms due to the high oxidation power and low substrate specificity of the oxidizing species generated in comparison with their ligninolytic enzymes (Marco-Urrea et al., 2009b). These oxidizing species are produced rapidly in the extracellular medium and almost total destruction of pollutants can be achieved during the first hours

water research 44 (2010) 521–532

of incubations without the need of adaptation of fungi to pollutants (Go´mez-Toribio et al., 2009a, b). Furthermore, degradation metabolites were elucidated, which were further oxidized either by the induced oxidizing agents or by the ligninolytic enzyme system of fungus. Studies concerning degradation of pharmaceuticals by fungi are very scarce, with 1- and 2-hydroxy ibuprofen, and 1,2-dihydroxy ibuprofen being the only fungal metabolites of pharmaceutical reported up to date (Marco-Urrea et al., 2009a). Nevertheless, the knowledge of these metabolites is necessary for safe application of fungi biocatalyst as bioremediation technology, as well as for the elucidation of the key steps of the degradation process.

2.

Materials and methods

2.1.

Chemicals

CA (CAS No. 882-09-7), CBZP (CAS No. 298-46-4), ATL (CAS No. 29122-68-7), PPL (CAS No. 3506-09-0) and DBQ (CAS No. 530-55-2) were obtained from Sigma–Aldrich Co. All other chemicals used were of analytical grade.

2.2.

Fungus and culture conditions

T. versicolor (ATCC#42530) was maintained by subculturing on 2% malt extract agar slants (pH 4.5) at room temperature. Subcultures were routinely made every 30 days. Pellets of T. versicolor were produced by inoculating 1 mL of a mycelial suspension, prepared as described previously (Marco-Urrea et al., 2008), in 1 L Erlenmeyer flask containing 250 mL of malt extract medium. This was shaken (135 rpm, r ¼ 25 mm) at 25 C for 5 days. Subsequent pellets formed by this process were transferred to another 1 L Erlenmeyer flask containing 250 mL of a defined medium described elsewhere (Marco-Urrea et al., 2008) and were also incubated for 2 days in shaking conditions.

2.3.

Degradation experiments

In time-course experiments, induction of oxidizing agents in T. versicolor via quinone redox cycling was routinely performed as follows. Mycelial pellets from each sample flask were collected by filtration and washed three times with MilliQ water. Appropriate amounts of 2-day old washed mycelium pellets were incubated in the reaction mixture (see figure legends). Reaction mixture contained 500 mM DBQ, 100– 300 mM Fe3þ-oxalate, and 100 mM Mn2þ in 25 mL 20 mM phosphate buffer, pH 5, based on previous optimization of ,OH production in quinone redox cycling (Go´mez-Toribio et al., 2009b). Twenty mL of a solution containing the corresponding pharmaceutical in acetonitrile was added into the flasks to give the desired final pharmaceutical concentration (approximately 10 mg L1). The flasks were incubated at 25 oC and 130 rpm and samples were taken at each point for analysis. The samples were filtered through a Millex-GV (Millipore) 0.22 mm filter and subsequently analyzed by HPLC. In order to investigate the degradation of pharmaceuticals present at environmentally relevant concentrations, CBZP was selected as a model compound and tests flasks were

523

amended with CBZP to a 50 mg L1 concentration. Erlenmeyer flasks containing 50 mL of the corresponding reaction mixture were sacrificed at each experiment time and were filtered through 0.45 mm glass fiber filter from Whatman. The target compound was extracted in one step by solid phase extraction with Oasis HLB cartridges (60 mg adsorbent, Waters, Barcelona, Spain) as is described elsewhere (Radjenovic et al., 2007). Briefly, the cartridges were preconditioned sequentially with 5 mL of methanol and 5 mL of deionized water at neutral pH. The cartridge was dried under vacuum and was eluted with two 2-mL portions of methanol and subsequently concentrated to dryness under a gentle nitrogen stream. The extracts were reconstituted with 0.5 mL 25:75 (v/v) acetonitrile-water. To obviate the possible influence of light on pharmaceuticals stability, all the experiments were carried out in the dark. Each experiment included control flasks (without Fe3þoxalate). All the results included in the text and shown in figures are the mean and standard deviation of duplicate experiments.

2.4.

Analytical procedures

2.4.1.

Analysis of pharmaceuticals and DBQ

Analysis of the pharmaceuticals and DBQ were performed using a Dionex 3000 Ultimate HPLC equipped with a UV detector at 230 nm. The column temperature was 30 C and a sample volume of 20 mL was injected from a Dionex autosampler. Chromatographic separation of CBZP, PPL, CA and DBQ was achieved on a GraceSmart RP 18 column (250 mm 4 mm, particle size 5 mm). The mobile phase consisted of 6.9 mmol L1 acetic acid adjusted to pH 4 (by NaOH) with 35% v/v acetonitrile. It was delivered isocratically at 1 mL min1 as was described elsewhere (Stafiej et al., 2007). ATL analysis was performed with a Ascentis C18 column (150 mm 4.6 mm, particle size 5 mm). Mobile phase of 0.01 M ammonium acetate (pH 7) and mobile phase B (acetonitrile) were delivered at flow rate of 1.2 mL min1 and used for gradient elution of ATL (t ¼ 0 min A ¼ 95%, t ¼ 20 min A ¼ 80%). The detection limit was calculated to be <0.125 mg L1.

2.4.2.

Identification of pharmaceutical metabolites

To identify the major metabolites of CBZP, CA, ATL and PPL, time-course experiments were carried out as described above. Accurate mass MS and MS/MS analyses of parent compounds and their microbial products were performed using a Waters/ Micromass QqToF-Micro system coupled to Waters ACQUITY UPLC system (Micromass, Manchester, UK). Samples from biodegradation experiments were analyzed on a Waters ACQUITY BEH C18 column (10 2.1 mm, 1.7 mm particle size). UPLC analysis was performed in the same way and with the same mobile phases on both instruments. After elution from the column, CBZP, ATL and PPL were analyzed in positive ion (PI) mode with mobile phases consisting of (A) 5 mM aqueous NH4Ac/acetic acid (pH 4.8); and (B) acetonitrile-methanol (2:1, v/v) at 350 mL min1. The elution of ATL and PPL started at 5% B for 1 min and then it was linearly increased to 80% of B in 10.5 min, further increased to 95% of B in the next 1 min, and then returned to initial conditions. Total run time, including the conditioning of the column to the initial conditions was 14 min. For CBZP after the first min at 5% of A the % of A was

524

water research 44 (2010) 521–532

Fig. 1 – Time-course of CBZP (a) and CA (b) degradation in T. versicolor cultures under quinone redox cycle conditions. Symbols: (B) CBZP and CA in the reaction mixture (25 mL 20 mM phosphate buffer containing 500 mM DBQ, 100–300 mM Fe3D-oxalate, 100 mM Mn2D, and 10 mg LL1 of the corresponding pharmaceutical); (C) CBZP and CA in the control treatment (non-containing DBQ and Fe3D-oxalate); (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative area (A/A0) where A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. In the case of CBZP where two hydroxylated isomers were found, black and white triangles refer to P254A and B, respectively. Incubations were carried out with 1.5 ± 0.1 and 1.7 ± 0.2 mg dry weight mLL1 for CBZ and CA, respectively.

increased to 60% at 11 min, further increased to 90% of A in the next 3 min and held isocratically for 2 min. Together with returning to initial conditions, the total run time was 20 min. CA was analyzed in the negative ion (NI) mode with mobile phases consisting of (A) methanol; and (B) water. The elution started at 5% of A for the first min, which was raised to 70% in the next 7 min, to 90% in the next 2 min, and then returned to initial conditions. The total run time was 13 min. The injection volume of the sample was 10 mL. The mass spectrometry analysis on the QqToF instrument was performed in wide pass quadrupole mode, for MS experiments, with the ToF data being collected between m/z 50–800. The capillary and cone voltages were set to 3000 and 25–30 V, respectively. Data were collected in the centroid mode, with a scan accumulation time of 1 s. The instrument was operated at a resolution of 5000 (FWHM). The nebulisation gas was set to 500 L h1 at a temperature of 350 C,

the cone gas was set to 50 L h1, and the source temperature to 120 C. All analyses were acquired using an independent reference spray via the LockSpray interference to ensure accuracy and reproducibility. Sulfaguanidine (C7H10N4O2S) was used as the internal lock mass in both PI mode ([M þ H]þ ¼ m/z 215.0602) and NI mode ([M H] ¼ m/z 213.0446). The LockSpray frequency was set at 11 s. Fragmentation of precursor ions was done by applying collision energies in the range of 10–35 eV, using argon as a collision gas at a pressure of ~20 psi. Elemental compositions of the molecular ions and their fragments were determined and exact masses were calculated with the help of MassLynx V4.1 software incorporated in the instrument. Since the software calculation of the accurate mass of cation is performed by adding a hydrogen atom instead of proton, mass of one electron (i.e., 0.0005) was subtracted from the calculated mass.

water research 44 (2010) 521–532

525

Fig. 2 – Time-course of PPL (a) and ATL (b) degradation in T. versicolor cultures under quinone redox cycle conditions. Culture conditions were as described in the legend for Fig. 1. Symbols: (B) PPL and ATL in the reaction mixture; (C) PPL and ATL in the control treatment; (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative area (A/A0) where A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. Incubations were carried out with 1.9 ± 0.1 mg dry weight mLL1 for both PPL and ATL.

2.4.3.

Mycelial dry weight

Mycelial dry weights were determined by vacuum filtering the cultures through reweighed glass filters (Whatman GF/C, Maidstone, England). The filters containing the mycelial mass were placed in glass dishes and dried at 100 C to constant weight.

3.

Results and discussions

3.1. Oxidation of CA, CBZP, PPL and ATL by T. versicolor under quinone redox cycling conditions The strategy proposed here to degrade the selected pharmaceuticals couples both AOP and biological strategies by inducing the production of oxidizing agents such as ,OH in the white-rot fungus T. versicolor. In the experiments conducted,

oxalate was used as chelant agent of Fe3þ instead of EDTA and Mn2þ was added into the incubation mixture, since the oxidizing power under these conditions was substantially improved as discussed previously (Go´mez-Toribio et al., 2009b). Figs. 1 and 2 show the time-courses degradation of the selected pharmaceuticals in test flasks amended with approximately 10 mg L1 of the target pharmaceuticals, when incubated under DBQ redox cycling conditions. Control treatments in the absence of DBQ and Fe3þ but containing fungus were performed to demonstrate both the positive involvement of oxidizing agents formed from the quinone redox cycling mechanism and to discard the role of fungal sorption in the removal of pharmaceuticals. Furthermore, control treatments permit us to rule out the intracellular and extracellular enzymatic system of T. versicolor in the degradation of the pharmaceuticals. Previously CA and CBZP were reported to be degraded by the cyt P450 system of T. versicolor grown in a chemically define medium (Marco-Urrea et al.,

526

water research 44 (2010) 521–532

experiment, adsorption and transformation of pharmaceutical by fungus cannot be clearly distinguished, since live cultures can also degrade adsorbed pollutants by intracellular mechanisms (Bla´nquez et al., 2004). In this case, adsorption was higher than that observed at 10 mg L1 probably due to the highest concentration of fungal mycelium in the experimental flasks (see figure legend). Similarly to that described for higher concentrations, the plateau in CBZP degradation under quinone redox cycling conditions was observed at the fifth hour, when DBQ(H2) was expected to be depleted under the tested conditions.

3.2. QqToF-MS2 fragmentation pathways of pharmaceuticals and their metabolites

Fig. 3 – Time-course of CBZP degradation added at low concentration of 50 mg LL1 in T. versicolor cultures under quinone redox cycle conditions. Symbols: (;) CBZP in the reaction mixture; (B) CBZP in the control treatment; (C) CBZP in uninoculated bottles. Incubations were carried out with 3.8 ± 0.1 mg dry weight mLL1. Values plotted are means ± standard deviations for triplicate cultures.

2009a) but as can be observed in Figs. 1 and 2 degradation did not occur in control treatments for the incubation period studied. Results in Figs. 1 and 2 show a steep decrease of the concentration of pharmaceuticals in the first hours of incubation when the fungus was incubated under quinone redox cycling conditions. Approximately 50% of CBZP, CA and ATL were degraded the first hour and a plateau was reached above 80% after the sixth hour of incubation. However, the incubation was continued to 24 h to investigate the further degradation of possible metabolites formed during the degradation process (see Section 3.3). The consumption of DBQ was also investigated in the same samples used for the determination of pharmaceutical degradation. This analysis included the determination of DBQ and its counterpart (DBQH2) levels and is shown in Figs. 1 and 2 as the sum of both: DBQ(H2). A high correlation between DBQ(H2) and pharmaceutical degradation was found, suggesting that DBQ(H2) consumption was due to the attack of the induced oxidizing agents. Furthermore, since quinone redox cycling and therefore the production of oxidizing species cannot be sustained without the presence of DBQ(H2), the plateau observed in the pharmaceutical degradation rate can be ascribed to DBQ(H2) depletion. In an effort to discuss the relevance of this strategy in regards to the normal concentrations of pharmaceuticals found in aqueous environments (that are frequently detected at levels from ng L1 to low mg L1), one experiment was carried out at 50 mg L1. For this purpose CBZP was chosen as model compound since it is found to be recalcitrant in sewer treatment facilities and in the environment (Zhang et al., 2008). As is shown in Fig. 3 the concentration of CBZP was decreased to approximately 10 mg L1, due to the mechanisms of fungal sorption and oxidation-induced degradation. In this

Table 1 summarizes the exact masses of molecular and fragment ions, together with recalculated mass errors and double bond equivalents (DBEs) given by the software (mass measurements accuracy threshold of 5 mg L1). Since all fragment ions had an even number of electrons (i.e., DBE was half integer number in all cases), all neutral losses had likewise even electron configurations. These data were obtained under optimized conditions of collision energy and cone voltage in ESI(þ) and ESI() MS2 experiments on the QqToF instrument. In ESI(þ) full-scan experiments performed at UPLC-QqToF instrument, ATL (molecular ion [M þ H]þ267) eluted at 3.4 min. Another peak was observed at 3.2 min, with the molecular ion [M þ H]þ283, denominated as metabolite P282. Considering that the mass of P282 was shifted 16 Da upwards relative to the parent compound, hydroxylation by ,OH radical attack was assumed. In Table 1 are presented calculated and measured masses of fragment ions of ATL and its metabolite P282, determined in ESI(þ)-MS2 experiments at the QqToF-MS instrument. The most abundant fragment ions in the MS2 spectrum of ATL were detected m/z 190 (loss of 77 Da, isopropylamine and water) and m/z 145 (further cleavage of CO and intermolecular cyclization and rearrangement) (see Fig. 4c). In the fragmentation pattern of the molecular ion [M þ H]þ283, characteristic losses of isopropyl group (42 Da) and ammonia (17 Da) afforded the fragment ions at m/z 241 and 224 (see Fig. 4d). The subsequent losses of water and CO from the fragment ion m/z 224 led to the formation of m/z 206 and 178 fragment ions. The fragment ion m/z 161 was formed through cleavage of isopropylamine, water, CO and ammonia and intramolecular cyclization, as suggested in the insert in Fig. 4d. The fragment ion m/z 116 was also present in the MS2 spectrum of P282, indicating that the hydroxylation had to occur at the p-hydroxyphenylacetamide fraction of ATL. From the comparison of the two fragmentation patterns, it was deduced that the –OH group was attached to the alkyl sidechain, at the C-atom next to the ether oxygen. In the case of PPL, MS2 fragmentation of the molecular ion [M þ H]þ260 rendered intense signals at m/z 183 and 157, corresponding to the subsequent cleavages of aminoisopropyl moiety and water, and C2H2, respectively (see Fig. 4a). The less intense fragment ions at m/z 242 and 218 were formed through cleavages of water and isopropylamine, respectively. The fungal metabolite of PPL eluted 0.6 min earlier, with the molecular ion [M þ H]þ276, and was denominated as P275. Besides the molecular ion with mass 16 Da greater than

Table 1 – Accurate mass measurement of product ions of carbamazepine (CBZP), clofibric acid (CA), propranolol (PPL), and atenolol (ATL) and their degradation products P254, P230, P275 and P282, respectively, as determined by UPLC-QqToF in ESI(D) MS2 mode for CBZP, PPL and ATL, and ESI(L) MS2 mode for CA. Comp.

