Functional Lipidomics

Functional

Lipidomics

Functional

Lipidomics edited by

Li Feng Glenn D. Prestwich

Boca Raton London New York

A CRC title, part of the Taylor & Francis imprint, a member of the Taylor & Francis Group, the academic division of T&F Informa plc.

Front cover image by: Joseph Shape and Daryll B. DeWald, Utah State University, Logan, Utah.

Published in 2006 by CRC Press Taylor & Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742 © 2006 by Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group No claim to original U.S. Government works Printed in the United States of America on acid-free paper 10 9 8 7 6 5 4 3 2 1 International Standard Book Number-10: 1-57444-467-0 (Hardcover) International Standard Book Number-13: 978-1-57444-467-4 (Hardcover) Library of Congress Card Number 2005043991 This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data Functional lipidomics / [edited by] Li Feng and Glenn D. Prestwich. p. cm. Includes bibliographical references and index. ISBN 1-57444-467-0 1. Lipids--Metabolism. 2. Lipids--Physiological effect. 3. Lipids--Synthesis. I. Feng, Li, Ph.D. II. Prestwich, Glenn D. QP751.F927 2005 612.3'97--dc22

2005043991

Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com Taylor & Francis Group is the Academic Division of T&F Informa plc.

and the CRC Press Web site at http://www.crcpress.com

Preface With the sequencing of the genomes of model organisms, humans, pathogens, and crops, the science of genomics paved the way for the systematic study of cellular functions of encoded proteins. Beyond proteomics, however, is the realm of nongenetically encoded information encompassing lipids, carbohydrates, glycopeptides, metabolic pathways, lipid–protein interactions, and more. The explosion of information in the postgenomic era challenges science to be integrative, as well as reductionist, in pursuing potential practical biomedical applications. Lipids, often considered as just fats and oils, are essential elements in cell signaling and subcellular structure. Scientists in the lipid research field have started to examine the discipline on a broader scale and to extract knowledge and patterns of various lipids in a more systematic way. Systematically analyzing the lipidome provides opportunities for therapeutic intervention and diagnosis. The expression patterns and levels of lipids and lipidbinding proteins could be important markers of a developmental stage, pathological condition, or disease treatment. Lipid metabolites may reveal biochemical patterns related to disease states such as cancer, immune function, and neurodegenerative disorders. Monitoring changes in the lipidome over time may lead to understanding of the effects of disease, medication, diet, or environmental factors on lipid metabolism and phenotype. Modeling the differences of the lipid metabolic pathways in prokaryotic versus mammalian cells and tissues may provide new targets for antibacterial drug design. Simultaneous monitoring of all lipid species at basal level or under stimulation may reveal cross-talk between major lipid-signaling pathways. Comprehensive analysis of lipids may uncover novel lipid species at lower quantities. By applying a systematic approach similar to that used for genomics and proteomics, an integrated knowledge base can be developed for bacteria, yeast, and the human lipidome. Such a knowledge base has the potential to identify and validate targets for improved, personalized medicine and health. In this edited volume, we intend to introduce the notion of “ functional lipidomics,” by which we mean the total profiling of lipids and the entities that interact functionally with lipids. We define the total lipid composition of a cell or organ as the “lipidome.” Lipidomics can include the functional genomics of lipid metabolism, i.e., understanding the biosynthetic pathways, the function and dysfunction of lipids, biological membranes, lipoproteins, and protein–lipid interactions. The biosynthesis, metabolism, and function of fatty acids, acylglyerols, glycerophospholipids, sphingolipids, terpenes including steroids, and lipids that form complexes with proteins and sugars will be uncovered in this book. The state of the art of lipid analysis approaches, diagnostics using lipid and lipid-binding protein patterns, and the therapeutic significance in targeting proteins involved with the lipid-signaling pathway will be elaborated from both academic and industrial perspectives.

We have presented topics at the forefront of current lipid research as viewed by an international set of authors from the United States, Europe, Canada, Singapore, and China. Our purpose was threefold: first, we hoped to capture the first snapshot of the current state of the art in functional lipidomics. Second, we wished to provide a comprehensive background for scientists new to the field and whose contributions will propel further understanding of lipidomics and biology. Finally, this monograph may serve as a resource for developing didactic courses in lipid cell signaling and lipid bioinformatics that could be employed for graduate study.

About the Authors Li Feng Li Feng is a scientist at Echelon Biosciences, Inc., an Aeterna-Zentaris Company, a leader in the field of lipid cell signaling research and product development. She earned her B.Sc. and M.Sc. in chemistry from Beijing University and a Ph.D. in medicinal chemistry from the University of Utah in 1999, followed by two years as a postdoctoral research associate at the Center for Cell Signaling, University of Utah. She has been involved in the technology development that furthered the growth of Echelon and facilitated research in the lipid signaling research field. Her research projects and interests include (1) reagents and assay development for lipid signaling research, (2) anticancer drug development targeting lipid-metabolizing enzymes, and (3) development of platform technology for profiling the interactions of lipids and lipid-associated proteins. Glenn D. Prestwich Glenn D. Prestwich, Presidential Professor of Medicinal Chemistry at the University of Utah since 1996, holds concurrent adjunct appointments in the departments of chemistry, biochemistry, and bioengineering. He is the director for two Utah “Centers of Excellence” for technology commercialization: the Center for Cell Signaling and the Center for Therapeutic Biomaterials. He chaired the University of Utah Department of Medicinal Chemistry from 1996 to 2002, and is currently program leader of the Molecular Pharmacology Program of the Huntsman Cancer Institute. Until its aquisition by Aeterna-Zentaris in 2005, he was the Chief Scientific Officer, Echelon Biosciences, Inc. (Salt Lake City, Utah), which he cofounded in 1997. In addition, he serves as Chief Scientific Officer, Sentrx Surgical, Inc. (Salt Lake City, Utah), a new therapeutic biomaterials start-up, which he cofounded in 2004. His research programs now include (1) new reagents for lipid signaling in cell biology, (2) anticancer drugs and diagnostics, and (3) biomaterials for wound repair, cartilage repair, tissue engineering, and prevention of postsurgical adhesions. Dr. Prestwich graduated with a B.Sc. (honors) in chemistry from the California Institute of Technology in 1970. He earned a Ph.D. in chemistry from Stanford University in 1974, followed by three years as an NIH postdoctoral fellow, first at Cornell University and then at the International Centre for Insect Physiology and Ecology in Nairobi, Kenya. From 1977 to 1996, he was at Stony Brook University in Stony Brook, New York, as professor of chemistry, professor of biochemistry and cell biology, and director of the New York State Center for Advanced Technology in Medical Biotechnology. He cofounded Clear Solutions Biotechnology, Inc. (Stony Brook), to commercialize biomaterials derived from hyaluronic acid. He is a recipient of the Alfred P. Sloan Research and Dreyfus Teacher–Scholar awards and was

honored with the 1998 Paul Dawson Biotechnology Award of the American Association of Colleges of Pharmacy. Dr. Prestwich has published over 490 technical papers and book chapters, including popular articles in National Geographic and Scientific American. Over 28 years, he has trained more than 66 graduate students and 52 postdoctoral scientists. He is an inventor on over 40 patents and patent applications, including the enabling technologies for his four start-up companies.

List of Contributors Michelle D. Armstrong The Alliance for Cellular Signaling Lipidomics Laboratory Department of Pharmacology and the Institute for Chemical Biology Vanderbilt University Medical Center Nashville, Tennessee Robert M. Barkley Department of Pharmacology University of Colorado Health Sciences Center Aurora, Colorado Rebecca C. Bowers–Gentry University of California, San Diego Department of Chemistry and Biochemistry and Department of Pharmacology La Jolla, California H. Alex Brown Department of Pharmacology Vanderbilt University Medical Center Nashville, Tennessee Pietro De Camilli Howard Hughes Medical Institute Department of Cell Biology Yale University School of Medicine Boyer Center for Molecular Medicine New Haven, Connecticut Leena Chakravarty Kwai Wa Cheng Department of Molecular Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas David Crotzer Department of Gynecologic Oncology The University of Texas M.D. Anderson Cancer Center Houston, Texas

Luciano De Petrocellis Istituto di Cibernetica “Eduardo Caianiello” Consiglio Nazionale delle Ricerche Pozzuoli (Napoli), Italy Raymond A. Deems Department of Chemistry and Biochemistry University of California, San Diego La Jolla, California Edward A. Dennis Department of Chemistry and Biochemistry and Department of Pharmacology University of California, San Diego La Jolla, California Vincenzo Di Marzo Istituto di Chimica Biomolecolare Consiglio Nazionale delle Ricerche Pozzuoli (Napoli), Italy Wynn Esch Mass Spectrometry Laboratory University of Kansas Lawrence, Kansas Edward A. Felix Department of Experimental Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas Li Feng Echelon Biosciences, Inc. Salt Lake City, Utah Colin Ferguson Echelon Biosciences, Inc. Salt Lake City, Utah Jeffrey S. Forrester The Alliance for Cellular Signaling Lipidomics Laboratory Department of Pharmacology and the Institute for Chemical Biology Vanderbilt University Medical Center Nashville, Tennessee

Christopher K. Glass Cellular and Molecular Medicine Department of Medicine University of California, San Diego La Jolla, California Richard W. Gross Division of Bioorganic Chemistry and Molecular Pharmacology Departments of Medicine, Molecular Biology and Pharmacology, and Chemistry Washington University School of Medicine St. Louis, Missouri Xianlin Han, Ph.D. Division of Bioorganic Chemistry and Molecular Pharmacology Department of Medicine Washington University School of Medicine St. Louis, Missouri Yusuf A. Hannun Department of Biochemistry and Molecular Biology Medical University of South Carolina Charleston, South Carolina Richard Harkewicz University of California, San Diego Department of Chemistry and Biochemistry and Department of Pharmacology La Jolla, California Yutaka Hasegawa Department of Molecular Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas Pavlina T. Ivanova The Alliance for Cellular Signaling Lipidomics Laboratory Department of Pharmacology and the Institute for Chemical Biology Vanderbilt University Medical Center Nashville, Tennessee Jessica Krank Department of Pharmacology University of Colorado Health Sciences Center Aurora, Colorado

John Lahad Department of Molecular Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas Rose Lapis Texas Children’s Hospital–Texas Children’s Pediatric Associates Baylor College of Medicine Houston, Texas Steven LeVine Department of Molecular and Integrative Physiology University Kansas Medical Center Kansas City, Kansas Alfred H. Merrill, Jr. Georgia Institute of Technology School of Biology Atlanta, Georgia Gordon B. Mills Department of Molecular Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas Stephen B. Milne The Alliance for Cellular Signaling Lipidomics Laboratory Department of Pharmacology and the Institute for Chemical Biology Vanderbilt University Medical Center Nashville, Tennessee Andrew Morris Department of Cell and Developmental Biology University of North Carolina Chapel Hill, North Carolina Robert C. Murphy University of Colorado Health Sciences Center Department of Pharmacology Aurora, Colorado Paul O. Neilsen Echelon Biosciences, Inc. Salt Lake City, Utah

Robert A. Newman Department of Experimental Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas Christian Pasquali Serono Pharmaceutical Research Institute Serono International S.A. Geneva, Switzerland Glenn D. Prestwich Department of Medicinal Chemistry The University of Utah Salt Lake City, Utah Christian R.H. Raetz Duke University Medical Center Department of Biochemistry Durham, North Carolina Christian Rommel Serono Pharmaceutical Research Institute Serono International S.A. Geneva, Switzerland David W. Russell The University of Texas Southwestern Medical Center at Dallas Department of Molecular Genetics Dallas, Texas Piotr W. Rzepecki Echelon Biosciences, Inc. Salt Lake City, Utah Jyoti Shah Division of Biology Kansas State University Manhattan, Kansas Walter Shaw Avanti Polar Lipids, Inc. Alabaster, Alabama

Shankar Subramaniam Department of Bioengineering University of California, San Diego La Jolla, California M. Cameron Sullards, Ph.D. Georgia Institute of Technology School of Chemistry and Biochemistry and School of Biology Atlanta, Georgia Makiko Umezu-Goto Department of Molecular Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas Mario van der Stelt NZ Organon The Netherlands Michael S. Van Nieuwenhze Department of Chemistry and Biochemistry University of California, San Diego La Jolla, California Xuemin Wang Department of Biology University of Missouri St. Louis, Missouri Ruth Welti Division of Biology Kansas State University Manhattan, Kansas

Markus R. Wenk Department of Biochemistry and Department of Biological Sciences National University of Singapore Singapore Stephen H. White Department of Physiology and Biophysics University of California at Irvine Irvine, California

Todd Williams Mass Spectrometry Laboratory University of Kansas Lawrence, Kansas Joseph L. Witztum Department of Medicine University of California, San Diego La Jolla, California Judith Wolf Department of Gynecologic Oncology The University of Texas M.D. Anderson Cancer Center Houston, Texas John Wooley University of California, San Diego La Jolla, California Yi Jin Xiao Department of Cancer Biology Cleveland Clinic Foundation Cleveland, Ohio Yan Xu Department of Cancer Biology Cleveland Clinic Foundation Cleveland, Ohio Shuangxing Yu Department of Molecular Therapeutics The University of Texas M.D. Anderson Cancer Center Houston, Texas

Table of Contents Chapter 1

The LIPID MAPS Approach to Lipidomics .......................................1

Edward A. Dennis, H. Alex Brown, Raymond A. Deems, Christopher K. Glass, Alfred H. Merrill, Jr., Robert C. Murphy, Christian R.H. Raetz, Walter Shaw, Shankar Subramaniam, David W. Russell, Michael S. VanNieuwenhze, Stephen H. White, Joseph L. Witztum, and John Wooley Chapter 2

LC/MS Methodology in Lipid Analysis and Structural Characterization of Novel Lipid Species...........................................17

Robert C. Murphy, Jessica Krank, and Robert M. Barkley Chapter 3

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids ..............................................................................57

Luciano De Petrocellis, Mario van der Stelt, and Vincenzo Di Marzo Chapter 4

Eicosanoid Lipidomics.......................................................................79

Rebecca C. Bowers-Gentry, Raymond A. Deems, Richard Harkewicz, and Edward A. Dennis Chapter 5

Functional Lipidomics: Lysophosphatidic Acid as a Target for Molecular Diagnosis and Therapy of Ovarian Cancer....................101

Janos L. Tanyi, David Crotzer, Judith Wolf, Shuangxing Yu, Yutaka Hasegawa, John Lahad, Kwai Wa Cheng, Makiko Umezu-Goto, Glenn D. Prestwich, Andrew Morris, Robert A. Newman, Edward A. Felix, Rose Lapis, and Gordon B. Mills Chapter 6

Analysis of Lysophospholipids in Human Body Fluids: Comparison of the Lysophospholipid Content in Malignant vs. Nonmalignant Diseases ............................................................................................125

Yi Jin Xiao and Yan Xu Chapter 7

Functional Lipidomics: Lessons and Examples from the Sphingolipids....................................................................................147

Yusuf A. Hannun

Chapter 8

Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) Analysis of Sphingolipids .........................................159

M. Cameron Sullards and Alfred H. Merrill, Jr. Chapter 9

Methods of Probing Phosphoinositides–Protein Interactions .........189

Li Feng, Colin Ferguson, Paul O. Neilsen, Leena Chakravarty, Piotr W. Rzepecki, and Glenn D. Prestwich Chapter 10 Fishing for Pharmaceutically Relevant phosphoinositide-Binding Proteins Using Chemical Proteomics.............................................................211 Christian Pasquali and Christian Rommel Chapter 11 Phosphoinositide Profiling in Complex Lipid Mixtures .................243 Markus R. Wenk and Pietro De Camilli Chapter 12 Multiplexed Lipid Arrays of Antiimmunoglobulin M–Induced Changes in the Glycerophospholipid Composition of WEHI-231 Cells .............................................................................263 Stephen B. Milne, Jeffrey S. Forrester, Pavlina T. Ivanova, Michelle D. Armstrong, and H. Alex Brown Chapter 13 Specific Lipid Alterations in Alzheimer’s Disease and Diabetes: Shotgun Global Cellular Lipidome Analyses by Electrospray Ionization Mass Spectrometry Using Intrasource Separation.........285 Xianlin Han and Richard W. Gross Chapter 14 High-Throughput Lipid Profiling to Identify and Characterize Genes Involved in Lipid Metabolism, Signaling, and Stress Response...........................................................................................307 Ruth Welti, Jyoti Shah, Steven LeVine, Wynn Esch, Todd Williams, and Xuemin Wang Index......................................................................................................................323

1

The LIPID MAPS Approach to Lipidomics Edward A. Dennis, H. Alex Brown, Raymond A. Deems, Christopher K. Glass, Alfred H. Merrill, Jr., Robert C. Murphy, Christian R.H. Raetz, Walter Shaw, Shankar Subramaniam, David W. Russell, Michael S. VanNieuwenhze, Stephen H. White, Joseph L. Witztum, and John Wooley

CONTENTS Contents .....................................................................................................................1 Abstract ......................................................................................................................1 Introduction................................................................................................................2 Background and Broad Context ................................................................................4 Significance for Biomedical Science and Disease....................................................5 Focus Areas................................................................................................................5 Introduction to Lipidomics Focus Areas...................................................................7 Fatty Acids and Eicosanoids Core...................................................................7 Neutral Lipids Core .........................................................................................8 Glycophospholipids Core.................................................................................9 Sphingolipids and Glycosphingolipids Core .................................................11 Sterols Core....................................................................................................12 Isoprenoids and Structural Lipidomics Core.................................................13 Summary and Cross Talk Among Lipid Classes ....................................................14 Acknowledgments....................................................................................................15

ABSTRACT A five-year, large-scale collaborative “Glue Grant” (funds providing the "glue" to bring investigators together, allowing them to work interactively) was funded by the National Institute of General Medical Sciences (NIGMS) in August 2003 to develop a LIPID Metabolites and Pathways Strategy, known as LIPID MAPS. The goals of the LIPID MAPS Consortium are: (1) to separate and detect all of the lipids in a specific cell and to discover and characterize novel lipids that may be present, (2) to quantitate each of the lipid metabolites present and determine the changes in their 1

2

Functional Lipidomics

levels and locations during cellular function, and (3) to define biosynthetic pathways for each lipid and develop lipid maps that define interaction networks.

INTRODUCTION Sequencing of the human genome has opened the way and provided the impetus for building a comprehensive picture of the mammalian cell. Significant efforts are underway in the fields of genomics and proteomics to identify all the genes and proteins in a given organism. The goal is a complete map of the genes, gene products, and their interaction networks in a functioning cell. The next step in establishing a comprehensive picture will be to integrate the cell’s metabolome into the rapidly developing genomic and proteomic maps. A cell’s metabolome, however, is an enormously complex and constantly changing entity that can only be approached in sections. We are focusing on the lipid section of the metabolome by developing an integrated system capable of characterizing the global changes in lipid metabolites (“lipidomics”). This approach is termed the LIPID Metabolites and Pathways Strategy or LIPID MAPS. Only recently has it been widely recognized by the scientific community that lipids are central to the regulation and control of cellular function and disease. This insight has enhanced our appreciation of the importance of lipids. Comprising an enormous fraction of cellular metabolites, lipids include such diverse classes as fatty acids, eicosanoids, phospholipids, neutral lipids, sphingolipids, glycosphingolipids, sterols, isoprenoids, and glycolipids. Because of the similarities in the chemical characteristics of this broad spectrum of compounds, the lipid metabolic pathways thus far identified are complex, and often a given lipid shuttles between several different pathways. Changes in the level of one member of a lipid class not only affect other members of that class but can also affect members of other classes. Although investigators are able to monitor concurrent changes in a few members of a lipid class, no comprehensive strategy has emerged for characterizing the global changes in lipid metabolites. LIPID MAPS is designed to fill this gap. The aim of LIPID MAPS is to develop the requisite technology and conduct an integrated research program that will establish lipidomics as a fully developed field. To accomplish this goal, we designed a strategy to identify and quantitate all of the lipids in a cell. The crucial first step in achieving this end will be to set a realistic scope for these studies. First, because of the complexity of metabolomics and lipidomics, we will focus on a single cell type. We have chosen to study macrophages because they carry out many functions known to be related to lipid metabolism, including the secretion of active lipid second messengers. Macrophages also possess many other characteristics that will aid in tackling the challenges presented by this project. In addition, all members of the consortium will focus on just one or two stimuli at a time. A second important factor is that we will employ a single detection method, mass spectrometry (MS), to qualitatively and quantitatively analyze lipids. Liquid chromatography/mass spectroscopy (LC/MS) is at present the only analytical technique that, by itself, can be used to separate, identify, and quantitate the vast number of closely related compounds that comprise the lipids.

The LIPID MAPS Approach to Lipidomics

3

An important objective of LIPID MAPS will be to locate and identify new minor lipid compounds. Although almost all of the major lipids of cells are known, many minor lipids remain to be described. Interestingly, minor lipids are often the most potent and profound biological entities and are associated with diseases. The LC/MS system is again ideally suited for identifying these rare lipids, even while quantitating the major ones. We will also employ genomics and proteomics to identify the enzymes and other proteins involved in lipid metabolism and to integrate them into the maps and networks of lipidomics. We have brought together a group of scientists who will operate ten cores and five bridges in close collaboration. These individuals will employ a common set of biochemical and cell culture protocols that will ensure that reproducible experiments are carried out in each of the consortium laboratories and that the data collected by one unit will be directly comparable to that collected by another. Six of the cores will develop techniques to quantitate a given class of lipids. To increase the reproducibility and to make certain that data from different lipid classes can be compared, each core will be able to rely on the results of the sterol care but also use a similar LC/MS system. Thus, for example, the glycerophospholipid core will be able to quantitate the level of sterols in a given experiment in their laboratory by employing the techniques developed in the sterol core. This shared functionality will allow the cross-checking of data. The data thus generated will be collected through a common laboratory information management system, stored in common databases, and analyzed by an informatics focus area. The lipidomics cores, as discussed in the section titled Introduction to Lipidomics Focus Areas, will not operate as traditional cores, i.e., in providing a narrowly defined function to outside researchers. Rather, the duties of each core will be to develop the techniques required to quantitate their specific class of lipids, to quantitate the levels and changes in levels of all of these lipids in the macrophages under the conditions set forward by the consortium, and to supply this mass of data to the bioinformatics and data coordination core for the new global approach to analysis. An equally important duty will be to follow any hypothesis-driven leads that they may discover in the course of these experiments. They will interact directly with the other cores, as appropriate, to test these hypotheses. LIPID MAPS, by measuring the changes in both major and minor lipid metabolites in response to stimuli and by developing a map of the routes these lipids take in cells, will provide the scientific community with a wealth of data concerning the coordinated responses of cellular lipids, at a level never before possible. By sharing these findings with all scientists, a much broader impact will be achieved. The goal of the first five-year period is to determine the complete lipidome of the mouse macrophage. Beyond the initial five-year effort, the long-term objective is to record the changes in the lipidome as the macrophage responds to various stimuli, and to continue building a database of results and query tools for the LIPID MAPS. At the same time, employing these techniques, studies on both human and mouse differentiated tissue macrophages will begin. All of this information will be freely accessible on the Internet and will establish a new global, discovery-driven approach in the field of lipids that will complement the traditional reductionist research approaches.

4

Functional Lipidomics

BACKGROUND AND BROAD CONTEXT Research in the biochemical and medical communities has entered a new era; one that is characterized by the generation of tremendous amounts of data, the volume of which derives from the use of high-throughput technologies by various genomic and proteomic studies. It also is due in part to the emerging use of universal systemswide approaches to cell biology in lieu of the traditional hypothesis-driven approaches. These new approaches focus on gathering experimental data from a variety of sources, methods, and scales, and on measuring as many components of the system as possible, without regard to possible relevance. Dealing with this enormous volume of data has driven the emergence of the bioinformatics field, which has now developed the tools to collect, store, retrieve, organize, and analyze this data. To date, the bulk of this effort has been centered around genomics and proteomics. The last member of the triad required to describe a cell in detail remains to be tackled, i.e., metabolomics. The intellectual challenge for biology, that of converting this information from raw data to organized, comprehensive knowledge, has been variously termed “beyond the genome,” “the new biology,” “bringing genomes to life,” or “21st century biology.” How challenging is this task? Over the last decade, various genome projects had to determine a linear sequence with only four nucleotides, and the success of these projects has been a major scientific coup. Proteomics presents a more challenging task, because the complete set of proteins, built by the combination of 20 amino acids and their posttranslational modifications, must be characterized in different states of development, aging, health, and disease, as well as in response to drugs and other stimuli. Besides differential splicing, the proteome of a cell changes with time and in response to environmental conditions. Even in comparison to proteomics, metabolomics presents a staggering challenge. Genes and proteins are composed of a single type of building block, whereas cellular metabolites represent a diverse range of structural moieties for which a single separation and identification system is unrealistic. In addition, the metabolic profile of a cell changes from second to second. The levels of metabolites, as well as their structure, are in constant flux, which is fundamental to metabolism and life. The levels of the metabolites vary by orders of magnitude. Developing the tools for quantitative analyses is a daunting task, and will require a wide variety of separation and analyses systems that reflect the diversity of metabolites. Collecting, organizing, and analyzing this data is equally intimidating. Correlating this data with that of genomics and proteomics is the ultimate challenge. Clearly, to have any chance at generating a cellular metabolome, it must be divided into more manageable units. The LIPID MAPS consortium feels that lipids are an advantageous class of molecules with which to start this move toward establishing a complete metabolome of a cell. This class of molecules can be easily separated from other functional classes by simple organic extraction of cells. Because of their solubility in organic solvents, good separation techniques exist. A further advantage is that most classes of lipids are metabolically related and result from anabolic and catabolic addition or subtraction of acetyl CoA or isoprenoid units, which make up practically all lipids. Therefore, we will approach the broader issues

The LIPID MAPS Approach to Lipidomics

5

of metabolomics by focusing on lipids. We will establish a core for each of the major functional subclasses of lipid metabolites with the goal of identifying and quantitating changes in metabolites by developing new techniques for detection and analysis, and by a systematic analysis of the massive amount of data generated. Yet, it should be recognized that the resulting lipidome will not be a static determination because the lipid metabolites will constantly be changing with time and location, and those changes need to be described as well. We will focus initially on the monocyte, or macrophage, a cell that has been widely studied in connection with lipid metabolism. Once the lipids of the monocyte are defined, we could readily progress to the analyses of Kupfer cells and peritoneal, alveolar, microglial, and other tissue macrophages, or to the study of foam cells and other disease implicated transformations of this cell type. Additionally, primary mouse and human monocytes are readily obtainable, thus providing the opportunity to study the lipids of cells harboring known mutations.

SIGNIFICANCE FOR BIOMEDICAL SCIENCE AND DISEASE Lipids play a central role in all cells; thus, it is clear that characterizing their metabolic changes in disease states will help in understanding the relationships among various lipid classes. One example, from among the many lipid agents active as second messengers, is the eicosanoids. The eicosanoids are a family of oxygenated derivatives of 20-carbon polyunsaturated fatty acids that mediate a wide variety of physiological and pathophysiological processes. Eicosanoid production by macrophages and macrophage-like cells plays a central role in many inflammatory diseases including rheumatoid arthritis, sepsis, asthma, and inflammatory bowel disease. Other diseases in which eicosanoids play a role include atherosclerosis, Alzheimer’s disease, cancer, and stroke. Transient, localized increases in the tissue concentration of eicosanoids influence differentiation, migration, and activation of cells in immunity and other integrated physiological responses. A large number of disorders are influenced by the regulation of eicosanoid production. Thus, tracking the synthesis and breakdown of fatty acids and eicosanoids will play a central role in any comprehensive characterization of lipid metabolism and should identify numerous opportunities for drug intervention in treating these diseases. To address impacts further upstream, it is essential to monitor changes in phospholipids and sphingolipids as well. Of course, there may be new, as yet undiscovered lipids that play important cellular roles, and these too will be identified and characterized with the advance of LIPID MAPS.

FOCUS AREAS Each major activity in the LIPID MAPS consortium is termed a focus area; individual contributing research efforts are comprised of various cores and bridges as indicated in the outline that follows. In the subsequent Sections, we describe lipidomics in

6

Functional Lipidomics

more detail to explain the collaborative approach being taken to tackle this challenging problem: 1. Administrative focus area 2. Informatics focus area a. Bioinformatics and data coordination b. LIPID MAPS networks 3. Cell biology focus area a. Macrophage biology and functional genomics b. Transcriptional regulation in macrophages 4. Lipid detection and quantitation focus area a. LC/MS analysis b. Novel detection technique development 5. Lipid synthesis and characterization focus area a. Lipid standards and production chemistry b. Novel lipid synthetic design c. Novel lipid biophysical characterization 6. Lipidomics focus areas a. Neutral lipids b. Fatty acids and eicosanoids c. Glycerophospholipids d. Sphingolipids and glycosphingolipids e. Sterols f. Isoprenoids and structural lipidomics g. Oxidized lipids in macrophages h. Lipid subcellular localization The core laboratories divide the major lipid classes expected in macrophages into six groups (Figure 1.1). Although these groups constitute the classically recognized major groupings of lipids, they also allow us to cover the full range of lipids. Each core will not only analyze their particular class of lipids but will also actively participate in translating and utilizing the results of their own work and that of the other cores for application to macrophage biology. It should be noted that because of the limited number of pure glycolipids known to be made by the macrophage (and mammalian cells in general), we have not included a separate glycolipid core; however, the existing cores will also monitor this class of lipids. The effectiveness of this effort in lipidomics will be made possible by the uniformity of the cellular approaches and the integration with functional genomics and contextual proteomics spearheaded by the efforts of the cell biology focus area. The contributions of the lipid detection and quantitation focus area will be crucial for the lipidomics cores to achieve their goals and to collect useful and comparative data. The lipid synthesis and characterization focus area will enable the lipidomics cores to carry out their mission by providing standards and by confirming and extending the utility and understanding of the properties of new lipids discovered in the lipidomics cores. The informatics focus area has the challenge of putting together all of the experimental input and results of the lipidomics cores, of devel-

The LIPID MAPS Approach to Lipidomics

Lipid class

Species

Fatty Acids/Eicosanoids

Free fatty acids, fatty acid amides, prostanoids; hydroxyl- and hydroperoxy-eicosaenoic acids, leukotrienes, and epoxyeicosatrienoic acids.

Neutral lipids

MAG, DAG, TAG, CE.

Glycerophospholipids

PC, PE, PG, PS, PI (and polyphospho derivatives), PA, cardiolipin, lysophospholipids, plasmalogens and other ether-linked phospholipids, prostanoid containing phospholipids.

Sphingolipids/ Glycosphingolipids

Sphingomyelin, glycosphingolipids, ceramides, sphingosine phosphate.

Sterols

Isoprenoids, cholesterol, oxidized steroids, sterols, and bile acids.

Isoprenoids

Polyisoprene-linked phosphate sugars, certain fat soluble vitamins and quinones.

7

FIGURE 1.1 Lipidomics Cores.

oping new networks and LIPID MAPS, and most importantly, of disseminating them. The administration focus area will coordinate the focus areas and their efforts, provide the infrastructure for success, manage the budgetary process and intellectual property issues, organize the advisory, steering, and operating committees, and ensure full dissemination of new data.

INTRODUCTION TO LIPIDOMICS FOCUS AREAS The lipidomics focus area is organized around six cores and two bridges. The thrust of each core will be discussed in turn. The oxidized lipids in the macrophages bridge will characterize various structurally defined oxidized lipids. The lipid subcellular localization bridge will develop methodology for the lipidomics cores that will allow them to determine the distribution of the various lipid classes in the macrophage and then to follow the changes in these pools as the cells are stimulated.

FATTY ACIDS

AND

EICOSANOIDS CORE

This core is directed by Dr. Edward A. Dennis (University of California, San Diego). The complexity of LIPID MAPS is clearly demonstrated by examining the metabolism of fatty acids. Once thought to be simply the building blocks of cellular membranes, fatty acids and their metabolites are now known to be potent second messengers. As outlined in the preceding text, eicosanoids are particularly notable for their roles in disease processes. Eicosanoids are not stored by cells, but are synthesized in response to cell-specific stimuli. These stimuli activate phospholipase A2 enzymes (PLA2s) that in turn increase the concentration of free 20-carbon polyunsaturated fatty acids by hydrolysis of glycophospholipids (GPLs). The majority of the eicosanoids produced by macrophages are derived from cleaved arachidonic acid (AA). Once freed, AA is converted into hundreds of active species, which

8

Functional Lipidomics

demonstrates one of the many aspects of the complexity of lipidomics. A second level of complexity stems from the fact that although a majority of eicosanoids are derived from AA, other 20-carbon polyunsaturated fatty acids also yield related compounds when acted upon by eicosanoid biosynthetic enzymes. Although all eicosanoid biosynthesis begins with the release of fatty acid from phospholipids by a PLA2, there are 14 different classes of PLA2s that release AA. Most cells express several PLA2s. The Dennis Laboratory has found numerous activation pathways in the P388D1 macrophage-like cell line that lead to significant PGE2 release. Most likely, other pathways exist that release different eicosanoids. Four pathways are relevant to the normal functions of macrophages, including responses to inflammation and acute injury. The first is mobilization and PGE2 production involving sequential exposure of the cells to two different stimuli, namely lipopolysaccharide (LPS) and platelet-activating factor (PAF). The second pathway involves the exposure of macrophages to LPS for long periods, up to 18 h, which also leads to AA release and PGE2 production. The third involves responses to zymosan as in yeast infections. In the fourth pathway, diacylglycerol pyrophosphate, which is not found in mammals, activates P388D cells; this unique lipid may elicit a defensive response from mammalian macrophages in reaction to the invasion by bacteria or yeast. The metabolism of fatty acids and the eicosanoid cascade represent a complex and intertwined metabolic system. It is important to know which phospholipid pools contribute the AA and which lysophospholipids result. This distribution will be determined by the glycerophospholipid core (see Section titled Glycerophospholipid Core). Current experiments tend to focus on a single metabolite in the pathway, usually either AA or PGE2. To understand how this system really works, how fatty acids flow through the pathways, and how various stimuli affect this metabolism will require that this core quantitate and monitor all fatty-acid-derived metabolites throughout the course of a given stimulus response. Severe limitations (in sensitivity and cost) hamper the use of most commonly employed techniques for detecting fatty acid metabolites via radiolabeling and enzyme immunoassays. Thus, the only currently available technology for a detailed definition of the cascade is the combined use of LC/MS, by which it is possible to monitor a majority of the fatty acid metabolites in an experiment without the need for radiolabels or monoclonal antibodies. Subcellular fractionation of the organelles and compositional analysis allow the mapping of changes and the determination of individual pools of phospholipids that are responsible for a given reaction, as well as determining the metabolites that are produced from each pool.

NEUTRAL LIPIDS CORE This core is directed by Dr. Robert Murphy (University of Colorado Medical Center). Neutral lipids include monoacylglycerols (MAGs), diacylglycerols (DAGs), and triacylglycerols (TAGs) as well as cholesteryl esters and wax esters. Neutral lipids are an abundant species in the mammalian cell and play important roles in normal cellular homeostasis as a storage medium for metabolic energy and as precursors from which a vast array of other lipids are derived. Some neutral lipids, such as

The LIPID MAPS Approach to Lipidomics

9

DAG, play an important role in signal transduction. The regulation of TAG content within a cell is complex and has been the focus of intense study for over 50 years Both the degradation and synthesis of neutral glyceryl and cholesteryl esters are intimately involved in common pathologies, including obesity, arteriosclerosis, diabetes, and alcohol-related liver diseases. Neutral lipids occur at concentrations ranging from copious to rare, but they are synthesized from a relatively small number of building blocks, the so-called molecular species. For example, individual TAGs are composed of one molecule of glycerol and three molecules of fatty acids whose chain lengths and degrees of saturation vary. How the individual molecular species that compose a TAG change in response to metabolic or activating stimuli remains unknown. Only recently have specific molecular species within an individual class of neutral lipids been studied in detail, the delay being largely due to the perceived lack of changes. The analysis of TAG, in the past and present, has been dominated by approaches that involve degradation of the lipid esters to their constituent fatty acids followed by analysis via gas-phase techniques such as gas chromatography. The emergence of electrospray ionization (ESI) and matrix-assisted laser desorption ionization approaches now make it possible to analyze the individual molecular species of these neutral lipids with respect to the precise fatty acyl groups esterified and their amounts. Because many neutral lipids are present at relatively high concentrations within a cellular membrane or lipid body, the major challenge in their analysis is in the development of qualitative and quantitative tools to measure relative and absolute changes in a large number of very closely related molecular species. In some instances, molecular species are isobaric yet not isomeric, which requires a new level of sophistication in glyceryl ester analysis. The use of tandem mass spectrometer and collisional activation of ions offers considerable promise in dealing with these obstacles. The overall goal of the neutral lipids core is the development of robust protocols to follow alterations in both abundant and rare neutral lipids as a function of cellular status, and to do so without altering their molecular structure.

GLYCOPHOSPHOLIPIDS CORE This core is directed by Dr. H. Alex Brown (Vanderbilt University). The plasma membrane of eukaryotic cells is composed of a glycerolipid bilayer of heterogeneous composition and immense chemical complexity. Dynamic changes in membrane lipid composition profoundly affect cell function by providing a rich source of potential cellular messengers and by modulating protein–protein interactions at the cell surface. The major GPLs found in membranes of mammalian cells include phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylinositol (PI), phosphatidylserine (PS), and phosphatidylglycerol (PG). These GPLs also exist as plasmalogens and ether lipids; as multiply phosphorylated species (e.g., polyPIPs); and with a diverse array of alkyl chains. Phospholipid metabolism is regulated by cell-surface receptors (e.g., G-protein-coupled receptors, growth factors, and integrins) in many ways and at many levels, including the generation of first and second messengers (IP3, PIP2, lysoPA, DAG, sphingosine, etc.) and the modifications associated with secretion, migration, and plasma membrane shape change. Further-

10

Functional Lipidomics

more, as yet poorly described changes in bilayer structure regulate the activities of enzymes, channels, and transport proteins. Defining the mechanisms behind these changes and how the phospholipid content changes in response to these inputs are the goals of the glycerphospholipid core. The core will determine basal and signaling-induced changes in macrophage phospholipid composition. These changes will have implications for understanding both the metabolic and signaling pathways. For example, phosphatidic acid is an important second messenger in many cell types and is also an intermediate at the branch point of glycerolipid biosynthesis. Thus, changes in phosphatidic acid levels may influence cellular responses to stimuli as well as the composition of the phospholipid bilayer. Although the intersecting pathways of GPL head groups, DAG, and TAG biosynthesis were elucidated in pioneering work by Kennedy and others during the last century, an understanding of how specific changes of a given species influence the global membrane lipid composition of the cell membrane is lacking. As important substrates for (and products of) lipid-signaling enzymes (e.g., PLA2, PLC, PLD, PI-3 kinase, and lipid phosphatases), GPL lipids directly modulate the pathways of cellular proliferation, cytoskeletal arrangement, migration, and membrane transport (e.g., endocytosis and phagocytosis) as well as many other essential processes. In some cases the in vivo substrates from which important signaling lipids that control these processes such as phosphatidic acid and lysoPA are generated have not been unambiguously identified. Using an ESI–MS approach, the Brown laboratory has identified more than 450 phospholipids in mammalian cells and described how these change as a result of cell-surface receptor stimulation; however, the analysis to date has been qualitative in nature. A central goal of the LIPID MAPS consortium is to develop a high-precision quantitative analysis procedure for membrane lipids, including changes in GPL composition using stable isotope dilution approaches. Such advances will facilitate the construction of lipid arrays, that is, the computational analysis of changes in specific acyl and head group lipid species in response to biological stimuli. The ways in which the relative concentrations of these lipids change in response to cellular signaling potentially provides an extraordinarily rich source of data for classifying cellular metabolic status in addition to evaluating the effects and interactions of incoming signaling information. Changes in lipid composition will also indicate the identities (and eventually help to determine subcellular localizations) of the metabolic enzymes involved. The glycerophospholipid core will collaborate with each of the lipidomics cores within the consortium. One of the reasons for choosing fundamentally similar instrumentation for each of the cores is to standardize the analytical procedures and develop strategic approaches that will make comprehensive analysis of cellular lipids a reality. This homogeneity will also facilitate the interactions between the cores. Changes in the cellular composition of phospholipids alter the biosynthetic pathways involving neutral lipids (i.e., DAG) and free fatty acids. Phospholipids serve as precursors for some glycolipids and, reciprocally, phospholipid degradation is regulated by sphingosine-derived compounds through modulation of lipid hydrolase activities. Similarly, the substrates from which many eicosanoids and free fatty acids are generated are frequently membrane phospholipids. Establishment of quantitative analysis and unambiguous identification of changes in cellular lipids in response to

The LIPID MAPS Approach to Lipidomics

11

important biological stimuli (e.g., LPS stimulation) are essential steps in establishing substrate–product relationships in lipid-signaling and metabolic pathways. A recent focus of the Brown laboratory is the resolution and measurement of changes in polyphosphate-PI-containing lipids (i.e., PI 3,4P2, PI 4,5P2, and PI 3,4,5P3) using LC/MS. The ability to make mass measurements of poly-PIPs and to identify the acyl composition of these lipid species represents a significant advance in the field of signaling and will facilitate the goal of defining the roles of bioactive lipids in the macrophage. It is highly probable that this improved quantitative analysis and sensitivity in detection will lead to the discovery of novel lipid species and the illumination of previously unappreciated roles of lipids in human diseases.

SPHINGOLIPIDS

AND

GLYCOSPHINGOLIPIDS CORE

This core is directed by Dr. Alfred H. Merrill, Jr. (Georgia Institute of Technology). Sphingolipids are composed of a ceramide backbone and head group, each of which has dozens to hundreds of structural variants with differing degrees of complexity (for example, macrophages contain ceramides with mainly a sphingosine backbone and several amide-linked fatty acids, which are present as free ceramides, sphingomyelins, glucosylceramides, lactosylceramides, ganglioside GM3’s, and more complex glycoconjugates). The functions of sphingolipids are also complex, encompassing: interactions with proteins at the cell surface to mediate cell–cell and cell–substratum recognition and to modify receptor behavior; influence on membrane dynamics and the properties of proteins associated with structures often referred to as rafts and caveolae (which include receptors and transporters); and mediation of cell signaling through production of bioactive lipid backbones (ceramide, sphingosine, and sphingosine-1-phosphate, inter alia) via both sphingolipid turnover and de novo biosynthesis. To add to the functional complexity, two closely related metabolites (such as sphingosine and sphingosine-1-phosphate) can have opposite regulatory effects (in this case, the former often induces apoptosis whereas the latter can stimulate proliferation); furthermore, changes in sphingolipid metabolism can affect other lipid-signaling pathways (e.g., ceramides and DAGs are interrelated through the pathway Cer + PC -> SM + DAG, and vice versa). Therefore, to understand the roles of sphingolipids in cell regulation, one must analyze a large number of species with the type of sensitivity and structural resolution that is provided almost exclusively by MS. Many of the sphingolipids of macrophages and other mammalian cell types can be analyzed by recently developed methods using liquid chromatography and electrospray ionization tandem mass spectrometry (LC ESI-MS/MS) with the appropriate internal standards. Using a triple-quadrupole mass spectrometer (the ABI 3000) and the multiple-reaction-monitoring protocol for quantitation, the Merrill laboratory has shown that the individual molecular species of sphingomyelins, ceramides, and neutral glycosphingolipids with up to two (or three) carbohydrates can be analyzed in small samples (typically 106 cells); furthermore, the method allows analysis of trace species such as sphingoid bases, lysosphingolipids (including sphingoid-base 1-phosphates and ceramide-1-phosphates), and N-methylsphingoid bases. As part of LIPID MAPS, Merrill and Sullards have shown that analysis of these compounds

12

Functional Lipidomics

as well as more complex glycosphingolipids (e.g., larger cerebrosides and gangliosides) can be achieved using two additional instruments: a higher-resolution Quadrupole time-of-flight mass spectrometer (the Q-STAR) and an ion-trapping instrument (the Q TRAP) to analyze connectivity during fragmentation of higher-order glycoconjugates. Using these instruments in combination with other biochemical and genetic tools, it will be possible to dissect the complex interrelationships among sphingolipids and their connection to other lipid metabolic and signaling pathways.

STEROLS CORE This core is directed by Dr. David W. Russell (University of Texas Southwestern Medical Center). Sterols, GPLs, and sphingolipids represent the three major classes of membrane lipids in mammalian cells. These molecules, each having very different structure, work together to surround cells and their organelles with a semipermeable barrier that functions as a scaffold for signaling, transport, and recognition proteins. Cholesterol and other sterols impart rigidity, and also control the microstructure of subdomains within the membrane. The last several decades have seen great progress towards understanding the pathways of sterol synthesis and catabolism that cells use to regulate the levels of this essential but toxic class of membrane lipids. How sterol homeostasis is maintained is of medical as well as scientific importance as disregulation of this process can lead to sterol accumulation and consequent heart attacks and strokes. All mammalian cells synthesize sterols via a pathway that involves at least 18 enzymes. This pathway is also responsible for the synthesis of biologically active intermediates, in particular the isoprenoids, which play important roles in the biosyntheses of heme, quinones, dolichols, and isoprenylated proteins such as small G proteins. Output from the pathway is regulated by a feedback mechanism that controls the activity of sterol regulatory element binding proteins (SREBPs), which are transcription factors required for the expression of a majority of the genes in the cholesterol biosynthetic pathway and many in the fatty acid biosynthetic pathways. As in synthesis, cholesterol catabolism involves at least 17 enzymes located in each of the major subcellular compartments of hepatocytes as elucidated in part by the Russell laboratory. Several of these enzymes are expressed in nonhepatic tissues and cells, including macrophages, demonstrating that cholesterol is actively metabolized in many cell types. Unlike synthesis, cholesterol catabolism is regulated by members of the nuclear receptor family of transcription factors via both feedback and feed-forward mechanisms. Several of these receptors are modulated by sterols that play multiple roles in the cell. For example, oxysterols are ligands for the liver X receptor (LXR), substrates for cholesterol catabolism, regulators of SREBPs, and vehicles of sterol transport and secretion. Many of the pleiotrophic roles of sterols in mammalian cells have only been recently described, suggesting that additional sterols with unique activities in intracellular signaling, metabolism, and the control of gene expression remain to be discovered. The LIPID MAPS approach will make innovative use of MS and molecular biology to identify and quantitate sterols with new biological functions in a well-defined cell type, the macrophage. The questions to be answered include: Which

The LIPID MAPS Approach to Lipidomics

13

sterols are present in the macrophage and in what amounts are they found? How does the macrophage sterol complement change in response to stimulation? What biosynthetic enzymes produce macrophage sterols? How do sterols regulate the output and metabolism of other lipid classes? Success in these endeavors requires a multipronged attack involving collaborators with diverse interests and research talents such as those assembled in the LIPID MAPS consortium.

ISOPRENOIDS

AND

STRUCTURAL LIPIDOMICS CORE

This core is directed by Dr. Christian Raetz (Duke University). There is considerable biochemical evidence for the existence of novel minor lipids. A key characteristic of mammalian lipids, including phospholipids, glycolipids, and sterols, is their extreme molecular heterogeneity. It is estimated that there are more than 1000 distinct chemical entities that comprise the lipidome of a typical animal cell. Heterogeneity occurs in both the polar and hydrophobic portions of these diverse molecules. When the most sensitive radiochemical-labeling techniques are combined with multidimensional separation procedures, it is clear that there are dozens of unidentified minor components, some of which may play key roles in signaling or regulatory networks that remain to be discovered. The recent explosion of information regarding the diversity and function of the multiple types of PI–phosphate derivatives nicely illustrates this point; it is very likely that many additional lipids belonging to the known, or even novel, structural classes are yet to be discovered. The isoprenoids and structural lipidomics core will carry out an in-depth analysis of novel lipids in the macrophage using technologies developed by the Raetz laboratory to explore minor lipids that are endotoxin precursors in Gram-negative bacteria. These efforts will be carried out in collaboration with the various cores of the lipidomics or lipid synthesis focus areas, as appropriate. This core will also devise methods to quantify important lipids not covered by the other lipidomics cores, such as the polysisoprene-linked phosphate sugars, certain fat-soluble vitamins, and quinones. Dolichol diphosphate oligosaccharides involved in protein N-glycosylation in animal cells have not previously been subjected to MS. Likewise, genomic evidence strongly supports the existence of novel lipids in prokaryotic and eukaryotic cells. Careful inspection of completed genomes indicates the presence of genes encoding for proteins that might be involved in lipid metabolism. For instance, in Escherichia coli, there are two distinct, but functionally uncharacterized, orthologs of the glycerol-3-phosphate acyltransferases, two additional members of the phosphatidylglycerophosphate synthase family, and two cardiolipin synthase homologues of unknown function. In animal cells, there is similar genomic evidence for a wealth of as yet uncharacterized proteins related to the known enzymes of lipid metabolism that likely account for some of the minor unknown lipids discussed earlier. There are important biological insights to be gained from the elucidation of novel lipid structures. The clues that will be provided by the study of these molecules in animal cells will immediately facilitate the search for novel enzymatic pathways required for their biosynthesis. The structural lipidomics core will be in a strong position to conduct initial enzymatic studies of this kind. The Raetz laboratory could

14

Functional Lipidomics

follow the same basic approach used in the case of Gram-negative endotoxins, which resulted in the identification of more than 25 new enzymes and their structural genes, over the past 15 years. The experimental approach is conceptually simple, and it may be summarized as follows: 1. The identification and elucidation of new lipid structures must come first. 2. The new lipid structures provide clues to the existence of plausible new enzymes that might be responsible for their biosynthesis. These plausible enzymes can often be deduced using the general rules of mechanistic enzymology derived from studies of unrelated metabolic pathways. The most difficult step is usually the synthesis of the several possible alternative substrate precursors that are needed to assay the proposed new enzyme. The availability of the lipid synthesis focus area will facilitate this step. 3. Once a new enzyme has been identified by in vitro assay, it can be expression cloned or purified to permit sequencing and matching with the genome. A match to a reading frame of unknown function is the ideal outcome, as it serves to define for the first time the function of that gene. Consequently, the structural lipidomics core will contribute in a systematic manner to the functional annotation of the human genome, using the same logic and methods validated with E. coli endotoxins. Both NMR spectroscopy and MS will be used to evaluate the structures of novel compounds. In some cases, chemical synthesis by the cognizant focus area will be undertaken for validation. The fractionation and structural determination methods developed previously for bacterial phospholipids and endotoxins are general in their applicability to animal lipid classes, and will be used here for macrophage studies.

SUMMARY AND CROSS TALK AMONG LIPID CLASSES The six lipidomics cores will explore the changes and interactions within their assigned class of lipids, but we anticipate the finding of many examples of cross talk between and among classes of lipids that will be analyzed as well. There already are many examples known of cross talk in lipid metabolic and signaling pathways. The interactions among lipids can be categorized into three major types: physical interactions in membranes and other structures (such as lipoproteins), direct interactions between lipids in coupled metabolic pathways (such as through shared enzymes or precursors), and cross talk between lipid-mediated signaling pathways. In addition, there are a number of interactions that have been linked to macrophage behavior that are important with regard to atherosclerosis, namely: the hydrolysis of sphingomyelia in low-density lipoproteins (LDL) (as revealed by the treatment of LDL particles with exogenous sphingomyelinase, which is thought to mimic the in vivo modification by a secreted sphingomyelinase); this induces the uptake of LDL by macrophages and increases foam cell formation and the aforementioned induction of PLA2 by ceramide in macrophages. These examples underscore the importance of knowing how changes in one

The LIPID MAPS Approach to Lipidomics

15

lipid metabolism or signaling pathway affects, and is affected by, the behavior of the other lipids. The comprehensive LIPID MAPS approach will allow the dissection of these interrelationships and clarify the consequences for the mammalian cell.

ACKNOWLEDGMENTS We wish to thank the NIH National Institute of General Medical Sciences largescale collaborative grants program (Glue Grant), 5 U54 GM069338, for the support of this project.

2

LC/MS Methodology in Lipid Analysis and Structural Characterization of Novel Lipid Species Robert C. Murphy, Jessica Krank, and Robert M. Barkley

CONTENTS Part I: Introduction ..................................................................................................18 Magnetic Sector and Electric Sector .............................................................19 Quadrupole Mass Filters and Ion Traps ........................................................19 Interface to High-Pressure Liquid Chromatography (HPLC).......................20 ESI Mechanism ..............................................................................................21 Molecular Ion Species ...................................................................................22 Collisional Activation of Lipid Ions ..............................................................25 Data Systems..................................................................................................28 Part II: Electrospray Mass Spectra of Sample Lipids ............................................28 Fatty Acids .....................................................................................................29 Glycerolipids ..................................................................................................31 Glycerophospholipids.....................................................................................34 Sphingolipids..................................................................................................41 Sterols ............................................................................................................45 Isoprenoids .....................................................................................................47 Saccharolipids ................................................................................................51 Polyketides .....................................................................................................51 Conclusion ...............................................................................................................53 Acknowledgments....................................................................................................53 References................................................................................................................53

17

18

Functional Lipidomics

PART I: INTRODUCTION The analysis of lipid substances by mass spectrometry is almost as old as the technique itself. The inventor of the mass spectrometer, J.J. Thompson, used his new instrument to investigate the positive rays of electricity generated from simple hydrocarbon gases derived from decaying lipids (1). The progress of mass spectrometry was rapid even a century ago, and one of the first biochemical applications involved the use of a magnetic-sector mass spectrometer to help unravel the complexity of steroid biosynthesis, using the newly discovered stable isotope, deuterium (2). However, many of the mass spectrometers used today for the analysis of lipids have evolved from the pioneering thoughts of Wolfgang Paul (3), who devised the quadrupole mass filter and ion trap. Another mass spectrometric technology rapidly being applied to lipid analysis is time-of-flight mass spectrometry. Although each mass separation device has its own strengths and weaknesses, all these instruments separate ions as to their mass-to-charge ratio (m/z), the fundamental unit of mass spectrometry. Often the choice of what type of physical instrument to use is predicated on the mass accuracy desired for the analysis, the ease of collisional activation of ions, the ease of coupling the flowing stream of liquid or gas containing the sample to the mass spectrometer, as well as the ease of operation and, not to the least extent, financial investment in the equipment. One general feature of all mass spectrometers is that they carry out separation of ions (each unique m/z) in a high vacuum, typically at 10−5 to 10−8 torr. At this reduced pressure, the mean free path of a particle (or ion) moving through this space is sufficiently long, so that it does not collide with another particle before it collides with the defining walls of the instrument. Under such conditions, an electrically charged positive or negative ion will interact with external physical forces such as electrical and magnetic fields in a manner predicted by well-known Newtonian physics to a first approximation. Thus, the maintenance of the high vacuum is essential, but it is also one element that makes this instrumental technique difficult in practice. Acceleration of ions is a fundamental process for all mass spectrometers as it is essential to move an ion from one region to another. This is typically achieved by generating an ion (by an ion source) in a region close to a conductive electrical surface such as a metal plate that has a fixed electrical voltage on it generated by a power supply. The ion then has a certain potential energy (PE) because of its charge and the voltage on the metal plate. PE = zV where z is the electronic charge V is the voltage supplied If the ion has the same fundamental charge sign as that of the metal plate, e.g., positive ion and a positive voltage on the plate, then the ion (mass, m) in a vacuum will respond by moving away or being repelled. In doing so, the ion picks up speed quickly (velocity, v) until it has converted all of its PE into kinetic energy (KE):

LC/MS Methodology in Lipid Analysis and Structural Characterization

19

KE = 1 2 mv 2 when PE = KE (at full acceleration ) zV = 1 2 mv 2 v=

2Vz m

The velocity of the ion is therefore related to the square root of the applied voltage and inversely proportional to the square root of the mass-to-charge ratio.

MAGNETIC SECTOR

AND

ELECTRIC SECTOR

The magnetic-sector instrument uses a strong magnetic field perpendicular to the direction of motion of ions to cause ions to travel in a circular path as described by the following equation:

m/z=

B2 r 2 2V

where a magnetic field B is the field experienced by the ion r is radius of curvature V is the accelerating potential of the ion Detailed monographs describing the derivation of this equation (4) and refinements to modern instruments are readily available (5,6). In some cases, ions are energy-resolved by an electric sector prior to entering the magnetic field. The energy resolution is achieved by an arrangement of two parallel plates curved into a specific geometry to effect energy focusing (7). Details of such sectors can be found in various monographs (6,7). The combination of energy focusing and magnetic focusing has been a powerful process by which high-resolution analysis of an ion can be achieved (7,8).

QUADRUPOLE MASS FILTERS

AND ION

TRAPS

The separation of ions passing through hyperbolic electrical fields composed of both time-dependent (AC) and fixed (DC) voltages can result in the separation of ions as to their mass-to-charge ratio. The equations of motions of ions in these fields require rather advanced mathematics to solve, and these solutions result in defining a stability region for ions of a specific m/z, dependent upon the frequency and voltage of the alternating voltage as well as the physical dimensions of the hyperbolic surface generating the field, usually a hyperbolic metal rod (9,10). If a set of four hyperbolic rods are placed in a square array with the rods at opposite corners connected to each other, then such a hyperbolic field is generated, and an m/z filtering process can be realized.

20

Functional Lipidomics

The frequency of the applied voltage is in the radio frequency (Rf) region. When a mixture of ions enter perpendicular to the square array, ions with a single m/z value emerge from the opposite end of the rod set if they have the required ion stability and do not hit or otherwise exit between the rod set. One advantage of this type of mass spectrometer is that fairly high pressure in the mass-filtering region can be tolerated, making the interface to a gas chromatograph or liquid chromatograph somewhat easier. One can use this device to scan all the ions sequentially by changing the applied voltage and obtain a mass spectrum if an appropriate detector is placed at the end of the rod sets. This filtering device can also be used to select a single ion that can then be passed on to a collision cell for tandem mass spectrometry. This same time-dependent hyperbolic electric field is a basis of the ion trap which is a closed arrangement with a short hyperbolic rod set enclosed by hyperbolic end caps. With the right set of Rf and applied voltages, ions can be made to undergo stable and somewhat circular orbits within the ion trap. Ions are ejected for detection by changing the electrical fields, causing the ions to become unstable in their orbits and eventually exit the ion trap. The entirely electronic ion trap is unusual in that it is a relatively small instrument, which can trap, ions of a broad range of m/z values and sequentially scan them out to generate a mass spectrum of the ion population within the ion trap. Alternatively, it can save a specific ion (m/z value) within the initial population of trapped ions while ejecting all other ions. This ion can be stored and forced to undergo a larger, stable orbit in which it can encounter neutral gases placed in the vacuum system (e.g., He) (11). After sufficient number of collisions, the trapped ions may decompose into one or more product ions. This collision-induced product ion spectrum can be produced by scanning out ions (MS/MS). Furthermore, this process can be repeated by selecting a population of product ions with a single m/z value out of a population of several product ions to generate MS3 spectra that are product ions of product ions, and the process repeated to generate MSn. A special case of the hyperbolic square array rod set is seen when the applied DC potential is left at zero, and the rods have only Rf applied. These Rf-only rod sets are used as collision cells because of their strong focusing characteristics. After collision of an ion within the cell with a neutral gas molecule such as nitrogen or argon, ions will change their trajectory because of collision vectors (collisional scattering). They are not lost but rather focused to exit the cell by the Rf-only quadrupole field. To impart sufficient energy to a precursor ion, three to four collisions may be necessary, which require pressures in the Rf-only quadrupole of 0.5 to 1 torr.

INTERFACE

TO

HIGH-PRESSURE LIQUID CHROMATOGRAPHY (HPLC)

One of the most important advances that has been made in the past 15 yr in the area of mass spectrometry of lipids is the development of electrospray ionization (ESI) as an efficient means to generate gas-phase ions from nonvolatile biomolecules dissolved in an electrically conductive solution and an interface between HPLC and the mass spectrometer. This advance has permitted direct analysis of lipids as they are separated by HPLC, thus combining the power of HPLC with mass spectrometric

LC/MS Methodology in Lipid Analysis and Structural Characterization

21

analysis. The electrospray process is able to generate ions free from the solvent at atmospheric pressure, and once formed, the ions can be directly introduced into the mass spectrometer using electrostatic focusing. Two different approaches have been commercialized; one in which a curtain of dry nitrogen gas is used to remove neutral solvent molecules, permitting the ions to enter the mass spectrometer through a skimmer–nozzle arrangement, in a process termed ionspray (12). The second process involves passing the electrospray droplets through a heated capillary that reduces the pressure from atmospheric (electrospray region) to vacuum conditions (mass spectrometer region) (13). Both approaches and variants on either theme have been very successful in interfacing the liquid chromatograph to the mass spectrometer.

ESI MECHANISM ESI generates gas-phase ions from an electrically conductive solution, generally as the effluent of an HPLC, in a process that involves several stages. The first stage in this process is the formation of droplets at the tip of a capillary tube held at a high potential (either positive or negative). The ions in the HPLC effluent, which carry the same charge as that applied to the capillary, will accumulate at the surface. The repulsion of the ions at the tip of the capillary overcomes the surface tension of the solvent and expands to form a Taylor cone that elongates into a liquid filament which breaks apart to form a fine jet of charged droplets (14). The second stage in the ionization process occurs as these droplets travel toward the entrance of the mass spectrometer, often with a countercurrent of dry nitrogen gas. The solvent evaporates from the droplets, reducing their diameter until they reach the Rayleigh limit, at which the repulsion of the charges within the droplet exceeds the surface tension. At this point, the droplets undergo a coulombic explosion, generating several smaller charged droplets that continue to travel toward the mass spectrometer. This cycle is repeated until droplets that carry a single ion are formed. The final stage in the electrospray process is the formation of gas-phase ions from these droplets. There are two proposed mechanisms for the formation of these ions. The first is the charged residue model, in which the solvent from the droplets is completely evaporated, leaving only the ion behind (15). The second is the ion evaporation model in which a single ion is ejected from highly charged droplets whose radius has been sufficiently reduced (16). Attention to the HPLC mobile phase is essential for successful liquid chromatography/mass spectrometry (LC/MS) of lipids by ESI. Various solvent-related parameters are critical, including volatility of the solvent system, electrical conductivity, and electrospray performance. Fortunately, these criteria are easily met with various solvent systems commonly used for lipid separations. The greatest amount of work has been carried out with methanol, acetonitrile, and water mixtures containing volatile salts, such as ammonium acetate, often 1 to 10 mM. ESI of lipids is also compatible with more organic solvent systems, including those containing chloroform, isopropanol, and even hexane. In addition, some lipid analyses have involved addition of reagents to the mobile phase after HPLC separation but before entering the electrospray needle. Additions have included alkali metals such as Li+ (17), K+ (18), as well as metal ions that form -complexes with lipids such as Ag+

22

Functional Lipidomics

(19). These metal ions are added to alter the nature of the ions formed by electrospray and, hence, the resultant ion chemistry. However, some HPLC solvent systems are not compatible with ESI, such as ion-pairing HPLC. The presence of detergents, even at trace levels, often leads to abundant unwanted ions that can even suppress the formation of the ions from a target lipid. Trifluoroacetic acid (TFA) is often used to improve the HPLC chromatography of peptides, and LC/MS, carried out in the positive ion mode, is usually performed. However, this additive causes considerable difficulty when subsequent analysis is attempted in the negative ion mode because the TFA present in the vacuum system forms an abundant ion at m/z 113, which suppresses the formation of other negative ions, including those from the target lipid. Another interface for LC/MS involves the process of atmospheric pressure chemical ionization (APCI), and this experimental approach has found particular application for generating ions from neutral lipids (20). In APCI, the HPLC effluent, composed of a mobile-phase system (with the same limitations criteria as for electrospray), is nebulized into a tube that is heated to 300 to 400˚C in order to volatilize the solvent. The flowing gases and droplets leaving the heated tube are exposed to a corona discharge generated from a sharp needle point held at high potential. A plasma of protons, ionic species, or mobile electrons (under negative ion conditions) are present in the corona discharge plume, which effect ionization of the lipid by an ion–molecule reaction. Ions are then swept into the mass spectrometer by way of a skimmer–nozzle or a heated capillary, as is the case for the ESI interface. The APCI process can generate positive and negative molecular ion species, depending on the voltage applied to the discharge needle. This method can tolerate relatively high HPLC flow rates, even up to 1 ml/min. A relatively new technique is atmospheric pressure photoionization (APPI), which is also used as the interface as well as ionization process for HPLC/MS (21). The physical apparatus for APPI is similar to that for APCI, except that the corona discharge needle is replaced with a light source of high-energy photons, e.g., a krypton lamp that emits photons with an energy of 10.1 eV. The HPLC effluent passes into the heated nebulizer after mixing with approximately 1% of a dopant such as toluene. The energy of the lamp wavelength is sufficient to photoionize the toluene and generate a radical cation from this dopant. As this occurs (at atmospheric pressure) within the interface, a rapid succession of ion–molecule reactions take place, eventually leading to a complex mixture of abundant proton-donating reactant ions such as [CH3OH⋅⋅⋅H]+. These reactant ions can transfer a proton to the analyte, for example, a neutral lipid, to form a protonated molecular ion [M+H]+, which enters the mass spectrometer by the skimmer–nozzle. The application of this technique to neutral lipid analysis is under active development, but only a few reports have emerged; these have involved the analysis of steroids (22) and wax esters (23).

MOLECULAR ION SPECIES The ESI technique typically generates a single ion species of either charge state from most lipids. A protonated species corresponding to the molecular weight plus one proton, designated [M+H]+, or an ion that has lost one proton, designated [M−

LC/MS Methodology in Lipid Analysis and Structural Characterization

23

H]−, are typical ions formed by ESI. Although not strictly molecular ions, because neither has the correct molecular weight of the lipid, these ions are often termed molecular ion species because they carry the molecular weight information plus or minus one proton depending upon the charge of the ion. ESI generates either positive or negative molecular ion species largely depending upon the solution chemistry of the lipid. For example, fatty acids readily form carboxylate anions [M−H] in solution by loss of the carboxyl proton through proton dissociation depending upon the pH of the solution; this dissociation also occurs while being sprayed in the ESI interface. Glycerophosphocholine lipids have a quaternary nitrogen atom that is always ionized [R-N(CH3)3]+ in solution. Thus, abundant quantities of [M+H]+ are readily observed from these complex lipids. In this case, addition of the proton neutralizes a phosphate anion present in the phosphodiester moiety. If simple acid–base solution chemistry predicts a facile proton addition or proton loss, then a cation or anion, as the case may be, is observed in the mass spectrometer. For neutral lipids, this situation is somewhat more complex and, correspondingly, these lipids are more difficult to ionize, often leading to somewhat less sensitivity in the MS experiment. One must consider cation affinities of the neutral lipid because it is possible to attach a proton, alkali metal, or ammonium ion and in this way, form a positive ion species in the gas phase. For a proton to associate with the lipid, a structural moiety within the lipid structure must have a higher proton affinity than that of the proton donor present in the electrospray droplet. For those neutral lipids that have only ester moieties as functional groups, cation attachment (H+, Li+, NH4+) usually predominates. Special cases of reactions leading to ionization have also been used with success, such as formation of complexes for lipids containing one or more double bonds, which have a high affinity for Ag+ ions (19). The statement that a single cation or anion is only generated by ESI is an oversimplification, as isotope-containing species are always observed; sometimes, a population of adducts is observed, too (Figure 2.1). The observed ESI-formed ions typically depend upon the presence of trace quantities of alkali metal ions such as sodium and potassium, which are often found in lipids isolated from biological origins. Moreover, when elevated lipid concentrations are analyzed, dimeric species with attachment-charging ions such as [2M+H]+ are observed. The natural isotopes of carbon and oxygen, the most abundant atoms with significant stable isotopes present in lipids, add an important and useful character to the molecular ion region. However, when complex mixtures of molecular species of lipids are present that differ only by one double bond or more, the presence of isotopes makes analysis somewhat less straightforward. Table 2.1 lists the most recent accepted values for the exact masses and isotope abundances of those elements of which lipids are typically composed. Note that the natural abundance of carbon13 is now assigned a value of 1.07%, somewhat less than the value of 1.1% used prior to 1995 (24). This drop in the accepted abundance of carbon-13 is a better reflection of the fact that this isotope of carbon present in the terrestrial biosphere is somewhat less then that of the carbon-13 isotope in deep-ocean carbonates, because of a small but finite isotope effect in carbon cycle biochemistry on a geological scale.

24

Functional Lipidomics

O O PO + O O N OH O 16:0a/20:4 GPCho O

100%

[M+H]+ 782.9

[M+H]+ 782

[M+K]+ 820.8 783.9

[M+K]+ 820

821.8 [M+Na]+ 804.8 805.8 806.8

Relative Intensity

784.9 785.9 770

780

790

800

810 m/z

822.9 823.8 820

830

840

[M+Na]+ 804

700

800

[2M+H]+ 1564.4 900

1000

1100 m/z

1200

1300

1400

1500

1600

FIGURE 2.1 Molecular ion region of the electrospray-generated ions from 1-hexadecanoyl2-arachidonoyl-glycerophosphocholine (16:0a/20:4-GPCho). The inset provides an expanded mass scale so that isotopes can be observed.

TABLE 2.1 Exact Mass and Isotopic Abundance of Common Elements Present in Lipids Exact Mass Carbon Hydrogen Nitrogen Oxygen

Sodium Phosphorous

12 (by definition) 13.003354 1.007825 2.014101 14.00307 15.000108 15.9949 16.99913 17.99916 22.98976 30.97376

Natural Abundance (Percent) 98.930 1.070 99.989 0.012 99.636 0.364 99.757 0.038 0.205 100.000 100.000

Note: Approved by the Commission on Atomic Weights and Abundances (IUPAC). Source: From de Laeter, J.R.; Böhlke, J.K.; De Bièvre, P.; Hidaka, H.; Peiser, H.S.; Rosman, K.J.R.; Taylor, P.D.P. Atomic weights of the elements. Review 2000 (IUPAC Technical Report). Pure Appl. Chem. 2003, 75, 683–799.

LC/MS Methodology in Lipid Analysis and Structural Characterization

25

Usually, the monoisotopic masses are used to assess the quantity of material present if a quantitative mass spectral assay is set up. Because complex lipids often have 50 or more carbon atoms, assignment of actual compositions using abundance of ions corresponding to [M+H]+ or [M−H]+ alone can be significantly in error owing to the natural abundance of carbon-13. This can be exemplified by calculating the percentage of the ion current carried by the monoisotopic ion species (containing no stable isotopes) compared to the entire ion current for all naturally occurring isotopic species. For example, the abundance and accurate mass of three isotopic species (M, M+1, and M+2) for several lipids is summarized in Table 2.2. A particular case to note is the [M+H] for 18:0/20:4-GPIns that has the monoisotopic [M−H] observed at m/z 885.12. To obtain an approximation of the relative amount of this lipid to that of the cardiolipin species (Ln4-Cl) in a mixture of these two lipids, one might assume that an equally abundant cardiolipin monoisotopic ion observed at m/z 1447.96 would be consistent with an equal amount of phosphatidylinositol in this sample. However, due to the natural abundance of carbon-13 and, to a much lesser extent, the other atoms that make up the elemental composition of each molecule, the ions at m/z 885.12 carry 61.2% of the ion current for all GPIns isotopic species (C 48 H 89 PO 11 ), whereas m/z 1447.96 from the cardiolipin (C81H141O17P2) carries only 44.3% of the ion current at this monoisotopic mass. Table 2.2 summarizes the total ion current carried by monoisotopic ion species for various lipids and the corresponding ion current carried by the isotopic ions, one and two mass units higher corresponding to the stable isotopic content in the naturally occurring compound. Naturally occurring stable isotopes are of particular importance when mass spectral data generated from biological samples are used to gain insight into the composition of isolated lipids

COLLISIONAL ACTIVATION

OF

LIPID IONS

The ionization techniques used to couple HPLC to the mass spectrometer for LC/MS analysis function by transmitting relatively low energy during the ionization process, resulting in the formation, essentially, of only molecular ion species. Abundant molecular ion species are observed with little decomposition or fragmentation for most lipids. On the one hand, this is a great advantage for determination of molecular weights or assessing the complexity of a lipid mixture, because all the ions observed are most likely of the molecular ion species. Molecular weight information, though necessary, is not sufficient if the goal of the mass spectrometric analysis is to gain insight into the structure of lipid species. Fundamental studies of gas-phase ion chemistry, beginning in the 1960s (25), opened an approach to reveal structural details from ions following collisional activation. In this experiment, an ion is forced to undergo collisions with a neutral gas atom (i.e., He or Ar) or a molecule such as nitrogen, thereby gaining energy from this collision by converting some of its translational energy into vibrational energy. In other words, the ions heat up following collision. When enough energy is absorbed by a single collision (high-velocity ions) or multiple collisions (low-velocity ions), covalent bonds can break, resulting in formation of a neutral species and a product ion at a lower mass.

C55H104NO6+

C81H141O17P2

18:0/20:4-GPInsf

POLn-TAGd,g

Ln4-CLh

Negative

Positive

Negative

1447.9649 (44.3)

874.7858 (56.5)

885.1205 (61.2)

758.5694 (64.3)

404.3887 (75.0)

Monoisotopic m/z (Relative Abundance)a 303.2330 (80.8)

b

Percentage of total molecular ion species only including +2 mass isotopes. M+1 isotope largely composed of one carbon-13 atom. cM+2 isotope largely composed of two carbon-13 atoms or one oxygen-18 atom. dNH + adduct. 4 e1-Hexadecanoyl-2-octadecadienoyl-glycerophosphocholine. f1-Octadecanoyl-2-arachidonoyl-glycerophosphoinositol. g1-Hexadecanoyl-2-octadecenoyl-3-octadecalienoyl-glycerol-ammonium adduct. hTetraoctadecadienoyl cardiolipin.

a

C48H89O11P+

16:0a/18:2-GPCho

Positive

C42H87NO10P

Cholesterold

Positive +

C27H50NO+

Name Arachidonate

Ion Polarity Negative

e

Elemental Composition C20H3102-

1448.9678 (38.8)

875.7897 (33.6)

886.5527 (31.1)

759.5733 (29.2)

405.3976 (21.9)

+1b amu m/z (Relative Abundance)a 304.2360 (17.4)

1449.9711 (16.8)

876.7931 (9.9)

887.5560 (7.7)

760.5767 (6.5)

406.3959 (3.1)

+2c amu m/z (Relative Abundance)a 305.2391 (1.8)

TABLE 2.2 Molecular Ion Species for Common Lipids with Exact Mass, Elemental Composition, and Percent Total Ionization for Major Isotopic Ions

26 Functional Lipidomics

LC/MS Methodology in Lipid Analysis and Structural Characterization

27

A major advance in this area was made with the design, construction, and commercialization of the tandem quadrupole mass spectrometer (26). In these instruments, an ion (precursor ion) is selected by the first quadrupole mass filter (Q1) and then directed into an Rf-only quadrupole mass filter (Q2), filled to relatively high pressure (0.5 to 1 torr) with a neutral gas, where multiple collisions take place. By virtue of the strong focusing field of the Rf-only quadrupole, product ions formed by the collisional process exit into a third quadrupole mass filter (Q3), operated under high-vacuum conditions, in which product ions can be measured as to their mass-to-charge ratio. One of the key advantages of this collision-induced decomposition (CID) process is that a strict precursor–product relationship is established. Any product ions observed in the third quadrupole must have originated from the initially selected precursor ion. This is extremely useful information when attempting to deduce structural information from mass spectrometric data. This arrangement also permits several powerful experiments: 1. Product ion scanning: as essentially described in the preceding text, determining all product ions following collisional activation of a single precursor ion (Q1 fixed–Q2 [CID]–Q3 scanned) 2. Precursor ion scanning: recording ion abundances at a single m/z set in Q3 while scanning Q1 (Q1 scan–Q2 [CID]–Q3 fixed) 3. Constant neutral loss scanning: Q1 and Q3 scanned at a fixed m/z offset equal to the mass of a neutral species loss (Q1 and Q3 offset linked–Q2 [CID]) 4. Multiple reaction monitoring (MRM): recording ions from a fixed m/z at Q3 while collisionally activating ions at another fixed m/z at Q1 and then switching to another pair of fixed m/z for Q1 and Q3 (Q1 fixed–Q2 [CID]–Q3 fixed) The MRM technique is employed to detect specific components eluted from the HPLC that satisfy the precursor–product relationship set up by the Q1 and Q3 mass filter settings. The MRM approach is often adopted for quantitative analysis of lipids, in which one set of precursor–product ions will detect the target lipid, and a second set of precursor–product ions will detect the stable isotope variant added to the sample as an internal standard. These techniques are often referred to as tandem mass spectrometry and abbreviated as MS/MS techniques. When coupled to HPLC, this can result in an LC/MS/MS experiment. ESI of the LC effluent permits the generation of abundant molecular ion species from essentially all nonvolatile lipids, and MS/MS can generate structurally relevant information specific to the lipid under study. A variant of the collisional activation experiments described earlier has been employed when only a single quadrupole mass analyzer is available. During ESI, one can apply a high voltage to a lens or cone device in the region in which ions first enter the vacuum system. This voltage (20 to 100 V) imparts to ions additional velocity and KE because the pressure is drastically dropping in this region, and the mean free path for ions is sufficiently long for significant acceleration to be achieved.

28

Functional Lipidomics

However, there are still neutral gas molecules from the HPLC solvents and atmospheric gases in this region of the interface so that multiple collisions take place, resulting in CID of ions. It must be kept in mind, however, that this experiment does not establish a precursor–product relationship, and that all observed ions might not arise from the molecular ion of interest but possibly from ionization of an impurity eluted from the HPLC. Such nonrelevant ions can cause problems in the interpretation of data in terms of structure.

DATA SYSTEMS An integral part of the LC/MS experiment is the mass spectrometer data system. The modern mass spectrometer is almost exclusively under computer control because of the quantity of the data being generated and the complexity of instrumental experiments. For example, the ion trap can operate in a very sophisticated manner because of the rapidity of computer control. Data-dependent operation is truly possible in that a mass spectrum from a single scan can be processed immediately (in less than 1 msec) to generate an instruction set to alter the ion trap’s fundamental mode of operation for the very next scan. Tandem quadrupole mass spectrometers use graphical user interfaces to perform complex scan operations, such as precursor ion scans, product ions scans, constant neutral loss scans, and MRM experiments. A second aspect of mass spectrometer data systems is the storage of data in experimental databases. Therefore, additional information such as time relationships of mass spectral data content, can be easily extracted. The time-base experiment is often driven by the HPLC elution time, thus rendering MS data directly relevant to HPLC elution. Powerful tools of database management and interrogation of this database are integral parts of all modern MS data systems.

PART II: ELECTROSPRAY MASS SPECTRA OF SAMPLE LIPIDS The analysis of lipids using ESI and tandem mass spectrometry is a rapidly growing area of research, largely because many of the biologically derived lipid substances that could not be rendered volatile even by derivatization techniques can now be readily analyzed by ESI. Thus, many biochemical questions that have been difficult to address before the advent of ESI, because of the absence of techniques to structurally characterize or quantitatively measure complex lipid substances, are now beginning to be addressed. There have been several recent reviews of MS/MS behavior of electrospray-generated lipid ions including eicosanoids (27), phospholipids (28,29), sphingolipids (30), and the general class of biologically derived lipids (31). Yet, much of the recent information concerning individual lipid classes remains in the primary research literature. The following mass spectral data of selected lipids is provided in this chapter as a glimpse into the tandem mass spectrometry of a wide range of diverse lipid structures. However, this compilation is not intended to be complete. All mass spectra, but one (as indicated), were obtained using a single instrument, a tandem

LC/MS Methodology in Lipid Analysis and Structural Characterization

29

quadrupole instrument that uses an Rf-only collision cell. Most spectra presented here were not obtained under LC/MS/MS conditions, but by flow injection; nonetheless, each of these lipids is amenable to either normal-phase or reversed-phase HPLC separation suitable for LC/MS/MS analysis.