CBZP

P254

CA

PPL

P275

ATL

P282

[M þ H]þ [M þ Na]þ [M þ H NH3]þ [M þ H CONH]þ [M þ H]þ [M þ Na]þ [M þ H NH3]þ [M þ H CONH]þ [M þ H CONH3]þ [M þ H CONH CO]þ [M H] [M H C4H6O2] [M H C6H4ClO] [M H] [M H C4H6O2] [M þ H]þ [M þ H H2O]þ [M þ H C(CH3)2]þ [M þ H H2O NH2CH(CH3)2]þ [M þ H H2O NH2CH(CH3)2 C2H2]þ [M þ H C10H8O]þ [M þ H]þ [M þ H H2O]þ [M þ H H2O O NH2CH(CH3)2]þ [M þ H H2O O NH2CH(CH3)2 C2H2]þ [M þ H C10H8O2]þ [M þ H]þ [M þ H C(CH3)2]þ [M þ H C(CH3)2 NH3]þ [M þ H C(CH3)2 H2O]þ [M þ H C(CH3)2 NH3 H2O CO NH3]þ [M þ H CHCONH2 H2O C(CH3)2 NH3]þ [M þ H C8H9NO2]þ [M þ H]þ [M þ H C(CH3)2]þ [M þ H C(CH3)2 NH3]þ [M þ H C(CH3)2 NH3 H2O]þ [M þ H C(CH3)2 NH3 H2O CO]þ [M þ H C(CH3)2 NH3 CO H2O NH3]þ [M þ H C8H9NO3]þ

Elemental formula

C15H13N2O C15H12N2ONa C15H10NO C14H12N C15H13N2O2 C15H12N2O2Na C15H10NO2 C14H12NO C14H10NO C13H12N C10H10ClO3 C6H4ClO C4H6O2 C10H10ClO4 C6H4ClO2 C16H22NO2 C16H20NO C13H16NO2 C13H11O C11H9O C6H14NO C16H22NO3 C16H20NO2 C13H11O C11H9O C6H14NO C14H23N2O3 C11H17N2O3 C11H14NO3 C11H12NO2 C10H9O C9H9O C6H14NO C14H23N2O4 C11H17N2O4 C11H14NO4 C11H12NO3 C10H12NO2 C10H9O2 C6H14NO

DBEa

Error

Exp.

Theor.

mDa

ppm

237.1019 259.0844 220.0761 194.0960 253.0976 275.0799 236.0710 210.0906 208.0762 182.0965 213.0327 126.9944 85.0294 229.0254 142.9909 260.1653 242.1547 218.1170 183.0814 157.0648 116.1066 276.1583 258.1492 183.0795 157.0664 116.1079 267.1696 225.1222 208.0975 190.0874 145.0650 133.0658 116.1070 283.1663 241.1186 224.0901 206.0807 178.0862 161.0617 116.1055

237.1022 259.0842 220.0757 194.0964 253.0972 275.0791 236.0706 210.0913 208.0762 182.0964 213.0324 126.9956 85.0290 229.0268 142.9905 260.1645 242.1539 218.1176 183.0804 157.0648 116.1070 276.1594 258.1489 183.0804 157.0648 116.1070 267.1704 225.1234 208.0969 190.0863 145.0648 133.0648 116.1071 283.1652 241.1183 224.0917 206.0812 178.0863 161.0597 116.1070

0.3 0.2 0.4 0.4 0.4 0.8 0.4 0.7 0 0.1 0.3 1.2 0.4 1.4 0.4 0.8 0.8 0.6 1.0 0 0.4 1.1 0.3 0.9 1.6 0.9 0.8 5.3 0.6 1.1 0.2 1.0 0.1 1.1 0.3 1.6 0.5 0.1 2.0 1.5

1.3 0.8 1.8 2.1 1.6 2.9 1.7 3.3 0 0.5 1.4 9.4 4.7 6.1 2.8 3.1 3.3 2.7 5.5 0 3.4 4.0 1.2 4.9 10.1 7.7 3.0 1.2 2.9 5.8 1.4 7.5 0.9 3.9 1.2 7.1 2.4 0.6 12.4 12.9

10.5 10.5 11.5 9.5 10.5 10.5 11.5 9.5 10.5 8.5 5.5 4.5 2.5 5.5 4.5 6.5 7.5 6.5 8.5 7.5 0.5 6.5 7.5 8.5 7.5 0.5 4.5 4.5 5.5 6.5 6.5 5.5 0.5 4.5 4.5 5.5 6.5 5.5 6.5 0.5

trb, min

8.93

7.24; 7.81

5.90

6.40 6.7

6.1

3.4

3.2

527

Only product ions with abundances higher than 10% are taken into account. a DBE, double bond equivalent. b tr-UPLC retention time.

Mass (m/z)

water research 44 (2010) 521–532

P230

Precursor ion/product ion

528

water research 44 (2010) 521–532

Fig. 4 – Spectra obtained in ESI(D)-MS/MS experiments at QqToF instrument (cone voltages 20–25 V, collision energies 15– 25 eV) for: (a) standard mixture of PPL in methanol/water (v/v, 25/75) at 10 mg LL1, (b) degradation product of PPL, P275, in the sample taken after 2 h, (c) standard mixture of ATL in methanol/water (v/v, 25/75) at 10 mg LL1, and (d) degradation product of ATL, P282, in the sample taken after 4 h.

[M þ H]þ260, fragment ion at m/z 258 was observed, equivalent to the m/z 242 in the MS2 spectrum of PPL (Fig. 4b). The rest of the observed fragment ions of the molecular ion [M þ H]þ276 were the same (i.e., m/z 116, 157 and 183), indicating that the naphthalene moiety was the site of ,OH radical attack. In full-scan experiments CBZP eluted at 8.94 min, whereas two new peaks emerged at 7.24 and 7.81 min. The identical fragmentation patterns of these two peaks and their molecular weight (MW) 16 Da higher than the parent compound (see Figs. 5c,d) suggested that the formed products were two hydroxylated isomers of CBZP, denominated as P254A and B. In Table 1 exact and measured masses were presented for a more intense M254B. The collision-induced-dissociation experiments with CBZP (molecular ion [M þ H]þ237) revealed the formation of only two fragment ions at m/z 220 and 194, as a consequence of the loss of ammonia and CONH group, respectively. Equivalent fragments were observed for P254A,B, at m/z 236 and 210. The strong signal observed at m/z 208 suggested that the –OH group was probably attached at the C4 and C7-atom for the two observed isomers, since the loss of 2 H-atoms could be favoured by intramolecular cyclization as presented in the insert of Fig. 5d. Further cleavage of CO in the

fragment ion m/z 208 afforded the signal at m/z 182. Nevertheless, based on the retention times only it could not be deduced which peak belongs to which product. Probably the formation of intramolecular hydrogen bond between the attached oxygen and hydrogen atoms of the amide group caused one of the isomers to elute ~0.6 min later than the other. Finally, ESI() full-scan screenings of samples from the experiments with CA (molecular ion [M þ H]þ213) revealed a new signal appearing 0.5 min after the CA peak, at 6.40 min. Similar to the other metabolites identified, its MW was deferring for 16 Da relative to the original drug, with the molecular ion [M þ H]þ229 (i.e., P230). Very poor MS2 fragmentation was observed for both parent compound and its fungal metabolite. The collision of [M þ H]þ213 rendered two fragment ions at m/z 126 and 85, representing the cleavage of the ether bond (see insert in Fig. 5a), whereas in the spectrum of [M þ H]þ229 only one fragment ion was detected at m/z 142, equivalent to the m/z 126 but with mass shifted upwards for 16 Da. Nevertheless, this was enough evidence to assume hydroxylation of the benzene ring, although the exact position of the attached –OH group could not be deduced.

water research 44 (2010) 521–532

529

Fig. 5 – Spectra obtained in MS/MS experiments at QqToF instrument (cone voltages 15–30 V, collision energies 15–25 eV) for: (a) standard mixture of CA in methanol/water (v/v, 25/75) at 10 mg LL1, (b) degradation product of CA, P230, in the sample taken after 2 h, (c) standard mixture of CBZP in methanol/water (v/v, 25/75) at 10 mg LL1, and (d) degradation product of CBZP, P254, in the sample taken after 2 h.

The degradation intermediates of each pharmaceutical detected in this section are found in Fig. 6.

3.3. Discussion of the results on the fungal metabolites of pharmaceuticals The formation of metabolites was depicted in Figs. 1 and 2 and was expressed as relative area (A/A0) measured by integration of LC-MS peaks of the corresponding metabolite (A) and the parent drug in the control treatments at time zero (A0), since due to the lack of authentic analytical standards for the newly identified products their quantitative determination was not possible. Examination of the profiles show qualitatively similar formation pattern of metabolites across the incubation time, obtaining a maximum value of A/A0 between the first and fourth hour and disappearing completely in all the cases after 24 h. The metabolites appeared to be formed at low concentration levels in comparison with the parent drug, with a maximum A/A0 values for each case in the range of 0.012 (PPL, Fig. 2a) to 0.27 (CA, Fig. 1b). The same compound that was identified in our experiments as the fungal metabolite of ATL, P282, was previously reported as one of the intermediate products of solar photo-

Fenton and TiO2 photocatalytic treatment of ATL in pilotscale compound parabolic collectors (Radjenovic et al., 2009). In the case of these photocatalytic treatments, P282 was observed together with its keto tautomer (P280), and was formed through ,OH-mediated reactions. Furthermore, Song et al. (2008) reported a product of ATL with MW 282, formed in the reaction of ATL with ,OH radicals. This product was identified by 60Co g-irradiation and LC-MS, although the ,OH attacks was assumed to occur at the benzene ring. In the same study hydroxylated derivative of PPL was also detected with MW 275 and –OH group attached at the naphthalene ring, which coincides with the structure of P275 identified in the present study. Hydroxylation of naphthalene moiety is phase I reaction in the human metabolism of PPL, in which 4hydroxy PPL is formed (Luan et al., 2005). Cytochrome P450 isozymes were found to be responsible for oxidative metabolic pathways of PPL in humans proceeding through naphthalene ring-hydroxylations at the 4-, 5-, and 7-positions and side-chain N-desisopropylation (Masubuchi et al., 1994). However, in our study no metabolites were found in fungal control treatments indicating that P275 was generated via induction of oxidizing agents instead of fungal cyt P450 mechanism.

530

water research 44 (2010) 521–532

Fig. 6 – Selected pharmaceuticals and proposed structures of their degradation products.

Sire´s et al. (2007) proposed the formation of a hydroxylated intermediate as the observed in our study after the hydroxylation of CA on its C2-position by electro-Fenton and photoelectro-Fenton. This proposed metabolite was not identified by pure standards in the mentioned study but it was assigned to a dehydrated species of 2-(4-chloro-2hydroxyphenoxy)-2-methylpropionic acid on the basis of its mass fragmentation spectra. On the other hand, by using an HPLC-DAD and FLD, Doll and Frimmel (2004) proposed TiO2 photocatalytic degradation pathway of CA, which did not include the hydroxylated product as the one observed in our study. During photocatalysis CA underwent substitution of –Cl by –OH group in the para-position, whereas products such as 2-(4-hydroxyphenoxy)-isobutyric acid and hydroquinone were formed. On the other side, cleavage of isobutyric acid from the side-chain afforded 4-chlorophenol and 4-chlorocatechol. Interestingly, in the abovementioned reports, substrates of laccase such as phenol and 4-chlorophenol have been described as typical metabolites of CA degradation by AOP and once produced they could therefore be mineralized by the ligninolytic enzymatic system of WRF. But perhaps even more interesting is the production of different

quinone metabolites (also described for CA degradation by , OH attack) that could theoretically be incorporated into the quinone redox cycling enhancing the degradation rates of the target pollutant. As far as CBZP is concerned, there are few studies regarding identification of products in ,OH-induced degradation pathway. Vogna et al. (2004) reported the formation of toxic acridine intermediates in UV/H2O2 treatment of CBZP, formed in ,OH and ,HO2-mediated reactions. Chiron et al. (2006) also found acridine as major photodegradation intermediate of CBZP after direct photolysis, but in the presence of Feþ3 they identified a hydroxycarbamazepine structure derived from a ,OH attack on the C6 aromatic ring. The formation of 2- and 3-hydroxylated derivatives of CBZP has been reported in the metabolism of CBZP by the cytochrome P450 in mammals (Mandrioli et al., 2001; Pearce et al., 2002) as well as in the two model fungi Cunninghamella elegans and Umbelopsis ramanniana used to study mammalian metabolism for many drugs (Kang et al., 2008). As far as we know, the two hydroxylated metabolites of CBZP in the 4- and 7-position (P254A and B) detected in this study are reported for the first time.

water research 44 (2010) 521–532

4.

Conclusions

Biological advanced oxidation of ATL, CA, CBZP and PPL was demonstrated for the first time by inducing extracellular oxidizing species in T. versicolor. Under quinone redox cycling conditions, pharmaceuticals degradation (added at 10 mg L1) reached a plateau after 6 h of incubation achieving degradation levels up to 80%. The plateau observed was due to DBQ depletion, which was shown to be concomitant to pharmaceutical degradation. The feasibility of this novel technology to degrade pharmaceuticals at low levels commonly found in the environment (50 mg L1) was also demonstrated. Hydroxylated intermediates of all pharmaceuticals were identified and they disappeared after the incubation period, suggesting the total mineralization of the target compound.