FATTY ACIDS Fatty acids constitute some of the important building blocks of more complex lipids, as well as being complex lipids themselves with various substituents. ESI generates abundant carboxylate anions for all fatty acids (32). Collisional activation of the carboxylate anion for arachidonic acid (Figure 2.2a) observed at m/z 303 leads not only to product ions corresponding to the loss of CO2 (m/z 259), but also interesting carbon–carbon bond cleavage ions observed at m/z 162, 177, 205, and 231. For fatty acids containing additional functional groups, collisional activation of this carboxylate anion can lead to useful structural information. Substitution of a single hydroxyl group to form an arachidonate metabolite called 5-hydroxyeicosatetraenoic acid (5HETE) (Figure 2.2b) results not only in the shift of the molecular anion to m/z 319, but also favors fragmentation of this molecule following collisional activation of this carboxylate anion. These are observed by the abundant product ions at m/z 115 and 203, corresponding to the cleavage of carbon-5 and carbon-6 immediately adjacent to the hydroxyl substituent in 5-HETE. Even though somewhat chemically unstable, fatty acid hydroperoxides can be analyzed by ESI and tandem mass spectrometry. The [M−H] ion is readily observed, and collisional activation of this ion is exemplified by the MS/MS mass spectrum of 5-hydroperoxyeicosapentaenoic acid (5-HpEPE, Figure 2.2c). An abundant loss of water is observed at m/z 315, which has been suggested to correspond to that of the keto product 5-oxo-eicosapentaenoic acid (33). Additional carbon–carbon bond cleavage ions are observed in this CID mass spectrum. More complex lipids are exemplified by the CID mass spectrum of leukotriene B4 (LTB4) and its carboxylate anion at m/z 335 (Figure 2.3a). The abundant formation of the product ion at m/z 195 is observed and has been shown to correspond to the cleavage between carbon-11 and carbon-12 of this structure. This facile decomposition product is likely the result of a complex of cyclization reactions, leaving this bond alpha to both hydroxyl and vinylic groups rendering it much more susceptible to collision-activated scission reactions (34). The CID of prostaglandin carboxylate anions also yields carbon–carbon bond cleavage reaction products (Figure 2.3b). Collisional activation of the molecular ion [M− H] at m/z 351 results in the complex product ion spectrum shown with almost complete decomposition of all the precursor ions. The ion at m/z 333 corresponds to the loss of water from the collisionally activated [M−H] and m/z 315 after an additional loss of water. The most abundant ion at m/z 271 corresponds to the loss of two molecules of water and CO2, which is not particularly structurally revealing. However, the ions at m/z 189 corresponds to the loss of the side chain starting at carbon-15 as hexanal, carbon dioxide, and water and is much more characteristic of the PGE2 structure (35).

30

Functional Lipidomics

259

COO-

[M-H]303

Arachidonic Acid

Relative Intensity

100%

59 259

83

60

80

162

177

160

180

109

100

120

140

205

231 285 301

267

200 m/z

220

240

260

280

300

320

340

(a) 115 OH

203

COO-

115

5-HETE

Relative Intensity

100%

203

[M-H]319

257 301

151 100

120

140

160

180

200

220

240 m/z

260

280

300

320

340

360

380

400

(b)

FIGURE 2.2 ESI and CID of (a) [M−H] of arachidonic acid at m/z 303; (b) [M−H] of 5hydroxyeicosatetraenoic acid (5-HETE) at m/z 319; and (c) [M−H] of 5-hydroperoxyeicosapentaenoic acid (5-HpEPE) at m/z 333.

LC/MS Methodology in Lipid Analysis and Structural Characterization

31

201 -H2O OOH

5(S)-HpEPE

129

100%

COO-

173 147

201

[M-H]-

59

333

Relative Intensity

155

71 161

271 315

167

109 121

60

80

243 253 237

296

100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 m/z

(c)

FIGURE 2.2 (continued)

GLYCEROLIPIDS ESI of glycerolipids offers a unique challenge in that these molecules are neutral in aqueous solutions and, thus, must be charged during the ionization process. This is typically accomplished by adding an attachment ion such as the ammonium cation (NH4+), an alkali metal (Na+ or Li+) (36), or by using more energetic ionization conditions such as with APCI. Collisional activation of the ammonium ion [M+NH4]+ results in product ions shown in Figure 2.4a. For the 1-palmitoylmonoacylglycerol having a molecular weight of 330 Da, the [M+NH4]+ ion is observed at m/z 348; when collisionally activated, this results in a protonated molecular ion at m/z 331 and the protonated molecular ion minus the loss of water (m/z 313). The acyl group present in this monoglyceride is indicated by the fragment ion at m/z 239 corresponding to the acylium ion from palmitic acid (C15H31CO+) and the ion at m/z 257 corresponding to protonated palmitic acid. Similar behavior is observed for diacyl glycerols and the collisional activation of [M+NH4]+ for dihexadecanoyl glycerol (molecular weight 568 Da). This CID mass spectrum is dominated by the loss of a single acyl substituent as the free carboxylic acid (palmitic acid) observed at m/z 313 (Figure 2.4b). The same behavior is observed for the triacylglycerol that has the most abundant ion following collisional activation of [M+NH4]+ (Figure 2.4c) observed at m/z 551. This ion corresponds to the protonated molecular ion losing one of the fatty acyl groups as the free fatty acid or [M+H−RCOOH]+.

32

Functional Lipidomics

195

OH

OH COO-

LTB4

195

Relative Intensity

100%

[M-H]335

317 123 59

70

60

151

129

181

135

80

203

273 245

100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 m/z

(a) O COO-

Relative Intensity

OH -CO2 -H2 O +H

PGE2

189

2O] 315 [M-H-H 333

175 113

100

HO 189

271

100%

120

207

163

140

160

180

200

217

220

235

240 m/z

260

280

300

320

340

360

380

400

(b)

FIGURE 2.3 ESI and CID of a carboxylate anion corresponding to (a) [M−H] from leukotriene B4 (LTB4) at m/z 335 and (b) [M−H] from prostaglandin E2 (PGE2) at m/z 351.

LC/MS Methodology in Lipid Analysis and Structural Characterization

33

OH

O 16:0 MAG

O

OH NH4+

313

Relative Intensity

100%

239

109

100

120

137

140

221

165

160

180

200

220

331 [M+NH4]+ 348

257

240 m/z

260

280

300

320

340

360

380

(a) O O H O 313

16:0/16:0 DAG

Relative Intensity

100%

O

OH

NH4+

551

[M+NH4]+ 586 569 100

150

200

250

300

350 m/z

400

450

500

550

(b)

FIGURE 2.4 ESI and CID of (a) [M+NH4]+ of 1-hexadecanoyl glycerol (16:0 MAG) at m/z 348; (b) 1,2-dihexadecanoyl glycerol (16:0/16:0 DAG) at m/z 586; and (c) 1,2,3-trihexadecanoyl glycerol (16:0/16:0/16:0 TAG) at m/z 824.

34

Functional Lipidomics

O O H O

O O

O

NH4+ 551

100%

Relative Intensity

16:0/16:0/16:0 TAG

[M+NH4]+ 824 100

150

200

250

300

350

400

450 m/z

500

550

600

650

700

750

800

850

(c)

FIGURE 2.4 (continued).

GLYCEROPHOSPHOLIPIDS The family of glycerophospholipids is exceedingly complex, differing by both fatty acyl substituents on the glycerol backbone as well as the polar head group, but all generate abundant ions by ESI (37). The simplest polar head group is that of a phosphate moiety itself observed for the phospholipids termed phosphatidic acids (Figure 2.5). Phosphatidic acids generate abundant negative ions as observed for the 1-octadecanoyl, 2-arachidonoyl glycerophosphate (18:1/20:4-GPA). Collisional activation of the negative ion that corresponds to the charge localized on the phosphate residue yields carboxylate anions observed at m/z 283 and 303, corresponding to the two fatty acyl substituents esterified to the glycerol backbone as well as more abundant ions from the [M−H] corresponding to the loss of the sn-2 fatty acid observed at m/z 419, the sn-2 fatty acid loss as ketene residue observed at m/z 437, and, to a lesser extent, loss of the carboxylic acid from sn-1 at m/z 439 and loss of the sn-1 fatty acyl substituent as a ketene at m/z 457 (Figure 2.5). More complex phospholipids are illustrated by those containing ethanolamine on the polar head group, and this class of phospholipids generates both positive and negative ions in abundance. ESI in the positive ion mode of phosphatidylethanolamine leads to the formation of abundant [M+H]+ ions, and one molecular species (Figure 2.6a) yields an abundant positive ion at m/z 718. Collisional activation of this ion results in the neutral loss of 141 Da, corresponding to cleavage of the polar head group and charge retention on the acylated glycerol portion of the molecule. Probably, the site of proton attachment to this phosphatidylethanolamine was at one of the two ester moieties with the primary amine being ionized but participating in

LC/MS Methodology in Lipid Analysis and Structural Characterization

O

303 O

18:0/20:4 GPA

O O P OH O-

O

[M-H]723

O

283

100%

35

[M-H-R2COOH]-

Relative Intensity

419

437 283

303 457

200

250

300

350

400

450

500 m/z

550

600

650

700

750

800

FIGURE 2.5 ESI and CID of the molecular anion [M−H] from 1-octadecanoyl-2-arachidonoyl glycerophosphate (18:0/20:4-GPA) at m/z 723.

an ion pair interaction with the ionized phosphate residue. This species therefore contained two cation sites and one anion site, making it a singly charged positive ion. Under negative ion conditions, an abundant [M−H] is observed for this molecular species at m/z 716 (Figure 2.6b). Collisional activation of this ion results in abundant carboxylate anions at m/z 255 and 281, corresponding to the sn-1 and sn-2 positions, respectively. Another interesting class of glycerophosphoethanolamine lipids is the plasmalogen class, which contains a vinyl ether substituent at sn-1 rather than ester moiety. Collisional activation of the corresponding [M−H] ion results in a single carboxylate anion observed at m/z 327 for the 16:0p/20:6 glycerophosphoethanolamine phospholipid molecular species (Figure 2.6c). Interestingly, the ion observed at m/z 283 should not be confused with a carboxylate anion for octadecanoic acid; rather it corresponds to the loss of CO2 from this polyunsaturated fatty acid carboxylate anion, a process typically observed for highly unsaturated fatty acids (38). Phosphatidylcholine-containing lipids form quite abundant positive ions corresponding to [M+H]+, for example, m/z 782 observed for 16:0a/20:4-GPCho. Collisional activation of this [M+H]+ (Figure 2.7a) results in a rather simple collisioninduced mass spectrum dominated by the product ion at m/z 184. This ion corresponds to the phosphocholine cation and is a clear indicator of the presence of phosphocholine in this molecule (29). Negative ions can be formed from glycerophosphocholine; however, a quaternary nitrogen atom must be either neutralized by an anion adduct or demethylated to form a dimethyl amino product. This latter species is observed at m/z 766 as shown in Figure 2.7b. CID at m/z 766 [M−15]

36

Functional Lipidomics

O

O H

O

O O

16:0a/18:1 GPEtn

577

O PO OH

NH3+

577 -H

Relative Intensity

100%

[M+H]+ 718

150

200

250

300

350

400

450 m/z

500

550

600

650

700

750

800

(a) O 281

O H

O

O P O -O O

NH2

O 255

281

100% 16:0a/18:1 GPEtn

Relative Intensity

[M-H]716

255

452 478

100

150

200

250

300

350

400

450 m/z

500

550

600

650

700

750

800

(b)

FIGURE 2.6 ESI and CID of molecular ion species from 1-hexadecanoyl-2-oleoyl-glycerophosphoethanolamine (16:0a/18:1-GPEtn). (a) [M+H]+ at m/z 718; (b) [M−H] at m/z 716; and (c) CID of the molecular anion from a plasmalogen glycerophosphoethanolamine lipid, 1-O-hexadec-1′enyl-2-docosahexaenoyl-glycerophosphoethanolamine (16:0p/22:6-GPEtn) at m/z 746.

LC/MS Methodology in Lipid Analysis and Structural Characterization

O

327

O

O

O P O -O O

NH2

[M-H]746

16:0p/22:6 GPEtn

Relative Intensity

100%

37

327

283

150

200

250

300

436

350

400

450 m/z

500

550

600

650

700

750

800

(c)

FIGURE 2.6 (continued)

from 16:0a/20:4-GPCho results in abundant product ions corresponding to the esterified carboxylate anions observed at m/z 255 and 303, corresponding to palmitate and arachidonate, respectively. There are also ions observed corresponding to the loss of the sn-2 carboxylic acid from this [M−15] ion at m/z 480 (Figure 2.7b). Phosphatidylserine is another phospholipid that forms abundant positive and negative ions. This can be exemplified as the positive ion from 18:0a/18:1-GPSer that has an [M+H]+ cation observed at m/z 790. CID of this molecular cation (Figure 2.8A) results in abundant product ion at m/z 605, corresponding to the loss of the phosphatidylserine as a neutral species [M+H−185]+. The corresponding negative ion is observed for this molecular ion species as seen at m/z 788, and collisional activation of this ion (Figure 2.8b) results in an abundant ion at m/z 701, corresponding to the loss of serine [M+H−89]. The carboxylate anions are observed at m/z 281 and 283, and the loss of the sn-2 carboxylate ester as the free acid from the [M+H− 89] ion at m/z 419. The phosphatidylglycerol class of phospholipids generates more abundant negative ions; however, they can be observed as positive ions as exemplified by 14:0/14:0-GPGro molecular species that generates the abundant [M+NH4]+ at m/z 684. Collisional activation of this positive ion (Figure 2.9a) results in the protonated molecular ion species at m/z 667 and the loss of the phosphatidylglycerol polar head group [M+H−172]+ observed at m/z 495. The [M−H] anion for 14:0/14:0-GPGro is observed at m/z 665, and collisional activation of this ion (Figure 2.9b) results in abundant carboxylate anions, in this case observed as a single ion at m/z 227, as both the sn-1 and sn-2 fatty acyl substituents are identical as tetradecanoic acid.

38

Functional Lipidomics

O O O PO OH

O

O

+ N

[M+H]+ 782

O

100%

Relative Intensity

16:0a/20:4 GPCho

184

150

200

250

300

350

400

450

500 m/z

550

600

650

700

750

800

850

(a) O

303 O

O

O O P- O O

N

O 255

303

100%

Relative Intensity

16:0a/20:4 GPCho

[M-15]766

255

480 259 150

200

250

462 300

350

400

450 m/z

500

550

600

650

700

750

800

(b)

FIGURE 2.7 ESI and CID of molecular ion species from 1-hexadecanoyl-2-arachidonoylglycerophosphocholine (16:0a/20:4-GPCho). (a) [M+H]+ at m/z 782 and (b) [M−15] derived from choline demethylation during ESI at m/z 766.

LC/MS Methodology in Lipid Analysis and Structural Characterization

O H O

39

O H P O O OH

O 605

OH O

18:0/18:1 GPSer

Relative Intensity

100%

+ NH3

[M+H]+ 790

100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 m/z

(a)

O H O 18:0/18:1 GPSer

100%

O H P O O OH

O 701

NH2

OO

Relative Intensity

[M-H]788

419

437 152

283

100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 m/z

(b)

FIGURE 2.8 ESI and CID of molecular ion species from 1-octadecanoyl-2-oleoyl-glycerophosphoserine (18:0/18:1-GPSer). (a) [M+H]+ at m/z 790 and (b) [M−H] at m/z 788.

40

Functional Lipidomics

+O

O

O

495

H O H O O P O OH

OH OH

14:0/14:0 GPGro

Relative Intensity

100%

[M+H]+ 667 [M+NH4]+ 684

100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 m/z

(a) O

O O

227

100%

O H O O P- O O

OH

OH

14:0/14:0 GPGro

Relative Intensity

[M-H]665

152 171

363

437 455

100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 m/z

(b)

FIGURE 2.9 ESI and CID of molecular ion species derived from 1,2-bistetradecanoyl glycerophosphoglycerol (14:0/14:0-GPGro). (a) [M+NH4]+ at m/z 684 and (b) [M−H] at m/z 665.

LC/MS Methodology in Lipid Analysis and Structural Characterization

41

Phosphatidylinositols represent an important class of phospholipids serving not only roles with respect to their physical properties, but also as precursors for important phosphorylated species or polyphosphoinositides. In the positive ion mode, the [M+H]+ is observed at m/z 863. Collisional activation of this cation for 18:1/18:1GPIns results in a single ion species observed at m/z 603 (Figure 2.10a) corresponding to loss of the phosphoinositol neutral species from [M+H]+. More abundant negative ions are typically observed for phosphatidylinositol molecular species. This is exemplified for the same molecular species that has an abundant [M−H] anion at m/z 861. Collisional activation of this anion (Figure 2.10b) leads to the carboxylate anions, in this case observed at m/z 281 because both are identical. An abundant ion, also corresponding to the loss of a neutral carboxylic acid from sn-2, is observed at m/z 579.

SPHINGOLIPIDS The sphingolipid class corresponds to those lipids with a long-chain base and various substituents on the amino group. The ceramide sphingolipids correspond to a longchain fatty acid acylated to the long-chain base. The ceramide molecular species having oleic acid acylated to sphingosine generates an abundant [M+H]+ ion at m/z 564 by ESI. Collisional activation of this cation results in an abundant ion corresponding to the loss of one and two molecules of water at m/z 528 and 546 (Figure 2.11a). A more structurally indicative ion is observed at m/z 264. This ion is characteristic of the long-chain base sphingosine and has been used in quantitative assays of ceramides (39). Negative ions for ceramides can also be generated by ESI as typified by the same molecular species that is observed at m/z 562 for [M−H]. CID of this molecular species (Figure 2.11b) yields an abundant ion at m/z 306. Sphingosine-1-phosphate can be analyzed either as a positive ion or negative ion corresponding to protonation of either the phosphate residue or of the primary amino group. Using positive ion ESI, an ion for sphingosine-1-phosphate is observed at m/z 380 [M+H]+ (Figure 2.12a). CID of this cation results in a single product ion at m/z 264, the same ion observed following collisional activation of the sphingosine ceramide (Figure 2.11a), probably corresponding to the same structure of this product ion, only in this case loss of phosphoric acid, with a loss of phosphoric acid and water. In the negative ion mode of ESI, sphingosine-1-phosphate yields an abundant [M−H] ion at m/z 378 (Figure 12b). Collisional activation of this anion results only in phosphate-related ions at m/z 79 and 97. Another abundant sphingolipid is that of sphingomyelin that corresponds to a ceramide with phosphocholine attached to the primary alcohol of the long-chain base. This phospholipid generates an abundant positive ion because of the quaternary nitrogen substituent; for the N-octadecanoyl sphingomyelin molecular species, it is observed at m/z 731. Not surprisingly, CID of this phosphocholine lipid results in the phosphocholine cation at m/z 184, which is again characteristic of phosphocholine-containing lipids (Figure 2.13).

42

Functional Lipidomics

+O

H

OH O HO P O O OH [M-259]+ 603 603

OH OH

O H O

O 18:1/18:1 GPIns

Relative Intensity

100%

OH

150

200

250

300

350

400

450

500 m/z

550

600

650

700

750

800

850

900

(a) O O H

O O

18:1/18:1 GPIns

OH

OH OH

OH

[M-H]-

603

861

Relative Intensity

100%

O HO P O -O O

241

579 281

150

200

250

300

350

400

450

500 550 m/z

600

650

700

750

800

850

900

(b)

FIGURE 2.10 ESI and CID of molecular ion species derived from 1,2-dioleoyl-glycerophosphoinositol. (a) [M+H]+ at m/z 863 and (b) [M−H] at m/z 861.

LC/MS Methodology in Lipid Analysis and Structural Characterization

+O

43

H N H OH

100%

546

H OH

18:1 Ceramide

Relative Intensity

264

[M+H]+ 564

528 282 252 100

150

200

250

516

397 300

350 m/z

400

450

500

550

600

(a) ON H OH

H OH

[M-H]562

18:1 Ceramide

100%

Relative Intensity

306

237 263

280

322

532 514

100

150

200

250

300

350 m/z

400

450

500

550

600

(b)

FIGURE 2.11 ESI and tandem mass spectrometry of molecular ion species derived from Noctadecenoyl sphingosine (18:1 ceramide). (a) [M+H]+ at m/z 564 and (b) [M−H] at m/z 562.

44

Functional Lipidomics

+ H3N H

O O P OH OH

H OH Sphingosine 1-Phosphate 264

Relative Intensity

100%

[M+H]+ 380

60

80

100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 m/z

(a) H2N H

H OH Sphingosine 1-Phosphate

O O P OH O[M-H]378

Relative Intensity

100%

79

97 60

80

100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 m/z

(b)

FIGURE 2.12 ESI and CID of molecular ion species derived from sphingosine-1-phosphate. (a) [M+H]+ at m/z 380 and (b) [M−H] at m/z 378.

LC/MS Methodology in Lipid Analysis and Structural Characterization

45

O NH H

Relative Intensity

100%

184

O P O O OH

+ N

H OH 18:0 Sphingomyelin

[M+H]+ 731

713 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 m/z

FIGURE 2.13 ESI and CID of the [M+H]+ ion derived from N-octadecanoyl sphingomyelin (18:0 sphingomyelin) at m/z 731.

STEROLS A large class of common lipid molecules are those having a steroid nucleus derived from the cholesterol biosynthetic pathway. The neutral lipid cholesterol is found in high abundance in all membranes of cells and, as a neutral lipid, is typically ionized by the formation of an adduct with an alkali metal or with the ammonium ion. The ammonium adduct ion of cholesterol is observed at m/z 404, and collisional activation of this cation (Figure 2.14a) results in an abundant ion at m/z 386 as well as an ion at m/z 369. These ions are quite characteristic of cholesterol. Another abundant lipid in most cells corresponds to the esters of cholesterol. This is also a neutral lipid, and formation of an ammonium adduct can result in a positive ion as observed for cholesterol, corresponding to oleic acid esterified to cholesterol and the [M+NH4]+ ion observed at m/z 668. Collisional activation of this cation results in the cholesterol nucleus at m/z 369 (Figure 2.14b). Although this ion corresponds to the cholesterol portion of the molecule, the loss corresponds to that of the neutral carboxylic acid. Modification of the cholesterol ring as exemplified by an epoxide between carbon-5 and carbon-6 (5,6-epoxycholesterol) does change the CID behavior of this molecule by a population of ions between m/z 150 and 273 (Figure 2.14c). The most abundant ions correspond to the CID of [M+NH4]+ at m/z 367, 385, and 403 for [M−2H2O]+, [M−H2O−NH3]+, and [M+H]+, respectively. Many sterol molecules exist, including steroid hormones and bile acids. Two examples are presented: Taurocholic acid forms an abundant negative ion at m/z 514

46

Functional Lipidomics

H3C CH3

CH3 CH3

CH3 CH3

100%

+ HO

[M+NH4]+ 404

H

H

369 Cholesterol

Relative Intensity

NH4

135

386

333

104

100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 m/z

(a) CH3 +O

NH4 CH3

C O 369

100%

369

Relative Intensity

Cholesteryl Oleate

[M+NH4]+ 668

100

150

200

250

300

350

m/z

400

450

500

550

600

650

700

(b)

FIGURE 2.14 ESI and CID of (a) [M+NH4]+ from cholesterol at m/z 404; (b) [M+NH4]+ from cholesterol oleate at m/z 668; and (c) [M+NH4]+ from 5,6-epoxycholesterol at m/z 420.

LC/MS Methodology in Lipid Analysis and Structural Characterization

H3C CH3

47

CH3 CH3

CH3 CH3

H

H

100%

HO

385

+O

NH4 5,6-Epoxycholesterol

Relative Intensity

367

159 177

[M+NH4]+ 403 420

219

60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 m/z

(c)

FIGURE 2.14 (continued)

[M−H], because it is a sulfonic acid. Collisional activation of this anion (Figure 2.15a) yields little information about the steroid nucleus; however, it does indicate that the taurine moiety is present by the ion at m/z 124. Cholic acid itself also generates an abundant [M−H] at m/z 407, as it is a carboxylic acid. Although lowenergy collisions result in little structural information, there are ions observed at m/z 289 and 343 (Figure 2.15b). The mechanism of formation of these ions has been discussed (28).

ISOPRENOIDS A very large class of lipids is derived from the isoprene pathway, including important vitamins such as -tocopherol (vitamin E). These lipids are readily analyzed by ESI; an example is the positive ion from -tocopherol succinate, a synthetic vitamin and an antioxidant. CID of [M+H]+ at m/z 531 (Figure 2.16a) results in three abundant ions at m/z 165, 265, and 431. Negative ions can also be generated by ESI, and for this isoprenoid, the [M−H] is observed at m/z 529. Collisional activation leads to abundant ions at m/z 429 and 162 (Figure 2.16b). The diterpene 12-O-tetradecanoyl-phorbol-13-acetate is another example of a complex isoprenoid that is neutral, yet can form both positive and negative ions by ESI. Collisional activation of the ammonium adduct, m/z 634, yields a dominant ion at m/z 311, probably corresponding to the core phorbol structure (see structure in Figure 2.17A). The CID of the [M−H] at m/z 615 has its most abundant product ion as the myristate carboxylate anion at m/z 277 (Figure 2.17b).

48

Functional Lipidomics

O HO

H3C CH3

124

CH3 H

H

100%

HO

H

O O S O-

NH

[M-H]514

H OH

Relative Intensity

Taurocholic acid

124

107

496

353 100

150

200

250

300

350 m/z

400

450

500

550

600

(a) O

H3C

HO [M-H]407

100%

CH3

H

H HO

H

O-

CH3

H OH

Relative Intensity

Cholic Acid

343 289 100

150

200

250

300

350 m/z

400

450

500

550

600

(b)

FIGURE 2.15 ESI and CID of (a) [M−H] of taurocholic acid at m/z 514 and (b) [M−H] of cholic acid at m/z 407.

LC/MS Methodology in Lipid Analysis and Structural Characterization

H

49

+O

OH

O

265

O

O

100% α -Tocopherol succinate

Relative Intensity

431 165

[M+H]+ 531 513

153 111 125 100

253 150

200

250

413 300

350 m/z

400

450

500

550

600

(a) O-

O α -Tocopherol succinate

O

100%

429

O

O

Relative Intensity

162

99 [M-H]529 50

100

150

200

250

300 m/z

350

400

450

500

550

(b)

FIGURE 2.16 ESI and CID of molecular ion species derived from the isoprenoid -tocopherol succinate. (a) [M+H]+ at m/z 531 and (b) [M−H] at m/z 529.

50

Functional Lipidomics

O

OH

H3C H

311 Phorbol 12-myristate 13-acetate

100% CH2OH

+O

O

H3C HO H

+O

H H

OH

m/z 311

CH2OH

NH4

Relative Intensity

H

O

O

371 389 293 50

100

150

200

250

300

[M+NH4]+ 599

353 350 m/z

400

450

500

550

600

634 650

(a) O

O O

H3C HO H

227

100%

O

OH

O H H

CH2OH

Relative Intensity

H Phorbol 12-myristate 13-acetate

283

309

59 217

109

50

100

[M-H]615

369

269 290 327

387 537

150

200

250

300

350 m/z

400

450

500

550

597

600

650

(b)

FIGURE 2.17 ESI and CID of the molecular ion species derived from the diterpene phorbol 12-myristate-13-acetate. (a) [M+NH4]+ at m/z 634 and (b) [M−H] at m/z 615.

LC/MS Methodology in Lipid Analysis and Structural Characterization

HO HO

OH OH O COOHOH HO OH O COOH O

100%

O O HN O O O

Kdo2 Lipid A - Prod 1118 (2-) 1796

O HO P O OH O O

51

O

Relative Intensity

O

O

HO O O O HN O P OH O O O HO HO

1244

1018 1036

783 679

[M-Kdo2]1796

1226

1488 1324

439 1000

557 670 690

1097

448

500

600

700

1568

1288

800

900

1470

1552

1000 1100 m/z 1200 1300 1400 1500 1600 1700 1800

FIGURE 2.18 ESI and CID of the doubly charged sacrolipid Kdo2–lipid A observed at m/z 1118. The mass spectrometry was carried out using a quadrupole time-of-flight (QaTOF) instrument. The indicated structure is for the singly charged species, leading to the abundant AU: Kdo2 as product ion at m/z 1796 for a loss of two Kdo sugars. elsewhere?

SACCHAROLIPIDS Some of the most complex lipids are those composed of both carbohydrate and fatty acyl substructures. Kdo2–lipid A is one such example that is part of a more complex bacterial endotoxin. The diphospho Kdo2–lipid A forms a doubly charged [M−2H]2 at m/z 1118 with ESI. Collisional activation of this dianion ( 2.18) results in a rich population of simply charged ions, including ions at m/z 1796 corresponding to the loss of Kdo2 disaccharide unit. Many of the product ions can be readily rationalized.

POLYKETIDES Perhaps the most structurally diverse lipid group is that of the polyketide. These lipids are synthesized by very large multifunctional enzymes and represent modular condensation products of acetyl CoA and malonyl CoA among other CoA esters, including methylmalonyl CoA, complex structures, including many bacterial products, and many antibiotics. Only one example of a polyketide is illustrated here: the calcium ionophore A23187, also known as calcimycin. ESI generates both abundant positive and negative molecular ion species because of the multiple functional groups on this polyketide. The CID of [M+H]+ at m/z 524 results in a large number of abundant product ions, suggesting many separate pathways for carbon–carbon bond cleavage in this activated [M+H]+ (Figure 2.19a). Somewhat less fragmentation is

52

Functional Lipidomics

O

OH

+

H3C NH2

N O

100%

O

CH2 O O

N H CHCH3

506 421

Calcimycin

Relative Intensity

488

383 395

188

137 160

50

439

300

100

150

207

200

470 462 480

318

233

282

250

300

[M+H]+ 524

413

365 357

393

350

m/z

400

450

500

550

600

(a) O

O-

H3C NH

122

100%

N O

245

Calcimycin

O

CH2 O O

N H CHCH3

399

Relative Intensity

399

[M-H]522 201

50

100

150

200

257 329

250

300

m/z

355371 381

350

400

450

500

550

600

(b)

FIGURE 2.19 ESI and tandem mass spectrometry of the polyketide calcimycin, also referred to as the calcium ionophore A23187. (a) [M+H]+ at m/z 524 and (b) [M−H] at m/z 522.

observed following collisional activation of the [M−H] (Figure 2.19b). The loss of the pyrrole side chain [M−H−123] by a McLafferty-type rearrangement at a chargeremote site is a possible mechanism for the ion at m/z 399.

LC/MS Methodology in Lipid Analysis and Structural Characterization

53

CONCLUSION ESI mass spectrometry forms abundant positive as well as negative ions, in many cases, of virtually all lipids found in both animal and plant kingdoms. Collisional activation of the corresponding molecular ion species often results in structurally useful information as well as abundant product ions that can used in the quantitative analysis of these lipids. Although a few examples of the tandem mass spectra of lipids are presented, this is a vast area for investigation. Probing the details of the biochemistry of these lipids often requires a detailed understanding of the actual molecules or molecular species involved in these processes, and mass spectrometry offers the potential for carrying out both qualitative and quantitative analyses of these lipid molecules.

ACKNOWLEDGMENTS This was supported in part by a grant from the National Institutes of Health (U54GM69338). The sample of Kdo2–lipid A was a generous gift from Professor Christian R.H. Raetz (Duke University).

References 1. Thomson, J.J. Rays of positive electricity. Phil. Mag. 1911, 6 (20), 752–767. 2. Urey, H.C., Brickwedde, F.G. and Murphy, G.M. A hydrogen isotope of mass. Phys. Rev. 1932, 39, 164–165. 3. Paul, W. and Steinwedel, H. A new mass spectrometer without a magnetic field. Z. Naurforsch. 1953, 8a, 448. 4. Murphy, R.C. Mass Spectrometry of Lipids. The Handbook of Lipid Research. Plenum Press: New York, 1993, vol. 5. Roboz, J. Introduction to Mass Spectrometry Instrumentation and Techniques. John Wiley & Sons: New York, 1968. 6. Busch, K.L., Glish, G.L., and McLuckey, S.A. Mass Spectrometry/Mass Spectrometry Techniques and Applications of Tandem Mass Spectrometry. VCH Publishers: New York, 1988; 333 pp. 7. de Hoffmann, E.; Stroobant, V. Mass Spectrometry: Principles and Applications. 2nd ed. John Wiley & Sons: Chichester, England, 2002.. 8. Biemann, K. Mass Spectrometry Organic Chemical Applications. McGraw-Hill: New York, 1962. 9. Dawson, P.H. Quadrupole Mass Spectrometry and Its Applications. Elsevier: Amsterdam, The Netherlands, 1976.. 10. Watson, J.T. Introduction to Mass Spectrometry. 3rd ed. Raven Press: New York, 1997. 11. March, R.E., and Todd, J.F.J. Practical Aspects of Ion Trap Mass Spectrometry. CRC Press: Boca Raton, FL, 1995. 12. Covey, T.R., Bonner, R.F., Shushan, B.I., and Henion, J. The determination of protein, oligonucleotide and peptide molecular weights by ion-spray mass spectrometry. Rapid Commun. Mass Spectrom. 1988, 249–256. 13. Chowdhury, S.K., Katta, V., Chait, B.T. An electrospray-ionization mass spectrometer with new features. Rapid Commun. Mass Spectrom. 1990, 4, 81–87.

54

Functional Lipidomics 14. Gaskell, S.J. Electrospray: principles and practice. J. Mass Spectrom. 1997, 32, 677–688. 15. Kebarle, P. A brief overview of the present status of the mechanisms involved in electrospray mass spectrometry. J. Mass Spectrom. 2000, 35, 804–817. 16. Iribarne, J.V., and Thomson, B.A. On the evaporation of small ions from charged droplets. J. Chem. Phys. 1976, 64, 2287–2294. 17. Hsu, F.F., Turk, J., Thukkani, A.K., Messner, M.C., Wildsmith, K.R., and Ford, D.A. Related articles, links abstract characterization of alkylacyl, alk-1-enylacyl and lyso subclasses of glycerophosphocholine by tandem quadrupole mass spectrometry with electrospray ionization. J. Mass Spectrom. 2003, 38, 752–763. 18. Ho, Y.P., Huang, P.C., and Deng, K.H. Related articles, links abstract metal ion complexes in the structural analysis of phospholipids by electrospray ionization tandem mass spectrometry. Rapid Commun. Mass Spectrom. 2003, 17, 114–121. 19. Seal, J.R., Havrilla, C.M., Porter, N.A., and Hachey, D.L. Analysis of unsaturated compounds by Ag+ coordination ionspray mass spectrometry: studies of the formation of the Ag+/lipid complex. J. Am. Soc. Mass Spectrom. 2003, 14, 872–880. 20. Byrdwell, W.C. Atmospheric pressure chemical ionization mass spectrometry for analysis of lipids. Lipids. 2001, 36, 327–346. 21. Raffaelli, A., Saba, A. Atmospheric pressure photoionization mass spectrometry. Mass Spectrom Rev. 2003, 22, 318–331. 22. Trosken, E.R., Straube, E., Lutz, W.K., Volkel, W., and Patten, C. Quantitation of lanosterol and its major metabolite FF-MAS in an inhibition assay of CYP51 by azoles with atmospheric pressure photoionization based LC-MS/MS. J. Am. Soc. Mass Spectrom. 2004, 15, 1216–1221. 23. Barkley, R.M., and Murphy, R.C. Photospray ionization and tandem mass spectrometry of wax esters. Proc. Ann. Conf. Mass Spectrom. 2004, 52, A042573. 24. de Laeter, J.R., Böhlke, J.K., De Bièvre, P., Hidaka, H., Peiser, H.S., Rosman, K.J.R., and Taylor, P.D.P. Atomic weights of the elements. Review 2000 (IUPAC Technical Report). Pure Appl. Chem. 2003, 75, 683–799. 25. Jennings, K.R. Collision-induced decompositions of aromatic molecular ions. Int. J. Mass Spectrom. Ion Phys. 1968, 1, 227–235. 26. Yost, R.A. and Enke, C.G. Selected ion fragmentation with a tandem quadrupole mass spectrometer. J. Am. Chem. Soc. 1978, 100, 2274–2275. 27. Dickinson, J.; Murphy, R.C. Analysis of leukotriene and lipoxygenase products by electrospray mass spectrometry. in The Encyclopedia Of Mass Spectrometry (Gross, M.L.; Caprioli, R., Eds.) Vol. 3, Elsevier: New York, in press. 28. Griffiths, W.J. Tandem mass spectrometry in the study of fatty acids, bile acids, and steroids. Mass Spectrom. Rev. 2003, 22, 81–152. 29. Pulfer, M., and Murphy, R.C. Electrospray mass spectrometry of phospholipids. Mass Spectrom. Rev. 2003, 22, 332–364. 30. Han, X., and Gross, R.W. Global analyses of cellular lipidomes directly from crude extracts of biological samples by ESI mass spectrometry: a bridge to lipidomics. J. Lipid Res. 2003, 44, 1071–1079. 31. Murphy, R.C., Fiedler, J., and Hevko, J. Analysis of nonvolatile lipids by mass spectrometry. Chem. Rev. 2001, 101, 479–526. 32. Kerwin, J.L., Wiens, A.M., and Ericsson, L.H. Identification of fatty acids by electrospray mass spectrometry and tandem mass spectrometry. J. Mass Spectrom. 1996, 31, 184–192.

LC/MS Methodology in Lipid Analysis and Structural Characterization

55

33. MacMillan, D.K. and Murphy, R.C. Analysis of lipid hydroperoxides and long-chain conjugated keto acids by negative ion electrospray mass spectrometry. J. Am. Soc. Mass Spectrom. 1995, 6, 1190–1201. 34. Wheelan, P. Zirrolli, J.A., and Murphy, R.C. Negative ion electrospray tandem mass spectrometric structural characterization of leukotriene B4 (LTB4) and LTB4-derived metabolites. J. Am. Soc. Mass Spectrom. 1996, 7, 129–139. 35. Hankin, J., Wheelan, P., and Murphy, R.C. Identification of novel metabolites of PGE2 formed by isolated rat hepatocytes. Arch. Biochem. Biophys. 1997, 340, 317–330. 36. Han, X., and Gross, R.W. Quantitative analysis and molecular species fingerprinting of triacylglyceride molecular species directly from lipid extracts of biological samples by electrospray ionization tandem mass spectrometry. Anal. Biochem. 2001, 295, 88–100. 37. Murphy, R.C. Mass Spectrometry of Phospholipids: Tables of Molecular and Product Ions. Illuminati Press: Denver, CO, 2002. 38. Zemski Berry, K. and Murphy, R.C. Free radical oxidation of plasmalogen glycerophosphocholine containing esterified docosahexaenoic acid. Structure determination by mass spectrometry. Antioxid. Redox Signal. submitted (2004). 39. Liebisch, G., Drobnik, W., Reil, M., Trumbach, B., Arnecke, R., Olgemoller, B., Roscher, A.;, and Schmitz, G. Quantitative measurement of different ceramide species from crude cellular extracts by electrospray ionization tandem mass spectrometry (ESI-MS/MS). J. Lipid Res. 1999, 40, 1539–1546.

3

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids Luciano De Petrocellis, Mario van der Stelt, and Vincenzo Di Marzo

CONTENTS The “Functional Plasticity” Concept: an Introduction ...........................................57 Endocannabinoids and Their Best-Known Molecular Targets, the Cannabinoid Receptors ............................................................................59 Metabolism of Endocannabinoids — A Key to Their Functional Plasticity .........60 Metabolic Plasticity of Endocannabinoids ....................................................62 Anandamide, the Most Promiscuous of the Endocannabinoids: A Typical Example of Cell-Compartment-Dependent Plasticity..................63 Neurotransmitter-Gated Channels..................................................................63 Calcium Channels ..........................................................................................63 Potassium Channels .......................................................................................64 TRPV1 Channels............................................................................................65 Anandamide as an Intracellular Messenger ..................................................65 Peculiarities of Bioactive Lipids and Endocannabinoids: Functional Plasticity and More......................................................................67 Acknowledgments....................................................................................................68 References................................................................................................................70

THE “FUNCTIONAL PLASTICITY” CONCEPT: AN INTRODUCTION Unlike most mediators, bioactive lipids are endowed with notable functional plasticity, i.e., their functional actions may be exerted via different molecular targets and sometimes with opposing outcomes, in different cell compartments. This “compartment-dependent activity” has been recently suggested, for example, for sphingosine-1-phosphate (S1P). This compound can exert both mitogenic and antimito-

57

58

Functional Lipidomics

genic actions by activating one of its cell membrane G-protein-coupled receptors (GPCR), the S1P1 receptor; alternatively, when produced intracellularly from the phosphorylation of sphingosine by sphingosine-1-kinases, it causes similar proproliferative and survival-stimulating effects via still-undefined mechanisms, possibly by targeting the ras oncogene product, the p21ras small G-protein (1–4). Furthermore, lipid mediators are characterized by metabolic plasticity in as much as their enzymatic conversion into other lipids can redirect activity towards other molecular targets. Thus, starting from a certain biosynthetic precursor, different targets can be activated in different compartments, again with possible opposing final effects. This may occur either at different time intervals, as when passing from ceramide to S1P [Figure 3.1(a)], or from phosphoinositide-4,5-phosphate to diacylglycerols (DAGs); or simultaneously, as with lysophosphatidic acids and arachidonic acid, when they are produced from the action of certain phospholipase A2 enzymes on phosphatidic acid (PA) [Figure 3.1(b)]. In this chapter, we will discuss the concept of functional plasticity of lipid mediators by using endocannabinoids as an example. Indeed, as will be reviewed here, these relatively novel eicosanoids, originally defined as endogenous agonists for the GPCRs of the major cannabis active principle 9-tetrahydrocannabinol, but now known to interact also with other molecular targets (5,6, for reviews), represent an important example, at once, of both compartment-dependent and metabolic plasticity. MAPKs

Ceramide Sphingosine

SIP1-3

Sphingosine-1-P (a) PC PA LPA

AA

LPA1-n

Ion channels (b)

FIGURE 3.1 (a) Ceramides, sphingosine, and spingosine-1-phosphate as an example of the metabolic functional plasticity of lipids, in which different signals with different molecular targets are generated sequentially in time. (b) Phospholipid metabolism as an example of the metabolic functional plasticity of lipids, in which different signals with different molecular targets are generated simultaneously. Legend: MAPK, proteins belonging to the family of the mitogen-activated protein kinases; S1P1-3, GPCRs selective for sphingosine-1-phosphate (S1P); PC, phosphatidylcholine; PA, phosphatidic acid; LPA, 2-lyso-phosphatidic acid; AA, arachidonic acid; and LPA1-n, GPCRs selective for LPA.

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

59

ENDOCANNABINOIDS AND THEIR BEST-KNOWN MOLECULAR TARGETS, THE CANNABINOID RECEPTORS To date, five putative endocannabinoids have been discovered. N-arachidonoylethanolamine (AEA, anandamide) (7), 2-arachidonoylglycerol (2-AG) (8,9), 2-arachidonylglyceryl ether (noladin) (10,11), O-arachidonoylethanolamine (virodhamine) (12), and N-arachidonoyldopamine (NADA) (13,14). However, only two of these endogenous compounds, AEA and 2-AG, have been fully characterized as physiological ligands of cannabinoid receptors, and their biosynthetic and metabolic pathways characterized. AEA belongs to the family of the N-acylethanolamines (NAEs), which has been long investigated before the identification of the endocannabinoids (15–19), leading to the conclusion that those compounds are biosynthesized via a phospholipid-dependent pathway consisting of the enzymatic hydrolysis of the corresponding N-acyl-phosphatidylethanolamines (NAPEs). The enzyme catalyzing this reaction, originally named phospholipase D selective for NAPEs (NAPE-PLD), and shown to exhibit catalytic properties quite different from other PLD enzymes, has been recently cloned and characterized as a hydrolase belonging to the zinc-metalloprotease family of the -lactamase fold (20). NAPE-PLD does not recognize phosphatidylcoline and phoshatidylethanolamine as substrates, and it is widely distributed in mouse organs with highest concentrations in the brain, kidney, and testis. Accordingly, the amino acid sequence reported for NAPE-PLD does not show homology with those of other PLDs, such as the mammalian PLD1 and PLD2 or the glycosylphosphatidylinositol-specific PLD (20). The phospholipidic precursors of NAEs are in turn produced from the enzymatic transfer of an acyl group from the sn-1 position of phospholipids to the N-position of phosphatidylethanolamine (PE), catalyzed by a Ca2+-dependent trans-acylase. This pathway does not generate a large amount of AEA because the levels of arachidonic acid esterified at the sn-1 position of phospholipids are usually very low. This is in agreement with the observation that AEA levels are generally lower than those of the other NAEs in most of the tissues analyzed so far. In fact, both the NAPE-PLD and the trans-acylase do not appear to be selective for the biosynthesis of a particular NAE or NAPE, respectively. Most importantly, these enzymes appear to be located preferentially in intracellular rather than plasma membranes. In unstimulated tissues and cells, the levels of 2-AG are higher than those of AEA, but they are probably overestimated due to artifactual 2-AG production (21). This simple observation suggests that only a part of 2-AG found in tissues is used to activate cannabinoid receptors. In fact, this endocannabinoid is an important precursor and/or degradation product of phospho-, di- and triglyceride pathways. In most cases, 2-AG is produced from the hydrolysis of DAGs containing arachidonate in the 2-position, catalyzed by a DAG lipase selective for sn-1 position. DAGs, in turn, can be produced from the hydrolysis of either phosphoinositides (PI), catalyzed by a PI-selective phospholipase C (PI-PLC), as in macrophages, platelets, and cortical neurons, or of PA, catalyzed by a PA phosphohydrolase, in mouse neuroblastoma cells N18TG2 and in a rat microglial RTMGL1 cell line (22–28). Regarding the enzymatic conversion of DAGs into 2-AG, two sn-1 DAG lipase isozymes (DAGLα and DAGLβ) have been cloned, enzymatically characterized, and found

60

Functional Lipidomics

to catalyze the formation of the endocannabinoid 2-AG in intact cells (29). Based on the amino acid sequence, it was possible to show that both enzymes contain a lipase-3 and a serine lipase motif and to suggest the presence of four transmembranespanning domains with the amino terminus in the citosolic side. Both proteins, transfected in COS-7 cells, are mostly localized in the plasma membrane, and this is at variance with the seemingly intracellular distribution of AEA-biosynthesizing enzymes. By definition, endocannabinoids, and AEA and 2-AG in particular, bind to and functionally activate the cannabinoid receptors (30). These are seven-trasmembrane spanning proteins coupled to G-proteins of the Gi/o family. Mammalian tissues contain at least two types of cannabinoid receptors, CB1 and CB2. The former receptors are expressed in the central nervous system and in peripheral tissues including immune cells, reproductive system, gastrointestinal tract, and lung. Although higher levels of CB1 usually are found in central and peripheral neurons, there is evidence for these receptors to be expressed in a large variety of cell types, including endothelial, epithelial, and glial cells, hepatocytes, adypocytes, and lymphocytes. Inside the brain, CB1 distribution correlates with the pharmacological properties reported for tetrahydrocannabinol (THC) and psychotropic cannabinoids. The localization of CB2 receptors is different from that of CB1 because these receptors are most abundant in the immune system, i.e., in tonsil, spleen, and immune cells (B-cells and natural killer cells). The CB1 and CB2 receptors share 44% overall identity and 68% identity for the transmembrane domains. Both receptors are coupled to pertussis-toxin-sensitive inhibition of cAMP formation and stimulate p42/p44 mitogen-activated protein kinase activity (31,32). CB1, but not CB2, receptors signal also via ion channels by inhibiting N- and P/Q-type calcium channels and by activating A-type and inwardly rectifying potassium channels (33–35). Furthermore, CB1 activation stimulates phosphatidylinositol 3-kinase and protein kinase B (36,37).

METABOLISM OF ENDOCANNABINOIDS — A KEY TO THEIR FUNCTIONAL PLASTICITY It is now clear that endocannabinoids are degraded by intracellular enzymes and through mechanisms depending on their chemical nature. One enzyme, fatty acid amide hydrolase (FAAH), has been identified as mostly responsible for AEA and, limited to in vitro preparations, 2-AG hydrolysis to arachidonic acid and ethanolamine or glycerol, respectively. FAAH was originally purified and cloned from rat liver, and catalyzes the hydrolysis also of long-chain primary fatty acid amides and glycerol esters. Its structural and catalytic properties have been fully investigated (38,39,40 for reviews). FAAH is an integral membrane protein of 597 amino acids, cloned from a wide range of species with a high degree of conservation between mouse and human, and it contains a short “amidase signature” sequence enriched in serine and glycine residues. Site-directed mutagenesis studies have identified the amino acid residues involved in the catalytic site of the enzyme, and the genomic loci containing human and mouse FAAH genes have been identified (41). FAAH is mainly expressed in microsomial membranes and has an alkaline-optimal pH. Hence,

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

61

as with AEA biosynthesis, AEA metabolism also occurs entirely at the level of intracellular membranes. Although FAAH can catalyze 2-AG hydrolysis, 2-AG levels, unlike AEA, are not increased in FAAH “knockout” mice (42,43). This observation is in agreement with the previously reported evidence regarding the existence of other enzymes catalyzing 2-AG inactivation different from FAAH (44,45). Other 2-AG hydrolases, known as monoacylglycerol lipases (MAGLs), and present in both membrane and cytosolic subcellular fractions, can catalyze 2-AG enzymatic hydrolysis. MAGLs also recognize other unsaturated monoacylglycerols as substrates, which in some cases compete with 2-AG inactivation (44,46,47). A MAG lipase, inactive on AEA and having high homology with other human and mouse MAGLs, has been cloned from human, mouse, and more recently, from rat (48–50). In the rat brain, this MAGL is present in highest levels in regions with medium–high expression of CB1 cannabinoid receptors (hippocampus, cortex, anterior thalamus, and cerebellum). Furthermore, immunohistochemical studies in the hippocampus suggested a presynaptic localization of the enzyme, supporting the role of rat MAGL in the degradation of 2-AG acting as a retrograde messenger at presynaptic CB1 receptors, and supplementing data showing that the DAGLs responsible for 2-AG biosynthesis are, instead, localized postsynaptically in the adult brain (29,50). 2-AG, unlike AEA, can be reesterified into phospholipids also before being enzymatically hydrolyzed, and this reesterification occurs through mechanisms involving phosphorylation or acylation of its hydroxyl groups (51). Most importantly, because of the presence of arachidonate moiety, the possibility that endocannabinoids can also be susceptible to oxidative mechanisms catalyzed by lipoxygenases, cyclooxygenases, and cytochrome P450 oxidases has been investigated (52). Regarding the lipoxygenase products of AEA and 2-AG, they can be formed through the action of 12- and 15- but not 5-lipoxygenases (52,53). The 12-hydroxy derivative of AEA still binds to cannabinoid receptors, whereas the 15-hydroxy derivative does not, but it inhibits FAAH (53,54). Unidentified hydroxy derivatives of AEA have been suggested to act, similar to AEA (see the following text), via vanilloid TRPV1 receptors (55). The 15-hydroxy derivative of 2-AG was recently shown to be formed in eukaryotic cells, and its potential biological actions as a peroxisome proliferatoractivated receptor (PPARα)- agonist was also investigated (56). Finally, a few investigations of P450-mediated endocannabinoid metabolism have been reported. These studies reported the production of monoxigenated AEA-derivatives through the activation of murine hepatic P450s (57,58). It has also been established that AEA and 2-AG can be enzymatically transformed into the corresponding prostaglandin ethanolamines (prostamides) and prostaglandin glyceryl esters, respectively, through the action of cyclooxygenase-2 and, subsequently, of prostaglandin synthases. Studies on the metabolism and possible interactions with cannabinoid and prostaglandin receptors of these compounds have been published (59–61). No specific molecular targets have been cloned for either prostamides or prostaglandin glyceryl esters, although it is clear that they cannot bind to, nor activate, any of the known prostanoid and cannabinoid receptors cloned to date. Very recently, we reported that prostamides stimulate cat iris contraction by a mechanism not due to transformation into prostaglandins, activation of prostanoid

62

Functional Lipidomics

receptors, enhancement of endogenous AEA levels, or gating of TRPV1 vanilloid receptors, and have suggested the interaction with novel receptors functionally expressed in the cat iris (61). Likewise, prostaglandin E2 glycerol ester was recently shown to activate, with high potency, a novel GPCR (62). This receptor is very efficaciously coupled to phosphoinositide breakdown, IP3 release, and intracellular Ca2+ mobilization, but, again, has not been characterized yet.

METABOLIC PLASTICITY

OF

ENDOCANNABINOIDS

From these data on the intracellular metabolism of endocannabinoids, it is clear that these compounds not only can be inactivated for what concerns their action at cannabinoid receptors, but can also be transformed, via oxidation reactions, into lipid mediators acting at noncannabinoid receptors, i.e., nuclear PPAR receptors, in the case of the 15-hydroxy derivative of 2-AG, or yet-to-be-identified, distinct GPCRs in the case of prostaglandin F2 ethanolamide and prostaglandin E2 glycerol ester. Furthermore, at least under certain pharmacological conditions, both AEA and 2-AG can act as precursors of arachidonic acid and eicosanoids such as: (1) cycloxygenase products, probably prostaglandins, whose formation underlies, for example, some of the cardiovascular, proemetic, and intraocular pressure-increasing actions of the two compounds (63–68); or (2) cytochrome P450 epoxide products, which act as mediators for AEA and 2-AG activation of transient receptor potential channel of the vanilloid type 4 (TRPV4) (69). All these observations, taken together, provide a highly representative example of metabolic plasticity, particularly in the case of 2-AG. In fact, this compound is itself the product of a metabolic pathway that sequentially and progressively transforms lipid mediators with certain molecular targets into other mediators with different sites of action (Figure 3.2).

15-hydroxy2-AG

PIP2

cPLA2 (+) TRPV1 (–)

DAG

PKC

2-AG

Cannabinoid CB1 & CB2 receptors

PGE2-G

New GPCR

FIGURE 3.2 Biosynthesis and metabolism of 2-arachidonoyl-glycerol (2-AG): a typical example of the metabolic functional plasticity of lipid mediators. Hydrolysis of phosphoinositides (PI), in particular, phosphatidylinositol-4,5-bisphosphate (PIP2), which has been reported to activate certain types of phospholipases A2 (129) and to tonically inhibit vanilloid TRPV1 channels (130), leads to diacylglycerols (DAGs), which are known activators of protein kinase C (PKC) enzymes. When the latter compounds contain an arachidonic acid moiety on the 2position, they can generate cannabinoid-receptor-active 2-AG via the action of sn-1-selective DAG lipases (29). Finally, through the action of cycloxygenase-2 and prostaglandin E2 synthase, 2-AG can be converted into prostaglandin E2 glycerol ester (PGE2-G), for which recent data indicate the existence of a specific high-affinity GPCR (62); whereas, via the 15-lipoxygenase, 2-AG is converted into 15-hydroxy-2-AG, which has been suggested to activate PPARα- (56).

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

63

Finally, particularly worth noting is the possibility that AEA and virodhamine, which exhibit opposing functional activity at the two cannabinoid-receptor types, the former compound being selective for CB1 receptors, and the latter for CB2 receptors (12), can in principle be converted one into the other in a nonenzymatic way by simply changing the pH of the solution (70). In vivo, under pathological conditions, such as inflammation, in which the pH changes to more acidic values, AEA might be converted into virodhamine, which, by activating CB2 receptors, might exert antiinflammatory actions more efficaciously than AEA. This phenomenon may provide yet another example of the metabolic plasticity of endocannabinoids.

ANANDAMIDE, THE MOST PROMISCUOUS OF THE ENDOCANNABINOIDS: A TYPICAL EXAMPLE OF CELLCOMPARTMENT-DEPENDENT PLASTICITY Apart from cannabinoid CB1 receptors, there is increasing evidence that AEA may interact with several other molecular targets, in particular with ion channels (5,6). This evidence is schematically summarized in the following subsections.

NEUROTRANSMITTER-GATED CHANNELS AEA has been shown to interact with a variety of neurotransmitter-gated ion channels, e.g., NMDA-, serotonin (5-HT3), and 7-nicotinic acetylcholine receptors (71–74). These studies were performed in Xenopus oocytes or HEK293 cells, expressing the cloned receptors, but not cannabinoid CB1 receptors, which suggests that there is a direct interaction of AEA with these channels. Furthermore, neither the cannabinoid CB1-receptor antagonist SR141716A nor the pertussis toxin blocks the effect of AEA on these channels (71,72,75). AEA augments the NMDA-induced current (71), whereas it inhibits the serotonin- and acetylcholine-induced currents (72,73,75). The binding site of AEA was suggested to be at the transmembrane or intracellular C-terminal site for the 7-nicotinic acetylcholine receptor (76). The effect of AEA was mimicked by arachidonic acid, but inhibitors of the metabolism of AEA did not prevent its action, whereas R-methanandamide, a more metabolically stable AEA analog, exhibited a higher potency, suggesting that the intact endocannabinoid was responsible for the modulation of nicotinic receptor function (72,76). At the moment, it is unknown whether these interactions also occurr in vivo, and little work has been done to assess the physiological importance of these effects.

CALCIUM CHANNELS There are three low-voltage-activated, T-type Ca2+ channels, which contribute to pacemaker activities in the central nervous system. All three channels are inhibited by AEA at submicromolar concentrations on a presumed intracellular site (77). It was shown that T-type Ca2+ currents, which could be blocked by AEA, are involved in neuritogenesis in neuroblastoma-glioma NG108-15 hybrid cells, a cell line that recapitulates early steps of neuronal differentiation (78). AEA also noncompetitively

64

Functional Lipidomics

inhibits L-type Ca2+ channels at low μM concentrations by binding to the 1,4dihydropyridine, phenylalkylamine, and 1,5-benzothiazepine–binding sites (79). Interestingly, the former two sites have been described to be located in the lipid bilayer (80,81).

POTASSIUM CHANNELS AEA has been shown to interact with both voltage-gated and background potassium channels: 1. Voltage-gated K+ channels shape the action potential by controlling its repolarization phase, and determine the membrane potential and duration of the interspike interval. AEA and arachidonic acid have been shown to be equipotent at converting noninactivating, delayed-rectifier voltagegated K+ channels into rapidly inactivating A-type K+ channels (82). AEA induced rapid and complete inactivation of these K+ channels when they were expressed in Xenopus oocytes. It was not tested whether this effect was due to hydrolysis to arachidonic acid. However, there was no lag time upon application of AEA, which suggested that the compound had a direct interaction with the channels and that it did not need to be hydrolyzed into arachidonic acid, nor did it act by altering the phosphorylation state of the channels. Indeed, AEA was shown previously to directly inhibit Shaker-related voltage-sensitive K+ channels at low μM concentrations (83). Although this effect was shared by THC and other polyunsaturated NAEs, it was insensitive towards blockade of CB1 receptors by SR141716A. At the moment it is unknown whether AEA influences the action potential of neurons through this mechanism in vivo. 2. Background K+ channels are not voltage-gated and play an essential role in the setting of the neuronal resting membrane potential and input resistance. AEA is a direct and selective blocker of the background K+ channel TASK-1. TASK-1 channels set the resting membrane potential of both cerebellar granule neurons and somatic motoneurons, and are sensitive to low pH and anesthetics (84). It is worth noting that TASK-1 is abundantly expressed in the brain stem, in which very low cannabinoid CB1 receptors are found. AEA and methanandamide block TASK-1-mediated currents independent of G-proteins at submicromolar concentrations in CHO cells overexpressing this K+ channel. In cerebellar granule neurons, the AEAmediated inhibition of TASK-1 standing-outward K+ current induces depolarization. Interestingly, TASK-1-like background K+ currents in motoneurons and cerebellar granule neurons are inhibited by the activation of several Gq-coupled receptors, including muscarinic receptors. Therefore, in light of our finding that muscarinic-receptor stimulation leads to AEA production (see the following text), it might be speculated that AEA acts as a messenger for the inhibition of TASK-1 channels by muscarinic agonists (85), thereby resulting in neuronal excitation (84).

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

65

TRPV1 CHANNELS Cation channels of the transient receptor potential (TRP) family are critically involved in the Ca2+ homeostasis of cells. In contrast to many neurotransmitter-gated channels, TRP channels are often gated by cytoplasmic ligands and integrate multiple chemical and physical stimuli. TRP channels are candidates to mediate store-operated Ca2+ entry, which is a process in which depletion of intracellular Ca2+ stores leads to the influx of extracellular Ca2+ into the cell (86,87). In 1999, it was reported that AEA could activate the vanilloid type 1 TRP channel (TRPV1) (88,89). TRPV1 is highly expressed in small-diameter primary-afferent fibers (90). It is a molecular integrator of noxious stimuli, such as heat and low pH, and can also be activated by the pungent ingredient of hot chilli peppers, capsaicin (91). In primary sensory neurons, TRPV1 is essential for the development of inflammatory hyperalgesia (92,93). Furthermore, TRPV1 has also been found in various brain areas, including dopaminergic neurons of the substantia nigra, hippocampal pyramidal neurons, hypothalamic neurons, the locus coeruleus in the brainstem, and in various layers of the cortex (94,95), where it might be involved in modulation of synaptic plasticity. The physiological role of TRPV1 in the central nervous system is unknown, but exogenous AEA induces, in hippocampal slices, a TRPV1-mediated enhancement of paired-pulse depression, which is a form of short-term synaptic plasticity (96). Tonically active TRPV1 receptors have been found in the substantia nigra compacta (97), the ventral tegmental area (98), and the paraventricular nucleus of the hypothalamus (99). AEA has a low micromolar affinity for TRPV1 in recombinant cell lines, which is similar to that of capsaicin (100,101). However, its potency in various assays is usually five to tenfold lower than that of capsaicin. For example, in highrecombinant-expression systems, the EC50 value for AEA-induced Ca2+ influx ranges from 0.4 to 0. 5 μM, and AEA appears to act as a full agonist, whereas in assays carried out native systems such as Ca2+ influx and inward current in sensory neurons, AEA is a partial agonist with a potency varying from 6 to 10 μM (102–104). Due to its low potency and partial agonism in some assays, AEA’s ability to be a physiologically relevant activator of TRPV1 was originally controversial (105). However, it is now well established that the potency and efficacy of (exogenous) anandamide at TRPV1 are influenced by a multitude of different factors, ranging from assay conditions and species differences to TRPV1 modification and the ability of AEA to reach the intracellular binding site on TRPV1 (5,106,107).

ANANDAMIDE

AS AN INTRACELLULAR

MESSENGER

In view of the possible role of TRP channels in store-operated Ca2+ entry, we hypothesize that AEA may function as a store-operated messenger signaling to TRPV1 to gate extracellular Ca2+. This hypothesis is based on our unpublished observations that intracellular Ca2+ mobilization by thapsigargin or by receptors coupled to the PLC/IP3 pathway leads to: (1) the formation of intracellular AEA in HEK293 cells and primary sensory neurons, and (2) AEA-induced TRPV1-depen-

66

Functional Lipidomics

dent influx of extracellular Ca2+ in these cells (M. van der Stelt et al., submitted). Thus, AEA can act as an intracellular messenger capable of modulating TRPV1 channel activity. AEA might behave as an intracellular messenger with respect to the actions of chronic nicotine, which is known to either increase or decrease AEA levels in rat brain, depending on the brain region under investigation (108). Given the inhibitory effects of AEA on 7-nicotinic receptors, effects that are exerted at the level of the plasma membrane bilayer interface (see the preceding text), nicotineinduced modulation of AEA levels may represent a feedback mechanism regulating the activity of these receptors. Indeed, AEA can also directly and noncompetitively inhibit M1 and M4 muscarinic receptors at low μM concentrations (109). Although it is not known whether this interaction occurs at intracellular or extracellular sites on these receptors, through this mechanism AEA could act again as a negative feedback signal on muscarinic receptors due the activation of AEA formation that follows the activation of these receptors, now documented by both pharmacological (110) and direct analytical (M. van der Stelt et al., submitted) experiments. Finally, the inhibition of T-type channels by AEA (described earlier) also appears to be exerted at an intracellular site (77). This effect does not necessarily to underlie the modulatory effects of muscarinic-receptor activation on T-type channels, because these effects may be both inhibitory (111) and stimulatory (112). In conclusion, AEA provides a typical example of cell-compartment-dependent plasticity as it can activate CB1 receptors when its extracellular concentrations are increased, and modulate the activity of ion channels when either its intracellular or extracellular levels, depending on the type of channel, are sufficiently elevated (Figure 3.3). Interestingly, AEA can also exert direct effects on other intracellular targets such as protein kinase C (PKC) (113) and phospholipase D (114). Extracellular: S1P1 Sphingosine-1-P Intracellular: Ras

Intracellular: TRPV1 Anandamide Extracellular: CB1

FIGURE 3.3 Pleiotropy of actions for anandamide and sphingosine-1-phosphate (S1P): a typical example of the cell-compartment-dependent plasticity of lipid mediators. Legend: Sphingosine-1-P, sphingosine-1-phosphate; S1P1, GPCR selective for S1P; Ras, product of the ras oncogene, suggested to be the target for some of the intracellular actions of S1P; TRPV1, transient receptor potential (TRP) channel of the vanilloid type 1; and CB1, cannabinoid receptor of type 1.

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

67

PECULIARITIES OF BIOACTIVE LIPIDS AND ENDOCANNABINOIDS: FUNCTIONAL PLASTICITY AND MORE In the previous chapters, we have described the experimental evidence accumulated so far showing how endocannabinoids provide a highly representative example of both metabolic and cell-compartment-dependent functional plasticity, two features that are emerging as being typical of many lipid mediator classes. However, AEA and 2-AG, as well as other endocannabinoid-related lipids, and in particular those belonging to the large family of the bioactive fatty acid amides, all possess other typical peculiarities of bioactive lipids, peculiarities that are being highlighted as the experimental work on these molecules continues. One of these properties is the capability of several compounds of the same family, defined as congeners, to be biosynthesized at the same time and: (1) to exert different biological functions at different molecular targets, or (2) to concur to the same biological effect by acting, in one way or the other, at the same receptor. In the former case, one could define as functional antagonism the actions exerted by the different congeners, a typical example of this phenomenon being offered by the prostaglandins, which are all originated from the action of the same enzymes, the cycloxygenases, and all act at different receptors, thereby producing often opposing actions. In the second case, the congeners that are not capable of directly activating the receptor exert what has been defined originally as an entourage effect (46). Many members of the NAE family are now known to act independently of cannabinoid receptors. In particular, N-palmitoyl-, N-stearoyl-, and N-oleoyl-ethanolamine (PEA, SEA, and OEA) are capable of activating targets different from either CB1 or CB2 receptors, and have been suggested to bind with high affinity to: (1) CB2-like receptors, in the case of PEA; (2) non-CB1, non-TRPV1 binding sites in the brain, in the case of SEA; and (3) PPAR-α, in the case of OEA (115–117). Because these NAEs are often biosynthesized, and in some cases released, at the same time, this may lead in principle to the functional antagonism defined earlier. This might be the case of AEA and OEA, which exert orexigenic and anorexant actions via CB1 and PPAR-α, respectively (117,118), and of AEA and SEA, which exert antiapoptotic and proapoptotic effects via CB1 and uncharacterized binding sites in the brain, respectively (116,119). On the other hand, many of the cannabinoid-receptor-inactive NAEs identified so far can also enhance the effects of AEA on both CB1 and TRPV1 receptors, thereby behaving as entourage compounds for AEA. In fact, PEA potentiates the CB1-mediated anticancer actions of AEA, in part by inhibiting the expression of FAAH (120), but it also enhances, together with other NAEs including OEA, the actions of AEA on TRPV1 (101,121,122). Similar entourage effects have been observed also for N-acyldopamines (13). Of these, only Narachidonoyl- and N-oleoyl-dopamine were found to potently activate TRPV1 receptors (14,123), whereas N-stearoyl- and N-palmitoyl-dopamine are inactive per se on these channels. Yet, these two latter compounds can significantly enhance the potency at TRPV1 of both NADA and AEA, as well as of some chemicophysical agents such as pH and, apparently, heat (124). Finally, entourage actions have been reported,

68

Functional Lipidomics

and indeed described for the first time, for cannabinoid-receptor-inactive 2-acylglycerols on 2-AG, both in vitro (46) and in vivo (125). Interestingly, only part of the entourage effects observed so far for the actions of AEA at its targets are due, as initially postulated for the primary amide of oleic acid (126), to inhibition of the degradation of AEA by its congeners, which are all substrates for FAAH. In fact, the molecular mechanisms of these entourage effects are not at all clear. On the other hand, in the absence of other explanations, inhibition of degradation needs always to be taken into consideration as the more plausible mechanism of action of AEA congeners and analogs that exert cannabimimetic or vanilloid-like activities without being capable of activating cannabinoid and vanilloid receptors directly. This is perhaps the case of some N-acylaminoacids, such as N-arachidonoyl-glycine and N-arachidonoyl-alanine (127), which, similar to AEA, exert analgesic actions, and yet do not bind to either CB1 or CB2 receptors, but are capable of inhibiting FAAH and, hence, AEA enzymatic hydrolysis (127,128). In conclusion, it is clear that endocannabinoids and the related lipids (i.e., the long-chain fatty acid primary amides and esters, and the long-chain fatty acid amides with dopamine and amino acids) exhibit most of the typical features of bioactive lipids, in terms of functional plasticity (both metabolic and cell-compartment-dependent), functional antagonism, and entourage effects (Table 3.1). Yet, these compounds also encompass the typical features of many other signals (such as peptides and proteins) in as much as they too are composed of chemical units available in the cell and often used for other purposes, such as arachidonic acid, ethanolamine, glycerol, amino acids, and bioactive amines. These observations should stimulate, in the future, more and more efforts aimed at fully understanding the molecular mechanisms regulating the levels and actions of these bioactive lipids, as well as their significance under both physiological and pathological conditions.

ACKNOWLEDGMENTS The authors’ work in this field is partly supported by a grant from the Volkswagen Stiftung (to Vincent Di Marzo).

Anandamide 2-AG Noladin Virodhamine NADA and other Nacyl-dopamines N-acyl-amino acids

Unknown

Metabolic Functional Plasticity Enzymatic modification redirects activity Yes Yes Unknown Yes (with anandamide) Unknown Functional Antagonism Congeners produced at the same time exert opposing actions Yes No Unknown Unknown No Unknown

Cell-Compartment- Dependent Functional Plasticity Cell compartimentalization redirects activity Yes No Unknown Unknown Yes Unknown

Yes

Entourage Effects Inactive congeners produced at the same time modulate the effect of the “active” congeners Yes Yes Unknown Unknown Yes

TABLE 3.1 A Summary of the Typical Features of Lipid Mediators Shared by Endocannabinoids and Related Lipid Mediators

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids 69

70

Functional Lipidomics

References 1. Spiegel, S., Milstien, S. Sphingosine-1-phosphate: an enigmatic signalling lipid. Nat. Rev. Mol. Cell Biol. 4, 397–407, 2003. 2. Watterson, K., Sankala, H., Milstien, S., Spiegel, S. Pleiotropic actions of sphingosine-1-phosphate. Prog. Lipid Res. 42, 344–357, 2003. 3. Ikeda, H., Satoh, H., Yanase, M., Inoue, Y., Tomiya, T., Arai, M., Tejima, K., Nagashima, K., Maekawa, H., Yahagi, N., Yatomi, Y., Sakurada, S., Takuwa, Y., Ogata, I., Kimura, S., and Fujiwara, K. Antiproliferative property of sphingosine 1-phosphate in rat hepatocytes involves activation of Rho via Edg-5. Gastroenterology 124, 459–469, 2003. 4. Kim, D. S., Kim, S. Y., Kleuser, B., Schafer-Korting, M., Kim, K. H., and Park, K.C. Sphingosine-1-phosphate inhibits human keratinocyte proliferation via Akt/protein kinase B inactivation. Cell Signal. 16, 89–95, 2004. 5. Di Marzo, V., De Petrocellis, L., Fezza, F., Ligresti, A., and Bisogno, T. Anandamide receptors. Prostaglandins Leukot. Essent Fatty Acids 66, 377–391, 2002. 6. Pertwee, R.G. Novel pharmacological targets for cannabinoids. Curr. Neuropharmacol. 2, 9–29, 2004. 7. Devane, W. A., Hanus, L., Breuer, A., Pertwee, R. G., Stevenson, L. A., Griffin, G., Gibson, D., Mandelbaum, A., Etinger, A., and Mechoulam, R. Isolation and structure of a brain constituent that binds to the cannabinoid receptor. Science 258, 1946–1949, 1992. 8. Mechoulam, R., Ben-Shabat, S., Hanus, L., Ligumsky, M., Kaminski, N. E., Schatz, A. R., Gopher, A., Almog, S., Martin, B. R., Compton, D. R., Pertwee, R. G., Griffin, G., Bayewitch, M., Barg, J., and Vogel, Z. Identification of an endogenous 2monoglyceride, present in canine gut, that binds to cannabinoid receptors. Biochem. Pharmacol. 50, 83–90, 1995. 9. Sugiura, T., Kondo, S., Sukagawa, A., Nakane, S., Shinoda, A., Itoh, K., Yamashita, A., Waku, K. 2-Arachidonoylglycerol: a possible endogenous cannabinoid receptor ligand in brain. Biochem. Biophys. Res. Commun. 215, 89–97, 1995. 10. Hanus, L., Abu-Lafi, S., Fride, E., Breuer, A., Vogel, Z., Shalev, D. E., Kustanovich, I., and Mechoulam, R. 2-arachidonyl glyceryl ether, an endogenous agonist of the cannabinoid CB1 receptor. Proc. Natl. Acad. Sci. U.S.A. 98(7), 3662–3665, March 27, 2001. 11. Fezza, F., Bisogno, T., Minassi, A., Appendino, G., Mechoulam, R., Di Marzo, and V. Noladin ether, a putative novel endocannabinoid: inactivation mechanisms and a sensitive method for its quantification in rat tissues. FEBS Lett. 513, 294–298, 2002. 12. Porter, A. C., Sauer, J. M., Knierman, M. D., Becker, G. W., Berna, M. J., Bao, J., Nomikos, G. G., Carter, P., Bymaster, F.P., Leese, A.B., Felder, C. C. Characterization of a novel endocannabinoid, virodhamine, with antagonist activity at the CB1 receptor. J. Pharmacol. Exp. Ther. 301, 1020–1024, 2002. 13. Bisogno, T., Melck, D., Bobrov, M. Yu., Gretskaya, N. M., Bezuglov, V. V., De Petrocellis, L., Di Marzo, V. N-acyl-dopamines: novel synthetic CB(1) cannabinoidreceptor ligands and inhibitors of anandamide inactivation with cannabimimetic activity in vitro and in vivo. Biochem. J. 351, 817–824, 2000. 14. Huang, S. M., Bisogno, T., Trevisani, M., Al-Hayani, A., De Petrocellis, L., Fezza, F., Tognetto, M., Petros, T. J., Krey, J. F., Chu, C .J., Miller, J. D., Davies, S. N., Geppetti, P., Walker, J. M., and Di Marzo, V. An endogenous capsaicin-like substance with high potency at recombinant and native vanilloid VR1 receptors. Proc. Natl. Acad. Sci. U.S.A. 99, 8400–8405, 2002.

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

71

15. Schmid, P. C., Reddy, P. V., Natarajan, V., and Schmid, H. H. Metabolism of Nacylethanolamine phospholipids by a mammalian phosphodiesterase of the phospholipase D type. J. Biol. Chem. 258, 9302–9306, 1983. 16. Schmid, H. H., Schmid, P. C., Natarajan, V. N-acylated glycerophospholipids and their derivatives. Prog. Lipid Res. 29, 1–43, 1990. 17. Schmid, H. H., Schmid, P. C., Natarajan, V. The N-acylation-phosphodiesterase pathway and cell signalling. Chem. Phys. Lipids 80, 133–142, 1996. 18. Schmid, H. H., Berdyshev, E. V. Cannabinoid receptor-inactive N-acylethanolamines and other fatty acid amides: metabolism and function. Prostaglandins Leukot. Essent. Fatty Acids 66, 363–376, 2002. 19. Hansen, H. S., Lauritzen, L., Moesgaard, B., Strand, A. M., and Hansen, H. H. Formation of N-acyl-phosphatidylethanolamines and N-acetylethanolamines: proposed role in neurotoxicity. Biochem. Pharmacol. 55, 719–725, 1998. 20. Okamoto, Y., Morishita, J., Tsuboi, K., Tonai, T., and Ueda, N. Molecular characterization of a phospholipase D generating anandamide and its congeners. J. Biol. Chem. 279, 5298–5305, 2004. 21. Sugiura, T., Yoshinaga, N., and Waku, K. Rapid generation of 2-arachidonoylglycerol, an endogenous cannabinoid receptor ligand, in rat brain after decapitation. Neurosci. Lett. 297, 175–178, 2001. 22. Di Marzo, V., De Petrocellis, L., Sugiura, T., Waku, K. Potential biosynthetic connections between the two cannabimimetic eicosanoids, anandamide and 2-arachidonoyl-glycerol, in mouse neuroblastoma cells. Biochem. Biophys. Res. Commun. 227, 281–288, 1996. 23. Stella, N., Schweitzer, P., and Piomelli, D. A second endogenous cannabinoid that modulates long-term potentiation. Nature 388, 773–778, 1997. 24. Kondo, S., Kondo, H., Nakane, S., Kodaka, T., Tokumura, A., Waku, K., and Sugiura, T. 2-Arachidonoylglycerol, an endogenous cannabinoid receptor agonist: identification as one of the major species of monoacylglycerols in various rat tissues, and evidence for its generation through Ca2+-dependent and -independent mechanisms. FEBS Lett. 429, 152–156, 1998. 25. Berdyshev, E. V., Schmid, P. C., Krebsbach, R. J., Schmid, H. H. Activation of PAF receptors results in enhanced synthesis of 2-arachidonoylglycerol (2-AG) in immune cells. FASEB J. 15, 2171–2178, 2001. 26. Stella, N., Piomelli, D. Receptor-dependent formation of endogenous cannabinoids in cortical neurons. Eur. J. Pharmacol. 425, 189–196, 2001. 27. Bisogno, T., Melck, D., De Petrocellis, L., and Di Marzo, V. Phosphatidic acid as the biosynthetic precursor of the endocannabinoid 2-arachidonoylglycerol in intact mouse neuroblastoma cells stimulated with ionomycin. J. Neurochem. 72(5), 2113–2119, 1999. 28. Carrier, E. J., Kearn, C. S., Barkmeier, A. J., Breese, N. M., Yang, W., Nithipatikom, K., Pfister, S. L., Campbell, W. B., and Hillard, C. J. Cultured rat microglial cells synthesize the endocannabinoid 2-arachidonylglycerol, which increases proliferation via a CB2 receptor-dependent mechanism. Mol. Pharmacol. 65, 999–1007, 2004. 29. Bisogno, T., Howell, F., Williams, G., Minassi, A., Cascio, M. G., Ligresti, A., Matias, I., Schiano-Moriello, A., Paul, P., Williams, E. J., Gangadharan, U., Hobbs, C., Di Marzo, V., and Doherty, P. Cloning of the first sn1-DAG lipases points to the spatial and temporal regulation of endocannabinoid signaling in the brain. J. Cell Biol. 163, 463–468, 2003.

72

Functional Lipidomics 30. Di Marzo, V. and Fontana, A. Anandamide, an endogenous cannabinomimetic eicosanoid: killing two birds with one stone.’ Prostaglandins Leukot. Essent. Fatty Acids 53, 1–11, 1995. 31. Vogel, Z., Barg, J., Levy, R., Saya, D., Heldman, E., and Mechoulam, R. Anandamide, a brain endogenous compound, interacts specifically with cannabinoid receptors and inhibits adenylate cyclase. J. Neurochem. 61, 352–355, 1993. 32. Bouaboula, M., Poinot-Chazel, C., Bourrie, B., Canat, X., Calandra, B., RinaldiCarmona, M., Le Fur, G., and Casellas, P. Activation of mitogen-activated protein kinases by stimulation of the central cannabinoid receptor CB1. Biochem. J. 312, 637–641, 1995. 33. Mackie, K. and Hille, B. Cannabinoids inhibit N-type calcium channels in neuroblastoma-glioma cells. Proc. Natl. Acad. Sci. U.S.A. 89, 3825–3829, 1992. 34. Mackie, K., Lai, Y., Westenbroek, R., and Mitchell, R. Cannabinoids activate an inwardly rectifying potassium conductance and inhibit Q-type calcium currents in AtT20 cells transfected with rat brain cannabinoid receptor. J. Neurosci. 15, 6552–6561, 1995. 35. McAllister, S. D., Griffin, G., Satin, L. S., and Abood, M. E. Cannabinoid receptors can activate and inhibit G protein-coupled inwardly rectifying potassium channels in a xenopus oocyte expression system. J. Pharmacol. Exp. Ther. 291, 618–626, 1999. 36. Gomez del Pulgar, T., Velasco, G., and Guzman, M. The CB1 cannabinoid receptor is coupled to the activation of protein kinase B/Akt. Biochem. J. 347, 369–373, 2000. 37. Molina-Holgado, E., Vela, J. M., Arevalo-Martin, A., Almazan, G., Molina-Holgado, F., Borrell, J., and Guaza, C. Cannabinoids promote oligodendrocyte progenitor survival: involvement of cannabinoid receptors and phosphatidylinositol-3 kinase/Akt signaling. J. Neurosci. 22, 9742–9753, 2002. 38. Cravatt, B. F., Giang, D. K., Mayfield, S. P., Boger, D. L., Lerner, R .A., and Gilula, N. B. Molecular characterization of an enzyme that degrades neuromodulatory fattyacid amides. Nature 384, 83–87, 1996. 39. Bisogno, T., De Petrocellis, L., and Di Marzo, V. Fatty acid amide hydrolase, an enzyme with many bioactive substrates. Possible therapeutic implications. Curr. Pharm. Des. 8, 533–547, 2002. 40. Cravatt, B. F. and Lichtman, A. H. Fatty acid amide hydrolase: an emerging therapeutic target in the endocannabinoid system. Curr. Opin. Chem. Biol. 7, 469–475, 2003. 41. Patricelli, M. P. and Cravatt, B. F. Clarifying the catalytic roles of conserved residues in the amidase signature family. J. Biol. Chem. 275, 19177–19184, 2000. 42. Di Marzo, V. and Deutsch, D. G. Biochemistry of the endogenous ligands of cannabinoid receptors. Neurobiol. Dis. 5, 386–404, 1998. 43. Lichtman, A. H., Hawkins, E. G., Griffin, G., and Cravatt, B. F. Pharmacological activity of fatty acid amides is regulated, but not mediated, by fatty acid amide hydrolase in vivo. J. Pharmacol. Exp. Ther. 302, 73–79, 2002. 44. Di Marzo, V., Bisogno, T., De Petrocellis, L., Melck, D., Orlando, P., and Wagner, J. A., Kunos, G. Biosynthesis and inactivation of the endocannabinoid 2-arachidonoylglycerol in circulating and tumoral macrophages. Eur. J. Biochem. 264, 258–267, 1999. 45. Goparaju, S. K., Ueda, N., Taniguchi, K., and Yamamoto, S. Enzymes of porcine brain hydrolyzing 2-arachidonoylglycerol, an endogenous ligand of cannabinoid receptors. Biochem. Pharmacol. 57, 417–423, 1999.

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

73

46. Ben-Shabat, S., Fride, E., Sheskin, T., Tamiri, T., Rhee, M. H., Vogel, Z., Bisogno, T., De Petrocellis, L., Di Marzo, V., and Mechoulam, R. An entourage effect: inactive endogenous fatty acid glycerol esters enhance 2-arachidonoyl-glycerol cannabinoid activity. Eur. J. Pharmacol. 353, 23–31, 1998. 47. Di Marzo, V., Bisogno, T., Sugiura, T., Melck, D., and De Petrocellis, L. The novel endogenous cannabinoid 2-arachidonoylglycerol is inactivated by neuronal- and basophil-like cells: connections with anandamide. Biochem. J. 331, 15–19, 1998. 48. Karlsson, M., Reue, K., Xia, Y. R., Lusis, A. J., Langin, D., Tornqvist, H., and Holm, C. Exon-intron organization and chromosomal localization of the mouse monoglyceride lipase gene. Gene 272, 11–18, 2001. 49. Ho, S.Y., Delgado, L., and Storch, J. Monoacylglycerol metabolism in human intestinal Caco-2 cells: evidence for metabolic compartmentation and hydrolysis. J. Biol. Chem. 277, 1816–1823, 2002. 50. Dinh, T. P., Carpenter, D., Lesile, F. M., Freund, T. F., Katona, I., Sensi, S. L., Kathuria, S., and Pomelli, D. Brain monoglyceride lipase participating in endocannabinoid inactivation. Proc. Natl. Acad. Sci. U.S.A. 99, 10819–10824, 2002. 51. Sugiura, T., Kobayashi, Y., Oka, S., and Waku, K. Biosynthesis and degradation of anandamide and 2-arachidonoylglycerol and their possible physiological significance. Prostaglandins Leukot. Essent. Fatty Acids. 66, 173–192, 2002. 52. Kozak, K. R. and Marnett, L .J. Oxidative metabolism of endocannabinoids. Prostaglandins Leukot. Essent. Fatty Acids 66, 211–220, 2002. 53. van der Stelt, M., van Kuik, J. A., Bari, M., van Zadelhoff, G., Leeflang, B. R., Veldink G. A., Finazzi-Agro, A., Vliegenthart, J. F., and Maccarrone, M. Oxygenated metabolites of anandamide and 2-arachidonoylglycerol: conformational analysis and interaction with cannabinoid receptors, membrane transporter, and fatty acid amide hydrolase. J. Med. Chem. 45, 3709–3720, 2002. 54. Edgemond, W. S., Hillard, C. J., Falck, J. R., Kearn, C. S., and Campbell, W. B. Human platelets and polymorphonuclear leukocytes synthesize oxygenated derivatives of arachidonylethanolamide (anandamide): their affinities for cannabinoid receptors and pathways of inactivation. Mol. Pharmacol. 54, 180–188, 1998. 55. Craib, S. J., Ellington, H. C., Pertwee, R. G., and Ross, R. A. A possible role of lipoxygenase in the activation of vanilloid receptors by anandamide in the guineapig bronchus. Br. J. Pharmacol. 134, 30–37, 2001. 56. Kozak, K. R., Gupta, R. A., Moody, J. S., Ji, C., Boeglin, W. E., DuBois, R. N., Brash, A. R., and Marnett, L. J. 15-Lipoxygenase metabolism of 2-arachidonylglycerol. Generation of a peroxisome proliferator-activated receptor alpha agonist. J. Biol. Chem. 277, 23278–23286, 2002. 57. Bornheim, L. M., Kim, K. Y., Chen, B., and Correia, M. A. The effect of cannabidiol on mouse hepatic microsomal cytochrome P450-dependent anandamide metabolism. Biochem. Biophys. Res. Commun. 197, 740–746, 1993. 58. Bornheim, L. M., Kim, K. Y., Chen, B., and Correia, M. A. Microsomal cytochrome P450-mediated liver and brain anandamide metabolism. Biochem. Pharmacol. 50, 677–686, 1995. 59. Woodward, D. F., Krauss, A. H., Chen, J., Lai, R. K., Spada, C. S., Burk, R. M., Andrews, S. W., Shi, L., Liang, Y., Kedzie, K. M., Chen, R., Gil, D. W., Kharlamb, A., Archeampong, A., Ling, J., Madhu, C., Ni, J., Rix, P., Usansky, J., Usansky, H., Weber, A., Welty, D., Yang, W., Tang-Liu, D. D., Garst, M. E., Brar, B., and Wheeler, L.A., Kaplan, L.J. The pharmacology of bimatoprost (Lumigan). Surv. Ophthalmol. 45, S337–345, 2001.

74

Functional Lipidomics 60. Ross, R. A., Craib, S. J., Stevenson, L. A., Pertwee, R. G., Henderson, A., Toole, J., and Ellington, H. C. Pharmacological characterization of the anandamide cyclooxygenase metabolite: prostaglandin E2 ethanolamide. J. Pharmacol. Exp. Ther. 301, 900–907, 2002. 61. Matias, I., Chen, J., De Petrocellis, L., Bisogno, T., Ligresti, A., Fezza, F., Krauss, A. H., Shi, L., Protzman, C. E., Li, C., Liang, Y., Nieves, A. L., Kedzie, K. M., Burk, R. M., Di Marzo, V., Woodward, D.F. Prostaglandin ethanolamides (prostamides): in vitro pharmacology and metabolism. J. Pharmacol. Exp. Ther. 309, 745–757, 2004. 62. Nirodi, C. S., Crews, B. C., Kozak, K .R., Morrow, J. D., and Marnett, L. J. The glyceryl ester of prostaglandin E2 mobilizes calcium and activates signal transduction in RAW264.7 cells. Proc. Natl. Acad. Sci. U.S.A. 101, 1840–1845, 2004. 63. Ellis, E. F., Moore, S. F., and Willoughby, K. A. Anandamide and delta 9-THC dilation of cerebral arterioles is blocked by indomethacin. Am. J. Physiol. 269, H1859–1864, 1995. 64. Pate, D. W., Jarvinen, K., Urtti, A., Jarho, P., Fich, M., Mahadevan, V., and Jarvinen, T. Effects of topical anandamides on intraocular pressure in normotensive rabbits. Life Sci. 58, 1849–1860, 1996. 65. Jarai, Z., Wagner, J. A., Goparaju, S. K., Wang, L., Razdan, R. K., Sugiura, T., Zimmer, A. M., Bonner, T. I., Zimmer, A., and Kunos, G. Cardiovascular effects of 2-arachidonoyl glycerol in anesthetized mice. Hypertension 35, 679–684, 2000. 66. White, R., Ho, W. S., Bottrill, F. E., Ford, W. R., and Hiley, C. R. Mechanisms of anandamide-induced vasorelaxation in rat isolated coronary arteries. Br. J. Pharmacol. 134, 921–929, 2001. 67. Darmani, N. A. The potent emetogenic effects of the endocannabinoid, 2-AG (2arachidonoylglycerol) are blocked by delta(9)-tetrahydrocannabinol and other cannnabinoids. J. Pharmacol. Exp. Ther. 300, 34–42, 2002. 68. Stanke-Labesque, F., Mallaret, M., Lefebvre, B., Hardy, G., Caron, F., and Bessard, G. 2-Arachidonoyl glycerol induces contraction of isolated rat aorta: role of cyclooxygenase-derived products. Cardiovasc. Res. 63, 155–160, 2004. 69. Watanabe, H., Vriens, J., Prenen, J., Droogmans, G., Voets, T., and Nilius, B. Anandamide and arachidonic acid use epoxyeicosatrienoic acids to activate TRPV4 channels. Nature 424, 434–438, 2003. 70. Markey, S. P., Dudding, T., and Wang, T. C. Base- and acid-catalyzed interconversions of O-acyl- and N-acyl-ethanolamines: a cautionary note for lipid analyses. J. Lipid Res. 41, 657–662, 2000. 71. Hampson, A. J., Bornheim, L. M., Scanziani, M., Yost, C. S., Gray, A. T., Hansen, B. M., Leonoudakis, D. J., and Bickler, P. E. Dual effects of anandamide on NMDA receptor-mediated responses and neurotransmission. J. Neurochem. 70, 671–676, 1998. 72. Oz, M., Ravindran, A., Diaz-Ruiz, O., Zhang, L., and Morales, M. The endogenous cannabinoid anandamide inhibits alpha7 nicotinic acetylcholine receptor-mediated responses in Xenopus oocytes. J. Pharmacol. Exp. Ther. 306, 1003–1010, 2003. 73. Fan, P. Cannabinoid agonists inhibit the activation of 5-HT3 receptors in rat nodose ganglion neurons. J. Neurophysiol. 73, 907–910, 1995. 74. Kimura, T., Ohta, T., Watanabe, K., Yoshimura, H., and Yamamoto, I. Anandamide, an endogenous cannabinoid receptor ligand, also interacts with 5-hydroxytryptamine (5-HT) receptor. Biol. Pharm. Bull. 21, 224–226, 1998. 75. Barann, M., Molderings, G., Bruss, M., Bonisch, H., Urban, B. W., and Gothert, M. Direct inhibition by cannabinoids of human 5-HT3A receptors: probable involvement of an allosteric modulatory site. Br. J. Pharmacol. 137, 589–596, 2002.

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

75

76. Oz, M., Zhang, L., Ravindran, A., Morales, M., and Lupica, C. L. Differential effects of endogenous and synthetic cannabinoids on {alpha}7-nicotinic acetylcholine receptor-mediated responses in Xenopus oocytes. J. Pharmacol. Exp. Ther. 310, 1152–1160, 2004. 77. Chemin, J., Monteil, A., Perez-Reyes, E., Nargeot, J., and Lory, P. Direct inhibition of T-type calcium channels by the endogenous cannabinoid anandamide. EMBO J. 20, 7033–7040, 2001. 78. Chemin J., Nargeot J., and Lory P. Neuronal T-type alpha 1H calcium channels induce neuritogenesis and expression of high-voltage-activated calcium channels in the NG108-15 cell line. J. Neurosci. 22, 6856–6862, 2002. 79. Johnson, D. E., Heald, S. L., Dally, R. D., and Janis, R. A. Isolation, identification and synthesis of an endogenous arachidonic amide that inhibits calcium channel antagonist 1,4-dihydropyridine binding. Prostaglandins Leukot. Essent. Fatty Acids 48, 429–437, 1993. 80. Bangalore, R., Baindur, N., Rutledge, A., Triggle, D. J., and Kass, R. S. L-type calcium channels: asymmetrical intramembrane binding domain revealed by variable length, permanently charged 1,4-dihydropyridines. Mol. Pharmacol. 46, 660–666, 1994. 81. Lipkind, G. M. and Fozzard, H. A. Molecular modeling of interactions of dihydropyridines and phenylalkylamines with the inner pore of the L-type Ca2+ channel. Mol. Pharmacol. 63, 499–511, 2003. 82. Oliver, D., Lien, C. C., Soom, M., Baukrowitz, T., Jonas, P., and Fakler, B. Functional conversion between A-type and delayed rectifier K+ channels by membrane lipids. Science 304, 265–270, 2004. 83. Poling, J. S., Rogawski, M. A., Salem, N., Jr., and Vicini, S. Anandamide, an endogenous cannabinoid, inhibits Shaker-related voltage-gated K+ channels. Neuropharmacology 35, 983–991, 1996. 84. Maingret, F., Patel, A. J., Lazdunski, M., and Honore, E. The endocannabinoid anandamide is a direct and selective blocker of the background K(+) channel TASK1. EMBO J. 20, 47–54, 2001. 85. Millar, J. A., Barratt, L., Southan, A. P., Page, K. M., Fyffe, R. E., Robertson, B., and Mathie, A. A functional role for the two-pore domain potassium channel TASK1 in cerebellar granule neurons. Proc. Natl. Acad. Sci. U.S.A. 97, 3614–3618, 2000. 86. Venkatachalam, K., van Rossum, D. B., Patterson, R. L., Ma, H. T., and Gill, D. L. The cellular and molecular basis of store-operated calcium entry. Nat. Cell Biol. 4, E263–272, 2002. 87. Benham, C. D., Davis, J. B., and Randall, A. D. Vanilloid and TRP channels: a family of lipid-gated cation channels. Neuropharmacology 42, 873–888, 2002. 88. Zygmunt, P. M., Petersson, J., Andersson, D. A., Chuang, H., Sorgard, M., Di Marzo, V., Julius, D. and Hogestatt, E.D. Vanilloid receptors on sensory nerves mediate the vasodilator action of anandamide. Nature 400, 452–457, 1999. 89. Smart, D., Gunthorpe M. J., Jerman J. C., Nasir S., Gray J., Muir A. I., Chambers J. K., Randall A. D., and Davis J. B. The endogenous lipid anandamide is a full agonist at the human vanilloid receptor (hVR1). Br. J. Pharmacol. 129, 227–230, 2000. 90. Caterina, M. J., Schumacher M. A., Tominaga M., Rosen T. A., Levine J. D., Julius D. The capsaicin receptor: a heat-activated ion channel in the pain pathway. Nature 389, 816–824, 1997. 91. Szallasi, A. and Blumberg, P. M. Vanilloid (Capsaicin) receptors and mechanisms. Pharmacol. Rev. 51, 159–212, 1999.

76

Functional Lipidomics

92. Caterina, M. J., Leffler, A., Malmberg, A. B., Martin, W. J., Trafton, J., PetersenZeitz, K. R., Koltzenburg, M., Basbaum, A. I., and Julius, D. Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science 288, 306–313, 2000. 93. Davis, J. B., Gray, J., Gunthorpe, M. J., Hatcher, J. P., Davey, P. T., Overend, P., Harries, M. H., Latcham, J., Clapham, C., Atkinson, K., Hughes, S. A., Rance, K., Grau, E., Harper, A. J., Pugh, P. L., Rogers, D. C., Bingham, S., Randall, A., and Sheardown, S. A. Vanilloid receptor-1 is essential for inflammatory thermal hyperalgesia. Nature 405, 183–187, 2000. 94. Mezey, E., Toth, Z. E., Cortright, D. N., Arzubi, M. K., Krause, J. E., Elde, R., Guo, A., Blumberg, P. M., and Szallasi, A. Distribution of mRNA for vanilloid receptor subtype 1 (VR1, and VR1-like inmmunoreactivity, in the central nervous system of the rat and human. Proc. Natl. Acad. Sci. U.S.A. 97, 3655–3660, 2000. 95. Roberts, J. C., Davis, J. B., and Benham, C. D. [3H]-Resiniferatoxin autoradiography in the CNS of wild-type and TRPV1 null mice defines TRPV1 (VR-1) protein distribution. Brain Res. 995, 176–183, 2004. 96. Al-Hayani, A., Wease, K. N., Ross, R. A., Pertwee, R. G., and Davies, S. N. The endogenous cannabinoid anandamide activates vanilloid receptors in the rat hippocampal slice. Neuropharmacology 41, 1000–1005, 2001. 97. Marinelli, S., Di Marzo, V., Berretta, N., Matias, I., Maccarrone, M., Bernardi, G. Mercuri, N. B. Presynaptic facilitation of glutamatergic synapses to dopaminergic neurons of the rat substantia nigra by endogenous stimulation of vanilloid receptors. J. Neurosci. 23, 3136–3144, 2003. 98. Melis, M., Pistis, M., Perra, S., Muntoni, A.L., Pillolla, G., and Gessa, G. L. Endocannabinoids mediate presynaptic inhibition of glutamatergic transmission in rat ventral tegmental area dopamine neurons through activation of CB1 receptors. J. Neurosci. 24, 53–62, 2004. 99. Li, D. P., Chen, S. R., and Pan, H. L. VR1 Receptor Activation Induces Glutamate Release and Postsynaptic Firing in the Paraventricular Nucleus. J. Neurophysiol. 92, 1807–1816, 2004. 100. Ross, R. A., Gibson, T. M., Brockie, H. C., Leslie, M., Pashmi, G., Craib, S. J., Di Marzo, V., and Pertwee, R. G. Structure-activity relationship for the endogenous cannabinoid, anandamide, and certain of its analogues at vanilloid receptors in transfected cells and vas deferens. Br. J. Pharmacol. 132, 631–640, 2001. 101. De Petrocellis, L., Davis, J. B., and Di Marzo, V. Palmitoylethanolamide enhances anandamide stimulation of human vanilloid VR1 receptors. FEBS Lett. 506, 253–256, 2001. 102. Smart, D., Jerman, J. C., Gunthorpe, M. J., Brough, S. J., Ranson, J., Cairns, W., Hayes, P. D., Randall, A. D., and Davis, J. B. Characterisation using FLIPR of human vanilloid VR1 receptor pharmacology. Eur. J. Pharmacol. 417, 51–58, 2001. 103. Hwang, S. W., Cho, H., Kwak, J., Lee, S. Y., Kang, C. J., Jung, J., Cho, S., Min, K. H., Suh, Y. G., Kim, D., and Oh, U. Direct activation of capsaicin receptors by products of lipoxygenases: endogenous capsaicin-like substances. Proc. Natl. Acad. Sci. U.S.A. 97, 6155–6160, 2000. 104. Roberts, L. A., Christie, M. J., and Connor, M. Anandamide is a partial agonist at native vanilloid receptors in acutely isolated mouse trigeminal sensory neurons. Br. J. Pharmacol. 137, 421–428, 2002. 105. Di Marzo, V., Bisogno, T., and De Petrocellis, L. Anandamide: some like it hot. Trends Pharmacol. Sci. 22, 346–349, 2001.

Functional Plasticity of Lipid Mediators: The Example of Endocannabinoids

77

106. Ross, R.A. Anandamide and vanilloid TRPV1 receptors. Br. J. Pharmacol. 140, 790–801, 2003. 107. van Der Stelt, M., and Di Marzo, V. Endovanilloids. Putative endogenous ligands of transient receptor potential vanilloid 1 channels. Eur. J. Biochem. 271, 1827–1834, 2004. 108. Gonzalez, S., Cascio, M. G., Fernandez-Ruiz, J., Fezza, F., Di Marzo, V., and Ramos, J. A. Changes in endocannabinoid contents in the brain of rats chronically exposed to nicotine, ethanol or cocaine. Brain Res. 954, 73–81, 2002. 109. Christopoulos, A., and Wilson, K. Interaction of anandamide with the M(1) and M(4) muscarinic acetylcholine receptors. Brain Res. 915, 70–78, 2001. 110. Fukudome, Y., Ohno-Shosaku, T., Matsui, M., Omori, Y., Fukaya, M., Tsubokawa, H., Taketo, M. M., Watanabe, M., Manabe, T., and Kano, M. Two distinct classes of muscarinic action on hippocampal inhibitory synapses: M2-mediated direct suppression and M1/M3-mediated indirect suppression through endocannabinoid signalling. Eur. J. Neurosci. 19, 2682–2692, 2004. 111. Wan, X., Desilets, M., Soboloff, J., Morris, C., and Tsang, B. K. Muscarinic activation inhibits T-type Ca2+ current in hen granulosa cells. Endocrinology 137, 2514–2521, 1996. 112. Pemberton, K. E., Hill-Eubanks, L. J., and Jones, S. V. Modulation of low-threshold T-type calcium channels by the five muscarinic receptor subtypes in NIH 3T3 cells. Pflugers Arch. 440, 452–461, 2000. 113. De Petrocellis, L., Orlando, P., Di Marzo, V. Anandamide, an endogenous cannabinomimetic substance, modulates rat brain protein kinase C in vitro. Biochem. Mol. Biol. Int. 36, 1127–1133, 1995. 114. Kiss, Z. Anandamide stimulates phospholipase D activity in PC12 cells but not in NIH 3T3 fibroblasts. FEBS Lett. 447, 209–212, 1999. 115. Lambert, D. M., and Di Marzo, V. The palmitoylethanolamide and oleamide enigmas: are these two fatty acid amides cannabimimetic? Curr. Med. Chem. 6, 757–773, 1999. 116. Maccarrone, M., Pauselli, R., Di Rienzo, M., and Finazzi-Agro, A. Binding, degradation and apoptotic activity of stearoylethanolamide in rat C6 glioma cells. Biochem. J. 366, 137–144, 2002. 117. Fu, J., Gaetani, S., Oveisi, F., Lo Verme, J., Serrano, A., Rodriguez De Fonseca, F., Rosengarth, A., Luecke, H., Di Giacomo, B., Tarzia, G., Piomelli, D. Oleylethanolamide regulates feeding and body weight through activation of the nuclear receptor PPAR-alpha. Nature 425, 90–93, 2003. 118. Di Marzo, V., Goparaju, S. K., Wang, L., Liu, J., Batkai, S., Jarai, Z., Fezza, F., Miura, G. I., Palmiter, R. D., Sugiura, T., and Kunos, G. Leptin-regulated endocannabinoids are involved in maintaining food intake. Nature 410, 822–825, 2001. 119. Maccarrone, M., Lorenzon, T., Bari, M., Melino, G., and Finazzi-Agro, A. Anandamide induces apoptosis in human cells via vanilloid receptors. Evidence for a protective role of cannabinoid receptors. J. Biol. Chem. 275, 31938–31945, 2000. 120. Di Marzo, V., Melck, D., Orlando, P., Bisogno, T., Zagoory, O., Bifulco, M., Vogel, Z., and De Petrocellis, L. Palmitoylethanolamide inhibits the expression of fatty acid amide hydrolase and enhances the anti-proliferative effect of anandamide in human breast cancer cells. Biochem. J. 358, 249–255, 2001. 121. Smart, D., Jonsson, K.O., Vandevoorde, S., Lambert, D. M., and Fowler, C. J. ‘Entourage’ effects of N-acyl ethanolamines at human vanilloid receptors. Comparison of effects upon anandamide-induced vanilloid receptor activation and upon anandamide metabolism. Br. J. Pharmacol. 136, 452–458, 2002.

78

Functional Lipidomics

122. Ahern, G. P. Activation of TRPV1 by the satiety factor oleoylethanolamide. J. Biol. Chem. 278, 30429–30434, 2003. 123. Chu, C. J., Huang, S. M., De Petrocellis, L., Bisogno, T., Ewing, S. A., Miller, J. D., Zipkin, R. E., Daddario, N., Appendino, G., Di Marzo, V., and Walker, J.M. Noleoyldopamine, a novel endogenous capsaicin-like lipid that produces hyperalgesia. J. Biol. Chem. 278, 13633–13639, 2003. 124. De Petrocellis, L., Chu, C., Schiano Moriello, A., Walker, M. J., Di Marzo, V. Actions of two naturally occurring saturated N-acyldopamines on Transient Receptor Potential Vanilloid 1 (TRPV1) channels. Br. J. Pharmacol. 143, 251–256, 2004. 125. Panikashvili, D., Simeonidou, C., Ben-Shabat, S., Hanus, L., and Breuer, A., Mechoulam, R., Shohami, E. An endogenous cannabinoid (2-AG) is neuroprotective after brain injury. Nature 413, 527–531, 2001. 126. Mechoulam, R., Fride, E., Hanus, L., Sheskin, T., Bisogno, T., Di Marzo, V., Bayewitch, M., and Vogel, Z. Anandamide may mediate sleep induction. Nature 389, 25–26, 1997. 127. Huang, S. M., Bisogno, T., Petros, T. J., Chang, S. Y., Zavitsanos, P. A., Zipkin, R. E., Sivakumar, R., Coop, A., Maeda, D. Y., De Petrocellis, L., Burstein, S., Di Marzo, V., and Walker, J.M. Identification of a new class of molecules, the arachidonyl amino acids, and characterization of one member that inhibits pain. J. Biol. Chem. 276, 42639–42644, 2001. 128. Cascio, M. G., Minassi, A., Ligresti, A., Appendino, G., Burstein, S., and Di Marzo, V.A structure-activity relationship study on N-arachidonoyl-amino acids as possible endogenous inhibitors of fatty acid amide hydrolase. Biochem. Biophys. Res. Commun. 314, 192–196, 2004. 129. Balsinde, J., Balboa, M. A., Li, W. H., Llopis, J., and Dennis, E.A. Cellular regulation of cytosolic group IV phospholipase A2 by phosphatidylinositol bisphosphate levels. J. Immunol. 164, 5398–5402, 2000. 130. Prescott, E. D. and Julius, D. A modular PIP2 binding site as a determinant of capsaicin receptor sensitivity. Science 300, 1284–1288, 2003.

4

Eicosanoid Lipidomics Rebecca C. Bowers-Gentry, Raymond A. Deems, Richard Harkewicz, and Edward A. Dennis

CONTENTS Introduction..............................................................................................................79 Eicosanoids ..............................................................................................................80 Pathways of Arachidonic Acid Metabolism............................................................82 Cyclooxygenase Products ..............................................................................82 Lipoxygenase Products ..................................................................................84 Other Oxidative Metabolites..........................................................................88 Pathways of Eicosanoid Metabolism ......................................................................90 ω–Oxidation ...................................................................................................90 β–Oxidation....................................................................................................90 Conjugation ....................................................................................................91 Detection and Quantitation......................................................................................92 Conclusion ...............................................................................................................92 Acknowledgment .....................................................................................................93 Reference .................................................................................................................94

INTRODUCTION Lipid mediators are a class of hormone-like compounds that regulate biological processes in paracrine and autocrine fashions. These compounds typically act through transmembrane receptors and initiate signaling mechanisms that have important biological consequences. Compounds that are included in this group of molecules are the eicosanoids, platelet-activating factor (PAF), lysophospholipids, diacylglycerides, ceramide, sphingosine-1-phosphate, anandamide, and other newly discovered bioactive autocoids. The sites of synthesis and action of these compounds are often different. Thus, they act as mediators carrying information from the cell, releasing them to other cells where they interact with specific receptors, thereby initiating a response in the target cells. The function of this diverse group of lipid molecules in vivo is quite complex, but it is clear that they act to modulate numerous biological functions. The complex network formed by lipid mediators is an example of the “systems biology” issues that are being addressed by the LIPID MAPS initiative. Lipid mediators are a set of numerous diverse, yet, in many cases, structurally similar molecules that can affect a large number of physiological functions. A given 79

80

Functional Lipidomics

molecule can affect a number of pathways and, vice versa, a given pathway can affect a number of lipid mediators. To understand the networks that are formed requires the quantitation of all components of the system at a given point in time as well as the tracking of changes in each component as the system is perturbed. To date, most research efforts have focused on a single component or class of mediator. In LIPID MAPS, we attempt to develop the tools to quantitate all of these components. Our efforts to date on eicosanoids are discussed briefly in the following sections.

EICOSANOIDS Eicosanoids are a subset of lipid mediators that are specifically derived from arachidonic acid. Eicosanoids include prostaglandins, leukotrienes, other lipoxygenase products, and the cytochrome P450 metabolites of arachidonic acid as illustrated in Figure 4.1. These cellular messengers mediate various biologically relevant processes that are critical for proper physiological function in tissues. Arachidonic acid is stored by the cell in the sn-2 ester of glycerophospholipids and triglycerides [1]. The cell tightly regulates the quantity of free arachidonic acid through the action of acyl CoA synthases, acyl CoA transferases, and phospholipid remodeling [2]. Typically, there is little free arachidonic acid in the cytosol of resting cells. Thus, there is normally very little arachidonic acid available for the enzymes that catalyze its oxygenation; these enzymes, in general, can only act on free arachidonic acid. Phospholipid membranes Phospholipase A2 COOH COX-1 COX-2

Arachidonic acid

Prostaglandins Prostacyclin Thromboxane 5-LO

Leukotrienes (LTB4, LTC4, LTD4, LTE4) 5-HETE and 5-oxo-ETE

5-LO 12-LO 15-LO

Non-enzymatic CYP450

Lipoxins 12-HETE 15-HETE

Epoxyeicosatetraenoic acids Hydroxyeicosatetraenoic acids Isoprostanes

FIGURE 4.1 The diverging pathways of eicosanoid formation from arachidonic acid present in cellular lipid membranes, which include the lipoxygenase, cycloxygenase, cytochrome P450, and nonenzymatic pathways.

Eicosanoid Lipidomics

81

Thus, the first step in the synthesis of eicosanoids is the release of arachidonic acid from its lipid storage sites. Arachidonic acid is liberated from phospholipids by various phospholipases. The most direct route is via phospholipase A2s (PLA2), which cleaves fatty acids directly from the sn-2 position of phospholipids. PLA2s constitute a superfamily of enzymes that contain 14 groups. These groups are gathered into three general classes: the cytosolic Ca2+-dependent (cPLA2), the Ca2+independent (iPLA2), and the secretory (sPLA2) enzymes. The members of this superfamily vary in molecular weight from the 13–19 kDa sPLA2s to the 85 kDa cPLA2s and iPLA2s. Neither the sPLA2s nor the iPLA2s are selective for arachidonate-containing phospholipids, whereas the cPLA2 preferentially hydrolyzes arachidonate-containing phospholipids [3,4]. This fact led investigators to identify cPLA2 as the most likely candidate for being the key enzyme responsible for agonist-induced arachidonic acid release and, thus, for regulating the subsequent production of eicosanoids. A significant body of data has confirmed that cPLA2 action is required for eicosanoid synthesis in most cells and tissues. Because of its potential importance in eicosanoid production, the characterization of cPLA2 has been aggressively researched. cPLA2 is ubiquitously expressed in cells and tissues [5]; it is regulated by the binding of Ca2+ to the C2 domain, which enables cPLA2 to translocate to the intracellular membranes to initiate arachidonic acid release [4]. Additionally, cPLA2 can bind to phosphatidylinositol-4,5-bisphosphate (PIP2) with high affinity that also leads to the translocation of the enzyme in cellular membranes [6,7]. PIP2 acts in parallel to Ca2+ and can sequester the enzyme to membranes even in the absence of Ca2+. Also, cPLA2 phosphorylation by p38 mitogen-activated protein kinase on a number of serine residues, specifically serine-505 and serine-727, is important for activation of this enzyme, though the mechanism of activation is not clear. All of these regulatory mechanisms are important for the membrane translocation and activation of cPLA2, and are presumably important for controlling arachidonic acid release and thus eicosanoid production [4]. Though cPLA2 appears to be the main enzyme that regulates the amount of free arachidonic acid, other phospholipases are also present in the cell and can release arachidonic acid [8]. The interaction of these phospholipases, including sPLA2 (specifically Group V) and cPLA2, continues to be an area of active investigation. A clear picture of the role of these and other auxiliary enzymes, such as transacylases [9], is only beginning to emerge. Nonetheless, these enzymes remain interesting as therapeutic targets for controlling eicosanoid biosynthesis within specific cells. Once arachidonic acid is released by PLA2, it can enter a number of divergent enzymatic pathways that lead to the production of a wide variety of eicosanoids. The enzymes that metabolize arachidonic acid include the lipoxygenases, cyclooxygenases, and cytochrome P450s. In nonenzymatic oxidative reactions, isoprostanes are formed in addition to racemic mixtures of the hydroxyeicosatetraenoic (HETE) acids.

82

Functional Lipidomics

PATHWAYS OF ARACHIDONIC ACID METABOLISM CYCLOOXYGENASE PRODUCTS The first pathway of oxygenation of arachidonic acid that was studied in detail was the prostaglandin pathway, which is mediated by the enzyme PGH synthase, for which two genes exist. These two genes code for similar yet significantly different proteins, PGH synthase-1 and PGH synthase-2, commonly referred to as the cyclooxygenases, COX-1 and COX-2 [10]. Cyclooxygenase isozymes exhibit both a bis-oxygenase activity, which incorporates two molecules of oxygen into arachidonic acid, and a peroxidase activity that reduces the 15-hydroperoxy group of prostaglandin G2 to PGH2 [11] (Figure 4.2). The ultimate fate of PGH2 is dependent on which enzymes are present to convert the PGH2 into downstream prostaglandins. COX-1 is constitutively expressed in most cells and tissues, leading to the production of prostaglandin H2 (PGH2). COX-2 is induced by inflammatory stimuli and cytokines in monocytes, macrophages, osteoblasts, and endothelial cells. Cyclooxygenase isozymes are the target of nonsteroidal antiinflammatory drugs (NSAIDs), which indicates that the products formed by this enzyme can induce inflammatory processes [12]. PGH2 can form a number of biosynthetically and biologically diverse molecules. This group includes, but is not limited to, prostaglandin I2 (PGI2, also called prosCOOH Arachidonic acid O

COOH

Cyclooxygenase

HO COOH

O

COOH

O O

OH

HO OOH PGG2

PGI2 Prostacyclin synthase

Cyclooxygenase

OH PGH2

Prostaglandin E synthase HO

HO

COOH HO

OH PGE2

Prostaglandin D synthase

Thromboxane synthase

COOH HO

Prostaglandin F synthase

COOH

O O

O

OH PGF2α

O OH TXB2

COOH O

OH PGD2

FIGURE 4.2 The cyclooxygenase pathway converts arachidonic acid into various prostaglandins and thromboxane B2, a stable metabolite of thromboxane A2.

Eicosanoid Lipidomics

83

tacyclin), thromboxane A2, prostaglandin E2 (PGE2), prostaglandin D2 (PGD2), and prostaglandin F2α (PGF2α). The final levels of these compounds in cells and tissues are determined by the levels of expression of the respective synthases [13]. Prostacyclin was first identified in 1976 as a vasodilative and antiaggregating mediator extracted from the endothelium [14]. This molecule is formed by converting PGH2 to PGI2 by prostacyclin synthase in an acid-catalyzed heterolytic cleavage of the endoperoxide, followed by reaction of the transiently positive oxygen at carbon 6 [11]. This molecule is unstable and is quickly metabolized to 6-keto-prostaglandin 1α (6-keto-PGF1α), a stable yet biologically inactive metabolite. Prostacyclin acts on a specific G-protein-coupled receptor, termed the IP receptor, and increases plasma cAMP. Prostacyclin is also a peroxisome proliferators-activated receptor (PPAR) agonist, which can induce a variety of physiological outcomes [15]. The synthesis of prostacyclin is reduced in hypertensive disease states. Synthetic stable prostacyclin analogs, such as iloprost, have been used in the treatment of pulmonary hypertension [16]. Thromboxane A2 (TxA2) is the physiological antagonist of prostacyclin; it causes vasoconstriction and platelet aggregation. TxA2 has been shown to be produced by platelets [17] and macrophages, and is present in the lung [18]. The conversion of PGH2 to TxA2 is catalyzed by thromboxane A synthase, and this conversion is mediated by a heme-dependent opening of the endoperoxide of PGH2 [19]. Like prostacyclin, TxA2 is unstable and is converted to thromboxane B2 (TxB2), a stable inactive metabolite [11]. The receptor for TxA2 (TX receptor) belongs to the G-protein-coupled seventransmembrane receptor family and activates protein kinase C (PKC) and also induces intracellular calcium flux, causing vasoconstriction [20]. The stable urinary prostacyclin and TxA2 metabolites, 2,3-dinor-6-keto-PG1α and 11-dihydro-TxB2, respectively, can be quantitated, and changes in the ratio between the two metabolites have been used as indicators in various diseases, including pulmonary hypertension [16], preeclampsia [21], diabetes [22], and renal diseases [23]. PGE2 is a potently biologically active molecule with many different activities. PGE2 is formed by isomerization of PGH2 by PGE2 synthase, which utilizes glutathione as an essential cofactor [24], and this synthase is recognized to be a member of the protein superfamily consisting of membrane-associated proteins involved in eicosanoid and glutathione metabolism (the MAPEG family) [25]. To accomplish its many physiological functions, PGE2 interacts with a number of G-protein-coupled receptors, termed EP1, EP2, EP3, and EP4, which are products of separate genes [26]. Activation of the EP1 receptor by PGE2 induces calcium mobilization, and in EP1, knockout studies have been implicated in colon carcinogenesis [27]. The EP2 receptor stimulation causes release of cAMP, and knockout studies connect signaling in this pathway to the PGE2 inflammatory response, cancer, bronchodilation, and the mechanism of parturition [26]. PGE2 has been shown to mediate pain responses including allodynia, and knockout studies suggest that this is through PGE2 interaction with the EP3 receptor [27]. The EP4 receptor activation by PGE2 has been investigated for its role in inflammation, rheumatoid arthritis, cancer, and bone formation [26]. Sites of action of PGE2 indicate the type of response that will be mediated, which is determined by where PGE2 is synthesized and where the receptors are located.

84

Functional Lipidomics

PGD2 is a major product present in the central nervous system and is also produced by mast cells. Prostaglandin D synthase catalyzes the isomerization of endoperoxide of PGH2 to the 9-hydroxy and 11-keto groups, directly opposite of the case in PGE2 that has 9-keto and 11-hydroxy groups formed from the endoperoxide ring opening. There are two types of PGD synthases, the lipocalin- and hematopoietic-types, which are very different enzymes that function in the same capacity, to form PGD2 [28]. The lipocalin-type PGD synthase is localized to the central nervous system, whereas the hematopoietic-type enzyme is widely expressed in peripheral tissues, including mast cells and lymphocytes [28]. The receptor for PGD2 (DP) is not very well characterized, but it has been shown that PGD2 increases cAMP generation. The physiological effects of PGD2 include mediation of vasodilation, platelet aggregation, moderation of the sleep–wake cycle, and body temperature [29]. Prostaglandin F2α (PGF2α) can be produced from PGH2 through a reduction mechanism by prostaglandin F synthase that forms hydroxyl groups at the 9 and 11 positions. In addition, PGF2α can be formed from PGE2 by PGE 9-ketoreductase and from PGD2 by PGD 11-ketoreductase [30]. The G-protein-coupled receptor for PGF2α (FP receptor) has been identified in many species as well as various cells types. FP receptor stimulation leads to phospholipase-C-mediated activation of PKC [31]. FP receptor expression as well as production of PGF2α are critical for parturition and leuteolysis [27]. As demonstrated, prostaglandins, perform diverse biological functions. These molecules can act in an autocrine fashion by agonist activation of PPARs causing changes in gene expression as exemplified by prostacyclin. They can also function in a paracrine fashion by exiting the cell through prostaglandin transporters and acting on G-protein-coupled receptors that instigate intracellular signaling pathways. The investigations into the in vivo physiological and pathological roles of prostaglandins as well as their biosynthetic enzymes and receptors continue to be stimulating and active areas of research.

LIPOXYGENASE PRODUCTS The three known lipoxygenase pathways that act on arachidonic acid, producing biologically active compounds, are the 5-, 12-, and 15-lipoxygenase pathways shown in Figure 4.3 and Figure 4.4. The 5-lipoxygenase (5-LO) pathway leads to a diverse group of compounds that are potentially biologically active, including 5-HETE, 5oxo-ETE, and the leukotrienes. Arachidonate metabolism through the 5-LO pathway was identified in 1979 when the first molecule in this pathway was characterized, from the slow-reacting substance of anaphylaxis, as leukotriene C4 (LTC4) [32]. The enzyme 5-LO, responsible for insertion of molecular oxygen at the 5 position, is the first critical protein involved in the oxidation of arachidonic acid, which ultimately results in the formation of leukotrienes [33]. An auxiliary protein, termed 5lipoxygenase-activating protein (FLAP), which is similar in primary sequence to LTC4 synthase [34,35], was found to be necessary for leukotriene biosynthesis [36]. FLAP is thought to function as an arachidonic acid transfer protein as found through competition binding experiments [37,38,39]. Furthermore, FLAP is localized in or

Eicosanoid Lipidomics

85

COOH

Arachidonic acid 5-Lipoxygenase OH

OOH COOH

Cellular reductases

COOH O

5-LO

COOH LTA4

5-HPETE LTA4 hydrolase

5S-HETE NADP+ -Depend. Dehydrogenase

OH

OH COOH

O

LTC4 synthase

OH

LTB4

COOH

LTC 4

γ -Glutamyl transpeptidase

S

S LTD4

NHCO(CH2)2CHCOOH NH2

OH

OH

COOH COOH Aminopeptidase NH2

COOH CONHCH2COOH

S

5-oxo-ETE

LTE4

+GSH

COOH CONHCH2COOH

NH2

FIGURE 4.3 5-LO metabolizes arachidonic acid through enzyme catalysis to a myriad of molecules including 5-HETE and 5-oxo-ETE, as well as the leukotrienes (LTB4, LTC4, LTD4, and LTE4).

near the nuclear envelope, which is consistent with earlier studies that demonstrated translocation of 5-LO to the nucleus during activation [40], where it likely encountered FLAP. The first stable product of 5-LO is 5-HPETE (Figure 4.3). 5-HPETE can be converted to 5(S)-HETE by cellular reductases [41]. Another possible metabolic fate for 5-HPETE is its direct conversion to 5-oxo-ETE, a potent eosinophilic chemotactic agent [42,44]. 5-HETE has also been converted into 5-oxo-ETE by an NADP+dependent dehydrogenase [43]. The receptor for 5-oxo-ETE [45] is shown to be distinct from the other 5-LO product receptors [46]. Because the expression of this receptor is specific to neutrophils, eosinophils, and lung macrophages, and has only been recently identified [47], many studies need to be done on the physiological role of 5-oxo-ETE and its receptor in vivo. A second pathway for 5-HPETE is a subsequent catalysis by 5-LO that converts 5-HPETE to LTA4 [40], an unstable intermediate with a half-life of less than 3 sec [48]. LTA4 hydrolase carries out the stereospecific hydrolysis of the conjugated triene epoxide, LTA4, to yield 5(S),12(R)-dihydroxy-6,14-cis-8,10-trans-eicosatetraenoic acid, LTB4. LTB4 is a potent chemotactic agent for neutrophils [49], which acts through specific G-protein-coupled receptors (BLT1 and BLT2) to elicit a cellular response. Both of these receptors have been cloned and expressed for pharmacological studies of LTB4 [50,51]. The conjugation of LTA4 with glutathione results in the formation of the myotropic leukotriene LTC4, which is the initial compound of the cysteinyl leukotrienes

86

Functional Lipidomics

COOH

Arachidonic acid

12-Lipoxygenase

15-Lipoxygenase

COOH

COOH

OOH

OOH 12S-HPETE

15S-HPETE

COOH

COOH

OH

OH 12S-HETE

15S-HETE

FIGURE 4.4 Alternative lipoxygenase pathways to 5-LO include 12- and 15-lipoxygenase, which form 12- and 15-HPETE, which is further converted to 12- and 15-HETE, respectively.

[32] (Figure 4.3). This enzymatic step is catalyzed by a unique glutathione-Stransferase (GST), termed LTC4 synthase, and requires millimolar levels of glutathione [34]. The resulting LTA4 glutathione adduct has been shown to be a potent agent capable of inducing smooth muscle contraction and has been implicated in the pathology of asthma [52]. Molecular recognition of the cysteinyl leukotrienes is mediated by at least two G-protein-linked-membrane-associated receptors [53], termed cysLT1 and cysLT2, which that have the highest affinity for cysteinyl leukotriene formed by the action of -glutamyl-transpeptidase on LTC4. This product, leukotriene D4, and the subsequent peptide hydrolysis product, leukotriene E4, complete the family of cysteinyl leukotrienes and are all potent activators of the cysLT receptors [52,54]. The transport of these unique peptidolipids has been studied by Keppler and coworkers [55], who have described involvement of the multidrug resistance receptor in this important process of export of LTC4 from the synthetic cell. The 12- and 15-lipoxygenase pathways are similar to 5-LO in that they can insert molecular oxygen stereospecifically into arachidonic acid; however, the end products differ from the 5-LO products. 12-lipoxygenase action on arachidonic acid results in the production of 12(S)-hydroperoxyeicosatetraenoic acid (12(S)-HPETE), which is subsequently reduced to 12(S)-HETE acid (see Figure 4.4). 12-lipoxygenase is found primarily in platelets and the epidermis [56,57], referred to as the platelettype, and in macrophages and leukocytes [56], called the leukocyte-type. 15-lipoxy-

Eicosanoid Lipidomics

87

genase catalyzes the conversion of arachidonic acid to 15(S)-HPETE acid with subsequent reduction to 15(S)-hydroxyeicosatetraenoic acid (15(S)-HETE). 15HETE may play a role in many pathological conditions, including oral cancer [57], asthmatic bronchitis [59], and atherogenesis [60]. Similar to many enzymes involved in the production of prostaglandins, 15-lipoxygenase has been reported to be suicide inactivated by covalent modification of the enzyme by 15(S)-HPETE [61]. Both 12and 15-lipoxygenase and their products have been implicated in atherogenesis [62] and cancer [63]. One subset of the lipoxygenase products, called lipoxins, are formed by the action of multiple lipoxygenases (Figure 4.5). Lipoxins are trihydroxylated arachidonic acid derivates that result from the action of 5-LO on arachidonic acid with subsequent action by 12- and 15-lipoxygenase. The lipoxins, which include lipoxin A4 and lipoxin B4, promote vasodilation and also counteract the immunomodulatory actions of most other eicosanoids [64]. An alternative biosynthetic pathway suggests that acetylation of prostaglandin H synthase changes its catalytic activity in the conversion of arachidonic acid to 15(R)-HPETE [65]. Then 15(R)-HPETE is acted on by 5-LO to form the 15-epi-lipoxins, which possess biological activities similar to lipoxins A4 and B4. A lipoxin-A4-specific receptor has been identified and cloned COOH

COX-2

5-LO

COOH

O

COOH

Arachidonic acid 15R-HETE

OH

15-LO

LTA4

5-LO COOH

O

15-LO or 12-LO

COOH 15S-HPETE OOH

OH 5S, 6S, 15R-Epoxytetraene

5-LO O

OH

HO

OH

OH 15-epi-Lipoxin A4

COOH

COOH

COOH

OH 5S, 6S, 15S-Epoxytetraene

HO OH 15-epi-Lipoxin B4 OH

COOH

HO OH Lipoxin B4

HO OH

COOH

OH Lipoxin A4

FIGURE 4.5 The divergent pathways of lipoxin formation to the lipoxin A4 and lipoxin B4 species as well as their respective 15-epi isomers are shown. The formation of lipoxins occurs by concerted lipoxygenase action; the 15-epi isomers are formed with an initial aspirinsensitive COX-2 catalysis followed by subsequent lipoxygenase catalysis.

88

Functional Lipidomics

[66]. In addition, experiments have indicated that the lipoxins can interact with other eicosanoid receptors [67]. Understanding the concerted efforts of these lipoxygenases and, in some cases, that of cyclooxygenase, to form many different lipoxins, and elucidating the biological processes that these molecules regulate in vivo will be a continuously developing field of research.

OTHER OXIDATIVE METABOLITES Though the metabolism of arachidonic acid was initially thought to be limited to the cyclooxygenase and lipoxygenase pathways, other oxidative pathways of metabolism have since emerged as important biochemical pathways. These alternative pathways of arachidonic acid metabolism include oxidation by cytochrome P450 isozymes and free-radical-catalyzed mechanisms as illustrated in Figure 4.6. Oxidation by cytochrome P450 results in the formation of 20-HETE acid as well as some other HETE acids and epoxyeicosatrienoic acids (EETs). Free-radical-derived metabolites of arachidonic acid include a complex family of compounds called isoprostanes. Various isozymes from the cytochrome P450 family metabolize arachidonic acid in the liver, kidney, heart, brain, and peripheral vasculature to 20-HETE and EETs [68]. It is thought that the ω-hydroxylases, also known as the cytochrome P450 4A, 4B, and 4F isoforms, are responsible for the conversion of arachidonic acid to 20HETE [69]. 20-HETE is a potent vasoconstrictor of small arterioles by inhibiting COOH

Arachidonic acid

CYP450

CYP450

Radical oxidation HO

COOH

COOH

OH O 14R, 15S-EET

HO

OH 15-F2t-isoP

COOH OH 20-HETE

FIGURE 4.6 Cytochrome P450 and nonenzymatic metabolites of arachidonic acid include 20-HETE, the epoxyeicosatrienoic acids (illustrated by 14,15-EET), and the isoprostanes (illustrated by 15-F2t-isoP, also known as 8-iso-iPGF2α).

Eicosanoid Lipidomics

89

calcium-activated potassium channels as well as by increasing L-type calcium channel conductance. Inhibition of the formation of 20-HETE has indicated that this mediator plays an important role in hypertension by regulating vascular tone and angiogenesis by promoting revascularization [68]. Also, 20-HETE is an endogenous inhibitor of the sodium/potassium ATPase in the proximal tubule due to phosphorylation by PKC, and it inhibits sodium-potassium-chloride transport in the thick ascending loop of Henle; these roles illustrate the importance of 20-HETE in the regulation of kidney function [70]. Other cytochrome P450 isoforms catalyze the formation of other HETE acids, including 5-HETE, 7-HETE, 8-HETE, 9-HETE, 10-HETE, 11-HETE, 12-HETE, 13-HETE, 16-HETE, and 17-HETE; the role of these arachidonic acid metabolites continues to be investigated [69]. The second set of metabolites formed from arachidonic acid metabolism by cytochrome P450 isozymes are the EETs. The cytochrome P450 1A, 2B, 2C, 2D, 2E, 2J, and 4A isoforms are responsible for the formation of EETs in a variety of tissue types, including brain, kidney, liver, lung, pancreas, and heart [71]. This group of molecules result from olefin epoxidation and include the cis-enantiomeric products 5,6-EET, 8,9-EET, 11,12-EET, and 14,15-EET [72]. EETs can hyperpolarize vascular smooth muscle by increasing the open-state probability potassium channels that cause dilation of coronary arteries. In addition, these arachidonic acid metabolites inhibit sodium transport and hemodynamics in the kidney [71]. Specific EETs also have anti-inflammatory action due to decrease in endothelial cell adhesion molecule expression, inhibit platelet aggregation, and possess mitogenic properties including calcium mobilization and activation of various kinases [72]. 20-HETE and EETs act in autocrine and paracrine fashions to elicit their actions, though receptors for these eicosanoids have yet to be identified. Isoprostanes are a group of molecules derived from arachidonic acid by nonenzymatic free-radical oxygenation. These molecules are formed in situ on phospholipids and once released, probably by PLA2s, circulate in the plasma, in which they can interact with membrane receptors, causing cell activation [73]. Isoprostanes are structurally analogous to the cyclooxygenase-derived prostaglandins and have a parallel naming system, but the stereochemistry of the isoprostanes and prostaglandins are notably different [74]. Because these molecules are formed from free-radical reactions derived from molecular oxygen, the isoprostane group consists of hundreds of molecules, some of which have been used as a quantitative measure of oxidative stress. This has been an important step in the identification of pathophysiological disease states in which free radicals damage tissue biomolecules. Isoprostanes have been used as indices of lipid peroxidation in such inflammatory diseases as atherosclerosis, Alzheimer’s disease, asthma, diabetes, and cancer [75]. Not only can this group of molecules be used an indicators of disease states but also have potent biological activities. Both 8-iso-iPGF2α and 8-iso-iPGE2 have been shown to elicit vasoconstriction in vascular beds as well as proliferation of vascular smooth muscles [74]. Though no specific isoprostane receptors have been identified, isoprostanes can activate thromboxane receptors and PGE2 receptors [73].

90

Functional Lipidomics

PATHWAYS OF EICOSANOID METABOLISM As previously discussed, eicosanoids are a vast group of potent molecules that act on G-protein-coupled receptors, and in some cases PPARs, to elicit various biochemical processes. Their biosynthesis is complicated by intricate steps of enzyme catalysis as well as cell-specific enzyme regulation. Final regulatory mechanisms of these molecules consist of metabolic enzymes that convert the eicosanoids into downstream metabolites. This is the conventional way of inactivation of the biological activity of these molecules; however, the metabolites formed are not always inactive. Eicosanoids can be metabolized through pathways of oxidation and conjugation with various substituents. These metabolic pathways are cell and tissue specific and lead to a myriad of metabolic products.

ω–OXIDATION There are subgroups of the cytochrome P450 isozymes that are responsible for the ω–oxidation of a number of eicosanoids along with the previously described formation of 20-HETE and EETs. Oxidation of eicosanoids by the cytochrome P450 enzymes occurs at the terminal methyl group to form 20-hydroxy and carboxy metabolites, as well as at the penultimate carbons that form the 18-hydroxy and19hydroxy metabolites. The oxidation of these molecules modifies their polarity, which in turn can alter their transport and excretion. However, metabolism by ω–oxidation does not always cause inactivation of the parent eicosanoid. The cytochrome P450 4F family is principally responsible for the ω–oxidation of eicosanoids. PGH2 can be metabolized into 19-hydroxy-PGH2 by the cytochrome P450 4F8 isozyme; this metabolite is unstable and is rapidly converted into 19hydroxy-PGE2 [76]. Lipoxin B4 also undergoes ω–oxidation by cytochrome P450 to 20-hydroxy-lipoxin B4, illustrated by carbon monoxide sensitivity [77]. Another ω–oxidation metabolism pathway is the conversion of LTB4 to 20-hydroxy-LTB4 in rat hepatocytes, with a second oxidation resulting in 20-carboxy-LTB4 [78]; these metabolites retain biological activity [79]. Both 18- and 19-hydroxy metabolites of 5-oxo-ETE have been identified in mouse macrophages and have resulted in the inactivation of this potently active molecule [80]. Thus, the ω–oxidation pathways metabolize a number of different eicosanoids into both biologically active and inactive products.

β–OXIDATION Mitochondrial and peroxisomal β–oxidation processes degrade eicosanoids after the initial ω–oxidation [81]. Acyl-CoA enzymes are responsible for the chain-shortening result of the β–oxidation of various eicosanoids. β–Oxidation, therefore also called chain shortening, allows for energy, in the form of ATP, to be produced and is the process by which carbon recycling occurs. Chain shortening by β–oxidation results in the formation of the parent molecule less 2, 4, and 6 carbons, referred to as dinor, tetranor, and hexanor compounds, respectively. β–oxidation occurs once the methyl terminal end of the eicosanoid is ω–oxidized to a hydroxyl moiety, with a subsequent

Eicosanoid Lipidomics

91

oxidation to a carboxyl group. Typically, β–oxidation inactivates the parent molecule because of the change in length of the molecule, though this is not always the case. Novel metabolites of PGE2 in rat hepatocytes were identified as dinor-PGE2, dinor-PGE1, and tetranor-PGE1 [82], which have also been identified as the human urinary metabolites of PGE2 [83]. Peroxisomal chain shortening of the cyclooxygenase metabolite thromboxane B2, a metabolite of thromboxane A2, and prostacyclin were determined from human urine measurements to be 2,3-dinor-thromboxane B2 and 2,3-dinor-6-keto-prostaglandin F2α, respectively. Similarly, the major human urinary metabolites of LTE4 were 14-carboxy-hexanor-LTE3 and 16-carboxy-Δ13tetranor-LTE4, which also illustrates that double-bond isomerization can occur [84]. One example of β–oxidation that does not result in inactivation is the metabolism of 8-epi-iPGF2α, a potent vasoconstrictor. The isoprostane 8-epi-iPGF2α can be metabolized to 2,3-dinor-5,6-dihydro-8-epi-iPGF2α, which retains the biological activity of the parent molecule [85].

CONJUGATION Metabolic transformation can also occur in eicosanoids by conjugation, which significantly alters its chemical structure, changing the bioavailability of these compounds. Conjugation of eicosanoids occurs with many different molecules and includes, but is not limited to, sulfation, glucuronidation, and glutathione addition. Glucuronide addition to eicosanoids occurs at free hydroxyl groups on the molecule and is catalyzed by UDP-glucuronosyltransferases (UGT). Identification of a variety of glucuronide conjugates of HETEs was done with stable transfected UGT isozymes [86]. Urinary excretion of 20-HETE was identified to be in the form of the glucuronide conjugate [87], and the increase in excretion of 20-HETE-glucuronide was correlated to liver cirrhosis, indicating that it can be extremely important to be able to detect these molecules in body fluids [88]. It has been shown that ω–oxidation and β–oxidation are important metabolic pathways for LTB4 metabolism [79,89]. In addition, LTB4-glucuronide metabolites have been identified in human urine [90]. Eicosanoids that possess α/β unsaturated ketones, conjugated dienones, epoxides, and conjugated epoxides are particularly susceptible to attack by glutathione, which occurs through a Michael addition reaction catalyzed by GST [91]. Two cyclopentenone prostaglandins, which have α/β unsaturated ketones, that undergo glutathione conjugations are prostaglandin A2 and prostaglandin J2; these conjugation reactions are catalyzed by various human GST isozymes [92]. A metabolite of prostaglandin J2, 15-deoxy-Δ12,14 prostaglandin J2, can be conjugated with glutathione, which terminates the PPAR agonist activity of this pathway of molecules. One isoprostane that is known to be conjugated with glutathione is 8-epiiPGA2 and may result in the inactivation of these highly reactive molecules [93]. Glutathione conjugation can also result in eicosanoid metabolites that have not lost their biological activities. LTA4 conjugation with glutathione results in the formation of LTC4 [32], which, as previously described, is a potent myotropic agent. Hepoxilins, or 12-lipoxygenase products of arachidonic acid metabolism, can also be conjugated with glutathione and still retain the ability to hyperpolarize hippocampal neurons [94]. 5-oxo-ETE, a potent eosinophil chemoattractant, can be glu-

92

Functional Lipidomics

tathione-conjugated to form a metabolite named FOG7 (5-oxo-7-glutathionyl8,11,14-eicosatrienoic acid), which is also potently biologically active [95].

DETECTION AND QUANTITATION The discussion in the preceding text clearly demonstrates the complexity of the eicosanoid network. The quantitation of all the components of this system is made even more difficult by the structural similarity among the members of this class of molecules. To date, the techniques used in investigations of this group of molecules for structural and quantitative purposes have included radioactivity detection, the use of stable isotopes, UV-absorbance detection, and separation by gas or liquid chromatography coupled with mass spectrometry for direct analysis of each molecule. Previous studies have focused primarily on one subclass, either the leukotrienes, prostaglandins, or the oxygenated metabolites of the eicosanoids, rather than all the arachidonic acid metabolites at the same time. Some of the studies that investigated products from various subclasses utilized gas chromatography and mass spectrometric techniques that required derivatization of the molecules [96,97], were performed with radioimmunassay [98], were analyzed by high-performance liquid chromatography (HPLC) [99,100] or radioactivity [101,102], or limited the molecules that were identified, for instance, to PGE2 and HETEs [103–105]. Previous research in this field has been hypothesis driven, in which the investigation centered on asking a question and pursuing the answer. The purpose of the research currently underway and planned for the future as part of the LIPID MAPS initiative is to use all the techniques necessary, including stable-isotope dilution, electrospray ionization mass spectrometry, radioactive labeling, and UV absorbance to identify and quantitate all of the eicosanoids secreted from an activated macrophage in a profiling, undirected manner that will allow the identification of networks via the new data mining techniques being developed in the field of bioinformatics. To this end, we have developed a liquid chromatography–mass spectrometry (LC–MS) technique that allows a varied mixture of different eicosanoid compounds to be separated and resolved in a single 16-min analysis. The technique involves online reverse-phase C18 LC coupled to triple-quadrupole tandem mass spectrometry. The mass spectrometer employs multiple-reaction ion-monitoring (MRM) detection, in which a unique parent or fragment ion pair is used to identify a specific compound. Unique LC retention times also aid in identifying different compounds. By combining the LC separation with MRM detection, we are able to resolve the very similar isomeric HETE compounds. Figure 4.7 shows a representative chromatogram of a mixture of 11 eicosanoid compounds (a set of commercial standards containing representatives of all three classes of eicosanoids, i.e., prostaglandins, leukotrienes, and HETEs) from a single LC–MS/MS analysis.

CONCLUSION The eicosanoid biosynthetic and metabolic pathways described herein are clearly complex. Concerted efforts of biosynthetic and metabolic enzymes regulate the

4

6

8 10 Time, min

5(s)HETE (319/115)

8(s)HETE (319/155) 12(s)HETE (319/179)

AA(303/259)

25%

2

11(s)HETE (319/167)

10. 8 11. 0 11. 2 11. 4 11. 6 11. 8

LTB4 (335/203)

50%

HETEs 15(s)HETE (319/167)

PGF2a (353/193)

Rel. intensity

75%

PGD2(351/271)

TXB2 (369/207)

100%

93

PGE2(351/271)

Eicosanoid Lipidomics

12

14

16

FIGURE 4.7 Liquid chromatography multiple-reaction ion-monitoring (MRM) mass spectrum obtained from a mixture of 11 eicosanoid standards. The MRM parent/daughter ion pairs used for each standard are shown. Note the ability to resolve the different classes of eicosanoids in a single 16-min LC analysis; particularly, the ability to resolve the five HETE isomers (the inset displays this area of the spectrum in more detail).

production and removal of these potently active molecules. Cell and tissue specificity is key to understanding the metabolic fate of these molecules as well as the biosynthetic pathways. Regulation of enzymes by cellular location and signaling mechanisms affect the production of the eicosanoids and constitutes an intense area of current research effort. Working toward an understanding of the pathophysiological relevance of this class of molecules that includes biosynthesis, transport, and metabolism is of paramount importance in the identification of the roles that eicosanoids play in disease states. Many questions have been answered, and there are many more to be resolved in the future.

ACKNOWLEDGMENT We wish to thank Dr. Jean Chin, Program Director, National Institutes of General Medical Sciences, for advice and the NIH for support of the LIPID MAPS initiative (Glue Grant) Number GM069338.

94

Functional Lipidomics

Reference 1. Irvine, R.F. How is the level of free arachidonic acid controlled in mammalian cells? Biochem. J. 1982, 204, 3–16. 2. Chilton, F.H. and Murphy, R.C. Remodeling of arachidonate-containing phosphoglycerides within the human neutrophil. J. Biol. Chem. 1986, 261, 7771–7777. 3. Balsinde, J., Winstead, M.V., and Dennis, E.A. Phospholipase A2 regulation of arachidonic acid mobilization. FEBS Lett. 2002, 531, 2–6. 4. Gijon, M.A. and Leslie, C.C. Regulation of arachidonic acid release and cytosolic phospholipase A2 activation. J. Leukoc. Biol. 1999, 65(3), 330–336. 5. Leslie, C.C. Properties and regulation of cytosolic phospholipase A2. J. Biol. Chem. 1997, 272(27), 16709–16712. 6. Mosior, M., Six, D.A., and Dennis, E.A. Group IV cytosolic phospholipase A2 binds with affinity and specificity to phosphatidyl 4,5-bisphosphate resulting in dramatic increases in activity. J. Biol. Chem. 1998, 273(4), 2184–2191. 7. Six, D.A. and Dennis, E.A. Essential Ca(2+)-independent role of the group IVA cytosolic phospholipase (2) C2 domain for interfacial activity. J. Biol. Chem. 2003, 278(26), 23842–23850. 8. Tougui, L. and Alaoui-El-Axher, M. Mammalian secreted phospholipases A2 and their pathophysiological in inflammatory diseases. Curr. Mol. Med. 2001, 1, 739–754. 9. Surette, M.E. and Chilton, F.H. The distribution and metabolism of arachidonatecontaining phospholipids in cellular nuclei. Biochem. J. 1998, 330, 915–921. 10. Smith, W.L. and Song, I. The enzymology of prostaglandin endoperoxide H synthases-1 and -2. Prostaglandins Other Lipid Mediat. 2002, 68–69, 115–128. 11. Smith, W.L. and Marnett, L.J. Prostaglandin endoperoxide synthase: structure and catalysis. Biochim. Biophys. Acta 1991, 1083, 1–17. 12. Bertolini, A., Ottani, A., and Sandrini, M. Dual acting anti-inflammatory drugs: a reappraisal. Pharmacol. Res. 2001, 44(6), 437–450. 13. Sales, K.J. and Jabbour, H.N. Cyclooxygenase enzymes and prostaglandins in reproductive tract physiology and pathology. Prostaglandins Other Lipid Mediat. 2003, 71, 97–117. 14. Whittaker, N., Bunting, S., Salmon, J., Moncada, S., Vane, J.R., Johnson, R.A., Morton, D.R., Ki, J.H., Gorman, R.R., McGuire, J.C., and Sun, F.F. The chemical structure of prostaglandin X (prostacyclin). Prostaglandins 1976, 12(6), 915–928. 15. Lim, H. and Dey, S.K. Minireview: a novel pathway of prostacyclin signaling — hanging out with nuclear receptors. Endocrinology 2002, 143, 3207–3210. 16. Olschewski, H., Rose, F., Schermuly, R., Ghofrani, H. A., Enke, B., Olschewski, A., and Seeger, W. Prostacyclin and its analogues in the treatment of pulmonary hypertension. Pharmacol. Ther. 2004, 102, 139–153. 17. Hamberg, M., Svensson, J., and Samuelsson, B. Thromboxanes: a new group of biologically active compounds from prostaglandin endoperoxides. Proc. Natl. Acad. Sci. USA 1975, 72(8), 2994–2998. 18. Pace-Asciak, C.R., Granstrom, E., and Samuelsson, B. Arachidonic acid epoxides. Isolation and structure of two hydroxy epoxide intermediate in the formation of 8,11,12- and 10,11,12-trihydroxyeicosatrienoic acids. J. Biol. Chem. 1983, 258, 6835–6840. 19. Wang, L.-H. and Kulmacz, R. J. Thromboxane synthase: structure and function of protein and gene. Prostaglandins Other Lipid Mediat. 2002, 68–69, 409–422.

Eicosanoid Lipidomics

95

20. Kinsella, B.T., O’Mahony, D.J., and Fitzgerald, G.A. The human thromboxane A2 receptor a isoform (TPα) functionally couples to the G proteins Gq in vivo and is activated by the isoprostane 8-epi prostaglandin F2α. J. Pharmacol. Exp. Ther. 1997, 281(2), 957–964. 21. Mills, J.L., DerSimonian, R., Raymond, E., Morrow, J. D., Roberts, L.J., II, Clemens, J.D., Hauth, J.C., Catalano, P., Sibai, B., Curet, L. B., and Levine, R. Prostacyclin and thromboxane changes predating clinical onset of preeclampsia. J. Am. Med. Assoc. 1999, 281(4), 356–362. 22. Hishinuma, T., Koseki, Y., Murai, Y., Yamakazi, T., Suzuki, K., and Mizugaki, M. Urinary thromboxane A2/prostacyclin balance reflecst the pathological state of a diabetic. Prostaglandins Other Lipid Mediat. 1999, 58(5–6), 263–271. 23. Hishinuma, T., Koseki, Y., Katayama, J., Murai, Y., Saito, T., and Mizugaki, M. Changes of the thromboxane A2/prostacyclin balance in the urine of patients with renal diseases analyzed by gas chromatography/selected ion monitoring. Prostaglandins Other Lipid Mediat. 2000, 60(1–3), 1–8. 24. Murakami, M. and Kudo, I. Recent advances in molecular biology and physiology of the prostaglandin E2-biosynthetic pathway. Prog. Lipid Res. 2004, 43, 3–35. 25. Jakobsson, P.-J., Thoren, S., Morgenstern, R., and Samuelsson, B. Identification of human prostaglandin E synthase: a microsomal, glutathione-dependent, inducible enzyme, constituting a potential novel drug target. Proc. Natl. Acad. Sci. USA 1999, 96, 7220–7225. 26. Regan, J.W. EP2 and EP4 prostanoid receptor signaling. Life Sci. 2003, 74, 143–153. 27. Kobayashi, T. and Narumiya, S. Function of prostanoid receptors: studies on knockout mice. Prostaglandins Other Lipid Mediat. 2002, 68–69, 557–573. 28. Urade, Y. and Eguchi, N. Lipocalin-type and hematopoietic prostaglandin D synthases as a novel example of functional convergence. Prostaglandins Other Lipid Mediat. 2002, 68–69, 375–382. 29. Breyer, R.M., Bagdassarian, C.K., Myers, S.A., and Breyer, M.D. Prostanoid receptors: Subtypes and signaling. Annu. Rev. Pharmacol. Toxicol. 2001, 41, 661–690. 30. Watanabe, K. Prostaglandin F synthase. Prostaglandins Other Lipid Mediat. 2002, 68–69, 401–407. 31. Negishi, M., Sugimoto, Y., and Ichikawa, A. Molecular mechanisms of diverse actions of prostanoid receptors. Biochim. Biophys. Acta 1995, 1259, 109–120. 32. Murphy, R. C., Hammarstrom, S., and Samuelsson, B. Leukotriene C: a slow reacting substance (SRS) from murine mastocytoma cells. Proc. Natl. Acad. Sci. USA 1979, 76, 4275–4279. 33. Samuelsson, B., Borgeat, P., Hammarstrom, S., and Murphy, R.C. Introduction of a nomenclature: leukotrienes. Prostaglandins 1979, 17, 785–787. 34. Penrose, J.F. LTC4 synthase. Enzymology, biochemistry, and molecular characterization. Clin. Rev. Allergy Immunol. 1999, 17, 133–152. 35. Jakobsson, P.-J., Morgenstern, R., Mancini, J., Ford-Hutchinson, A.W, and Persson, B. Membrane-associated proteins in eicosanoid and glutathione metabolism (MAPEG). Am. J. Respir. Crit. Care Med. 2000, 161, S20–S24. 36. Shimizu, T., Izumi, T., Honda, Z., Seyama, Y., Kurachi, Y., and Sugimoto, T. Biosynthesis and functions of leukotriene C4. Adv. Prostagl. Thrombx. Leukot. Res. 1990, 20, 46–53. 37. Mancini, J.A. and Evans, J.F. Coupling of recombinant 5-lipoxygenase and leukotriene A4 hydrolase activities and transcellular metabolism of leukotriene A4 in Sf9 cells. Eur. J. Biochem. 1993, 218, 477–484.

96

Functional Lipidomics 38. Hatzelmann, A., Fruchtmann, R., Mohrs, K.H., Raddatz, S., Matzke, M., Pleiss, U., Keldenich, J., and Muller-Peddinghaus, R. Mode of action of the leukotriene synthesis (FLAP) inhibitor BAY X 1005: implications for biological regulation of 5-lipoxygenase. Prostaglandins Other Lipid Mediat 1994, 43(1–2), 64–68. 39. Charleson, S., Prasit, P., Leger, S., Gillard, J.W., Vickers, P.J., Mancini, J.A., Charleson, P., Guay, J., Ford-Hutchinson, A.W., and Evans, J.F. Characterization of a 5lipoxygenase-activating protein binding assay: correlation of affinity for 5-lipoxygenase-activating protein with leukotriene synthesis inhibition. Mol. Pharmacol. 1992, 41(5), 873–879. 40. Rouzer, C.A., Matsumoto, T., Shimizu, T., and Samuelsson, B. The human leukocyte 5-lipoxygenase:enzyme purification, multi-component regulatory system, and LTA4 synthase activity. Adv. Prostaglandin Thromboxane Leukot. Res. 1986, 16, 3–16. 41. Chiba, N., Imai, H., Narashima, K., Oshima, G., Kunimoto, M., and Nakagawa, Y. Cellular glutathione peroxidase as a predominant scavenger of hydroperoxyeicosatetraenoic acids in rabbit alveolar macrophages. Biol. Pharm. Bull. 1999, 22(10), 1047–1051. 42. Zarini, S. and Murphy, R.C. Biosynthesis of 5-oxo-6,8,11,14-eicosatetraenoic acid from 5-hydroperoxyeicosatetraenoic acid in the murine macrophage. J. Biol. Chem. 2003, 278(13), 11190–11196. 43. Powell, W.S., Gravelle, F., and Gravel, S. Phorbol myristate acetate stimulates the formation of 5-oxo-6,8,11,14-eicosatetraenoic acid by human neutrophils by activating NADPH oxidase. J. Biol. Chem. 1994, 269(41), 25373–25380. 44. Guilbert, M., Ferland, C., Bosse, M., Flamand, N., Lavigne, S., and Laviolette, M. 5-Oxo-6,8,11,14-Eicosatetraenoic acid induced important eosinophil transmigration through basement membrane components: comparison of normal and asthmatic eosinophils. Am. J. Respir. Crit. Care Med. 1999, 21, 97–104. 45. Hosoi, T., Koguchi, Y., Sugikawa, E., Chikada, A., Ogawa, K., Tsuda, N., Suto, N., Tsunoda, S., Taniguchi, T., and Ohnuki, T. Identification of a novel human eicosanoid receptor coupled to Gi/o. J. Biol. Chem. 2002, 277(35), 31459–31465. 46. Powell, W.S., Gravel, S., MacLeod, R.J., Mills, E., and Hashefi, M. Stimulation of human neutrophils by 5-oxo-6,8,11,14-eicosatetraenoic acid by a mechanism independent of the leukotriene B4 receptor. J. Biol. Chem. 1993, 268(13), 9280–9286. 47. Jones, C.E., Holden, S., Tenaillon, L., Bhatia, U., Seuwen, K., Tranter, P., Turner, J., Kettle, R., Bouhelal, R., Charlton, S., Nirmala, N.R., Jarai, G., and Finan, P. Expression and characterization of a 5-oxo-6E,8Z,11Z,14Z-eicosatetraenoic acid receptor highly expressed on human eosinophils and neutrophils. Mol. Pharmacol. 2003, 63, 471–477. 48. Fitzpatrick, F.A., Morton, D.R., and Wynalda, M.A. Albumin stabilizes leukotriene A4. J. Biol. Chem. 1982, 257, 4680–4683. 49. Ford-Hutchinson, A.W, Bray, M.A., Doig, M.V., Shipley, M.E., and Smith, M.J. Leukotriene B, a potent chemokinetic and aggregating substance released from polymorphonuclear leukocytes. Nature 1980, 286, 264–265. 50. Yokomizo, T., Izumi, T., Chang, K., Takuwa, Y., and Shimizu, T. A G-protein-coupled receptor for leukotriene B4 that mediate chemotaxis. Nature 1997, 387, 620–624. 51. Yokomizo, T., Masuda, K., Kato, K., Toda, A., Izumi, T., and Shimizu, T. Leukotriene B4 receptor. Cloning and intracellular signaling. Am. J. Respir. Crit. Care Med. 2000, 161, S51–S55. 52. Nicosia, S., Capra, V., and Rovati, G.E. Leukotrienes as mediators of asthma. Pulmon. Pharmacol. Therap. 2001, 14, 3–19.

Eicosanoid Lipidomics

97

53. Christie, P.E. and Henson, P.M. Lipid inflammatory mediators: leukotrienes, prostaglandins, platelet-activating factor. Clin. Allergy Immunol. 2002, 16, 233–254. 54. Nicosia, S., Capra, V., Ravasi, S., and Rovati, G.E. Binding to cysteinyl-leukotriene receptors. Am. J. Respir. Crit. Care Med. 2000, 161, S46–S50. 55. Keppler, D., Leier, I., and Jedlitschky, G. Transport of glutathione conjugates and glucuronides by the multidrug resistance proteins MRP1 and MRP2. Biol. Chem. 1997, 378(8), 787–791. 56. Chen, X.S., Brash, A.R., and Funk, C.D. Purification and characterization of recombinant histidine-tagged human platelet 12-lipoxygenase expressed in a baculovirus/insect cell system. Eur. J. Biochem. 1993, 214(3), 845–852. 57. Takahashi, Y., Reddy, G.R., Ueda, N., Yamamoto, S., and Arase, S. Arachidonate 12lipoxygenase of platelet-type in human epidermal cells. J. Biol. Chem. 1993, 268(22), 16443–16448. 58. Metzger, K., Angres, G., Maier, H., and Lehmann, W.D. Lipoxygenase products in human saliva: Patients with oral cancer compared to controls. Free Radical Biol. Med. 1995, 18(2), 185–194. 59. Profita, M., Sala, A., Riccibono, L., Paterno, A., Mirabella, A., Bonanno, A., Guerrera, D., Pace, E., Bonsignore, G., Bousquet, J., and Vignola, A.M. 15-Lipoxygenase expression and 15-(S)-hydroxyeicosatetraenoic acid release and reincorporation in induced sputum of asthmatic subjects. J. Allergy Clin. Immunol. 2000, 105(4), 711–716. 60. Henriksson, P., Hamberg, M., and Diczfalusy, U. Formation of 15-HETE as a major hydroxyeicosatetraenoic acid in the atherosclerotic vessel wall. Biochim. Biophys. Acta 1985, 834(2), 272–274. 61. Wiesner, R., Suzuki, H., Walther, M., Yamamoto, S., and Kuhn, H. Suicide inactivation of the rabbit 15-lipoxygenase by 15S-HpETE is paralleled by covalent modification of active site peptides. Free Radical Biol. Med. 2003, 34(3), 304–315. 62. Funk, C.D. Prostaglandins and leukotrienes: Advances in eicosanoid biology. Science 2001, 294, 1871–1875. 63. Wallace, J.M. Nutritional and botanical modulation of the inflammatory cascadeeicosanoids, cyclooxygenases, and lipoxygenases-as an adjunct therapy. Integrative Cancer Therapies 2002, 1(1), 7–37. 64. Fierro, I.M. and Serhan, C.N. Mechanisms in anti-inflammation and resolution: the role of lipoxins and aspirin-triggered lipoxins. Braz. J. Med. Biol. Res. 2001, 34, 555–566. 65. Rowlinson, S.W., Crews, B.C., Goodwin, D.C., Schneider, C., Gierse, J.K., and Marnett, L.J. Spatial requirements for 15-(R)-hydroxy-5Z,8Z,11Z,13E- eicosatetraenoic acid synthesis within the cyclooxygenase active site of murine COX-2. Why acetylated COX-1 does not synthesize 15-(R)-HETE. J. Biol. Chem. 2000, 275(9), 6586–6591. 66. Fiore, S. and Serhan, C.N. Lipoxin A4 receptor activation is distinct from that of the formyl peptide receptor in myeloid cells: inhibition of CD11/18 expression by lipoxinA4-lipoxin A4 receptor interaction. Biochemistry 1995, 34(51), 16678–16686. 67. McMahon, B. and Godson, C. Lipoxins: endogenous regulators of inflammation. Am. J. Physiol. Renal Physiol. 2004, 286, F189–F201. 68. Sarkis, A., Lopez, B., and Roman, R.J. Role of 20-hydroxyeicosatetraenoic acid and epoxyeicosatetraenoic acids in hypertension. Curr. Opin. Nephrol. Hypertens. 2004, 13, 205–214. 69. Roman, R.J. P-450 metabolites of arachidonic acid in the control of cardiovascular function. Physiol. Rev. 2001, 82, 131–185.