Acknowledgements This work was supported by the Spanish MICINN (project CTM2007-60971/TECNO) and MMAMRM (project 010/PC08/3-04). The Department of Chemical Engineering of the UAB is the Unit of Biochemical Engineering of the XRB de la Generalitat de Catalunya. E. Marco-Urrea, T. Vicent and G. Caminal are members of a Consolidated Research Group of Catalonia (2005SGR 00220). Jelena Radjenovic´ gratefully acknowledges the JAE Program (CSIC-European Social Funds).

references

Alder, A.C., Bruchet, A., Carballa, M., Clara, M., Joss, A., Lo¨ffler, D., McArdell, C.S., Miksch, K., Omil, F., Tuhkanen, T., Ternes, T.A., 2006. Consumption and occurrence. In: Ternes, T.A., Joss, A. (Eds.), Human Pharmaceuticals, Hormones and Fragrances. IWA Publishing, London, pp. 15–54. Bla´nquez, P., Casas, N., Font, X., Gabarrell, X., Sarra`, M., Caminal, G., Vicent, T., 2004. Mechanism of textile metal dye biotransformation by Trametes versicolor. Water Research 38 (8), 2166–2172. Buser, H.-R., Mu¨ller, M.D., Theobald, N., 1998. Occurrence of the pharmaceutical drug clofibric acid and the herbicide mecoprop in various Swiss Lakes and in the North Sea. Environmental Science and Technology 32 (1), 188–192. Cabana, H., Jones, J.P., Agathos, S.N., 2007. Elimination of endocrine disrupting chemicals using white-rot fungi and their lignin modifying enzymes: a review. Engineering in Life Sciences 7 (5), 429–456. Chiron, S., Minero, C., Vione, D., 2006. Photodegradation processes of the antiepileptic drug carbamazepine, relevant to estuarine waters. Environmental Science and Technology 40 (19), 5977–5983. Doll, T.E., Frimmel, F.H., 2004. Kinetic study of photocatalytic degradation of carbamazepine, clofibric acid, iomeprol and iopromide assisted by different TiO2 materials – determination of intermediates and reaction pathways. Water Research 38 (4), 955–964. Dollery, C., 1991. Therapeutic Drugs. Churchill Livingstone, Edinburgh. Escher, B.I., Bramaz, N., Richter, M., Lienert, J., 2006. Comparative ecotoxicological hazard assessment of beta-blockers and their

531

human metabolites using a mode-of-action-based test battery and a QSAR approach. Environmental Science and Technology 40 (23), 7402–7408. Ferrari, B., Paxe´us, N., Giudice, R.L., Pollio, A., Garric, J., 2003. Ecotoxicological impact of pharmaceuticals found in treated wastewaters: study of carbamazepine, clofibric acid, and diclofenac. Ecotoxicology and Environmental Safety 55 (3), 359–370. Go´mez-Toribio, V., Garcı´a-Martı´n, A.B., Martı´nez, M.J., Martı´nez, A.T., Guille´n, F., 2009a. Induction of extracellular hydroxyl radical production by white-rot fungi through quinone redox cycling. Applied and Environmental Microbiology 75 (12), 3944–3953. Go´mez-Toribio, V., Garcı´a-Martı´n, A.B., Martı´nez, M.J., Martı´nez, A.T., Guille´n, F., 2009b. Enhancing the production of hydroxyl radicals by Pleurotus eryngii via quinone redox cycling for pollutant degradation. Applied and Environmental Microbiology 75 (12), 3954–3962. Grillo, M.P., Benet, L.Z., 2001. Interaction of gglutamyltranspeptidase with clofibryl- S-acyl-glutathione in vitro and in vivo in rat. Chemical Research in Toxicology 14 (8), 1033–1040. Guille´n, F., Go´mez-Toribio, V., Martı´nez, M.J., Martı´nez, A.T., 2000. Production of hydroxyl radical by the synergistic action of fungal laccase and aryl alcohol oxidase. Archives of Biochemistry and Biophysics 383 (1), 142–147. Hug, S.J., Leupin, O., 2003. Iron-catalyzed oxidation of arsenic(III) by oxygen and by hydrogen peroxide: pH-dependent formation of oxidants in the Fenton reaction. Environmental Science and Technology 37 (12), 2734–2742. Joss, A., Repetto, G., Rios, J.C., Hazen, M.J., Molero, M.L., del Peso, A., Salguero, M., Fernandez-Freire, P., Perez-Martin, J.M., Camean, A., 2003. Ecotoxicological evaluation of carbamazepine using six different model systems with eighteen endpoints. Toxicology in Vitro 17 (5-6), 525–532. Kang, S.-I., Kang, S.-Y., Hur, H.-G., 2008. Identification of fungal metabolites of anticonvulsant drug carbamazepine. Applied Microbiology and Biotechnology 79 (4), 663–669. Kerr, B.M., Thummel, K.E., Wurden, C.J., Klein, S.M., Kroetz, D.L., Gonzalez, F.J., Levy, R.H., 1994. Human liver carbamazepine metabolism: role of CYP3A4 and CYP2C8 in 10,11-epoxide formation. Biochemical Pharmacology 47 (11), 1969–1979. Klavarioti, M., Mantzavinos, D., Casinos, D., 2009. Removal of residual pharmaceuticals from aqueous systems by advanced oxidation processes. Environment International 35 (2), 402–417. Luan, L.-J., Shao, Q., Ma, J.-Y., Zeng, S., 2005. Stereoselective urinary excretion of S-()- and R-(þ)-propranolol glucuronide following oral administration of RS-propranolol in Chinese Han subjects. World Journal of Gastroenterology 11 (12), 1822–1824. Mandrioli, R., Albani, F., Casamenti, G., Sabbioni, C., Raggi, M.A., 2001. Simultaneous high-performance liquid chromatography determination of carbamazepine and five of its metabolites in plasma of epileptic patients. Journal of Chromatography B 762 (2), 109–116. Marco-Urrea, E., Parella, T., Gabarrell, X., Caminal, G., Vicent, T., Reddy, C.A., 2008. Mechanistics of trichloroethylene mineralization by the white-rot fungus Trametes versicolor. Chemosphere 70 (3), 404–410. Marco-Urrea, E., Pe´rez-Trujillo, M., Vicent, T., Caminal, G., 2009a. Ability of white-rot fungi to remove selected pharmaceuticals and identification of degradation products of ibuprofen by Trametes versicolor. Chemosphere 74 (6), 765–772. Marco-Urrea, E., Aranda, E., Caminal, G., Guille´n, F., 2009b. Induction of hydroxyl radical production in Trametes versicolor to degrade recalcitrant chlorinated hydrocarbons. Bioresource Technology 100 (23), 5757–5762.

532

water research 44 (2010) 521–532

Masubuchi, Y., Hosokawa, S., Horie, T., Suzuki, T., Ohmori, S., Kitada, M., Narimatsu, S., 1994. Cytochrome P450 isozymes involved in propranolol metabolism in human liver microsomes. The role of CYP2D6 as ring-hydroxylase and CYP1A2 as N-desisopropylase. Drug Metabolism and Disposition 22 (6), 909–915. Pearce, R.E., Vakkalagadda, G.R., Steven Leeder, J., 2002. Pathways of carbamazepine bioactivation in vitro I. Characterization of human cytochromes P450 responsible for the formation of 2- and 3-hydroxylated metabolites. Drug Metabolism and Disposition 30 (11), 1170–1179. Pfluger, P., Dietrich, D.R., 2001. Effects on pharmaceuticals in the environmentdan overview and principle considerations. In: Ku¨mmerer, K. (Ed.), Pharmaceuticals in the EnvironmentdSources, Fate, Effects and Risks. Springer, Berlin, pp. 11–17. Radjenovic, J., Petrovic, M., Barcelo´, D., 2007. Analysis of pharmaceuticals in wastewater and removal using a membrane bioreactor. Analytical and Bioanalytical Chemistry 387 (4), 1365–1377. Radjenovic, J., Sirtori, C., Petrovic, M., Barcelo´, D., Sixto, M., 2009. Solar photocatalytic degradation of persistent pharmaceuticals at pilot-scale: Kinetics and characterization of major intermediate products, Applied catalysis B: Environmental 89 (1–2), 255–264.

Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., Gonza´lez, S., Barcelo´, D., Ventura, F., Knepper, T.P., 2006. Polar pollutants entry into the water cycle by municipal wastewater: a European perspective. Environmental Science and Technology 40 (17), 5451–5458. Sire´s, I., Arias, C., Cabot, P.Ll., Centellas, F., Garrido, J.A., Rodrı´guez, R.M., Brillas, E., 2007. Degradation of clofibric acid in acidic aqueous medium by electro-Fenton and photoelectro-Fenton. Chemosphere 66 (9), 1660–1669. Song, W., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M., 2008. Free radical destruction of b-blockers in aqueous solution. Environmental Science and Technology 42 (4), 1256–1261. Stafiej, A., Pyrzynska, K., Regan, F., 2007. Determination of antiinflammatory drugs and estrogens in water by HPLC with UV detection. Journal of Separation Science 30 (7), 985–991. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32 (11), 3245– 3260. Vogna, D., Marotta, R., Andreozzi, R., Napolitano, A., D’Ischia, M., 2004. Kinetic and chemical assessment of the UV/H2O2 treatment of antiepileptic drug carbamazepine. Chemosphere 54 (4), 497–505. Zhang, Y., Geissen, S.U., Gal, C., 2008. Carbamazepine and diclofenac: removal in wastewater treatment plants and occurrence in water bodies. Chemosphere 73 (8), 1151–1161.

water research 44 (2010) 533–544

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Loadings, trends, comparisons, and fate of achiral and chiral pharmaceuticals in wastewaters from urban tertiary and rural aerated lagoon treatments Sherri L. MacLeod a,1, Charles S. Wong a,b,* a

Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2, Canada Environmental Studies Program and Department of Chemistry, Richardson College for the Environment, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9, Canada b

article info

abstract

Article history:

A comparison of time-weighted average pharmaceutical concentrations, loadings and

Received 1 June 2009

enantiomer fractions (EFs) was made among treated wastewater from one rural aerated

Received in revised form

lagoon and from two urban tertiary wastewater treatment plants (WWTPs) in Alberta,

14 September 2009

Canada. Passive samplers were deployed directly in treated effluent for nearly continuous

Accepted 23 September 2009

monitoring of temporal trends between July 2007 and April 2008. In aerated lagoon effluent,

Available online 27 September 2009

concentrations of some drugs changed over time, with some higher concentrations in winter likely due to reduced attenuation from lower temperatures (e.g., less microbially

Keywords:

mediated biotransformation) and reduced photolysis from ice cover over lagoons; however,

Pharmaceuticals

concentrations of some drugs (e.g. antibiotics) may also be influenced by changing use

Passive samplers

patterns over the year. Winter loadings to receiving waters for the sum of all drugs were

Wastewater

700 and 400 g/day from the two urban plants, compared with 4 g/day from the rural plant.

Loadings

Per capita loadings were similar amongst all plants. This result indicates that measured

Trends

loadings, weighted by population served by WWTPs, are a good predictor of other effluent

Chiral drugs

concentrations, even among different treatment types. Temporal changes in chiral drug EFs were observed in the effluent of aerated lagoons, and some differences in EF were found among WWTPs. This result suggests that there may be some variation of microbial biotransformation of drugs in WWTPs among plants and treatment types, and that the latter may be a good predictor of EF for some, but not all drugs. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

Pharmaceutical occurrence, fate, and toxicity in aquatic ecosystems has been studied for more than 10 years (Daughton and Ternes, 1999; Halling-Sorensen et al., 1998; Kolpin

et al., 2002; Stamm et al., 2008). While drugs have been consistently detected in surface water, the effects of these complex mixtures of bioactive pollutants on non-target aquatic organisms are not fully understood (Fent et al., 2006). Accurate exposure scenarios are essential to assess toxicity

Abbreviations: CR, Capital Region; DL, method detection limit; EF, enantiomer fraction; GB, Gold Bar; LLB, Lac La Biche; PEC, predicted environmental concentration; POCIS, polar organic chemical integrative sampler; SD, standard deviation; TWA, time-weighted average; WWTP, wastewater treatment plant. * Corresponding author. Richardson College for the Environment, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9, Canada. E-mail addresses: [email protected] (S.L. MacLeod), [email protected] (C.S. Wong). 1 Department of Chemistry, Universite´ de Montre´al, Montre´al, Quebec, H3C 3J7, Canada. 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.056

534

water research 44 (2010) 533–544

and assign risk. Exposure, in turn, can be determined by either prediction or by measurement of wastewater or surface water concentrations. Prescription and sales information are not readily available, and may not be ideal for predicting environmental concentrations of some drugs (Coetsier et al., 2009) because of the uncertainty associated with use and metabolism (Letzel et al., 2009). In addition, removal of drugs from wastewater may range from 0 to 100% depending on the drug and the treatment process (Batt et al., 2007; Letzel et al., 2009, Lishman et al., 2006), even for the same compound such as diclofenac (Letzel et al., 2009), further complicating prediction of environmental concentrations. In Europe, where a tiered approach is taken for requirements of fate and effects studies, predicted environmental concentrations (PECs) are calculated as a worst-case scenario, without accounting for any metabolism, biodegradation, or retention by treatment (EMEA, 2006). For PEC values above 10 ng/L, higher tiered assessment is required (EMEA, 2006), and while in a French study, PECs were accurate for some drugs (carbamazepine, diclofenac, propranolol), local measured environmental concentrations of others were found to be much lower than those predicted, even very close (10 m) to effluent outfalls (Coetsier et al., 2009). Wastewater and surface water concentrations may be heavily affected by abiotic transformation processes such as photolysis, which depends on light intensity over the seasons (Vieno et al., 2005); as well as biotransformation, which would depend on microbial activity and on temperature. The latter is particularly significant for the numerous chiral drugs. As with other chiral pollutants (Wong, 2006), the enantiomers of pharmaceuticals are affected identically by abiotic processes, but may be affected differently by biologically mediated processes in wastewater treatment (Fono and Sedlak, 2005; MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006). Drug enantiomers may also exhibit differential toxicity to aquatic life (Stanley et al., 2006). Thus, a robust means by which to predict environmental loadings and fate of drugs from wastewater effluents is valuable for exposure and risk assessment. Direct measurement of drug loadings from wastewater treatment plants (WWTPs) has proven accurate for prediction of drug concentrations in receiving waters (Letzel et al., 2009; MacLeod, 2009). Such measurements can be accomplished via repeated grab sampling (Sacher et al., 2008), which is tedious and time-consuming, or continuous flow-proportional effluent collection (Letzel et al., 2009). However, passive sampling devices such as the Polar Organic Chemical Integrative Sampler (POCIS) (Alvarez et al., 2004) can provide a simple means of obtaining time-weighted average (TWA) loadings of polar contaminants, such as drugs, over longer time periods with lower detection limits (Jones-Lepp et al., 2004; MacLeod et al., 2007a; Zhang et al., 2008). While a high-frequency, discrete sampling approach may capture the real-time heterogeneity of contaminant loadings, passive samplers have the advantage of providing integrated, continuous sample collection, and are thus useful as an alternative or complementary sampling approach for dissolved phase chemicals. Most research on drugs in the environment has, thus far, focused on measurement of drugs in the waste or receiving waters from larger (10,000 to millions) population centers (Carballa et al., 2007; Chen et al., 2006; Hua et al., 2006; Kimura et al., 2007; Lajeunesse et al., 2008; Letzel et al., 2009; Lissemore

et al., 2006; Loraine and Pettigrove, 2006; Managaki et al., 2007; Sacher et al., 2008; Vieno et al., 2005), with little attention paid to the drug output from the wastewater of smaller (<10,000) communities (Batt et al., 2007; Lishman et al., 2006; MacLeod et al., 2007a; Vasskog et al., 2008; Vieno et al., 2005). Thus, the risk of exposure to pharmaceuticals in receiving waters of smaller communities is unclear. In this study, we used POCIS to assess pharmaceutical fate via TWA loadings and temporal trends of achiral and chiral pharmaceuticals at three WWTPs (Fig. 1). The first plant serves Lac La Biche, a rural community of 4000 (Statistics Canada, 2009) in northeastern Alberta, Canada. As of 1999, 84% of Canadian inland municipalities use secondary or tertiary treatment (Environment Canada, 2009). Treatment effectively equivalent to secondary treatment is used in the Lac La Biche WWTP (LLB) through long retention (90 days) in waste stabilization ponds or aerated lagoons (Duigou, 2006). We compared winter drug output from the LLB WWTP with that from two urban WWTPs in central Alberta, Canada, with tertiary UV treatment: Edmonton Gold Bar (GB), serving 750,000 (City of Edmonton, 2009) and Capital Region (CR), serving 250,000 (Alberta Capital Region Wastewater Commission, 2009) in parts of Edmonton and surrounding

Fig. 1 – Map showing study sites in central and northeastern Alberta, Canada. GB, Gold Bar WWTP in Edmonton; CR, Capital Region WWTP in Fort Saskatchewan; LLB, Lac La Biche WWTP in Lac La Biche. Figure prepared by Charlene Nielsen, Geographic Information Systems, Department of Biology, University of Alberta.

water research 44 (2010) 533–544

communities. Treated effluent from LLB is discharged continuously to Field Lake which discharges into Lac La Biche (the lake) via Red Deer Brook, while GB and CR both discharge treated effluent to the North Saskatchewan River and we have previously measured drugs in both of these receiving waters (MacLeod, 2009; MacLeod et al., 2007a). To our knowledge, this is the first study to use POCIS to undertake long-term, nearly continuous, monitoring of drugs, particularly chiral drugs, in municipal effluents from a small population center.

2.

Materials and methods

2.1.

Materials

Table 1 – Pooled winter time-weighted average concentrations for drugs detected at all three WWTPs in each of the three deployment periods.

Atenolol Carbamazepine Celecoxib Citalopram Clarithromycin Codeine Diclofenac Erythromycin Gemfibrozil Metoprolol Naproxen Paroxetine Propranolol Sotalol Temazepam Triclosan Trimethoprim Sum of drugs (mg/L)

Gold Bar (ng/L) 72 200 17 23 220 450 1200 13 150 58 66 0.7 4.7 27 130 1.5 38 2.2

SD 270 130 8 5 140 380 680 10 59 47 58 0.9 3.0 17 71 2.3 39

Capital Region (ng/L) 120 290 28 67 420 920 2,000 19 190 65 530 1.4 14 66 160 53 65 5.0

SD 440 190 13 8 270 710 1300 15 72 52 510 1.8 8.5 22 76 69 70

Lac la Biche (ng/L) 36 63 22 5.7 120 990 460 3.1 72 29 280 0.2 1.4 16 120 6.9 10 2.2

SD 160 39 10 1.7 90 710 280 2.8 34 23 250 0.2 1.1 11 60 8.7 12

Pooled values calculated from n ¼ 6 for Gold Bar and Capital Region WWTPs, n ¼ 9 for Lac la Biche WWTP.

Passive sampling in WWTP effluent

Sampling of WWTP effluent took place between July 2007 and April 2008 (Table S1). During each sampling period at the CR and LLB WWTPs, two and three POCIS, respectively, were protected in perforated stainless steel cages secured with nylon rope, and deployed in final effluent. During each sampling period at GB WWTP, two caged samplers were deployed in a tank into which treated effluent was continuously pumped. Field blanks were brought to all sites at deployment and retrieval, at which time POCIS were individually put on ice until return to the laboratory, where they were stored at 20 C until chemical extraction.

2.3.

We used POCIS (Environmental Sampling Technologies, St. Joseph, MO) in the ‘pharmaceutical’ configuration: 200 mg Oasis HLB sorbent (Waters, Milford, MA) between two polyethersulfone membranes (45.8 cm2 total surface area) held together by stainless steel rings and attached to a spindle holder. This configuration was used as POCIS sampling rates are available for the target analytes (Tables 1 and S2), all commonly used drugs. Chemical analytes (>98% purity) and used as received (sources in Table S2). Solvents and reagents for chromatography (Fisher Scientific, Ottawa, Canada) were of HPLC grade; nanopure water (18 MU cm) was supplied by a Nanopure Ultrapure system (Barnsteam/Thermolyne, Dubuque, IA). Details are published elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b).

Analyte

2.2.

535

Chemical extraction and instrumental analysis

Details on chemical extraction and instrumental analysis are provided elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b) and in Supplemental Information. In brief, POCIS sorbent was washed into a glass column, and methanol used to elute target analytes. The extract was filtered, reduced to 5–10 mL via rotary evaporation, filtered through 0.22 mm Acrodisc polytetrafluoroethylene syringe filters (Pall Life Sciences, Ann Arbor, MI, USA), and nitrogen-evaporated to dryness. Internal standards (Table S2) were added, and extracts were reconstituted to 1 mL with methanol. Achiral chromatography for determining concentrations (MacLeod et al., 2007a) was performed with an Ultra C18 reversed-phase column (150 mm 4.6 mm internal diameter (i.d.) 5 mm particle size (dp), Restek, Bellefonte, PA, USA), while enantioselective chromatography for determining enantiomer compositions was performed with a Chirobiotic V column (250 mm 4.6 mm i.d. 5 mm dp, Advanced Separation Technologies, Whippany, NJ, USA) for all chiral analytes except temazepam, for which a Chiralpak AD-RH column (MacLeod, 2009) was used (150 mm 4.6 mm i.d. 5 mm dp, Daicel Chemical Industries, West Chester, PA). Analytes were detected by electrospray ionization-tandem mass spectrometry using an Applied Biosystems QTrap triple-quadruple instrument (Foster City, CA) in multiple reaction monitoring mode.

2.4. Quality assurance/quality control and data handling Detection limits (DLs, Table S2) were determined as the analyte concentration in 1 mL POCIS extract providing a signal-to-noise ratio of 3, and ranged from 0.001 to 1.6 ng/ POCIS. Drugs were not detected in any blanks. Calculated time-weighted average concentrations and loadings (hereafter called ‘‘concentrations’’ and ‘‘loadings’’, respectively) are presented as mean standard deviation (SD). Data analysis was carried out as previously described elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b). Statistical comparisons between two groups were carried out by t-test, while three or more groups were compared by ANOVA, with either Tukey’s test for multiple comparisons or with Dunnett’s test for comparison with a control, as indicated. Differences noted are statistically significant (all p < 0.05). As expected, analytes for which the POCIS sampling rate was known with higher

536

water research 44 (2010) 533–544

precision (Table S2) were generally more conducive to statistical significance testing. For chiral drugs atenolol, citalopram, fluoxetine, metoprolol, nadolol, pindolol, propranolol, salbutamol, sotalol, and temazepam, enantiomer fractions (EFs) described enantiomer composition (Harner et al., 2000): EF ¼

E1 ðþÞ ¼ E1 þ E2 ð þ Þ þ ð Þ

(1)

where E1 and E2 are the first and second eluted enantiomers, respectively, when the enantiomer optical rotation was unknown (i.e., all analytes except atenolol, propranolol, and fluoxetine). Further information on data quality and handling is available in Supplementary Data.

3.

Results and discussion

3.1. Time-weighted average pharmaceutical concentrations in treated effluents 3.1.1.

Aerated lagoons: Lac la Biche

Most studies to date have relied on grab or short-term composite sampling to assess temporal trends in pharmaceutical concentrations in effluents (Batt et al., 2007; Chen et al, 2006; Hua et al., 2006; Loraine and Pettigrove 2006, Stu¨lten et al., 2008; Vieno et al., 2005). However, temporal trends assumed from grab sampling may not be representative of longer time periods. Because POCIS can provide longer and continuous temporal integration, it may provide more accurate assessment of temporal trends. However, passive samplers cannot provide reliable information on concentration fluctuations over time scales shorter than sample integrative periods, which can result in very different conclusions about temporal changes in concentration over time (Shaw and Mueller, 2009). Drug concentrations at LLB (Fig. 2) measured over six sampling periods of 35–51 days (Table S1) ranged from 0.07 ng/L (paroxetine, propranolol) to 1100 ng/L (codeine). Fenoprofen, fluoxetine, ketoprofen, omeprazole, roxithromycin, sildenafil, sulfadimethoxine, sulfamethazine, sulfapyridine, sulfisoxazole, tadalafil and vardenafil were not detected in LLB treated effluent above their respective DLs in any sampling period (Table S2). Concentrations of atenolol and propranolol were similar to effluent grab sampled from LLB in September 2005 (Nikolai et al., 2006); however, metoprolol was higher (200 ng/L). Diclofenac and naproxen concentrations at LLB were also similar to that in effluents of the activated sludge WWTP of Aura, Finland, which has a similar population (4000) to LLB (Vieno et al., 2005). Concentrations were also generally similar to those in an aerated lagoon WWTP in Louisiana, USA (Conkle et al., 2008) for drugs common to both studies: atenolol, carbamazepine, gemfibrozil, metoprolol, naproxen, and sotalol. Thus, our TWA concentrations are likely representative of those expected in small community WWTP effluents. There were statistically significant temporal concentration changes (Fig. 2) for celecoxib, citalopram, clarithromycin, codeine, diclofenac, erythromycin, gemfibrozil, naproxen, propranolol, sotalol, and temazepam. Temporal concentration trends were similar for clarithromycin, codeine, diclofenac,

erythromycin, naproxen and propranolol, with higher concentrations during winter sampling periods (December to April) than at other times (July to December, Fig. 2). For atenolol, carbamazepine, metoprolol, triclosan and trimethoprim, the effluent concentrations did not change significantly over the sampling periods (Fig. 2). For drugs with temporal differences in concentration, the changes in concentration ranged from 2 (propranolol) to 960 ng/L (codeine), whereas drugs without statistically significant temporal differences had smaller concentration changes over time, from 8 (triclosan) to 85 ng/L (atenolol). While the average daily volume of treated effluent discharged decreased over the six sampling periods (Table S1), this decrease was not statistically significant. Thus, TWA loadings are reflective of concentrations. Of the drugs studied, photodegradation likely plays a major role in the dissipation of diclofenac (Andreozzi et al., 2003), naproxen (Lin et al., 2006; Lin and Reinhard, 2005), and propranolol (Andreozzi et al., 2003) in aquatic systems. The higher effluent concentrations of those drugs in winter (Fig. 2) may be at least partially the result of decreased photolysis, due to ice cover on lagoons from November to March (Siebold, 2008) blocking sunlight from the water column (Vieno et al., 2005), and fewer hours of sunlight during those months (MacLeod, 2009; Vieno et al., 2005). It is impossible to ascertain the amount of photodegradation in the lagoon during ice-free periods with our data. While the LLB lagoon wastewater was not clear in color, it is likely that photodegradation of drugs took place at least at the lagoon surface during ice-free periods, and continuous aeration of lagoon waters for oxygenation purposes would bring water at depth to the surface to provide some photodegradation treatment. Lower winter temperatures also likely influenced elimination of pharmaceuticals, through lower reaction rate constants for abiotic transformation, and reduced microbial activity for biotransformation (Kim and Carlson, 2007). Effluent pH was 8 0.5 over the six sampling periods and effluent temperatures ( C) were 21 3, 16 2, 4 3, 1 0, 1.4 0.5, 4 2, chronologically for the six sampling periods (Siebold, 2008). The concentrations of those three drugs decreased again in March/April, as temperatures and number of daylight hours increased, and ice cover was reduced or absent on the lagoons (Fig. 2). Since naproxen can be readily biodegraded in sewage (Yu et al., 2006), decreased biodegradation may also contribute to the higher effluent concentrations for naproxen in winter. It is possible that changes in temperature may affect calculated TWA concentrations, as sampling rates used in this study were determined at room temperature (22–28 C, MacLeod et al., 2007a), and WWTP temperatures during colder months were considerably less. However, we cannot presently quantify the compound-dependent effect of such changes on our observed trends. Lower temperatures would result in lower analyte aqueous diffusion coefficients and hence lower sampling rates, estimated to be roughly 75% over a 20 C ambient temperature change (MacLeod et al., 2007a). When applied to water colder than under the calibration condition, sampling rates would then provide an underestimate of the calculated TWA concentrations, the effect of which would be compound-dependent. Such an effect may be pronounced in the case of drugs with higher

water research 44 (2010) 533–544

537

Fig. 2 – Time-weighted average concentrations (ng/L ± SD; ) and enantiomer fractions (EF) of chiral drugs (,) in LLB treated wastewater during six sampling periods from July 2007 to August 2008 (Table S1). Hash marks on x-axis indicate break in continuous sampling between September and October 2008. Significant concentration differences are indicated by lowercase letters while EF differences are indicated by upper-case letters. Plots without letters do not have concentrations or EFs, as appropriate, statistically different from one another over sampling periods. Non-racemic EFs are indicated by an asterisk, and EFs that were corrected for enantiomer-specific matrix effects are indicated by ^. RS, racemic standards. Error bars (SD) represent propagated error from replicate POCIS, analytical processing and instrumental uncertainty, and error associated with POCIS sampling rates.

concentrations in winter than in other times (e.g., antibiotics such as clarithromycin and erythromycin, Fig. 2). Temporal effects on concentrations were less for other analytes, and may either be insignificant (e.g., triclosan and gemfibrozil) or of little practical significance even if statistically significant, given the relatively narrow range (e.g., from a factor of two, to an order of magnitude) in concentrations over the seasons studied (Fig. 2).

Without measurements of influent concentration, we cannot discern whether temporal concentrations in effluent are more influenced by changes in removal efficiency, as suggested as likely by many studies (Hua et al., 2006; Kim and Carlson, 2007; Loraine and Pettigrove, 2006; Sacher et al., 2008; Vieno et al., 2005) or in use patterns over time, which may be the case for antibiotics. In a Swiss study (McArdell et al., 2003), higher macrolide concentrations were found in winter than in

538

water research 44 (2010) 533–544

other periods, consistent with our observations for the macrolides clarithromycin and erythromycin (Fig. 2). In the Swiss study, as with this study, raw wastewater concentrations were not measured so removal efficiency could not be assessed. However, the Swiss authors cited a personal communication which indicated that macrolide sales were higher in winter, indicating that increased use may contribute to higher environmental concentrations. The winter concentrations in LLB treated effluent, which were sampled mostly at the same time as those at GB and CR (see Section 3.1.3), changed over time for some but not all of our target analytes (Fig. 2). Concentration changes were less prevalent among the three winter sampling months, compared with other periods. Among the sampling periods between December and April (Tukey’s test), there were statistically significant changes in effluent concentration for only three drugs. The celecoxib concentration in December/ January was three times that in January/March and twice that in March/April with magnitudes of difference of 20 and 15 ng/L, respectively. Sotalol had December/January concentrations were eight times that in both January/March and March/April and magnitudes of 33 ng/L for both comparisons. While the concentration of citalopram in January/March was 50% higher than that in March/April, the magnitude of this difference was only 2 ng/L. The general lack of effluent concentration change over the winter indicates that there is no change in drug use patterns and/or in the ability of aerated lagoons to remove drugs effectively over this season. It is important to note that concentration changes in WWTPs serving small communities may well be sensitive to changes in use patterns of the population served. If a drug was completely excreted from the body in unchanged form and did not undergo any removal during wastewater treatment, then the addition of a single additional individual’s twice-aday, 100 mg active ingredient dose at LLB would result in an effluent concentration change of 100 ng/L, which would rival existing concentrations at LLB (Fig. 2)! In actuality, only a fraction of most drugs are excreted as parent compounds, and WWTP treatment provides some elimination. For example, 3% of carbamazepine is excreted unmetabolized from humans, with 30% removal based on a survey of Canadian WWTPs (Miao et al., 2005). Assuming these conditions hold for LLB, then a 2 ng/L change in effluent concentration, or a 3–10% increase (Fig. 2), would result from the same dosage. However, without specific usage information, excretion data, and WWTP-specific removal efficiencies, we cannot make more specific predictions of the sensitivity of drug concentrations in wastewater on usage in small communities.

3.1.2.