98

Functional Lipidomics 70. Maier, K.G. and Roman, R.J. Cytochrome P450 metabolites of arachidonic acid in the control of renal function. Curr. Opin. Nephrol. Hypertens. 2001, 10, 81–87. 71. Sacerdoti, D., Gatta, A., and McGiff, J. C. Role of cytochrome P450-dependent arachidonic acid metabolites in liver physiology and pathophysiology. Prostaglandins Other Lipid Mediat. 2003, 72, 51–71. 72. Zeldin, D. C. Epoxygenase pathways of arachidonic acid metabolism. J. Biol. Chem. 2001, 276(39), 36059–36062. 73. Crackowski, J.-L. Isoprostanes: an emerging role in vascular physiology and disease? Chem. Phys. Lipids 2004, 128, 75–83. 74. Fam, S. S. and Morrow, J. D. The isoprostanes: unique products of arachidonic acid oxidation — a review. Curr. Med. Chem. 2003, 10, 1723–1740. 75. Morrow, J.D. and Roberts, L.J. The isoprostanes: their role as an index of oxidant stress status in human pulmonary disease. Am. J. Respir. Crit. Care Med. 2002, 166, S25–S30. 76. Bylund, J., Hidestrand, M., Ingelman-Sundberg, M., and Oliw, E.H. Identification of CYP4F8 in human seminal vesicles as a prominent 19-hydroxylase of prostaglandin endoperoxides. J. Biol. Chem. 2000, 275(29), 21844–21849. 77. Mizukami, Y., Sumimoto, H., Isobe, R., and Minakami, S. Omega-hydroxylation of lipoxin B4 by human neutrophil microsomes: identification of omega-hydroxy metabolite of lipoxin B4 and catalysis by leukotriene B4 omega-hydroxylase (cytochrome P-450LTB omega). Biochim. Biophys. Acta 1993, 1168(1), 87–93. 78. Hankin, J.A., Clay, C.E., and Murphy, R.C. The effects of ethanol and acetaldehyde on the metabolism of prostaglandin E2 and leukotriene B4 in isolated rat hepatocytes. J. Pharmacol. Exp. Ther. 1998, 285, 155–161. 79. Wheelan, P., Hankin, J.A., Bilir, B., Guenette, D., and Murphy, R.C. Metabolic transformation of leukotriene B4 in primary cultures of human hepatocytes. J. Pharmacol. Exp. Ther. 1999, 288, 326–334. 80. Hevko, J.M., Bowers, R.C., and Murphy, R.C. Synthesis of 5-oxo-6,8,11,14-eicosatetraenoic acid and identification of novel omega-oxidized metabolites in the mouse macrophage. J. Pharmacol. Exp. Ther. 2001, 296, 293–305. 81. Diczfalusy, U. β-Oxidation of eicosanoids. Prog. Lipid Res. 1994, 33, 403–428. 82. Hankin, J.A., Wheelan, P., and Murphy, R.C. Identification of novel metabolites of prostaglandin E2 formed by isolated rat hepatocytes. Arch. Biochem. Biophys. 1997, 340, 317–330. 83. Hamberg, M. and Samuelsson, B. On the metabolism of prostaglandins E1 and E2 in man. J. Biol. Chem. 1971, 246, 6713–6721. 84. Sala, A., Voelkel, N.F., Maclouf, J., and Murphy, R.C. Leukotriene E4 elimination and metabolism in normal human subjects. J. Biol. Chem. 1990, 265(35), 21771–21778. 85. Hou, X., Roberts L.J., II, Taber, D.F., Morrow, J.D., Kanai, K., Gobeil, F., Jr., Beauchamp, M.H., Bernier, S.G., Lepage, G., Varma, D.R., and Chemtob, S. 2,3Dinor-5,6-dihydro-15-F2t-isoprostane: a bioactive prostanoid metabolite. Am. J. Physiol. Regulatory Integrative Comp. Physiol. 2001, 281, R391–R400. 86. Turgeon, D., Chouinard, S., Belanger, P., Picard, S., Labbe, J.-F., Borgeat, P., and Belanger, A. Glucuronidation of arachidonic acid and linoleic acid metabolites by human UDP-glucuronosyltransferases. J. Lipid Res. 2003, 44, 1182–1191. 87. Prakash, C., Zhang, J.Y., Falck, J.R., Chauhan, K., and Blair, I.A. 20-Hydroxyeicosatetraenoic acid is excreted as a glucuronide conjugate human urine. Biochem. Biophys. Res. Commun. 1992, 185(2), 728–733.

Eicosanoid Lipidomics

99

88. Sacerdoti, D., Balazy, M., Angeli, P., Gatta, A., and McGiff, J.C. Eicosanoid excretion in hepatic cirrhosis. J. Clin. Invest. 1997, 100, 1264–1270. 89. Kikuta, Y., Kusunose, E., Sumimoto, H., Mizukami, Y., Takeshige, K., Sakaki, T., Yabusaki, Y., and Kusunose, M. Purification and characterization of recombinant human neutrophil leukotriene B4 omega-hydroxylase (cytochrome P450 4F3). Arc. Biochem. Biophys. 1998, 355(2), 201–205. 90. Zemksi Berry, K.A., Borgeat, P., Gosselin, J., Flammand, L., and Murphy, R.C. Urinary metabolites of leukotriene B4 in the human subject. J. Biol. Chem. 2003, 278(27), 24449–24460. 91. Murphy, R.C. and Zarini, S. Glutathione adducts of oxyeicosanoids. Prostaglandins Other Lipid Mediat. 2002, 68–69, 471–482. 92. Bogaards, J.J.P., Venekamp, J.C., and van Bladeren, P.J. Stereoselective conjugation of prostaglandin A2 and prostaglandin J2 with glutathione, catalyzed by the human glutathione S-transferases A1-1, A2-2, M1a-1a, and P1-1. Chem. Res. Toxicol. 1997, 10, 310–317. 93. Hubatsch, I., Mannervik, B., Gao, L., Roberts L.J., II, Chen, Y., and Morrow, J.D. The cylcopentenone product of lipid peroxidation, 15-A2t-isoprostane (8-isoprostaglandin A2), is efficiently conjugated with glutathione by human and rat glutathione transferase A4-4. Chem. Res. Toxicol. 2002, 15, 1114–1118. 94. Pace-Asciak, C.R., Laneuville, O., Su, W.-G., Corey, E.J., Gurevich, N., Wu, P., and Carlen, P.L. A glutathione conjugate of hepoxilin A3: Formation and action in the rat central nervous system. Proc. Natl. Acad. Sci. USA 1990, 87, 3037–3041. 95. Bowers, R.C., Hevko, J., Henson, P.M., and Murphy, R.C. A novel glutathione containing eicosanoid (FOG7) chemotactic for human granulocytes. J. Biol. Chem. 2000, 275, 29931–29934. 96. Livio, M., Chiabrando, C., Macconi, D., Benigni, A., Zimei, M., De Pietro, M.T., and Remuzzi, G. Metabolism of arachidonic acid in isolated glomeruli from pig kidney. Biochim. Biophys. Acta 1988, 961(1), 110–121. 97. Alber, D., Moussard, C., Toubin, M., Henry, J.C., Ottignon, Y., and Deschamps, J.P. Gas chromatographic/mass spectrometric quantitative analysis of eicosanoid in human oesophageal mucosa. Biomed. Environ. Mass Spectrom. 1988, 16(1–12), 299–304. 98. Hollenberg, S.M., Tong, W., Shelhamer, J.H., Lawrence, M., and Cunnion, R.E. Eicosanoid production by human aortic endothelial cells in response to endothelin. Am. J. Physiol. 1994, 267(6 Pt. 2), H2290–H2296. 99. Rademacher, D.J., Kearn, C.S., Carrier, E.J., Patel, S., Delgado, M.A., Barkmeier, A., Klick, D.E., Breese, N.M., Pfister, S.L., Nithipatikom, K., and Hillard, C.J. Production of hydroxyeicosatetraenoic acid and prostaglandins by a novel rat microglial cell line. J. Neuroimmunol. 2004, 149, 130–141. 100. Wu, T., Rieves, R.D., Larivee, P., Logun, C., Lawrence, M.G., and Shelhamer, J.H. Production of eicosanoids in response to endothelin-1 and identification of specific endothelin-1 binding sites in airway epithelial cells. Am. J. Respir. Cell Mol. Biol. 1993, 8(3), 282–290. 101. Revtyak, G. E., Hughes, M. J., Johnson, A. R., and Campbell, W. B. Histamine stimulation of prostaglandin and HETE synthesis in human endothelial cells. Am. J. Physiol. 1988, 255(2 Pt. 1), C214–C225. 102. King, K. L., Delamere, N. A., Csukas, S. C., and Pierce Jr., W. M. Metabolism of arachidonic acid by isolated rabbit ciliary epithelium. Exp. Eye Res. 1992, 55(2), 235–241.

100

Functional Lipidomics

103. Kempen, E.C., Yang, P., Felix, E., Madden, T., and Newman, R.A. Simultaneous quantification of arachidonic acid metabolites in cultured tumor cells using highperformance liquid chromatography/electrospray ionization tandem mass spectrometry. Anal. Biochem. 2001, 297, 183–190. 104. Newby, C.S. and Mallet, A.I. Rapid Simultaneous analysis of prostaglandin E2, 12hydroxyeicosatetraenoic acid and arachidonic acid using high performance liquid chromatography/electrospray ionization mass spectrometry. Rapid Commun. Mass Spectrom. 1997, 11, 1723–1727. 105. Chen, X., Li, N., Wang, S., Hong, J., Fang, M., Yousselfson, J., Yang, P., Newman, R.A., Lubet, R.A., and Yang, C.S. Aberrant arachidonic acid metabolism in esophageal adenocarcinogenesis, and the effects of sulindac, nordihydroguaiaretic acid, and a-difluoromethylornithine on tumorigenesis in a rat surgical model. Carcinogenesis 2002, 23 (12), 2095–2102.

5

Functional Lipidomics: Lysophosphatidic Acid as a Target for Molecular Diagnosis and Therapy of Ovarian Cancer Janos L. Tanyi, David Crotzer, Judith Wolf, Shuangxing Yu, Yutaka Hasegawa, John Lahad, Kwai Wa Cheng, Makiko Umezu-Goto, Glenn D. Prestwich, Andrew Morris, Robert A. Newman, Edward A. Felix, Rose Lapis, and Gordon B. Mills

CONTENTS Summary ................................................................................................................101 Introduction............................................................................................................102 Physiologic Role of LPA.......................................................................................103 LPA Receptors .......................................................................................................104 LPA Metabolism ....................................................................................................105 Pathophysiology of LPA in Ovarian Cancer.........................................................106 LPA Production or “Activity” as a Target for Therapy of Ovarian Cancer .........107 Targeting LPA Function in Patients ......................................................................115 Conclusions............................................................................................................116 References..............................................................................................................116

SUMMARY Epithelial ovarian cancer has the highest mortality rate of all gynecologic malignancies owing to late diagnosis and a lack of effective tumor-specific therapeutics. Ovarian carcinogenesis and metastasis occurs as a consequence of an orchestrated cascade of genetic, molecular, and biochemical events. Indeed, over the last several years, an extensive array of aberrations have been identified in this tumor; however, 101

102

Functional Lipidomics

their roles in the pathophysiology of ovarian cancer remain to be elucidated. Abnormal lysophosphatidic acid (LPA) production, receptor expression, and signaling are frequently found in ovarian cancers, suggesting that LPA plays a role in the pathophysiology of the disease. Although LPA is the simplest lipid found in nature, it contains a number of structure components providing important informational content. High-affinity LPA receptors of the G-protein-coupled receptor family provide evidence for the importance of the molecule in normal cellular functions. LPA levels and the levels of related lysopholipids have been reported to be elevated in patient fluids including ascitic fluid and peripheral blood. The recent identification of the enzymes that mediate the degradation and production of LPA and the development of receptor-selective analogs may lead to new approaches in the treatment of this deadly disease. The LPA pathway may contain novel molecular targets, illustrating the potential role of functional lipidomics in the development of new therapeutic and diagnostic strategies for disease management.

INTRODUCTION Ovarian cancer has the highest mortality rate of all gynecologic cancers. This gloomy prognosis results from an inability to diagnose the tumor at an early, curable stage because there is no well-established screening method. Although current therapeutic approaches, which consist of a wide variety of chemotherapeutic regimens preceded by radical debulking surgery, radiation therapy, or both, result in high response rates that translate into modest increases in survival time and improvement in quality of life, no notable increase in the cure rate has occurred in the last 25 yr (Penson et al., 1998; Greenlee et al., 2001). Most patients rapidly develop resistance to chemotherapy and eventually die of their disease. The overall 5-yr survival rate in patients with advanced disease remains less than 30% (Penson et al., 1998; Greenlee et al., 2001). Ovarian cancer genomes are remarkably unstable; an advanced cancer usually harbors multiple genomic changes, an unknown fraction of which influences biologic and clinical behavior (Mills et al., 2003a; Suzuki et al., 2000; Gray et al., 2003). This genomic variability, both within and between tumors, likely contributes to drug resistance and the lack of effective therapeutic approaches that can convert the high response rate into a high cure rate. The outcomes of ongoing genomic analyses may improve ovarian cancer management by revealing early events in ovarian cancer oncogenesis and progression that can be investigated as markers for early detection and identification of specific genetic aberrations that can be targeted therapeutically. Alternatively, global genomic analysis may identify patterns of genetic aberrations that predict outcome or response to particular therapies, allowing tailoring of disease management to the specific genetic aberrations in the tumor — individualized molecular medicine. Regional delivery of gene therapy constructs was heralded as an attractive alternative for the treatment of ovarian cancer because ovarian cancer tends to remain localized in the peritoneal cavity, with local tumor growth contributing to morbidity and mortality. However, most gene therapy approaches currently use nonspecific and nonselective prokaryotic promoters (e.g., CMV and SV40) that allow a high expression of the gene in normal cells, potentially limiting the therapeutic index

Functional Lipidomics: Lysophosphatidic Acid

103

because of toxicity to normal tissues. Further, it has proven difficult to introduce target genes into all tumor cells, and the long-hoped-for “bystander effect,” in which nontransfected cells die because of interactions with their neighbors, has not proven to be sufficiently effective to alter disease outcomes. Additional studies are needed to clarify optimal gene therapy targets and to identify effective methods for tumorselective delivery, such as tumor-specific promoters or cell-surface markers to deliver gene therapy preferentially and effectively to tumor cells (Tanyi et al., 2002; Lee et al., 2004; Bao et al., 2002). LPA, which was originally considered a precursor and component of lipid remodeling, is now known to constitute an important extracellular bioactive lipid, mediating cell proliferation, migration, and survival in almost every cell type, both normal and malignant (Mills and Moolenaar, 2003b). In this review, we summarize recent studies demonstrating the key role of LPA in ovarian carcinogenesis, supported by our recent findings of the therapeutic potential of LPA and its function and metabolism in this deadly disease. In the context of functional lipidomics, the biosynthesis, metabolism, and receptor-mediated interactions of LPA constitute an excellent case study of how disorders of lipid metabolism may offer important therapeutic targets (Feng et al., 2003).

PHYSIOLOGIC ROLE OF LPA Despite being one of the simplest of all serum phospholipids, LPA exerts pleomorphic effects on many cell lineages (Mills and Moolenaar, 2003b). LPA and its actions are highly conserved through ontogeny, and affect diverse organisms (Moolenaar et al., 1992). Yeast and insect cells, which lack functional receptors for LPA, have proven to be powerful models for understanding the mechanisms by which LPA mediates its effects (Hu et al., 2001). Although LPA induces cellular proliferation and survival in most cell lineages, it can induce differentiation or death in some. For example, in neurons, LPA can induce necrosis and apoptosis (Nietgen and Durieux, 1998). LPA also induces a number of early cellular responses, such as motility, chemotaxis, gap-junction opening, and invasion and morphologic changes that do not require new protein synthesis. Longer-term effects such as changes in cellular shape, increased cell viability, improved wound healing, and production of endothelin and proangiogenic factors may require, at least in part, new gene transcription (Mills and Moolenaar, 2003b). LPA may additionally function as part of an autocrine signaling loop by increasing the secretion or activation of multiple peptide growth factors, including transforming growth factors α and β, heparinbinding epidermal growth factor, insulin-like growth factor II, and endothelin 1 (Nakano et al., 1994; Pustilnik et al., 1999; Laffargue et al., 1999; Fang et al., 2000a). LPA also induces the production of several paracrine growth factors active on blood vessel endothelial cells, such as interleukin-6, interleukin-8, GRO alpha, and vascular endothelial growth factor; it also increases neovascularization in physiologic (wound healing) and pathologic (tumor) environments (Jalink et al., 1994; Xu et al., 1995a; Levine et al., 1997, Fang et al., 2004). LPA exerts many other vascular effects, including alteration of monocyte attachment to blood vessel walls, plaque formation, increased endothelial permeability, and vascular smooth muscle cell contraction.

104

Functional Lipidomics

These changes can alter blood pressure in animal models and humans and may also play a role in the development of atherosclerosis (Shulcze et al., 1997; van Nieuw Amerongen et al., 2000). Further, LPA is a precursor for other, more complex lipid syntheses, some of which may also function as bioactive mediators. For example, conversion of LPA to phosphatidic acid by endophilin and lysophospholipid acyltransferase is known to influence membrane curvature and the process of endocytosis (Schmidt, 1999). Most of the signaling effects of LPA appear to be due to binding to specific cell-surface receptors, but intracellular LPA has been demonstrated to activate the peroxisome proliferating activating receptor, potentially contributing to atherosclerosis, adipogenesis, and insulin signaling (McIntyre et al., 2003).

LPA RECEPTORS The extracellular activities of LPA are mediated through its binding to a group of G-protein-coupled receptors on the surface of mammalian cells. The first three members of the LPA-receptor family are members of the endothelial differentiation gene (EDG) family (An et al., 1998; Bandoh et al., 1999). The EDG2, EDG4, and EDG7 LPA receptors were recently renamed LPA1, LPA2, and LPA3, respectively. A fourth LPA receptor, LPA4, a member of the purinergic family of protein-coupled receptors, was recently identified (Noguchi et al., 2003). The roles of specific LPA receptors in the outcomes of extracellular LPA remain complex and elusive. The frequently contradictory results likely represent the spectrum of LPA receptors on the cell surface, the cell lines and potential lineages studied, and the interaction of LPA receptors with other receptors and intracellular signaling molecules. LPA1, which is widely expressed in the placenta, brain, small intestine, and colon, has a lower level of expression in pancreatic and normal ovarian tissue (Hecht et al., 1996). In some cell lineages, LPA1 is a major regulator of cellular motility by initiating Rho-dependent changes in cytoskeletal function, including cell rounding and stress-fiber formation (Fukushima et al., 1998; Van Leeuwen et al., 2003b). LPA2, however clearly mediates motility in at least some cellular lineages. LPA2 and LPA3 have more constrained distribution patterns than LPA1 does in terms of normal tissues. LPA1 and LPA2 are aberrantly expressed in prostatic and ovarian cancers (Fang et al., 2002) and in other cancer lineages, suggesting that they may be appropriate targets for therapy. LPA2 exhibits a higher affinity for LPA than the other family members, suggesting that it is a major contributor to the functions of LPA, particularly at the low levels of LPA that are present in normal tissues and plasma. LPA2 appears to be a major regulator of the production of vasculogenic factors (Fang et al., 2004; Huang et al., 2004). The lower-affinity LPA receptors may function in pathologic or stress states, such as those following cellular injury. These include, for example, wound healing, reperfusion after ischemia, and blood clotting (Okusa et al., 2003). The receptors through which LPA mediates cellular proliferation and viability remain controversial, although LPA3 is a strong candidate because LPA3-selective LPA homologues are potent activators of proliferation and viability (Hasegawa et al., 2003). LPA4 is expressed at a very low level in most human tissues, but a substantial level of expression has been detected in normal

Functional Lipidomics: Lysophosphatidic Acid

105

ovarian tissue (Noguchi et al., 2003). The roles of LPA4 in physiologic and pathologic states have not been elucidated. LPA activates at least three different G proteins via the LPA receptors, which can, in turn, stimulate multiple intracellular signal transduction systems. How the integration of these signaling events leads to the functional outcome of LPA-receptor ligation remains to be fully determined. Activation of Gq leads to activation of phospholipase C, producing diacylglycerol and inositol triphosphate, second messengers that contribute to the activation of protein kinase C and to increases in cytosolic calcium level, respectively. Activation of Gi feeds into three different and important signaling pathways: adenylate cyclase, by increasing CAMP levels and activating protein kinase A; RAS and the mitogen-activated protein kinase cascade; and activation of the phosphatidylinositol-3-kinase (PI3K) pathway (Van Leeuwen et al., 2003a; Fang et al., 2000a; Fang et al., 2000b). LPA also activates G12/13, contributing to activation of the small GTPase RhoA, which leads to cell rounding and cytoskeletal contraction (Etienne-Manneville and Hall, 2002).

LPA METABOLISM The concentration of LPA in plasma is normally low (100–200 nM), suggesting that the production, metabolism, and clearance of LPA are strictly controlled in vivo (Xu et al., 1998). LPA can be produced by activated platelets, adipocytes, leukocytes, fibroblasts, and endothelial cells, during clotting and, of particular importance, by ovarian cancer cells (Moolenaar et al., 1992; Goetzl et al., 1998; Fang et al., 2000b; Eder et al., 2000; Sano et al., 2002; Aoki et al., 2002). The primary pathway for extracellular LPA production is the activity of autotaxin/lysophospholipase D (ATX/lysoPLD), which removes a choline from lysophosphatidylcholine (LPC) (Sano et al., 2002; Aoki et al., 2002; Umezu-Goto et al.; 2002; Tokumura et al., 2002). LPC is produced by the removal of a fatty acyl chain from phosphatidylcholine (PC) by phospholipase A1 (PLA1) or A2 (PLA2). PLA2 has limited ability to hydrolyze lipids in intact membranes, suggesting that the major source of LPA production may be vesicles or apoptotic cells in which normal membrane structure is compromised. In addition, LPC is generated by the lecithine cholesterol acyltransferase pathway in which cholesteryl esters are generated by acyl transfer from PC. PC secreted by the liver and bound to albumin or low-density lipoproteins could be hydrolyzed by the sequential activity of PLA2 and ATX/lysoPLD, producing bioactive LPA (Croset et al., 2000). LPA can also be produced by the removal of a fatty acyl chain from PA by PLA1 or PLA2 (Aoki et al., 2002; Sano et al., 2002); however, the physiological relevance of this pathway remains to be determined. ATX/lysoPLD was originally identified as a major regulator of motility, metastasis, and tumor aggressiveness (Nam et al., 2000; Nam et al., 2001; Yang et al., 2002); however, the mechanism by which it mediates these processes was unknown. The discovery that ATX and lysoPLD were encoded by the same molecule resulted in a convergence of these two major areas of research (Umezu-Goto et al., 2002; Tokumura et al., 2002). ATX/lysoPLD is widely expressed, with the highest mRNA levels in the ovary, intestine, lung, and brain (Bachner et al., 1998). The fact that ATX/lysoPLD levels are markedly increased in many different cancers supports the

106

Functional Lipidomics

idea that it has a role in the pathophysiology of malignant diseases (Umezu-Goto et al., 2004). LPA is efficiently metabolized, being maintained at low levels under physiologic conditions. A family of lipid phosphate phosphohydrolases (LPPs) dephosphorylates LPA, decreasing the duration of LPA-mediated signaling events (Imai et al., 2000; Tanyi et al., 2003a; Tanyi et al., 2003b). This family consists of three members LPP1 (PAP2A), LPP-2 (PAP2C), and LPP-3 (PAP2B) (Hu et al., 2001). Other routes of LPA metabolism include acylation by lysophospholipid acyltransferase or endophilin to create phosphatidic acid and deacylation by lysophospholipases to produce glycerol phosphate (Wang and Dennis, 1999). In addition to its metabolism, LPA is cleared rapidly from circulation, contributing to its maintenance at appropriately low levels under physiologic conditions.

PATHOPHYSIOLOGY OF LPA IN OVARIAN CANCER Ascitic fluid from patients with ovarian cancer frequently contains elevated concentrations of LPA (up to 80 μM) (Eder et al., 2000; Xu et al., 2003). This suggests that ovarian cancer cells are exposed to an LPA-rich environment in vivo. Because ovarian cancer cells can produce LPA (Eder et al., 2000; Sengupta et al., 2003; Shen et al., 1998; Luquain et al., 2003), the cancer cells themselves are potential sources of elevated concentrations of LPA. LPA can increase the proliferation of ovarian cancer cells (Xu et al., 1995b), and it contributes to the multistep process of metastasis and invasion by increasing the production of urinary plasminogen activator (Pustilnik et al., 1999). Furthermore, LPA markedly decreases anoikis (a form of apoptosis) in ovarian cancer cell lines, suggesting that it plays a role in preventing anoikis and increasing metastasis in vivo. LPA also is a potent inducer of neovascularizing factors, vascular endothelial growth factor, and production of interleukin6 and interleukin-8 by ovarian cancer cells, potentially contributing to neovascularization and the aggressiveness of ovarian cancer (Hu et al., 2001; Fang et al., 2004; Schwartz et al., 2001), and to the elevated levels of these factors in ovarian cancer (Zebrowski et al., 1999). LPA is not produced in notable amounts by normal ovarian surface epithelial cells but is produced in considerable amounts by some ovarian cancer cell lines (Eder et al., 2000; Sengupta et al., 2003; Shen et al., 1998; Luquain et al., 2003). The levels of ATX/lysoPLD mRNA and protein levels are modestly elevated in malignant diseases. In at least half of ovarian cancers, patients exhibit at least a twofold increase, and one quarter exhibit up to a three-fold increase (Umezu-Goto et al., 2004). The production of LPA by ovarian cancer cells can be increased by phorbol esters, nucleotides, laminin, and even LPA itself (Eder et al., 2000; Sengupta et al., 2003; Shen et al., 1999; Luquain et al., 2003). In addition to increased rates of LPA production, LPA metabolism is decreased, with LPP-1 mRNA levels and activity being consistently higher in ovarian cancer samples as compared to normal ovarian epithelium (Tanyi et al., 2003a). The decreased LPA inactivation, the increased levels of ATX/lysoPLD protein, and the LPA autocrine loop likely contribute to the increased levels of LPA in ascitic fluid.

Functional Lipidomics: Lysophosphatidic Acid

107

LPA has modest activity on normal ovarian surface epithelium, which expresses LPA1 predominantly. Overexpression of LPA1 in ovarian cancer cell lines decreases cellular proliferation and increases apoptosis, suggesting that LPA1 can act as a negative regulator (Furui et al., 1999). In contrast, both LPA2 and LPA3 are aberrantly overexpressed in ovarian cancer cells, suggesting that LPA plays an important role in the pathophysiology of this cancer (Furui et al., 1999; Goetzl et al., 1999; Eder et al., 2000; Fang et al., 2002). Expression of LPA2 and LPA3 receptors in ovarian cancer cells greatly increases their ability to proliferate, form colonies, grow under anchorage-independent conditions, and to grow in vivo (unpublished data), all of which are compatible with a role for the receptor overexpression in the pathophysiology of ovarian cancer. In addition to increased LPA production and receptor expression, the genetic aberrations in ovarian cancer cells may contribute to increased responsiveness to LPA-receptor ligation. Both the PI3K and Ras/Erk pathways are stimulated by LPA receptors in ovarian cancer cells (Xu et al., 1995c; Fang et al., 2000b), and contribute to LPA-induced cell proliferation and survival. It is intriguing that the p110β catalytic subunit of PI3K is selectively activated by LPA in ovarian cancer cells (Roche et al., 1998). Both the PI3K and Ras/Erk pathways are highly activated in ovarian cancer cells, at least in part because of the presence of LPA in ascitic fluid. A recent review highlighted the importance of the PI3K pathway as a second target of lipidomics in developing new targeted signal transduction or targeted therapeutics for the treatment of ovarian cancer (Drees et al., 2003). If LPA is present at an increased concentration in ascitic fluid, it could diffuse into the systemic blood circulation, in which it could serve as a tumor marker for early disease detection. Indeed, plasma levels of LPA in patients with ovarian cancer are lower than they are in matched samples of ascitic fluid, a finding that is compatible with the theory that the levels of LPA in plasma represent diffusion from the peritoneal cavity (Eder et al., 2000). High LPA levels have been reported in the plasma of approximately 90% of patients with ovarian cancer (Xu et al., 1998). Similarly, aberrations of particular PC isoforms have been detected in the plasma of patients with ovarian cancer (Okita et al., 1997). However, considerable controversy exists about whether LPA levels are elevated in the plasma of patients with ovarian cancer and whether such elevations can predict the presence of ovarian cancer (Xiao et al., 2000; Baker et al., 2002; Yoon et al., 2003, Sutphen et al., 2004)

LPA PRODUCTION OR “ACTIVITY” AS A TARGET FOR THERAPY OF OVARIAN CANCER Elevated LPA concentrations in ascites could be a consequence of altered levels or activity of enzymes involved in LPA production, increased numbers of LPA-producing cells (i.e., tumor cells), or altered clearance. Using public transcriptional-profiling databases, the CGAP SAGE database, and our own transcriptional-profiling databases, we analyzed the transcriptional profiles of genes involved in LPA production and metabolism in ovarian cancer tissue samples and cell lines and compared them with those of normal ovarian surface epithelium and cell lines. We found

108

Functional Lipidomics

marked increases in levels of ATX/lysoPLD, the key enzyme regulating LPA production (Umezu-Goto et al., 2002; Tokumura et al., 2002). ATX/lysoPLD mRNA levels were elevated up to 200 times in a proportion of ovarian cancer cell preparations obtained directly from patients (Umezu-Goto et al., 2004). These increases were concurrent with decreases in the levels of LPPs (Tanyi et al., 2003a), which degrade LPA in ovarian cancer. The expected combined effects of the changes in these enzymes in ovarian cancer would be to increase LPA levels in the local tumor environment. It is important to note, however, that approximately 40% of ovarian cancer samples do not express markedly elevated levels of ATX/lysoPLD mRNA. Compatible with the increase in mRNA levels, ATX/lysoPLD activity is increased in the ascitic fluid of patients with ovarian cancer, compared with normal plasma (12.6 units vs. 6.9 units; p < 0.01), albeit at a lower level than mRNA levels. As with mRNA levels, ATX/lysoPLD activity was markedly elevated in a subset of patients; approximately 20% of patients did not demonstrate alterations in ATX/lysoPLD activity compared with plasma (Umezu-Goto, 2004). In contrast to ATX/lysoPLD and compatible with the increased LPA levels in ascitic fluid, the overall total levels of all three LPP mRNAs were lower in ovarian cancer tissue samples and cell lines than they were in normal ovarian epithelial tissues and cell lines (Tanyi et al., 2003a). Most of the difference in total LPP mRNA levels could be attributed to the decreased concentration of LPP-1, which was, on average, five times lower in ovarian cancers than in normal epithelial tissues. Again, it is important to note that only some patients’ samples demonstrated marked aberrations in LPP-1 mRNA levels. In contrast to LPP-1, LPP-2 and LPP-3 expressions were similar in ovarian cancer and normal epithelium. The same pattern of enzyme expression (i.e., decreased LPP-1 with normal LPP-2 and LPP-3 levels) was observed in the HEY, OVCAR-3, and SKOV3 ovarian cancer cell lines (Tanyi et al., 2003a). Together, these data suggest that ovarian cancer cells, compared with normal ovarian epithelium, have an increased ability to produce LPA. This is indeed the case with ovarian cancer cells producing high levels of LPA constitutively or in response to stimuli such as LPA itself, phorbol esters, laminin, and nucleotides (Eder et al., 2000; Sengupta et al., 2003; Shen et al., 1999; Luquain et al., 2003). The ability of ovarian cancer cell lines to produce LPA constitutively or inducibly depends on the action of PLA isozymes and PLD (Eder et al., 2000; Sengupta et al., 2003; Shen et al., 1998; Luquain et al., 2003). Whether the levels and activities of these enzymes are aberrant in ovarian cancer and contribute to elevated levels of LPA in the tumor microenvironment remains to be elucidated. Secretory PLA2 (sPLA2) has a limited ability to hydrolyze lipids in intact cell membranes but is highly efficient at hydrolyzing lipids in vesicles and apoptotic cells (Fourcade et al., 1998; Kudo et al., 1993; Fourcade et al., 1995). Ascitic fluid from patients with ovarian cancer contains elevated levels of vesicles accessible to sPLA2, potentially increasing the amount of substrate available for ATX/lysoPLD (Ginestra et al., 1999; Andre et al., 2002). In addition to alterations in the production or action of enzymes involved in LPA production, even a modest ability of ovarian cancer cells to produce LPA could result in marked aberrations in levels of LPA in patients with ovarian cancer. It is not

Functional Lipidomics: Lysophosphatidic Acid

109

unusual for these patients to present with more than 1 kg of tumor in the peritoneal cavity and several liters of ascitic fluid, which can contain up to 108 tumor cells per milliliter. Thus, the initial numbers of ovarian cancer tumor cells in patients can be very high. Together, these studies suggest that the ovarian cancer environment is LPA-rich because of aberrations in both LPA production and metabolism. We have begun to explore the LPA transcriptome in model systems and have identified a number of genes regulated by LPA. It is important that many of the LPA-regulated genes are coordinately altered in patients with ovarian cancer, compatible with the ovarian cancer microenvironments being LPA-rich and with LPAs playing a role in the pathophysiology of ovarian cancer. In this regard, the availability of LPA-affinity reagents from one of our laboratories (G.D.P.) and from Echelon Biosciences (Salt Lake City, UT) will facilitate the discovery of LPA-binding proteins and other physically associated proteins that constitute an essential part of the signaling complexes. Combined with the observations described in the preceding text, that the expression of LPA receptors is aberrant in ovarian cancer, the evidence that LPA levels and metabolizing enzymes are aberrant in ovarian cancer suggests that LPA production or activity could provide novel and effective targets for therapy. We have validated LPA production, metabolism, and action as potential therapeutic targets in ovarian cancer both in vitro and in vivo by evaluating the effects of decreasing LPA production by ovarian cancer cells by transfecting them with LPP. Ovarian cancer cell lines and normal control epithelial cells were transiently and stably transfected with LPP-1- and LPP-3-expressing vectors, and enzyme overexpression was confirmed by semiquantitative reverse transcriptionase polymerase chain reaction, immunoprecipitation, Western blotting, and immunofluorescence (Figure 5.1) (Tanyi et al., 2003a; 2003b). Overexpression of either LPP-1 or LPP-3 isozymes decreased proliferation and survival as indicated by the colony-forming activity of ovarian cancer cells and by MTT dye reduction (Figure 5.2). The ability of LPP-3 and LPP1 to decrease the colony-forming activity of ovarian cancer cells depended on LPA degradation because mutant, biologically inactive LPP-3 had no effect on colonyforming activity (unpublished data). To further determine whether the effects of LPP-3 on the growth of ovarian cancer cells were due to hydrolysis of extracellular LPA, we assessed the ability of a nonhydrolyzable LPA3-receptor-selective agonist OMPT (Hasegawa et al., 2003), to reverse the effects of LPP-3 expression. OMPT substantially reversed the inhibition of both colony-forming activity and apoptosis by LPP-3 (Tanyi et al., 2003b). To determine whether OMPT was acting as an agonist for LPA receptors or inhibiting LPP activity, we assessed OMPT’s ability to inhibit LPP-mediated LPA hydrolysis. High concentrations of OMPT (>10 μM) can inhibit LPPs, but the concentrations used in these studies (100 nM) failed to alter LPP activity. Recently, the activity of OMPT on LPA3 was shown to be enantioselective, with the “unnatural” (2S) enantiomer being fourfold to eightfold more potent than the 2R isomer (Qian et al., 2003). The ability of exogenous OMPT to reverse the effects of LPP-3 suggests that the major effect of hLPP-3 on the growth of ovarian cancer cells is due to hydrolysis of extracellular LPA. LPP-1 and LPP-3 overexpression markedly increased the apoptosis rate in ovarian carcinoma cell lines, with lesser effects on cell cycle

110

Functional Lipidomics

(a)

(b)

FIGURE 5.1 (See color insert following Page 174) LPP-3 overexpression in the stably transfected SKOV3 ovarian carcinoma cell line. (a) Parental SKOV3 cells and (b) LPP-3 enzyme-overexpressing stably transfected SKOV3 cells were fixed and stained with antibodies against LPP-3. Immunoreactive proteins were visualized using a fluorescein-conjugated secondary antibody as described by Smyth et al. (2003).

Functional Lipidomics: Lysophosphatidic Acid

111

8 7

Absorption 450 nm

6

Actual value Predicted value

5 4 3 2 1 0

Control

1 U pc

1 U LPP3

2 U pc

2 U LPP3 1 U pc+1 U 1 U pc+2 U 1 U pc+4 U LPP3 LPP3 LPP3

FIGURE 5.2 LPP-3 inhibits cell proliferation in stably transfected SKOV3 ovarian cancer cells Different units (U) of cells (1 U = 2 × 103) from stably hLPP-3-transfected SKOV3 cells (LPP-3) and SKOV3 parental cells (pc) were seeded onto 96-well plates and cultured for 24 h in RPMI 1640 medium. Thereafter, the cells were starved for 48 h. Cell proliferation induced by 0.1% bovine serum albumin containing mixed LPA was evaluated by MTT hydrolysis. The predicted value is the sum of the absorption values when the control and LPP-3-expressing cell lines were seeded separately. The actual value is that obtained when the control cells and hLPP-3-expressing cells were plated together.

progression, suggesting that the decreased growth rate was a consequence of both decreased proliferation and increased cell death. However, the recent demonstration that XY-14, a difluorophosphonate analog of PA (Xu et al., 2002), acts as a submicromolar inhibitor of LPP-1 (Smyth et al., 2003) but has no activity on the LPA Gprotein-coupled receptors, offers an additional tool for determining the importance of LPA metabolism and signaling in tumorigenesis. The amount of LPA produced by ovarian cancer cells depends on the rates of production and catabolism. ATX/lysoPLD and LPPs appear to be the major contributors to this process (Umezu-Goto et al., 2002; Imai et al., 2000), which suggests that overexpression of ATX/lysoPLD could potentially increase LPA production to levels sufficient to overcome the effects of LPP transfection. Indeed, ATX/lysoPLD reversed the growth-inhibitory (Figure 5.3) and apoptosis-inducing effects (Figure 5.4) of LPP-3 overexpression in ovarian carcinoma cell lines. This suggests that high levels of LPA production can override the effects of LPP expression. We evaluated the role of ATX/lysoPLD in tumorigenesis by increasing expression through transfection or decreasing expression through the use of RNAi in cancer cell lines. ATX/lysoPLD RNAi markedly decreased cell proliferation and survival in model

112

Functional Lipidomics

80

Average number of colonies/dish

70 60 50 40 30 20 10 0

pcDNA3.1

pc+ATX

LPP–3

LPP–3+ATX

Transfected constructs

FIGURE 5.3 Growth-inhibitory effects of LLP-3 overexpression after transfection of hLPP-3 are opposed by ATX/lysoPLD transient transfection. Two days after transfection, the cells were trypsinized, washed twice in PBS, and counted. Cells (3 × 104) were seeded into 30-mm, sixwell plates. Selection medium supplemented with G418 was changed every third day. Two weeks later, colonies were stained with 0.1% Coomassie blue (Serva, Heidelberg, Germany) in 30% methanol and 10% acetic acid. The mean number of colonies per dish is shown. The data represent the means ± standard error of three samples in three separate experiments.

systems (Umezu-Goto et al., 2004). Conversely, overexpression of ATX/lysoPLD can increase colony-forming cell activity and anchorage-independent growth (unpublished data). Together, these findings suggest that autocrine loops involving LPA play a critical role in the proliferation and survival of ovarian cancer cells. LPA has been shown to markedly increase cellular migration, which could contribute to tumor aggressiveness or metastasis (Van Leeuwen et al., 2003b; Stahle et al., 2003). We therefore assessed whether overexpression of LPP-1 in ovarian cancer cells would alter LPA-induced migration. LPA (10 μM) stimulated migration of control-transfected SKOV3 cells in a transwell assay. In contrast, LPA failed to induce migration in LPP-1-overexpressing SKOV3 ovarian cancer cells (data not shown), indicating that the increased extracellular hydrolysis of LPA by LPP-1 is translated into alterations of cellular motility and proliferation, colony formation, and survival. These findings are compatible with the hypothesis that decreases in LPP-1 levels and activity in ovarian cancer cells contribute to the pathophysiology of ovarian cancer through increased local LPA levels and subsequent increases in proliferation, survival, and activation of the metastatic cascade. This was also supported by previous study results indicating that LPA increased the activity and

Functional Lipidomics: Lysophosphatidic Acid

113

30

25

Apoptosis rate

20

15

10

5

0

No transfection

GFP only

GFP+LacZ

GFP+LPP–3

GFP+LPP– 3+ATX

Transfection conditions

FIGURE 5.4 Transient transfection of hLPP-3 increases apoptosis, but transient transfection with ATX/lysoPLD opposes this effect. Forty-eight hours after transient transfection with LPP and a green fluorescent protein-containing vector, cells were harvested and fixed with 0.25% paraformaldehyde in PBS solution followed by propidium iodide (10 μg/ml) DNA staining. Two-color cytometric analysis was performed, and the percentages of hypodiploid cells as an indication of apoptosis were determined (CellQuest software; Becton Dickinson, Franklin Lakes, NJ). The data shown are the means ± standard error of three separate experiments.

amount of a number of proteases involved in invasion and metastasis (Pustilnik et al., 1999; Fishman et al., 2001). LPP-1 and LPP-3 are transmembrane enzymes whose catalytic surface is on the exterior of the cell membrane. Thus, overexpression of these enzymes should decrease LPA concentrations in cell culture media. Compatible with this hypothesis, we demonstrated that cells transfected with LPPs had an increased ability to hydrolyze radiolabeled LPA (Tanyi et al., 2003a). To confirm this ability to hydrolyze extracellular LPA, we determined the concentration of LPA isoforms in supernatants of transfected and nontransfected cells using a MicroMass QuattroUltima triplequadrupole mass spectrometer (Waters Corporation, Milford, MA). Nontransfected SKOV3 cells induced moderate hydrolysis of extracellular LPA: LPA levels decreased to 76% of the original concentration at the end of the first hour. In contrast, transfection of LPP-1 and LPP-3 into SKOV3 cells resulted in decreases in the extracellular concentration of LPA to 52% and 28%, respectively, of the original concentrations after 1 h. If expression of LPP-1 and LPP-3 decreases extracellular LPA, then overexpression of LLP-1 and LLP-3 in one population of ovarian cancer cells should also

114

Functional Lipidomics

influence the colony-forming ability of nontransfected tumor cells. This might prove very important in gene therapy approaches because LPP-1 or LPP-3 could then induce the death of nontransfected bystander cancer cells. To evaluate this hypothesis, we mixed LPP-1- and LPP-3-transfected cells with an equal number of parental cells (Tanyi et al., 2003a,b). Both types of transfected cells decreased the proliferation of nontransfected tumor cells, a finding compatible with the hypothesis that the effect of LPP-1 or LPP-3 is related to degradation of an extracellular mediator, likely LPA. This also suggests that LPP-mediated gene therapy could result in effective bystander activity. Overexpression of LPP isoenzymes appears to alter cellular function through degradation of LPA and limitation of LPA-induced signaling. LPA induces changes in cytosolic calcium within seconds of its addition to the medium. In contrast, other signaling events such as phosphorylation of Erk kinases are delayed and prolonged. If the effects of LPPs primarily result from degradation of LPA, LPP overexpression should not alter the changes in cytosolic calcium that occur prior to LPA degradation but , however, LPP overexpression can be expected to decrease Erk phosphorylation. In contrast, if LPPs alter LPA-receptor function directly as has been suggested, then both responses should be inhibited. As indicated in Figure 5.5, there was no difference in the ability of LPA to induce increases in intracellular calcium mobilization in parental or LPP-transfected cells. In contrast, stable transfection with LPP-3 resulted in a decrease in maximal levels of Erk phosphorylation, and further in a more rapid decline in Erk phosphorylation levels to baseline. The differential rate of decline in Erk phosphorylation was compatible with the rate of degradation of LPA by LPP-transfected and parental cells (Tanyi et al., 2003a). The ability of LPPs to decrease cellular proliferation and apoptosis in vitro was reflected by a marked decrease in tumor “take” rates and growth rates in vivo. After subcutaneous injection of SKOV3 cells, tumors developed in 90% of mice compared with only 40% of mice injected with LPP-3-expressing SKOV3 cells. Thus, the take rate of tumors expressing LPP-3 was markedly less than that of the parental line. In addition, the growth rate of the hLPP-3-expressing SKOV3 tumor cells was markedly less than that of the parental cell lines (Tanyi et al., 2003b). At the termination of the study (mandated by tumor size and Association for Assessment and Accreditation of Laboratory Animal Care guidelines), the average weight of SKOV3 parental tumors was four times higher than that of hLPP-3-expressing SKOV3 tumors. When SKOV3 cells were injected into the orthotopic site for ovarian cancer (the peritoneal cavity), LPP-3 expression decreased the take rate to 20% in the injected mice, whereas all parental cell injections resulted in tumor growth. LPP3 expression decreased intraperitoneal growth, as assessed by both weight and abdominal circumference of the animals (Tanyi et al., 2003b). Thus, LPP-3 resulted in a marked decrease in tumor take rates and a decrease in growth rates in those cases in which tumors formed. However, after a delay in growth, the LPP-3-expressing tumors appeared to enter a more rapid growth phase. Because the hLPP-3 construct was not under selective pressure in vivo, it was possible that the eventual increase in growth rate was due to the loss of LPP-3. It was striking that when the rapidly growing tumors were excised and assessed for the presence of the LPP-3 transgene, LPP-3 levels in the transfected cell lines were

Functional Lipidomics: Lysophosphatidic Acid

115

PAP2B overexpressing SKOV3 cells 4000 Changes in [Ca2+] (nM)

Control (N=4) PAP2B (N=5)

3000

2000

1000

0

−12

−11

−10

−9 −8 −7 Log [18:1 LPA] (M)

−6

−5

−4

FIGURE 5.5 LPA induces intracellular calcium mobilization independently of LPP-3 expression. Parental and LPP-3-overexpressing SKOV3 ovarian cancer cell lines were grown to subconfluence, starved in serum-free medium, and harvested for cytoplasmic [Ca2+]I assay. Cytoplasmic [Ca2+]I was determined at an excitation wavelength of 331 nm and an emission wavelength of 410 nm using a fluorescence spectrophotometer (F-4000, Hitachi, Tokyo, Japan). Approximately 3 × 106 cells were used for [Ca2+]I determination, stirred in a quartz cuvette, and kept at 37ºC as detailed by Hasegawa et al. (2003).

markedly decreased. Furthermore, LPP-3 levels were decreased in the parental lines in vivo (Tanyi et al., 2003b). It therefore appears that the in vivo growth of SKOV3 cells is associated with the downregulated expression of transfected LPP-3. This suggests that a very strong negative selection exists against LPP-3 expression in ovarian cancer cells in vivo. Together, these studies suggest that an autocrine LPA loop is critical for the proliferation, survival, motility, and in vivo growth of ovarian cancer cells. A similar autocrine loop appears to be present in prostate cancer (Xie et al., 2002) and renalcell carcinoma (Umezu-Goto et al., 2004). Further, the results suggest that this autocrine loop is a potential target for therapy of ovarian and possibly other cancer cell lineages.

TARGETING LPA FUNCTION IN PATIENTS A wide variety of LPA-receptor-selective agonists and antagonists have been investigated (Fischer et al., 1998; Bandoh et al., 1999; Fischer et al., 2001; Okusa et al., 2003; Lynch et al., 2002; Hasegawa et al., 2003; Feng et al., 2003), but a highly effective therapeutic agent with drug-like characteristics that could be used in the treatment of ovarian cancer is not yet available. It is not yet clear whether a panLPA-receptor inhibitor would be toxic, precluding its utility and, thus, whether LPAreceptor-selective analogs would be required (Feng et al., 2003). However, mice lacking specific LPA receptors are viable, albeit with variable adult phenotypes, and LPP-overexpressing mice are also viable (Yang A. H. et al., 2002; Contos et al.,

116

Functional Lipidomics

2002; Yue et al., 2004). The effects of altering the expression of LPA receptor or LPP on tumorigenesis remain to be determined. The structural basis of LPA-receptor selectivity is beginning to be determined through structure–activity relationships and molecular modeling, suggesting that rational drug design could contribute to the creation of new and more selective receptor agonists and antagonists to serve as therapeutic mediators (Feng et al., 2003; Tigyi et al., 2003; Virag et al., 2003; Sardar et al. 2002; Wang et al., 2001). ATX/lysoPLD is a particularly attractive target for therapy. Its enzyme face is external and should be accessible to small-molecule drugs or inhibitory antibodies. Its role in cancer, both before it was found to be a lysoPLD and through studies of LPA in cancer, is well validated. Indeed, RNAi to ATX/lysoPLD potently downregulates cellular proliferation and survival (Umezu-Goto et al., 2004). We are currently exploring preclinical gene therapeutic approaches using LPPs on the basis of the effects of overexpressing these moieties in cells. We have developed a series of adenoviral vectors with broadly active promoters and with ovarian-cancer-specific promoters such as telomerase, survivin, and ceruloplasm (Tanyi et al., 2002; Bao et al., 2002; Lee et al., 2004) and are also exploring liposomal delivery approaches. Preliminary data indicates marked effects on tumor growth supporting the potential of gene therapy targeting the LPA pathway.

CONCLUSIONS Understanding the physiologic aberrations that originate from genetic alterations of ovarian cancer will lead to the development of new therapeutic approaches for treating the disease. Functional lipidomics is an essential part of developing signal transduction modifiers suited to the improvement of patient outcomes. Through its production, metabolism, and receptors, LPA may provide an excellent target for the development of molecular therapeutics, and the early detection of molecular forms of LPA and other lysolipids and the activities of LPA-pathway receptors and enzymes may facilitate both diagnosing and monitoring patient responses to therapy. The impressive development of knowledge about the pathway that regulates LPA production and the identification of selective LPA-receptor agonists suggest that targeting the LPA cascade could be a valuable approach to the management and treatment of this deadly disease. Additional studies of the LPA cascade and other phospholipids in ovarian cancer are essential to further elucidate their critical roles in ovarian cancer.

References 1. An, S., Bleu. T., Hallmark, O.G., and Goetzl, E.J. (1998). Characterization of novel subtype of human G protein-coupled receptor for lysophosphatidic acid. J. Biol. Chem. 273, 7906–7910. 2. Andre, F., Schartz, N.E., Movassagh, M., Flament, C., Pautier, P., Morice, P., Pomel, C., Lhomme, C., Escudier, B., Le Chevalier, T., Tursz, T., Amigorena, S., Raposo, G., Angevin, E., and Zitvogel, L. (2002). Malignant effusions and immunogenic tumour-derived exosomes. Lancet 360, 295–305.

Functional Lipidomics: Lysophosphatidic Acid

117

3. Aoki, J., Taira, A., Takanezawa, Y., Kishi, Y., Hama, K., Kishimoto, T., Mizuno, K., Saku, K., Taguchi, R., and Arai, H. (2002). Serum lysophosphatidic acid is produced through diverse phospholipase pathways. J. Biol. Chem. 277, 48737–48744. 4. Bachner, D., Ahrans, M., Schroder, D., Hoffmann, A., Lauber, J., Batat, N., Steinert, P., Flohe, L., and Gross, G. (1998). Bmp-2 downstream targets in mesenchymal development identified by subtractive cloning from recombinant mesenchymal progenitors (C3H10T1/2). Dev. Dyn. 213, 398–411. 5. Bandoh, K., Aoki, J., Hosomo, H., Kobayashi, T., Murakami-Murofushi, K., Tsujimoto, M., Arai, H., and Inoue, K. (1999). Molecular cloning and characterization of a novel human G-protein-coupled receptor, Edg-7, for lysophosphatidic acid. J. Biol. Chem. 274, 27776–27785. 6. Bandoh, K., Aoki, J., Taira, A., Tsujimoto, M., Arai, H., and Inoue, K. (2000). Lysophosphatidic acid (LPA) receptors of the Edg family are differentially activated by LPA species. Structure-activity relationship of cloned LPA receptors. FEBS Lett. 478, 159–165. 7. Baker, D.L., Morrison, P., Miller, B., Riely, C.A., Tolley, B., Westermann, A.M., Bonfrer, J.M., Bais, E., Moolenaar, W.H., and Tigyi, G. (2002). Plasma lysophosphatidic acid concentration and ovarian cancer. JAMA 287, 3081–3082. 8. Bao, R., Connolly, D.C., Murphy, M., Green, J., Weinstein, J.K., Pisarcik, D.A., and Hamilton, T.C. (2002). Activation of cancer-specific gene expression by the survivin promoter. J. Natl. Cancer Inst. 94, 522–528. 9. Contos, J.J., Ishii, I., Fukushima, N., Kingsbury, M.A., Ye, X., Kawamura, S., Brown, J.H., and Chun, J. (2002). Characterization of lpa(2) (Edg4) and lpa(1)/lpa(2) (Edg2/Edg4) lysophosphatidic acid receptor knockout mice: signaling deficits without obvious phenotypic abnormality attributable to lpa(2). Mol. Cell Biol. 22, 6921–6929. 10. Croset, M., Brossard, N., Polette, A., and Lagarde, M. (2000). Characterization of plasma unsaturated lysophosphatidylcholine in human and rat. Biochem. J. 345, 61–67. 11. Drees, B.E., Weipert, A., Hudson, H., Fergoson, C.G., Chakrararty, L., Prestwich, G.D. (2003). Competitive pluorescence polarization assays for the detection of phosphoinositide kinase and phosphatase activity. Comb. Chem. High Throughput Screen 4, 321–330. 12. Eder, A.M., Sasagawa, T., Mao, M., Aoki, J., and Mills, G.B. (2000). Constitutive and lysophosphatidic acid (LPA)-induced LPA production: role of phospholipase D and phospholipase A2. Clin. Cancer Res. 6, 2482–2491. 13. Etienne-Manneville, S., and Hall, A. (2002). RhoGTPases in cell biology. Nature 420, 629–635. 14. Fang, X., Yu, S., Lapushin, R., Lu, Y., Furui, T., Pena, L.Z., Stokor, D., Erickson, J., Bast, R.C., Jr., and Mills, G.B. (2000a). Lysophosphatidic acid prevents apoptosis in fibroblast via G(i) protein-mediated activation of mitogen activated protein kinase. Biochem. J. 352, 135–143. 15. Fang, X., Gaudette, D., Furui, T., Mao, M., Esterella, V., Eder, A.M., Pustilnik, T., Sasagawa, T., Lapushin, R., Yu, S., Jaffe, R., Wiener, J., Erickson, J., and Mills, G.B. (2000b). Lysophospholipid growth factor in the initiation, progression, metastasis, and management of ovarian cancer. Ann. N Y Acad Sci. 905, 188–208. 16. Fang, X., Schummer, M., Mao, M., Yu, S., Tabassam, F.H., Swaby, R., Hasegawa, Y., Tanyi, J.L., Lapushin, R., Eder, A., Jaffe, R., Erickson, J., and Mills, G.B. (2002). Lysophosphatidic acid is a bioactive mediator in ovarian cancer. Biochim. Biophys. Acta 1582, 257–264.

118

Functional Lipidomics

17. Fang, X., Yu, S., Bast, R.C., Liu, S., Lapushin, R., Claret, F.X., Aggarwal, B.B., Lu, Y., and Mills G.B. (2004). Mechanisms of lysophosphatidic acid-induced cytokine production in ovarian cancer cells. J. Biol. Chem. 279, 9653–9661. 18. Feng, L., Mills, G.B., and Prestwich, G.D. (2003). Modulators of lysophosphatidic Acid Signaling. Expert Opinions on Therapeutic Patents 13, 1619–1634. 19. Fishman, D.A., Liu, Y., Ellerbroek, S.M., and Stack, M.S. (2001). Lysophosphatidic acid promotes matrix metalloproteinase (MMP) activation and MMP-dependent invasion in ovarian cancer cells. Cancer Res. 61, 3194–3199. 20. Fischer, D.J., Liliom, K., Guo, Z., Nusser, N., Virag, T., Murakami-Murofushi, K., Kobayashi, S., Erickson, J., Sun, G., Miller, D.D., and Tigyi, G. (1998). Naturally occurring analogs of lysophosphatidic acid elicit different cellular responses through selective activation of multiple receptor subtypes. Mol. Pharmacol. 54, 979–988. 21. Fischer, D.J., Nusser, N., Virag, T., Yokohama, K., Wang, D.A., Baker, D.L., Bautista, D., Parril, A.L., and Tigyi, G. (2001). Short-chain phosphatidates are subtype selective antagonists of lysophosphatidic acid receptors. Mol. Pharmacol. 60, 776–784. 22. Fourcade, O., Le Balle, F., Fauvel, J., Simon, M.F., and Chap, H. (1998). Regulation of secretory type-II phospholipase A2 and of lysophosphatidic acid synthesis. Adv. Enzyme Regul. 38, 99–107. 23. Fourcade, O., Simon, M.F., Viode, C., Rugani, N., Leballe, F., Ragab, A., Fournie, B., Sarda, L., and Chap, H. (1995). Secretory phospholipase A2 generates the novel lipid mediator lysophosphatidic acid in membrane microvesicles shed from activated cells. Cell 80, 919–927. 24. Fukushima, N., Kimura, Y., and Chun, J. (1998). A single receptor encoded by vzg1/LPA1/Edg-2 couples to G proteins and mediates multiple cellular responses to lysophosphatic acid. Proc. Natl Acad. Sci. U S A 95, 6151–6156. 25. Furui, T., Lapushin, R., Mao, M., Khan, H., Watt, S.R., Watt, M.A., Lu, Y., Fang, X., Tsutsui, S., Siddik, Z.H., Bast, R.C. Jr., and Mills, G.B. (1999). Overexpression of Edg-2/vzg-1 induces apoptosis and anoikis in ovarian cancer cells in a lysophosphatidic acid-independent manner. Clin. Cancer Res. 5, 4308–4318. 26. Frisch, S.M., and Rouslahti, E. (1997). Integrins and anoikis. Curr. Opin. Cell Biol. 9, 701–706. 27. Gijsberg, R., Aoki, J., Arai, H., and Bollen, M. (2003). The hydrolysis of lysophospholipids and nucleotides by autotoxin (NPP-2) involves a single catalytic site. FEBS Lett. 538, 60–64. 28. Ginestra, A., Miceli, D., Dolo, V., Romano, F.M., and Vittorelli, M.L. (1999). Membrane vesicles in ovarian cancer fluids: a new potential marker. Anticancer Res. 19, 3439–3445. 29. Goetzl, E.J., and An, S. (1998). Diversity of cellular receptors and functions for the lysophospholipid growth factor lysophosphatidic acid and sphingosine 1-phosphate. FASEB J. 12, 1589–1598. 30. Goetzl, E.J., Dolezalova, H., Kong, Y., Hu, Y.L., Jaffe, R.B., Kalli, K.R., and Conover, C.A. (1999). Distinctive expression and function of the type 4 endothelial differentiation gene-encoded G protein coupled receptor for lysophosphatidic acid in ovarian cancer. Cancer Res. 59, 5370–5375. 31. Gray, J.W., Suzuki, S., Kuo, W.L., Polikoff, D., Smith McCune, K., Berchuck, A., Pinkel, D., Albertson, D., and Mills, G.B. (2003). Genome copy number abnormalities in ovarian cancer. Gynecol. Oncol. (suppl) S16–S21, 88. 32. Greenlee, R.T., Hill-Harmon, M.B., Murray, T., and Thun, M. (2001). Cancer statistics 2001. CA Cancer J. Clin. 51, 15–36.

Functional Lipidomics: Lysophosphatidic Acid

119

33. Hasegawa, Y., Erickson, J.R., Goddard, G.J., Yu, S., Liu, S., Cheng, K.W., Eder, A., Bandoh, K., Aoki, J., Jarosz, R., Schrier, A.D., Lynch, K.R., Mills, G.B., and Fang, X. (2003). Identification of phosphothionate analogue of lysophosphatidic acid (LPA) as a selective agonist of the LPA3 receptor. J. Biol. Chem. 278, 11962–11969. 34. Haseruck, N., Erl, W., Pandey, D., Tigyi, G., Ohlmann, P., Ravanat, C., Gachet, C., and Siess, W. (2003). The plaque lipid lysophosphatidic acid stimulates platelet activation and platelet-monocyte aggregate formation in whole blood: involvement of P2Y1 and P2Y12 receptors. Blood 103:2585–2592. 35. Hecht, J.H., Weiner, J.A., Post, S.R., and Chun, J. (1996). Ventricular zone gene-1 (vzg-1) encodes a lysophosphatidic acid receptor expressed in neurogenic regions of the developing cerebral cortex. J. Cell. Biol. 135, 1071–1083. 36. Hu, Y.L., Tee, M.K., Goeztl, E.J., Auersperg, N., Mills, G.B., Ferrara, N., and Jaffe, R.B. (2001). Lysophosphatidic acid induction of vascular endothelial growth factor expression in human ovarian cancer cells. J. Natl. Cancer Inst. 93, 762–768. 37. Huang, M.C., Lee, H.Y., Yeh, C.C., Kong, Y., Zaloudek, C.J., and Goetzl, E.J. (2004). Induction of protein growth factor systems in the ovaries of transgenic mice overexpressing human type 2 lysophosphatidic acid G protein-coupled receptor (LPA2). Oncogene 23, 122–129. 38. Imai, A, Furui, T, Tamaya, T, and Mills, GB (2000). A gonadotropin-releasing hormone-responsive phosphatase hydrolyses lysophosphatidic acid within the plasma membrane of ovarian cancer cells. J. Clin. Endocrinol. Metab. 85, 3370–3375. 39. Jalink, K., Hordijk, P.L., and Moolenaar, WH. (1994). Growth factor like effect of lysophosphatidic acid, a novel lipid mediator. Biochim. Biophys. Acta 1198, 185–196. 40. Kudo, I., Murakami, M., Hara, S., and Inoue, K. (1993). Mammalian non-pancreatic phospholipases A2. Biochim. Biophys. Acta 1170, 217–231. 41. Laffargue, M., Raynal, P., Yart, A., Peres, C., Wetzker, R., Roche, S., Payraste, B., and Chap, H. (1999). An epidermal growth factor receptor/Gab1 signaling pathway is required for activation of phosphoinositide 3-kinase by lysophosphatidic acid. J. Biol. Chem. 274, 32835–32841. 42. Lee, C.M., Lo, H.W., Shao, R.P., Wang, S.C., Xia, W., Gershenson, D.M., and Hung, M.C. (2004). Selective activation of ceruloplasmin promoter in ovarian tumors: potential use for gene therapy. Cancer Res. 64, 1788–1793. 43. Levine, J.S., Koh, J.S., Triaca, V., and Lieberthal, W. (1997). Lysophosphatidic acid a novel growth and survival factor for renal proximal tubular cells. Am. J. Physiol. 273, 6575–6585. 44. Lynch, K.R., and Macdonald, T.L. (2002). Structure-activity relationships of lysophosphatidic acid analogs. Biochim. Biophys. Acta 1582, 289–294. 45. Luquain, C., Singh, A., Wang, L., Natarajan, V., and Morris, A.J. (2003). Role of phospholipase D in agonist-stimulated lysophosphatidic acid synthesis by ovarian cancer cells. J. Lipid. Res. 44, 1963–1975. 46. Yang, S.Y., Lee, J., Park, C.G., Kim, S., Hong, S., Chung, H.C., Min, S.K., Han, J.W., Lee, H.W. and Lee, H.Y. (2002). Expression of autotaxin (NPP-2) is closely linked to invasiveness of breast cancer cells. Clin. Exp. Metastasis 19, 603–608. 47. Yang, A.H., Ishii, I., and Chun, J. (2002). In vivo roles of lysophospholipid receptors revealed by gene targeting studies in mice. Biochim. Biophys. Acta 1582, 197–203. 48. Yoon, H.R., Kim, H., and Cho, S.H. (2003). Quantitative analysis of acyl-lysophosphatidic acid in plasma using negative ionization tandem mass spectrometry. J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci. 788, 85–92.

120

Functional Lipidomics

49. Yue, J., Yokoyama, K., Balazs, L., Baker, D.L., Smalley, D., Pilquil, C., Brindley, D.N., and Tigyi, G. (2004). Mice with transgenic overexpression of lipid phosphate phosphatase-1 display multiple organotypic deficits without alteration in circulating lysophosphatidate level. Cell Signal. 16, 385–399. 50. McIntyre, T.M., Pontler, A.V., Silva, A.R., St. Hilaire, A., Xu, Y., Hinshaw, J.C., Zimmerman, G.A., Hana, K., Aoki, J., Arai, H., and Prestwich, G.D. (2003). Identification of intracellular receptor for lysophosphatidic acid (LPA): LPA is a transcellular PPARgamma agonist. Proc. Natl. Acad Sci. U S A 100, 131–136. 51. Mills, G.B., Bast, R.C., and Srivastava, S. (2001). Future for ovarian cancer screening: novel markers for emerging technologies of transcriptional profiling and proteomics. J. Natl. Cancer Inst. 93, 1437–1439. 52. Mills, G.B., May, C., McGill, M., Roifman, C., and Mellors, A. (1988). A putative new growth factor in ascetic fluid from ovarian cancer patients: identification, characterization and mechanism of action. Cancer Res. 48, 1066–1071. 53. Mills, G.B., Fang, X., Lu, Y., Hasegawa, Y., Eder, A., Tanyi, J., Tabassam, F.H., Mao, M., Wang, H., Cheng, K.W., Nakayama, Y., Kuo, W., Erickson, J., Gershenson, D., Kohn, E.C., Jaffe, R., Bast, R.C. Jr., and Gray, J. (2003). Specific keynote: Molecular therapeutics in ovarian cancer. Gynecol. Oncol. 88 (Suppl), S88–S92. 54. Mills, G.B. and Moolenar, W.H. (2003). The emerging role of lysophosphatidic acid in cancer. Nat. Rev. 3, 582–591. 55. Moolenaar, W., Jalink, K., and Van Corven, E. (1992). Lysophosphatidic acid: a bioactive phospholipid with growth factor-like properties. Rev. Physiol. Biochem. Pharmacol. 119, 7–65. 56. Nakano, T., Raines, E.W., Abraham, J.A., Klagsburn, M., and Ross, R. (1994). Lysophosphatidylcholine upregulates the level of heparin-binding epidermal growth factor-like growth factor mRNA in human monocytes. Proc. Natl. Acad. Sci. U S A 91, 1069–1073. 57. Nam, S.W., Clair, T., Campo, C.K., Lee, H.Y., Liotta, L.A., and Stracke, M.L. (2000). Autotaxin (ATX), a potent tumor mitogen, augments invasive and metastatic potential of ras-transformed cells. Oncogene 19, 241–247. 58. Nam, S.W., Clair, T., Kim, Y.S., McMarlin, A., Schiffmann, E., Liotta, L.A., and Stracke, M.L. (2001). Autotaxin (NPP-2), a metastasis-enhancing mitogen, is an angiogenic factor. Cancer Res. 61, 6938–6944. 59. Nietgen, G.W. and Durieux, M.E. (1998). Intercellular signaling by lysophosphatide, recent developments. Cell Adhes. Commun. 5, 221–235. 60. Noguchi, K., Ishii, S., and Shimizu, T. (2003). Identification of P2y9/GPR23 as a novel G protein-coupled receptor for lysophosphatidic acid, structurally distant from the Edg family. J. Biol. Chem. 278, 25600–25606. 61. Okita, M., Gaudette, D.C., Mills, G.B., and Holub, B.J. (1997). Elevated levels and altered fatty acid composition of plasma lysophosphatidylcholine (lysoPC) in ovarian cancer patients Int. J. Cancer 71, 31–34. 62. Okusa, M.D., Ye, H., Huang, L., Sigismund, L., Macdonald, T., and Lynch, K.R. (2003). Selective blockade of lysophosphatidic acid LPA3 receptors reduces murine renal ischemia-reperfusion injury. Am. J. Physiol. Renal Physiol. 285, F565–574. 63. Penson, R.T., Shannon, N.E., and Seiden, M.V. (1998). Ovarian cancer: an update of genetics and therapy. Compr. Ther. 24, 477–487. 64. Pustilnik, T.B., Esterella, V., Wiener, J., Mao, M., Eder, A., Watt, M.A.V, Bast, R.C., and Mills, G.B. (1999). Lysophosphatidic acid induces urokinase secretion in ovarian cancer cells. Clin. Cancer Res. 5, 3704–3710.

Functional Lipidomics: Lysophosphatidic Acid

121

65. Qian, L., Xu, Y., Hasegawa, Y., Aoki, J., Mills, G.B., and Prestwich, G.D. (2003). Enantioselective responses to a phosphorothioate analogue of lysophosphatidic acid with LPA3 receptor-selective agonist activity. J. Med. Chem. 46, 5575–5578. 66. Roche, S., Downward, J., Raynal, P., and Courtneidge, S.A. (1998). A function of phosphatidylinositol 3-kinase (p85–p110) in fibroblast during mitogenesis: requirement for insulin- and lysophosphatidic acid-mediated signal transduction. Mol. Cell. Biol. 18, 7119–7129. 67. Sano, T., Baker, D., Virag, T., Wada, A., Yatomi, Y., Kobayashi, I., Igarashi, Y., and Tigyi, G. (2002). Multiple mechanisms linked to platelet activation result in lysophosphatidic acid and sphingosine 1-phosphate generation in blood. J. Biol. Chem. 277, 21197–21206. 68. Sardar, V.M., Bautista, D.L., Fischer, D.J., Yokoyama, K., Nusser, N., Virag, T., Wang, D.A., Baker, D.L., Tigyi, G., and Parrill, A.L. (2002). Molecular basis for lysophosphatidic acid receptor antagonist selectivity. Biochim. Biophys. Acta 1582, 309–317. 69. Schmidt, A., Wolde, M., Thiele, C., Fest, W., Kratzin, H., Rodtelejnikov, A.V., Witke, W., Huttner, W.B., and Boling, H.D. (1999). Endophilin I mediates synaptic vesicle formation by transfer of arachidonate to lysophosphatidic acid. Nature 401, 133–141. 70. Schwartz, B.M., Hong, G., Morrison, B.H., Wu, W., Baudhuin, L.M., Xiao, Y.J., Mok, S.C. and Xu, Y. (2001). Lysophospholipids increase interleukin-8 expression in ovarian cancer cells. Gynecol. Oncol. 81, 291–300. 71. Sengupta, S., Xiao, Y.J. and Xu, Y. (2003). A novel laminin-induced LPA autocrine loop in the migration of ovarian cancer cells. FASEB J. 17, 1570–1572. 72. Shen, Z., Belinson, J., Morton, R.E., Xu, Y., and Xu, Y. (1998). Phorbol 12-myristate 13-acetate stimulates lysophosphatidic acid secretion from ovarian and cervical cancer cells but not from breast cancer or leukemia cells. Gynecol. Oncol. 71, 364–368. 73. Shulcze, C., Smales, C., Rubin, L.L., and Staddon, J.M. (1997). Lysophosphatidic acid increases tight junction permeability in cultured brain endothelial cells. J. Neurochem. 68, 991–1000. 74. Sutphen, R., Xu, Y., Wilbanks, G.D., Fiorica, J., Grendys, E.C. Jr., LaPolla, J.P., Arango, H., Hoffman, M.S., Martino, M., Wakeley, K., Griffin, D., Blanco, R.W., Cantor, A.B., Xiao, Y.J., Krischer, J.P. (2004). Lysophospholipids are potential biomarkers of ovarian cancer. Cancer Epidemiol Biomarkers Prev. 13: 1185-1191. 75. Suzuki, S., Moore, D., II, Ginzinger, D., Godfrey, T., Barclay, J., Powell, B., Pinkel, D., Zaloudek, C., Lu, K., Mills, G.B., Berchuck, A., and Gray, J.W. (2000). An approach to analysis of large-scale correlations between genome changes and clinical endpoints in ovarian cancer Cancer Res. 60: 5382–5385. 76. Smyth, S.S., Sciorra, V.A., Sigal, Y.J., Wang, Z., Xu, Y., Prestwich, G.D., and Morris, A.J. (2003). Lipid phosphate phosphatase 1 (LPP1) regulates lysophosphatidic acid signaling in platelets. J. Biol. Chem. 278, 43214–43223. 77. Stahle, M., Veit, C., Bachfischer, U., Schierling, K., Skripczynski, B., Hall, A., Gierschik, P., and Giehl, K. (2003). Mechanisms in LPA-induced tumor cell migration: critical role of phosphorylated ERK. J. Cell Sci. 116, 3835–3846. 78. Tanyi, J., Lapushin, R., Eder, A., Auersperg, N., Tabassam, F.H., Roth, J.A., Gu, J., Fang, B., Mills, G.B., and Wolf, J. (2002). Identification of tissue- and cancer-selective promoters for the introduction of genes into human ovarian cancer cells. Gynecol Oncol. 85, 451–458.

122

Functional Lipidomics

79. Tanyi, J.L., Hasegawa, J., Lapushin, R., Morris, A.J., Wolf, J.K., Berchuk, A., Lu, K., Smith, D.J., Kalli, K., Hartmann, L.C., McCune, K., Fishman, D., Broaddus, R., Cheng, K.W., Atkinson, E.N., Yamal, J.M., Bast, R.C. Jr., Felix, E.A., Newman, R.A., and Mills, G.B. (2003a). Role of decreased level of LPP-1 in accumulation of lysophosphatidic acid (LPA) in ovarian cancer. Clin. Cancer Res. 9, 3534–3545. 80. Tanyi, J., Morris, A.J., Wolf, J.K., Fang, X., Hasegawa, J., Lapushin, R., Auersperg, N., Sigal, Y.J., Newman, R.A., Felix, A.E., Atkinson, E.N., and Mills, G.B. (2003b). The human lipid phosphate phosphatase-3 decreases the growth, survival and tumorigenesis of ovarian cancer cells: validation of the lysophosphatidic acid signaling cascade as a target for therapy in ovarian cancer. Cancer Res. 63, 1073–1082. 81. Tigyi, G. and Parrill, A.L. (2003). Molecular mechanisms of lysophosphatidic acid action. Prog. Lipid Res. 42, 498–526. 82. Tokumura, A., Majima, E., Kariya, Y., Tominaga, K., Kogure, K., Yasuda, K., and Fukuzawa, K. (2002). Identification of human plasma lysophospholipase D, a lysophosphatidic acid-producing enzyme, as autotaxin, a multifunctional phosphodiesterase. J. Biol Chem. 277, 39436–39442. 83. Van Nieuw Amerongen, G.P., Wermeer, M.A., and van Hinsberg, V.W. (2000). Role of RhoA and Rho-kinase in lysophosphatidic acid-induced endothelial barrier dysfunction. Thromb. Vasc. Biol. 20, 127–133. 84. Van Leeuwen, F.N., Olivo, C., Grivell, S., Griepnans, B.N., Collard, J.G., and Moolenaar, W.H. (2003a). Rac activation by lysophosphatidic acid LPA1 receptors through the guanine nucleotide exchange factor Tiam 1. J. Biol. Chem. 278, 400–406. 85. Van Leeuwen, F.N., Giepmans, B.N., van Meeteren, L.A., and Moolenaar, W.H. (2003b). Lysophosphatidic acid: mitogen and motility factor. Biochem Soc. Trans. 31, 1209–1212. 86. Virag, T., Elrod, D.B., Liliom, K., Sardar, V.M., Parrill, A.L., Yokoyama, K., Durgam, G., Deng, W., Miller, D.D., and Tigyi, G. (2003). Fatty alcohol phosphates are subtype-selective agonists and antagonists of lysophosphatidic acid receptors. Mol. Pharmacol. 63, 1032–1042. 87. Wang, A. and Dennis, E.A. (1999). Review: mammalian lysophospholipases. Biochim. Biophys. Acta 1001, 282–285. 88. Xiao, Y, Chen, Y, Kennedy, AW, Belinson, J, and Xu, Y (2000). Evaluation of plasma lysophospholipids for diagnostic significance using electrospray ionization mass spectrometry (ESI-MS) analyses. Ann. N Y Acad. Sci. 905, 242–259. 89. Wang, D.A., Lorincz, Z., Bautista, D.L., Liliom, K., Tigyi, G., and Parrill, A.L. (2001). A single amino acid determines lysophospholipid specificity of the S1P1 (EDG1) and LPA1 (EDG2) phospholipid growth factor receptors. J. Biol. Chem. 276, 49213–49220. 90. Xie, Y., Gibbs, T.C., Mukhin, Y.V., and Meier, K.E. (2002). Role for 18:1 lysophosphatidic acid as an autocrine mediator in prostate cancer cells. J. Biol. Chem. 277, 32516–32526. 91. Xu, Y., Casey, G., and Mills, G.B. (1995a). Effects of lysophospholipids on signaling in the human Jurkat T cell line. J. Cell. Physiol. 163, 441–450. 92. Xu, Y., Gaudette, D., Boynton, J.D., Frankel, A., Fang, X.J., Sharma, A., Hurteau, J., Casey, G., Goodbody, A., Mellors, A., Holub, B., and Mills, G.B. (1995b). Characterization of an ovarian cancer activating factor (OCAF) in ascites from ovarian cancer patients. Clin. Cancer Res. 1, 1223–1232. 93. Xu, Y., Fang, X.F., Casey, G., and Mills, G.B. (1995c). Lysophospholipids activate ovarian and breast cancer cells. Biochem. J. 309, 933–940.