Tertiary treatment: Gold Bar and Capital Region

The concentration of detectable drugs (Table 1, Table S3) in GB effluent ranged from 0.03 (omeprazole) to 1350 ng/L (diclofenac) while that in CR ranged from 0.09 (roxithromycin) to 2520 ng/L (diclofenac). These concentrations are similar to grab sample measurements for our analytes in other Canadian WWTP effluents (Hua et al., 2006; Lishman et al., 2006; Miao et al., 2005), indicating that our passive sampler derived TWA concentrations were comparable with previous work. However, concentrations of chiral drugs in this study were generally lower by a factor of at least 10 compared with a grab

sample from GB in April 2007 (Macleod et al., 2007b), suggesting possible annual or other temporal changes in concentrations as discussed below. Since only two samplers were deployed in each of GB and CR during each sampling period (Table S1, Table S3), temporal trends could not be statistically assessed. However, few temporal trends were evident. In GB effluent, the temazepam concentration in March was double (range >100 ng/L) that in December/January, and in CR effluent, the naproxen concentration increased seven-fold over the winter (range >700 ng/L). Also in CR effluent, the roxithromycin concentration in December/January was nine times higher than that in both February/March and March (range 0.8 ng/L), whereas the opposite trend was seen for sulfamethazine (increased three times over time, range 3 ng/L). Fenoprofen, ketoprofen, sulfadimethoxine, sulfisoxazole, and tadalafil were not detected in either urban WWTP above their respective DLs. When the TWA concentrations for each drug were pooled for each WWTP (n ¼ 6 for each WWTP), statistical comparisons revealed no significant differences between urban effluent concentrations for any drug (Table 1). As with the aerated lagoon effluent, there was little evidence of changes in drug concentration over the monthly integrated sampling periods. Seasonality in removal efficiency and/or or drug use patterns could not be assessed for the tertiary plants as only winter data was collected. The concentrations of each drug were similar between the two tertiary plants (Table 1), likely a reflection of similar per capita drug and water use between the populations since treatment processes are nearly identical in the two plants (Letzel et al., 2009). We do not expect concentrations in these large-community WWTPs to be influenced significantly by use changes in a small number of individuals; using the same assumptions as in the previous section, the same individual’s carbamazepine dose change would result in a 0.02 ng/L increase in GB concentrations, an insignificant amount.

3.1.3. Comparison of concentrations in aerated lagoon and tertiary treated effluents in winter Drug concentrations in effluents depend on a number of factors. Removal efficiency during treatment is known to vary by drug and by WWTP (Batt et al., 2007; Letzel et al., 2009). Differences in sewage retention time (Batt et al., 2007; Go¨bel et al., 2007; Kimura et al., 2007), and/or the co-metabolic activity of microbes (Joss et al., 2006; Kim and Carlson, 2007; Kimura et al., 2007; Yu et al., 2006) can produce differences in the efficacy of wastewater treatment for removal of some drugs. For other chemicals, sorption capacity, rather than retention time or biodegradation, can be most important for removal during treatment (Carballa et al., 2007; Joss et al., 2006). Differences in sunlight attenuation while wastewater is in outdoor tanks (tertiary) and lagoons may also play a role for some drugs, as the turbidity of the wastewater may vary, thus varying the amount of sunlight attenuation and thus the amount available for photolysis. Drugs in GB and CR were exposed to sunlight for less than 24 h, and received at least 8 s of low power UV exposure (City of Edmonton, 2009) for disinfection of effluent before release. The drugs in LLB are exposed to sunlight for as long as the daylight hours of 90 days (Duigou, 2006), providing more time for photolysis.

water research 44 (2010) 533–544

Temperature can also play a role, as previously noted. In addition, both GB and CR use secondary activated sludge treatment. These tanks are warmed in the winter to maintain efficiency, especially in cold climates such as that in Alberta. Thus, the water temperatures in those tanks may be higher than that in the LLB lagoon pond under ice (w4 C). As a result, biological activity capable of biodegrading drugs may be more active in GB and CR compared with LLB. Notwithstanding these potential differences in incidental removal during treatment, the concentrations of several drugs detected at all three plants were generally comparable, with CR having the highest total concentration (Table 1). Most drugs (17 out of 24 detectable) were found in all three WWTPs in winter (Table 1). No differences in winter pooled TWA concentrations (three winter sampling periods with n ¼ 9 for LLB, and n ¼ 6 for GB and for CR) among the three effluents for atenolol, celecoxib, codeine, metoprolol, naproxen, paroxetine, temazepam, triclosan and trimethoprim. For eight other drugs, the concentrations in urban effluents were from 2 to 12 times higher than that in rural effluent, with concentration differences that ranged from 9 to 1600 ng/L (Table 1). For citalopram and gemfibrozil, the concentrations at GB were greater than those at LLB. For carbamazepine, citalopram, clarithromycin, diclofenac, erythromycin, gemfibrozil, propranolol, and sotalol, the concentrations in CR effluent were greater than those in LLB effluent. Between the two urban WWTPs, the concentrations of citalopram, propranolol and sotalol at CR were greater than those at GB. Several drugs (fluoxetine, paroxetine, roxithromycin, sildenafil, sulfamethazine, sulfapyridine, vardenafil), infrequently or not detected in the rural WWTP effluent, were consistently detected in the urban WWTP effluents (Tables S2–S3). These drugs were generally present at low concentrations (<4 ng/L) in urban WWTP effluents, with the notable exception of sulfapyridine (>100 ng/L). The discrepancy with sulfapyridine may be explained by differences in drug use among the communities, but is more likely caused by treatment differences. Sulfapyridine is generally administered via sulfasalazine, which is cleaved in the colon to release sulfapyridine (10–35%), while another 30–50% is released as sulfasalazine and a metabolite (Neumann, 1989), both of which may be cleaved in biological wastewater treatment to release sulfapyridine (Go¨bel et al., 2005). It is possible that such cleavage does not occur in the aerated lagoons, leading to a lower concentration of sulfapyridine in LLB effluent. For the other drugs, differences reflect differences between the communities in terms of water and/or drug use, and/or a difference in the fate of drugs in the WWTPs (Hua et al., 2006; Kim and Carlson, 2007; Loraine and Pettigrove, 2006; Sacher et al., 2008; Vieno et al., 2005).

3.2. Loadings of pharmaceuticals to surface waters from aerated lagoon and tertiary treated effluents Average daily loadings (Table S4) in winter were calculated for each plant by multiplying the average daily water discharge by the TWA concentration of each drug during each sampling period. Total winter loadings for GB, CR, and LLB were 700, 400 and 4 g/day, respectively. For the winter sampling periods, average daily loadings from GB were generally twice that from CR. Average daily loadings from LLB were on the order

539

of mg/day. This was at least an order of magnitude lower than loadings from GB and CR, as expected based on the populations served (Letzel et al., 2009). The above result suggests that population-weighted per capita TWA daily loadings would be similar among the three WWTPs. Indeed, this was the case for most drugs, for which there were no statistically significant differences among population-weighted average daily loadings (Fig. 3), e.g., atenolol, celecoxib, clarithromycin, codeine, gemfibrozil, metoprolol, naproxen, temazepam, triclosan, and trimethoprim. Urban per capita average daily loadings were greater than those from the rural WWTP in a few cases: CR was greater than LLB (carbamazepine, citalopram, diclofenac, propranolol, erythromycin, sotalol), GB was greater than LLB (citalopram), while the per capita loading from CR was greater than GB for three drugs (citalopram, propranolol, sotalol). These per capita loading differences, although statistically significant, may not be practically significant, as all were all less than an order of magnitude (two to eight times) and the absolute differences ranged from 2 (propranolol, CR > GB) to 390 mg/person per day (diclofenac, CR > LLB). All of these differences in loadings between WWTPs were also noted as differences in concentration. Loading differences were not found for clarithromycin and gemfibrozil, for which differences in concentration were found among WWTPs. Thus, we conclude that population is a good predictor of human-use drug release into receiving waters, at least for over-the-counter or long-term prescription drugs such as those in this study. Such predictions are likely to be more accurate when based on direct measurements, particularly from demographically similar regions (Stamm et al., 2008), and most accurate when also based on comparable wastewater treatment. In a ten-year study of drugs in the Rhine, concentrations of diclofenac and carbamazepine were found to be relatively constant and given consistency in use patterns, and although raw wastewater concentrations were not measured, the authors attributed any temporal variations to differences in WWTP removal efficiency (Sacher et al., 2008). Their conclusion was that any measures that had been implemented to reduce loadings were thus far ineffective. Because our estimates are based on continuous measurements of pharmaceutical concentrations (Letzel et al., 2009; Miao et al., 2005; Stamm et al., 2008), rather than instantaneous grab sampling, they are likely to be more robust as TWA concentrations from passive samplers are less susceptible to short-term fluctuations in concentration over periods shorter than the sampler deployment periods. Efforts have been undertaken to find a suitable mechanism for removal of drugs from the waste stream (Ikehata et al., 2008), and such efforts should continue, because as our results further highlight, even sophisticated WWTPs like GB and CR produce wastewater with drug concentrations that are comparable with water treated by aerated lagoons. As previously noted, loadings at WWTPs serving small communities are potentially more sensitive to small changes in use patterns than larger communities. The twice-a-day, 100 mg/dose from one individual at LLB would result in a loading increase of 50 mg/person per day at 100% excretion and no WWTP removal, or 1 mg/person per day (4% increase, Fig. 3) using the carbamazepine assumptions previously given. At GB, the equivalent increase in loadings would

540

water research 44 (2010) 533–544

Fig. 3 – Time-weighted average per capita daily loadings (:, mg/person per day) and enantiomer fractions (,, EF), both ± SD, of drugs at three WWTPs: GB, Gold Bar; CR, Capital Region (n [ 6), LLB, Lac La Biche. Differences in daily loadings are indicated by lower-case letters while EF Statistically significant differences are indicated by upper-case letters. Plots without letters do not have loadings or EFs, statistically different from one another over sampling periods. Asterisks indicate EFs that are significantly non-racemic compared with racemic standards. Error bars (SD) represent propagated error from replicate POCIS, analytical processing and instrumental uncertainty, and error associated with POCIS sampling rates.

water research 44 (2010) 533–544

generally be insignificant (Fig. 3), at 0.27 and 0.006 mg/person per day, respectively.

3.3. Enantiomer fractions of chiral drugs in treated effluents 3.3.1. Temporal trends in enantiomer fractions of chiral drugs in aerated lagoon effluent We measured EFs in LLB to discern whether temporal changes existed in enantiomer composition of chiral drugs released by WWTPs. Changes in EF may arise from differential interaction of chiral drug enantiomers with other chiral chemicals (e.g., enzymes). These may result from temporal differences in enantiomer-specific metabolism of patients (Mehvar and Brocks, 2001; Schmekel et al., 1999), and/or from enantiomerspecific, microbial enzyme mediated biotransformations during wastewater treatment, as previously noted for atenolol (MacLeod et al., 2007b; Nikolai et al., 2006), citalopram (MacLeod et al., 2007b) and propranolol (Fono and Sedlak, 2005; MacLeod et al., 2007b; Nikolai et al., 2006), but not metoprolol (Fono and Sedlak, 2005; MacLeod et al., 2007b; Nikolai et al., 2006), salbutamol (MacLeod et al., 2007b), or sotalol (MacLeod et al., 2007b). Because of enantiomer-specific metabolism by the target organism, chiral pharmaceuticals may not necessarily enter the environment with the same enantiomeric composition as the dose. This is in contrast to chiral persistent organic pollutants, which generally enter the environment as a racemate, or sometimes as a single enantiomer. Thus, changes in enantiomeric composition from the excreted EF, not just from a racemic value of 0.5, would be environmentally significant as such a change would indicate post-release biochemical weathering. However, given that few environmental measurements of drug EFs exist to date, our discussion here focuses on changes in EF from the racemate, the most likely enantiomeric composition of formulated chiral drugs not sold as single enantiomers (e.g., naproxen). With the exception of tempazepam, chiral drugs were generally non-racemic in LLB effluent (Fig. 2) compared with racemic standards (Dunnett’s test). Atenolol was enriched in ()-atenolol during some sampling periods, with changes over time, consistent with previous observations in 2005 at LLB (Nikolai et al., 2006). Citalopram was mainly enriched in the first-eluted enantiomer except during August/September. Metoprolol was enriched in the E2-enantiomer when nonracemic, which was not previously observed at LLB in individual grab samples (Nikolai et al., 2006). Salbutamol was always enriched in the second eluted enantiomer, except during January/March wherein it was E1-salbutamol that was enriched. Sotalol was always non-racemic, with no change in EF over time. Temazepam was racemic with one exception, wherein E2-temazepam was enriched. Propranolol EFs could not be measured (supplemental data). Temporal changes in EF were observed (Tukey’s test) for all drugs except sotalol in LLB effluent (Fig. 2, Table S5). Without direct measurement of the EF in influent and during the treatment process, we cannot conclusively determine whether EFs in effluent arose from metabolism prior to bodily excretion or from processes during treatment. The latter may be plausible, given changes in activity and composition of microbial consortia during biological sewage treatment from

541

temperature changes over the seasons. For drugs such as citalopram and salbutamol, changes in EF may also reflect differences in availability or use of single-enantiomer formulations (Maier et al., 2001). Because of the temporal changes in EF observed, our results show that an EF measured at one time point may not necessarily be characteristic of the enantiomer composition of chiral drugs discharged from a WWTP, and hence insufficient to characterize exposures in receiving waters.

3.3.2. Comparison of chiral drug EFs from aerated lagoons and tertiary treated wastewater Winter EFs from LLB and from the urban tertiary WWTPs were compared to gain insight on whether EFs could be predicted by treatment type, and may possibly be caused by similar human metabolism and release and/or microbial degradation during WWTP treatment. Due to low sample sizes, statistical tests and trend analyses were not carried out on the temporal EF data from the urban tertiary WWTPs. Instead, winter EFs were pooled for each drug from GB and CR and compared with EFstandard (Dunnett’s test), to each other (t-test, with Welch’s correction for unequal variances as needed), and to the pooled EFs from the three winter sampling periods at LLB (Tukey’s test). Standard deviations in pooled data reflect propagated variability from both temporal changes in EFsample and from replicates. The two tertiary WWTP effluents had similar EFs during winter for all chiral drugs except sotalol (Fig. 3), suggesting that both treatment plants were processing those drugs in the same stereoselective manner during microbial treatment. At all three WWTPs (Fig. 3), metoprolol EFs were racemic in effluents, while sotalol was enriched in the E2-enantiomer and differed in EF between the tertiary effluents. Racemic metoprolol was previously observed in both GB and LLB effluents (MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006) and in the effluent of a WWTP in Dallas TX (Fono et al., 2006), with no change in EF from raw to treated wastewater (MacLeod et al., 2007b). After correction for enantiomer-specific matrix effects (MacLeod et al., 2007b), sotalol was previously found to be racemic in GB effluent (MacLeod et al., 2007b) via grab samples. This observation suggests that changes in sotalol EFs might have occurred over time in GB effluent; however, we cannot unequivocally conclude that this must be the case given our earlier caveats that a single EF cannot necessarily characterize the enantiomer composition. No statistically significant differences were observed for salbutamol EFs in the three effluents, despite enrichment of E2-salbutamol in both tertiary effluents. This observation is largely a result of the high variability in salbutamol EFs in LLB and CR. In turn, the variability may be a function of the drug’s input to WWTPs, given that salbutamol is available as an R-salbutamol-only formulation, while S-salbutamol is metabolized slower than its antipode (Schmekel et al., 1999). Citalopram (Fig. 3) was enriched in the E1-enantiomer in all effluents, with differences in EF between tertiary and aerated lagoon output. As with salbutamol, this difference could be influenced by differences in prescription patterns, given the availability of escitalopram, the S-enantiomer formulation. Citalopram was previously found with similar EFs (0.55–0.65) in GB effluent (MacLeod et al., 2007b). Temazepam was enriched in the E2-enantiomer at both GB and LLB, but racemic in CR,

542

water research 44 (2010) 533–544

wherein its EF was different from that in LLB, but not from that in GB. No information regarding temazepam EFs in treated effluent is available, except our previous results (MacLeod, 2009). For atenolol, all effluents had EFs that were different from one another, with GB containing racemic atenolol, CR containing more (þ)-atenolol and LLB containing more ()-atenolol. The EF of atenolol from healthy renal excretion is 0.476 (Mehvar and Brocks, 2001), and 90% of atenolol is excreted unchanged in urine (Bourne, 1981). Thus, a process other than healthy renal excretion, and potentially specific to microbial communities of WWTPs, was influencing atenolol EFs in effluent. Likewise, the differences in EFs observed among WWTP effluents is likely due, at least in part, to variations in enantioselective microbial biotransformation of drugs during wastewater treatment. As we have noted, it is not clear what the enantiomeric composition of drugs may be upon excretion from humans, but non-racemic release is likely. As such, we must delineate between environmental biotransformation and in changes in enantiomeric composition of the source material in assessing the environmental fate of drug enantiomers, as has been done for chiral pesticides (e.g., metolachlor) also released in nonracemic amounts (Kurt-Karakus et al., 2008).