Functional Lipidomics: Lysophosphatidic Acid

123

94. Xu, Y., Shen, Z., Wiper, D.W., Wu, M., Morton, R.E., Elson, P., Kennedy, A.W., Belinson, J., Markman, M., and Casey, G. (1998). Lysophosphatidic acid as a potential biomarker for ovarian and other gynecologic cancer. JAMA 280, 719–723. 95. Xu, Y., and Prestwich, G.D. (2002). Synthesis of chiral (α,α-difluoroalkyl)phosphonate analogues of (lyso)phosphatidic acid via hydrolytic kinetic resolution. Org Lett. 4, 4021–4024. 96. Xu, Y., Xiao, Y.J., Zhu, K., Baudhuin, L.M., Lu, J., Hong, G., Kim, K.S., Cristina, K.L., Song, L., Williams, F., Elson, P., Markman, M., and Belinson, J. (2003). Unfolding the pathophysiological role of bioactive lysophospholipids. Curr. Drug Targets Immune Endocr. Metabol. Disord. 3, 23–32. 97. Zhang, C., Baker, D.L., Yasuda, S., Makarova, N., Balazs, L., Johnson, L.R., Marathe, G.K., McIntyre, T.M., Xu, Y., Prestwich, G.D., Byun, H.S., Bittman, R., and Tigyi, G. (2004). Lysophosphatidic acid induces neointima formation through PPAR (gamma) activation. J. Exp. Med. 199, 763–774. 98. Zebrowski, B.K., Liu, W., Ramirez, K., Akagi, M.D., Mills, G.B., and Ellis, L.M. (1999). Markedly elevated levels of vascular endothelial growth factor in malignant ascites. Ann. Surg. Oncol. 6, 373–378. 99. Umezu-Goto, M., Kishi, Y., Taira, A., Hama, K., Dohmae, N., Takio, K., Yamori, T., Mills, G.B., Inoue, K., Aoki, J., and Arai, H. (2002). Autotaxin has lysophospholipase D activity leading to tumor cell growth and motility by lysophosphatidic acid production. J. Cell. Biol. 158, 227–233. 100. Umezu-Goto, M., Tanyi, J.L., Lahad, J., Liu, S., Yu, S., Lapushin, R., Hasegawa, Y., Lu, Y., Trost, R., Bevers, T., Jonasch, E., Aldape, K., Liu, J., James, R.A., Ferguson, C.G., Xu, Y., Prestwich, G.D., and Mills, G.B. (2004). Lysophosphatidic acid production and action: validated targets in cancer. J. Cell. Biochem. 92:1115–1140.

6

Analysis of Lysophospholipids in Human Body Fluids: Comparison of the Lysophospholipid Content in Malignant vs. Nonmalignant Diseases Yi Jin Xiao and Yan Xu

CONTENTS Abstract ..................................................................................................................126 Introduction............................................................................................................126 Lipid Preparation from Biological Samples .........................................................127 The Biological Questions to Be Addressed and Sample Selection ............127 The Major Steps Involved in Lipid Analysis ..............................................128 Sample Collection and Processing ..............................................................128 The General Principles in Sample Collection and Processing .......128 The Effect of Temperature and Time between Blood Withdrawal and Processing (Centrifugation) on the Lipid Levels in Blood Samples ........................................128 The Centrifugation Conditions ........................................................129 Lipid Extraction ...........................................................................................131 Lipid Separation...........................................................................................131 Thin-Layer Chromatography (TLC): a Classical and Powerful Method to Separate Lipids ...............................................132 High-Performance Liquid Chromatography (HPLC)......................133 Solid-Phase Extraction.....................................................................134 Lipid Analyses .......................................................................................................134 The Importance of Lyso-PL Analyses .........................................................134

125

126

Functional Lipidomics

Comparisons of Different Lipid Analysis Methods ....................................134 Classical Methods for Lipid Detection............................................134 Biological Methods ..........................................................................135 GC Analyses.....................................................................................135 MALDI-MS and ESI-MS-Based Methods ......................................135 ESI-MS-Based Lyso-PL Analyses ...............................................................136 The Negative and Positive Detection Modes ..................................136 Internal Standards ............................................................................136 The Detection Mode and Conditions ..............................................138 The Standard Curves and Quantitative Analyses ............................138 Future Perspective..................................................................................................138 Reference ...............................................................................................................141

ABSTRACT Qualitative and quantitative analyses of lysophospholipids (lyso-PLs) in human body fluids and tissues have become critically important for our understanding of human physiology and pathology. As a new class of signaling molecules, these lipids play important regulatory roles in cellular functions. Thus, it is not surprising to find that a number of lyso-PLs are potential biomarkers for a variety of diseases, including cancers. Traditional lipid analytical methods, such as thin-layer chromatography (TLC), phosphorous determination, gas chromatography (GC), and high-performance liquid chromatography (HPLC) have been successfully used in phospholipid analysis. However, due to either the lack of sensitivity or the cumbersome analytical procedures, these methods are not highly effective in analyzing most lyso-PLs, some of which compose <1% of the total phospholipids. In addition, a complex array of enzymes and their regulators are involved in the metabolism of lyso-PLs. Therefore, analysis of lyso-PLs in human body fluids and tissues is not only an analytical, but also a biological issue. A good understanding of the biology of these lipids is essential for pathophysiologically relevant analyses of lyso-PLs. We will focus our discussion on: (1) a comparison of the electrospray ionization mass spectrometry (ESI-MS)-based method with other methods of lyso-PL analysis; (2) the influences of sample preparation on the analyses results; and (3) the issues related to sensitivity and reproducibility of the analyses, which are extremely important for biomarker identification and potential clinical use. The chapter includes discussions on the methods for qualitative and quantitative analysis of lyso-PLs in human body fluids (plasma, ascitic fluid, and urine) and in cells or cell culture media, as well as in animal tissues. The important factors that significantly affect the assays are emphasized. As future perspectives, we discuss the importance of further increasing the sensitivity and accuracy of the lyso-PL assays, as well as potential methods to establish high-throughput assays for lyso-PLs.

INTRODUCTION As one of the five major classes of biological molecules (DNA, RNA, protein, carbohydrates, and lipids), the chemical structures of the molecules of the lipid class

Analysis of Lysophospholipids in Human Body Fluids

127

are most diverse. By definition, lipids are a heterogeneous group of water-insoluble (hydrophobic) organic molecules that can be extracted from tissues by nonpolar solvents. Major types of lipids include hydrocarbons, alcohols (including sterols), long-chain amino alcohols (including sphingolipids), aldehydes, ketones and quinines, fatty acids, esters of sterols and vitamin alcohols, glycerides, glycerol ethers, phosphoglycerides (phospholipids), and glycolipids. Their biological functions range from energy storage to forming cell membrane building blocks, metabolic intermediates, hormones, neurotransmitters, and intracellular signaling molecules. In the past two to three decades, considerable information has been accumulating to show that a group of lyso-PLs, including lysophosphatidic acid (LPA), lysophosphatidylcholine (LPC), sphingosine-1-phosphate (S1P), and sphingosylphosphorylcholine (SPC), can be secreted or produced extracellularly and function as extracellular signaling molecules. These lyso-PLs are involved in a very broad range of biological processes [1–4]. The connection between lipid metabolism and human diseases has long been recognized, and the correlation of high cholesterol levels with the risk of heart disease is one of the best examples. One of the emerging areas of interest in the studies on the biological functions of lipids, and the signaling lyso-PLs in particular, is cancer. Cancers, as genetic diseases with dysregulated cellular signaling, show abnormal lipid metabolisms and signaling transduction. Thus, it is not surprising to observe that a number of lyso-PLs have been shown to have potential diagnostic or prognostic roles in cancers [5–9]. Abnormal expression of phospholipids has also been detected in many different tumors detected by nuclear magnetic resonance (NMR) techniques [10]. Therefore, it becomes increasingly interesting and important to analyze lipids both qualitatively and quantitatively. In particular, lyso-PLs are secreted and play important roles in regulating many biological functions of cancer cells and other cellular systems [4,11]. In this chapter, we will summarize and review the recent developments in lipid analyses, focusing on the analysis of lyso-PLs using mass spectrometry (MS)-based methods.

LIPID PREPARATION FROM BIOLOGICAL SAMPLES THE BIOLOGICAL QUESTIONS SAMPLE SELECTION

TO

BE ADDRESSED

AND

It is very important to bear in mind the biological or clinical questions to be addressed and to design the experiments accordingly. For example, if developing a method for the detection of a diagnostic marker for a disease is the goal, blood or urine samples are the first choice for a number of reasons: (1) procedures for obtaining these samples are non- or minimally invasive; (2) early detection is almost always critical in disease detection. If a marker can only be detected in tumor tissues, it often requires that the tumor has already attained a certain mass, and thus may represent a late-stage marker; and (3) it is relatively easy to obtain reliable, consistent, and reproducible results, because these samples are homogeneous. However, when cancer prognostic markers are being sought, tumor tissues could be better sources,

128

Functional Lipidomics

because they are from the site of the disease, although the host–environment interactions also play a critical role in patient prognosis. In general, lipids are present in all tissues and body fluids. They can be analyzed in samples from any part of human or animal tissues or fluids, if a specific issue needs to be addressed. For in vitro studies, lipids, including the lyso-PLs, are present in cell culture supernatant, on the cell surface, or inside cells, which can be fractionated and analyzed, depending on the biological questions to be studied. For blood sample analysis, serum and plasma are the two major categories used. For the former, blood is collected into collecting tubes without any additive and the platelets are aggregated. This process releases LPA and S1P [12–15], and thus should be prevented if measuring these lipids is a goal of the study. In collecting plasma samples, an anticoagulant is included. This can be EDTA, heparin, or citrate. For lipid analyses, EDTA is preferred, because it plays a dual role: preventing platelet aggregation and inhibiting cation-dependent lipid enzymatic activities.

THE MAJOR STEPS INVOLVED

IN

LIPID ANALYSIS

The important steps involved in lyso-PL analysis are shown in Figure 6.1. Detailed discussions on each of these steps are presented in the following sections.

SAMPLE COLLECTION

AND

PROCESSING

The General Principles in Sample Collection and Processing Our experience in lipid analyses over the past 10 yr indicates that one of the most important rules of lipid analyses in biological samples is that the samples have to be collected and processed consistently. Using blood samples as an example, the consistencies should be with regard to the time window and the temperature at which the samples are to be kept between blood withdrawal and processing, the centrifuge conditions, the storage conditions, and the cycles of freeze and thaw. The sensitivity to the conditions of sample collection and processing may reflect the inherited properties of lipids. As metabolites, lipids are likely to be produced, degraded, and converted to other lipids quickly in tissues, body fluids, and cells, which results in temperature- and time-dependent changes in lipid contents. In addition, lipids may be associated with different cellular and noncellular compartments, and physical separation conditions (such as centrifugation) may affect their distribution. These conditions will be discussed in detail in the following text. The Effect of Temperature and Time between Blood Withdrawal and Processing (Centrifugation) on the Lipid Levels in Blood Samples We found that fresh blood samples left at different temperatures gradually produced lyso-PLs. In particular, LPA content increased with time. Samples kept at room temperature produced more LPA than that at 4˚C. At 4˚C, the lyso-PLs levels were relatively stable for 4–8 h. Thus, it is important to keep the blood sample cool (at 0–4˚C) and process it within a few hours. The whole-blood samples should not be

Analysis of Lysophospholipids in Human Body Fluids

129

Body fluids Centrifugation Cell-free fluids Extraction by organic solvent HPLC

Protein precipitation

Lipid extract TLC LPC

LPA Lyso-PLs eluted from silica plate Hydrolysis

Injection

Fatty acid Methylation Fatty acid methyl esters

ESI-MS/MS

Sample UV detector Carrier gas GC

FIGURE 6.1 Lyso-PL analysis in human body fluids. The major steps involved in lyso-PL analyses include sample collection, processing (centrifugation and storage), lipid extraction, separation, and analyses. Different strategies (such as TLC versus HPLC method for lipid separation) may be employed. ESI-MS/MS: tandem electrospray ionization mass spectrometry; GC: gas chromatography; HPLC, high-performance liquid chromatography; MALDIMS, Matrix-assisted laser desorption ionization mass spectrometry; TLC: thin-layer chromatography.

frozen before centrifugation. After freeze and thaw, platelets (and possibly other blood cells) are lysed and high levels of lipids are released. The Centrifugation Conditions Blood samples are routinely collected and processed in hospitals and laboratories, however, no standard blood processing procedure is defined. To obtain serum or plasma, centrifugation conditions range from 1,000g for 10 min to 25,000g for 30 min at room temperature or 4˚C [16–18]. Because the lipid levels are affected by the centrifugation conditions [5,19], it is very important to keep them consistent for lipid analysis. Table 6.1 summarizes the conditions that we currently use in collecting and processing different biological samples.

4oC; 4–8 h

4oC; 4–8 h 1750 g; 15'; 20oC Siliconized tubes aliquoted

1750 g; 15'; 20oC Siliconized tubesa aliquotedb

3000 g; 20'; 4oC Siliconized tubes aliquoted

Ascitic Fluid Glass tubes

Serum Red top tubes

Plasma Purple top EDTAcontaining tubes 4oC; 4–8 h 3000 g; 20'; 4oC Lipid extraction without storage

4oC; < 1 h

Urine Glass tubes

No spin Siliconized tubes after homogenization

4oC; 4–8 h

Tissues Glass tubes

No spin Lipid extraction without storage

4oC; < 1 h

Cellsd Glass tubes

aIt is of pivotal importance to store extracted lipids in siliconized tubes (SafeSeal microcentrifuge tubes, low binding, 1.7 ml; Cat. # 505–201; PGC Scientifics, Frederick, MD). In regular Eppendorf centrifuge tubes, more than 90% LPA and other negatively charged lyso-PLs would stick to the tubes [19]. Thus, glass and siliconized plasticware are highly recommended in all steps of lipid processing. bCycles of freeze and thaw will change the lipid contents in samples. To prevent this, samples should be aliquoted to prevent repetitive freeze–thaw cycles. cThe lyso-PL contents are low in urine and cell supernatant samples (except under certain pathological conditions). Lipids should be extracted from fresh samples, because we have observed that long-term storage, as well as freeze and thaw, affect the levels of lyso-PLs in these samples. dThree fractions can be collected for lipid analyses: (1) cell supernatant (cell-free fragment) after removing cells, (2) cell-surface-bound lipids; can be eluted by washing cells with 0.1% BSA or a buffer with low pH (pH 2 to 3), and (3) cell pellet or lysate.

Collecting containers Storage before processing Centrifugation Storage

TABLE 6.1 Our Current Method for Sample Processing for Lipid Analysis

130 Functional Lipidomics

Analysis of Lysophospholipids in Human Body Fluids

131

LIPID EXTRACTION Extraction methods for the major phospholipids, such as phosphatidylcholine (PC), phosphatidylethanolamine (PE), and phosphatidylserine (PS), as well as LPC have been established for decades. The most commonly used methods are the Bligh and Dyer [20] and the Folch [21] methods, based on solvent extraction using chloroform and methanol. One of the important chemical properties of lyso-PLs is that they are relatively less hydrophobic due to the lack of acylation on one of their hydroxyl groups or the primary amine group. This property makes them more suitable for extracellular signal transduction. The second feature of these lipids is their low concentrations in biological systems. As signaling molecules, their production, degradation, and secretion are tightly controlled, and most of them (except LPC, which has functions other than signaling transduction) are present at low concentrations (<1 μM) in tissues and body fluids. When the controlling system is dysfunctional under pathological conditions, abnormal concentrations of lyso-PLs will be detected. These two features (relatively higher solubility in aqueous solutions and lower physiological concentrations) of lyso-PLs require modifications to the classical extraction methods used for phospholipids. For negatively charged lyso-PLs, such as LPA and lysophosphatidylinositol (LPI), acidification during the chloroform–methanol extraction method is one way to increase the yield. We have described this method in detail previously [22,23]. In acidic conditions, the charges on the phosphate group will be neutralized, which enhances the enrichment of these lipids in the organic phase. For zwitterion lyso-PLs, such as LPC, SPC, PAF, and lyso-PAF, neutralized solvent extraction is preferred. Butanol has been used to extract polar lipids for decades [24] and has been shown to be effective in extracting lyso-PLs [25]. However, this solvent is difficult to evaporate. In addition, it tends to extract more impurities (as it is rather polar solvent), which results in difficulties in subsequent lipid analysis using ESI-MS or other methods of analysis. Other organic solvents have also been tested in lyso-PL extraction. However, the yields are usually lower than that from the chloroform–methanol or butanol extraction. It is worth noting that internal standards for both negative and positive modes in MS detection are added to all samples prior to extraction [22,23], and thus the yield will be justified, even when it is low. However, a higher yield is still preferred for the higher levels of accuracy in measurement and for the advantage of using lower volumes of test samples.

LIPID SEPARATION Unlike the major phospholipid species, lyso-PLs compose less than 5% of the total lipids extracted by the chloroform–methanol–HCl method described earlier. Currently, there are no specific extraction methods for individual lipid species. For GC analyses, separation of lyso-PLs is absolutely required before transmethylation because GC analysis is unable to distinguish where the fatty esters are originally derived from. Although it is possible to detect individual lipid species using massbased methods without further purification of lipids, our experience indicates that

132

Functional Lipidomics

it is very difficult to achieve reproducible and quantitative analyses of lyso-PLs in the presence of large amounts of other lipids. Even though many of these lipids have different molecular weights and can be separately detected, an overwhelming excess of other lipids raise many concerns. To name a few, they may induce ion suppression and oversaturation of signals. The large amount of lipids and associated impurities tend to block the sample capillary of the electrospray tube, in addition to the problems associated with leftover contaminations, which shifts the background of detection. In particular, separation of LPC from LPA prior to MS detection is necessary. We have found that LPC could lose the choline group during ionization and be detected as LPA (unpublished results). Therefore, for lyso-PL analyses, a separation and purification step is necessary. Several methods have been tested in our lab, as well as in other labs. Thin-Layer Chromatography (TLC): a Classical and Powerful Method to Separate Lipids For lyso-PLs, the Silica gel 60 TLC plates (EM Science) are most commonly used. For separation of different lipid species, different solvent systems can be used. Avanti Polar Lipid Inc. provides a table in their catalog (under Technical Information) for the solvent systems suitable for separation of different lipid species with Rf values. We have shown that all major lyso-PL species, including LPC, LPI, lysophosphatidylethanolamine (LPE), lysophosphatidylglycerol (LPG), lysophosphatidylserine (LPS), lyso-platelet activating factor (lyso-PAF), and PAF, can be separated from LPA on one-dimensional (1-D) TLC plates, using the solvent system of chloroform–methanol–AmOH in the proportions of 65:35:5.5 (v:v:v), although the separation of LPI and LPS from LPA were not optimal [22,23]. In most cases, 1-D TLC is performed, which can separate lipid species from multiple samples simultaneously. In addition, standard lipids can be applied to the same TLC plate for the accurate localization of the lipid of interest. However, the resolution on 1-D TLC is limited. For example, it is difficult to clearly separate LPI and LPS from LPA using the solvent system mentioned earlier [22,23]. In contrast, two-dimensional (2-D) TLC provides much better separation. However, it is very time consuming and cost-inefficient, because only one sample can be separated on each plate. More importantly, one cannot apply a standard lipid on the same plate. No matter how carefully the quality of TLC plates and the running conditions are controlled, a particular lipid may not run to the exactly same position every time, and therefore, it is almost impossible to locate the spot of the lipid of interest accurately. Staining the spot is not an option, because staining is far less sensitive than the subsequent lipid analysis by MS or GC. In most cases, LPA, S1P, and other lyso-PLs are undetectable in biological samples using staining on TLC plates. In addition, the dye used for staining will interfere with subsequent analyses. It is critically important to run TLC plates at high quality. Lipid bands should be straight (Figure 6.1 and Figure 6.2). The TLC tanks should be saturated by solvent. Our experience suggests that two TLC plates (facing each other) running in the same tank give the best results, probably due to the balanced and even solvent environment in the tank. Nevertheless, both 1-D and 2-D TLC separations are time consuming

Analysis of Lysophospholipids in Human Body Fluids

A Plasma

LPC

B Ascites

C Tissue

133

D Urine

E Cell lysate

LPA

FIGURE 6.2 Lipid separation on TLC plates. (a) A representative TLC separation of lipids extracted from a plasma sample from a patient with ovarian cancer. Standard LPC and LPA were used for localize the LPA band and the LPC band; (b) a representative TLC separation of lipids extracted from an ascitic fluid sample from a patient with ovarian cancer; (c) lipids extracted from mouse liver tissue; (d) lipids extracted from a urine sample; and (e) lipids extracted from cell lysate from HEY ovarian cancer cells.

(usually it takes 1 to 2 h to run a plate). An additional problem associated with TLC is the potential for oxidation during lipid separation on TLC plates. High-Performance Liquid Chromatography (HPLC) The most appealing method for lipid separation for MS analyses is the online HPLC separation. HPLC has long been used in the separation, quantification, and characterization of phospholipids [26–29]. However, due to the lack of a sensitive detection system, lipid separation using HPLC becomes useful only after it is coupled to MS detection. In general, in developing LC-MS methods for lyso-PLs, it is necessary to take two factors into consideration: the best separation conditions when using HPLC (the column selection, the mobile phase, and the elution conditions), and the most favorable conditions for the MS detection. For example, lipid separations using normal-phase HPLC (including silica gel chromatography) usually have lower resolution than those of reverse-phase HPLC (such as ODS, or C18-based columns). However, normal phase is preferred for acidic lyso-PLs such as LPA and related lyso-PLs because the basic component (such as ammonium hydroxide [AmOH]) present in the solvents commonly used for normal-phase elution favors the negative ionization of acidic lyso-PLs in MS detection. In contrast, reverse-phase HPLC is preferred for the separation of choline-containing lyso-PLs, such as LPC, SPC, PAF, and lyso-PAF. Addition of formic acid in the mobile phase for elution of cholinecontaining lyso-PLs not only improves the separation but also increases the ionization efficiency of these analytes. Various mobile-phase and elution methods have been established. Generally, an elution solution for normal-phase HPLC is composed of a nonpolar solvent mixed with a low percentage of more polar solvents. For example, CHCl3 : MeOH : H2O : AmOH in the proportions 250 : 100 : 15 : 0.3 (v : v) [30] was used in silica gel column separation. In reverse-phase HPLC, mixtures of H2O and methanol are commonly used for lipid elution. As mentioned in the preceding text, a minor pH modifier (such as AmOH or formic acid) will enhance the ionization efficiency in

134

Functional Lipidomics

the relative MS detection mode. A higher pH favors negative ionization of acidic lipids, and a lower pH favors the positive ionization of basic lipids. Solid-Phase Extraction Solid-phase (both normal and reverse phase) extraction and separation of lipid classes, for example, separation of phospholipid using silica cartridges, were described decades ago [31]. Neutral lipids, free fatty acids, and polar lipids can be rapidly separated using silica Sep-Pak columns [32]. We have found that although lyso-PLs could potentially be separated and eluted from these cartridges, two main problems are associated with this method: (1) a large amount of MS-detectable impurities were eluted, and (2) it is difficult to obtain reproducible results. Further investigation of solid-phase-extraction-based separation is needed to evaluate the usefulness of this technology in lyso-PL analyses.

LIPID ANALYSES THE IMPORTANCE

OF

LYSO-PL ANALYSES

Among the five major classes of biological molecules, quantitative determination of lipids is relatively difficult. For DNA, RNA, and protein molecules, UV or visible absorbance is a highly convenient and quantitative method. The complementary hybridization between DNA and RNA strands provides a powerful method for detection of these molecules. In addition, proteins and carbohydrates are usually antigenic, and in most cases, antibodies can be readily generated against them. Although certain subclasses of lipids can be detected and quantified easily (such as sterols via their UV absorbance), many other lipids, including lyso-PLs, do not have strong UV or visible absorbance. They are not antigenic, due to their simple chemical structures and promiscuous existence in the animal kingdom. When present at high concentrations, the abundant phospholipid species can be quantitatively analyzed using well-developed traditional methods, such as dye staining of TLC-separated lipids, phosphorus determination, and UV detection of HPLC-separated lipids. However, these methods are hardly effective in quantitative analyses of lyso-PL, except for LPC, which is the most abundant lyso-PL in biological systems. Yet, qualitative and quantitative analyses of lyso-PLs has become essential because of the pathophysiological importance of lyso-PLs.

COMPARISONS

OF

DIFFERENT LIPID ANALYSIS METHODS

Classical Methods for Lipid Detection Phosphorus analyses, dye staining of specific subgroups of lipids separated on TLC plates, and HPLC separation using UV (205–206 nm) or light scattering detection are some classical methods for lipid detection. In addition, 1H and 31P NMR has been developed for phospholipid detection, and abnormal phospholipid levels associated with human diseases have been reported [10,33,34]. Although numerous papers on lipid analysis using these techniques have been published, we will not

Analysis of Lysophospholipids in Human Body Fluids

135

review these methods here because of two major reasons: (1) they are not sensitive enough for quantitative analyses of lyso-PLs and (2) none of these methods can be easily adapted to separate and analyze subspecies of lyso-PLs. For example, there are three subclasses of LPAs: acyl-, alkyl-, and alkenyl-LPAs; in each subclass, there are several different LPAs with different fatty acid chains [35]. Currently, MS-based methods are the best to meet both of these requirements (see subsections that follow). Biological Methods Various biological methods to analyze lyso-PLs have been developed [36–44]. These methods are dependent on either a biological activity of the lipid (such as receptormediated signaling activation), or an enzymatic activity converting the lipid to a photomerically or radioactively detectable substance. Although these methods are usually convenient to conduct, they may suffer from a number of shortcomings. First, they may not be sensitive enough to detect a pM range of lipids, which is essential for identifying low levels of lyso-PLs in small volumes of body fluids, tissues, or other biological systems. Secondly, it is difficult to use these methods to detect different molecular species of lyso-PLs, such as LPAs with varying fatty acid side chains. Thirdly, the results may be influenced by endogenous factors, which affect either the biological activity of the lipid or the enzymatic activity necessary for converting the lipid. Finally, the activity of the converting enzyme may vary from batch to batch. Gas Chromatography (GC) Analyses Though GC analyses are highly sensitive, they can only analyze gaseous molecules. Esterification is necessary to convert fatty acid to fatty esters, which can then be detected in GC. In addition, GC analysis cannot detect lyso-PLs such as sphingolipids that do not have esterifiable groups. As mentioned earlier, GC analyses require high levels of lipid purification. Unlike MS-based methods, which can detect the intact molecule, GC analyzes fatty acid derivatives from lyso-PLs and is not able to distinguish their origin. Owing to this property, internal standards for GC analyses can only be added after fatty acid hydrolysis, and the yields of the lipid extraction and separation steps cannot easily be assessed or justified among different samples. MALDI-MS and ESI-MS-Based Methods The development of soft ionization methods in MS, the matrix-assisted laser desorption ionization (MALDI), and electrospray ionization (ESI) have revolutionized the detection of biological molecules. Although vast and rapid advances have been made in the detection and characterization of proteins and other macromolecules, these MS methods, and the ESI-MS in particular, have been mostly used in the detection and identification of small organic molecules, including lipids [45,46]. It becomes essential for pharmacological studies. Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry is an essential tool for proteomics. Several groups have explored its potential for application in lipid analysis. Major phospholipids and LPC can be

136

Functional Lipidomics

analyzed using MALDI [47]. Two major concerns are associated with lyso-PL analysis using MALDI. First, MALDI is not highly quantitative, although it is highly sensitive. Secondly, spectra of phospholipid mixtures containing PC or LPC are strongly dominated by their signals, which prevents the detection of other phospholipids [47,48] ESI-MS-based methods for analyzing major species of phospholipids, including LPC, have been developed [9,49–55]. A few groups have been working on the development of methods for lyso-PLs analysis [9,22,25,30,56,57]. ESI-MS provides the most important features required for the quantitative analysis of lyso-PL, including high accuracy, reproducibility, sensitivity, and the applicability to complex lysoPLs without extensive separation and derivatization. We will discuss ESI-MS in detail in the next section.

ESI-MS-BASED LYSO-PL ANALYSES The Negative and Positive Detection Modes The MS negative scan mode detects negatively charged ions and the positive scan mode detects positively charges ions. Acidic lyso-PLs, such as LPA, LPI, LPS, LPE, and LPG can be detected in the negative mode. In contrast, the positive mode detects choline-containing lyso-PLs much more efficiently. Internal Standards Ideally, an isotope-labeled standard for each testing compound is the best internal standard. In practice, however, it is almost impossible if multiple lipids species are to be tested. We routinely test more than 30 different lyso-PL species in a sample and use two internal standards in most of our assays: one for the negative ion, and one for the positive ion detection mode. There are several criteria for the selection of internal standards: The molecule should be chemically similar to the analytes, must have a distinct molecular weight, and it should not be present in the biological samples to be analyzed. In addition, this molecule should share the daughter ion to be used for detection (see discussion that follows). Moreover, if an endogenous molecule exists with the same molecular weight and the daughter ion as that of the chosen compound, this compound is not suitable for being used as an internal standard. Based on these criteria, we initially chose 17:0-LPA (m/z = 423) and 17:-0-LPC (m/z = 510) as internal standards for the negative and positive ion detection modes, respectively. Even numbers of carbon atoms are present in the fatty acid chain of naturally occurring phospholipids. However, we detected a signal at m/z 423, with a daughter ion of 79 derived from a phosphate group, in ascitic fluid samples from patients with ovarian cancer (Figure 6.3). Tandem MS analysis identified this compound as 18:0-alkyl-LPA [57], which has the same molecular weight, and the daughter ion as 17:0-LPA. Therefore, we have had to choose a different internal standard. We found that in all biological samples that we analyzed, endogenous 14:0-LPA (m/z 381) is basically undetectable and thus, we have chosen this compound as our internal standard in the negative-mode lipid analyses [57]. Similarly, we have found that 12:0-LPC (m/z 440) is a better

Analysis of Lysophospholipids in Human Body Fluids LPAs

LPIs

409.5

100

137

599.3

437.1

619.6

456.9

%

571.7 481.1 394.9

644.9

0 (a) 437.3

100 16:0-Alkyl-LPA

395.2

%

18:0-Alkenyl-LPA

18:0-Alkyl-LPA

421.3

599.3

409.3 458.9

495.6 516.8

0 350

375

400

425

450

475

500

525 550 m/z (b)

575

600

625

650

675

700

FIGURE 6.3 Lyso-PLs detected in the LPA band isolated from TLC plates by using parent scan mode. (a) A representative MS spectrum of the LPA band lyso-PLs in a blood sample from a patient with ovarian cancer and (b) a representative MS spectrum of the LPA band lyso-PLs in an ascitic fluid sample from a patient with ovarian cancer. The presence of etherlinked LPAs is indicated by arrows.

internal standard for the positive mode, because we have detected an endogenous molecule at m/z 510 from some biological samples (derived from 18:0-lyso-PAF; unpublished results). The internal standards should be added before lipid extraction, so that the yield during extraction and purification will be justified. We routinely add 1 nmol of each internal standard to the samples to be analyzed.

138

Functional Lipidomics

The Detection Mode and Conditions We have found that to detect lyso-PLs, a parent mode setting is necessary. A normal negative (or positive) scan detects any negatively (or positively) charged molecules in the m/z range selected. A parent negative scan selectively detects those parent molecules containing the fragment chosen, such as m/z 79 (the phosphate group in the negative mode) or m/z 184 (the phosphocholine group in the positive mode). As shown in Figure 6.4a and Figure 6.4d, normal scans in the negative or the positive mode detected many molecules in materials eluted from the LPA band or the LPC band, but not the lyso-PLs. Only when the parent modes are used are these lipids specifically detected (Figure 6.4b and Figure 6.4e), indicating that even though these lipids are considered to be purified and enriched in these samples, large amounts of MS-detectable impurities still remain, and therefore, specific parent modes need to be used for lyso-PL detection. Because all lyso-PLs contain the phosphate group (m/z 79) and a glycerol backbone (m/z 153), and these groups can be fragmented efficiently and consistently, either the 79 or the 153 parent mode can be used for the negative scan mode. For the positive scan mode, the parent model at m/z 184 is most effective in detecting all phosphocholine-containing lyso-PLs. The Standard Curves and Quantitative Analyses The best way to set up standard curves is to use both standard lipids and the internal standard as we described previously [22,57]. It is interesting to note that the slopes of the standard curves (reflecting both ionization and fragmentation efficiencies) are different for lyso-PLs with different fatty acid chains within the same group [22]. Figure 6.5 shows that when LPCs with different fatty acid chains (all at the same concentration) were detected under the same MS conditions, different intensities were observed, with an interesting bell-shaped curve when this was plotted against the number of carbons in the fatty acid chain. Thus, one cannot simply assume that structurally similar compounds will have the same or similar ionization or fragmentation efficiency, and standard curves should be established for each individual compound under the same conditions that are to be used for testing samples.

FUTURE PERSPECTIVE Although ESI-MS-based methods are highly sensitive and effective in both qualitative and quantitative analysis of lyso-PLs, this technique is rather complex, and the equipment is rather expensive. Development of simpler and high-throughput assays of lyso-PL is a major challenge for researchers. The potential approaches include development of antibody-based analyses, elimination of some time-consuming steps in lipid analysis, and automation. Antibodies against certain phospholipids have been detected in humans [58–61]. Antibodies against PAF have been developed [62–66], and an antibody against LPA has been reported [67]. However, antibodies useful for the quantitative analysis of lyso-PLs have yet to be developed. As mentioned in the preceding text, its simple

Analysis of Lysophospholipids in Human Body Fluids

100

139

532.0

100

Normal scan (–)

Normal scan (+)

381.2 356.0

%

404.7 432.8

619.6 647.2

%

597.1

353.3 637.1

562.7

0

0

413.4

325.1

(d)

(a) 332.6

100

590.9 566.2

449.3 473.0

Parents of 79

496.6

100

Parents of 184

510.7

599.0 408.5 %

%

433.3 457.4

570.9

480.3

544.3

643.7 690.1

569.0 0

0 300

350

400

450

500 m/z

550

600

650

300

700

350

400

450

500 m/z (e)

(b) 153

152.

10

HOCH C HO CH

C R

O

H2C HO

Daughter of 409

O

171

104.3

O

CH

H2C O P OH OH 79

%

259

184.1

100

255

550

Daughter of 496

C

R

184 O P

O

CH2CH2

170

O

86.3

0

+ N(CH3)3

104

259.1

408

255

700

86

97

79

650

O

CH2 O

%

600

496.5 478.2

313.4

0 5

10

15

20

25 m/z (c)

30

35

40

45

50

100

150

200

250

300 m/z

350

400

450

500

(f)

FIGURE 6.4 Representative ESI-MS spectra of plasma lyso-PLs. The negative scan mode (a–c): (a) Normal scan detected molecules negatively charged, (b) the parent scan of m/z 79 detected lyso-PLs, and (c) a representative example of the structural identification of ions using tandem MS analysis. The positive scan mode (d–f): (d) Normal scan detected molecules positively charged, (e) the parent scan of m/z 184 detected lyso-PLs, and (f) a representative example of the structural identification of ions using tandem MS analysis.

structure, low physiological concentrations, and presence in all animals makes antibody production against lyso-PLs very difficult. However, it is still possible to make such antibodies through conjugation, micelle preparation, or other methods that enhance the antigenicity of the lipids. Should these antibodies become available, they would be very useful in lipid analyses. Eliminating certain time-consuming steps of lipid analysis may be an alternative approach for developing a high-throughput method. Among the different steps shown in Figure 6.1, lipid extraction in general is a necessary step. Lipid analyses in crude biological samples have been performed using MALDI or ESI-MS techniques [52,68–70]. However, quantitative determination of lyso-PLs is currently impossible using these methods. Lipid separation steps may represent an area where important improvements can be expected in the near future. Currently, LC-MS-based methods

Functional Lipidomics 18:0-LPC

140

523

20:0-LPC

16:0-LPC

100

496

24:0-LPC

12:0-LPC

10:0-LPC

440

384

580

608

510

413 Impurity

6:0-LPC

8:0-LPC

%

22:0-LPC

468

17:0-LPC internal standard

14:0-LPC

552

356

568

0 300

350

400

450

500 m/z

550

600

650

700

FIGURE 6.5 MS intensities of different LPCs. The same amount (80 pmol) of LPCs with different fatty acid chains and 50 pmol 17:0-LPC (the internal standard) was detected under the same MS conditions: the ion-spray interface was maintained at 70°C with a nitrogen nebulization flow of 10 l/h. The ESI drying gas (N2) was at 250 l/h. The argon (as collision gas) was at a pressure of 10 × 105 bar; the ion-spray voltage was 3500 V; the counter electrode potential was 500 V; the sample cone potential was 20 V; and the collision energy was 50 eV.

Analysis of Lysophospholipids in Human Body Fluids

141

are the most effective. However, they are still quite time consuming, and further improvements are required. Recently, Waters has developed an ACQUITY Ultra Performance LC (UPLC) system, based on the great technical strides made in particle chemistry performance, system optimization, detector design, and data processing and control. This system has the potential to provide improvement in speed, sensitivity, and resolution several times over, when compared to conventional HPLC. An additional area for investigation is the development of automated lipid extraction and purification procedures.

ACKNOWLEDGEMENT: This work was supported in part by RO1 HL68804, RO1 CA89228, by RO1 CA 095042 and a Ralph C. Wilson, Sr. and Ralph C. Wilson, Jr. Medical Research Foundation grant (to Y.X.).

Reference 1. Moolenaar, W.H. Mitogenic action of lysophosphatidic acid. Adv Cancer Res 1991, 57, 87–102. 2. Moolenaar, W.H. Development of our current understanding of bioactive lysophospholipids. Ann N Y Acad Sci 2000, 905, 1–10. 3. Spiegel, S. and Milstien, S. Sphingosine-1-phosphate: an enigmatic signalling lipid. Nat Rev Mol Cell Biol 2003, 4, 397–407. 4. Xu, Y., Xiao, Y.J., Zhu, K., Baudhuin, L.M., Lu, J., Hong, G., Kim, K.S., Cristina, K.L.; Song, L., Williams, F.S., Elson, P., Markman, M., and Belinson, J. Unfolding the pathophysiological role of bioactive lysophospholipids. Curr Drug Targets Immune Endocr Metabol Disord 2003, 3, 23–32. 5. Xu, Y., Shen, Z., Wiper, D.W., Wu, M., Morton, R.E., Elson, P., Kennedy, A.W., Belinson, J., Markman, M., and Casey, G. Lysophosphatidic acid as a potential biomarker for ovarian and other gynecologic cancers. JAMA 1998, 280, 719–723. 6. Sasagawa, T., Suzuki, K., Shiota, T., Kondo, T., and Okita, M. The significance of plasma lysophospholipids in patients with renal failure on hemodialysis. J Nutr Sci Vitaminol (Tokyo) 1998, 44, 809–818. 7. Okita, M., Gaudette, D.C., Mills, G.B., and Holub, B.J. Elevated levels and altered fatty acid composition of plasma lysophosphatidylcholine(lysoPC) in ovarian cancer patients. Int J Cancer 1997, 71, 31–34. 8. Sasagawa, T., Okita, M., Murakami, J., Kato, T., and Watanabe, A. Abnormal serum lysophospholipids in multiple myeloma patients. Lipids 1999, 34, 17–21. 9. Yoon, H.R., Kim, H., and Cho, S.H. Quantitative analysis of acyl-lysophosphatidic acid in plasma using negative ionization tandem mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci 2003, 788, 85–92. 10. Podo, F. Tumour phospholipid metabolism. NMR Biomed 1999, 12, 413–439. 11. Mills, G.B. and Moolenaar, W.H. The emerging role of lysophosphatidic acid in cancer. Nat Rev Cancer 2003, 3, 582–591.

142

Functional Lipidomics

12. Watson, S.P., McConnell, R.T., and Lapetina, E.G. Decanoyl lysophosphatidic acid induces platelet aggregation through an extracellular action. Evidence against a second messenger role for lysophosphatidic acid. Biochem J 1985, 232, 61–66. 13. Mauco, G., Chap, H., Simon, M.F., and Douste-Blazy, L. Phosphatidic and lysophosphatidic acid production in phospholipase C-and thrombin-treated platelets. Possible involvement of a platelet lipase. Biochimie 1978, 60, 653–661. 14. Eichholtz, T., Jalink, K., Fahrenfort, I., and Moolenaar, W. H. The bioactive phospholipid lysophosphatidic acid is released from activated platelets. Biochem J 1993, 291 (Pt. 3), 677–680. 15. Yatomi, Y., Ruan, F., Hakomori, S., and Igarashi, Y. Sphingosine-1-phosphate: a platelet-activating sphingolipid released from agonist-stimulated human platelets. Blood 1995, 86, 193–202. 16. Lepage, N., Miron, P., Hemmings, R., Roberts, K.D., and Langlais, J. Distribution of lysophospholipids and metabolism of platelet-activating factor in human follicular and peritoneal fluids. J Reprod Fertil 1993, 98, 349–356. 17. oist, J.H., Dolezel, G., Lloyd, J.V., and Mustard, F. Phospholipid transfer between plasma and platelets in vitro. Blood 1976, 48, 199–211. 18. Vecchione, J. and Zucker, M. B. Procoagulant activity of platelets in recalcified plasma. Br J Haematol 1975, 31, 423–428. 19. Sutphen, R., Xu, Y., Wilbanks, G., Fiorica, J., Grendys, E., LaPolla, J., Arango, H.; Hoffman, M., Martino, M., Wakefield, K., Griffin, D., Xiao, Y.J., and Krischer, J.P. Lysophospholipids are useful biomarkers in ovarian cancer. Cancer Epidemiol Biomarkers Prev 2004, 13, 1185–1191. 20. Bligh, E.G. and Dyer, W.J. A rapid method of total lipid extraction and purification. Can J Med Sci 1959, 37, 911–917. 21. Folch, J., Lees, M., and Sloane Stanley, G.H. A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem 1957, 226, 497–509. 22. Xiao, Y., Chen, Y., Kennedy, A.W., Belinson, J., and Xu, Y. Evaluation of plasma lysophospholipids for diagnostic significance using electrospray ionization mass spectrometry (ESI-MS) analyses. Ann N Y Acad Sci 2000, 905, 242–259. 23. Xiao, H., Siddiqui, R.A., Al-Hassani, M.H., Sliva, D., and Kovacs, R.J. Phospholipids released from activated platelets improve platelet aggregation and endothelial cell migration. Platelets 2001, 12, 163–170. 24. Furbish, F.S., Blair, H.E., Shiloach, J., Pentchev, P.G., and Brady, R.O. Enzyme replacement therapy in Gaucher’s disease: large-scale purification of glucocerebrosidase suitable for human administration. Proc Natl Acad Sci U S A 1977, 74, 3560–3563. 25. Baker, D.L., Morrison, P., Miller, B., Riely, C.A., Tolley, B., Westermann, A.M., Bonfrer, J.M., Bais, E., Moolenaar, W.H., and Tigyi, G. Plasma lysophosphatidic acid concentration and ovarian cancer. JAMA 2002, 287, 3081–3082. 26. Yandrasitz, J.R., Berry, G., and Segal, S. High-performance liquid chromatography of phospholipids: quantitation by phosphate analysis. Anal Biochem 1983, 135, 239–243. 27. McCluer, R.H., Ullman, M.D., and Jungalwala, F.B. HPLC of glycosphingolipids and phospholipids. Adv Chromatogr 1986, 25, 309–353. 28. Patton, G.M. and Robins, S.J. Separation and quantitation of phospholipid classes by HPLC. Methods Mol Biol 1998, 110, 193–215. 29. Lesnefsky, E.J., Stoll, M.S., Minkler, P.E., and Hoppel, C.L. Separation and quantitation of phospholipids and lysophospholipids by high-performance liquid chromatography. Anal Biochem 2000, 285, 246–254.

Analysis of Lysophospholipids in Human Body Fluids

143

30. Baker, D.L., Desiderio, D.M., Miller, D.D., Tolley, B., and Tigyi, G.J. Direct quantitative analysis of lysophosphatidic acid molecular species by stable isotope dilution electrospray ionization liquid chromatography-mass spectrometry. Anal Biochem 2001, 292, 287–295. 31. Juaneda, P. and Rocquelin, G. Rapid and convenient separation of phospholipids and non phosphorus lipids from rat heart using silica cartridges. Lipids 1985, 20, 40–41. 32. Hamilton, J.G. and Comai, K. Rapid separation of neutral lipids, free fatty acids and polar lipids using prepacked silica Sep-Pak columns. Lipids 1988, 23, 1146–1149. 33. Daly, P.F., Lyon, R.C., Faustino, P.J.;, andCohen, J.S. Phospholipid metabolism in cancer cells monitored by 31P NMR spectroscopy. J Biol Chem 1987, 262, 14875–14878. 34. Schiller, J. and Arnold, K. Application of high resolution 31P NMR spectroscopy to the characterization of the phospholipid composition of tissues and body fluids — a methodological review. Med Sci Monit 2002, 8, MT205–222. 35. Xu, Y., Xiao, Y.J.,Baudhuin, L.M., and Schwartz, B.M. The role and clinical applications of bioactive lysolipids in ovarian cancer. J Soc Gynecol Invest 2001, 8, 1–13. 36. Bast, R.C., Jr., Urban, N., Shridhar, V., Smith, D., Zhang, Z., Skates, S., Lu, K., Liu, J.,Fishman, D., and Mills, G. Early detection of ovarian cancer: promise and reality. Cancer Treat Res 2002, 107, 61–97. 37. Mills, G.B., Eder, A., Fang, X., Hasegawa, Y., Mao, M.,Lu, Y., Tanyi, J., Tabassam, F.H., Wiener, J., Lapushin, R., Yu, S., Parrott, J.A., Compton, T., Tribley, W., Fishman, D., Stack, M. S., Gaudette, D., Jaffe, R., Furui, T., Aoki, J., and Erickson, J.R. Critical role of lysophospholipids in the pathophysiology, diagnosis, and management of ovarian cancer. Cancer Treat Res 2002, 107, 259–283. 38. Saulnier-Blache, J.S., Girard, A., Simon, M.F., Lafontan, M., and Valet, P. A simple and highly sensitive radioenzymatic assay for lysophosphatidic acid quantification. J Lipid Res 2000, 41, 1947–1951. 39. Sciorra, V. A. and Morris, A. J. Roles for lipid phosphate phosphatases in regulation of cellular signaling. Biochim Biophys Acta 2002, 1582, 45–51. 40. Yatomi, Y., Ruan, F., Ohta, J., Welch, R.J., Hakomori, S., and Igarashi, Y. Quantitative measurement of sphingosine 1-phosphate in biological samples by acylation with radioactive acetic anhydride. Anal Biochem 1995, 230, 315–320. 41. Edsall, L.C. and Spiegel, S. Enzymatic measurement of sphingosine 1-phosphate. Anal Biochem 1999, 272, 80–86. 42. dsall, L., Vann, L., Milstien, S., and Spiegel, S. Enzymatic method for measurement of sphingosine 1-phosphate. Methods Enzymol 2000, 312, 9–16. 43. Kralik, S.F., Du, X., Patel, C., and Walsh, J.P. A method for quantitative extraction of sphingosine 1-phosphate into organic solvent. Anal Biochem 2001, 294, 190–193. 44. Min, J.K., Yoo, H.S., Lee, E.Y., Lee, W.J., and Lee, Y.M. Simultaneous quantitative analysis of sphingoid base 1-phosphates in biological samples by o-phthalaldehyde precolumn derivatization after dephosphorylation with alkaline phosphatase. Anal Biochem 2002, 303, 167–175. 45. Murphy, R.C., Fiedler, J., and Hevko, J. Analysis of nonvolatile lipids by mass spectrometry. Chem Rev 2001, 101, 479–526. 46. Ivanova, P.T., Milne, S.B., Forrester, J.S., and Brown, H.A. LIPID Arrays: New Tools in the Understanding of Membrane Dynamics and Lipid Signaling. Mol Interv 2004, 4, 86–96.

144

Functional Lipidomics

47. Petkovic, M., Schiller, J., Muller, M., Benard, S., Reichl, S., Arnold, K., and Arnhold, J. Detection of individual phospholipids in lipid mixtures by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry: phosphatidylcholine prevents the detection of further species. Anal Biochem 2001, 289, 202–216. 48. Petkovic, M., Schiller, J., Muller, J., Muller, M., Arnold, K., and Arnhold, J. The signal-to-noise ratio as the measure for the quantification of lysophospholipids by matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry. Analyst 2001, 126, 1042–1050. 49. Hsu, F.F., Turk, J., Thukkani, A.K., Messner, M.C., Wildsmith, K.R., and Ford, D.A. Characterization of alkylacyl, alk-1-enylacyl and lyso subclasses of glycerophosphocholine by tandem quadrupole mass spectrometry with electrospray ionization. J Mass Spectrom 2003, 38, 752–763. 50. Hsu, F.F. and Turk, J. Structural determination of sphingomyelin by tandem mass spectrometry with electrospray ionization. J Am Soc Mass Spectrom 2000, 11, 437–449. 51. Hsu, F.F., Bohrer, A., and Turk, J. Formation of lithiated adducts of glycerophosphocholine lipids facilitates their identification by electrospray ionization tandem mass spectrometry. J Am Soc Mass Spectrom 1998, 9, 516–526. 52. Han, X. and Gross, R.W. Global analyses of cellular lipidomes directly from crude extracts of biological samples by ESI mass spectrometry: a bridge to lipidomics. J Lipid Res 2003, 44, 1071–1079. 53. Kerwin, J.L., Tuininga, A.R., and Ericsson, L.H. Identification of molecular species of glycerophospholipids and sphingomyelin using electrospray mass spectrometry. J Lipid Res 1994, 35, 1102–1114. 54. Han, X. Characterization and direct quantitation of ceramide molecular species from lipid extracts of biological samples by electrospray ionization tandem mass spectrometry. Anal Biochem 2002, 302, 199–212. 55. Forrester, J.S., Milne, S.B., Ivanova, P.T., and Brown, H.A. Computational lipidomics: a multiplexed analysis of dynamic changes in membrane lipid composition during signal transduction. Mol Pharmacol 2004, 65, 813–821. 56. Fang, N., Yu, S., and Badger, T.M. LC-MS/MS analysis of lysophospholipids associated with soy protein isolate. J Agric Food Chem 2003, 51, 6676–6682. 57. Xiao, Y.J., Schwartz, B., Washington, M., Kennedy, A., Webster, K., Belinson, J., and Xu, Y. Electrospray ionization mass spectrometry analysis of lysophospholipids in human ascitic fluids: comparison of the lysophospholipid contents in malignant vs nonmalignant ascitic fluids. Anal Biochem 2001, 290, 302–313. 58. Wu, R.; Huang, Y.H., Elinder, L.S., and Frostegard, J. Lysophosphatidylcholine is involved in the antigenicity of oxidized LDL. Arterioscler Thromb Vasc Biol 1998, 18, 626–630. 59. Wu, R., Svenungsson, E., Gunnarsson, I., Andersson, B., Lundberg, I., Schafer Elinder, L., and Frostegard, J. Antibodies against lysophosphatidylcholine and oxidized LDL in patients with SLE. Lupus 1999, 8, 142–150. 60. Wu, R., Svenungsson, E., Gunnarsson, I., Haegerstrand-Gillis, C., Andersson, B., Lundberg, I., Elinder, L.S., and Frostegard, J. Antibodies to adult human endothelial cells cross-react with oxidized low-density lipoprotein and beta 2-glycoprotein I (beta 2-GPI) in systemic lupus erythematosus. Clin Exp Immunol 1999, 115, 561–566. 61. Ponzin, D., Mancini, C., Toffano, G., Bruni, A., and Doria, G. Phosphatidylserineinduced modulation of the immune response in mice: effect of intravenous administration. Immunopharmacology 1989, 18, 167–176.

Analysis of Lysophospholipids in Human Body Fluids

145

62. Sticherling, M., Schroder, J.M., and Christophers, E. Production and characterization of monoclonal antibodies against the novel neutrophil activating peptide NAP/IL-8. J Immunol 1989, 143, 1628–1634. 63. Karasawa, K. and Nojima, S. Production and application of antibodies to PAF. Tanpakushitsu Kakusan Koso 1991, 36, 516–523. 64. Smal, M.A., Baldo, B.A., and Redmond, J.W. Synthesis of a PAF immunogen and production of PAF-specific antibodies. Lipids 1991, 26, 1130–1135. 65. Tatsumi, N., Terano, Y., Hashimoto, K., Hiyoshi, M., and Matsuura, S. An anti-platelet activating factor antibody and its effects on platelet aggregation. Osaka City Med J 1993, 39, 167–174. 66. Macpherson, J.L., Spur, B., Pyne, S.G., Heymans, F., Cox, M.F., Godfroid, J.J., and Krilis, S.A. Production and characterization of antibodies to platelet-activating factor. J Lipid Mediat 1992, 5, 49–59. 67. Chen, J.H., Zou, F., Wang, N.D., Xie, S.W., and Zhang, X. Production and application of LPA polyclonal antibody. Bioorg Med Chem Lett 2000, 10, 1691–1693. 68. Drobnik, W., Liebisch, G., Audebert, F.X., Frohlich, D., Gluck, T.,Vogel, P., Rothe, G., and Schmitz, G. Plasma ceramide and lysophosphatidylcholine inversely correlate with mortality in sepsis patients. J Lipid Res 2003, 44, 754–761. 69. Liebisch, G., Drobnik, W., Lieser, B., and Schmitz, G. High-throughput quantification of lysophosphatidylcholine by electrospray ionization tandem mass spectrometry. Clin Chem 2002, 48, 2217–2224. 70. Kim, H.K., Yoon, H.R., and Pyo, D. Quantitative analysis of lysophosphatidic acid in human plasma by tandem mass spectrometry. Bull Korean Chem Soc 2002, 23, 1139–1143.

7

Functional Lipidomics: Lessons and Examples from the Sphingolipids Yusuf A. Hannun

CONTENTS Introduction and Historical Perspective ................................................................147 Overview of Sphingolipid Structure, Metabolism, and Function.........................148 Pathways of Sphingolipid Metabolism as a Network of Bioactive Lipids ..........152 Bioactive Lipids Regulate Cell Function: Notion of a Module..................152 Metabolic Interconversion of Two or More Bioactive Lipids: Notion of Switches...........................................................................152 Metabolically Interconnected Bioactive Lipids: Foundations for Networks ....................................................................................152 Function as an Integrated Output ................................................................154 Scope and Status of Lipidomics............................................................................155 Future Directions ...................................................................................................156 Developing Mechanistic Insights.................................................................156 Compartmentalization, Subcellular Localization, Topology, and Transporters: Significance of Physical Properties of Lipids...........157 Acknowledgments..................................................................................................157 References..............................................................................................................158

INTRODUCTION AND HISTORICAL PERSPECTIVE Undoubtedly, experimental biology has entered the era of “omics,” and with a vengeance — somewhat to the dismay of the more focused mechanistic scientist. This era was ushered in by adopting “big” science in biology with the initial target of sequencing entire genomes. To distinguish this pursuit as a novel activity, the term genomics was introduced and quickly gained wide acceptance, especially in the lay media. Among experimental biologists, current usage suggests that genomics refers to the study of entire genomes, comparative genomes, and, more importantly, the elucidation of biology through the study of genes, their expression, regulation, and function. These goals are highly laudable, and many of them have become feasible 147

148

Functional Lipidomics

(e.g., the use of microarrays for gene expression studies and genotyping); nevertheless, it has become obvious that the study of genomics, which by definition is centered on, and restricted to, the study of genes, is not sufficient to formulate all the laws governing biology or to reconstruct biological systems, even at the level of the single cell. At a very obvious level, the functions of genes are most often carried out by the protein products of these genes, and hence, a subdiscipline of protein science emerged that was dubbed proteomics, which aimed at elucidating the expression profile of an entire set of proteins (e.g., all the proteins expressed in a given cell at a given stage of development or function). The success of proteomics came from both the appreciation of the significance of proteins as the functional molecules in gene expression as well as the feasibility of studying entire protein data sets through the development of advanced mass spectrometry techniques and sophisticated analytical software. Even with the evolution of proteomics, many scientists, including (and perhaps especially) lipidologists, were not satisfied with the implications that the study of genes and proteins is sufficient to explain biology, and for obvious reasons. In addition to proteins, cells contain thousands of small molecules, and cells also interact with other extraneous molecules and agents from the environment. Therefore, to understand how a cell functions and responds, a biologist must have insights into its specific molecules and their interactions. To impress on the scientific and lay communities the need to incorporate all biological molecules in any proposed overarching biological construct, terms such as lipidomics, glycomics, and metabolomics were introduced and gained wide popularity. (The term lipidomics was first used in the Orlando ASBMB meeting in 2001, primarily with this in mind.) It appears that lipidomics currently denotes the study of entire classes or subclasses of lipids in an attempt to integrate their functions. As with proteomics and other “omics”, the study of lipidomics has been enabled by the evolution of sophisticated analytical methods (see other chapters) and by the significant growth in the understanding of the biochemical foundations of lipidology. It should be noted that although all omics emphasize the study of entire classes in an integrative fashion, such studies do create tension between classical reductive biochemistry and integrative biology. Perhaps a major challenge in current research is to reconcile these seemingly opposing approaches to the biological sciences. In this chapter, the study of sphingolipids is presented as a subset of lipidomics. Hopefully, this will illustrate what is unique about sphingolipids, their significance, and the progress in integrative approaches and also the further reductionist and mechanistic developments that are required.

OVERVIEW OF SPHINGOLIPID STRUCTURE, METABOLISM, AND FUNCTION The reader is referred to a number of excellent reviews that focus on sphingolipid metabolism, structures of the various sphingolipids, and the emerging understanding of the function of individual bioactive sphingolipids [1–3].

Functional Lipidomics: Lessons and Examples from the Sphingolipids

149

Sphingolipids are based on the unique structure of long-chain amino alcohols, the sphingoid bases. They are present in all known eukaryotic organisms as well as in the distinct genus Sphingomonas among the prokaryotes. Sphingolipid metabolism (Figure 7.1) commences with the condensation of two readily available metabolites, serine and palmitoyl CoA, by the action of serine palmitoyl transferase. This results in the formation of keto-dihydrosphingosine, which is then reduced to dihydrosphingosine (sphinganine) followed by N-acylation to dihydroceramide by (dihydro) ceramide synthase. Introduction of a double bond at the 4-5 position by a desaturase results in the formation of ceramide in mammals. In turn, ceramide serves as a precursor for the biosynthesis of complex sphingolipids, including sphingomyelin, cerebroside, glycosphingolipids, gangliosides, sulfatides, and ceramide-1-phosphate through the action of specific synthases (e.g., sphingomyelin synthase and glucosylceramide synthase). The breakdown of complex sphingolipids proceeds stepwise through the actions of various hydrolytic enzymes that cleave off the head groups of sphingolipids, ultimately resulting in the formation of ceramide. Ceramide may then be deacylated by one of several ceramidases to generate sphingosine, which can in turn be phosphorylated to sphingosine-1-phosphate (S1P) through the action of sphingosine Serine + Palmitoyl-CoA Serine Myriocin palmitoyltransferase

Ethanolamine 1-P+ Palmitaldehyde

3-Ketodihydro-sphingosine NADPH-Dependent Reductase Sphinganine + Fatty acyl CoA (Dihydrosphingosine) Dihydroceramide synthase

LYASE CH2OPOH

OH

Sphingosine 1-P

D-MAPP

O

S-1-P Phosphatase

Desaturase

Ceramide 1-P

Sphingosine kinase

Fatty acid

Ceramide kinase

OH CH OH 2

OH

Ceramidase

CH2OH

NH O

Ceramide PC

SM synthase

GC S

CER EB

RO

Smase

OH

Sphingosine

Ceramide synthase

Choline phosphate

DAG

O-

NH3+

CH2OH NH

Dihydroceramide

O

OH

Fumonisin B1

O

NH3+

PDMP/P4

SID ASE

Glucosylceramide

Lactosylceramide

CH2OPOCH2CH2N+(CH3)3 – O NH

Sphingomyelin

O

Glycosphingolipids

FIGURE 7.1 (See color insert following page 174.) Shown are the pathways of sphingolipid metabolism with the structures of sphingosine, sphingosine-1-phosphate (S1P), dihydroceramide, ceramide, and sphingomyelin. Highlighted are the established bioactive lipids, and shown in red are the representative inhibitors of specific enzymes in the pathway.

150

Functional Lipidomics

kinases. S1P lyase then cleaves this molecule to generate ethanolamine phosphate and a fatty aldehyde. The study of sphingolipids has witnessed a dramatic resurgence with the increasing appreciation of the various functions of these molecules. In membrane bilayer structure and function, sphingolipids, especially sphingomyelin and perhaps glycosphingolipids, are involved in organizing membranes into more rigid microdomains that include rafts and caveolae. These specialized membrane structures are increasingly appreciated to play critical roles in transmembrane signal transduction and in endocytosis. In addition, it is now well established that the levels of several sphingolipids, including sphingosine, ceramide, ceramide-1-phosphate, and S1P, are regulated and that these molecules function as bioactive lipids involved in various processes of signal transduction and cell regulation. Ceramide levels can be regulated by at least two pathways (Figure 7.2a). In the first pathway, neutral and acid sphingomyelinases respond to the action of various cytokines and stress signals, and their activation results in hydrolysis of sphingomyelin and the generation of ceramide. Alternatively, substantial evidence also shows that increase in ceramide levels can be induced through the activation of the de novo pathway of sphingolipid synthesis (Figure 7.2a). In both cases, the generated ceramide has been implicated in the regulation of a number of signaling pathways including activation of phosphatases and kinases that in turn couple the formation of ceramide to specific cellular responses (Figure 7.2a). For example, activation of protein phosphatase 1 causes dephosphorylation of the retinoblastoma gene product (Rb), resulting in cell cycle arrest. Multiple mechanisms induce a proapoptotic response to ceramide, and these include dephosphorylation of the antiapoptotic proteins Bcl-2, Akt, and protein kinase C. The regulation of ceramide and its functions have been extensively reviewed [4–7]. Likewise, S1P plays critical roles in cell signaling, primarily through interaction with specific high-affinity membrane receptors, the edg (or S1P) receptors [8,9]. Binding to these receptors initiates classical G-protein-coupled signaling responses (Figure 7.2b). These receptors are highly expressed on endothelial and mesenchymal cells and play critical roles in vasculogenesis and cell migration. Moreover, S1P appears to exert intracellular actions (through as yet unknown means) to induce antiapoptotic cell responses. Sphingosine and other sphingoid bases also play key roles in signaling, including regulation of apoptosis and inhibition of various signaling pathways, such as the protein kinase C pathway (Figure 7.2b). Their roles in cell regulation are best defined in yeast in which they play essential roles in the response and adaptation to heat stress and in the regulation of endocytosis [10,11]. Ceramide-1-phosphate has recently been shown to play roles in cytokine responses, especially in coupling cytokine action to the activation of phospholipase A2 (Figure 7.2b) and the induction of eicosanoids, as well as in vesicle formation and function. Therefore, at one level, the complex pathways of sphingolipid metabolism may be considered as a collection of specific pathways that regulate the levels of specific functional molecules (Figure 7.1).

Functional Lipidomics: Lessons and Examples from the Sphingolipids

151

Sphingomyelin

Serine + Palmitate Enzymes of de novo pathwaye

SMases Ceramide PKCζ

Raf PP2A

Cathepsin D

PP1

Akt; Bcl-2; C-jun; PKC

Rb; pSR

Bid

(a)

Sphingomyelin

Serine + Palmitate

DAG

SMases Enzymes of de novo pathway

SMS Ceramide

Cerebroside GCS

Ceramide kinase

Ceramide synthase

Ceramidase Lyase

SK C-1-P

Sphingosine

S-1-P

EtPi + Fatty aldehyde

S1Pase

cPLA2

Kinases/PLD

Edg/S1P receptors

(b)

FIGURE 7.2 Individual and interacting signaling pathways of bioactive lipids. (a) Two main pathways contribute to ceramide formation, the de novo pathway and activation of sphingomyelinases (SMases). The ceramide thus generated may activate protein phosphatases, PP1 and PP2A, a protease, cathepsin D, or protein kinases, Raf and protein kinase C . These in turn transmit a “ceramide signal” further downstream through modulation of specific targets. For example, cathepsin D may mediate proteolysis of the proapoptotic protein Bid. Activation of phosphatases results in dephosphorylation of a number of critical proteins including Rb, Akt, protein kinase C , and SR proteins. (b) Further metabolism of ceramide generates additional bioactive lipids such as DAG, C-1-P, sphingosine, and S1P. C-1-P activates cytosolic phospholipase A2 (cPLA2), whereas sphingosine regulates a number of protein kinases and other signaling molecules. S1P acts on transmembrane protein receptors that initiate pathways of G-protein signaling. Thus, pathways of sphingolipid metabolism create a network of bioactive lipids.

152

Functional Lipidomics

PATHWAYS OF SPHINGOLIPID METABOLISM AS A NETWORK OF BIOACTIVE LIPIDS BIOACTIVE LIPIDS REGULATE CELL FUNCTION: NOTION OF A MODULE In a broad definition of bioactivity, one may consider each lipid as a bioactive molecule either for its structural, metabolic, or regulatory functions. However, for the sake of this discussion, we will restrict the definition to those lipids that transduce specific cell responses through the modulation of their levels. As such, one can define the simplest module of lipid-mediated cell regulation (Figure 7.3a) as involving an enzyme that regulates the level of a bioactive lipid (e.g., sphingomyelinase regulating the levels of ceramide), the bioactive lipid itself (e.g., ceramide), and one or more targets that respond to changes in the levels of the bioactive lipid (e.g., protein phosphatases), as illustrated in Figure 7.2a. Thus, for a metabolic module to be functional, the requirements are: (1) the presence of a regulated enzyme that responds to specific stimuli (input) and (2) the presence of a target whose function is modulated by the product (and/or substrate) of the lipid pathway (output) (Figure 7.3a).

METABOLIC INTERCONVERSION OF TWO OR MORE BIOACTIVE LIPIDS: NOTION OF SWITCHES In contrast to the classical paradigm of a signaling module, such as the cyclase–cyclic AMP–protein kinase A module, which functions as a self-contained unit dedicated for a signaling function, modules of lipid signaling are often complicated by the presence of enzymes that act on those bioactive lipids [12]. For example, although the action of a sphingomyelinase may generate ceramide, the concomitant action of ceramidases would decrease ceramide levels while increasing the levels of sphingosine (Figure 7.2b). Thus, ceramidase activity may potentially convert a ceramide signal to a sphingosine one [13]. Likewise, sphingosine kinase may convert sphingosine functions to those of sphingosine phosphate [8], whereas the action of sphingomyelin synthase [14] decreases ceramide and increases the levels of diacylglycerol (Figure 7.2b), the prototypic bioactive glycerolipid (which acts primarily through activation of protein kinase C). Moreover, in most of these scenarios, the lipids involved in these interconversions exert distinct, and often opposing, effects; for example, S1P often acts to inhibit ceramide-mediated apoptosis. Thus, the enzymes involved in these reactions function in effect as switches in cell signaling and cell regulation (Figure 7.3b).

METABOLICALLY INTERCONNECTED BIOACTIVE LIPIDS: FOUNDATIONS FOR NETWORKS At an even more complex level, pathways of sphingolipid metabolism (and lipids in general) are highly connected. Thus, it is readily apparent that perturbations of these pathways (e.g., by changes in the activity of one of the enzymes) may initiate a cascade of changes in the levels of multiple lipids. For example, activation of sphingomyelinases may result in elevation in ceramide, then sphingosine, sphingosine phosphate, cerebroside, and diacylglycerol, and all this, through the action

Functional Lipidomics: Lessons and Examples from the Sphingolipids

153

Substrate 1

Target 1

Enzyme 1

Input 1

Output 1

Product 1 (a)

Substrate 1

Enzyme 1

Input 1

Product 1/ substrate 2

±

Enzyme 2

Target 2

Output 2

Product 2 (b)

FIGURE 7.3 Enzymes as switches. (a) In a simple module, an input causes a change in the activity of an enzyme (enzyme 1) that in turn changes the levels of a product (e.g., product 1). The presence of a target that senses changes in these levels then constitutes the output of the module. (b) When another enzyme (enzyme 2) acts on the product of the first enzyme to generate another product (product 2) with a distinct target, a distinct output is generated. As such, enzyme 2 acts as a switch in the pathway. For example, in Figure 7.2b, sphingomyelin synthase, ceramidase, ceramide kinase, and sphingosine kinase may function as switches.

of only those enzymes acting directly on the generated ceramide (Figure 7.2b). After that, changes in ceramide may lead to changes in almost any other sphingolipid or glycerolipid through direct metabolic connections (i.e., without invoking additional regulatory and feedback effects). Thus, the biochemical interconnectedness of lipids forms a network of pathways that regulate the levels of multiple bioactive lipids (Figure 7.1), each with its own important cell functions [3]. This presents a daunting problem in any experiment attempting to decipher the regulation and function of any specific signaling module involving bioactive lipids. For example, the addition of many sphingolipids (e.g., ceramide, sphingosine, S1P, or ceramide-1-phosphate) can result in the induction of eicosanoid production [15]. In a recent study using interference RNAi to silence the various enzymes that connect these lipids, we were able to implicate S1P as the key endogenous molecule responsible for mediating cytokine induction of cyclooxygenase and eicosanoid production [16].

154

Functional Lipidomics

Interestingly, one can begin to discern some specific properties of these networks of bioactive lipids. In system analysis of networks, scientists and mathematicians define two main architectural designs of such systems; individual nodes that connect directly to each other (regular network) or a hub-and-spoke architecture in which one or more nodes serve as central connections to several nodes (scale free) [17]. A simple connection map of bioactive sphingolipids and glycerolipids (Figure 7.4) demonstrates that whereas most bioactive lipids act as individual modules, both ceramide and diacylglycerol stand at the center of a sphingolipid and a glycerolipid hub, respectively.

FUNCTION

AS AN INTEGRATED

OUTPUT

Because of these metabolic considerations, coupled with the bioactivity of several of the lipids involved in these pathways, we have proposed the concept of integrated networks of lipid-mediated signaling [12]. According to this evolving concept, perturbation of the lipid network at any of its nodes (e.g., regulated enzymes) would result in multiple changes in various lipids and not just its direct substrate and product. Because many of these lipids exert specific cell functions, such perturbations would influence multiple pathways simultaneously. Therefore, to predict the effects of the activation of any one enzyme, one must adopt an integrative approach that S-1-P

NAC

GalCer

D

en

ov

o

Sp h

Ceramide

Glu

C1

P

FA Cer

PLs

SM

PGs

LPs

Ps

AG

G

M

TA

AA

PI

De novo

PA

DAG

LTs

FIGURE 7.4 Networking of bioactive sphingolipids and glycerolipids. Each pathway is represented by a spoke. Ceramide, DAG, and arachidonic acid (AA) act as hubs, whereas most other lipids may act as nodal points. Some of the nodes are connected by more than one metabolic pathway (e.g., sphingosine can be generated by one of several ceramidases that show distinct subcellular localization. (Sph, sphingosine; NAC, N-acyl-ceramide; TAG, triacylglycerol; SM, sphingomyelin; PGs, prostaglandins; LP, lipoxins; LTs, Leukotrienes; MAG, monoacylglycerol; PA, phosphatidic acid; PIPs, phosphatidylinositols; PL, phospholipids; and FA, fatty acids.)

Functional Lipidomics: Lessons and Examples from the Sphingolipids

155

incorporates the various components that are directly and indirectly affected through their metabolic interconnections.

SCOPE AND STATUS OF LIPIDOMICS The foundation of lipidomics is based on the global analysis of all the lipids of a system (e.g., a cell) in a specific phase or during a sequence of events. The analytical output of such an analysis has been referred to as the lipidome. Such analytical approaches have become feasible in the past few years due to significant methodological developments. Equally important, and as discussed earlier in the definition of lipidomics (Section titled Introduction and Historical Perspective), the scope of lipidomics includes the functional aspects of lipids. Accordingly, the overarching question becomes: What is the functional significance of a specific profile of all lipids (the lipidome) or of changes in such profiles? Dealing with this question requires: 1. Analysis of lipid levels (the lipidome) analogous to microarray profiling of gene expression or protein profiling in proteomics: This is now feasible with mass spectroscopic approaches, but given the scope of lipid metabolism and the huge numbers of distinct molecular species (as distinguished by mass spectroscopy), such an endeavor requires parceling this goal into smaller components for the examination of the various lipid subclasses and then reassembling those components (see chapter by E. Dennis). The collation and analysis of such huge data sets has also become possible with the development of specific mathematical models and software programs. 2. Creating mathematical models for lipid metabolism to enhance the prediction of changes in lipid profiles that arise in response to perturbations in specific components (e.g., enzyme activities or addition or loss of specific lipids): Along these lines, we have recently mapped pathways of sphingolipid metabolism in the yeast Saccharomyces cerevisiae and then developed a mathematical model that recapitulates (at first blush) sphingolipid metabolism in yeast [18]. This model was validated through experimentation whereby predictions from the mathematical simulations were tested in de novo designed experiments. The close concordance of the results served to validate the model and promises to allow more simplified approaches in predicting changes in lipid metabolism [19]. Thus, once the biochemical components and parameters that underlie a specific lipid network (or a subcomponent) are sufficiently known (including the regulation of enzymes and the kinetics governing fluxes), it is possible to predict changes in the lipidome starting from any given configuration, if the specifics of a perturbation are also known. 3. Determining or predicting the functional significance of changes in each of the examined lipids: The goal here is to be able to predict the biochemical and cellular consequences arising from specific changes in the levels

156

Functional Lipidomics

of specific bioactive lipids. For example, can one predict what the consequence would be of doubling the levels of ceramide over a period of 2 h in any given cell type under specified conditions? In effect, this is the answer being sought in the majority of the ongoing biochemical and biological investigations that aim at defining the mechanisms of action of specific bioactive lipids, downstream responses, and their roles in various aspects of cell biology. 4. Integrating the functional consequences that arise from the simultaneous (or coordinated) changes in lipids and their function: once any lipid profile is determined (goal 1) or predicted (goal 2), and if biochemical knowledge is sufficiently advanced to predict the biochemical and functional consequences of the changes in each bioactive lipid (goal 3), then it is theoretically possible to assemble all these pieces into an overall model that should predict the integrated sum of all of the biochemical and cellular responses that occur in response to changes in all the related bioactive lipids.

FUTURE DIRECTIONS Experimental research into lipids has developed sufficiently to allow inroads into integrative approaches (i.e., lipidomics). Mass spectroscopy methods increasingly provide large data sets on the levels of entire subclasses of lipids, and mathematical models of lipid metabolism have been established and validated at least in the yeast cell. Therefore lipidologists cannot ignore the power of these approaches or the potential significance of these results, and investigators stand to gain by incorporating these developing aspects of lipidomics into their understanding of lipid metabolism and function. On the other hand, this initial evolution of lipidomics equally underscores areas of major deficiencies in our understanding of lipid biology.

DEVELOPING MECHANISTIC INSIGHTS Developing comprehensive lipidomics approaches that incorporate function into the understanding of lipids necessitates a high degree of mechanistic understanding of the enzymes that act on lipids, their regulation, and their specific mechanisms. This is not a trivial goal as many of these proteins are hydrophobic integral membrane proteins that defy the usual methods of purification, crystallization, and in vitro biochemical study. In sphingolipids, the availability of the yeast model has been instrumental in the functional cloning of many of the eukaryotic enzymes of sphingolipid metabolism. Equally, there is a need to define the mechanisms of action of bioactive lipids. What are their direct targets? How do these targets transmit the signal? Many of these targets are unknown, but even when they are known, there is a paucity of information on their mechanisms of regulation and function. For example, S1P interacts with transmembrane receptors, but appears to also have a heretofore undefined intracellular target. Ceramide activates protein phosphatases and cathepsins; yet, how they interact in vivo remains to be defined.

Functional Lipidomics: Lessons and Examples from the Sphingolipids

157

Thus, at present, it is clear that integrative approaches (e.g., lipidomics) are here to stay, and it is equally clear that reductionist approaches provide the solid foundations on which integration is to be based, and this is in need of significant development.

COMPARTMENTALIZATION, SUBCELLULAR LOCALIZATION, TOPOLOGY, AND TRANSPORTERS: SIGNIFICANCE OF PHYSICAL PROPERTIES OF LIPIDS The physical properties of lipids impose several restrictions and add extra dimensions to the reconstruction of the metabolic pathways of lipid metabolism and lipidmediated function. Lipids are usually very hydrophobic molecules (most sterols, glycerolipids, and sphingolipids); however, several lipid subclasses include amphipathic lipids with variable aqueous solubility. These include sphingosine, S1P, other lysophospholipids, fatty acids, and various eicosanoids. The most soluble lipids readily transfer among membranes, whereas the more hydrophobic molecules are restricted to the membrane compartments in which they are produced unless specific mechanisms exist to transport these lipids between various compartments. For example, ceramide formed from the de novo pathway in the endoplasmic reticulum (ER) is restricted to that compartment, but can reach other compartments such as the Golgi through the action of specific transporters such as CERT, which delivers ER ceramide to the Golgi specifically for the synthesis of sphingomyelin. Ceramide may also reach the Golgi for the synthesis of cerebroside, probably in a mechanism involving vesicular trafficking. Ceramide may also reach other compartments such as the nuclear membrane or mitochondria through sites of direct contact, but this has not been demonstrated yet. This compartmentalization of lipid metabolism is further underscored by the recent recognition of compartment-specific enzymes of lipid metabolism. Again in the case of ceramide, distinct ceramidases have been localized to specific compartments (acid ceramidase to lysosomes, alkaline to the ER, another distinct alkaline to the Golgi, and a neutral ceramidase to the mitochondria and plasma membrane). In addition to compartmentalization, many lipids exhibit a distinct asymmetric distribution between the two leaflets of biological bilayers. This is perhaps best appreciated with sphingomyelin and complex glycolipids in the plasma membrane, in which they concentrate (although not exclusively) in the outer leaflet. These considerations on subcellular restrictions, topology, and biophysical properties have to be taken into account (at least eventually) in any attempt to reconstruct the networks of bioactive lipids and their functional significance. In conclusion, the future is very bright for the study of lipids and their function. This will increasingly incorporate basic biochemical studies (mechanisms), cell studies (localization and function), and integrative approaches.

ACKNOWLEDGMENTS I would like to thank Kellie Sims for the careful review of the manuscript and for helpful suggestions. This work was supported by NIH grant GM63625.

158

Functional Lipidomics

References 1. Sandhoff, K. and T. Kolter, Biosynthesis and degradation of mammalian glycosphingolipids. Philos Trans R Soc Lond B Biol Sci, 2003. 358(1433): pp. 847–861. 2. Merrill, A.H., Jr., De novo sphingolipid biosynthesis: a necessary, but dangerous, pathway. J Biol Chem, 2002. 277(29): pp. 25843–25846. 3. Hannun, Y.A. and L.M. Obeid, The Ceramide-centric universe of lipid-mediated cell regulation: stress encounters of the lipid kind. J Biol Chem, 2002. 277(29): pp. 25847–25850. 4. Ogretmen, B. and Y.A. Hannun, Biologically active sphingolipids in cancer pathogenesis and treatment. Nat Rev Cancer, 2004. 4(8): pp. 604–616. 5. Levade, T., et al., Ceramide in apoptosis: a revisited role. Neurochem Res, 2002. 27(78): pp. 601–607. 6. Bleicher, R.J. and M.C. Cabot, Glucosylceramide synthase and apoptosis. Biochim Biophys Acta, 2002. 1585(2-3): pp. 172–178. 7. Kester, M. and R. Kolesnick, Sphingolipids as therapeutics. Pharmacol Res, 2003. 47(5): pp. 365–371. 8. Maceyka, M., et al., Sphingosine kinase, sphingosine-1-phosphate, and apoptosis. Biochim Biophys Acta, 2002. 1585(2-3): pp. 193–201. 9. Kluk, M.J. and T. Hla, Signaling of sphingosine-1-phosphate via the S1P/EDGfamily of G-protein-coupled receptors. Biochim Biophys Acta, 2002. 1582(1-3): pp. 72–80. 10. Obeid, L.M. and Y.A. Hannun, Ceramide, stress, and a “LAG” in aging. Sci Aging Knowledge Environ, 2003. 2003(39): p. PE27. 11. Dickson, R.C. and R.L. Lester, Sphingolipid functions in Saccharomyces cerevisiae. Biochim Biophys Acta, 2002. 1583(1): pp. 13–25. 12. Hannun, Y.A., C. Luberto, and K.M. Argraves, Enzymes of sphingolipid metabolism: From modular to integrative signaling. Biochemistry, 2001. 40(16): pp. 4893–4903. 13. el Bawab, S., et al., Ceramidases in the regulation of ceramide levels and function. Subcell Biochem, 2002. 36: pp. 187–205. 14. Albi, E. and M.P. Viola Magni, The role of intranuclear lipids. Biol Cell, 2004. 96(8): pp. 657–667. 15. Ballou, L.R., et al., Interleukin-1-mediated PGE2 production and sphingomyelin metabolism. Evidence for the regulation of cyclooxygenase gene expression by sphingosine and ceramide. J Biol Chem, 1992. 267: pp. 20044–20050. 16. Pettus, B.J., et al., The sphingosine kinase 1/sphingosine-1-phosphate pathway mediates COX-2 induction and PGE2 production in response to TNF-alpha. Faseb J, 2003. 17(11): pp. 1411–1421. 17. Bray, D., Molecular networks: the top-down view. Science, 2003. 301(5641): pp. 1864–1865. 18. Alvarez-Vasquez, F., et al., Integration of kinetic information on yeas sphingolipid metabolism in dynamical pathway models. J Theor Biol, 2004. 226(3): pp. 265–291. 19. Alvarez-Vasquez, F., Sims, K.J., Cavart, L.A., Okamoto, Y., Voit, E.O., and Hannun, Y.A., (2005). Simulation and Validation of Modelled Sphingolipid Metabolism in Sacchaomeyces Cerevisiae. Nature 433: pp. 425–430.

8

Liquid ChromatographyTandem Mass Spectrometry (LC-MS/MS) Analysis of Sphingolipids M. Cameron Sullards and Alfred H. Merrill, Jr.

CONTENTS Abstract ..................................................................................................................160 Introduction............................................................................................................160 Biosynthesis and Turnover...........................................................................160 LC and MS...................................................................................................162 Materials and Methods ..........................................................................................165 Extraction .....................................................................................................165 LC.................................................................................................................166 MS and MS/MS ...........................................................................................167 Results and Discussion..........................................................................................167 MS ................................................................................................................167 Product, MS3, Precursor, and Neutral Loss Scans ......................................170 Product Ion Scan..............................................................................170 MS3 ...................................................................................................174 Precursor Ion Scan ...........................................................................175 Constant Neutral Loss Scans ...........................................................177 MRM ............................................................................................................177 Reverse-Phase Chromatography ..................................................................178 Normal-Phase Chromatography...................................................................182 Quantitation — Internal Standards ..............................................................183 Conclusions............................................................................................................185 Acknowledgments..................................................................................................186 References..............................................................................................................186

159

160

Functional Lipidomics

ABSTRACT Sphingolipids are among the most complex and structurally diverse lipids found in eukaryotes. This is a result of thousands of possible combinations of head group, fatty acid, and sphingoid base in addition to variations in chain length, substitution, or unsaturation of these moieties. Moreover, sphingolipids and their metabolites are highly bioactive compounds, which affect a wide variety of cellular functions including growth, differentiation, and apoptosis (programmed cell death) through involvement in signaling, membrane trafficking, or other processes. These cellular functions perturb either the de novo biosynthetic pathway or turnover of more complex sphingolipids, resulting in changes to either the type or quantity of endogenous sphingolipid species. Therefore, to determine which sphingolipids are being modulated, it is critical to precisely identify, quantify, and elucidate the structures of all species in these pathways. Liquid chromatography (LC) allows complex mixtures of sphingolipids from crude extracts to be trapped, focused, and selectively eluted prior to introduction to the mass spectrometer. This serves to minimize interferences from matrix as well as ionization suppression effects. Tandem mass spectrometry (MS/MS) uniquely identifies various types of sphingolipids by characteristic fragmentations of either the head group, fatty acid, or sphingoid base. This information is then used in multiple reaction monitoring (MRM) transitions to generate quantitative data, greatly increasing sensitivity and minimizing errors associated with other MS/MS techniques. Thus, LC used in conjunction with MS/MS provides a rapid, robust, sensitive, and highly specific way to determine the metabolic profile of sphingolipids in a given cellular state at a given time.

INTRODUCTION BIOSYNTHESIS

AND

TURNOVER

Sphingolipids, which were discovered by Thudichum [1], are among the most complex, structurally diverse, and enigmatic lipids found in nature. This high degree of structural variety is a direct result of the thousands of possible combinations of head groups, fatty acids, and sphingoid base backbones, which include variations in chain length, substitution, unsaturation, or branching in these moieties. Sphingolipids are highly bioactive species, which affect a wide variety of cellular functions including growth, differentiation, and apoptosis (programmed cell death) through modulation of membrane structure, trafficking, signaling, and other processes [2]. Changes in the amounts and types of cellular sphingolipids arise from both de novo biosynthesis as well as the turnover of more complex sphingolipids (Figure 8.1 and Figure 8.2). Therefore, to determine which sphingolipids are being perturbed from their basal state, it is critical to precisely identify, elucidate the structures of, and quantify all species in these pathways. Sphingolipids may be generally categorized as either simple or complex. Simple sphingolipids are commonly referred to as free sphingoid long-chain bases and contain the core structure, 2-amino-1, 3-dihydroxyoctadecane, or sphinganine

Liquid Chromatography-Tandem Mass Spectrometry

161

Palmitoyl-CoA + Serine

CH3(CH2)10CH 2

OH

3-Ketosphinganine

Sphinganine-1-phosphate CH2OH

NH3(+)

CH3(CH2)10CH 2

HO H

CH2OPO3(–2)

NH3(+)

4-Hydroxysphinganine CH3(CH2)10CH 2

HO H

HO H

CH2OH

NH

CH3(CH2)9-19*CH 2

Dihydro-ceramide

CH3(CH2)10CH 2

NH3(+)

Sphinganine

CH3(CH2)10CH 2

CH2OH

O

HO H

CH2OH

HO NH2

CH3(CH2)10CH 2 CH3(CH2)9-19*CH 2

Phytoceramide

HO H

CH2OH

HO NH O

FIGURE 8.1 (See color insert following Page 174.) Metabolic pathways giving rise to biosynthesis (green arrows) and turnover of simple sphingolipids sphinganine, sphingosine, and their associated 1-phosphate derivatives up to dihydroceramide.

(d18:0, in which d = dihydroxy, 18 = carbon chain length, and 0 = units of unsaturation) (Figure 8.1). This structure may be modified by having double bonds at various locations along the carbon chain. For example, sphingosine (d18:1Δ4, 4sphingenine, or sphing-4-enine), 8-sphingenine (d18:1Δ8), and 4, 8-sphingadiene (d18:2Δ4,8) have double bonds between C4–C5, C8–C9, and both C4–C5 and C8–C9, respectively. The core structure may also be modified via addition of a hydroxyl group at C4 yielding 4-hydroxy sphinganine (t18:0, in which t = trihydroxy), or it may contain both a double bond and an additional hydroxyl group giving rise to 4hydroxy-8-sphingenine (t18:1Δ8). Sphingoid bases can be modified by phosphorylation at C1 (e.g., sphingosine-1-phosphate) or methylation of the amino group. The majority of the sphingoid bases in biological samples are N-acylated with long-chain fatty acids (to produce ceramides) and conjugated with a polar head group at C1 (to produce complex sphingolipids). The fatty acids of ceramide vary in chain length (14 to 30 carbon atoms), degree of unsaturation (but are mostly saturated), and presence or absence of a hydroxyl group on the α- or ω-carbon atom. The head groups are mainly linked by phosphodiester bonds (e.g., sphingomyelins [SMs], ceramide phosphoethanolamines [CPEs], and ceramide phosphoinositols) and glycosidic bonds (glycosphingolipids). Glycosphingolipids are classified into broad types on the basis of carbohydrate composition. The neutral glycosphingolipids of mammals contain uncharged sugars such as glucose (Glc), galactose (Gal), N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), and fucose (Fuc) (and in other organisms, mannose and other sugars). Acidic glycosphingolipids

162

Functional Lipidomics

HO H

CH3(CH2)10CH2 CH3(CH2)9-19*CH2

CH2OH

NH O

Dihydroceramide

CH3(CH2)9-19*CH2

CH3(CH2)9-19*CH2 Phytoceramide

HO H

CH3(CH2)10CH2

CH3(CH2)10CH2

HO H HO

CH2OH

NH

O

Sphingomyelins, Glucosyl-& GalacatosylCeramides

CH2OH

NH O

Ceramide

Lacatosylceramides HO H

CH3(CH2)10CH2

CH2OH

NH2(+)

Sphingosine

Globosides

CH3(CH2)10CH2

HO H

Gangliosides

CH2OPO3(–2)

NH3(+) Sphingosine 1-phosphate

FIGURE 8.2 (See color insert following Page 174.) Metabolic pathways associated with biosynthesis (green arrows) and turnover of the complex sphingolipid ceramides, glucosyl/galactosylceramides, lactosylceramides, SM, globosides, and gangliosides.

contain ionized functional groups such as phosphate, sulfate (sulfatoglycosphingolipids), or charged carbohydrate residues such as sialic acid (N-acetylneuraminic acid) in gangliosides or glucuronic acid in some plant glycosphingolipids. Gangliosides are often denoted by the Svennerholm nomenclature that is based on the number of sialic acid residues (e.g., GM1 refers to a monosialo-ganglioside) and a number reflecting, in many instances, the relative position of the ganglioside upon thin-layer chromatography (for example, the Rf of GM3 > GM2 > GM1). A number of sphingolipids are referred to by their historic names as antigens and blood group structures, such as Forssman antigen (a pentosylceramide that is found in many mammals) and the Lewis blood group antigens, which correspond to a family of a13-fucosylated glycan structures (Lewis x, sialyl Lewis x, etc.).

LC

AND

MS

Because there are many sphingolipids that have the same elemental composition but different structures, LC (sometimes also called high-performance LC or HPLC) is a powerful analytical technique used for their separation prior to analysis by a

Liquid Chromatography-Tandem Mass Spectrometry

163

detection method such as mass spectrometry (MS). In most cases, LC involves the flow of a complex mixture of biomolecules in solution (mobile phase) through a column packed with small silica particles (the stationary phase). The surface of the particles may or may not be functionalized, depending on the mechanism to be used for partitioning between the mobile and stationary phases. Two of the more widely used are either based on hydrophobicity (reverse phase) [3,4] or hydrogen-bonding interactions (normal phase) [3,4]. Once applied to the column, the molecules then distribute themselves between the two phases according to their affinity toward either phase. As additional mobile phase is added, those molecules partitioned more into the mobile phase will be carried further along the length of the column than those molecules partitioned more into the stationary phase. Continued addition of mobile phase will further separate the mixture into discrete bands of molecules, which continue to traverse the column until they pass from the end of the column. The order of elution from the column occurs with regard to each species’ preference for the mobile versus stationary phase. The compounds of interest are detected in the eluate by a variety of methods. MS has long been used as a method of detection because it provides levels of sensitivity that are several orders of magnitude higher, allowing infinitesimal quantities of sphingolipids to be examined. In addition, MS yields a high degree of specificity with regard to identification of sphingolipids by molecular mass, especially when analyzed with high resolution and accurate mass measurement. Furthermore, when analyzed by MS/MS, intact molecular ions of sphingolipids may be induced to dissociate to yield structurally diagnostic fragment ions (Figure 8.3), which add another level of specificity. Furthermore, a system of nomenclature has been developed that describes these fingerprint fragmentations, and has been thoroughly reviewed elsewhere [5–8]. These fragmentations may in turn be used to uniquely identify specific species in complex mixtures at very low abundances. Finally, MS may be used to provide quantitative data when the proper ionization and detection parameters are used because the magnitude of the resulting signal generated may be correlated to analyte concentration. Coupling the separation and concentration power of LC with the sensitivity and specificity of molecular identification by MS has proven to be challenging. Early work, using dynamic fast atom bombardment (FAB) and liquid secondary ionization mass spectrometry (LSIMS), sought to continuously infuse a mixture of solvent and matrix onto a probe tip for subsequent ionization [9,10]. Although the matrix was diluted by a factor of ~100, background chemical noise arising from ionization of the matrix served to limit sensitivity of analyte detection. Additionally, the rate of sample infusion was limited to 10 μ l min1, owing to the restrictions of the instrument vacuum system and its ability to remove solvent while maintaining high vacuum. This dictated that eluents from HPLC flowing at 200 μl min1 had to be split, greatly reducing the amount of sample being introduced and further limiting sensitivity. The emergence of electrospray ionization (ESI) has directly addressed the limitations in coupling HPLC to MS by generating intact molecular ions of sphingolipids directly from solution without chemical modification [11,12]. Therefore, direct information regarding the molecular mass of a wide variety of sphingolipids may be readily obtained. Because ESI is a soft ionization technique with little fragmentation,

164

Functional Lipidomics

Z0 Y0

OH

HO

HO

1,5

HO

5

4

R' HN

C2

R"

B2 O

OH 2,5A

C1

B1

2

1 2

1

3

0

3

O

O

O

OH

X2 O 0

OH

5

Z1 4

Y1

2

OH O

OH

N

P

R

S R

OH

T R'

R'

O HN

R''

U R''

HN

C

O E O

V

O

FIGURE 8.3 Nomenclature for cleavages of complex sphingolipids via MS/MS. R = H (Cer), glucose (GlcCer), galactose (GalCer), lactose (LacCer), phosphoethanolamine ceramide phosphoethanolamine (CPE), phophorylcholine (SM), etc. R’ = -CH=CH(CH2)nCH3 where n = 12 for d18:1 sphingoid bases, and R” = -(CH2)nCH3 where typically n = 14–22.

MS/MS may be employed to determine the structures of various species. Furthermore, the signal response is proportional to analyte concentration in ESI, so it may be used to generate quantitative data. The ESI process involves infusing an analyte in solution into an ion source consisting of an inner metalized needle held at a high potential (1 to 6 kV) surrounded by an outer needle and is positioned a few millimeters from the grounded inlet orifice of the mass spectrometer. The high potential causes formation of highly charged droplets at the needle tip, which contain both solvent and analyte. The potential difference between the needle and the orifice serves to draw the charged droplets into the mass spectrometer. In the transition from atmospheric pressure in the ion source to high vacuum inside the mass spectrometer, the charged droplets undergo rapid desolvation. In this process, neutral solvent molecules are pumped away, leaving the analyte bearing the charge, resulting in its soft ionization. This process results in greatly reduced background chemical noise from solvent ions and yields sensitivity several orders of magnitude higher than was possible with FAB. Interfacing eluents from HPLC columns with an ESI ion source requires some modifications. This is because typical ESI flow rates are ~1 to 20 μl min−1, and multiple stages of vacuum pumping are necessary to remove excess solvent molecules while retaining high vacuum. Increased flow rates of ~100 μl min−1 can be

Liquid Chromatography-Tandem Mass Spectrometry

165

attained by addition of a cocurrent, dry nitrogen bath gas flow through the outer needle surrounding the inner needle to aid in droplet formation, dispersal, and desolvation. Additional increases in solvent flow rates of up to 1.0 ml min−1 into the mass spectrometer can be achieved by heating of the ion source, bath gas, or both to further aid the desolvation process. Thus, eluents from HPLC columns may be directly introduced to the ion source of the mass spectrometer, yielding rapid ionization and mass analysis of compounds separated from complex biological mixtures. LC, coupled with single-stage mass analysis, can show great improvement with regard to sensitivity and specificity for the identification and quantitation of sphingolipids vs. classical methods of analysis. However, it should be noted that these methods still possess limitations, and structural information is not readily available. Although background chemical noise is limited in ESI, solvent ions as well as other coeluting species may interfere with detection of sphingolipids of interest. This is particularly true at lower m/z ratios at which solvent ions predominate, and the longchain base sphingolipids are typically observed. Interference may also arise from isotopic contributions from ions having a slightly different molecular mass but possessing an isotope having the same molecular mass as the analyte of interest and, thus, give rise to a false signal. Isobaric or isomeric compounds may also yield anomalous results in single-stage mass analysis of sphingolipids by their having the same mass but different structure vs. the analyte of interest. LC, used in conjunction with tandem mass spectrometry (LC-MS/MS), directly addresses the specificity and structure limitations associated with analysis of sphingolipids via LC-MS. Here MS/MS (product ion scan) is used to identify highly abundant, structure-specific fragmentations associated with the various sphingolipid species. Next, both ionization and dissociation conditions are optimized to yield maximum precursor and fragment ion formation for each individual molecular species. After that, either a precursor ion scan or a neutral loss scan is performed to confirm that the identified fragment ion uniquely identifies the molecular ion of interest. The instrument is then set up so that the first mass analyzer passes only the ion of interest, and the second mass analyzer passes to the detector only the fragment ion arising from the molecule of interest. Thus, only those molecular species that have the appropriate retention time, molecular mass, and fragment to yield the product ion of interest will be detected. The additional stage of selectivity provided by MS/MS serves to greatly reduce background chemical noise and thereby increase sensitivity.

MATERIALS AND METHODS EXTRACTION Mass spectral analysis of sphingolipids in biological samples can be complicated by the presence of other molecules. These other biomolecules may preclude the identification, structure elucidation, or quantitation of sphingolipids via isobaric, isomeric, or isotopic interferences. They may also suppress ionization or be present in large molar excess, making measurements on very small quantities of sphingolipids difficult owing to background chemical noise. Therefore, it is highly desirable

166

Functional Lipidomics

to remove interfering species without altering or modifying the endogenous sphingolipids. To this end, crude biological mixtures of sphingolipids can be extracted by a variety of established methods [13]. For example, cells, tissue, or other biological materials may typically be suspended in a 2:1 mixture of CH3OH and CHCl3. The biomaterial is then sonicated, homogenized, or otherwise uniformly broken down and exposed to the organic liquid phase to promote maximal solubility and removal of the endogenous lipid material. If quantitative analyses are to be performed, the sample is spiked with a known quantity of internal standard for each species to be quantitated. Bulk lipid and protein that are also coextracted in this organic mixture are removed by base hydrolysis, which is performed by addition of a few hundred microliters of 1-M KOH in CH3OH. This solution is sonicated for a few seconds to ensure distribution of the base and then incubated for 2 h at 37˚C in a shaking water bath to promote hydrolysis. After base hydrolysis, an aliquot consisting of approximately 50% of the sample volume may be removed and dried under vacuum for subsequent long-chain base sphingolipid analysis. The remaining solution is then neutralized via addition of a few microliters of glacial acetic acid and dried under vacuum. Once dry, the sphingolipids may be reconstituted in a small volume of a 2:1 mixture of H2O and CHCl3 and sonicated or vigorously shaken. Afterwards, this solution is centrifuged at 4000 g for 10 min, and the upper water layer is then carefully removed, leaving behind the CHCl3 layer and the interface having a small volume of H2O. This bottom layer is also dried under vacuum and will be used for subsequent complex sphingolipid analysis.

LC Free long-chain sphingoid bases sphingosine, sphinganine, phytosphingosine, analogous sphingoid base 1-phosphate, and chain-length-variant standards were analyzed via reverse-phase chromatography. Here, a 2.1 × 50 mm Supelco Discovery C18 column, packed with 5-um particles, was used at a flow rate of 1.0 ml min–1. The eluent from the HPLC column was directly connected to the ABI 4000 QTrap for subsequent dispersal via 6.0 L min–1 nitrogen bath gas flow at 500˚C. Mobile phase A consisted of 74:25:1 CH3OH/H2O/HCOOH (v/v/v) and mobile phase B was 99:1 CH3OH/HCOOH. A mixture of the free long-chain sphingoid bases containing C17 sphingosine-1-phosphate, phytosphingosine, sphingosine, sphinganine, sphingosine1-phosphate, phytosphingosine-1-phosphate, C20 sphingosine, and C20 sphinganine was made from stock solutions such that the final concentration of each was on the order of 100 to 300 fmol μl –1 in 80:20 A/B. The elution protocol consists of 0.5min column preequilibration with 80:20 A/B (v/v), followed by 20-μl sample injection, a 0.6-min sample load and wash on column with 80:20 A/B (v/v), a 1.8-min linear gradient to 100% B, and a 0.6-min hold at 100% B. Afterwards, a 0.3-min postrun column equilibration with 80:20 A/B (v/v) was performed. Complex sphingolipid extracts from Drosophila, containing Cer, HexCer, and CPEs, were analyzed via reverse-phase chromatography. Here a 2.1 × 150 mm Supelco Discovery C18 column packed with 5-um particles was used at a flow rate of 0.500 ml min–1. The eluent from the HPLC column was directly connected to the API 3000 for subsequent dispersal and ionization, using the same dry nitrogen bath

Liquid Chromatography-Tandem Mass Spectrometry

167

gas flow rate and temperature listed earlier. Mobile phase A consisted of 58:41:1 CH3OH/H2O/H3CCOOH (v/v/v), and mobile phase B was 99:1 CH3OH/H3CCOOH with both solutions having a 5-mM ammonium acetate buffer. The dried complex sphingolipid–containing fraction was reconstituted in a 25:75 A/B mixture of the mobile phases with 150 μl of B added first followed by 50 μl of A; this was then centrifuged to pellet any insoluble material. The elution protocol consists of 1.5min column preequilibration with 25:75 A/B (v/v), followed by 20-μl sample injection, a 3.0-min sample load and wash on column with 25:75 A/B (v/v), a 5.0-min linear gradient to 100% B, and a 10.0-min hold at 100% B. Afterwards, a 1.5-min postrun column equilibration with 25:75 A/B (v/v) was performed. Complex sphingolipid standards C12 ceramide, glucosylceramide, SM, and lactosylceramide were separated via normal-phase chromatography. Here, a 2.1 × 50 mm Supelco amino column was used at a flow rate of 1.5 ml min1. As in the longchain base analysis, the column eluent was connected directly to the ion source of the 4000 QTrap. Mobile phase A consisted of 97:2:1 CH3CN/CH3OH/CH3COOH (v/v/v), and mobile phase B was 99:1 CH3OH/CH3COOH (v/v); both contained 5mM ammonium acetate. The complex-sphingolipid-containing fraction was reconstituted in 200 μl of mobile phase A and centrifuged to pellet any insoluble material. The elution protocol consists of 0.5-min column preequilibration with 100% A, followed by 10-μl sample injection, a 0.5-min column load and wash, 0.2-min linear gradient to 90:10 A/B (v/v), 0.5-min hold, 0.4-min linear gradient to 82:18 A/B (v/v), 0.6-min hold, 0.4-min linear gradient to 100% B. Afterwards, a 0.3-min postrun column equilibration with 100% A was performed.

MS

AND

MS/MS

For the purposes of this chapter, discussions and examples will involve the use of ESI and either an API 3000 triple quadrupole or a unique hybrid quadrupole–linear ion trap tandem mass spectrometer, the ABI 4000 Q Trap. The latter instrument will be used to demonstrate several mass spectral scanning techniques with regard to their role in identification and structure elucidation of sphingolipids prior to analysis via LC-MS/MS. This instrument was chosen because of the versatility of tandem mass spectrometric methods such as neutral loss, precursor, and product ion scans, which can be performed using this instrument. HPLC-MS/MS methods utilizing MRM for the analysis of both simple and complex sphingolipids will be discussed with regard to high throughput and automation. Additionally, data for enhanced resolution of higher-order glycosphingolipids, enhanced product ion analysis, and more complete structure elucidation using MS/MS/MS (MS3) will be presented.

RESULTS AND DISCUSSION MS A simple mass spectrum of several long-chain base sphingolipids acquired with nominal resolution displays species that have been volatilized, ionized, and separated by their mass-to-charge (m/z) ratios (Figure 8.4a). This mass spectrum can provide

168

Functional Lipidomics

some clues as to the sphingolipids that are present. However, interferences from other species present, such as salts, can either suppress ionization of the sphingolipids or form adducts that complicate the spectrum. Problems also arise when examining sphingolipids at low relative abundance (Figure 8.4b). Here, the background signal generated from either the ESI process or other species may be of much greater abundance (chemical noise) than that of the sphingolipids; therefore, their detection may precluded. Long-chain bases and complex sphingolipids readily ionize via positive ion ESI to form primarily (M + H)+ ions (Figure 8.4a). Some sphingolipids such as the sphingoid base-1-phosphates, SM, sulfatides, and gangliosides may also form (M – H) ions via negative ion ESI. Sphingolipid molecular ions are generally detected at m/z 250 to 450 for long-chain bases, m/z 450 to 700 for Cer’s, m/z 650 to 900 for GlcCer’s and SMs, and m/z 800 to 1050 for LacCer’s. The higher-order glycosphingolipids such as gangliosides may be observed at m/z 1200 to 3000 depending on the degree of glycosylation and charge state. However, if multiply charged, the m/z values will of course be lower (Figure 8.5a), and enhanced-

Relative ion abundance

Sa 302.4 So 300.4

C20So 328.4

Phyto Sa 318.3

C20Sa 330.4 So-1-P 380.4

(a)

Relative ion abundance

Sa 302.4 So 300.4

275

C20So 328.4 C20Sa 330.4

Phyto Sa 318.3

300

325

350

So-1-P 380.4

375

400

m/z (b)

FIGURE 8.4 MS analysis of a mixture of free long-chain base sphingolipids that consumed approximately (a) 70 pmol and (b) 0.70 pmol, respectively.

Liquid Chromatography-Tandem Mass Spectrometry

169

resolution scans will be required for isotopic resolution, so that charge state as well as neutral mass may be determined (Figure 8.5b). The difference in sensitivity between the quadrupole and ion-trapping modes was observed to be quite striking. In this case, the ion-trapping mode yielded approximately 25% of the absolute ion intensity for the m/z 917.5 species vs. the quadrupole mode (1.8 × 107 and 7.5 × 107, respectively). However, the enhancedresolution spectrum was acquired using a 1/20 dilution of the solution used to acquire the quadrupole spectrum so as to prevent overfilling the trap and yielding poorly resolved peaks. Over 160 × more sample was required to obtain a low-resolution spectrum via quadrupole scanning. Given the number of scans acquired and the associated scan times, it was calculated that this roughly translates to almost an order of magnitude enhancement in sensitivity for ion-trapping vs. quadrupole modes for this ganglioside species. 932.0

Relative ion abundance

918.0

7.5 e7

925.5 940.0 946.0 953.0

904.0 890

910

930

950

970

990

m/z (a) 917.5

1.8 e7

Relative ion abundance

917.9 918.5 918.9

910

914

918

922

926

930

m/z (b)

FIGURE 8.5 MS analysis of ganglioside GD1a via (a) Q1 scan (~9 pmol consumed), and (b) enhanced-resolution scan (~0.055 pmol consumed).

170

Functional Lipidomics

Other instruments capable of high resolution and accurate mass measurement can determine the molecular mass of a species to within a few parts per million (ppm). Such mass accuracy can provide information with regard to the type and number of atoms present (empirical formula), which yields the degree of unsaturation (rings and double bonds) in a molecule. However, it is important to note that a mass spectrum generated using a soft ionization technique such as ESI primarily provides information regarding molecular mass. Little information regarding structure or quantitation may be accurately determined from a single stage of mass analysis. Thus, the usefulness of MS for the analysis of sphingolipids in biological samples is limited because they are often present in very low quantities (nmol to pmol in 1 to 5 × 106 cells in culture) and are often buried in the background chemical noise.

PRODUCT, MS3, PRECURSOR,

AND

NEUTRAL LOSS SCANS

MS/MS may be generally defined as relating to ions that, once initially formed in the ion source, subsequently fragment during their transit to the detector. The resulting fragment ions provide important clues as to the structure and reactivity of the intact molecular ion. Fragment ions may be formed either by metastable dissociation (MD) or by collision-induced dissociation (CID). Metastable dissociations arise via unimolecular decomposition, which means that molecular ion decomposition occurs from internal energy inherent to ions when they are formed. Fragment ions that arise by this process are generally considered low energy, revealing cleavage of the weakest chemical bonds. CID arises from a bimolecular decomposition reaction in which a portion of translational energy is converted into internal energy via collisions between analyte ions in motion and neutral gas molecules. The amount of energy deposited, and therefore the degree of fragmentation, can be controlled by changing either the velocity of the ions or the mass of the target gas. Product Ion Scan In a product ion scan, the first mass analyzer (MS1 or, in this case, Q1) is set to pass a single ion of interest (m/z). This precursor ion is transmitted to a second quadrupole (Q2), which is filled with a neutral gas, such as N2 or Ar, and operated in an Rf-only mode to pass all ions. Here, the precursor ions undergo multiple collisions with the target gas and are induced to decompose. Subsequently formed product ions are transmitted to MS2 (Q3), which is scanned across a range of m/z values, passing fragment ions sequentially to the detector, thus constituting a second stage of mass analysis. The resulting tandem mass spectrum shows the fragmentation pattern of the selected precursor ions and yields structural information in the form of both product ions detected and neutral species lost. The relative ion abundances of the product ions detected are reflective of the kinetics of the various dissociation pathways and vary with collision energy, which, in this case, is typically the potential difference between Q1 and Q2. MS/MS of the long-chain bases such as So (d18:1), Sa (d18:0), Phyto Sa (t18:0), C20 So (d20:1), and C20 Sa (d20:0) show that they fragment via single and double dehydration to yield product ions of m/z 282/264, 284/266, 282/264, 310/292, and

Liquid Chromatography-Tandem Mass Spectrometry

171

312/294, respectively (Figure 8.6). The single-dehydration products are much more abundant than the double-dehydration products over a range of collision energies. The free long-chain base-1-phosphates derivatives also undergo a similar dehydration and cleavage of the head group, yielding the same m/z product ions already described. It should be noted that although different free sphingoid base species such as So and SoP fragment to yield the same m/z product ions, they are distinguished by their precursor ion mass and retention time on HPLC. Product ion analysis of the (M + H)+ ions of Cer’s reveals that these ions undergo cleavage of the amide bond and dehydration of the sphingoid base to form highly abundant, structurally specific O” fragment ions (Figure 8.3) [14]. These product ions are characteristic of the sphingoid base, and their m/z yields information regarding the number of carbon atoms in the chain, degree of hydroxylation, unsaturation, or other structural modifications of the long-chain base. Therefore, these types of complex sphingolipids contain commonly occurring sphingoid bases, such as So, Sa, phyto Sa, C20 So, and C20 Sa, which give rise to product ions of m/z 264, 266, 264, 292, and 294, respectively. Product ion scans of the (M + H)+ ions of GlcCer’s and LacCer’s reveal that these ions undergo dissociation via two pathways. At low collision energies, bond cleavage occurs at the glycosidic linkages. The sugar head group is lost as a neutral species with charge retention remaining on the ceramide moiety forming the Yn/Zntype (where n = 0, 1) ions (Figure 8.7a and Figure 8.7b). At higher collision energies, OR

OH

H+

NH2 –H2O OR +

NH2 –HOR

+

NH2 R = H, H2PO3

FIGURE 8.6 Major route of dissociation of free sphingoid bases So and Sa occurs via single and double dehydration, and 1-phosphate species fragment similarly.

Functional Lipidomics

Relative ion abundance

172

N" 264.4

Z0 464.4

Z0' 446.5

–H2O 626.6 (M+H)+ 644.6

N' 282.3

200

400

600

800

m/z

Relative ion abundance

(a)

Z0 464.4 N'' 264.4

(M+H)+ 807.0

Z0' 446.5

–H2O 788.6

N' 282.3

200

Z1 626.6

400

600

800

m/z (b)

FIGURE 8.7 Product ion spectra of (a) C12 glucosylceramide at high energy and (b) C12 lactosylceramide at low energy.

both the sugar head group and the fatty acid acyl chain are cleaved, and the charge is retained on the dehydrated sphingoid base yielding predominantly N” ions. (Note: The N” ions are structurally identical to the O” ions from ceramides. The difference in nomenclature is attributed to the GlcCer and LacCer having a head group other than a hydrogen atom.) Sphingolipids containing phosphodiester-linked head groups, such as in SM, fragment in a much different manner. In these (M + H)+ species, cleavage occurs at the phosphate–ceramide bond with charge retention on the phospho-head group resulting in highly abundant C ions of m/z 184 being formed [14]. A similar phosphate–ceramide bond cleavage occurs in CPE. However, the head group is lost as a neutral species of mass 141 u. This difference in charge retention is most likely the result of the difference in gas-phase basicity between the primary nitrogen and quaternary nitrogen of the phosphoethanolamine and phosphocholine, respectively.

Liquid Chromatography-Tandem Mass Spectrometry

173

1.4e6

Intensity, cps

1.2e6

C1β – H2O 290.2

1.0e6 8.0e5 6.0e5 4.0e5

Y0 564.7

Y4α/Y2β 1254.3

2.0e5 200

1.6e8

Intensity, cps

1.4e8

600

Y2β 1545.5

Y2α/Y2β 890.0

1000 m/z (a)

1400

1800

C1β – H2O 290.1

1.2e8 1.0e8 8.0e7 6.0e7 4.0e7

Y0 564.6

Y2α/Y2β 888.8 Y1 726.6

Y2β 1544.8 Y4α/Y2β 1253.8

2.0e7 200

600

1000 m/z (b)

1400

1800

FIGURE 8.8 MS/MS spectra of (M – 2H)2– ions (m/z 917.5) of ganglioside GD1a acquired using 4000 QTrap (a) product ion scan in triple quadrupole mode and (b) enhanced product ion scan in ion-trapping mode.

Higher-order glycosphingolipids, such as gangliosides, (M – 2H)2– ions fragment in triple quadrupole analyses to yield highly abundant C1β – H2O ions of m/z 290, indicative of the sialic acid residue (Figure 8.8a). Interestingly, enhanced product ion scans using the ion-trapping function on the 4000 Q Trap provides much better sensitivity and more abundant high-mass product ions than are observed in the triple quadrupole mode (Figure 8.8b). Other prominent fragment ions observed arise via cleavages at glycosidic bond linkages to yield characteristic Yn-type ions (Figure 8.8b and Figure 8.9). Furthermore, product ions that arise through ring cleavages such as 2,4X2α and 2,4X3α (m/z 1323.8 and 1484.8, respectively) are also observed (though not highlighted) and are useful for determination of glycosidic bond linkage.

174

Functional Lipidomics

Y1

Y2β HO HO AcHN

OH

COOH O

O O

OH

O O

Y2α

2–

Y0 HO

OH

O O

O HO

OH

Ceramide

OH 2,4X NHAc 2α

HO HO

O

2,4 OH

X3α

O

Y3α

HO

HO HO

O

NHAc OH

O COOH OH

Y4α

OH

FIGURE 8.9 Structure and position of glycosidic bond cleavages of fragment ions observed in product ion spectra of the (M – 2H)2– ions of ganglioside GD1a (m/z 917.5). (Note that several combinations of positions may be deprotonated to give rise to the (M–2H)2− species.)

MS3 An MS3 scan is performed in much the same way as a product ion scan. Again the first mass analyzer (MS1 or, in this case, Q1) is set to pass a selected ion of interest (m/z). This precursor ion is transmitted to a second quadrupole (Q2), which is filled with a neutral gas, such as N2 or Ar, and operated in an Rf-only mode. The precursor ions undergo multiple collisions with the target gas and are induced to decompose. Subsequently formed product ions are transmitted to MS2, which in this case is the linear ion trap (LIT). The LIT is set to trap and hold a 2-m/z unitwide window centered on the product ion of interest. The LIT then irradiates the trapped ions with a single wavelength, amplitude, and frequency to induce further fragmentation. The subsequently formed secondary product ions are then scanned out of the LIT to the detector for a third stage of mass analysis. The resulting MS3 spectrum shows the fragmentation pattern of the selected product ion and yields additional structural details regarding the primary product ion that may not be observed in an MS/MS spectrum. The relative ion abundances of the secondary product ions detected are reflective of the kinetics of the various dissociation pathways and vary with the amplitude of the Rf and the length of time for which it is applied. MS3 scans provide critical structural information regarding sphingolipids such as the gangliosides. As mentioned earlier, the (M – 2H)2– ions fragment primarily at glycosidic bonds, which give rise to highly detailed information regarding the carbohydrate head group of the glycosphingolipid. The MS/MS spectra, however, do not reveal any information about the sphingoid base or fatty acid moieties in the ceramide portion of the molecule. MS3 analyses of the Y0 product ions (m/z 564.6), which comprise the core lipid part of the molecule, are necessary to determine the composition of the ceramide. Here, highly abundant S, T, U, and V + 16 ions (m/z 324, 308, 282, and 283, respectively) demonstrate that the fatty acid is C18:0 (Figure 8.3 and Figure 8.10). The complementary P and Q ions (m/z 237 and 263, respec-

Liquid Chromatography-Tandem Mass Spectrometry

175

Relative ion abundance

Y0 564.6 U 282.3

V + 16 283.3

Q 263.2

T 308.3 S 324.3

P 237.2

200

300

400 m/z

–H2CO 534.6

516.5

500

600

FIGURE 8.10 MS3 spectrum of the core lipid Y0 ions (m/z 564.6) arising from fragmentation of ganglioside GD1a (M – 2H)2– ions (m/z 917.5).

tively), which are characteristic of the sphingoid base, indicate that their composition is d18:1 (Figure 8.3 and Figure 8.10). These results clearly demonstrate that MS3 scans provide an additional level of structural analysis, yielding critical information regarding sphingoid base, fatty acid, and head groups in sphingolipids. Precursor Ion Scan In a precursor ion scan, the second mass analyzer (MS2 or, in this case, Q3) is set to pass an m/z value that corresponds to a structure-specific product ion. Precursor ions are sequentially allowed to enter the collision chamber (Q2) by scanning MS1 across a range of m/z values. Subsequently formed product ions are transmitted to MS2 (Q3). However, only those precursor ions that decompose to yield the specified product ion of interest will be transmitted by Q3 to the detector. This serves to greatly reduce background chemical noise and identify specific molecular species by structure in complex mixtures. Precursor ion scans of structure-specific fragmentations are highly useful for the identification of free sphingoid bases at low analyte concentrations. For example, precursor ion scans specific for m/z 264.3 clearly reveal the presence of So-1-P at low concentration, which was not observed in single-MS scans (Figure 8.11 and Figure 8.4B, respectively). Precursor ion scans of m/z 282/284 and m/z 264/266 reveal that the single-dehydration products yield approximately an order of magnitude more signal and suppress chemical noise to a greater extent than the doubledehydration products. Optimization of ionization and dissociation conditions for the free sphingoid bases revealed some interesting points regarding their accurate quantitation. First, the free sphingoid bases were not as susceptible to dehydration in the ABI 4000 QTrap as in the API 3000 triple quadrupole. Increased declustering potentials could be used in the QTrap without fragmentation, yielding improved sensitivity. Second, species containing a Δ4 double bond again yield more abundant dehydration products than do saturated species of similar concentration. It is hypoth-

176

Functional Lipidomics

Relative ion abundance

So-1-P 380.6

275

So 300.6 Phyto So 318.4

300

325

350

375

400

m/z

FIGURE 8.11 Precursor ion spectrum of those ions that fragment to yield product ions of m/z 264.3 over the range m/z 275–400.

esized that the difference in reaction rate is a result of dehydration allylic to the double bond, resulting in the formation of a stable conjugated carbocation (Figure 8.6). Owing to this difference in dissociation rates, internal standards for both the saturated and unsaturated species are necessary for accurate quantitation of both. Additionally, the saturated species (Sa/C20Sa) yielded a prominent fragment ion of m/z 60 at higher collision energies, which was not as sensitive, but more specific than the single-dehydration product. New conditions were determined for optimization of ionization and dissociation conditions for Cer’s, HexCer’s, and LacCer’s using the ABI 4000 QTrap. It was observed that the ceramides, unlike the free sphingoid bases, were still susceptible to in-source dehydration, whereas the HexCer’s and LacCer’s were not. Declustering potentials must still be kept low for the ceramide species, so that decomposition of (M + H)+ ions prior to entry to the mass spectrometer will not occur, and accurate quantitative data will result. All species containing Δ4 double bonds show much more “O” and N” ion abundance (m/z 264 for d18:1) than do the fully saturated species (m/z 266 for d18:0) at similar concentrations. Thus, internal standards for both the saturated and unsaturated species are necessary so that they may be quantitated accurately. Quantitative information regarding multiple species of sphingolipids (i.e., all chain-length variants of Cer’s, HexCer’s, SM’s, etc.) using precursor ion scans is complicated by multiple factors. First, ionization suppression could play a key role because a wide variety and quantity of molecules are present during charged droplet formation and desolvation. The heterogeneous nature of the solution may alter the extent to which a given analyte of interest will be ionized, and may yield a much different signal response than if it were ionized as a homogeneous species. Second, the kinetic differences in the rates of dissociation result in different signal responses for varied molecular species. For example, at a given collision energy, only a few similarly sized ions will fragment optimally. Ions that have fewer atoms, and thus fewer degrees of freedom, will have more collision energy per degree of freedom

Liquid Chromatography-Tandem Mass Spectrometry

177

and may undergo additional fragmentation. This is in contrast to larger ions, which have more atoms and more degrees of freedom, and will have less collision energy per degree of freedom. These ions may, therefore, fragment less efficiently, yielding lower-abundance product ions, and thus, reduced signal response in the precursor ion scan. Finally, precursor ion scans are very inefficient with regard to duty cycle, or the amount of instrument time spent detecting ions of interest. For example, sphingolipids often comprise approximately 10 distinct individual molecular species in a given 200 u mass range. A precursor ion scan over this mass range will result in the instrument spending only 5% of scan time detecting ions of interest. This yields poor ion statistics and, therefore, reduced sensitivity, which greatly limits confidence in any quantitative data acquired in this manner. Constant Neutral Loss Scans In a constant neutral scan, the first and second mass analyzers (MS1 and MS2, or in this case, Q1 and Q3) are scanned together. However the m/z ratios that either one will pass are offset from one another by a fixed m/z value corresponding to the mass of a structure-specific loss of a neutral molecule. Only those precursor ions that decompose, losing the correct mass fragments as a neutral, will be passed to the detector. This also has the effect of identifying specific molecular species in complex mixtures by their structures and greatly reducing background chemical noise. However, it should be noted that constant neutral loss scans, such as precursor ion scans, have limitations with regard to their use for quantitation, which include ionization suppression, kinetics of dissociation, and sensitivity.

MRM In MRM analyses, the first mass analyzer (Q1) is set to pass a specific precursor ion m/z, and the second mass analyzer is set to pass a specific product ion m/z. Only those ions that meet both precursor and product ion m/z conditions simultaneously will be transmitted to the detector. After a user-defined period of time, detection of this m/z transition from precursor to product ion is halted, a new set of precursor and product ion m/z’s may be set, and signal detection for this transition is begun. This process can continue onto other transitions or cycles repeatedly. Thus, it is possible to monitor multiple transitions corresponding to numerous analytes very rapidly on an HPLC time frame. Using this technique, the amount of instrument time spent detecting ions of interest is greatly increased. The mass spectrometer now resides on the proper precursor and product ion m/z’s rather than scanning broad regions, which contain no ions of interest. The signal detected is greatly enhanced relative to either precursor ion or neutral loss scans and provides lower limits of detection. The resulting data possess a high degree of specificity with regard to unequivocal identification in that individual molecular species must meet three criteria: proper retention time, mass, and structure. Furthermore, sensitivity can be further enhanced by optimization of ionization and dissociation conditions for each individual molecular species’ m/z transition pair. Thus, MRM transitions, used in conjunction with LC and internal standards, provide more accurate quantitative data

178

Functional Lipidomics

as a result of addressing critical issues with regard to ionization suppression, kinetics of ion dissociation, and sensitivity.

REVERSE-PHASE CHROMATOGRAPHY Reverse-phase chromatography utilizes hydrophobic interactions as the basis for separating complex mixtures of molecules. Here, the stationary phase has a hydrophobic surface typically composed of C18 covalently linked to a silica particle support. The mobile phase is less hydrophobic, typically containing an aqueous–organic solvent, which serves to solubilize the analyte and prevent the collapse of the hydrophobic stationary phase. Upon introduction of a complex sample mixture onto the column, molecules partition into and out of the stationary phase. As more mobile phase is passed over the column, the more hydrophilic compounds are carried further along the column, whereas the hydrophobic compounds are more retained. Continued addition of mobile phase further separates the mixture into discrete bands, which continue to traverse the column until they pass from the end of the column. Polar or hydrophilic compounds elute first, followed by compounds with increasing organic nature or hydrophobicity. Sphingolipids may be eluted either isocratically [15] or by a gradient [14,16]. In an isocratic elution, mobile phase is continuously added at a constant aqueous–organic ratio. Species that elute early exhibit sharp, focused peaks, whereas more hydrophobic compounds have much greater retention, elute later, and give rise to more broad and diffuse peaks. This phenomenon can be addressed by use of a gradient elution, whereby the ratio of the aqueous–organic solvent changes over time so that the mobile phase becomes more organic. As the mobile phase added to the column is increasingly organic, the more hydrophobic compounds begin to partition more into the mobile phase. This serves to elute the lipids from the column as more sharp, focused bands. Resolution, peak shape, and retention time in a gradient HPLC run are affected by multiple factors, including the rate of change of aqueous–organic solvent ratio in the mobile phase. Analysis of free long-chain base sphingolipids can be performed using reversephase LC-MS/MS, which provides several advantages vs. other methods. First, crude mixtures of free sphingoid bases may be loaded and retained well, as a result of their hydrophobic tail interaction with the stationary phase, even in solutions having up to 80% organic solvent. This allows the column and sample to be washed thoroughly, which serves to remove contaminating salts and other polar interferences. Second, individual molecular species are separated by hydrophobicity, so species that vary by chain length, hydroxylation, or unsaturation will have different retention times. Moreover, when used in conjunction with MS/MS, each long-chain base can be identified by their unique MRM transitions. These precursor–product ion m/z pairs are optimized with regard to both ionization and dissociation conditions using the 4000 Q Trap (Table 8.1) providing a high degree of sensitivity and specificity. The reverse-phase HPLC analysis of several free sphingoid base standards reveals several points regarding their separation. Here, the extracted ion chromatograms of each individual molecular species show that the compounds elute in the order of carbon chain length from least to most C17 < C18 < C20 compounds (Figure 8.12). Interestingly,

Liquid Chromatography-Tandem Mass Spectrometry

179

TABLE 8.1 Summary of Precursor–Product Ion m/z’s, and Associated Collision Energy, Declustering Potential, and Dwell Time Used for MRM Detection of Individual Molecular Species of Free Sphingoid Bases Free Sphingoid Bases So Sa Phyto Sa c20 So c20 Sa c17SoP SoP Phyto SaP

Precursor–Product Ion m/z 300.3–282.3 302.3–60.00 318.3–300.3 328.4–310.3 330.3–312.3 366.4–250.4 380.3–264.4 398.4–300.4

Collision Energy (eV) 18.0 45.0 22.0 20.0 23.0 23.0 25.0 22.0

Declustering Potential (eV) 40.0 50.0 50.0 40.0 50.0 50.0 50.0 50.0

Dwell Time (msec) 25 25 25 25 25 25 25 25

Relative ion abundance

C20 So

So Phyto So

C20 Sa Sa

C17So-1-P 1

1.25

So-1-P Phyto So-1-P

1.5 Time (min)

1.75

2

FIGURE 8.12 (See color insert following Page 174.) Reverse-phase HPLC–MS/MS–extracted ion chromatograms of free long-chain base sphingolipid standards C17 sphingosine-1-phosphate (dark red), phytosphingosine (green), sphingosine (black), sphinganine (red), sphingosine-1-phosphate (purple), phytosphingosine-1-phosphate (black), C20 sphingosine (blue), and C20 sphinganine (brown).

So and Sa differ only by a Δ4 double bond, yet they are baseline-resolved chromatographically, as are C20So and C20Sa (Figure 8.12). The phytosphingoid bases differ from sphingosine in that water is added across the Δ4 double bond, fully saturating the sphingoid base backbone and resulting in a third hydroxyl group on carbon 4. The phytosphingosine is now observed to be much more hydrophilic than Sa and elutes slightly ahead of So. This is in contrast to phytosphingosine-1-phosphate, which possesses an additional hydroxyl group, yet elutes slightly later than So-1-P. Additionally, it is observed that So-1-P, although more hydrophilic than So, is retained more strongly and elutes much later than So. This raises the possibility that the mechanism of retention of the phosphorylated compounds is driven by more than hydrophobicity,

180

Functional Lipidomics

and the phosphate groups may be interacting with the silica surface. However, it should be noted that the phosphate-containing species still exhibit sharp, nontailing peaks, which is often not the case when using other HPLC columns. The mass spectral analysis via LC-MS/MS also provides several points of interest in the mass spectral analysis of the free sphingoid bases. First, it is of analytical importance to note that an adequate number of data points are taken across each eluting peak. Here the dwell time is 25 msec per transition with a 5-msec intertransition delay yielding a 0.240 sec cycle time total. The LC peak widths of each sphingolipid are ~6 sec wide, resulting in approximately 24 data acquisition points across each analyte of interest. Second, most transitions are unique, which means that although some species are not chromatographically resolved and their masses are similar, the second stage of mass analysis is sufficient to distinguish individual species without interference. Only one transition appears to generate a false signal, which corresponds to the C20Sa. This transition yields two peaks, one occurring simultaneously with the elution of C20So, and another afterwards (Figure 8.12, brown trace). The second peak is consistent with the elution pattern of So and Sa and is the true C20Sa signal. The false signal is generated because this transition differs from C20So in both the precursor and product ion transition by only 2 m/z units, so there may be isotopic contribution from the eluting C20So into the C20Sa transition. The natural isotopic contribution of 13C in the C20So ion is ~2.7%, which closely approximates the ratio of the area of this peak relative to the area of C20So. The isotopic contribution is easily eliminated by choosing a product ion transition of m/z 60, similar to that found for Sa, which does not display this behavior. Quantitation may still be performed in either case as the two peaks observed for the C20Sa are baseline-resolved, and there is no overlap of their peak areas. Reverse-phase chromatography may also be used to separate mixtures of complex sphingolipids. In this case, the samples are loaded onto the column in an even more organic mobile phase (typically only 10% H2O) because these species are retained by the column much more strongly even at very high organic solvent content. This is a result of these lipids containing two hydrophobic tails vs. one found in the long-chain base species, thereby greatly increasing hydrophobic interactions. Therefore, long retention times and increasingly broad and diffuse peaks for complex sphingolipids containing longer sphingoid base or fatty acid chains are expected. This effect may be lessened by use of either a less hydrophobic stationary phase such as C8 or stronger organic solvents such as ethanol or isopropanol. However, use of the latter two solvents will result in greatly increased column backpressure as a result of increased solvent viscosity. Additional complications arise from complex sphingolipids containing multiple polar functional groups such as hydroxyls or phosphates. These moieties have a high affinity for alkali metal ions such as sodium and potassium, which may cause ionization suppression or formation of adducts greatly reducing sensitivity. Therefore, the complex sphingolipids should be initially washed thoroughly to remove these interferences. LC-MS/MS analysis of a mixture of the complex sphingolipids extracted from Drosophila Melongaster cells provides some insights into the elution of Cer, CPE, and HexCer species via reverse-phase chromatography (Table 8.2). As expected,

Liquid Chromatography-Tandem Mass Spectrometry

181

TABLE 8.2 Summary of Precursor–Product Ion m/z’s and Associated Collision Energy, Orifice Potential, and Dwell Time for MRM Detection of Individual Molecular Species of Complex Sphingolipids from Drosophila Containing a d14:1 Sphingoid Base Precursor–Product ion (m/z) 538.6–208.3 566.6–208.3

Collision Energy (eV) 40.0 44.0

Orifice Potential (eV) 30.0 30.0

Dwell Time (msec) 20 20

GlcCer C20:0 C22:0

700.6–208.3 728.8–208.3

45.0 45.0

50.0 50.0

20 20

CPE C20:0 C22:0

689.5–548.3 723.5–582.3

35.0 35.0

50.0 50.0

20 20

Cer C20:0 C22:0

C20: 0

d14: 1

Relative ion abundance

Cer HexCer

CPE C22: 0 HexCer CPE

9

10

11 12 Time (min)

Cer

13

14

FIGURE 8.13 (See color insert following Page 174.) Reverse-phase HPLC-MS/MS–extracted ion chromatograms of the d14:1 complex sphingolipids monohexosylceramide (red), CPE (green), and ceramide (black) extracted from Drosophila cells.

species containing shorter lipid tails (d14:1 / C20:0) elute before species containing longer lipid tails (d14:1/C22:0) for all complex sphingolipids (Figure 8.13). Interestingly, the head group attached to the core lipid affects the retention and subsequent elution of these complex sphingolipids as well. In this case, it is observed that addition of the carbohydrate head group yielded the greatest decrease in hydrophobicity and resulted in HexCer species (red trace) eluting before either CPE (green trace) or Cer species (black trace) containing similar core lipid structures (Figure 8.13). However, the phosphoethanolamine head group was only slightly more hydro-

182

Functional Lipidomics

philic than Cer, resulting in its elution between similar HexCer and Cer species (Figure 8.13). Quantitation of complex sphingolipids using reverse-phase chromatography is complicated by at least two factors, ionization efficiency and suppression. In the former case, the ability of a sphingolipid species to become ionized can be greatly influenced by its solubility or its gas-phase basicity relative to the composition of the solvent it is in. Under reverse-phase HPLC conditions, closely related complex sphingolipid species such as C16 and C24 ceramides, which differ by their carbon chain length, will be in markedly different solvent compositions at the time of their elution and subsequent ionization. Thus, the conditions in which one species may be ionized can yield a much different signal response from the other, when both may be present in equimolar quantities. Furthermore, this difference in elution time may also lead to additional complications as later-eluting peaks become more broad and diffuse. Another related factor to consider is suppression of ionization resulting from coeluting species. Here, other molecular species may be present in much greater abundance or may possess a much greater gas-phase basicity relative to the analyte of interest. These species may therefore be ionized preferentially, either greatly reducing or eliminating the ionization of the analyte of interest. Therefore, it is highly desirable to have species widely different in either abundance or basicity separated chromatographically from the analyte of interest to avoid these issues.

NORMAL-PHASE CHROMATOGRAPHY Normal-phase chromatography provides a complementary means of separating complex mixtures of biomolecules. In this case, the mechanism of separation can be described by hydrogen-bonding interactions between functional groups on the analyte and the stationary phase. Typically, the stationary phase has a hydrophilic surface composed primarily of the free hydroxyl groups on the silica particle support. However, these may be modified to other functional groups such as primary amines to enhance specific interactions. The mobile phase is more hydrophobic, consisting primarily of an organic solvent, which may serve to solubilize the analyte, and upon introduction of a complex sample mixture onto the column, allow more polar molecules to interact with the surface of the stationary phase. As more mobile phase is passed over the column, more hydrophobic compounds have little interaction with the stationary phase and are carried further along the column, whereas compounds having functional groups capable of hydrogen bonding are more strongly retained. Continued addition of mobile phase will further separate the mixture, based on the strength and number of interactions, into discrete bands, which continue to traverse the column until they pass from its end. Nonpolar or hydrophobic compounds elute first, followed by compounds with increasing polarity or hydrophilicity. Analysis of complex sphingolipids can be performed using normal-phase LCMS/MS, which provides several advantages vs. reverse-phase process. First, use of normal-phase HPLC allows crude mixtures to be loaded onto column in organic solvents in which the lipids are readily soluble, thus avoiding sample losses and reduced sensitivity resulting from poor solubility. Second, interferences arising from the formation of adducts of alkali metal salts are reduced as these are not as soluble

Liquid Chromatography-Tandem Mass Spectrometry

183

in some organic solvents such as methanol. Most important, complex sphingolipids elute by head group, which means that all chain lengths within a given class of sphingolipids elute simultaneously. This allows all individual molecular species within a given class (in addition to any internal standard) to be ionized under identical solvent composition conditions, avoiding ionization efficiency issues. Here, the use of MRM transitions is of critical importance as this allows rapid detection of many individual precursor–product ion m/z pairs, which have been optimized with regard to both ionization and dissociation conditions. The resulting data provides a high level of both sensitivity and specificity that may be used for quantitation. Additionally, this process may be repeated for classes of complex sphingolipids eluted later. Therefore, dozens of complex sphingolipids may be quantitated in a single normalphase LC-MS/MS run. This yields a more comprehensive and detailed view of the flux of molecules through either biosynthetic or turnover pathways. Normal-phase LC-MS/MS analysis of a mixture of C12 sphingolipid standards demonstrates the unique aspects of this type of analysis. First, it is observed that the lipids elute in the order Cer, GlcCer, SM, and LacCer (Figure 8.14), and reveals the increasing affinity of each molecule toward the amino stationary phase. Interestingly, SM is retained more than GlcCer. This may indicate that interaction between the phosphate and the amino stationary phase is much stronger than the multiple interactions between the carbohydrate hydroxyls and the amino stationary phase. Addition of a galactose to the GlcCer to form the LacCer serves to further increase the number of hydroxyl interactions to the point that it is now retained more strongly than SM. Second, it is also observed that the relative ion intensity decreases with retention. This may be the result of decreasing ionization efficiency of each lipid in the particular solvent composition required for elution. Finally, the time required for analysis is very rapid at ~2 min compared to previous LC-MS/MS methods, which required ~7 to 20 min per run [13,14,16]. This rapid LC-MS/MS method, when used in conjunction with an autosampler, allows automated, high-throughput analyses of dozens of complex sphingolipids from a large number of samples in a short period of time. Thus, the flux of individual molecular species of complex sphingolipids may be followed through various metabolic signaling pathways.

QUANTITATION — INTERNAL STANDARDS Accurate and precise quantitation of either simple or complex sphingolipids requires the use of internal standards to control sample losses in extraction, differences in chromatographic retention, ionization efficiency, and fragmentation. The most ideal standard sphingolipids would be stable-isotope-labeled species, which would possess nearly identical physical and chemical properties as the lipids of interest. Unfortunately, difficulties in the synthesis, position of the label, and obtaining a large enough mass shift make standards of this type either very expensive or not commercially available. In the absence of stable-isotope-labeled compounds, it is critical to select an internal standard that most closely approximates the various aspects of the analyte’s chromatography, ionization, and dissociation. This will serve to minimize potential sources of error and yield more accurate and precise quantitative data.

184

Functional Lipidomics

Relative ion abundance

Cer

GlcCer SM LacCer 0

0.5

1

1.5

2

Time (min)

FIGURE 8.14 (See color insert following Page 174.) Normal-phase HPLC-MS/MS–extracted ion chromatograms of the complex sphingolipid C12 standards ceramide (black), glucosylceramide (red), SM (green), and lactosylceramide (blue).

Long-chain sphingoid base quantitation is typically performed using reversephase LC-MS/MS. Here, the early-eluting C17 So/Sa-1-P and later-eluting C20 So/Sa are used as internal standards for the quantitation of So/Sa-1-P and So/Sa, respectively (As noted in the product and precursor ion analyses of these molecules, the presence or absence of the Δ4 double bond yields much different signal responses for these species, necessitating separate internal standards). Quantitation of the desired sphingolipid is performed by simply setting up a ratio of the areas under the peak of the unknown over the internal standard and multiplying by the quantity of standard added. It should also be noted that the internal standards closely resemble the analytes in mass, structure, and fragmentation behavior. Slight differences in retention do result from the shorter- and longer-chain-length species and may affect quantitative accuracy. However, in the absence of isotopically labeled standards, these species were chosen so as to bracket the analytes of interest with regard to retention on HPLC and polarity in extraction. Complex sphingolipid quantitation is performed in an analogous fashion using normal-phase LC-MS/MS. Here, short-chain C12 analogs of Cer, GlcCer, LacCer, and SM are used as internal standards for quantitation. As occurred with the free long-chain bases, separate internal standards are required for quantitation of Cer, GlcCer, and LacCer species in which the sphingoid base is fully saturated. However, this is not necessary for SM species as they dissociate via a different mechanism, which is not affected by the presence or absence of the Δ4 double bond. In this case, quantitation of each individual chain-length variant within a given class of sphingolipids is performed by simply setting up a ratio of the areas under the peak of the unknown over the internal standard and multiplying by the quantity of standard added. The total quantity in a given class would therefore be the sum of all the individual molecular species detected. In this case, the internal standards closely approximate the analytes in retention time, mass, and structure, lending a significant degree of confidence to quantitative accuracy. This accuracy is further enhanced

Liquid Chromatography-Tandem Mass Spectrometry

185

when the quantity of internal standard added is within ±100-fold of the analyte concentration, so that ionization suppression of either the standard or analyte does not result.

CONCLUSIONS MS/MS provides highly sensitive and structurally specific information regarding sphingolipids. Protonated free sphingoid bases such as So, Sa, and their analogous 1-phosphate derivatives fragment to yield product ions characteristic of the sphingoid base. Similarly, (M + H)+ ions of complex sphingolipids such as Cer, GlcCer, and LacCer also fragment to form these structurally distinctive product ions. For commonly observed mammalian sphingoid bases having either d18:1 or d18:0 sphingoid bases, these product ions are observed at m/z 264 or m/z 266, respectively. These m/z values may be observed to shift in species in which modifications occur in the sphingoid base chain length, degree of unsaturation, hydroxylation, or other characteristics. Additionally, it has been determined that the presence of the Δ4 double bond greatly affects the signal response in the formation of these product ions. Therefore, to obtain accurate quantitation of both the saturated and unsaturated species, internal standards for each type of sphingoid base are necessary. SMs, CPEs, and the gangliosides are unique in that they fragment to yield other structurally distinctive species. For (M + H)+ ions of SM, these product ions are characteristic of the phosphorylcholine head group and are detected at m/z 184. This is in contrast to CPE, which dissociates via loss of the phosphoethanolamine head group as a neutral species (141 u) with the charge remaining on the ceramide portion of the molecule. Gangliosides are typically analyzed in the negative mode and give rise to deprotonated species such as (M – H)–, (M2H)2–, or (M3H)3–. These ions are observed to dissociate to form highly abundant ions of m/z 290 indicative of the sialic acid moiety. Structurally diagnostic fragmentations allow use of either precursor ion or neutral loss scans to identify sphingolipids in complex mixtures with a high degree of specificity and sensitivity. In these scanning techniques, only those ions that fragment to yield the prescribed product ion or neutral loss may be transmitted to the detector. This serves to greatly reduce background chemical noise, thus enhancing detection of specific sphingolipid species at very low concentrations. Furthermore, these scans can reveal the exact combinations of head groups, sphingoid base, and fatty acid by deduction given the intact mass and the structure-specific fragment ions. Unfortunately, these types of scans do not provide accurate quantitative data owing to issues regarding ionization, dissociation, and instrument duty cycle. Quantitation of sphingolipids, however, can be accomplished via use of MRM transitions. Here, both ionization and dissociation conditions are optimized for each individual molecular species’ precursor–product ion pair. Maximal duty cycle and, therefore, sensitivity is obtained by merely detecting these transition pairs rather than scanning wide m/z ranges. Furthermore, the resulting signal response for each individual molecular species is not biased with regard to instrumental parameters. LC may be used in conjunction with MS/MS for obtaining unambiguous data with regard to both structure elucidation and quantitation of dozens of metabolically

186

Functional Lipidomics

active sphingolipids. Here, LC is used to concentrate and separate complex mixtures of sphingolipids into discrete bands based on either hydrophobic or hydrogenbonding interactions. The eluents are then directed into the mass spectrometer via an ESI source, which serves to disperse and remove neutral solvent molecules and generate intact molecular ions of the sphingolipids. MRM transitions using optimal ionization and dissociation conditions are used to uniquely identify each individual sphingolipid species. Quantitative information regarding the sphingolipids may then be obtained by comparison of the areas of the peaks of unknowns with the areas of judiciously chosen internal standards. The result is a structure-specific, quantitative metabolic profile of dynamically interconverting sphingolipid species. These LC-MS/MS techniques are not without limitations, however. One of these is the inability to distinguish the origin of key intermediate sphingolipids as arising from either biosynthesis or turnover. For example, if Cer is observed to be elevated under a given set of conditions, is this a result of an increase in de novo biosynthesis or turnover of more complex sphingolipids? MS may be a powerful analytical tool to differentiate these two sources. Methods are currently under development whereby an isotopically labeled precursor such as uniformly carbon-13-labeled (U-[13C]) palmitic acid may be introduced exogenously. Upon cellular uptake, its incorporation into the sphingoid base backbone and/or fatty acid would yield a distinctive m/z shift in precursor and/or product ions observed indicating species produced via de novo biosynthesis.

ACKNOWLEDGMENTS We acknowledge the Georgia Research Alliance for acquisition of the API 3000 LCMS/MS triple quadrupole mass spectrometer, and the NIH and Lipid MAPS (GM 69338) for acquisition of the ABI 4000 QTrap hybrid quadrupole–ion trap tandem mass spectrometer. We also would like to thank Prof. Joe Goldstein and Iriana Dobrosotskaya for the donation of the Drosophila cells.

References 1. Thudichum, J.L.W. A Treatise on the Chemical Constitution of Brain, Bailliere, Tindall, and Cox, London, 1884. 2. Merrill, A.H., Jr., Liotta, D.C., and Riley, R.T. Lipid second messengers. in Handbook of Lipid Research, Bell, R.M., Exton, J.H., and Prescott, J.M., Eds.; Plenum Press: New York, 1996; Vol. 8, 205–237. 3. Gazzotti, G., Sonnino, S., Ghidoni, R., Kirschner, G., and Tettamanti, G. analytical and preperative high-performance liquid chromatography of gangliosides. J. Neurosci. Res. 1984, 12, 179–192. 4. Palestini, P., Masserini, M., Sonnino, S., Giuliani, A.;, and Tettamanti, G. Changes in the ceramide composition of rat forebrain gangliosides with age. J. Neurochem. 1990, 54(1), 230–235. 5. Domon, B. and Costello, C.E. Structure elucidation of glycosphingolipids and gangliosides using high-performance tandem mass spectrometry. Biochem. 1988, 27, 1534–1543.

Liquid Chromatography-Tandem Mass Spectrometry

187

6. Domon, B., Vath, J.E., and Costello, C.E. Analysis of Derivatized Ceramides and Neutral Glycosphingolipids by High-Performance Tandem Mass Spectrometry. Anal. Biochem. 1990, 184, 151–164. 7. Adams, J. and Ann, Q. Structure determination of sphingolipids by mass spectrometry. Mass Spectrom. Rev. 1993, 12, 51–85. 8. Ann, Q. and Adams, J. Structure determination of ceramides and neutral glycosphingolipids by collisional activation of (M+Li)+ Ions. J. Am. Soc. Mass Spectrom. 1992, 3, 260–263. 9. Suzuki, M., Sekine, M., Yamakawa, T., and Suzuki, A. High-performance liquid chromatography-mass spectrometry of glycosphingolipids: I. Structural characterization of molecular species of GlcCer and IV3 bGal-Gb4Cer. J. Biochem. 1989, 105, 829–833. 10. Suzuki, M., Yamakawa, T., Suzuki, A. High-performance liquid chromatographymass spectrometry of glycosphingolipids: II. Application to neutral glycolipids and monosialogangliosides. J. Biochem. 1990, 108, 92–98. 11. Mano, N., Oda, Y., Yamada, K., Asakawa, N., and Katayama, K. Simultaneous quantitative determination method for sphingolipid metabolites by liquid chromatography/ionspray ionization tandem mass spectrometry. Anal. Biochem. 1997, 244, 291–300. 12. Gu, M., Kerwin, J.L., Watts, J.D., and Aebersold, R. Ceramide profiling of complex lipid mixtures by electrospray ionization mass spectrometry. Anal. Biochem. 1997, 244, 347–356. 13. Sullards, M.C., and Merrill, A.H., Jr. Analysis of Sphingosine-1-phosphate, Ceramides, and Other Bioactive Sphingolipids by High-Performance Liquid Chromatography-Tandem Mass Spectrometry. Sci. STKE 2001. http://stke.sciencemag.org/cgi/ content/full/OC_sigtrans:2001/67/pl1. 14. Sullards, M.C. Analysis of sphingomyelin, glucosylceramide, ceramide, sphingosine, and sphingosine-1-phosphate by tandem mass spectrometry. in Methods of Enzymology, Merrill, A.H., Jr., Hannun, Y.A.; Academic Press: San Diego, 2000, Vol. 312, 32–45. 15. Merrill, A.H., Jr., Wang, E., Mullins, R.E., Jamison, W.C.L., Nimkar, S., Liotta, D.C. Quantitation of free sphingosine in liver by high-performance liquid chromatography. Anal. Biochem. 1988, 171, 373–381. 16. Sullards, M.C., Wang, E., Peng, Q., and Merrill, A. H., Jr. Metabolomic profiling of sphingolipids in human glioma cell lines by liquid chromatography tandem mass spectrometry. Cell Mol. Biol. 2003, 49(5), 789–797.

9

Methods of Probing Phosphoinositides– Protein Interactions Li Feng, Colin Ferguson, Paul O. Neilsen, Leena Chakravarty, Piotr W. Rzepecki, and Glenn D. Prestwich

CONTENTS Introduction to Phosphoinositide Signaling ..........................................................189 Modification Strategies of Phosphoinositide Probes ............................................191 Cellular Studies with Phosphoinositide Probes ....................................................194 Mimicry of Physiological Effects of PIPn in Cells .....................................194 Visualization of PIPn in Cells ......................................................................196 Development of Nonradioactive Lipid Assays......................................................197 In Vitro Assays .............................................................................................197 Cell-Based Assays........................................................................................200 Probing Protein–Phosphoinositide Interactions Using Immobilized Lipids ........202 Closing Remarks....................................................................................................204 References..............................................................................................................205

INTRODUCTION TO PHOSPHOINOSITIDE SIGNALING Over the past two decades, phosphoinositides have emerged as an important class of phospholipids that modulate diverse cellular functions [1–4]. The eight subspecies of phosphoinositides (Figure 9.1) are generated and interconverted by phosphorylation and dephosphorylation at the D3, D4, and D5 hydroxyls of the inositol head group by specific kinases and phosphatases [5–7]. The phosphoinositide (PIPn)signaling networks, essential elements in the tyrosine kinase growth factor receptor and G-protein-receptor signaling pathways [8,9], are dynamically modulated by proteins with lipid recognition, kinase, phosphatase, and phospholipase activities. Activation of these cellular signaling pathways often results from specific PIPn

189

190

Functional Lipidomics

−O O P O O

O

O

O

O

O

OH

O O

PI

R2

OH P

O O

O R1

PI(3)P

R2

−O O P O O

O O O

O

O R1

−O

R1

PI(4)P

R2

OPO32− 2− OPO3 OH HO

HO

HO

OH

−O O P O O

OPO32− OH OH

OH OH 2− OPO3 OH 2− OH OPO3 HO

OH OH OH

HO

OH

−O O

O

O

OH P

O

R1

PI(5)P

R2

O

O

O R2

PI(4,5)P2

R1

HO

OH

−O O P O O

O O

O R1

OPO32− OPO 32− OPO32− OH OPO32− HO OPO32−

OPO32− OPO32−

−O O P O O

O O O

O

O

OH HO

R2

PI(3,4)P2

OH

−O O

O

O

O O O

O

O

OH P

O O

O R1

R2

PI(3,5)P2

R1

R2

PI(3,4,5)P3

FIGURE 9.1 (See color insert following Page 174.) Eight scaffolds of phosphatidylinositol. Naturally occurring phosphoinositides have acyl chains (R1, R2) of varied lengths and saturation degrees. The acyl chains are primary sites for chemical modifications.

production in response to a stimulus and, in turn, recruits an assemblage of proteins, now referred to as a signalosome, to a given subcellular location to initiate a cascade of signaling events [10,11]. PIPn have important roles in membrane traffic [12,13], including endocytosis, exocytosis, Golgi vesicle movement and protein trafficking [14,15], in cell adhesion and migration, in remodeling of the actin cytoskeleton, and in mitogenesis and oncogenesis [16–18]. Despite the deceptively minor variation of the phosphorylation state of the 8 subspecies of the PIPn, each specific PIPn has quite distinctive and specific roles in signaling pathways, in the control of cytoskeletal architecture [1], and in membrane trafficking for a given cell-type [14]. A prominent example of PIPn signaling is the signaling role of PI(3,4,5)P3, the cellular level of which is modulated by two pivotal enzymes, phosphatidylinositol 3-kinase (PI 3-K) and the 3-phosphatase PTEN. PI 3-K is implicated in numerous signaling systems that convey information from mammalian cell-surface receptors including growth factors, oncoproteins, and nonmitogenic stimuli [19]. In response to growth factor stimulation, tyrosine kinase receptors activate PI 3-K to catalyze the formation of PI(3,4,5)P3 via phosphorylation of PI(4,5)P2 [3,4,8,20]. The transient increase of PI(3,4,5)P3 induces the formation of defined molecular complexes. Through binding to their PH domains, PI(3,4,5)P3 recruits downstream effector enzymes such as Akt/protein kinase B (PKB), phosphoinositide-dependent kinase 1 (PDK1), and Bruton’s tyrosine kinase 1(Btk1) to the plasma membrane. Activation of PKB/Akt [9,13–15] suppresses apoptosis and promotes cell survival. The activity of PI 3-Ks and regulation of the level of PI(3,4,5)P3 is often defective in tumorigenesis [16,20]. Elevated PI 3-K levels have been observed in some cancers [17], and experiments have indicated that cellular transformation is PI 3-K dependent [18]. The importance of the PI 3-K pathway in tumor progression [21] makes it a potential target in drug development [22,23]. The levels of PI(3,4,5)P3 are negatively regulated by PTEN. The PTEN tumor suppressor gene (phosphatase and tensin homologue deleted on chromosome ten) is among the most frequently mutated genes in high-grade brain tumors. Loss of PTEN activity results in accumulation of PI(3,4,5)P3 [24], abnormal activation of PKB/Akt, suppression of apoptosis [25,26], and increased tumorigenesis in a number of human tissues [25,26]. Alterations in the enzymes that dysregulate PI(3,4,5)P3 levels are associated with a variety of cancer types [27–29]. The PIPn 5-phosphatase SHIP2 is involved in insulin signaling and diabetes [30,31]. SHIP2 appears to be

Methods of Probing Phosphoinositides– Protein Interactions

191

an essential negative regulator of insulin signaling and sensitivity, and altered SHIP2 activity may be a contributing factor to the insulin resistance associated with NIDDM and obesity. Moreover, SHIP2 — as with PI 3-K — is also a crucial regulator of immune function [32–34]. The PIPns exert their effect as signaling molecules and second messengers by directing protein translocation and the formation of macromolecular-signaling complexes at specific subcellular locations [35]. This is mediated by the presence of a number of highly conserved modular protein domains that bind to various phosphoinositides [11]. A variety of PIPn-binding motifs have been identified and characterized, including the PH, PX, FYVE, C2, Dix, and ENTH domains, as well as short, basic, amino-acid rich sequences [12,36,37]. The structural basis for PH domain discrimination of the head group has been explored by NMR and x-ray crystallographic studies by the Lietzke et al. (Grp1) [38], Ferguson et al. (Grp1, DAPP1/PHISH) [39], and Baraldi et al. (Btk) [40]. The mode of phosphoinositide binding by different domains also appears to reflect their distinct functions [37]. Yet, genome-wide analysis in yeast show that only a fraction of proteins with PH domains bind to PIPs [41,42]. Mammalian cells express more than 25 different FYVEdomain-containing proteins, and only a fraction of these proteins have been shown to be involved in membrane trafficking, cytoskeletal regulation, and receptor signaling [43,44]. Moreover, several other proteins that bind to PIPn do not contain the small protein modules as defined in the preceding text. Thus, in combination with proteomic analyses, mapping of the interactions between phosphoinositides and their potential binding proteins will not only facilitate the understanding of the PIPnsignaling networks, but will also contribute to the understanding of how this network fits into the global context of cell signaling [37]. Phosphoinositide chemical probes have facilitated the identification and characterization of numerous proteins important in the biology described in the preceding text [47]. These chemical probes are crucial components of the bioassays developed for monitoring enzymatic activity and drug-candidate screening. These compounds also form the basis for systematic mapping of protein–lipid interaction. The sections that follow will review the chemistries developed for preparing tethered InsPn and PIPn ligands that display reporter groups with different molecular topologies, the use of the probes in assays and biological contexts, and tools for lipid-proteomics studies.

MODIFICATION STRATEGIES OF PHOSPHOINOSITIDE PROBES Chemically-modified derivatives of the natural PIPn [45–47] have proven extremely valuable in isolating new PIPn-binding proteins, identifying binding sites, unraveling the roles of specific molecular species in cell signaling, and screening for potential drug leads targeting the PIPn-signaling pathway [46,47]. The design and application of PIPn-affinity probes has been extensively reviewed previously and, therefore, a more general overview will be presented here. The evolution of PIPn-affinity probes has been in response to evidence that PIPn-binding proteins are sensitive to the spatial

192

Functional Lipidomics

location of the probe, the characteristics of the linker, and mimicry of the diacylglycerol (DAG) component. An analogy to fishing has been proposed in that different baits, i.e., the types and “flavors” of the probes, will catch different fish. Examples of PI(4,5)P2 probes employing different architectures are shown in Figure 9.2, and each format will be discussed in the text that follows. The earliest probes were IPn analogs (Figure 9.2). Two such cases were the P1-tethered IPn and ring-modified IPn probes. The P-1 (and P-2) probes were first developed in the Prestwich lab and employ a simple alkyl linker off the indicated phosphate (P-1 refers to the phosphate at the 1-position of the inositol ring) [48]. Modification at the P-1 position is generally well tolerated by many proteins because the DAG in endogenous PIPn is linked at the same position. The clearest advantage of this type of probe is the simplicity in synthesizing the alkyl linker rather than preparing a functionalized DAG. Synthesizing an aminoalkyl precursor allowed a great degree of freedom in selecting the type of probe, which was introduced via an amide linkage. In addition, these probes are generally very water soluble and easy to handle in aqueous buffers. P-1 probes have been extensively used as affinity resins, photoaffinity probes [46,47], fluorescent markers for cellular studies [49] and fluorescence polarization (FP) assays [50], and as ligands in surface plasmon resonance (SPR) [51]. The most important disadvantages of these probes are the lack of DAG, precluding insertion into membranes, and the tag being close to the head group, which can potentially interfere with binding interactions. Analogous to the P-1 compounds are the ring-modified derivatives, in which the tag is attached to a nonphosphorylated hydroxyl group (Figure 9.2). A number of P-1 tethered affinity resin

−2O PO 3 HO

2-ester tethered photoaffinity resin

OPO3−2 −2O PO OH 3 O O P OHO O OH O

OPO3−2 OH

Acyl-linked BODIPY-TMR

−2O PO 3

OPO3 −2

HO

O

Triester-linked photophore

OPO3−2 −2O PO OH 3 O O P OHO O OH

OPO3−2 OH O O P O O OH

O

O

N

O

HO NH

OPO 3−2 OH O − O P O O OH O

T

O

O

−2O PO 3

O

O

O HN

Biotin labeled Pea-PIP

O

T

O

O

O

O − O P O O O

N OH

O

HN O

NH O

HN O

O NH O

S

HN F N BF N+

O

N H

MeO

FIGURE 9.2 Chemical modification strategies of phosphoinositides as exemplified by PI(4,5)P2 analogs.