4.

Conclusions

Although other studies have used POCIS to monitor drugs in treated wastewater, none have done so for more than 30 days (Jones-Lepp et al., 2004; Zhang et al., 2008). Herein, we monitored LLB effluent for 271 days and found that the concentration of drugs in effluent varied over seasonal time scales. Other studies have mainly focused on larger urban centers. By comparing drugs in both rural and urban wastewater effluents, we find that pharmaceutical loadings to aquatic ecosystems scale with population. However, we also found that the concentrations of drugs in effluent from WWTPs with biological activated sludge treatment and UV disinfection were comparable with concentrations in effluent from simpler treatment methods like aerated lagoons. Although absolute loadings were generally lower, drugs were still detectable in effluent from small communities. Thus, if pharmaceutical contamination of surface water presents a risk to aquatic life, that risk is also present near smaller centers, especially those which are landlocked and not discharging to a large surface water body to provide effluent dilution. Overall, temporal changes in EF were found for all chiral drugs except sotalol, with temporal changes in concentration for all chiral drugs except atenolol and metoprolol. The concentration of atenolol and metoprolol in LLB effluent did not change over time, but the EFs of these drugs did change. Conversely, although the concentration of sotalol in LLB effluent changed over time, the EF remained constant. Although temporal changes in concentration may not be occurring, the relative proportions of enantiomers being released to the environment may be changing over time. Also, for some drugs the EF may remain constant, but the concentration of the drug may change. Similar conclusions can be drawn from the comparison between EFs and per capita daily loadings among the three

WWTPs. Loading and EF differences were found among WWTPs for citalopram and sotalol, while for atenolol and temazepam, there were no differences in loading but there were changes in EF, and for metoprolol there were no differences in loading or EF among the three WWTPs. Simply put, although concentrations and loadings may be accurately predicted based on population-weighted measured effluent concentrations in other treatment plants, the EFs of chiral drugs released by one WWTP may not necessarily reflect that released by another, probably from site-specific biochemical activity. Thus, enantiomer-specific monitoring of chiral drugs may be useful, particularly for drugs with enantiomer-specific toxicity to aquatic life (Stanley et al., 2006).

Acknowledgements We thank D. Bleackley, V. Cooper, M. Ross, B. Asher and E. McClure (University of Alberta), R. Litwinow and D. Seehagel (GB WWTP), D. Cikaluk (CR WWTP) and G. Siebold (LLB WWTP) for sampling assistance, and H. Li (Trent University) for providing unpublished POCIS sampling rates for citalopram and sotalol. Funding was provided to C.S.W. from the Canada Research Chairs Program, Canada’s Natural Sciences and Engineering Research Council (NSERC), and the Society of Environmental Toxicology and Chemistry (SETAC) Early Career Award for Applied Ecological Research, cosponsored by the American Chemistry Council; and to S.L.M. as fellowships from NSERC, Alberta Ingenuity Fund, and the American Chemical Society Division of Analytical Chemistry, sponsored by DuPont.

Appendix. Supplementary data Information about sampling locations and times, analytical and data treatment procedures, concentrations, loadings, and EFs are available in the online supplementary data to Supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2009.09.056.

references

Alberta Capital Region Wastewater Commission, 2009. http:// www.acrwc.ab.ca accessed February 25. Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L., Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environ. Toxicol. Chem. 23, 1640–1648. Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in STP effluents and their solar photodegradation in the aquatic environment. Chemosphere 50, 1319–1330. Batt, A.L., Kim, S., Aga, D.S., 2007. Comparison of the occurrence of antibiotics in four full-scale wastewater treatment plants with varying designs and operations. Chemosphere 68, 428–435. Bourne, G.R., 1981. The metabolism of b-adrenoreceptor blocking drugs. In: Bridges, J.W., Chasseaud, L.F. (Eds.), Progress in Drug Metabolism 6. John Wiley & Sons Ltd, pp. 77–110.

water research 44 (2010) 533–544

Carballa, M., Omil, F., Ternes, T., Lema, J.M., 2007. Fate of pharmaceutical and personal care products (PPCPs) during anaerobic digestion of sewage sludge. Water Res. 41, 2139–2150. Chen, M., Ohman, K., Metcalfe, C., Ikonomou, M.G., Amatya, P.L., Wilson, J., 2006. Pharmaceuticals and endocrine disruptors in wastewater treatment effluents and in the water supply system of Calgary, Alberta, Canada. Water Qual. Res. J. Can. 41, 351–364. City of Edmonton, 2009. http://www.edmonton.ca. accessed February 25. Coetsier, C.M., Spinelli, S., Lin, L., Roig, B., Touraud, E., 2009. Discharge of pharmaceutical products (PPs) through a conventional biological sewage treatment plant: MECs vs PECs? Environ. Int 35, 787–792. Conkle, J.L., White, J.R., Metcalfe, C.D., 2008. Reduction of pharmaceutically active compounds by a lagoon wetland wastewater treatment system in Southeast Louisiana. Chemosphere 73, 1741–1748. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health. Perspect 107, 907–938. Duigou, E., 2006. Town of Lac La Biche, Personal communication. EMEA, 2006. Note for guidance on environmental risk assessment of medicinal products for human use. The European Agency for the Evaluation of Medicinal Products (EMEA) CMPC/SWP/4447/00. Environment Canada, 2009. http://www.ec.gc.ca/soer-ree/ English/Indicator_series/new_issues.cfm?issue_id¼7&tech_ id¼28#bio_pic. accessed February 27. Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of human pharmaceuticals. Aquat. Toxicol 76, 122–159. Fono, L.J., Sedlak, D.L., 2005. Use of the chiral pharmaceutical propranolol to identify sewage discharges into surface waters. Environ. Sci. Technol 39, 9244–9252. Fono, L.J., Kolodziej, E.P., Sedlak, D.L., 2006. Attenuation of wastewater-derived contaminants in a effluent-dominated river. Environ. Sci. Technol 40, 7257–7262. Go¨bel, A., McArdell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate of sulfonamides, macrolides, and trimethoprim in different wastewater treatment methodologies. Sci. Total Environ 372, 361–371. Go¨bel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005. Occurrence and sorption behavior of sulfonamides, macrolides, and trimethoprim in activated sludge treatment. Environ. Sci. Technol 39, 3981–3989. Halling-SBrensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F., Holten-Lu¨tzhBft, H.C., JBrgensen, S.E., 1998. Occurrence, fate and effects of pharmaceutical substances in the environment – a review. Chemosphere 36, 357–393. Harner, T., Wiberg, K., Norstrom, R., 2000. Enantiomer fractions are preferred to enantiomer ratios for describing chiral signatures in environmental analysis. Environ. Sci. Technol 34, 218–220. Hua, W.Y., Bennett, E.R., Miao, X.-S., Metcalfe, C.D., Letcher, R.J., 2006. Seasonality effects on pharmaceuticals and s-triazine herbicides in wastewater effluent and surface water from the Canadian side of the Upper Detroit River. Environ. Toxicol. Chem. 25, 2356–2365. Ikehata, K., El-Din, M.G., Snyder, S.A., 2008. Ozonation and advanced oxidation treatment of emerging organic pollutants in water and wastewater. Ozone Sci. Eng 30, 21–26. Jones-Lepp, T.L., Alvarez, D.A., Petty, J.D., Huckins, J.N., 2004. Polar organic chemical integrative sampling and liquid chromatography-electrospray/ion-trap mass spectrometry for assessing selected prescription and illicit drugs in treated sewage effluents. Arch. Environ. Contam. Toxicol 47, 427–439. Joss, A., Zabcynski, S., Go¨bel, A., Hoffmann, B., Lo¨ffler, D., McArdell, C.S., Ternes, T.A., Thomsen, A., Siegrist, H., 2006. Biological degradation of pharmaceuticals in municipal

543

wastewater treatment: proposing a classification scheme. Water Res. 40, 1686–1696. Kim, S.-C., Carlson, K., 2007. Temporal and spatial trends in the occurrence of human and veterinary antibiotics in aqueous and river sediment matrices. Environ. Sci. Technol 41, 50–57. Kimura, K., Hara, H., Watanabe, Y., 2007. Elimination of selected acidic pharmaceuticals from municipal wastewater by an activated sludge system and membrane bioreactors. Environ. Sci. Technol 41, 3708–3714. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in U.S. streams, 1999–2000: a national reconnaissance. Environ. Sci. Technol 36, 1202–1211. Kurt-Karakus, P.B., Bidleman, T.F., Muir, D.C.G., Cagampan, S.J., Struger, J., Sverko, E., Small, J.M., Jantunen, L.M., 2008. Chiral current-use herbicides in Ontario streams. Environ. Sci. Technol 42, 8452–8458. Lajeunesse, A., Gagnon, C., Sauve´, S., 2008. Determination of basic antidepressants and their N-desmethyl metabolites in raw sewage and wastewater using solid-phase extraction and liquid chromatography-tandem mass spectrometry. Anal. Chem. 80, 5325–5333. Letzel, M., Metzner, G., Letzel, T., 2009. Exposure assessment of the pharmaceutical diclofenac based on long-term measurements of the aquatic input. Environ. Int 35, 363–368. Lin, A.Y.-C., Reinhard, M., 2005. Photodegradation of common environmental pharmaceuticals and estrogens in river water. Environ. Toxicol. Chem. 24, 1303–1309. Lin, A.Y.-C., Plumlee, M.H., Reinhard, M., 2006. Natural attenuation of pharmaceuticals and alkylphenol polyethoxylate metabolites during river transport: photochemical and biological transformation. Environ. Toxicol. Chem. 25, 1458–1464. Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J., Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006. Occurrence and reductions of pharmaceuticals and personal care products and estrogens by municipal wastewater treatment. Sci. Total Environ 367, 544–588. Lissemore, L., Hao, C., Yang, P., Sibley, P.K., Mabury, S., Solomon, K.R., 2006. An exposure assessment for selected pharmaceuticals within a watershed in Southern Ontario. Chemosphere 64, 717–729. Loraine, G.A., Pettigrove, M.E., 2006. Seasonal variations in concentrations of pharmaceuticals and personal care products in drinking water and reclaimed wastewater in Southern California. Environ. Sci. Technol 40, 687–695. MacLeod, S.L., 2009. Pharmaceutical occurrence and fate in wastewater and receiving surface waters in two Alberta watersheds. PhD dissertation. University of Alberta, Edmonton, Alberta. MacLeod, S.L., McClure, E.L., Wong, C.S., 2007a. Laboratory calibration and field deployment of the Polar Organic Chemical Integrative Sampler for pharmaceuticals and personal care products in wastewater and surface water. Environ. Toxicol. Chem. 26, 2517–2529. MacLeod, S.L., Sudhir, P., Wong, C.S., 2007b. Stereoisomer analysis of wastewater-derived beta-blockers, selective serotonin re-uptake inhibitors, and salbutamol by highperformance liquid chromatography-tandem mass spectrometry. J. Chromatogr. A 1170, 23–33. Maier, N.M., Franco, P., Lindner, W., 2001. Separation of enantiomers: needs, challenges, perspectives. J. Chromatogr. A 906, 3–33. Managaki, S., Murata, A., Takada, H., Tuyen, B.C., Chiem, N.H., 2007. Distribution of macrolides, sulfonamides, and trimethoprim in tropical waters: ubiquitous occurrence of

544

water research 44 (2010) 533–544

veterinary antibiotics in the Mekong Delta. Environ. Sci. Technol 41, 8004–8010. McArdell, C.S., Molnar, E., Suter, M.J.-F., Giger, W., 2003. Occurrence and fate of macrolide antibiotics in wastewater treatment plants and in the Glatt Valley Watershed. Switzerland. Environ. Sci. Technol 37, 5479–5486. Mehvar, R., Brocks, D.R., 2001. Stereospecific pharmacokinetics and pharmacodynamics of beta-adrenergic blockers in humans. J. Pharm. Pharm. Sci. 4, 185–200. Miao, X.-S., Yang, J.-J., Metcalfe, C.D., 2005. Carbamazepine and its metabolites in wastewater and in biosolids in a municipal wastewater treatment plant. Environ. Sci. Technol 39, 7469–7475. Neumann, J., 1989. Untersuchungen zur Bioverfu¨gbarkeit und Pharmakokinetik von Sulfasalazin und seinen Metaboliten. PhD thesis. Free University of Berlin. Nikolai, L.N., McClure, E.L., MacLeod, S.L., Wong, C.S., 2006. Stereoisomer quantification of the b-blocker drugs atenolol, metoprolol, and propranolol in wastewaters by chiral highperformance liquid chromatography-tandem mass spectrometry. J. Chromatogr. A 1131, 103–109. Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008. Pharmaceutical residues in the river Rhine-results of a onedecade monitoring programme. J. Environ. Monit. 10, 664–670. Schmekel, B., Rydberg, I., Norlander, B., Sjo¨swa¨rd, K.N., Ahlner, J., Anderson, R.G.G., 1999. Stereoselective pharmacokinetics of S-salbutamol after administration of the racemate in healthy volunteers. Eur. Respir. J 13, 1230–1235. Shaw, M., Mueller, J.F., 2009. Time integrative passive sampling: how well do Chemcatchers integrate fluctuating pollutant concentrations? Environ. Sci. Technol 43, 1443–1448. Siebold, G., 2008. Lac La Biche County, Personal communication.