Methods of Probing Phosphoinositides– Protein Interactions

193

2-substituted Ins(1,4,5)P3 compounds were synthesized and used to prepare affinity matrices for protein purifications and photoaffinity labeling of IPn-binding protein active sites [52]. In addition, an Ins(1,2,6)P3 analog, esterified at the 5-position with a photoaffinity label, was used to search for molecular targets of the antiinflammatory drug α-trinositol [53]. The next generation of probes were acyl-tethered PIPn analogs in which the tag was incorporated into the sn-1 acyl chain via a 6-aminohexanoyl linker (Figure 9.2). Though they were more complex synthetically, these probes more accurately mimicked natural PIPn. The probe was placed further from the head group therefore reducing the chance of interference in interactions with binding proteins. This type of probe has been most extensively utilized for investigating protein–lipid interactions. Various probes have been synthesized for all eight known PIPn head groups: PI, PI(3)P, and PI(4)P [54], PI(5)P and PI(3,5)P2 [55], PI(3,4)P2 [56], and PI(4,5)P2 and PI(3,4,5)P3 [57]. Detailed applications of these compounds have been reviewed [47]. Some specific examples include a BODIPY-TMR-labeled PI(4,5)P2 compound that has been used to study the binding of the myristoylated-alanine-rich C kinase substrate (MARCKS) by fluorescence resonance energy transfer [58]. Biotinylated PI(3,4)P2 and PI(3,4,5)P3 analogs were coupled to streptavidin-coated beads and employed to affinity-purify the dual-specificity 3-phosphoinositide-interacting src homology–containing protein (PHISH) from an in vitro expression–cloning (IVEC) library [59]. Spin-labeled PI(4,5)P2 has been used to study MARCKS by EPR [60], and a PI(3)P derivative has been used in NMR-binding experiments with the FYVE domain [61]. The first anti-PI(3,4,5)P3 mAb was elicited from keyhole limpet hemocyanin coupled to over 100 acyl-tethered PI(3,4,5)P3 molecules The antibody has subsequently been used to visualize transient PDGF-stimulated PI(3,4,5)P3 formation in NIH 3T3 fibroblasts and overproduction of PI(3,4,5)P3 in uterine cancer tissue [63], PI(3,4,5)P3 induction by insulin in mouse liver [64] and hypothalamus [65,66], human breast epithelial tumors [67], and human neutrophils undergoing chemotactic migration [68]. In addition to immunocytochemistry and histochemical visualization, this antibody has also been used in intact cells to determine PI(3,4,5)P3 levels in ICAM-ligated monocytes by flow cytometry [69] and to modulate the activation state of a G-protein-gated potassium channel upon microinjection [70]. However, some disadvantages with acyl-modified PIPn have been noted. First, modifications can alter size, shape, and physiochemical properties and interfere with the lipid’s ability to act as a substrate or ligand. Second, short-chain, hydrophilic-acyltethered PIPn cannot often insert into membranes adequately. Third, in a lipid bilayer, the tag may be too far to interact with the protein when it binds to the head group. The “triester”-modified PIPn analogs were designed to try and address some of the difficulties associated with P-1 and acyl-labeled PIPn. In these derivatives, the P-1 phosphate is modified with an aminopropyl group, making it a triester, to which the tag is attached through an amide linkage (Figure 9.2). By bringing the tag closer to the head group, the environment around the lipid–water interface could be investigated. Triester photoaffinity probes were prepared for PI(3)P and PI(4)P [54], PI(3,4)P2 [56], and PI(4,5)P2 and PI(3,4,5)P3 [71]. In photolabeling studies of profilin I, BZDC-triester-PI(4,5)P2 covalently derivatized both the monomeric and oligomeric forms of the protein whereas BZDC-acyl-PI(4,5)P2 showed little labeling [72].

194

Functional Lipidomics

The obvious disadvantage is that normal anionic phosphodiester linkage to the DAG is altered to a neutral one and the area around the head group becomes sterically crowded. To circumvent problems associated with the previous examples, new tools — hybrid lipids with both functionalized phosphatidylethanolamine (PE or Pea) and PIPn head groups, the so-called “Pea-PIPs,” — were developed (Figure 9.2) [73]. The strategy involved using a 2,3-diacylthreitol (four-carbon) backbone in place of the three-carbon 1,2-diacylglycerol backbone. These hybrid lipids have a PE head group at the 1-position and a PIPn head group at the 4-position. The reporter group can then be covalently attached to the free PE amino group, and is thus situated at the lipid–water interface at a site distant from the key PIPn head group that is recognized by enzymes and binding proteins. The diacyl moiety permits insertion and retention in a lipid bilayer. The first derivatives were NBD, fluorescein, biotin, and proxyl-linked Pea-PI(4,5)P2 [73], and the technology has subsequently been expanded to include Pea-PIPn with all eight head groups and an expanded number of fluorescent and affinity probes appended to the Pea moiety. Preliminary experiments have demonstrated that Pea-PIPs posses features that have advantages over acyl-modified PIPns, including increased water solubility, improved translocation into cells without intracellular delivery methods, and improved binding kinetics with PH domains using SPR methods. Moreover, Pea-PIPs appear to be better substrates for many lipid kinases and phosphatases than the acyl-modified PIPs.

CELLULAR STUDIES WITH PHOSPHOINOSITIDE PROBES As described in the preceding text, PIPn analogs have been used to discover novel PIPn-binding proteins in cells, but this is only the beginning of how synthetic PIPn and their analogs aid cell studies. The following subsections briefly review how exogenous PIPn are used to mimic physiological function, visualize location, and measure enzyme activity in cells.

MIMICRY

OF

PHYSIOLOGICAL EFFECTS

OF

PIPN

IN

CELLS

Three methods have been used to directly increase the intracellular PIPn concentration in cells to determine a given PIPn function: (1) microinjection and via membranepermeant acetoxymethyl esters, (2) addition of micellar PIPn to cell media, and (3) lipid transfection via Shuttle or Signal PIP system. First, the most direct method for introducing PIPn into cells is microinjection, and Ikonomov et al. used this technique to show that PI(3,5)P2, but not PI(5)P corrects endomembrane defects [74] and an excessive vacuole phenotype [75] in cells expressing PIKfyve mutants. Also, Feranchak et al. measured an increased Cl− conductance when PI(3,4)P2 and PI(3,4,5)P3, but not PI(3)P or PI(4,5)P2, were delivered intracellularly by a patch-clamp pipette [76]. In addition to microinjection, Li et al. and Rudof et al. accomplished delivery of PIPn and IPn into cells with membrane-permeant acylmethoxy esters (AM), which, after traversing the plasma membrane, are rendered biologically active by the action of intracellular esterases

Methods of Probing Phosphoinositides– Protein Interactions

195

[77,78]. Di-C8 PI(3,4,5)P3/AM delivered in this way to adipocytes was an incomplete substitute for insulin, but di-C12-PI(3,4,5)P3/AM delivered to T84 carcinoma cells did mimic the effects of epidermal growth factor (EGF) stimulation [79]. Greenwood and colleagues found that rat embryonic fibroblast cells incubated for 30 min with either PI(3,4,5)P3 vesicles or PI(3,4,5)P3/AM induced restructuring of focal adhesion complexes [80]. Finally, Niggli described how PI(3,4,5)P3/AM induced polarity and F-actin via Rho kinase and a positive feedback loop in primary human neutrophils [81]. The second PIPn delivery method combines PIPn with or without carrier lipids which are then sonicated to form micelles or vesicles that in turn are taken up by cells. Cantley’s group first demonstrated this in the late 1990s using mixed vesicles of di-C16 PI(3,4,5)P3 micelles or di-C8 PI(3,4,5)P3. They were able to stimulate 3T3 cell migration at 20–33% of the level induced by PDGF [82]. And in the same year, Franke et al. reported that synthetic PtdIns-3,4-P2 activated Akt both in vitro and in vivo [83]. Di-C16 synthetic PI(4,5)P2 triggered full spreading of platelets pretreated with wortmannin whereas PI(3,4)P2 was only able to restore partial platelet spreading [84]. More recently, this approach has been used to at least partially recapitulate cell activation by IGF-1 in 3T3-L1 preadipocytes [85], Ca2+ influx in T cells [86] and mast cells [87], and to restore agonist-dependent activation of Rap1B in wortmannin-treated human platelets [88]. The third method introduces synthetic PIPn and analogs into cells with cationic carriers, e.g., Shuttle PIP™ or Signal PIP™. Ozaki et al. first demonstrated that cells could have a physiological response to phosphoinositides when synthetic PIPn formed complexes with polyamine carrier molecules [89]. This seminal paper describes the delivery of PIPn analogs to a wide variety of cells, and shows that exogenous PIPn delivered in this way are biologically active — inducing Ca2+ mobilization in 3T3-L1 and NIH3T3 cells. Since this report was published in 2000 and commercialized by Echelon, there have been more than a dozen papers published using this approach to determine the effect of an individual PIPn on a given biological system. Several of these findings are briefly summarized in the next paragraph. Weiner et al. utilized this method to demonstrate that one product of PI 3-kinase, PI(3,4,5)P3, contributed to neutrophil polarization and chemotaxis [90]. Studies conducted in bone-marrow-derived mast cells from SHIP knockout mice demonstrated that both PI(3,4)P2 and PI(3,4,5)P3 contribute to the cellular activation of Akt [91]. Wang et al. delivered synthetic PI(4)P to show that PI(4)P-rich membranes help define the Golgi in mammalian cells and aids in the recruitment of AP-1 complexes [92]. Larsen et al. delivered PI(3,4,5)P3 to an organ culture system to show that this lipid product is important in mouse submandibular gland epithilium branching morphogenesis during development [93]. Finally, Klip and Sweeney and their colleagues showed that PI(3,4,5)P3 can directly cause glucose transporter 4 (GLUT4) translocation to the plasma membrane in fibroblast and myoblast cells, but full activation of the glucose transporter requires additional PI 3-kinase-dependant signals [94]. The combination of multiple methods to directly deliver exogenous PIPn/PIP analogs inside cells, along with improved techniques for measuring cellular responses, will lead to new discoveries of individual PIPn functions in lipid-signaling pathways.

196

Functional Lipidomics

VISUALIZATION

OF

PIPN

IN

CELLS

Fluorescent NBD-PI(4,5)P2 was delivered to mammalian cells (MDCK, CHO, and NIH 3T3 fibroblasts), plant root tip cells (A. thaliana), yeast (S. cerevisiae), protists (Cryptosporidium parvum), and bacteria (E. coli) with the aid of a stoichiometric amount of polyamine carrier [89]. In mammalian cells, NBD-PI(4,5)P2 was localized to the plasma membrane, punctate cytoplasmic structures, and nuclear speckles, whereas NBD-PI(3,4,5)P3 localized at the plasma membrane and cytoplasmic vesicular structures, but remained outside of the nucleus. Several years later, di-C6, BODIPY-FL PI(3,4,5)P3 delivered to primary human neutrophils was over 80% localized to the inner leaflet of the plasma membrane with minor but distinct regions of exclusion that corresponded to immobilization of the phosphoinositide [95]. These exclusion zones were specific for 3’-phosphorylated phosphoinositides because both fluorescent PI(3,4)P2 and PI(3,5)P2 showed similar patches of exclusion and immobility, whereas PI(4,5)P2 and PA were uniformly distributed and mobile. A third study using fluoresecent PIPn analogs demonstrated that annexin 2 bound PI(4,5)P2 in vitro, and annexin 2-GFP and anti-PI(4,5)P2 antibody colocalized at sites of pinosome formation [96]. Furthermore, TMR-labeled PI(4,5)P2 introduced into living cells also colocalized with pinosome-forming vesicles. Due to their intermediate aqueous solubility, Pea-PIPs with hydrophobic fluors may enter cells without the aid of carrier molecules. Figure 9.3 shows Pea-PI(4,5)P2BODIPY-TR and Pea-PI(3,4,5)P3-BODIPY-TR, added without facilitation of any carrier molecules, localized to intracellular compartments with bright staining associated in specific regions of the plasma membrane. This pattern of intracellular localization positions these PIPn analogs correctly in the cell to substitute for endogenous PIPn in lipid-signaling pathways. In support of this, Akihiro Kusumi and his Pea-PI(3,4,5)P3

Pea-PI(4,5)P2

FIGURE 9.3 (See color figure following page 174.) Localization of BODIPY-TR Pea-PIPn in 3T3-L1 preadipocyte cells after adding to cell medium. BODIPY-TR Pea-PI(4,5)P2 and Pea-PI(3,4,5)P3 were added to 50% confluent 3T3-L1 preadipocyte cells seeded onto 8-well cover-glass chamber slides. The cells were imaged within 10–20 min on a Zeiss confocal microscope.

Methods of Probing Phosphoinositides– Protein Interactions

197

coworkers at Nagoya University and ERATO, Japan, have used total internal reflectance fluorescence (TIRF) have used BODIPY-TR Pea-PI(4,5)P2 and BODIPY-TRPI(3,4,5)P3 incorporated within intracellular vesicles diffusing rapidly just beneath the cell membrane in the cytoplasm. TIRF enables high resolution, and single PeaPIP molecules are visualized undergoing lateral diffusion in the membrane with diffusion rates comparable to those of single phospholipid molecules. Together, these images and video established the ability of fluorescent Pea-PIP analogs to be taken up by cells and to behave biophysically like endogenous lipids. Pea-PIP analogs are rapidly taken up by cells (cells begin to brighten in less than 1 min) without the aid of carriers, which eliminates constraints due to cytotoxicity of some carriers. The full scope of applications for this new class of analogs by cells for physiological processes and cell-based assays remains to be discovered.

DEVELOPMENT OF NONRADIOACTIVE LIPID ASSAYS IN VITRO ASSAYS The phosphoinositide probes, described above are crucial of Phosphoinositide Probes, are critical components in developing assays to monitor the activity of lipidmetabolizing enzymes. Such assays are also useful for screening compound libraries targeting these enzymes and to measure the mass production of a certain phosphoinositide in a biological matrix. Because the lipid products of PI 3-K enzymes serve as second messengers responsible for proliferation, survival, adhesion, and several other cellular responses, PI 3-Ks have emerged as an attractive target for drug development. Several compounds with inhibitory effects on PI 3-K activity have been identified, including wortmanin and LY294002 with reasonable specific activity compared to other kinases. However, these compounds do not exhibit selectivity within the PI 3-K family and thus have lesser therapeutic potential. Conventionally used lipids assays involve the usage of radioactive isotopes. The radioactive method involves tedious steps of extraction of labeled substrate and product fractions. This method is neither environment- nor user-friendly for detection of PI 3-K activity. A nonradioactive method was needed. By using fluorescently labeled and biotinylated PIPn probes, two assay formats have been developed for PI 3-K: an amplified luminescent proximity homogenous assay (ALPHA) and a fluorescence polarization assay (FP). The PI 3-K assay in the ALPHA format is composed of six major components: synthetic PI(4,5)P2 as a substrate, biotinylated PI(3,4,5)P3 as the probe, a GST-tagged PI(3,4,5)P3 detector, the PI 3-K enzyme and streptavidin-coated donor beads, and anti-GST-coated acceptor beads (Perkin Elmer) (Figure 9.4). Biotinylated PI(3,4,5)P3 and streptavidin-coated donor beads form a complex that releases singlet oxygen upon exposure to laser beam at 680 nm. In addition, the GST-tagged PI(3,4,5)P3 detector forms a complex with the anti-GST-coated acceptor beads. The binding affinity of biotinylated PI(3,4,5)P3 and the GST-tagged PI(3,4,5)P3 detector bring the donor and acceptor beads in proximity. Singlet oxygen released from donor beads reacts with acceptor beads and causes chemiluminescence molecules of the acceptor beads to produce a luminescent signal. Production of PI(3,4,5)P3 by PI 3-

198

Functional Lipidomics

FIGURE 9.4 Schematic illustration of PI 3-K assay in the ALPHA format. When biotinylated PI(3,4,5)3-coated with donor beads of ALPHA Screen interact with acceptor beads coated with PI(3,4,5)P3-detector protein in close proximity, the diffusion of singlet oxygen produced by the photosensitive donor beads activate chemiluminescent molecules on the acceptor beads, producing luminescence. Addition of PI 3-kinase products PI(3,45)P3 in the ALPHA Screenmix displaces the biotinylated PI(3,4,5)P3 from lipid-protein-bead complex causes decreased luminescence.

K from PI(4,5)P2 competes for the binding to the PI(3,4,5)P3 detector and disrupts the bead complex and hence causes a reduced signal. The FP assay is readily adapted to a homogenous “mix and measure” format. The sensitivity of the assay is comparable to the radioactive assay and, thus, is becoming a popular option for HTS assay implementation [97]. In addition to the detection of phosphoinositides, FP assays have been used for determination of protein kinase activity [98–102]. These assays are based on competitive immunoassays in which a phosphopeptide formed by the activity of a kinase competes with a fluorescently labeled phosphopeptide for interaction with phosphor-specific protein or antibody [50]. The FP format of the PI 3-K assay involves fewer components (Figure 9.5). It contains

Methods of Probing Phosphoinositides– Protein Interactions

199

FIGRUE 9.5 Schematic illustration of PI 3-K assay in the FP format. Randomly oriented fluorophore-labeled PI(3,4,5)P3 molecules do not rotate plane polarized light. However, the ligand complexed to a detector protein, rotates anisotropizally rotation and causes increased polarization. Addition of unlabeled product of PI 3-kinase, PI (3,4,5)P3 in the mix displaces fluorophore-labeled PI(3,4,5)P3 and causes decreased fluorescence.

synthetic PI(4,5)P2 as a substrate, fluorescently labeled PI(3,4,5)P3 as the probe, a PI(3,4,5)P3 detector, and the PI 3-K enzyme. The PI(3,4,5)P3 detector, a 40-kD protein, binds to fluorescently labeled PI(3,4,5)P3 probe and causes the polarization value of the fluorescent PI(3,4,5)P3 to increase. The production of PI(3,4,5)P3 by PI 3-K from the PI(4,5)P2 substrate will compete for binding to the protein detector and free the fluorescently labeled PI(3,4,5)P3. The competition leads to a decrease in mP values as a result of the reduction in the anisotropiz fraction. These assays are sensitive, faster, and robust. They can be used to monitor the enzymatic activity of the PI 3-K, to screen compound libraries targeting the PI 3K, and to measure the presence of PI(3,4,5)P3 in a biological matrix. Due to the wealth of synthetic phosphoinositide probes, similar assays have also been developed for SHIP2 and PTEN enzymes. Given the importance of lipid kinases, phosphatases, and phospholipases in human pathogenesis, these assays carry significant value to further research in these areas and also for drug discovery and development.

200

Functional Lipidomics

CELL-BASED ASSAYS Modified lipid analogs are also successfully used in cell-based lipid assays. The following is a short description of using inositol phosphate analogs to develop a cell-based assay for PLC activation as a direct readout for G-protein-coupled receptor (GPCR) activation in intact cells. GPCRs are a large class of membrane-spanning heptahelical receptors with diverse ligands [103]. GPCRs are classified into four subfamilies based on the G proteins to which they couple and the corresponding second-messenger system effected. All members of the Gαq subfamily activate the beta isoform of phosphoinositide-specific phospholipase C, which cleaves PI(4,5)P2 to produce DAG and Ins(1,4,5)P3. Ins(1,4,5)P3 is subsequently metabolized by both kinases and phosphatases, and several of these inositol phosphates are used to assess receptor activation. A simple assay has been developed to quantify Ins(1,3,4,5)P4 in activated cells (Figure 9.6). This assay utilizes a P1-inositol phosphate probe to bind a specific inositol phosphate-binding protein (IPBP). A GST-modified PH domain and biotinylated Ins(1,3,4,5)P4 were used as binding partners to measure cell-derived Ins(1,3,4,5)P4 at 0 to 4 min after addition of the GPCR ligand, platelet-activating factor (PAF), to triplicate wells according to the procedure in Figure 9.6. Figure 9.7 (top panel) shows the standard curve generated by adding increasing concentrations of Ins(1,3,4,5)P4 that generates a competitive signal. The competitive Ins(1,3,4,5)P4 signal generated from activation of these cells is shown as ALPHA luminescent values lower than the “no competitor” control in Figure 9.7 (lower panel). Figure 9.8 shows the concentration of Ins(1,3,4,5)P4 generated, as calculated from nonlinear regression of the standard curve. The levels of Ins(1,3,4,5)P4 remain elevated for the entire time course, establishing a lengthened window of time for activating and stopping the cell activation portion of the assay compared to other nonradioactive cell-based inositol phosphate assays for GPCR activation. The successful automation of this assay for HT screening will provide researchers in drug discovery with a viable alternative to image-based Ca2+ reading systems.

Cells + Agonist

Cells + Standard 1 to 5 min

PCA 15 min

Neutralize, Binding protein, & Tracer mix 0.5 to 2 hrs

Read with compatible instrument

FIGURE 9.6 Protocol for measuring inositol phosphate production in cells.

Methods of Probing Phosphoinositides– Protein Interactions

201

125,000 S/B IC50 580 nM 30

Alpha RLU

100,000

R2 0.986

75,000 50,000 25,000 0 10-10

10-9

10-8

10-7

10-6

10-5

10-4

(IP4 competitor), M

125,000

No competitor cont.

Alpha RLU

100,000 75,000 50,000 25,000 0 0

1

2 3 4 PAF activation time (min)

5

FIGURE 9.7 Activation of a specific receptor in an engineered cell line (PAF-R HEK-293 cells) generates measurable Ins(1,3,4,5)P4. Top panel shows the standard curve generated by adding increasing concentrations of Ins(1,3,4,5)P4 which generates a competitive signal. The competitive Ins(1,3,4,5)P4, signal generated from activation of these cells is shown as ALPHA luminescent values lower than the “no competitor” control in the lower panel.

IP4 generated (nM)

400 300 200 100 0 0

1 2 3 Activation time (min)

4

FIGURE 9.8 Demonstration of the concentration of Ins(1,3,4,5)P4 generated in PAF-R cells as a function of time calculated from the standard curve.

202

Functional Lipidomics

PROBING PROTEIN–PHOSPHOINOSITIDE INTERACTIONS USING IMMOBILIZED LIPIDS The task of profiling the interactions between phosphoinositides and their associated binding proteins has two main aspects. First, for a given protein, we seek to determine the binding profile of the protein to each of the subspecies of the phosphoinositides. Second, for a given set of the phopshoinositides, we seek the identities and functions of the binding proteins that are bound to a given PIPn. The second problem is the more complex one. In a biological context, a protein that complexes with a phosphoinositide may not be the one that has direct binding contact with the lipid molecule, but rather may be indirectly recruited to the membrane through protein–protein interactions. This adds an additional layer of complexity. Other chapters have given detailed accounts of using liquid chromatography and mass spectrometry in lipidomics research. Here we describe how lipid chemistry will facilitate proteomics research in the the subdiscipline we refer to as proteolipidomics. Three tools based on immobilized lipids are particularly useful for lipid-proteomics research: phosphoinositides immobilized on beads, phosphoinositides in a stabilized liposomal structure, and phosphoinositides immobilized on solid surfaces in an array format (Figure 9.9). Phosphoinositides immobilized on beads with cleavable linkages have been used to profile the lipid-associated proteins in a defined biological context, namely the PI 3-K signaling pathway [104], as detailed in Chapter 10 in this volume. Phosphoinositide-binding proteins are pulled down by the beads, and the proteins are identified by tandem mass spectrometry (MS/MS) and analyzed by other assay methods such as immunoblots, lipid-binding assays, and protein kinase assays. The beads pull down a complex that is supposedly aggregated to the lipid molecule in response to a cell signal. The mere identification of the presence of a protein in the complex may shed light on the interaction network along the signaling pathway. Novel phosphoinositide-binding proteins may also be discovered in this process. Previously, various PIPn beads have been applied to affinity-purify PIPn-binding proteins and facilitate the identification of a single protein of interest. It is the coupling with mass spectrometry that made the identification of specific PIPn-binding proteins possible. Another approach to lipid-proteomics research is the use of immobilized PIPn in a stablized liposomal structure. Because PIPn–protein interactions occur at the cell’s inner leaflet, employing PIPn-containing liposomes is a common practice to model the natural membrane. However, instability and the difficulty in producing reproducible liposomes are two potential drawbacks with conventional vesicles. To address these problems, stabilized polymerized PIPn-containing liposomes have been prepared, taking advantage of diyne photochemistry. P-1-tethered aminopropylInsPns were coupled to 10,12-pentacosadiynoic acid and incorporated into vesicles containing diyne-phosphatidylcholine, diyne-phosphatidylethanolamine, and a synthetic biotinylated diyne-phosphatidylethanolamine. Irradiation with UV light (254 nm) polymerizes the acyl chains into an ene-yne system producing suspensions of pale red nanoparticles, named PolyPIPosomes™, which are stable for many months at 4°C. Experiments with PolyPIPosomes showed similar results as with conventional liposomes when binding to nitrocellulose-immobilized proteins. In addition,

Methods of Probing Phosphoinositides– Protein Interactions

203

FIGURE 9.9 Schematic presentation of immobilized phosphoinositides. Top: biotinylated reductively cleavable PI(3,4,5)P3 streptavidin beads. Middle: biotinylated polymerized PI(4,5)P2 PolyPIPosomes™. Bottom: amino-linked PI(4,5)P2 immobilized to epoxy chip surface.

PolyPIPosomes bound to PH domains with expected specificities in amplified luminescenct proximity homogeneous assays and performed better than acyl-modified PIPn in SPR [105]. A biotin tag can be easily attached to PolyPIPosomes™, and this stabilized liposomal structure immobilized on beads may provide better support for lipidomics research when coupled with other tools such as MS and bioassay. The third tool for lipid-proteomics research is to employ an array of lipids spotted on solid surfaces in an addressable format. A primitive example for lipid arrays is the use of “Fat Westerns”[106–108] for biochemical applications on subcellular localization and lipid selectivity. Protein overlay blots provide a quick interaction profile of the specificity and relative affinity of a given protein for specific phosphoinositides. This simple tool has facilitated the study of lipid-binding specificity of many proteins, including a PX domain in Vam7 t-SNARE towards PI(3)P [109], the PX domain of the P40phox subunit of NADPH towards PI(3)P [110], the COOHterminal domain of tubby protein towards PI(4,5)P2 [111], a PI(4)P-specific domain (FAPP1), two PI(3)P-specific binders (PEPP1 and AtPH1), two PI(3,4)P2-specific binders (TAPP1 and TAPP2), a PI(3,5)P2-specific PH domain of centaurin β2 [108], and the PIPn-binding nuclear receptor ING2 [112]. Modification steps of the protein overlay blots were made to improve reproducibility, throughput, and sensitivity. Phospholipids can be spotted on solid chip surfaces in uniform and microscale spots, via both covalent and noncovalent strategies [113]. Most of the phosphoinostides immobilized through their acyl hydrophobic

204

Functional Lipidomics

chains are still recognizable by their cognate-binding proteins. Functionalized surfaces, including epoxy- and carbonyldiimidazole-activated surfaces, have proven to be reliable support for phospholipid immobilization. The phosphoinositide analogs delineated in the Section titled Introduction to Phosphoinositide Signaling provide a variety of chemical means to covalently link the lipid molecule to a solid substrate. However, covalent linkages between surfaces and lipids require laborious chemical modification of every single lipid molecule, creating a problem in making all the materials needed for automation in a high-throughput system. Moreover, covalent modification of lipids may alter their ability to be recognized by specific lipidbinding proteins. There may be a better alternative to covalent attachment. As amphiphilic molecules, lipids can be noncovalently immobilized to hydrophobic solid surfaces, e.g., hydrocarbon-chain-derivatized glass slides, or to porous surfaces, e.g., polymer-coated surfaces. Employing amphiphilicity, other lipid classes in addition to phosphoinositides can be spotted on a single lipid chip, increasing the density and information content of an array. Data acquisition and analysis of a lipid array may employ the tools used in the DNA microarray technology. Different sets of images of the lipid chip layered with lipid-associated proteins can be obtained, depending on the primary purpose of the study. Either the total lipid-binding protein patterns or the subpatterns of the proteins captured by the lipid chip can be visualized via generic protein stains or antibody recognition. The identity of the proteins that are associated with a single lipid spot may be characterized by mass spectrometry.

CLOSING REMARKS Recognition of the integral role of lipid signaling in normal cell biology and pathophysiology has led to the conclusion that lipid–protein interactions control the majority of signal transduction cascades. The sequence of the human genome available, combined with the growing ability to conduct massive proteomic analyses, makes proteolipidomics a key research frontier for the next 10 years. We have previously indicated that genes are but the blueprints, and proteins but the structures built from these blueprints. Extending this analogy, it is the lipids that provide a functional home [114]. As a result, functional lipidomics, or proteolipidomics, is the key to the identification of novel molecular targets for the discovery of more selective therapeutics that can modify signal transduction in a tissue- and cell-specific manner. Lipid molecules are not composed of repeating building blocks such as DNA and proteins. The complexity of lipid structures makes the systematic analysis of lipids and lipid-interacting entities a daunting task. Mass spectrometry is probably the most powerful tool in the lipidomics research realm. Yet, it is unlikely that a single-platform technology would be applicable to all lipid species which would limit the information content obtained. For example, although comprehensive lipid contents have been profiled [115–117], the information is limited to the composition of fatty acids in the total lipids. The use of synthetic lipid probes in hypothesisbased research has facilitated the understanding of many cellular functions. When lipidomics research, a more information-driven field, is carried out in a defined functional context, tools based on chemical probes are even more powerful, as

Methods of Probing Phosphoinositides– Protein Interactions

205

demonstrated by the profiling of PIPn-binding proteins along the PI 3-K pathway utilizing the cleavable PIPn beads. The arsenal of synthetic lipid probes will undoubtedly still play an important role in lipidomics research. The role of lipid chemistry is crucial because creative lipid analogs and attachment protocols will provide the tools from which important biological conclusions can be achieved in proteolipidomic research.

References 1. Martin, T.F.J., Phosphoinositide lipids as signaling molecules: common themes for signal transduction, cytoskeletal regulation, and membrane trafficking. Annu Rev Cell Develop Biol, 1998. 14:. 231–264. 2. Czech, M.P., PIP2 and PIP3: Complex roles at the cell surface. Cell, 2000. 100(6): 603–606. 3. Simonsen, A. and H. Stenmark, PX domains: attracted by phosphoinositides. Nat Cell Biol, 2001. 3(8):. E179–182. 4. Simonsen, A. et al., The role of phosphoinositides in membrane transport. Curr Opin Cell Biol, 2001. 13(4): 485–492. 5. Anderson, R.A., et al., Phosphatidylinositol phosphate kinases, a multifaceted family of signaling enzymes. J Biol Chem, 1999. 274(15): 9907–9910. 6. Majerus, P.W., M.V. Kisseleva, and F.A. Norris, The role of phosphatases in inositol signaling reactions. J Biol Chem, 1999. 274(16): 10669–10672. 7. Leslie, N.R., R.M. Biondi, and D.R. Alessi, Phosphoinositide-regulated kinases and phosphoinositide phosphatases. Chem Rev, 2001. 101(8): 2365–2380. 8. Toker, A. and L.C. Cantley, Signalling through the lipid products of phosphoinositide3-OH kinase. Nature, 1997. 387(6634): 673–676. 9. Toker, A., The synthesis and cellular roles of phosphatidylinositol 4,5-bisphosphate. Curr Opin Cell Biol, 1998. 10(2): 254–261. 10. Overduin, M., M.L. Cheever, and T.G. Kutateladze, Signaling with phosphoinositides: better than binary. Mol Interv, 2001. 1(3): 150–159. 11. Cullen, P.J. et al., Modular phosphoinositide-binding domains--their role in signalling and membrane trafficking. Curr Biol, 2001. 11(21): R882–93. 12. Wenk, M.R. and P. De Camilli, Protein-lipid interactions and phosphoinositide metabolism in membrane traffic: insights from vesicle recycling in nerve terminals. Proc Natl Acad Sci U S A, 2004. 101(22): 8262–8269. 13. Corvera, S., A. Darrigo, and H. Stenmark, Phosphoinositides in membrane traffic. Curr Opin Cell Biol, 1999. 11(4): 460–465. 14. De Camilli, P. et al., Phosphoinositides as regulators in membrane traffic. Science, 1996. 271(5255): 1533–15399. 15. Odorizzi, G., M. Babst, and S.D. Emr, Phosphoinositide signaling and the regulation of membrane trafficking in yeast. Trends Biochem Sci, 2000. 25(5): 229–235. 16. Hall, A., Rho GTPases and the actin cytoskeleton. Science, 1998. 279(5350): pp. 509–514. 17. Rameh, L. and L. Cantley, The role of phosphoinositide 3-kinase lipid products in cell function. J Biol Chem, 1999. 274: 8347–8350. 18. Downward, J., Mechanisms and consequences of activation of protein kinase B/Akt. Curr Opin Cell Biol, 1998. 10(2): 262–267.

206

Functional Lipidomics

19. Wymann, M.P. et al., Lipids on the move: phosphoinositide 3-kinases in leukocyte function. Immunol Today, 2000. 21(6): 260–264. 20. Chang, Y.T., G.R. Rosania, and S.K. Chung, Inositol phospholipid pathway inhibitors and regulators. Opin Ther Patents, 2001. 11: 45–59. 21. Roymans, D. and H. Slegers, Phosphatidylinositol 3-kinases in tumor progression. Eur J Biochem, 2001. 268: 487–498. 22. Scheid, M.P. and J.R. Woodgett, Phosphatidylinositol 3 kinase signaling in mammary tumorigenesis. J Mammary Gland Biol Neoplasia, 2001. 6(1): 83–99. 23. Katso, R. et al., Cellular function of phosphoinositide 3-kinases: implications for development, homeostasis, and cancer. Annu Rev Cell Dev Biol, 2001. 17: 615–675. 24. Maehama, T. and J. Dixon, The tumor suppressor, PTEN/MMAC1, dephosphorylates the lipid second messenger, phosphatidylinositol-3,4,5-trisphosphate. J Biol Chem, 1998. 273: 13375–13378. 25. Tamura, M. et al., PTEN interactions with focal adhesion kinase and suppression of the extracellular matrix-dependent phosphatidylinositol 3-kinase/Akt cell survival pathway. J Biol Chem, 1999. 274(29): 20693–20703. 26. Haas-Kogan, D. et al., Protein kinase B (PKB/Akt) activity is elevated in glioblastoma cells due to mutation of the tumor suppressor PTEN/MMAC. Curr Biol, 1998. 8: pp. 1195–1198. 27. Phillips, W.A. et al., Increased levels of phosphatidylinositol 3-kinase activity in colorectal tumors. Cancer, 1998. 83(1): 41–47. 28. Ma, Y.Y. et al., PIK3CA as an oncogene in cervical cancer. Oncogene, 2000. 19(23): 2739–2744. 29. Shayesteh, L. et al., PIK3CA is implicated as an oncogene in ovarian cancer. Nat Genet, 1999. 21(1): 99–102. 30. Blero, D. et al., The SH2 domain containing inositol 5-phosphatase SHIP2 controls phosphatidylinositol 3,4,5-trisphosphate levels in CHO-IR cells stimulated by insulin. Biochem Biophys Res Commun, 2001. 282(3): 839–843. 31. Habib, T. et al., Growth factors and insulin stimulate tyrosine phosphorylation of the 51C/SHIP2 protein. J Biol Chem, 1998. 273(29): 18605–18609. 32. Bolland, S., et al., SHIP modulates immune receptor responses by regulating membrane association of Btk. Immunity, 1998. 8(4): 509–516. 33. Krystal, G., Lipid phosphatases in the immune system. Semin Immunol, 2000. 12(4): 397–403. 34. Rauh, M.J. et al., Role of Src homology 2-containing-inositol 5-phosphatase (SHIP) in mast cells and macrophages. Biochem Soc Trans, 2003. 31(Pt. 1): 286–291. 35. Mannervik, B. and U.H. Danielson, Glutathione transferases-structure and catalytic activity. CRC Crit Rev Biochem, 1988. 23(3): 283–337. 36. Hurley, J.H. and T. Meyer, Subcellular targeting by membrane lipids. Curr Opin Cell Biol, 2001. 13(2): 146–152. 37. Lemmon, M.A., Phosphoinositide recognition domains. Traffic, 2003. 4(4): 201–213. 38. Lietzke, S.E. et al., Structural basis of 3-phosphoinositide recognition by pleckstrin homology domains. Mol Cell, 2000. 6(2): 385–394. 39. Ferguson, K.M. et al., Structural basis for discrimination of 3-phosphoinositides by pleckstrin homology domains. Mol Cell, 2000. 6(2): 373–384. 40. Baraldi, E. et al., Structure of the PH domain from Bruton’s tyrosine kinase in complex with inositol 1,3,4,5-tetrakisphosphate. Struct. Fold Des, 1999. 7(4): 449–460. 41. Lemmon, M.A., Pleckstrin homology domains: not just for phosphoinositides. Biochem Soc Trans, 2004. 32(Pt. 5): 707–711.

Methods of Probing Phosphoinositides– Protein Interactions

207

42. Yu, J.W. et al., Genome-wide analysis of membrane targeting by S. cerevisiae pleckstrin homology domains. Mol Cell, 2004. 13(5): 677–688. 43. Gillooly, D.J., A. Simonsen, and H. Stenmark, Cellular functions of phosphatidylinositol 3-phosphate and FYVE domain proteins. Biochem J, 2001. 355(Pt. 2): 249–258. 44. Misra, S., G.J. Miller, and J.H. Hurley, Recognizing phosphatidylinositol 3-phosphate. Cell, 2001. 107(5): 559–562. 45. Prestwich, G.D. et al., Probing phosphoinositide polyphosphate binding to proteins, in Phosphoinositides: Chemistry, Biochemistry and Biomedical Applications, K.S. Bruzik, Editor. 1999, American Chemical Society: Washington, D.C. 24–37. 46. Prestwich, G.D., Touching all the bases: Inositol polyphosphate and phosphoinositide affinity probes from glucose. Acc Chem Res, 1996. 29: 503–513. 47. Prestwich, G.D., Phosphoinositide signalling: from affinity probes to pharmaceutical targets. Chem Biol, 2004. 11: 619–637. 48. Estevez, V.A. and G. Prestwich, Synthesis of Enantiomerically Pure, P-1-Tethered inositol tetrakis(phosphate) Affinity labels via a Ferrier Rearrangement. J Am Chem Soc, 1991. 113: 9885–9887. 49. Inoue, M. et al., Homogeneous Ca2+ stores in rat adrenal chromaffin cells. Cell Calcium, 2003. 33: 19–26. 50. Drees, B.E. et al., Competitive fluorescence polarization assays for the detection of phosphoinositide kinase and phosphatase activity. Comb Chem High Throughput Screen, 2003. 6: 321–330. 51. Mehrotra, B., D.G. Myszka, and G.D. Prestwich, Binding kinetics and ligand specificity for the interactions of the C2B domain of synaptogmin II with inositol polyphosphates and phosphoinositides. Biochemistry, 2000. 39(32): 9679–9686. 52. Ozaki, S. et al., Synthesis and biological properties of 2-substituted myo-inositol 1,4,5-trisphosphate analogues directed toward affinity chromatography and photoaffinity labeling. Carbohydr Res, 1992. 234: 189–206. 53. Chaudhary, A. and G.D. Prestwich, Photoaffinity analogue for the anti-inflammatory drug alpha-trinositol: Synthesis and identification of putative molecular targets. Bioconjugate Chem, 1997. 8(5): 680–685. 54. Chen, J., L. Feng, and G.D. Prestwich, Asymmetric total synthesis of phosphatidylinositol 3-phosphate and 4-phosphate derivatives. J Org Chem, 1998. 63(19): 6511–6522. 55. Peng, J. and G.D. Prestwich, Synthesis of L-α-phosphatidyl-D-myo-inositol 5-phosphate and L-α-phosphatidyl-D-myo-inositol 3,5-bisphosphate. Tetrahedron Lett, 1998. 39: 3965–3968. 56. Thum, O., J. Chen, and G.D. Prestwich, Synthesis of a photoaffinity analogue of phosphatidylinositol 3,4-bisphosphate, an effector in the phosphoinositide 3-kinase signaling pathway. Tetrahedron Lett, 1996. 37(50): 9017–9020. 57. Chen, J., A.A. Profit, and G.D. Prestwich, Synthesis of photoactivatable 1,2-O-diacylsn-glycerol derivatives of 1-L-phosphatidyl-D-myo-inositol 4,5-bisphosphate (PtdInsP2) and 3,4,5-trisphosphate (PtdInsP3). J Org Chem, 1996. 61(18): 6305–6312. 58. Gambhir, A. et al., Electrostatic sequestration of PIP2 on phospholipid membranes by basic/aromatic regions of proteins. Biophys J, 2004. 86(4): pp. 2188–2207. 59. Rao, V.R. et al., Expression cloning of protein targets for 3-phosphorylated phosphoinositides. J Biol Chem, 1999. 274(53): 37893–37900. 60. Rauch, M. et al., MARCKS sequesters spin-labeled phsophatidylinositol-4,5-bisphosphate in lipid bilayers. J Biol Chem, 2002. 277: 14068–14076.

208

Functional Lipidomics

61. Kutateladze, T.G. et al., Multivalent mechanism of membrane insertion by the FYVE domain. J Biol Chem, 2004. 279: 3050–3057. 62. Chen, R. et al., A monoclonal antibody to visualize PtdIns(3,4,5)P3 in cells. J Histochem Cytochem, 2002. 50(5):697–708. 63. Prestwich, G.D., et al., In situ detection of phospholipid and phosphoinositide metabolism. Adv Enzyme Reg, 2002. 42: 19–38. 64. Niswender, K.D. et al., Immunocytochemical detection of phosphatidylinositol 3kinase activation by insulin and leptin. J Histochem Cytochem, 2003. 51(3): 275–283. 65. Niswender, K.D. et al., Insulin activation of phosphatidylinositol 3-kinase in the hypothalamic arcuate nucleus: a key mediator of insulin-induced anorexia. Diabetes, 2003. 52(2): 227–231. 66. Schubert, M. et al., Role for neuronal insulin resistance in neurodegenerative diseases. Proc Natl Acad Sci U S A, 2004. 101(9): 3100–3105. 67. Liu, H. et al., Polarity and proliferation are controlled by distinct signaling pathways downstream of PI3-kinase in breast epithelial tumor cells. J Cell Biol, 2004. 164(4): 603–612. 68. Wang, F. et al., Lipid products of PI(3)Ks maintain persistent cell polarity and directed motility in neutrophils. Nat Cell Biol, 2002. 4(7): 513–518. 69. Perez, O.D. et al., Activation of the PKB/AKT pathway by ICAM-2. Immunity, 2002. 16(1): 51–65. 70. Ishii, M., A. Inanobe, and Y. Kurachi, PIP3 inhibition of RGS protein and its reversal by Ca2+/calmodulin mediate voltage-dependent control of the G protein cycle in a cardiac K+ channel. Proc Natl Acad Sci U S A, 2002. 99(7): 4325–4330. 71. Gu, Q.-M. and G.D. Prestwich, Synthesis of phosphotriester analogues of the phosphoinositides PtdIns(4,5)P2 and PtdIns(3,4,5)P3. J Org Chem, 1996. 61: 8642–8647. 72. Chaudhary, A. et al., Probing the phosphoinositide 4,5-bisphosphate binding site of human profilin I. Chem Biol, 1998. 5: 273–281. 73. Rzepecki, P.W. and G.D. Prestwich, Synthesis of hybrid lipid probes: derivatives of phosphatidylethanolamine-extended phosphatidylinositol 4,5-bisphosphate (PeaPIP2). J Org Chem, 2002. 67: 5454–5460. 74. Ikonomov, O.C. et al., Functional dissection of lipid and protein kinase signals of PIKfyve reveals the role of PtdIns 3,5-P2 production for endomembrane integrity. J Biol Chem, 2002. 277(11): 9206–9211. 75. Ikonomov, O.C. et al., PIKfyve Kinase and SKD1 AAA ATPase define distinct endocytic compartments. Only PIKfyve expression inhibits the cell-vacoulating activity of Helicobacter pylori VacA toxin. J Biol Chem, 2002. 277(48): pp. 46785–46790. 76. Feranchak, A.P. et al., The lipid products of phosphoinositide 3-kinase contribute to regulation of cholangiocyte ATP and chloride transport. J Biol Chem, 1999. 274(43): 30979–30986. 77. Li, W. et al., Cell-permeant caged InsP3 ester shows that Ca2+ spike frequency can optimize gene expression. Nature, 1998. 392(6679): 936–941. 78. Rudolf, M.T., A.E. Traynor-Kaplan, and C. Schultz, A membrane-permeant, bioactivatable derivative of Ins(1,3,4)P3 and its effect on Cl(-)-secretion from T84 cells. Bioorg Med Chem Lett, 1998. 8(14): 1857–1860. 79. Jiang, T. et al., Membrane-permeant esters of phosphatidylinositol 3,4,5-trisphosphate. J Biol Chem, 1998. 273(18): 11017–11024. 80. Greenwood, J.A. et al., Restructuring of focal adhesion plaques by PI 3-kinase. Regulation by PtdIns (3,4,5)-p(3) binding to alpha-actinin. J Cell Biol, 2000. 150(3): 627–642.

Methods of Probing Phosphoinositides– Protein Interactions

209

81. Niggli, V., A membrane-permeant ester of phosphatidylinositol 3,4, 5-trisphosphate (PIP(3)) is an activator of human neutrophil migration. FEBS Lett, 2000. 473(2): 217–221. 82. Derman, M.P. et al., The lipid products of phosphoinositide 3-kinase increase cell motility through protein kinase C. J Biol Chem, 1997. 272(10): 6465–6470. 83. Franke, T.F. et al., Direct regulation of the Akt proto-oncogene product by phosphatidylinositol-3,4-bisphosphate. Science, 1997. 275(5300): 665–668. 84. Heraud, J.M. et al., Lipid products of phosphoinositide 3-kinase and phosphatidylinositol 4′,5′-bisphosphate are both required for ADP-dependent platelet spreading. J Biol Chem, 1998. 273(28): 17817–17823. 85. Gagnon, A. et al., Phosphatidylinositol-3,4,5-trisphosphate is required for insulin-like growth factor 1-mediated survival of 3T3-L1 preadipocytes. Endocrinology, 2001. 142(1): 205–212. 86. Hsu, A.L. et al., Novel function of phosphoinositide 3-kinase in T cell Ca2+ signaling. A phosphatidylinositol 3,4,5-trisphosphate-mediated Ca2+ entry mechanism. J Biol Chem, 2000. 275(21): 16242–16250. 87. Ching, T.T. et al., Phosphoinositide 3-kinase facilitates antigen-stimulated Ca(2+) influx in RBL-2H3 mast cells via a phosphatidylinositol 3,4,5-trisphosphate-sensitive Ca(2+) entry mechanism. J Biol Chem, 2001. 276(18): 14814–14820. 88. Lova, P., et al., A selective role for phosphatidylinositol 3,4,5-trisphosphate in the Gi-dependent activation of platelet Rap1B. J Biol Chem, 2003. 278(1): 131–138. 89. Ozaki, S. et al., Intracellular delivery of phosphoinositides and inositol phosphates using polyamine carriers. Proc Natl Acad Sci USA, 2000. 97(21): 11286–11291. 90. Weiner, O.D., Regulation of cell polarity during eukaryotic chemotaxis: the chemotactic compass. Curr Opin Cell Biol, 2002. 14(2): 196–202. 91. Scheid, M.P. et al., Phosphatidylinositol(3,4,5)P3 is essential but not sufficient for PKB activation: phosphatidylinositol(3,4)P2 is required for PKB phosphorylation at Ser473. studies using cells from SHIP knockout mice. J Biol Chem, 2002. 277: 9027–9035. 92. Wang, Y.J., et al., Phosphatidylinositol 4 phosphate regulates targeting of clathrin adaptor AP-1 complexes to the Golgi. Cell, 2003. 114(3): 299–310. 93. Larsen, M. et al., Role of PI 3-kinase and PIP3 in submandibular gland branching morphogenesis. Dev Biol, 2003. 255(1): 178–191. 94. Sweeney, G. et al., Intracellular delivery of phosphatidylinositol (3,4,5)-trisphosphate causes incorporation of glucose transporter 4 into the plasma membrane of muscle and fat cells without increasing glucose uptake. J Biol Chem, 2004. 279(31): 32233–32242. 95. Tian, W. et al., Exclusion of exogenous phosphatidylinositol-3,4,5-trisphosphate from neutrophil-polarizing pseudopodia: stabilization of the uropod and cell polarity. EMBO Rep, 2003. 4(10): 982–988. 96. Hayes, M.J. et al., Annexin 2 binding to phosphatidylinositol 4,5-bisphosphate on endocytic vesicles is regulated by the stress response pathway. J Biol Chem, 2004. 279(14): 14157–14164. 97. Pope, A.J., U.M. Haupts, and K.J. Moore, Homogeneous fluorescence readouts for miniaturized high-throughput screening: theory and practice. Drug Discov Today, 1999. 4(8): 350–362. 98. Parker, G.J. et al., Development of high throughput screening assays using fluorescence polarization: nuclear receptor-ligand-binding and kinase/phosphatase assays. J Biomol Screen, 2000. 5(2): 77–88.

210

Functional Lipidomics

99. Wu, J.J. et al., Identification of a high-affinity anti-phosphoserine antibody for the development of a homogeneous fluorescence polarization assay of protein kinase C. J Biomol Screen, 2000. 5(1):23–30. 100. Seethala, R. and R. Menzel, A homogeneous, fluorescence polarization assay for srcfamily tyrosine kinases. Anal Biochem, 1997. 253(2): 210–218. 101. Seethala, R. and R. Menzel, A fluorescence polarization competition immunoassay for tyrosine kinases. Anal Biochem, 1998. 255(2): 257–262. 102. Turek, T.C. et al., Development and validation of a competitive AKT serine/threonine kinase fluorescence polarization assay using a product-specific anti-phospho-serine antibody. Anal Biochem, 2001. 299(1): 45–53. 103. Liebmann, C., G protein-coupled receptors and their signaling pathways: classical therapeutical targets susceptible to novel therapeutic concepts. Curr Pharm Des, 2004. 10(16): 1937–1958. 104. Pasquali, C. et al., Mining cellular phosphoinositide-binding proteins by signalling pathway proteomics. Nature Chem. Biol., 2005. 105. Ferguson, C.G. et al., Phosphoinositide-containing polymerized liposomes (PolyPIPosomes): stable membrane-mimetic vesicles for protein-lipid binding analysis. Chem. Biol, 2004: Submitted. 106. Stevenson, J., I. Perera, and W. Boss, A phosphatidylinositol 4-kinase pleckstrin homology domain that binds phosphatidylinositol 4-monophosphate. J Biol Chem, 1998. 273: 22761–22767. 107. Dowler, S. et al., DAPP1: a dual adaptor for phosphotyrosine and 3-phosphoinositides. Biochem J, 1999. 342: 7–12. 108. Dowler, S. et al., Identification of pleckstrin-homology-domain-containing proteins with novel phosphoinositide-binding specificities. Biochem J, 2000. 351: 19–31. 109. Cheever, M.L. et al., Phox domain interaction with PtdIns(3)P targets the Vam7 tSNARE to vacuole membranes. Nat Cell Biol, 2001. 3(7): 613–618. 110. Zhan, Y. et al., The p40phox and p47phox PX domains of NADPH oxidase target cell membranes via direct and indirect recruitment by phosphoinositides. J Biol Chem, 2002. 277(6): 4512–4518. 111. Santagata, S. et al., G-protein signaling through tubby proteins. Science, 2001. 292(5524): 2041–2050. 112. Gozani, O. et al., The PHD finger of the chromatin-associated protein ING2 functions as a nuclear phosphoinositide receptor. Cell, 2003. 114(1): 99–111. 113. Feng, L., Lipid microarrays. Prostaglandins Other Lipid Mediat, 2005: in press. 114. Prestwich, G.D., Visualization and perturbation of phosphoinositide and phospholipid signaling. Prostaglandins Other Lipid Mediat, 2005: in press. 115. Watkins, S.M. and J.B. German, Metabolomics and biochemical profiling in drug discovery and development. Curr Opin Mol Ther, 2002. 4(3): 224–228. 116. Watkins, S.M. et al., Individual metabolism should guide agriculture toward foods for improved health and nutrition. Am J Clin Nutr, 2001. 74(3): 283–286. 117. Watkins, S.M. et al., Lipid metabolome-wide effects of the PPAR gamma agonist rosiglitazone. J Lipid Res, 2002. 43(11): 1809–1817.

10

Fishing for Pharmaceutically Relevant PhosphoinositideBinding Proteins Using Chemical Proteomics Christian Pasquali and Christian Rommel

CONTENTS Summary ................................................................................................................211 PI3K and Phosphoinositide Signal Transduction: Regulation, Function, and Therapeutic Value..................................................................................212 Interface between Phosphoinositide Signaling and Proteomics ...........................216 Functional Proteomics: Potential and Limitation for Signal-Transduction-Oriented Protein Discovery............................217 Phosphoinositide–Protein Interactions: Specific or Promiscuous?.......................226 Protein-Domains-Mediated Signaling: An Example of the PH Domain....227 Phosphoinositide-Dependent Protein Kinases: the Example of Akt/PKB..229 Mining Phosphoinositide-Binding Proteins by Chemical Proteomics .................230 Selective Capture of Cellular Phosphoinositide Signaling Proteins ...........233 Outlook ..................................................................................................................235 Acknowledgments..................................................................................................235 References..............................................................................................................235

SUMMARY In contrast to proteins and transcripts, intracellular lipids are not genetically encoded messengers but are, nevertheless, of fundamental importance in cell physiology and pathology. Although for a long period, phospholipid signaling remained poorly understood, recent findings have created a tremendous momentum for further inves-

211

212

Functional Lipidomics

tigation of the biochemical secrets of phospholipid synthesis, metabolism, and signal transduction. Today, it is recognized that almost every cellular event requires the use of phospholipids. Among them is phosphatidylinositol (PtdIns), the basic building block for the intracellular inositol lipids in eukaryotic cells, which consists of Dmyo-inositol-1-phosphate (Ins1P) linked via its phosphate group to diacylglycerol. (For reviews on generic PtdIns metabolism, see Reference 1 to Reference 5). Upon esterification of the hydroxyl groups with phosphate (usually single or combined phosphorylations), the resulting phosphorylated phosphoinositides (PIPs) serve as second lipid messengers that promote the recruitment, relocation, and activation of cellular proteins through the assembly of intracellular signalosomes.2,6 Although different phosphoinositides preferentially localize at different sites within the cell, many of the known PIP-binding proteins signal via the inner leaflet of the plasma membrane. In particular, phosphoinositide 3-kinase (PI3K), a lipid kinase that was intensively studied during the past decade, signals via the generation of 3-PIP products.6 This signaling mediates plasma membrane morphology, plasticity, polarity, endosomal fusion, receptor turnover, vesicular trafficking comprising the endocytic (early and late endosomes and lysosomes) and the biosynthetic routes (e.g., Golgi to plasma membrane), and retrograde transport; lipid signaling may also take place in the nuclei (see Figure 10.1). Here, we will concentrate mainly on PI3K signaling as well as the genetic and pharmacological validation of PI3Ks as drug targets and therapeutic pathways. Following this, we will discuss the variety, specificity, and importance of different mechanisms by which cellular PIPs can interact with downstream protein targets. Finally, as an example of an attempt to discover novel PIP-interacting proteins in a cellular signaling context, we will present a recent study that led to the identification of novel potential PIP-binding proteins. More precisely, we will describe an integrated chemical proteomics strategy that uses: (1) cleavable soluble affinity ligands for the large-scale identification of PIP-associated proteins, (2) a specific assay for the validation of lipid–protein interactions, and (3) an activity-based technique for protein kinase profiling of isolated PIP–protein complexes.

PI3K AND PHOSPHOINOSITIDE SIGNAL TRANSDUCTION: REGULATION, FUNCTION, AND THERAPEUTIC VALUE Class-I PI3Ks control a variety of cellular functions including survival, proliferation, migration, apoptosis, and glucose metabolism. During the last decade, it became evident that this class of enzymes plays pivotal roles in the pathology of metabolic, inflammatory, allergic, and autoimmune diseases, as well as in cancer. For these reasons, small molecule inhibitors targeting Class-I PI3K isoforms have an immense potential for use in the treatment of such diseases. PI3Ks are dual-specific lipid and protein kinases that exert their function by triggering numerous intracellular signaling pathways. PI3K primarily phosphorylates inositol-containing phospholipids at the 3′-OH position in the inositol ring. Upon

Fishing for Pharmaceutically Relevant PIP-Binding

Cellular Trafficking Vesicles

Exocytosis

213

Differentiation

Cytohesins Centaurins

Arp

Glucose Metabolism

P50/P55 SHIP2

Development Amphiphysin

?

Phagocytosis 3

Polarity VAV Rac

4 5

Chemotaxis P-rex1/2

Glucose Apoptosis and Survival

PDK/Akt

Growth Proliferation Protein Turnover

Rho-GT Pase

Actin Polarity Chemotaxis Phagocytosis Motility

Tec kinases (Itk, Etk, Btk, Ilk) Erythroid Apoptosis Proliferation T/B Cell Signalling Progenitor Signalling

Rhok Myosin

? Actin

Mobility

FIGURE 10.1 The importance of phosphorylated phosphoinositide (PIP) second messenger–mediated signaling. Phosphoinositides are generated by differential phosphorylation of the inositol ring. Depending on the addition or removal of phosphate or hydroxyl groups (balls) on the 3–5 inositol carbons, different signaling proteins will associate and initiate cascade signaling that ultimately controls various cellular responses. The question marks represent as yet unidentified PIP-binding proteins.

ligand–receptor binding, plasma membrane PI3K becomes activated and transiently generates the 3′-phospholipid product PI(3,4,5)P3 (termed PIP3) by phosphorylating the D-3 position of the resident membrane phospholipid PI(4,5)P2. This latter phosphoinositide is the most abundant PIP in the cell and also serves as the leading front for signaling molecules involved in many cellular processes. A representative example of the different cellular functions mediated by 3′-PIPs products is illustrated in Figure 10.1 Along with other phosphoinositides, the 3′-PIPs play important roles as second messengers by interacting with various lipid-binding domains expressed by a variety of cellular proteins, notably the PDK1-PKB/Akt-mTOR-p70S6K kinase cascade (see Section titled Phosphoinositide–Protein Interactions: Specific or Promiscuous? for more details). Depending on the number and position of the phosphate groups, different proteins will associate with the inositol ring of the newly formed PIPs and initiate downstream signaling throughout the cell.7 PI3Ks are divided into three main classes on the basis of sequence homology and substrate preference in vitro.1,3,8 In the present section, we will focus mostly on PI3K Class I, separated into two subclasses (A and B) due to different mechanisms of activation. Following tyrosine phosphorylation, Class IA can be activated by different receptors such as integrins or receptor tyrosine kinases (RTKs). In contrast,

214

Functional Lipidomics

Class IB activation is strictly dependent on seven transmembrane G-protein-coupled receptors (GPCRs). Their composition and link to phosphoinositide signaling are illustrated in Figure 10.2. Heterodimeric Class IA consists of three (p110α, p110β, or p110δ) catalytic subunits complexed to the SH2-containing p85 or p55 regulatory subunits responsible for tyrosine kinase receptor signaling (P-Y in Figure 10.2). Binding of ligand induces receptor dimerization and subsequent transphosphorylation of Tyr residues in the cytoplasmic tail of the receptor to create a docking site for SH2–domain interaction. Heterodimeric Class IB is composed of the single catalytic subunit p110γ, which is activated by GPCRs via Gβγ and its regulatory subunit p101 (Figure 10.2 and Figure 10.3). Interestingly, both PI3Ks use PI(4,5)P2 as substrate. PI(4,5)P2 is at the center of two lipid routes (anabolic and catabolic) that both lead to different lipid products and downstream signaling. This equilibrium is required for cell homeostasis and, therefore, creates the pressing need for a better understanding and comprehensive elucidation of the different ligands targeting individual PI3K isoforms as well as of their downstream signaling partners. Figure 10.3 illustrates the activation of phosphoinositide signal transduction mediated by PI3Kγ, leading to the activation of downstream signaling mediators. Following liganded receptor activation, the Gprotein complex dissociates from the receptor into α and βγ subunits and from each other, following exchange of GDP for GTP of the Gα subunit. Consequently, various target enzymes are recruited and activated. This is the case for PI3Kγ, PLCβ, and adenylyl cyclases together with the production of a variety of intracellular second messengers such as diaclylglycerol, inositol triphosphate, and PIPs, to name a few. The kinase activation is probably the consequence of the translocation of PI3K from the cytosol to the membrane where its substrate PI(4,5)P2 is located,3 and where other regulatory mechanisms take place. In contrast to PI3Kα and β that are ubiquitously expressed, the expression of PI3Kγ and δ is mainly restricted to the hematopoietic system and probably some additional tissues such as endothelium, muscle, and brain in the case of PI3Kγ. Unfortunately, the lack of specificity, isoform selectivity, and pharmacological profile of the early PI3K inhibitors hampered rigorous in vivo physiological analysis as well as disease-relevant target validation.8 Fortunately, gene targeting in mice has filled the gap and uncovered some of the physiological roles of these powerful signaling enzymes. All four catalytic subunits, as well as the adaptor proteins p85 and p101, have been genetically manipulated. Phenotype analysis of p110α and β knockout mice only reached the level of embryonic development, due to severe early-stage lethality.9 In contrast, mice lacking expression or activity of PI3Kγ and δ are viable and do not show any overt adverse phenotype. Also, mice expressing inactive p110δ (kinase-dead knockin) show selective attenuation of adaptive immune functions, whereas PI3Kδ mutant mice show impaired B-cell and T-cell antigen receptor signaling, suggesting that a selective small molecule kinase inhibitor could effectively suppress B-cell- and T-cell-mediated autoimmunity.10 In addition, mutations in PI3Kα catalytic subunits were recently identified in a significant fraction (25 to 30%) of colorectal cancer, as well as in gastric and a smaller fraction of breast and lung cancers.11 The consistency of hotspot mutations in p110α across diverse

Fishing for Pharmaceutically Relevant PIP-Binding

215

L R

P-Y PTK

β

γ Ras P101

P85

p110γ

p110α, β, δ PI(4,5)P2

PI(3,4,5)P3 PTEN

PIP-BPs

PIP-signal transduction:

PI(3,4)P2 SHIP

PIP-BPs

Location Concentration (ratio) Duration

PIP-BPs

Specificity

FIGURE 10.2 Schematical 3-D representation of the activation of heterodimeric PI3K class IA (α, β, δ) and IB (γ) linked to PIP signaling. Depending on the class, ligand-receptor-binding promotes PI3K activation either by tyrosine phosphorylation (left) or Gβγ subunits association (right). This is followed by the local and transient generation of different PIP products, all timely regulated by the 5’-SHIP and bona fide 3′-PTEN phosphatases. PIP-BPs indicates phosphoinositide-binding proteins.

types of tumour indicates the critical role of PI3Kα in tumorgenesis as much as it establishes the need for selective antagonists of the mutant kinase alleles that will spare normal cells exhibiting normal activity. Although inhibition of the PI3Kδ isoform could have the beneficial effect of reducing autoimmune responses, the additional targeting of the gamma isoform could add therapeutic value. As a matter of fact, PI3K knockout mice show reduced chemoattractant-induced neutrophil respiratory burst and migration to the site of infection.12–14 GPCR-mediated amplification of Fc Receptor I–mediated mast cell degranulation is also severely impaired in these genetically modified mice, resulting in the absence of passive systemic anaphylaxis and clearly linking PI3Kγ to GPCRinduced mast cell activation and function.15 PI3Kγ-defective mice are also protected from angiotensin-mediated high blood pressure, ADP-mediated thromboembolism, and show improved cardiac function in response to stress compared to wild-type animals.9,16 In summary, PI3Kγ and δ play a crucial role in mediating several aspects of innate and adaptive immunity — suggesting that isoform-specific pharmacological

216

Functional Lipidomics

L

R

β

Degradation

γ

Synthesis

PI(4,5)P2 P13K

PLCβ

PI(4,5)P2

PI(4,5)P2 DAG

PIP3

IP3 Tec GAB1/2

Ras

PKC

Akt . . . Ca2+ (family TyrK) GRP1PDK1

PI(3,4)P2

PI(3,5)P2

TAPP1

Vps24p Ent3p

FIGURE 10.3 Schematic 3-D representation of small-Gβα-mediated lipid signaling in response to ligand receptor activation. Depending on the cascade mode (PLCβ or class IB PI3Kγ ), directional catabolic or anabolic routes will be taken via the degradation or recycling of PI(4,5)P2, respectively.

blockade of PI3K would be beneficial for the treatment of acute and chronic inflammatory disease.

INTERFACE BETWEEN PHOSPHOINOSITIDE SIGNALING AND PROTEOMICS Phosphoinositides comprise a family of eight minor membrane lipids that play important roles in many intracellular signal transduction pathways. Moreover, their spatiotemporal production and assembly of lipid–protein complexes is a highly regulated process involving several enzymes and scaffold proteins. Although signaling originating from PI3Ks activity has focused attention on 3′-PIPs,2 all cellular phosphoinositides play a major role in signaling.4,17 Signal transduction through various PIPs has been shown to mediate cell growth and proliferation, apoptosis, and cytoskeletal changes that also lead to cell motility, insulin action, and vesicle trafficking, and recently, also a link to T-cell activation by lipopeptide antigens.18,19 As a matter of fact, phosphoinositide signaling is recognized as a network per se and can lead to severe diseases when perturbed.20 Therefore, to gain insight into phosphoinositide signal transduction

Fishing for Pharmaceutically Relevant PIP-Binding

217

mechanisms in the postgenomic era, an important challenge is the mapping of the celltype-specific proteomes that interact with these lipids.

FUNCTIONAL PROTEOMICS: POTENTIAL AND LIMITATION FOR SIGNALTRANSDUCTION-ORIENTED PROTEIN DISCOVERY Compared to nucleotide-based genetic structures, proteins are several orders of magnitude more challenging to prepare for analysis. Moreover, their functionality often depends on protein–protein interactions, posttranslational modifications (PTM), three-dimensional (3-D) structure, and a myriad of other variables. For a mechanistic yet functional understanding of how proteins interact within a given cell context, the proteins need to be identified. However, despite some progress in protein identification by microsequencing, a significant number of the studies failed when the isolated proteins were submitted to the identification procedure. In particular, it was often not possible to identify proteins that had pivotal roles in the regulation of different cellular processes and were expressed only in low copy numbers. Such critical proteins include regulated enzymes such as kinases, phosphatases, and small G-proteins. For a long time, the main reason for this failure was the limitations in protein microsequencing based on Edman degredation analysis. The real breakthrough in protein identification in the last 10 years came with the introduction of nanoscale technologies derived from gel-based or gel-free capillary electrophoresis21 that allowed proteomics,22,23 previously termed large-scale protein identification,24–26 to become the reference standard for the identification of the dynamic protein content of a given proteome at any given time. It is now recognized that proteomics, similar to largescale DNA sequencing, is making an important contribution to our understanding of biology, diseases, and drug actions and, ultimately, medicine through the global analysis of gene products.27,28 Researchers have realized that merely having complete sequences of the genome is not sufficient to elucidate biological function. The verification of a gene product’s identity and function by proteomics methods has become an important first step in “annotating the genome.” To emphasize further the importance of proteomics in the postgenomic era, the 30,000 to 40,000 genes in the human genome encodes between 10 × 5 to 10 × 6 proteins, largely through alternative mRNA splicing. As a matter of fact, about 60% of human genes seem to produce splice variants (Genomics and Proteomics, September 2003, Vol. 3, No. 7). Moreover, the diversity and complexity that is introduced by PTMs, protein–protein and second messenger–protein interactions is enormous and further increases this percentage by several magnitudes. Even though the statement is provocative and unproven, one could say that the genome-encoded information is unidimensional, whereas the complementary proteome and interactome are multidimensional. Proteomics can be divided into four main areas: 1. Protein display, which results in extensive cataloging of protein spots that originate most often from a given organelle, a cell line, or a tissue of a given organism. It is based on 2-D gel electrophoresis29 used as a support in the separation of the proteins according to their isoelectric point and

218

Functional Lipidomics

molecular weight, as X and Y axes, respectively. Although only descriptive, this technique is still recognized as giving the best representation of the overall protein distribution at a given time. The so-called “2-D PAGE maps references” aim to represent the proteome content of a given protein mixture at a given time. An example of such 2-D maps is accessible at http://us.expasy.org, the first public Web access with interlinked 2-D map databases pioneered by Hochstrasser and colleagues almost 15 yr ago.30–32 2. Microcharacterization for large-scale identification of proteins and their PTMs.21,33 3. Differential display proteomics for comparison of protein levels and modifications33,34 or for archiving data with potential application in a wide range of diseases.35 Today, this approach is less and less combined with protein display (1), but has merged with the second area using real qualitative and quantitative MS-based technologies. 4. Protein–protein interactions, which will be discussed in more depth as it covers much of the functional analysis of gene products or functional genomics, including large-scale identification of proteins21,36–38 and interaction studies (interactome).39,40 Among them, affinity-based approaches have garnered more interest, as they can be coupled to large-scale proteomics identification and do not affect enzymatic activity.41–44 Thus, the link between signaling and proteomics, similar to genomic-linked technologies, is becoming an attractive interdisciplinary research topic.45 Recently, a large-scale affinity-based proteomics approach has identified the complete set of protein interactions involved in a key human immune system signaling pathway namely the TNF-alpha/NF-kappa B pathway. This pathway is implicated in a variety of disease states, including inflammation and cancer. This snapshot of the hundreds of interactions involved provides the most complete and realistic picture to date of how this regulatory system functions, and identifies many potential targets for therapeutic intervention.46 Although this work has paved a new route to the rigorous identification of protein–protein interactions in a given subproteome, it suffers from a significant experimental limitation. Small- or large-scale affinity-based mass spectrometry (MS) proteomics methods used to identify protein–protein interactions generate nonrelevant associations, which consequently make it difficult to dissociate the nonspecific from the functionally relevant protein interactions. In addition, more sophisticated experimental setups are required when intrinsic enzymatic activity needs to be preserved. Needless to say, in proteomics, the identification of proteins per se is not an end, but the starting point of a marathon experiment. In contrast to protein–protein interaction studies, the identification and confirmation of novel PIP-binding proteins has not delivered the same level of information, and is less likely to generate reliable results on a large scale in the near future by following conventional approaches. Up to now, several site-specific mutagenesis studies examining the proper binding of a protein to a given phosphoinositide have been successfully performed in vivo. Conducted by specialists in the lipid field, they were performed at the one-to-one lipid–protein scale and required different and complex biological systems. An illustrative example that combines both protein–pro-

Fishing for Pharmaceutically Relevant PIP-Binding

219

tein interaction and lipid–protein interaction in an environment in which enzymatic activity was conserved was recently published by McManus et al.47 Using the power of molecular genetics to probe the physiological function of these interactions, they established the binding role for the regulatory domains of 3-phosphoinositide-dependent protein kinase 1 (PDK1)-PIP3 towards Akt and downstream effectors. Specifically, they generated homozygous knockin embryonic stem (ES) cells that only express a form of PDK1 with a mutation in its pleckstrin homology (PH) domain that abolishes PIP3 binding without affecting catalytic activity. Akt/PKB was not activated by IGF1, whereas ribosomal S6 kinase (RSK) was activated normally, indicating that PIP3 binding to PDK1 is required for PKB, but not for RSK activation. Moreover, employing PDK1 knockin cells in which either the PIP3-binding or the substrate-docking site, termed the PIF pocket, was disrupted, they established the roles that these domains play in regulating phosphorylation and stabilization of protein kinase C isoforms. In the cellular context of phosphoinositides and protein interactions, the means by which one could isolate, confirm, and map these interactions on a large proteomics scale is rather limited in vitro, and not yet accessible in vivo. When compared with protein–protein interactions, lipid–protein interactions are more difficult to investigate due to their often weaker electrostatic-charge-dependent or hydrophobic associations. Besides a few phosphoinositide–protein interactions that have shown high affinity, most are weak and prone to dissociation.48 Also, the intrinsic low aqueous solubility of these amphiphilic lipids further complicates the assays and limits the biochemical tools that are suitable for the analyses of PIP–protein interactions. From the bioinformatics point of view, the low degree of primary sequence homology of phosphoinositide-interacting protein modules, combined with a limited number of lipid–protein domains, and the relative absence of biological knowledge complicates the application of sequence searching of genomewide databases for proteins that are important in PIP signaling.49 Although successful studies were performed in yeast on PH and PX domains, they are not representative of the overall number of PH domain proteins in humans (there are 33 in yeast compared to 142 known and another probable 252 candidates retrieved from the human genome).50 In short, in silico searches for lipidbinding domains performed on genomewide databases do not allow either the largescale identification or the identification of individual novel PIP-binding proteins. Finally, although unambiguously contributing to large-scale protein identification, bioinformatics methods most often require novel predictive algorithms that also necessitate significant amount of work for the experimental confirmation of the putative candidates. For these reasons, to identify or confirm PIP–protein interactions, researchers have explored alternative experimental strategies. An attempt to summarize the methods used to identify or confirm the interaction between lipid and proteins is shown in Table 10.1. Among them, affinity-based lipid chromatography, coupled with modern proteomics approaches for protein identification, offers considerable promise for the discovery and validation of novel therapeutic targets. To date, two large-scale phosphoinositide-affinity-based proteomics studies have led to the identification of numerous novel potential phosphoinositide-binding proteins. The first study used phosphoinositides containing long acyl chains together with a highly heterogeneous protein source (leukocytes and tissue samples composed

PIP-overlay

PIP–Array/Strips and Protein lipid overlay (PLO)

Approach

Bilayer liposomes used on proteins immobilized on a filter. (Synthetic lipids mostly made of multilamellar or small or large unilamellar vesicles.) Confirmation

Strategy Recombinant protein used on pure lipids immobilized on nitrocellulose filters Preliminary screen Advantages Simple and rapid Simultaneous study of various PIP–protein interactions Specific for strong interacting ligands Commercially available (expensive) Does not requires any lipid protein modules Depending on the assay, provides qualitative information on the relative protein–lipid affinity Can be adapted to a high-throughput format PLO compatible with fluorescent technologies Reproducible Less prominent to binding artifacts Significant cellular relevance when liposomes are in solution. Avidity effect arises from the proximity of multiple PIPs presenting head groups toward the membrane-bound protein target Allow qualitative and semiquantitative analyses Requires expertise and challenging protocols: e.g., following in vitro kinase assay Low throughput Requires homogenous liposomes and/or micelles (containing detergent)

Disadvantages Reproducibility Interaction depends on intrinsic ligand property Does not allow quantitative determination of affinity or physiological lipid composition Low specificity for weak interactions

(Zhu et al., Science 2001.) (Snyder et al., JBC 2001.)

References (Dowler et al., STKE Science 2002) (Buser et al., Meth. Mol. Biol. 1998) (Kavran et al., JBC 1998) (Dowler et al., Biochem. J. 2000)

TABLE 10.1 Illustration of the Pros and Cons of the Various Approaches Used to Identify or Confirm Novel Lipid-Binding Proteins

220 Functional Lipidomics

Affinity chromatography

In vitro expression cloning (IVEC)

Applicable for the isolation of different lipid products or different PIP-binding proteins depending on the method Selective by combining cDNA library technology with affinity isolation techniques Qualitative

Expression library screen with ATPradiolabelled phospholipids or biotinylated PIP-affinity probes and PI3K Isolation and purification of binding Non byes approach nor restrictive to proteins. soluble lipids Phosphoinositide homologues are Highly specific and sensitive used as baits in a cellular Whole-lipid-proteome screening allows preparation comprehensive representation Suitable for proteomics and MS protein identification Applicable to any cell system or tissue Cellular relevance for endogenous molecular interactions Isolation of signaling PIP-protein complexes Commercially available lipid baits Qualitative information of the relative affinity

Synthetic biotinylated 3′phosphoinositide analogs used in conjunction with libraries of radiolabeled proteins that are produced by coupled in vitro transcription/translation reactions (Rao et al., JBC 1999) (Klarlund et al., Science 1997) (Whitley et al., JBC 2003) (Zhou et al., J. of Immunol. 2004) (Gozani et al., Cell 2003) (Chaudhary et al., Biotechnique 1997)

(Kanematsu et al., JBC 1992) (Krugmann s et al., Mol. Cell 2002) (Hirata et al., Biochem. Biophys. Res. Comm. 1990) (Theibert et al., PNAS 1991) (Greenwood et al., J. Cell Biol 2000) (Stricker et al., BBA 2003) (Pasquali et al., PhD thesis, Compiegne University, 2004)

Complicate approach Requires chemistry Restrictive to PI3K lipid products

Requires expertise and challenging protocol settings Heavy data mining due to isolation of massive protein-complex mixtures Requires validation of direct PIP–protein interactions Limited to short soluble baits when analysis is performed in batch Does not allow direct cell-specific assignment of PIP-interacting protein when biological source is made of mixed cells or tissues

Fishing for Pharmaceutically Relevant PIP-Binding 221

Proteome chips

Bioinformatic

Approach

Proteome microarray used as template to screen for interacting proteins and phospholipids

Strategy Quantitative method using statistic to test a hypothesis. Use any known lipid-binding domain to perform genomewide searches. Disadvantages Due to lack of biological knowledge, restricted to a limited number of lipid-protein domains (mostly PX and PH domains) Prediction tool that ultimately requires biochemical confirmation Genomewide studies mostly performed on S.cerevisiae (see Yu et al. and Vollert et al.) Currently structural and sequence alignments could not predict what specific lipid associates to a putative lipid-binding domain Allow to screen for diverse biochemical Most often restricted to yeast activities such as protein–drug, Requires the cloning of thousands protein–lipids, protein–protein, as well of open reading frames and corresponding proteins as enzymatic assays and PTMs Heavy data mining due to isolation of massive protein targets that all requires biochemical confirmations Libraries are tedious to screen Specificity issue: negatively charged lipids tend to associate with basic stretches of proteins

Advantages Allow genomewide screen for lipidbinding proteins when consensus sequences of protein module exist Applicable for the identification of membrane-targeting motifs for lipidbinding proteins containing covalent lipid modifications

(Zhu et al., Science 2001)

References (Dowler et al., Biochem. J. 2000) (Yu et al., JBC 2001) (Yu et al., Mol. Cell 2004) (Vollert et al., MCP 2004) (Eizenberg et al., Nature 2000) (Bateman et al., Nucl. Acids. Res. 2004) Hurley et al., Ann. Rev. Biophys. Struct. 2000)

TABLE 10.1 (continued) Illustration of the Pros and Cons of the Various Approaches Used to Identify or Confirm Novel Lipid-Binding Proteins

222 Functional Lipidomics

Gel Filtration Chromatography (SEC)

SELDI

Photoaffinity labeling

Enables the direct probing of target Identification of interacting partners in protein through a covalent bond protein mixtures introduced between a ligand and Determination of the relative affinity of its specific receptor different PIPs for a given protein and the localization of the PIP-binding sites Allow identification of binding protein Use of photoprobe to selectively modify the hydrophobic binding subunits in oligomeric complexes regions in target macromolecules. Selective and sensitive. Large panel of affinity probes Often combined with affinity chromatograhy Great potential in drug discovery and development processes: i) screening of early leads ii) structural information at the contact point of drugs with receptors Surface-enhanced laser desorption Rapid capture of specific binding ionization-mass proteins from complex samples. spectrometry(SELDI-MS). Ideal for small sample volumes (microliter) Based on sensor chip chemistries: use of biospecific interactive Biophysical measurement of binding surfaces. affinity and ligand selectivity Radiolabeled soluble inositol head Easy readout (radiolabeled Pic) group incubated with a ligand and Qualitative fractionated by SEC Selective Mostly used to confirm an interaction Identification of captured proteins is Reviewed in (Tang et al., Mass difficult. Spectrom. Rev 2004) Cannot handle large sample volumes Requires dedicated reader Expensive (Virbasius, PNAS 2001) Low throughput So far, limited use with the PX and (Lemmon, PNAS 1995) PH domains.

Sometime difficult to reconcile with (Theibert et al., JBC 1992) theory (Mourey et al., Biochemistry 1993) (Sciora et al., EMBO 1999) (Whitley et al., Eur. J. Cell Biol. 2002) (Li et al., J. Cell Biol. 2002) Reviewed in (Gartner et al., Cuu Med. Chem. 2003) Reviewed in (Hatanaka, Curr Top Med Chem. 2002)

Fishing for Pharmaceutically Relevant PIP-Binding 223

isothermal titration calorimetry

and

Surface plasmon resonance (SPR)

Also, allows measurement of kinetic/thermodynamic association and dissociation of two components as well as enthalpies and entropies of binding for (calorimetry)

Advantages Selective and sensitive Identification of new PIP-binding motives. Identification of new downstream targets of PI3K. Qualitative Uses Biosensor chips coated with a Highly selective. lipid monolayer of the test lipid. Provide qualitative and quantitative Affinity interaction measured by solution affinities the change in the defractive index Cellular relevance for molecular across the chip. interactions Optimized for large-scale analysis of interaction proteomics Allow assessment of binding and dissociation constants Protein–lipid interaction detected Biophysical measurement of binding by measuring temperature changes affinity and ligand selectivity

Strategy Screen for new cDNAs that bind Targets of PI3K Identification PI3K products System (TOPIS)

Approach

References (Isakoff et al., EMBO 1998)

(Kavran et al., JBC 1998) Reviewed in (McDonnell, Curr. Opi. Chem. Biol. 2001) (James et al., Biochem J. 1996) (Currie et al., Biochem. J. 1999) Hodgkin et al., Curr. Biol. 2000) (Stahelin et al., JBC 2003)

Disadvantages Restrictive to PI3K products and PH-domain-containing proteins Heterologous system for nonyeast PIP-binding proteins interactions

Limited to interactions with head groups or soluble lipid analogs Requires expertise and expensive device Can generate false negative due to surface change interferences or misinterpretation Requires millig. (isothermal) and microg. (SPR) of protein.