Stamm, C., Alder, A.C., Fenner, K., Hollender, J., Kraus, M., McArdell, C.S., Ort, C., Schneider, M.K., 2008. Spatial and temporal patterns of pharmaceuticals in the aquatic environment. Geog. Compass 2/3, 920–955. Stanley, J.K., Ramirez, A.J., Mottaleb, M., Chambliss, C.K., Brooks, B.W., 2006. Enantiospecific toxicity of the b-blocker propranolol to Daphnia magna and Pimephales promelas. Environ. Toxicol. Chem. 25, 1780–1786. Statistics Canada, 2009. http://www12.statcan.ca. accessed February 25. Stu¨lten, D., Zu¨hlke, S., Lamsho¨ft, M., Spiteller, M., 2008. Occurrence of diclofenac and selected metabolites in sewage effluents. Sci. Total Environ 405, 310–316. Vieno, N.M., Tuhkanen, T., Kronberg, L., 2005. Seasonal variation in the occurrence of pharmaceuticals in effluents from a sewage treatment plant and in the recipient water. Environ. Sci. Technol 39, 8220–8226. Vasskog, T., Anderssen, T., Pedersen-Bjergaard, S., Kallenborn, R., Jensen, E., 2008. Occurrence of selective serotonin reuptake inhibitors in sewage and receiving waters at Spitsbergen and in Norway. J. Chromatogr. A 1185, 194–205. Wong, C.S., 2006. Environmental fate processes and biochemical transformations of chiral emerging organic pollutants. Anal. Bioanal. Chem. 386, 544–558. Yu, J.T., Bouwer, E.J., Coelhan, M., 2006. Occurrence and biodegradability studies of selected pharmaceuticals and personal care products in sewage effluent. Agric. Water Mgmt 86, 72–80. Zhang, Z., Hibberd, A., Zhou, J.L., 2008. Analysis of emerging contaminants in sewage effluent and river water: comparison between spot and passive sampling. Anal. Chim. Acta 607, 37–44.

water research 44 (2010) 545–554

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Degradation of fifteen emerging contaminants at mg LL1 initial concentrations by mild solar photo-Fenton in MWTP effluents N. Klamerth a,b, L. Rizzo c, S. Malato a,*, Manuel I. Maldonado a, A. Agu¨era b, A.R. Ferna´ndez-Alba b a

Plataforma Solar de Almerı´a-CIEMAT. Carretera Sene´s km 4, 04200 Tabernas (Almerı´a), Spain Pesticide Residue Research Group, University of Almerı´a, 04120 Almerı´a, Spain c Department of Civil Engineering, University of Salerno, via Ponte don Melillo, 84084 Fisciano (SA), Italy b

article info

abstract

Article history:

The degradation of 15 emerging contaminants (ECs) at low concentrations in simulated and

Received 9 March 2009

real effluent of municipal wastewater treatment plant with photo-Fenton at unchanged pH

Received in revised form

and Fe ¼ 5 mg L1 in a pilot-scale solar CPC reactor was studied. The degradation of those

23 September 2009

15 compounds (Acetaminophen, Antipyrine, Atrazine, Caffeine, Carbamazepine, Diclofe-

Accepted 28 September 2009

nac, Flumequine, Hydroxybiphenyl, Ibuprofen, Isoproturon, Ketorolac, Ofloxacin, Proges-

Available online 4 October 2009

terone, Sulfamethoxazole and Triclosan), each with an initial concentration of 100 mg L1, was found to depend on the presence of CO2 3 and HCO3 (hydroxyl radicals scavengers) and

Keywords:

on the type of water (simulated water, simulated effluent wastewater and real effluent

Emerging contaminants

wastewater), but is relatively independent of pH, the type of acid used for release of

Pharmaceuticals treatment

hydroxyl radicals scavengers and the initial H2O2 concentration used. Toxicity tests with

Photo-Fenton

Vibrio fisheri showed that degradation of the compounds in real effluent wastewater led to

Solar photocatalysis

toxicity increase. ª 2009 Elsevier Ltd. All rights reserved.

1.

Introduction

Due to their growing use, pharmaceuticals, like the antiinflammatory ibuprofen, the antibiotic flumequine and the antiepileptic carbamazepine, endocrine disruptors like bisphenol A and atrazine, personal care products like oxybenzone and parabens (PHBA), synthetic musks and fragrances like musk xylene and galaxolide, pesticides like isoproturon and endosulphan and illicit drugs like THC aznd cocaine, to name just a few, and other xenobiotic substances, are found in increasing quantities in wastewater, surface water, and even in drinking water (Kasprzyk-Hordern et al., 2009; Kim et al., 2007; Mitch et al., 2003; Esplugas et al., 2007; Ternes, 1998). Since the use of these substances cannot be controlled or eliminated as they are ever present in our daily

lives, their release into the environment has to be optimized and restricted, as they pose risks to the environment, public health and aquatic systems and they are responsible for building up microbiological resistance, feminisation of higher organisms and ecotoxicological issues (Laville et al., 2004). Particularly relevant examples of such emerging contaminants (ECs), such as those mentioned above, which are ubiquitously present in influents and effluents of MWTPs in the high ng L1 to low mg L1 range, do not need to be persistent to be hazardous, because they are introduced continuously into the environment (Fono et al., 2006; Jackson and Sutton, 2008; Nakada et al., 2008; Petrovic et al., 2003). Conventional MWTPs, typically based on biological processes, are capable of removing some substances, but nonbiodegradable compounds may escape the treatment and be

* Corresponding author. Tel.: þ34 950387940; fax: þ34 950365015. E-mail address: [email protected] (S. Malato). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.059

546

water research 44 (2010) 545–554

released into the environment (Go¨bel et al., 2007; Carballa et al., 2004; Ternes et al., 2007). ECs have been found in the MWTP effluents at mean concentrations ranging from 0.1 to 20 mg L1 (Castiglioni et al., 2006; Martı´nez Bueno et al., 2007; Radjenovic´ et al., 2007; Richardson, 2007; Zhao et al., 2009). Concern about the growing problem of the continuously rising concentrations of these compounds must be emphasized, and therefore, the application of more thorough wastewater treatment protocols, including the use of new and improved technologies, is a necessary task. Reusable water should be free of these persistent, toxic, endocrine-disrupting or non-biodegradable substances, (Radjenovic´ et al., 2007; Teske and Arnold, 2008), and therefore, an effective tertiary treatment is required to remove these substances completely. Among the advanced technologies that may be used to remove these pollutants, (Gogate and Pandit, 2004; Saritha et al., 2007; Huber et al., 2005) advanced oxidation processes (AOPs), through the generation of hydroxyl radicals which are able to mineralise most organic molecules yielding CO2 and inorganic ions as final products, are a particularly attractive option. (Farre´ et al., 2005; Gebhardt and Schro¨der, 2007; Gu¨ltekin and Ince, 2007; Ning et al., 2007; Rosenfeldt and Linden, 2004; Rosenfeldt et al., 2007; Ternes et al., 2003) The generation of the OH radicals can be achieved: electrochemically (Can˜izares et al., 1999; Pelegrini et al., 2001; Zhou et al., 2005), sonochemically (Mantzavinos et al., 2004; Papadaki et al., 2004; Lesko et al., 2006), photochemically (Esplugas et al., 2005; Bali et al., 2003; Bremner et al., 2006), and by homogeneous or heterogenous catalysis (Zepp et al., 1992; Martinez et al., 2005) in acid or basic media (Glaze et al., 1987; Hislop and Bolton, 1999; Neyens and Baeyens, 2003). Most of the AOPs make use of a combination of either oxidants and irradiation (O3/H2O2/ UV), or a catalyst and irradiation (Fe2þ/H2O2; UV/TiO2). The drawbacks which make them economically disadvantageous depend on the specific AOP: (i) High electricity demand (e.g. ozone and UV-based AOPs), (ii) the relatively large amounts of oxidants and/or catalysts (e.g. ozone, hydrogen peroxide and iron-based AOPs), and (iii) the pH operating conditions (e.g. Fenton and photo-Fenton). This is why, although AOPs are well known for their capacity for oxidising and mineralising almost any organic contaminant, commercial applications are still scarce. Processes like photo-Fenton may be applied to commercial applications by using solar energy as a light source, optimizing the pH range and the amounts of oxidant/ catalyst required. AOP efficiency in the removal of ECs has typically been studied in demineralised water and bench scale, at initial concentrations in the milligram-to-gram range, which is not realistic compared to the concentrations detected in real water and wastewater (Farre´ et al., 2005; Lapertot et al., 2007; Kassinos et al., 2009; Malato et al., 2007). This work focused on solar photo-Fenton degradation of the ECs typically found in the effluents of MWTPs, leaving the treated wastewater suitable for reuse. Moreover, to make the process of interest for practical applications high iron concentrations (mM range), excessive amounts of H2O2 and a pH under 3 must be avoided (Pignatello et al., 2006). A new approach aimed at finding a very mild photo-Fenton treatment (low iron concentration and H2O2 dose at neutral pH), has been proposed (Moncayo-

Lasso et al., 2008; Klamerth et al., 2009). In this paper, a pilotscale solar photo-Fenton treatment was run at a pH between 6 and 7, with starting concentrations of 5 mg L1 Fe, and 50 mg L1 H2O2. Synthetic water (SW), simulated effluent wastewater (SE) and real effluent wastewater (RE) were tested in this study to which a mixture of 15 ECs, consisting of pharmaceuticals, pesticides and personal care products, selected from a list of 80 compounds found in MWTP effluents in previous studies (Martı´nez Bueno et al., 2007), were added at low concentrations (100 mg L1 each).

2.

Material and methods

2.1.

Reagents

All reagents used for chromatographic analyses, acetonitrile, methanol, and ultrapure (MilliQ) water, were HPLC grade. Analytical standards for chromatography analyses were purchased from Sigma-Aldrich. Table 1 and Scheme 1 list the 15 compounds (pharmaceuticals, pesticides and personal care products) used. Photo-Fenton experiments were performed using iron sulphate (FeSO4$7H2O), reagent-grade hydrogen peroxide (30% w/v), sulphuric acid and hydrochloric acid for carbonate stripping, all provided by Panreac. The filters used were syringe driven 0.2 mm Millex nylon membrane filters from Millipore. The reagents used in the preparation of the synthetic water (NaHCO3, CaSO4$2H2O, MgSO4, and KCl), and simulated effluent wastewater (Peptone, Meat extract, Urea, K2HPO4, CaCl2$2H2O, MgSO4$7H2O and NaCl) were provided by Panreac.

2.2.

Polluted waters

Taking into consideration that typical EC concentrations in the effluent are in the 0.1–20.0 mg L1 range, it was decided to work at 100 mg L1, which is a compromise between (i) a high enough concentration to characterise kinetics using conventional

Table 1 – LOD, LOQ and absorption l of the selected compounds. Name Acetaminophen Antipyrine Atrazine Caffeine Carbamazepine Diclofenac Flumequine Hydroxybiphenyl Ibuprofen Isoproturon Ketorolac Ofloxacin Progesterone Sulfamethoxazole Triclosan

LOD (mg L1)

LOQ (mg L1)

Absorption l [nm]

1.4 2.1 0.6 0.7 0.8 2.7 2.1 1.6 2.5 1.5 2.1 1.3 2.0 1.4 5.0

2.8 4.2 1.2 1.5 1.6 5.5 4.3 3.2 5.0 3.0 4.2 2.6 4.0 2.7 10.0

245 205 223 205 211 277 248 243 222 205 321 295 248 267 280

547

water research 44 (2010) 545–554

Name

Structure

Name Ibuprofen

Acetaminophen HO

analgesic

Structure

NH

nonsteroidal

/

O

antipyretic

anti-

OH

inflammatory

O

O

Isoproturon

Antipyrine

H N N

analgesic

phenylurea

N

O

herbicide

H N

N

O

H N

N

Ketorolac

Atrazine

N N

N

OH

anti-

herbicide

inflammatory

Cl

O N

O

Caffeine

F

Ofloxacin

N

gramnegative

O

N

N

O

stimulant

O

N

N

antibiotic

N

O

OH

O

Progesterone

Carbamazepine N

steroid

anticonvulsant

hormone

NH2

O

O O

Diclofenac Cl

O

Sulfamethoxaz OH

ole

H2N

S

H N

anti-

H N N

O

O

bacteriostatic

inflammatory

antibiotic

Cl O

Flumequine

F

COOH

Cl

Triclosan

OH O

broadspectrum

anti-bacterial/ N

antibiotic

Hydroxybiphenyl

fungal agent

Cl

OH

biocide

Scheme 1 – Name and structure of the 15 selected compounds.

Cl

548

water research 44 (2010) 545–554

1.0

Acetaminophen Caffeine Ofloxacine Antipyrine Sulfamethoxazole

illumination

0.5 Fenton

Carbamazepine Flumequine Ketorolac Atrazine Isoproturon

0.5

0.0 1.0

Hydroxybenzone Diclofenac Ibuprofen Progesterone Triclosan

0.5

0.0 -40

-20

0

20

tEXP, min

40

60

80

100

t30W, min

Fig. 1 – Degradation of the 15 ECs (0.1 mg LL1 each) by photo-Fenton with 5 mg LL1 Fe without pH adjustment in SW acidified for release of carbonates.

2.3.

analytical techniques, and (ii) low enough to simulate real conditions. Synthetic water (SW), simulated effluent wastewater (SE) and real effluent wastewater (RE) to which a mixture of the 15 ECs previously listed had been added at low concentrations (100 mg L1) were tested. The experimental protocol was designed to study solar photo-Fenton with a relatively uncomplicated aqueous matrix (SW) first, before gradually increasing complexity by studying SE and finally RE, in order to acquire information on process operation. Demineralised water used for preparing SW and SE in the pilot plant was supplied by the Plataforma Solar de Almerı´a

1.0

Solar photo-Fenton pilot plant

Photo-Fenton experiments were performed at the Plataforma Solar de Almerı´a in a pilot compound parabolic collector (CPC) solar plant designed for solar photocatalytic applications. This reactor is composed of two 11dL modules with twelve Pyrex glass tubes (30 mm O.D.) mounted on a fixed platform tilted 37 (local latitude). The water flows (20 L min1) directly from one module to the other and finally to a 10dL reservoir. The piping and valves (3 L) between the reactor and the tank are black HDPE, which is highly resistant to chemicals, weatherproof and opaque, preventing any photochemical effect from outside the collectors. The total illuminated area is 3 m2, the total volume (two modules þ reservoir tank þ piping and valves) is 35 L (VT) and the irradiated volume is 22 L (Vi). Solar ultraviolet radiation (UV) was measured by a global UV radiometer (KIPP&ZONEN, model CUV 3) mounted on a platform tilted 37

Acetaminophen Caffeine Ofloxacine Antipyrine Sulfamethoxazole

illumination

0.5

C/C0

0.0 1.0

Carbamazepine Flumequine Ketorolac Atrazine Isoproturon

0.5

0.0 1.0

Fenton 120

consumed H2O2

0.5

Hydroxybenzone 90 Diclofenac 60 Ibuprofen Progesterone 30 Triclosan

0.0

0 0

tEXP, min

60

120

180

240

300

consumed H2O2, mg L-1

C/C0

0.0 1.0

(PSA) distillation plant (conductivity < 10 mS cm1, Cl ¼ 0.7– 1 1 0.8 mg L1, NO 3 ¼ 0.5 mg L , organic carbon < 0.5 mg L ). The preparation of standard moderately-hard freshwater (SW) was carried out taking into account the characteristics of ground water in the province of Almerı´a (southern Spain). SW was prepared by mixing 96 mg L1 of NaHCO3, 60 mg L1 of CaSO4$2H2O, 60 mg L1 of MgSO4, and 4 mg L1 of KCl (Standard Methods, 1998). The simulated effluent wastewater (SE) consists of SW and Peptone (32 mg L1), Meat extract (22 mg L1), Urea (6 mg L1), K2HPO4 (28 mg L1), CaCl2$2H2O (4 mg L1), MgSO4$7H2O (2 mg L1) and NaCl (7 mg L1) which derives to an initial DOC (dissolved organic carbon) of 25 mg L1 (OECD, 1999). RE was taken downstream of the Almerı´a MWTP secondary biological treatment and used as received within the next 2 days. Initial COD and DOC were at 60 mg L1 and 25 mg L1, respectively.

t30W, min

Fig. 2 – Degradation of the 15 ECs (0.1 mg LL1 each) by photo-Fenton (5 mg LL1 Fe, no pH adjustment) and hydrogen peroxide consumption in SE acidified for carbonate release.

549

water research 44 (2010) 545–554

Table 2 – Final concentration of ECs in two different experiments with SE after similar irradiation times, with [Fe]0 [ 5 mg LL1: 300 min with [H2O2]0 [ 50 mg LL1 and addition of 50 mg LL1 when the initial dose was almost consumed and so on; 336 min with [H2O2]0 [ 5 mg LL1 and addition of H2O2 every 30–60 min until the end of the experiment. Fenton degradation (%) in the dark and first order kinetic constants (k) for photo-Fenton are also included. [H2O2]0 ¼ 50 mg L1

[H2O2]0 ¼ 5 mg L1

t30W

C/C0

Fenton deg. X [%]

k [min1]

t30W

C/C0

Fenton deg. X [%]

k [min1]

Acetaminophen Caffeine Ofloxacin Antipyrine Sulfamethoxazole Carbamazepine Flumequine Ketorolac Atrazine Isoproturon Hydroxybiphenyl Diclofenac Ibuprofen Progesterone Triclosan

261 300 56 300 300 300 56 191 300 300 300 111 261 300 300

16.1 21.7 63.2 10.5 10.2 18.0 39.3 21.8 6.1 12.4 15.6 35.2 11.3 18.3 14.9

0.016 0.008 0.063 0.008 0.012 0.014 0.078 0.022 0.004 0.016 0.012 0.032 0.011 0.011 0.005

248 336 114 336 336 336 114 248 336 336 336 133 294 336 336

< LOD 0.09
14.1 1.2 56.7 3.8 4.5 6.6 30.0 11.6 0.0 6.9 6.2 20.1 11.3 8.9 15.6

0.019 0.007 0.034 0.007 0.010 0.013 0.036 0.017 0.004 0.011 0.010 0.028 0.010 0.011 0.005

Conditions TOC [mg L1] Used H2O2 [mg L1]

Initial 37.8

End 31.1 110

Initial 39.9

(the same as the CPCs). The temperature inside the reactor was continuously recorded by a temperature probe (Crioterm PT-100 3H) inserted in the piping. A plant diagram has been published elsewhere (Kositzi et al., 2004). With Eq. (1), combination of the data from several days’ experiments and their comparison with other photocatalytic experiments is possible,

lasted as long as there was H2O2 to be consumed (usually 3–4 h), in the case of experiments which lasted longer than that, the experiment was covered at the end of the day and the next day, after recirculation for 15 min, a sample was taken, peroxide was added (50 mg L1) and the covers again removed.

2.5. t30W;n

UV Vi ¼ t30W;n1 þ Dtn ; Dtn ¼ tn tn1 ; t0 ¼ 0ðn ¼ 1Þ 30 VT

(1)

where tn is the experimental time for each sample, UV is the average solar ultraviolet radiation (l < 400 nm) measured between tn1 and tn, and t30W is a ‘‘normalized illumination time’’. In this case, time refers to a constant solar UV power of 30 Wm2 (typical solar UV power on a perfectly sunny day around noon).

2.4.

Experimental setup

The mixture of the 15 compounds (100 mg L1 each) dissolved in Methanol at 2.5 g L1 (mother solution) was added directly (1.4 mL) into the pilot plant and well homogenized by turbulent recirculation for 15 min. Methanol was used to dissolve the contaminants (TOC from methanol was 12 mg L1). The pH in SW was 7.6, in SE 7.8 and in RE 8.0, which favoured bicarbonate ions, so at the beginning of the process, when the collectors and HCO were still covered, enough acid to remove CO2 3 3, known OH radical scavengers, was added. Recirculation time for this process was between 15 and 60 min. When the total inorganic carbon (TIC) concentration was below the desired level (10 mg L1), 50 mg L1 of peroxide was added and homogenised by recirculation for 15 min. Finally, Fe2þ as FeSO4$7H20 was added (5 mg L1). After 15 min of recirculation, in which the Fenton reaction started, the collectors were uncovered and photo-Fenton began. Hydrogen peroxide and iron were measured in every sample taken. The experiments normally

End 38.7 52

Analytical measurements

Experiments were performed with 100 mg L1 of each of 15 different compounds in synthetic water (SW), simulated effluent wastewater (SE) and real effluent wastewater (RE). Therefore, no study of degradation intermediates or mineralisation was done, as under these conditions, it is really almost impossible to monitor degradation intermediates. Wastewater background DOC was in the range of tens of mg L1, and all 15 compounds together at concentrations of 100 mg L1 produce no more than 1 mg L1 of DOC. Therefore, results of DOC degradation with EC mineralisation would be meaningless. Dissolved Organic Carbon (DOC) and total inorganic carbon (TIC) were measured by direct injection of samples filtered with 0.2 mm syringe-driven filters into a Shimadzu – 5050A TOC analyser. Colorimetric determination of total iron concentration was used with 1,10-phenantroline according to ISO 6332. Hydrogen peroxide was analyzed by a fast, simple spectrophotometric method using ammonium metavanadate, which allows the H2O2 concentration to be determined immediately based on a red-orange peroxovanadium cation formed during the reaction of H2O2 with metavanadate, with a maximum absorption at 450 nm. The peroxide concentrations are calculated from absorption measurements according to the method proposed by Nogueira et al. (2005). The concentration profile of each compound during degradation was determined by UPLC-UV (series 1200, Agilent technologies, Palo Alto, CA) using a method specifically developed for this application (Martı´nez-Bueno et al., 2007).

550

water research 44 (2010) 545–554

Table 3 – Final EC concentration in RE with [Fe]0 [ 5 mg LL1, [H2O2]0 [ 50 mg LL1 after an irradiation time of t30W [ 276 min. Fenton degradation (%) in the dark and first order kinetic constant (k) for photo-Fenton are also included. t30W

C/C0

Fenton deg. X [%]

k [min1]

Acetaminophen Caffeine Ofloxacin Antipyrine Sulfamethoxazole Carbamazepine Flumequine Ketorolac Atrazine Isoproturon Hydroxybiphenyl Diclofenac Ibuprofen Progesterone Triclosan

109 158 102 276 198 158 102 43 276 122 109 64 109 158 276

16.1 36.4 11.4 21.3 20.5 19.7 23.1 33.3 1.4 15.3 14.7 38.0 10.1 23.7 15.0

0.030 0.020 0.042 0.015 0.025 0.026 0.040 0.107 0.009 0.032 0.033 0.066 0.030 0.024 0.005

Conditions TOC [mg L1] Used H2O2 [mg L1]

Initial 55

the wavelength of maximum light absorption of each compound with the response limits of detection (LODs) and limits of quantification (LOQs) as shown in Table 1.

2.6.

Toxicity test with Vibrio fisheri

Photo-Fenton experiments with SE and RE were repeated for toxicity test sampling. During the experiments, 0.5 mL of H2O2 (which corresponds to the original H2O2 concentration of 5 mg L1) were added every 30 min. The time had been previously estimated to be long enough to make H2O2 residual totally disappear; so, after 30 min a sample was taken for toxicity testing and then more H2O2 was added. The pH of the samples was adjusted to 7. Toxicity tests based on the inhibition of the luminescence emitted by the marine bacteria V. fisheri, were carried out using a Biofix Lumi-10. The reagent is a freeze-dried preparation of a specially selected strain of the marine bacterium V. fisheri (formerly known as Photobacterium phosphoreum, NRRL number B-11177). The drop in light emission is measured after 30-min contact periods.

End 34.5 45

Analytes were separated using a reversed-phase C-18 analytical column (XDB-C-18 1.8 mm, 4.6 50 mm) using acetonitrile (mobile phase A) and ultrapure water (25 mM formic acid, mobile phase B) at a flow rate of 1 mL min1. A linear gradient progressed from 10% A (original conditions) to 82% A in 12 min. Re-equilibration time was 3 min. A 25 mL sample was filtered through a 0.2dmm syringe-driven filter, and the filter was washed with 3 mL acetonitrile in order to remove any adsorbed compounds. The two solutions were mixed and a 100dmL aliquot was injected into the UPLC. The UV signal was recorded at different wavelengths according to

3.

Results and discussion

3.1.

Photo-Fenton tests using SW

Previous experiments have confirmed that low iron concentrations without pH adjustment can lead to extremely slow degradation kinetics because of the presence of carbonate and HCO species (CO2 3 3 ) which compete with organic contaminants for hydroxyl radicals (Klamerth et al., 2009). Therefore, SW was acidified (41 mg L1 HCl) to remove carbonates prior to photo-Fenton. 15 min recirculation in the pilot plant was necessary to remove carbonates (final TIC ¼ 5 mg L1) with pH between 5.2 and 6.0. Degradation of pollutants (Fig. 1) started in the dark (Fenton), 15 min before uncovering the

100

2

1

3

inhibition, %

75 50 25 0

SE RE

-25 -50 1.0

0

10

20

30

40

Antrazine Diclofenac Carbamazepine Ketorolac

0.5

C/C0

50 SE

0.0 1.0

RE Atrazine Diclofenac Carbamazepine Ketorolac

0.5 0.0 0

10

20

30

40

50

dose H2O2, mg L-1 Fig. 3 – Toxicity tests and correlation of toxicity with concentrations of 4 selected compounds.

water research 44 (2010) 545–554

reactor (tExP ¼ 15 min). Removal of all compounds but atrazine (18%) was in the range between 30% (caffeine) and 60% (ibuprofen). Fenton degradation was very quick after adding just 5 mg L1 of Fe2þ but it did not proceed further until the reactor was uncovered, as Fe3þ reduction to Fe2þ was inefficient without illumination. All the pollutants but triclosan (87%) and atrazine (77%) were efficiently degraded (removal over 94%) under photoFenton conditions, at t30W ¼ 90 min. Moreover, some pollutants (ofloxacin and diclofenac) were degraded to below the LOD as early as t30W ¼ 32 min. The total amount of H2O2 consumed was 37 mg L1 and the iron concentration decreased due to precipitation from 5 mg L1 to 3.1 mg L1, probably due to colloidal aggregation, while the pH varied from an original 7.6–5.3 (after adding the FeSO4$7H2O) to a final pH ¼ 3.8 at t30W ¼ 90 min. Similar results were found by adding H2SO4 (56 mg L1) instead of HCl. H2SO4 has certain advantages (handling, volatility, etc.) over HCl, and Cl ions may act as radical scavengers. Experiments did not show any significant difference between using HCl or H2SO4. Indeed, the total concentration of chloride and sulphate measured in SW was 43 mg L1 Cl when HCl was added and 145 mg L1 SO2 4 when H2SO4 was added respectively. It is well known that inorganic species (chloride, sulphate, phosphates, etc) are usually detrimental to the photo-Fenton reaction rate (Pignatello et al., 2006) due to complexation of these ions with Fe2þ or Fe3þ, or to the scavenging of hydroxyl radicals and formation of less reactive inorganic ions (De Laat et al., 2004), but it has also been observed that they significantly reduce the photoFenton reaction rate only if they occur at concentrations over 500 mg L1, usually very far from the concentration found in natural water (Bacardit et al., 2007; Zapata et al., 2009).

3.2.

Photo-Fenton tests using SE

The following experiment dealt with the photo-Fenton treatment of SE, which was characterised by original TOC and COD of 37 mg L1 and 80 mg L1, respectively. Two different approaches to the H2O2 dose were studied: (i) 50 mg L1 were added at the beginning and again when the initial dose was almost consumed and so on; (ii) 5 mg L1 were added every 30–60 min (also depending on consumption) until the end of the experiment. In this case, 56 mg L1 H2SO4 acid were added. A recirculation time of 15 min was necessary for carbonates to be released (TIC < 5 mg L1). As can be seen in Fig. 2, degradation is much slower with both Fenton (t < 0) and photoFenton (t30W > 0) than in the experiment with SW shown in Fig. 1. The residual concentration of the contaminants at the end of the experiments (t30W ¼ 300 min and 336 min, respectively) can be seen in Table 2. The pH varied during the experiment with high H2O2 concentrations from the original 7.6–4.9 at the end, while the iron concentration varied from 5 mg L1 to 3.2 mg L1. During the experiment with low H2O2 concentrations the pH varied from 8.1 to 4.9, while iron concentration went from 4.8 to 3.1 mg L1. The EC degradation behaviour did not change significantly, no matter which hydrogen peroxide dose approach was used, as can be seen in Table 2, although the overall amount of peroxide consumed was much lower (110 mg in 50-mg-L1 additions compared to 52 mg with the 5-mg-L1 additions). It should be mentioned, although not relevant to the purpose of the experiments,

551

because the main purpose was to degrade ECs, that TOC mineralisation was rather low in both cases (around 25%).

3.3.

Photo – Fenton treatment of RE

The initial DOC, TIC and COD were 36 mg L1, 106 mg L1 and 60 mg L1, respectively. In this case, 406 mg L1 H2SO4 were added to reach < 20 mg L1 TIC and recirculation time in the pilot plant necessary to remove carbonate species was 30 min. According to previous tests, photo-Fenton treatment at a TIC over 20 mg L1 leads to very slow degradation. Interestingly, although the degradation rate of all ECs was faster than in the previous experiments in SE, as observed in Table 3, DOC degradation was similar to SE (around 20%). This can be explained by the presence of humic acids in RE which produce solvated electrons and hydroxyl radicals upon irradiation (Fukushima and Tatsumi, 2001; Prosen and Zupancˇicˇ-Kralj, 2005; Auger et al., 2002).

3.4. The effect of photo-Fenton treatment on V. fisheri toxicity In Fig. 3, the results of V. fisheri toxicity tests with SE and RE after photo-Fenton experiments are compared. Three different steps in toxicity behaviour (marked 1, 2 and 3 in Fig. 3) can be described. Before photo-Fenton starts, SE toxicity (20–30% inhibition) may be due to the mixture of pollutants as well as to the formation of intermediates during Fenton in the dark, particularly from the fast oxidation of some of the ECs. On the contrary, simultaneous activation (negative inhibition) is also observable in the RE experiment. This completely different behaviour may be explained by the RE – nutrients and salt contents, which in someway favoured the growth of V. fisheri until the toxic effect of pollutants and their oxidation intermediates was neutralized. In both SE and RE, when photo-Fenton starts up, from an of H2O2 dose ¼ 0 mg L1 up to H2O2 dose ¼ 50 mg L1, the degradation of parent pollutants begins forming more toxic intermediates, which drastically increase the inhibition rate in SE (Fig. 3, Step 2). The previously favourable conditions of RE probably can no longer efficiently balance the increasingly toxic effect, so inhibition start to gradually increase in RE too. In the last step (3) inhibition with RE is quite constant, probably because toxic organic intermediates take longer to mineralise; on the other hand, toxicity in SE almost totally disappeared. If we compare these results with residual concentration of the four parent ECs selected for comparison (selected for their representativeness of the behaviour of the 15 different compounds) it may be observed that photo-Fenton treatment was more efficient in the RE experiment than in the SE. So in terms of toxicity, the higher removal rate in RE experiment may result in a correspondingly larger amount of more toxic intermedi