TABLE 10.1 (continued) Illustration of the Pros and Cons of the Various Approaches Used to Identify or Confirm Novel Lipid-Binding Proteins

224 Functional Lipidomics

Ultracentrifugation

Measure the binding of peptides and Qualitative and quantitative assessment proteins to loaded phospholipid of the partitioning of lipid-binding vesicles domains with bilayer vesicles of defined composition Suitable for any phospholipid vesicles High in vivo cellular relevance Highly selective Can be combined with in vitro kinase assay

Requires expertise (Buser et al., Methods Mol. Long and complex method that first Biol. 1998) requires the enrichment of targeted (Stephens et al., Science 1998) membrane organelle Low throughput

Fishing for Pharmaceutically Relevant PIP-Binding 225

226

Functional Lipidomics

of different cell types).51 The second study used soluble phosphoinositides with shorter acyl chains together with well-defined primary cell material. The affinity isolation step was combined with protein-activity-profiling methods to generate a comprehensive PIP-signaling proteomics technology platform (see Section titled Mining Phosphoinositide-Binding Proteins by Chemical Proteomics).

PHOSPHOINOSITIDE–PROTEIN INTERACTIONS: SPECIFIC OR PROMISCUOUS? The affinity of a phosphoinositide for a given protein is dominated by the stereoelectronic complementarities of the head group with a binding pocket. The affinities and selectivities vary from high and specific to weak and promiscuous. Phosphoinositides can relocalize the protein, e.g., from the cytosol to a docking site at the plasma membrane, can facilitate the vesicular trafficking via the interaction of specific protein modules, can initiate the severing or capping effects of some proteins to trigger cytoskeletal changes, as well as facilitate the assembly of large, multicomponent protein complexes.52–55 The association of phosphoinositides with proteins is usually based on electrostatic interactions that lack known primary sequence determinants.48 Mostly, lipid–protein interactions are mediated by two known structural mechanisms, both strictly dependent on protein conformation. One mechanism is based on the accessible electrostatic charges of the protein that most often display promiscuous lipid–binding tendencies, which can vary from one to another depending on its structure or the steric environment. Although the association can be purely based on a signaling cascade, e.g., turning on gene transcription, it can also take part in structural protein–protein association, e.g., mediated by the highly expressed cytoskeletal, multifunctional PI(4,5)P2 lipid substrate. This first mechanism applies for actin cytoskeletal reorganization, in which PI(4,5)P2 is bound to severing and capping proteins that comprise focal adhesion, vesicle trafficking, filament crosslinking, and F-actin-binding proteins. The association can take place with different cellular PIPs or with PI3K itself, like for profiling, which bind to the p85 regulatory subunit of PI3K and enhances the ability of the kinase to generate PIPs.56,57 Conversely, the second structural mechanism of lipid–protein interaction is based on the spatiotemporal assembly of signaling-protein complexes or cascades. Signaling complexes form by translocation of proteins that contain single or multiple lipidbinding domains that recognize specific phosphoinositides.49,58 These include PH and Dbl homology (DH) domains; plant homeodomain (PHD); the Zn-finger family originating from Fab1-p, YOTB, Vac1p, and early endosome antigen 1 EEA1 (FYVE); PSD-95, Dlg, Zo-1 (PDZ); Ca2+-dependent or -independent (C2); epsin N-terminal homology (ENTH); gelsolin; profilin; PI-kinase and PIP phosphatases; Phox homology (PX); and phosphotyrosine-binding domains (PTB), respectively.48,49,58,59 In some cases, the interaction of phosphoinositides with these binding modules can exclusively depend on an organelle compartment that fulfils different functions. An example in which head group–protein and acyl chain–protein interactions contribute to affinity is the association of phosphatidylinositol 3-phosphate (PI3P) with FYVE-domain-containing proteins.60 PI3P is generated on early endosomal and

Fishing for Pharmaceutically Relevant PIP-Binding

227

phagosomal membranes by PI3Ks and serves important regulatory functions in phagocytosis, endocytic traffic, receptor signaling, and microbial killing through the recruitment and activation of a number of effector proteins, almost all containing FYVE or PX domains.61–63 Recently, in a study examining the fate of the downregulated EGF receptor (EGFR) and bulk-transport markers in the endosomal pathway, Gruenberg and colleagues showed that transport and sorting can be uncoupled in the pathway. Their data showed that PI3P signaling did not regulate the core machinery of endosome biogenesis and transport, but controlled the sorting of downregulated receptor molecules in early endosomes via hepatocyte-growth-factor-regulated tyrosine kinase substrate (Hrs).64 Apart from the example of PI3P, lipid-binding protein modules do not necessarily bind to the same PI nor do they need to be targeted to the plasma membrane as a prerequisite. In the case in which membrane targeting is not required for protein activation, the nonspecific electrostatic attraction to the membrane surface will also participate significantly in binding the protein to the membrane surface. Although this was already modeled more than 15 yr ago by Ferguson and colleagues65 using PLC-δ PH binding to the surface of a membrane containing PI(4,5)P2, this is still true for many proteins that contain PH domains.66 To add one more level of complexity, the protein function or the primary protein domain structure composition can also vary depending on its origin. For instance, one isoform of the mouse-p21-activated kinase (Pak2), recently proposed as a new phosphoinositide-binding protein kinase,67 lacks the PH domain homology module found in the corresponding Dictyostelium discoideum protein68 but still binds to the PI3K product PI(3,4)P2. Lipid–protein interactions may also differ among cells from different tissues. Here, this diversity again depends on the presence of additional proteins, the pH, and the cellular protein distribution. Independent of the presence of specific lipid–protein modules, the lipid–protein interaction can induce a conformational change in the protein, thereby modulating its activity or downstream signaling cascades, such as those mediated by phosphorylation. For example, the PIP-interacting PH-domain-containing protein kinases and nucleotide exchange factors regulating small GTPases have multiple functions. The activities of these bona fide signaling proteins are in many cases regulated by PI3K signaling.

PROTEIN-DOMAIN-MEDIATED SIGNALING: AN EXAMPLE THE PH DOMAIN

OF

As discussed in the preceding text, phosphoinositides and their derivatives interact with many different cellular proteins via a variety of lipid-binding modules. Among them are PH domains, to date the most characterized, and composed of 120 amino acid residue motifs; it has been recognized in over 140 proteins derived from all eukaryotes.48,50,69 Composed of a small β-sandwich and a single alpha helix, this domain was first identified as a twice-repeated pattern in pleckstrin, the major protein kinase C in platelets.70 From the primary sequence point of view, there are few positions at which the identity of the amino acid is reasonably well conserved between PH domains, and only one residue (the tryptophan in the alpha helix) is absolutely conserved. Rather, as shown in the Figure 10.4, PH domains are defined

228

Functional Lipidomics

FIGURE 10.4 Ribbon representation of DAPP1/PHISH-PH domain associating with PIP3. The affinity for a PH domain of a given phopshoinositide compared to another can vary considerably depending on the extended β-sheets forming the PH domain. Usually, these extensions restrict binding access to a given inositol- or phosphoinositide-phosphate group.

and identified on the basis of six sequence-homology blocks (5 to 7 β-sheets and one α-helix at the C-terminal) that have a reasonably conserved pattern of hydrophobic and hydrophilic amino acids, further separated by various stretches of amino acids. For instance, dynamin, spectrin, pleckstrin, Btk, and phospholipase D, all share the same basic structure of seven antiparallel β-sheets forming a hydrophobic pocket that is capped by a carboxy-terminal amphipathic helix.71 The structures of only a few lipid-selective PH domains have been elucidated by NMR and x-ray crystallography. The vast majority are not well characterized and commonly exhibit a weak affinity and low specificity for a given phosphoinositide.50,72 Indeed, the majority of PH domains have quite different functions that may not involve protein–lipid interactions.73 PH domains can vary greatly in specificity and affinity depending on the protein. At one extreme, some bind very strongly and specifically to a single phosphoinositide, and most often, are necessary and sufficient for signal-dependent recruitment of their host proteins to the plasma membrane. This is the case for the centaurin-α, Grp1, Akt/PKB, and phosphoinositide-dependent kinase 1 (PDK1) that bind specifically PI(3,4)P2 and PIP3,49,74 Bruton tyrosine kinase family (Btk) that bind specifically to PIP3, PLC-δ, and β-ARK, and spectrin that bind to PI(4,5)P2 as well as to PI(3,4)P275–77 to name only few. In the case of Grp1, this is even reinforced due to an extended β6/β7 loop that selectively restricts its binding to 5’-lipids by covering the 5-phosphate group.4 Conversely, as shown in the Figure 10.4, this is not the case for some other lipid-binding proteins such as DAPP1, which has a dual selectivity for PI(3,4)P2 and PIP3.78 Also, in some cases, phosphoinositide-binding enhances or

Fishing for Pharmaceutically Relevant PIP-Binding

229

inhibits functional activity. Thus, Vav, a member of the guanine nucleotide exchange factor family, binds to, and is activated by, PIP3 and PI(3,4)P2, whereas it is inhibited by PI(4,5)P2.79 At the other extreme, there are PH domains (such as the N-terminal domain from pleckstrin) that bind many different phosphoinositides with both low affinity and poor selectivity.71 In contrast to the high-affinity binding proteins, these protein modules are necessary but not sufficient for membrane targeting and may require cooperation with other parts of the protein (e.g., oligomerization) or the cooperation of other associated proteins (H.L. Yin, personal communication). Finally, PH domains show strong electrostatic polarization that allows the binding to negatively charged surfaces.65,80,81 They associate with lipids by means of electrostatic interactions within a small hydrophilic pocket that recognizes the arrangement and number of phosphates on the phosphoinositide. The affinity of a PH domain for a particular phosphoinositide can have significant physiopathological consequences, as previously demonstrated for Akt/PKB with a PI3K lipid product.82 By inducing a mutation in the PH domain of Akt/PKB, Stocker and colleagues demonstrated that a reduction in its affinity was sufficient to restore viability in flies lacking the 3′ phosphatase PTEN (see also Figure 10.6). Taken together, the specific and promiscuous interactions that exist between phosphoinositides and proteins depend on various parameters. Phosphoinositide binding to PH domains or other lipid-binding modules regulates protein activity, protein targeting, and signaling-complex formation. Also, phosphoinostides coordinate events in actin cytoskeletal reorganization, e.g., PI(4,5)P2 modulation of severing and capping via interactions with profilin and gelsolin. Finally, due to the heterogeneity of function of some lipid-binding protein modules such as the DH and PH domains, it has been difficult to unambiguously assign their binding specificity and/or promiscuity towards phosphoinositides. Nevertheless, whatever the means by which a given protein associates with lipids, we should not forget that PIPs are as dynamic as signaling-protein complexes and are also transient. As a matter of fact, PIPs can mediate both, specific and promiscuous protein interactions.

PHOSPHOINOSITIDE-DEPENDENT PROTEIN KINASES: OF AKT/PKB

THE

EXAMPLE

Akt was originally identified as the oncogene transduced by the acute transforming retrovirus (Akt-8) that was isolated from AKR thymoma.83–85 The sequence analysis of the viral oncogene and its cellular homolog revealed that it encoded a serine–threonine protein kinase, composed of a carboxy-terminal kinase domain and an aminoterminal PH domain.85 Almost at the same time, two other groups cloned it as well and named it RAC or PKB.86 Akt/PKB includes three known members (Akt 1–Akt 3), and the kinase is evolutionarily conserved in all eukaryotes from Dictyostelium to humans. The finding that mutation of the Akt/PKB PH domain blocked its activation by growth factors in vivo and by PI3K-generated phosphoinositides in vitro87 suggested that the activation of Akt/PKB may be regulated via the binding of 3′-phosphoinositides to its PH domain. Therefore, the discovery nearly a decade ago that

230

Functional Lipidomics

Akt/PKB is a target of PI3K88 generated enormous interest because it provided a direct link to the lipid products of PI3K and the downstream targets of PI3K signaling. Importantly, phosphoinositides enabled Akt/PKB activation by facilitating its phosphorylation in the activation loop. Concomitant with this, PDK1, which also carries a PH domain, is constitutively active and bound to the plasma membrane. Whereas it binds to the same 3′-phosphoinositides, it was also found to phosphorylate and partially activate Akt/PKB at Thr308 in its kinase domain in response to growth factor after its translocation from the cytosol to the plasma membrane.89 The final Akt/PKB activation was shown to be mediated by a second phosphorylation at Ser473 in the tail by an as yet unknown kinase termed PDK2.89,90 Subsequent studies that revealed several other PH-domaincontaining kinases similarly regulated by PI3K via phosphoinositides opened new avenues for seeking and understanding better phosphoinositide signaling, and in particular, 3′-phosphoinositide second messengers and immediate downstream activators. A list of most known PIP3 interacting proteins is accessible at http://stke.sciencemag.org (free access on the address on the Web at: http://stke.sciencemag.org/content/content/vol0/issue2002/images/data/CMP_6557/DC2/CMP_6 557.jpg). In summary, the important role played and the equilibrium maintained by the interaction of cellular phosphoinositide with specific proteins, together with the evidence of a yet unidentified pool of PIP-interactome, provide good reasons to search for novel lipid-binding proteins. To this end, the combination of PIP signaling with proteomics is a valuable strategy to identify and characterize potential, pharmaceutically relevant, signaling targets.

MINING PHOSPHOINOSITIDE-BINDING PROTEINS BY CHEMICAL PROTEOMICS Having reviewed a restricted portion of the phosphoinositide signaling field, the interaction of PIPs with specific proteins as well as the pros and cons of proteomics, we will now discuss a comprehensive approach we recently developed in our laboratory. It combined chemistry, PIP-signaling, and proteomics to selectively capture and identify novel PIP-binding proteins. In contrast to the previous sections, we will only focus on the results we obtained.67 For a broader view of the means by which lipid–protein interactions can be identified or confirmed, refer to Table 10.1. The establishment of a strategy suited to the identification and validation of novel PIP-binding proteins led us to the generation of an interactive skeleton combining different biochemical tools, some of which were new. The core represents the affinity matrices and required the chemical generation of new soluble affinity ligands suitable for large-scale identification of PIP-binding proteins. To preserve the enzymatic activity of the affinity-captured proteins, we further implemented an activity-based technique for the protein kinase profiling of the isolated lipid protein complexes. Finally, for the validation of the PIP–protein interactions, a novel highly specific assay that uses radiolabeled micelles was done. The strategy used for this approach is summarized in Figure 10.5a.

Fishing for Pharmaceutically Relevant PIP-Binding

231

Validation of lipid–protein interaction micelles protein

protein

protein PIP-BP protein

PIP-BP

wash wash PIP-BP

Lipid to protein

Protein to lipid (b)

FIGURE 10.5 (a) Flow diagram of the integrated functional PIP-signaling proteomics strategy. Following homogenization of the biological source, the sample is partitioned by size exclusion chromatography (SEC). The resulting proteome fractions containing protein complexes are subsequently submitted to a two-step lipid-affinity chromatography (LAC): (1) using soluble biotinylated PIP ligands to capture phosphoinositide-binding proteins and (2) using streptavidin-coated beads to immobilize PIP-liganded protein complexes. An example of a PIP bait is PI(3,4)P2. The ball represents the biotin. Purified proteins are then eluted by reductive cleavage from the disulfide bond connecting the biotinylated phosphoinositide to the affinity matrix. Resulting protein complexes are either submitted for protein identification by MS/MS or activity-based assay. Validated novel PIP-binding proteins are further characterized according to standard procedures. (b) Experimental concept of the strategy used to confirm direct PIP-protein interaction. The assessment of the binding of the phosphoinositidebinding protein (PIP-BP) is performed using a membrane coated with different lipids further incubated with the recombinant protein (left). The opposite assay consists of the use of micelles containing specific radiolabeled 3′-PIP messengers produced following in vitro kinase assay using recombinant PI3K. See Table 10.1 for additional details.

232

Functional Lipidomics

(4,

5)P 2

PIP-Ks

PI

PT

3

(3,4

5

)P 2

EN

PIP3

PI

SH

PI5K

P I( 4,5

)P

2

PIP-PPTs

IP

P I(3,4

)P 2

FIGURE 10.6 PIP3, a lipid product of the interplay of phosphoinositide-kinases (PIP-Ks) and phosphatases (PIP-PPTs). PI(4,5)P2 and PI(3,4)P2 serve as PIP-K substrates for the generation of PIP3. Conversely, they also serve as PIP-PPT substrates for the generation of the original kinase substrate that will again be used through the PIP-Ks cycle.

After establishing the concept, the first step was to orient our choice towards a proteome source corresponding to the biological model under study. Two reasons motivated the selection of primary murine macrophages as the model system, the first being their link to PI3K. PI3K-dependent phosphoinositide signaling controls many processes including myeloid lineage differentiation, proliferation, cell survival, motility and directed cell migration, growth, cytokine production, radical oxygen production, and both pinocytosis and phagocytosis.91–94 The second reason was related to the relevance of the use of macrophages in the context of PI3K. As a matter of fact, macrophages that lack either expression or activity of myeloid-specific PI3K isoforms are unable to produce 3′-phosphorylated phosphoinositides (PI(3,4)P2 and PIP3) when stimulated with a variety of cytokines and growth-factor agonists.4,12,16 Therefore, it was anticipated that a signaling-proteomics survey of primary macrophages would not only reveal known and novel, but also physiologically relevant 3′-PIP-interacting proteins. Although the strategy depicted in Figure 10.5a should be suitable for any tissue (and most cellular) proteomes, there is a limitation for its use in primary cell culture, which barely yields 2% of the standard starting material used in previous PIP–protein interaction studies.51,95 This limitation, together with others that are not mentioned here, brought us to develop a more efficient method for the capture and detection of phosphoinositide-interacting proteins. Namely, a two-step lipid-affinity chromatography (LAC) that consists of using a soluble, biotinylated, cleavable, PIP-affinity ligand that can specifically associate with PIP-binding protein complexes and is subsequently released from the streptavidin-coated beads to significantly abolish nonspecific beads interactions (for the PIP chemistry, see Figure 10.5A, left-hand side). This mild elution protocol afforded protein complexes suitable for MS-based technologies (MS/MS) because only the proteins specifically associated with the biotinylated PIP ligands were eluted.

Fishing for Pharmaceutically Relevant PIP-Binding

233

Following the experimental settings, and to confirm the efficacy of the approach, we interrogated the content of the captured PIP–protein complexes. This was done by several means that all linked or added value to the complementary approach depicted in the Figure 10.5. For instance, we first searched for known, endogenouslyexpressed PIP-binding proteins. Here, for the first time, and irrespective of the approach that was used in previous studies, we were able to identify several known PI3K PIP-signaling effectors using specific antibodies (see the following subsection). Also, based on the fact that affinity chromatography is compatible with the capture of enzymatically active protein kinases, we employed different activity-based assays to investigate if our means to release PIP–protein complexes from the matrices did not harm the enzymatic activity of known PIP-protein kinases (with the help of specific phosphoantibodies directed against a given phosphorylated site and/or following in vitro or in-gel kinase experiments). This allowed us to not only confirm the presence of known, but also to identify novel, PIP-binding protein kinases. Nevertheless, the best use of this integrated approach may still be for the discovery of novel PIP-binding proteins using MS-based analysis. In contrast to nonlimited sources, the use of primary macrophages had its own limitation and required an additional step, (i.e., protein partitioning, Figure 10.5a). This allowed us to specifically enhance the protein-complex content used as the source for the LAC prior to microsequencing of the captured signaling proteins. Altogether, more than 80 potential novel PIP-mediated signalosomes were identified and functionally clustered as enzymes, cytoskeletal, and membrane proteins.67 Not only does this result add value to the approach, because most of the PIP-interactome are known to be distributed in these families of proteins, but it also confirms the high dynamics of PIP-signaling that was already illustrated in Figure 10.2 and Figure 10.3. Once novel potential PIP-binding proteins had been identified, subsequent confirmation of direct lipid-binding protein interaction was performed using overlay procedures (see Table 10.1). A scheme illustrating the experimental concept of both assays is shown in the Figure 10.5b. Here, in an attempt to mirror the protein–lipid interactions that takes place at the inner leaflet of the plasma membrane, we describe a novel procedure that uses micelles containing radiolabeled 3′-PIP messengers produced following in vitro kinase assay. To conclude this chapter, in the following subsection we will present examples of selective identification of known and novel PIP-binding proteins kinases done by two different means, namely using immunoblot analysis and activity-based kinase profiling.

SELECTIVE CAPTURE PROTEINS

OF

CELLULAR PHOSPHOINOSITIDE SIGNALING

Among the different proteins that can arise using this complementary approach, the capture of novel PIP-binding protein sequences as well as 3′-PIP-binding protein kinases (PIP-Ks) and phosphatases (PIP-PPTs) certainly represents the most attractive signaling targets. As a matter of fact, besides the significant impact of novel protein sequences, the interplay of PIP-Ks and PIP-PPTs puts 3′-PIPs, and in particular PIP3 (see Figure 10.6), at the center of physiopathologically relevant signaling processes, most of which have been extensively documented in the previous sections.

234

Functional Lipidomics

Although both would require the proper setting of more cellular and experimental parameters, the pivotal role they could play in PI3K-mediated phosphoinositide signaling and their known low expression (in particular for PKs and PPTs) posed a challenge to validate our new experimental strategy. We will discuss the successful identification of some of them in the following text. Among them are the oncogenetransformed protein kinase Alt/PKB, and also PDK1, integrin-linked kinase (Ilk), as well as Tec kinase. Although they all contain a PH domain that could be of use for their capture by different approaches (see Table 10.1), until the present study, there has been no report of their isolation in mammalian cells either using affinity-based techniques 5 1 , 7 8 , 9 6 or other global genomewide or protein-chip-based approaches.49,97–99 Accordingly, the LAC of Akt/PKB and its status and properties upon binding to the soluble small PIP molecules will be discussed as the first example. For a more detailed analysis, refer to the original work.67 Following the endogenous capture of Akt/PKB by the two-steps LAC (see Figure 10.5a), two key parameters were further investigated, namely sensitivity and selectivity of the new chemically modified thiol-containing phosphoinositides (SS-PIPs). To this end, and using Akt/PKB as readout, a competition experiment was performed to test the binding for three different PIP baits using increasing concentrations of their corresponding nonbiotinylated derivatives. In accordance with earlier literature,100 Akt/PKB selectively associated in a concentration-dependent manner with SS-PI(3,4)P2 and SS-PIP3 ligands, but not with SS-PI(4,5)P2 used as the internal negative control. To further evaluate the SS-PIP LAC efficacy, we confirmed the affinity isolation of other known PI3K-regulated phosphoinositide-interacting proteins. PDK1 was successfully identified with a lipid selectivity for PI(3,4)P2 and PIP3,89 as well as Ilk, which was not previously identified as a PIP3-binding protein in a cell-based assay,66 and Vav-1 exchange factor that is exclusively expressed in hematopoietic cells and preferentially binds to PIP3.79 Similarly, Tec kinase as well as the Tec-family kinase Btk72 displayed selective binding for PIP3. Only Btk had previously been purified using lipid-affinity matrices on a preparative scale from porcine leukocyte cytosol.51 In summary, several independent experiments showed that Akt/PKB and PDK1 recognized PI(3,4)P2 and PIP3 with similar affinity, whereas Tec, Btk, Ilk, and Vav preferentially, if not selectively, associated with PIP3. Following this, and knowing that the catalytic domain of Akt/PKB is distal from the PH domain, we also wondered what form of Akt/PKB was binding to the small molecule baits. Using the activity-based approach, we could show that Akt/PKB was able to bind SS-PIP independently of its enzymatic status. More precisely, following ligand receptor activation (see Figure 10.2 and Figure 10.3), affinity-isolated PIP–protein complexes containing active Akt/PKB showed the ability to phosphorylate GSK3, its known direct substrate. A second example of selective PIP-protein kinase identification was also successfully conducted using a more sophisticated activity-based approach (see Figure 10.5a). Using a technique that immobilizes a substrate on the gel matrix, we could localize the presence of p21-activated kinase (Pak2) by its ability to phosphorylate p47-phox peptide, its direct substrate, identifying at the same time a new kinase able to bind phosphoinositides. A further assessment of its interaction with PI(3,4)P2 and phosphatidic acid (PA) by using overlay techniques (see Figure 10.5b) confirmed

Fishing for Pharmaceutically Relevant PIP-Binding

235

the previous finding. Following these two different means of PIP-binding protein identification, we were then ready to use the strategy for the identification of novel PIP-binding proteins following MS-based analysis. Here, the critical point was the evaluation of the cleavable bait. Therefore, we first compared the efficiency of the newly synthesized soluble SS-PIPs with the analogous ligand lacking the cleavable disulfide bond already which had been used in a previous study.78 As expected, only the cleavable PIP bait (SS-PIPs) allowed the selective isolation of potential protein candidate bands confirming its practical use for proteomic-based approaches. In contrast, the conventional bait displayed virtually the same protein pattern as the all-carbon tethered bait control, and thus could not provide samples suitable for protein identification by analytical MS-based analysis. Finally, and as already mentioned earlier in the text, the use of this strategy permitted the identification of more than 80 different PIP-binding protein complexes some of which have been confirmed as novel PIP-interacting proteins. Although this example was based on the use of 3′-PIPs and provided selective access to cellular subproteomes, it should also be suitable for the use of other phosphoinositides. Finally, and of equal importance, the greatly enhanced signal-to-noise ratio and direct access to analytical MS-analysis achieved using reductively-cleavable LAC matrices may bypass the current requirement for scale-up sample preparation prior to the affinity separation and subsequent microsequencing.

OUTLOOK Altogether, we hope this review will help to establish a conceptual framework to extend our current thinking in studying signal transduction beyond protein- or DNAbased technologies towards functional proteomics using “small molecules” for baitdesign. More insights into the lipid-interacting proteome will ultimately strengthen our understanding of how molecular components such as phospholipids, functionally as well as mechanistically, work together with target proteins in complex biological systems. Elucidation of the secret of genetic information has been a successful endeavor of the last millennium. Now, the challenge is to deconvolute the variety and complexity of the proteome. Although the proteomic information in the life sciences might be infinite, the task is definitely fascinating.

ACKNOWLEDGMENTS We would like to thank Christopher Hebert for preparing the artwork, in particular Figure 10.1 to Figure 10.3, Figure 10.5, and Figure 10.6, as well as Jasna Klicic for preparing Figure 10.4.

References 1. Fruman, D.A., Meyers, R.E., Cantley, L.C. Phosphoinositide kinases. Annu. Rev. Biochem. 1998, 67, 481–507.

236

Functional Lipidomics

2. Leevers, S.J., Vanhaesebroeck, B., Waterfield, M.D. Signalling through phosphoinositide 3-kinases: the lipids take centre stage. Curr. Opin. Cell Biol. 1999, 11(2), 219–225. 3. Vanhaesebroeck, B., Waterfield, M.D. Signaling by distinct classes of phosphoinositide 3-kinases. Exp. Cell Res. 1999, 253(1), 239–254. 4. Vanhaesebroeck, B., Leevers, S.J., Ahmadi, K. et al. Synthesis and function of 3phosphorylated inositol lipids. Annu.Rev.Biochem. 2001, 70, 535–602. 5. Tolias, K.F., Cantley, L.C. Pathways for phosphoinositide synthesis. [Review]. Chem. Phys. Lipids 1999, 98(1-2), 69–77. 6. Toker, A., Cantley, L.C. Signalling through the lipid products of phosphoinositide-3OH kinase. Nature 1997, 387(6634), 673–676. 7. Cantley, L.C. The phosphoinositide 3-kinase pathway. Science 2002, 296(5573), 1655–1657. 8. Drees, B.E., Mills, G.B., Rommel, C., Prestwich, G.D. Therapeutic potential of phosphoinositide 3-kinase inhibitors. Expert Opinion on Therapeutic Patents 2004, 14(5) 703–732. 9. Vanhaesebroeck, B., Rohn, J.L., Waterfield, M.D. Gene targeting: attention to detail. Cell 2004, 118(3), 274–276. 10. Okkenhaug, K., Vanhaesebroeck, B. PI3K in lymphocyte development, differentiation and activation. Nat. Rev. Immunol 2003, 3(4), 317–330. 11. Samuels, Y., Wang, Z., Bardelli, A. et al. High frequency of mutations of the PIK3CA gene in human cancers. Science 2004, 304(5670), 554. 12. Hirsch, E., Katanaev, V.L., Garlanda, C. et al. Central role for G protein-coupled phosphoinositide 3-kinase gamma in inflammation [see comments]. Science 2000, 287(5455), 1049–1053. 13. Li, Z., Jiang, H., Xie, W., Zhang, Z., Smrcka, A.V., Wu, D. Roles of PLC-2 and -3 and PI3K in chemoattractant-mediated signal transduction. Science 2000, 287(5455), 1046–1049. 14. Sasaki, T., Irie-Sasaki, J., Horie, Y. et al. Colorectal carcinomas in mice lacking the catalytic subunit of PI(3)Kgamma. Nature 2000, 406(6798), 897–902. 15. Laffargue, M., Calvez, R., Finan, P. et al. Phosphoinositide 3-kinase gamma is an essential amplifier of mast cell function. Immunity. 2002, 16(3), 441–451. 16. Patrucco, E., Notte, A., Barberis, L. et al. PI3Kgamma modulates the cardiac response to chronic pressure overload by distinct kinase-dependent and -independent effects. Cell 2004, 118(3), 375–387. 17. Toker, A. Phosphoinositides and signal transduction. Cell Mol Life Sci 2002, 59(5), 761–779. 18. Zhou, D., Cantu, C., III, Sagiv, Y. et al. Editing of CD1d-bound lipid antigens by endosomal lipid transfer proteins. Science 2004, 303(5657), 523–527. 19. Moody, D.B., Young, D.C., Cheng, T.Y. et al. T cell activation by lipopeptide antigens. Science 2004, 303(5657), 527–531. 20. Pendaries, C., Tronchere, H., Plantavid, M., Payrastre, B. Phosphoinositide signaling disorders in human diseases. FEBS Lett. 2003, 546(1), 25–31. 21. Aebersold, R., Mann, M. Mass spectrometry-based proteomics. Nature 2003, 422(6928), 198–207. 22. Wilkins, M.R., Pasquali, C., Appel, R.D. et al. From proteins to proteomes: large scale protein identification by two-dimensional electrophoresis and amino acid analysis. Biotechnology (N.Y.) 1996, 14(1), 61–65.

Fishing for Pharmaceutically Relevant PIP-Binding

237

23. Wilkins, M.R., Williams, K.L, Appel, R.W, and Hochstrasser, D.F. (Eds.). MRWKLARDHDF. Proteome Research: New Frontiers in Functional Genomics. Springer-Verlag, Berlin, 1997. 24. Celis, J.E., Gromov, P., Ostergaard, M. et al. Human 2-D PAGE databases for proteome analysis in health and disease: http://biobase.dk/cgi-bin/celis. FEBS Lett. 1996, 398(2-3), 129–134. 25. Anderson, N.G., Anderson, N.L. Twenty years of two-dimensional electrophoresis: past, present and future. Electrophoresis 1996, 17(3), 443–453. 26. Yarmush, M.L., Jayaraman, A. Advances in proteomic technologies. Annu. Rev. Biomed. Eng 2002,4: 349–373. Epub March 22. 27. Pandey, A., Mann, M. Proteomics to study genes and genomes. Nature 2000, 405(6788), 837–846. 28. de Hoog, C.L., Mann, M. Proteomics. Annual Review of Genomics and Human Genetics 2004, 5(1), 267–293. 29. O’Farrell, P.H. High resolution two-dimensional electrophoresis of proteins. J. Biol. Chem. 1975, 250(10), 4007–4021. 30. Hochstrasser, D.F., Appel, R.D., Vargas, R. et al. A clinical molecular scanner: the Melanie project. MD Comput. 1991, 8(2), 85–91. 31. Hochstrasser, D.F., Frutiger, S., Paquet, N. et al. Human liver protein map: a reference database established by microsequencing and gel comparison. Electrophoresis 1992, 13(12), 992–1001. 32. Appel, R.D., Sanchez, J.C., Bairoch, A. et al. The SWISS-2DPAGE database of twodimensional polyacrylamide gel electrophoresis, its status in 1995. Nucleic Acids Res. 1996, 24(1), 180–181. 33. Mann, M., Jensen, O.N. Proteomic analysis of post-translational modifications. Nat. Biotechnol. 2003, 21(3), 255–261. 34. Gygi, S.P., Rist, B., Gerber, S.A., Turecek, F., Gelb, M.H., Aebersold, R. Quantitative analysis of complex protein mixtures using isotope-coded affinity tags. Nat. Biotechnol. 1999, 17(10), 994–999. 35. Taylor, C.F., Paton, N.W., Garwood, K.L. et al. A systematic approach to modeling, capturing, and disseminating proteomics experimental data. Nat. Biotechnol. 2003, 21(3), 247–254. 36. Shevchenko, A., Jensen, O.N., Podtelejnikov, A.V. et al. Linking genome and proteome by mass spectrometry: large-scale identification of yeast proteins from two dimensional gels. Proc. Natl. Acad. Sci U.S.A 1996, 93(25), 14440–14445. 37. Pandey, A., Andersen, J.S., Mann, M. Use of mass spectrometry to study signaling pathways. Sci STKE 2000, 2000(37), L1. 38. Andersen, J.S., Rappsilber, J., Steen, H. et al. Multi-protein complexes studied by mass spectrometry. ScientificWorld J. 2002, 2(1 Suppl. 2), 91–92. 39. Blagoev, B., Kratchmarova, I., Ong, S.E., Nielsen, M., Foster, L.J., Mann, M. A proteomics strategy to elucidate functional protein-protein interactions applied to EGF signaling. Nat. Biotechnol. 2003, 21(3), 315–318. 40. Patterson, S.D., Aebersold, R.H. Proteomics: the first decade and beyond. Nat. Genet. 2003, 33 Suppl: 311–323. 41. Leonoudakis, D., Conti, L.R., Anderson, S. et al. Protein trafficking and anchoring complexes revealed by proteomic analysis of inward rectifier potassium channel (Kir2.x)-associated proteins. J. Biol. Chem. 2004, 279(21), 22331–22346. 42. Marles, J.A., Dahesh, S., Haynes, J., Andrews, B.J., Davidson, A.R. Protein-protein interaction affinity plays a crucial role in controlling the sho1p-mediated signal transduction pathway in yeast. Mol Cell 2004, 14(6), 813–823.

238

Functional Lipidomics

43. Godl, K., Wissing, J., Kurtenbach, A. et al. An efficient proteomics method to identify the cellular targets of protein kinase inhibitors. PNAS 2003, 100(26), 15434–15439. 44. Kalume, D.E., Molina, H., Pandey, A. Tackling the phosphoproteome: tools and strategies. Curr. Opin. Chem. Biol 2003, 7(1), 64–69. 45. Stagljar, I. Finding partners: Emerging protein interaction technologies applied to signaling networks. Sci. STKE 2003, 2003(213), e56. 46. Bouwmeester, T., Bauch, A., Ruffner, H. et al. A physical and functional map of the human TNF-alpha/NF-kappa B signal transduction pathway. Nat. Cell Biol 2004, 6(2), 97–105. 47. McManus, E.J., Collins, B.J., Ashby, P.R. et al. The in vivo role of PtdIns(3,4,5)P(3) binding to PDK1 PH domain defined by knockin mutation. EMBO J 2004, May 19; 23(10), 2071–2082. 48. Lemmon MA. Phosphoinositide recognition domains. Traffic. 2003, 4(4), 201–213. 49. Yu, J.W., Lemmon, M.A. Genome-wide analysis of signaling domain function. Curr. Opin. Chem. Biol. 2003, 7(1), 103–109. 50. Yu, J.W., Mendrola, J.M., Audhya, A. et al. Genome-wide analysis of membrane targeting by S. cerevisiae pleckstrin homology domains. Mol. Cell 2004, 13(5), 677–688. 51. Krugmann, S., Anderson, K.E., Ridley, S.H. et al. Identification of ARAP3, a novel PI3K effector regulating both Arf and Rho GTPases, by selective capture on phosphoinositide affinity matrices. Mol. Cell 2002, 9(1), 95–108. 52. Brazil, D.P., Yang, Z.Z., Hemmings, B.A. Advances in protein kinase B signalling: AKTion on multiple fronts. Trends Biochem Sci 2004, 29(5), 233–242. 53. Ellson, C.D., Andrews, S., Stephens, L.R., Hawkins, P.T. The PX domain: a new phosphoinositide-binding module. J Cell Sci 2002, 115(6), 1099–1105. 54. McGough, A.M., Staiger, C.J., Min, J.K., Simonetti, K.D. The gelsolin family of actin regulatory proteins: modular structures, versatile functions. FEBS Lett. 2003, 552(2-3), 75–81. 55. Janmey, P.A., Lindberg, U. Cytoskeletal regulation: rich in lipids. Nat. Rev. Mol Cell Biol. 2004, 5(8), 658–666. 56. Bhargavi, V., Chari, V.B., Singh, S.S. Phosphatidylinositol 3-kinase binds to profilin through the p85 alpha subunit and regulates cytoskeletal assembly. Biochem Mol Biol Int. 1998, 46(2), 241–248. 57. Singh, S.S., Chauhan, A., Murakami, N., Chauhan, V.P. Profilin and gelsolin stimulate phosphatidylinositol 3-kinase activity. Biochemistry 1996, 35(51), 16544–16549. 58. Fruman, D.A., Rameh, L.E., Cantley, L.C. Phosphoinositide binding domains: embracing 3-phosphate 2. Cell 1999, 97(7), 817–820. 59. DiNitto, J.P., Cronin, T.C., Lambright, D.G. Membrane recognition and targeting by lipid-binding domains. Sci. STKE 2003, 2003(213), re16. 60. Kutateladze, T.G., Capelluto, D.G.S., Ferguson, C.G. et al. Multivalent Mechanism of Membrane Insertion by the FYVE Domain. J. Biol. Chem. 2004, 279(4), 3050–3057. 61. Birkeland, H.C., Stenmark, H. Protein targeting to endosomes and phagosomes via FYVE and PX domains. Curr. Top. Microbiol. Immunol. 2004, 282, 89–115. 62. Kanai, F., Liu, H., Field, S.J. et al. The PX domains of p47phox and p40phox bind to lipid products of PI(3)K. Nat. Cell Biol. 2001, 3(7), 675–678. 63. Sato, T.K., Overduin, M., Emr, S.D. Location, Location, Location: Membrane Targeting Directed by PX Domains. Science 2001,294(5548),1881–85.

Fishing for Pharmaceutically Relevant PIP-Binding

239

64. Petiot, A., Faure, J., Stenmark, H., Gruenberg J. PI3P signaling regulates receptor sorting but not transport in the endosomal pathway. J. Cell Biol. 2003, 162(6), 971–979. 65. Ferguson, K.M., Kavran, J.M., Sankaran, V.G. et al. Structural basis for discrimination of 3-phosphoinositides by pleckstrin homology domains. Mol. Cell 2000, 6(2), 373–384. 66. Shamshad, C. Biology of Phosphoinositides. New York: Oxford University Press, 2000. 67. Pasquali C. Identification and Characterization of Novel Phosphoinositide-Binding Proteins by Functional Signaling Proteomics. Ph.D. thesis, Compiegne University, UTC, Compiegne, France, 2004. 68. Merlot, S., Firtel, R.A. Leading the way: directional sensing through phosphatidylinositol 3-kinase and other signaling pathways. J. Cell Sci. 2003, 116(Pt. 17), 3471–3478. 69. Dowler, S., Currie, R.A., Campbell, D.G. et al. Identification of pleckstrin-homologydomain-containing proteins with novel phosphoinositide-binding specificities. Biochem. J. 2000, 351(Pt. 1), 19–31. 70. Tyers, M., Rachubinski, R.A., Stewart, M.I. et al. Molecular cloning and expression of the major protein kinase C substrate of platelets. Nature 1988, 333(6172), 470–473. 71. Chan, T.O., Rittenhouse, S.E., Tsichlis, P.N. AKT/PKB and other D3 phosphoinositide-regulated kinases: kinase activation by phosphoinositide-dependent phosphorylation. Annu. Rev. Biochem. 1999, 68(1), 965–1014. 72. Lemmon, M.A., Ferguson, K.M. Signal-dependent membrane targeting by pleckstrin homology (PH) domains. Biochem J. 2000, 350(Pt. 1), 1–18. 73. Hurley, J.H., Misra, S. Signaling and subcellular targeting by membrane-binding domains. Annu. Rev. Biophys. Biomol. Struct. 2000, 29: 49–79. 74. Falasca, M., Logan, S.K., Lehto, V.P., Baccante, G., Lemmon, M.A., Schlessinger, J. Activation of phospholipase Cgamma by PI 3-kinase-induced PH domain-mediated membrane targeting. EMBO J. 1998, 17(2), 414–422. 75. Rameh, L.E., Arvidsson, A., Carraway, KL, III et al. A comparative analysis of the phosphoinositide binding specificity of pleckstrin homology domains. J. Biol. Chem. 1997, 272(35), 22059–22066. 76. Salim, K., Bottomley, M.J., Querfurth, E. et al. Distinct specificity in the recognition of phosphoinositides by the pleckstrin homology domains of dynamin and Bruton's tyrosine kinase. EMBO J 1996, 15(22), 6241–6250. 77. Harlan, J.E., Hajduk, P.J., Yoon, H.S., Fesik, S.W. Pleckstrin homology domains bind to phosphatidylinositol-4,5-bisphosphate. Nature 1994, 371(6493), 168–170. 78. Rao, V.R., Corradetti, M.N., Chen, J. et al. Expression cloning of protein targets for 3-phosphorylated phosphoinositides. J. Biol. Chem. 1999, 274(53), 37893–37900. 79. Han, J., Luby-Phelps, K., Das, B. et al. Role of substrates and products of PI 3-kinase in regulating activation of Rac-related guanosine triphosphatases by Vav. Science 1998, 279(5350), 558–560. 80. Lemmon, M.A., Ferguson, K.M., O’Brien, R., Sigler, P.B., Schlessinger, J. Specific and high-affinity binding of inositol phosphates to an isolated pleckstrin homology domain. PNAS 1995, 92(23),10472–10476. 81. Pacold, M.E., Suire, S., Perisic, O. et al. Crystal structure and functional analysis of Ras binding to its effector phosphoinositide 3-kinase gamma. Cell 2000, 103(6), 931–943.

240

Functional Lipidomics

82. Stocker, H., Andjelkovic, M., Oldham, S. et al. Living with lethal PIP3 levels: viability of flies lacking PTEN restored by a PH domain mutation in Akt/PKB. Science 2002, 295(5562), 2088–2091. 83. Staal, S.P., Hartley, J.W., Rowe, W.P. Isolation of transforming murine leukemia viruses from mice with a high incidence of spontaneous lymphoma. Proc. Natl. Acad. Sci. U.S.A 1977, 74(7), 3065–3067. 84. Staal, S.P. Molecular cloning of the akt oncogene and its human homologues AKT1 and AKT2: amplification of AKT1 in a primary human gastric adenocarcinoma. Proc. Natl. Acad. Sci. U.S.A 1987, 84(14), 5034–5037. 85. Bellacosa, A., Testa, J.R., Staal, S.P., Tsichlis, P.N. A retroviral oncogene, akt, encoding a serine-threonine kinase containing an SH2-like region. Science 1991, 254(5029), 274–277. 86. Coffer, P.J., Woodgett, J.R. Molecular cloning and characterisation of a novel putative protein-serine kinase related to the cAMP-dependent and protein kinase C families. Eur. J Biochem. 1991, 201(2), 475–481. 87. Franke, T.F., Yang, S.I., Chan, T.O. et al. The protein kinase encoded by the Akt proto-oncogene is a target of the PDGF-activated phosphatidylinositol 3-kinase. Cell 1995, 81(5), 727–736. 88. Cross, D.A., Alessi, D.R., Cohen, P., Andjelkovich, M., Hemmings, B.A. Inhibition of glycogen synthase kinase-3 by insulin mediated by protein kinase B. Nature 1995, 378(6559), 785–789. 89. Alessi, D.R., Deak, M., Casamayor, A. et al. 3-Phosphoinositide-dependent protein kinase-1 (PDK1): structural and functional homology with the Drosophila DSTPK61 kinase. Curr. Biol. 1997, 7(10), 776–789. 90. Alessi, D.R., James, S.R., Downes, C.P. et al. Characterization of a 3-phosphoinositide-dependent protein kinase which phosphorylates and activates protein kinase Balpha. Curr.Biol. 1997, 7(4), 261–269. 91. Araki, N., Hatae, T., Furukawa, A., Swanson, J.A. Phosphoinositide-3-kinase-independent contractile activities associated with Fc{gamma}-receptor-mediated phagocytosis and macropinocytosis in macrophages. J. Cell Sci. 2003, 116(2), 247–257. 92. Xu, D., Kyriakis, J.M. Phosphatidylinositol 3-Kinase-dependent Activation of renal mesangial cell Ki-Ras and ERK by advanced glycation end products. J. Biol. Chem. 2003, 278(41), 39349–39355. 93. Vanhaesebroeck, B., Jones, G.E., Allen, W.E. et al. Distinct PI(3)Ks mediate mitogenic signalling and cell migration in macrophages. Nat. Cell Biol. 1999, 1(1), 69–71. 94. Katso, R., Okkenhaug, K., Ahmadi, K., White, S., Timms, J., Waterfield, M.D. Cellular function of phosphoinositide 3-kinases: implications for development, homeostasis, and cancer. Annu. Rev. Cell Dev.Biol 2001, 17, 615–675. 95. Stricker, R., Vandekerckhove, J., Krishna, M.U., Falck, J.R., Hanck, T., Reiser, G. Oligomerization controls in tissue-specific manner ligand binding of native, affinitypurified p42(IP4)/centaurin alpha1 and cytohesins-proteins with high affinity for the messengers D-inositol 1,3,4,5-tetrakisphosphate/phosphatidylinositol 3,4,5-trisphosphate. Biochim. Biophys. Acta 2003, 1651(1-2), 102–115. 96. Shirai, T., Tanaka, K., Terada, Y. et al. Specific detection of phosphatidylinositol 3,4,5-trisphosphate binding proteins by the PIP3 analogue beads: an application for rapid purification of the PIP3 binding proteins. Biochim. Biophys. Acta 1998, 1402(3), 292–302. 97. Zhu, H., Bilgin, M., Bangham, R. et al. Global analysis of protein activities using proteome chips. Science 2001, 293(5537), 2101–2105.

Fishing for Pharmaceutically Relevant PIP-Binding

241

98. Isakoff, S.J., Cardozo, T., Andreev, J. et al. Identification and analysis of PH domaincontaining targets of phosphatidylinositol 3-kinase using a novel in vivo assay in yeast. EMBO J 1998, 17(18), 5374–5387. 99. Klarlund, J.K., Guilherme, A., Holik, J.J., Virbasius, J.V., Chawla, A., Czech, M.P. Signaling by phosphoinositide-3,4,5-trisphosphate through proteins containing pleckstrin and Sec7 homology domains [see comments]. Science 1997, 275(5308), 1927–1930. 100. Stephens, L., Anderson, K., Stokoe, D. et al. Protein kinase B kinases that mediate phosphatidylinositol 3,4,5-trisphosphate-dependent activation of protein kinase B. Science 1998, 279(5351), 710–714.

11

Phosphoinositide Profiling in Complex Lipid Mixtures Markus R. Wenk and Pietro De Camilli

CONTENTS Abstract ..................................................................................................................243 Introduction............................................................................................................244 Biological Significance of PIs .....................................................................244 The Analytical Problem: Low-Abundance, Rapid-Turnover, and Postmortem Degradation..................................................................246 PI Analysis .............................................................................................................246 PI Extraction from Tissues and Cells..........................................................247 Chromatographic Methods...........................................................................248 Thin-Layer Chromatography ...........................................................248 High-Performance Liquid Chromatography of Intact PIs...............249 High-Performance Liquid Chromatography of Deacylated PIs......249 Receptor Approaches ...................................................................................250 Radioligand Displacement Assays...................................................250 Fluorescence Techniques .................................................................250 Mass-Spectrometry-Based Detection of PIs................................................251 Measurement of PI Kinase and Phosphatase Activities..............................256 Conclusions and Outlook ......................................................................................257 Acknowledgments..................................................................................................258 Abbreviations .........................................................................................................258 References..............................................................................................................258

ABSTRACT Phosphoinositides (PIs, which are phosphorylated derivatives of phosphatidylinositol [PtdIns]) are a class of phospholipids concentrated at cytosolic leaflets of cellular membranes that function as major regulators of a wide variety of cellular processes. A unique feature of these phospholipids is that their hydrophilic inositol head group can be reversibly phosphorylated by lipid kinases and phosphatases, at the various hydroxyl positions, thus leading to the generation of seven naturally occurring isomeric 243

244

Functional Lipidomics

species. Their differentially phosphorylated head groups function as ligands for cytoplasmic proteins as well as for cytosolic domains of membrane proteins and thus are main mediators of the signaling function of PIs. In addition, PIs serve as precursors of other intracellular second messengers. Their subcellular localization, which is spatially and temporally controlled, can be revealed in living cells by fluorescent fusion proteins of protein modules that bind specific stereoisomers. Commonly used methods for their detection in cell extracts include receptor-based assays and radiometric measurements of chromatographically separated products. These approaches report the position of the phosphate(s) on the inositol head group (isomer profiles), but are not easily applicable to large-scale studies and low-abundance samples. Recently, electrospray ionization mass spectrometry (ESI/MS) has been developed as a rapid and sensitive method to detect, identify, and quantify phosphatidylinositol phosphate (PIP) and phosphatidylinositol bisphosphate (PIP2) with different fatty acid compositions (PI fatty acid profiles) in tissue and cell extracts. Here we review and discuss the current technologies available for PI profiling in complex lipid mixtures with an emphasis on chromatography and mass spectrometry.

INTRODUCTION BIOLOGICAL SIGNIFICANCE

OF

PIS

Cellular membranes are composed of lipids with a bewildering diversity of chemical compositions. The different head group and fatty acid moieties have physiological actions based on their effects on the physical–chemical properties of the bilayer, their metabolic roles, and their signaling functions. PIs are the most important membrane-signaling lipids and play an important regulatory role in a wide variety of cellular processes (1–3). Unique features of PIs are the variable phosphorylation state of their inositol head group and the abundance of polyunsaturated fatty acids (PUFAs) in their acyl chain positions (Figure 11.1). PIs play important signaling functions, as first recognized in the mid-1950s. For many years, they were best known for the signaling function of their phospholipasegenerated metabolites, IP3, diacylglycerols (DAGs), and arachidonic acid (AA) (4,5). IP3 and DAG, generated by phospholipase C (PLC), regulate intracellular calcium levels and protein phosphorylation, whereas AA, generated by phospholipase A2, is metabolized to form eicosanoids that are involved in inflammatory responses (see also Chapter 4 and Chapter 14 of this volume). More recently, it has become clear that PIs themselves are potent signaling molecules. Their phosphorylated inositol head groups interact with variable affinity and specificity with domains and modules of cytosolic and membrane proteins (6). Indeed, the reversible phosphorylation of inositol phospholipids may rival in importance the reversible tyrosine-phosphorylation of membrane proteins as a mechanism to control recruitment and regulation of proteins at the membrane interface. PIs are not distributed randomly inside the cell but instead are localized to specific subcellular sites. PI(4)P, for example, is enriched in the Golgi whereas the bulk of PI(4,5)P2 is believed to reside at the cell periphery (see Reference 3, Reference 6, and Reference 7 for recent reviews; see also Chapter 9 and Chapter 10 in this volume).

Phosphoinositide Profiling in Complex Lipid Mixtures

A 6 1

5 2

PIP

B 4

Sac1/MT

3

PI3K II&III

PI(3)P

245

PIP2 PIKfyve SJ

PIP3

PI(3,5)P2

PIPK I&II INPP4 PI4K

PI(4)P

SJ/Sac1

PI3K I&II

PI(3,4)P2

PIPK I 5-PT/SJ

PIPK I 20 carbon atoms 4 double bonds

Arachidonic acid (20:4)

18 carbon atoms 0 double bonds

Stearic acid (18:0)

PI

5-PT/SJ PIPK I

PI(5)P

PIPK II

PI(4,5)P2

LPA

PI3K I&II PTEN/ SJ/SHIP

PI(3,4,5)P3

LPAAT

PC

PLD

CDP-DAG synthase

PA

PAP DAGK

DAG

PLC

CDP-DAG PI synthase

Inositol

Membrane cytosol

I(1,4,5)P3

IP4

IP5

IP6

IP7

IP8

I(1,3,4)P3

FIGURE 11.1 PIs and their metabolic chain. In many mammalian cells, the most abundant molecular species of PIs, at least in unstimulated conditions, are those that carry SA and AA in their acyl chains. The corresponding PI(4,5)P2, 1-stearoyl-2-arachidonoyl-sn-glycero-3phosphatidylinositol-4,5-bisphosphate, is shown in panel A. Reversible phosphorylation of the various hydroxyl moieties of the inositol head group leads to the generation of seven naturally occurring isomers, with mono (PIP), bis (PIP2), or tris (PIP3) phosphorylation. A simplified summary of these metabolic pathways is shown in panel B. Phospholipase C (PLC)–catalyzed breakdown of PI(4,5)P2 generates soluble inositolpolyphosphates (I(1,4,5)P3 and IPx) and diacylglycerols (DAG). The latter are rapidly phosphorylated to phosphatidic acids (PAs), which are signalling lipids themselves, and which also represent important metabolites in phospholipid synthesis. Note that each class of lipid metabolite (e.g., PI(4,5)P2) will be represented by a number of molecular species with different fatty acid compositions (cf. also Figure 11.4). Not included in the scheme are phospholipase A–catalyzed reactions that lead to the generation of lysolipids and fatty acids. LPA, lysophosphatidic acid; PLD, phospholipase D; PAP, phosphatidic acid phosphatase; LPAAT, lysophosphatidic acid acyltransferase; DAGK, DAG kinase; PT, phosphatase; SJ, synaptojanin; and MT, myotubularin.

Dysfunctions in the control of PI metabolism have already been implicated in a variety of pathological conditions in humans and model organisms. For example, excess PI(3,4,5)P3 signaling may promote cancer (e.g., PTEN and PI3Kinases [1]) as well as enhanced sensitivity to insulin (SHIP2 [8]). Patients with Lowe syndrome harbor a mutation in the OCRL1 gene, which encodes for a PI 5-phosphatase (9). Mutations in the polyphosphoinositide phosphatase synaptojanin, in the PIP kinase type 1γ and in type 1 inositol 4-phosphatase, lead to perinatal or early postnatal lethality in mice (10–12). It has been suggested that aberrant PI metabolism might be implicated in neurological disorders such as dipolar disorders and schizophrenia (13). One of the targets of lithium, which is widely used to treat manic depression, is inositol polyphosphate phosphatase Inpp1 (14). Growing evidence also points to important roles of inositol lipids in the immune system and in infectious diseases, though many of the precise molecular details and mechanisms are still poorly characterized. Lipids, and in particular inositol lipids, play an important role during

246

Functional Lipidomics

bacterial invasion, intracellular replication, and persistence (15) as well as in the immune response via CD1 receptors (16).

THE ANALYTICAL PROBLEM: LOW-ABUNDANCE, RAPID-TURNOVER, AND POSTMORTEM DEGRADATION Phospholipids make up the bulk of the lipid mass in a typical biological membrane. PIs, however, constitute a rather minor fraction of this mass compared with other phospholipids such as phosphatidylcholines or phosphatidylserines. In fact, PIs were first described by Folch in 1949 as an additional component of brain cephalin, a lipid fraction rich in phosphatidylserine and phosphatidylethanolamine (17). The average content of inositol phospholipids in biomembranes is 5 to 10% of total lipids. Approximately 10 to 20% of PtdIns exists in the form of the mono- phosphorylated (PIP) or bi-phosphorylated (PIP2) derivatives, which, as a consequence, make up ~ 1 to 2% each of total lipids (18). PI(4)P and PI(4,5)P2 represent the most abundant species. PIs phosphorylated at the 3 position (3-PIs) are present at only very low concentrations in resting conditions, and their levels can undergo rapid changes upon stimulation by growth factors. Even during stimulation, however, PI(3,4,5)P3 (PIP3) accounts for only approximately 120 of the amount of PIP2 (19). These rather low levels are one of the characteristics that led, early on, to the assumption that PIs might be involved in cell signaling. This hypothesis was subsequently supported by studies on the turnover of PIs. Treatment of secretory tissues with acetylcholine led to a rapidly increased incorporation of phosphate into PIs (20). This “inositide effect,” was then observed in a wide variety of tissues, including nerve tissue, and in response to various stimuli (21,22). Eventually, PI metabolism was recognized as a major mechanism of cellular signaling. The precise levels of each of the PIs are continuously controlled by a tight balance between their ATPdependent synthesis and their hydrolysis by lipases and phosphatases. Not surprisingly, PIs levels rapidly fall postmortem (23) due to the loss of ATP and the predominance of hydrolysis under these conditions. Low abundance, rapid turnover, and degradation are complicating factors for the chemical analysis of PIs. Thus, highly sensitive methods and manipulations that achieve instant and efficient quenching of enzymatic activity during extraction are required.

PI ANALYSIS Methods for PI analysis have greatly expanded since their discovery 50 years ago, mainly driven by advances in analytical chemistry and cell biology. There are now a number of methods available that are summarized and discussed in the following text (Figure 11.2, from the Subsection titled Chromatographic Methods to the Subsection titled Measurement of PI Kinase and Phosphatase Activities). Each approach has its advantages and intrinsic limitations, which are briefly highlighted in each section. Approaches for visualization in living cells using protein modules that specifically interact with PIs are presented in different chapters of this volume and are not discussed in detail here (24).

Phosphoinositide Profiling in Complex Lipid Mixtures

Detection method

Separation method

PtdIns

PIs headgroup isomers

247

Fatty acids

Quantification

Sensitivity

V

E

G N

R

Fluorescent Probes FRET probes

−

−

+

−

−

−

−

+

−

−

−

+

−

+

+

+2

Lipid extraction ESI-MS

PIs

Very high

Lipid stains

TLC

+

−

+

Low

Radiometric

TLC

+

−

+2

+3

High

FRET probes

−

−

+

−

+4

Very high

Radiometric

HPLC

+

+

−

+3

High

Conductivity

HPLC

−

+

−

+

Very high

IP receptors

−

−

+

−

+4

High

Deacylation

gPIs Deglyceration

IPs

FIGURE 11.2 Schematic overview of PI analysis in complex mixtures. The currently available methodologies for analysis of PIs have distinct advantages and limitations, which are summarized here. Fluorescent probes that bind to PIs are a powerful tool for in vitro applications, though dual specificities (e.g., recognition of multiple PIs and overlap with protein binding) complicate interpretation of the fluorescence signals. Their use will not be described in detail here (see also footnote 1). The first step in the biochemical analysis of PIs is their extraction from cell and tissue samples. PIs can then be measured directly in the complex lipid mixture (ESI-MS), after sample cleanup (FRET probes) or chromatographic separation (TLC). Determination of the full isomer spectrum requires additional reactions (deacylation) and separation by HPLC. Notes: (1) See also Chapter 9 and Chapter 10 of this volume. For reviews please see Reference 6 and Reference 24. Not included in the present discussion are antibody-based detection methods of PIs (78). (2) Molecular species of PIs with different fatty acid compositions are generally not resolved by TLC. The fatty acid composition can however be determined by gas chromatography after chemical derivatization of chromatographically separated and hydrolyzed PIs (62). (3) Specific activity difficult to determine; signal might not reflect mass levels (see also Section titled Thin-Layered Chromatography). (4) Indirect measurement based on displacement reaction.

PI EXTRACTION

FROM

TISSUES

AND

CELLS

PIs are very polar. In fact, they are so polar that they are not extracted efficiently from tissues and cells by neutral solvents. Instead, acidified chloroform-methanolHCl solvents are commonly used (25,26). Under these extraction conditions, protonation of the phosphoryl groups leads to more efficient partitioning of the PIs into

248

Functional Lipidomics

the nonpolar organic phase. To further enhance recovery, in particular for the quantitative extraction of PIP3, tetrabutyl ammonium sulphate was used as an additive during extraction (27). Alternatively, a two-step protocol can be employed which (1) efficiently precipitates (and inactivates) cellular proteins and membranes (using TCA or PCA), and in which (2) PIs are extracted from this pellet (using chloroform methanol) (25). It is crucial in either case to follow a procedure that will rapidly (within seconds) destroy enzymatic activity upon cell disruption. Either precipitation or extraction media or both can be applied directly to cultured cells. In the case of tissues, the method of choice is to first flash-freeze the sample, then mechanically powder it into liquid nitrogen, e.g., by using a mortar, followed by extraction using chloroform-methanol-HCl. In metabolic labeling experiments involving radioactive PI precursors, “cold,” i.e., unlabeled, PIs are added during the extraction as carriers to enhance recovery (26,28). Siliconized glass- and plasticware should be used to minimize surface adsorption of PIs. From 10 g of brain tissue, a few milligrams of PIP and PIP2 can be expected (29).

CHROMATOGRAPHIC METHODS A large number of chromatographic methods have been developed for the analysis of PIs. In addition to separating PIs from other lipids, some of these methods separate the different inositol head group isomers, but molecular species with different fatty acid composition are generally not resolved. The greatest challenge with most of these approaches is their low sensitivity. Thin-Layer Chromatography Among the many chromatographic methods available, which include column and cartridge chromatography (e.g., ion exchange cellulose, silicic acid, and C18 resins), thin-layer chromatography (TLC) is the most convenient for the separation of PIs and requires no sophisticated equipment. Typically, PIs are seeded onto silicagel plates that have been pretreated with potassium oxalate and EDTA to chelate divalent ions that are present in the binding material of the TLC plate. If this step is ignored, PIs do not migrate well as they themselves are potent sensors of divalent cations (30). Plates are then developed with chloroform-based solvent systems that contain either ammonia or acetic acid and which will separate PIs from other faster-running phospholipids (26,31). TLC analysis of PIs is a low-resolution method. Head group isomers (e.g., PI(3,4)P2 and PI(4,5)P2) are not easily separated except for synthetic and short-chain PIs (32). Information on the fatty acid composition is also not revealed as molecular species with different fatty acids comigrate, and, generally, only lyso-derivatives can be resolved from their corresponding parent PIs. Lipids, including PIs, can be visualized by a number of ways such as iodine vapor (requires microgram quantities of PIs), cupric acetate (33), Periodate-Shiff (34), or alphanapthol (35), though none of these stains are specific for PIs. Radiolabeling with inositol isotopes (e.g., 3H-myo-inositol) is a convenient way to selectively mark PIs, as inositol is not further metabolized in cells. As a consequence, only very few spots will be visible on an autoradiograph of a TLC plate that was used to separate complex

Phosphoinositide Profiling in Complex Lipid Mixtures

249

lipid extracts from cells labeled with 3H-myo-inositol. In contrast, labeling with other tracers, such as phosphate or carbon isotopes, yields much more complex spot patterns due to signals that stem from other phospholipids present in the extract. When radiolabeled precursors are used, the specific activity associated with each PI species may vary, thus complicating quantification of the results. Radioactivity signals reflect total mass of a given PI species only if metabolic equilibrium between cold and hot species has been reached. In the case of 3H-myo-inositol and mammalian cells, days of incubation with 3H-myo-inositol are needed (25). For precise quantitative analysis, TLC spots can be scraped and processed for phosphate determination (36) or radioactivity counts. High-Performance Liquid Chromatography of Intact PIs Typically, intact phospholipids are separated by high-performance liquid chromatography (HPLC) on reversed-phase columns and quantified by inorganic phosphate analysis or, if the solvent system permits, absorption at 200 nm (31). These separations are inefficient, however, and PIs are not recovered well under these conditions. Furthermore, detection of PIs requires derivatization to increase sensitivity for detection by UV (37) or analysis of inorganic phosphate (36). Early after the discovery of PIs, ion exchange chromatography using DEAE cellulose and salt gradient elution were used for the separation of PI, PIP, and PIP2 from contaminating lipids present in tissue fractions (38,39). These methods have since been used and continuously improved for the purification of milligram quantities of PIs for biochemical studies and enzyme assays (40). Neomycin, a primary amine antibiotic with high affinity for PIs, is an excellent ligand in affinity chromatography and yields highly enriched PI fractions (41). Sensitive detection of PIs remains the major challenge in analytical approaches based on liquid chromatography (LC). LC followed by mass spectrometry promises to be an ideal method to achieve optimal PI detection with high sensitivity. It can be expected that improved LC conditions suitable for these applications will be developed in the future (42). (See also Subsection titled MassSpectrometry-Based Detection of PIs.) High-Performance Liquid Chromatography of Deacylated PIs Strong anion exchange (SAX) chromatography is the method of choice for the separation of the different PI head group isomers (26,28,43). This requires prior deacylation of PIs in a chemical reaction with methylamine that yields the corresponding glycerol-phosphoryl head groups (gPIs) (Figure 11.2). HPLC with shallow phosphate gradients are then used to separate gPIs, on ionic exchange columns, and the elution of the test sample is compared with the elution of authentic internal standards (43). The major advantage of this method is that the full isomer spectrum of PIs can be obtained, though for some gPIs, this requires very long chromatographic runs. In fact, PI(5)P elutes on the shoulder of PI(4)P in most gradient conditions. This explains, at least in part, why relatively little is known about the cellular levels of PI(5)P (44). As in the case of TLC analysis, HPLC generally requires the use of radioisotopes for sensitive detection of gPIs by scintillation

250

Functional Lipidomics

counting (28). In addition, deacylation of PIs involves additional sample-handling steps and is therefore relatively laborious. In many cases, such as in the analysis of intact tissues, radiolabeling — in particular, radiolabeling to isotopic equilibrium — is not feasible. For example, this greatly limits the application of these methods to the characterizations of genetically modified animals or in the analysis of human specimens. More recently, a conductivity-based method was developed that allows direct quantification of gPIs without prior metabolic labeling of cells (45).

RECEPTOR APPROACHES Receptor approaches depend on the availability of modules that specifically interact with PIs and their phosphorylated head groups. Traditionally, these have been preparations of the IP3 receptor. The discovery of novel PI-binding domains has greatly enhanced receptor approaches in recent years (see also Chapter 9 and Chapter 10). Radioligand Displacement Assays Radioligand displacement assays provide a sensitive and direct method for the determination of levels of IP3 and other inositol polyphosphates such as, for example, inositol-(1,3,4,5)P4 (IP4) (46, 47). The same assays have been used to determine the levels of PIP2 and PIP3 in cellular extracts following chemical cleavage of the inositol head group from PIs. Briefly, IP3-specific (e.g., IP3 receptor from bovine adrenals) or IP4-specific (e.g., GAP1IP4BP) binding proteins are incubated with cellular PI extracts after alkaline hydrolysis that liberates the inositol head groups. Endogenous IP3 (or IP4) present in this material are next displaced with known amounts of radiolabeled IP3 (or IP4), and their levels are determined using a calibration curve. These assays are, in general, fairly sensitive (approximately 0.3 pmol in the case of IP4), and some kits are commercially available. They do, however, remain technically demanding, laborious, and expensive, and require the use of radioactivity, albeit at low doses. Fluorescence Techniques The inventory of PI-isomer-specific protein modules (e.g., PH, PX, FYVE, ENTH, etc.) is growing steadily. Whereas some of them have promiscuous specificity, others recognize, selectively, one isomer. It is conceivable that, in the future, domains highly specific for each of the PI isomer will be discovered (see also Chapter 9 and Chapter 10 of this volume) (6,24). The discovery of these modules offers new possibilities for the analysis of PIs. Fluorescence methods (based on these modules) for the qualitative and quantitative detections of PIs in living cells are discussed elsewhere in this volume. Here, we will focus primarily on biochemical approaches that capitalize on these modules, i.e., assays that measure PIs in extracts from cells and tissue rather than in vivo measurements. The emission property of a fluorescent probe is critically dependent on its environment. In fluorescence resonance energy transfer (FRET), this characteristic is used to monitor the relative distance of two probes of interest. Sensitive optical assays for PI detection based on this principle have recently been designed. For example, fluorescence quenching was used to screen for compounds that inhibit the binding of a GFP-PH fusion protein to PI

Phosphoinositide Profiling in Complex Lipid Mixtures

251

containing membrane bilayers that harbor a FRET acceptor (48). In a further development of this approach, PI kinase and phosphatase activities, as well as levels of PIs in cellular extracts, were measured (49). This assay is based on the displacement of a PH domain from a sensor complex that leads to a change in the FRET signal. Interference of the displacement reaction with components in the lipid extract (e.g., other lipids and metal ions) remains a technical complication, however (49). A major advantage of such nonradioactive approaches is that they may be developed into high-throughput formats.

MASS-SPECTROMETRY-BASED DETECTION

OF

PIS

Inositol phospholipids were among the first phospholipids to be analyzed by fast atom bombardment (FAB) mass spectrometry. These early techniques, however, did not result in major advantages over existing methodology such as the gas chromatographic analysis of the chemically derivatized fatty acids esters (50). The high chemical background observed in FAB and chemical ionization was a limiting disadvantage particularly for the analysis of complex mixtures found in crude lipid extracts (50,51). Nevertheless, these early studies showed that fragmentation of phosphatidylinositol species leads to the generation of ions that are specific for this class of lipids (52,53). A breakthrough for the analysis of phosphatidylinositols in complex mixtures came, as in the case of other phospholipids, with the development of electrospray ionization (ESI) and tandem mass spectrometry (see also other chapters in this volume). ESI is orders of magnitude more sensitive than FAB and circumvented the problems inherent to FAB mentioned in the preceding text. In ESI, phosphatidylinositols can be detected in both positive and negative modes, though they are generally analyzed as singly charged [M−H]− ions (54,55). The negative mode is generally preferred because, in a mixture, their ion yield is comparable to that of other major anionic phospholipids such as phosphatidylglycerols or phosphatidylserines (56). Furthermore, the negative-ion spectra are easier to interpret because they avoid the overlap with sodium adducts [M+Na]+ and protonated molecular ions [M+H]+ (56,57). The fragmentation pathways of phosphatidylinositol and PIs have been studied in detail by Hsu and Turk using deuterated compounds (58). The fragmentation of phosphatidylinositol upon low-energy-collision-induced dissociation (CID) in ESI yields abundant product ions that reflect (1) the fatty acid carboxylate ions, (2) ions that result from the neutral loss of the fatty acids, and (3) head group fragments (Figure 11.3). With respect to the loss of fatty acids, the fragmentation of phosphatidylinositol is similar to that of phosphatidic acid (PA) (58). Neutral loss of the sn2 substituent of 38:4 PI as a fatty acid or ketene moiety leads to ions at m/z 599 and 581, respectively, which upon additional loss of inositol produces m/z 419 and 439 (Figure 11.3) (58). The ions resulting from the loss of the fatty acid from sn-2 are more intense than those from sn-1, consistent with the preferential hydrolysis at the sn-2 position. Among the head group fragments, m/z 223, 241, 259, 297, and 315 are unique for phosphatidylinositols. Thus, they may be used as identifier ions for phosphatidylinositols in complex mixtures.

252

Functional Lipidomics

MS/MS m/z 885 38:4 PI

Productions

Precursor ion 18:0

SA 283 AA 303 IP 241

38:4 PI 885 20:4

200

Relative intensity

581 297 315

300

500 m/z (a)

153

223

259

223 153

700

800

259

315

283

223 153 150

283

321

241 200

250

MS/MS m/z 965 38:4 PIP

IP2

303 321

303

0.5

900

MS/MS m/z 885 38:4 PI

241 +HPO 3

0.0 1.0

0.0

600

SA 283 AA 303

241

0.0 1.0 0.5

400

18:0

599

439

IP

1.0 0.5

20:4

419 223

259

gP 153

300

+HPO3350

MS/MS m/z 1045 38:4 PIP2 401 IP3 400

m/z (b)

FIGURE 11.3 Electrospray ionization tandem mass spectrometry of PIs. Panel A shows MS/MS analysis of m/z 885 (38:4 PI), that upon collision-induced fragmentation, yields product ions derived from the fatty acids (m/z 303, AA, and m/z 283, SA), the inositol head group (m/z 241, IP, which is the dehydrated phosphoryl inositol head group), and the glycerol backbone (m/z 153, gP, dehydrated glycerolphosphate). PIP and PIP2 produce fragments at 321 (IP2) and 401 (IP3), respectively, characteristic for their additional head group phosphorylations (panel B). Ions at m/z 223, 241, and 259 are additional inositol head group fragments common to all of these PIs. Panel B is modified and reproduced with permission (From Wenk, M.R., Lucast, L., Di Paolo, G., Romanelli, A.J., Suchy, S.F., Nussbaum, R.L., Cline, G.W., Shulman, G.I., McMurray, W., and De Camilli, P. Phosphoinositide profiling in complex lipid mixtures using electrospray ionization mass spectrometry. Nat. Biotechnol. 2003, 21, 813–817.).

Phosphoinositide Profiling in Complex Lipid Mixtures

253

Indeed, a number of studies have used this information, in particular, m/z 241 (the dehydrated inositol head group, Figure 11.3), to confirm the identity of phosphatidylinositols in complex lipid mixtures (54,55,59,60). The existence of specific fragment ions allows for additional tandem mass spectrometric analysis. Precursor ion scanning is a mode of tandem mass spectrometry in which two mass analyzers undergo linked scanning operations. It is conveniently performed on so called “triple quadrupole” instruments but can also be achieved on other types of instruments (61). Two mass analyzers separated by a collision cell (in which ions are induced to fragment) are operated so that the first mass analyzer (usually a quadrupole, Q1) is stepped (i.e., performs a step scan function over a certain mass range), while the second mass filter (which is either fixed on a set mass in the case of a “triple quadrupole instrument” or which collects all ions in the case of a TOF) analyzes the fragments that originate from each step of the scan in Q1. If fragment ions of m/z 241 are selected in the second mass filter, all compounds in the mixture that have the ability to generate this ion will be detected as precursors, and hence identified as inositol containing lipids (54,60,61) (see also Figure 11.4). This is a very powerful method to “profile” phosphatidylinositols (i.e., to identify and quantify inositol phospholipids with different fatty acid distributions) in complex mixtures derived from cell and tissue extracts. Using this approach it was found that, in mammalian membranes, PI species with stearic acid (SA) (18:0) and AA (20:4), i.e., 38:4 PI (Figure 11.1 and Figure 11.3) dominate over less abundant molecular species with different fatty acids (54,60). This result is consistent with traditional analysis of fatty acids by GC-MS (62). Phosphorylation of phosphatidylinositols adds mass (80 units per phosphate moiety, HPO3) and hydrophilic phosphate groups that make PIs chemically distinct from the bulk of phospholipids. Both are complicating factors with respect to mass spectrometric analysis. As a consequence, PIs have been missed out in many studies involving mass spectrometry of lipid extracts. The major limiting complication for their detection is suppression of ionization (60,63) that can, however, be overcome, at least to some extent, by the addition of organic buffers (such as triethyl ammonium acetate or piperidine) to the lipid extract prior to infusion into the mass spectrometer (60). The mechanism of action of these organic buffers is not entirely understood, but part of their effect might be due to the neutralization of residual acid from the extraction procedure (Subsection titled PI Extraction from Tissues and Cells). Suppressed ionization of PIs is reminiscent of similar effects observed with phosphorylated peptides. Figure 11.3 shows product ion spectra of PIs in crude extracts from brain tissue and identifies ions at m/z 965 and 1045 as the singly and doubly phosphorylated derivatives of 38:4 PI (m/z 885) (60). If retained during fragmentation, addition of a single phosphate (HPO3) to the inositol ring should yield an ion at m/z 321, 80 mass units higher than the inositol fragment from PI (m/z 241). This is indeed what happens experimentally (Figure 11.3B). An additional shift to m/z 401 is observed in the case of PIP2. Whereas the major molecular species (e.g., m/z 965 and 1045) can be detected in single-stage ESI-MS, the quantitative profiling of less abundant species and the analysis of extracts that contain a high degree of isobaric overlap (such as, for example, the yeast S. cerevisiae) requires tandem MS/MS analysis

254

Functional Lipidomics

38:4 PIP

38:4 PI

ESI-MS-

+HPO3

885 965

600

700

800

900

1000 38:4 PIP

C38

36:4 PIP 34:1 PIP

Precursor ion scan for m/z 321 (PIP) 800

C34

C36

40:6 PIP C40

900

1000

FIGURE 11.4 Profiling of PIs in a brain lipid extract by ESI-MS. Top panel: negative-ion single-stage mass spectrum of a total rat brain lipid extract. A large number of ions are detected in the mass range of m/z 700–900 that mainly represent the phospholipids. Phosphorylation of the major phosphatidylinositol species, 38:4 PI shifts the corresponding PIP (38:4 PI, m/z 965) by 80 units, the mass of a phosphate moiety. Less abundant PIP species are detected by precursor ion scanning (bottom panel). A precursor ion scan for m/z 321 (the inositol head group of PIP) yields clusters of PIP species with 34, 36, 38, and 40 fatty acid carbons (lower panel), whose structures are shown in color (blue, C16:0, palmitic acid; green, C18:0, SA; magenta, C18:1, oleic acid; yellow, C20:4, AA; red, C22:6, docosahexaenoic acid). (From Wenk, M.R., and De Camilli, P. Inaugural Article: Protein–lipid interactions and phosphoinositide metabolism in membrane traffic: Insights from vesicle recycling in nerve terminals. Proc Natl Acad Sci USA. 2004, 101, 8262–8269.)

(Table 11.1) (60). A precursor ion scan experiment for m/z 321 (401) delivers the molecular species profile of PIP (PIP2). In brain lipid extracts, the total number of fatty acid carbon atoms ranges from 32 to 40, with increasing numbers of double bonds in the longer acyl chains (Figure 11.4). Table 11.1 contains the calculated mass values of PI species over a wide range of theoretical compositions. Many of these theoretical molecular species may not exist at all or may be present only at very low abundance in vivo. As an alternative to the precursor scanning mode, selected PI species of interest can be specifically monitored using multiple reaction

Phosphoinositide Profiling in Complex Lipid Mixtures

255

TABLE 11.1 Calculated Mass Values of Molecular Species of Phosphatidylinositol and PIs with Acyl-Linked Fatty Acids

Total Number of Carbon Atoms (x)

Total Number of Carbon Atoms (x)

Total Number of Carbon Atoms (x)

Total Number of Carbon Atoms (x)

PI 24 26 28 30 32 34 36 38 40 42 PIP 24 26 28 30 32 34 36 38 40 42 PIP2 24 26 28 30 32 34 36 38 40 42 PIP3 24 26 28 30 32 34 36 38 40 42

0 697 725 753 781 809 837 865 893 921 949 0 777 805 833 861 889 917 945 973 1001 1029 0 857 885 913 941 969 997 1025 1053 1081 1109 0 937 965 993 1021 1049 1077 1105 1133 1161 1189

Number of 1 2 695 693 723 721 751 749 779 777 807 805 835 833 863 861 891 889 919 917 947 945 1 2 775 773 803 801 831 829 859 857 887 885 915 913 943 941 971 969 999 997 1027 1025 1 2 855 853 883 881 911 909 939 937 967 965 995 993 1023 1021 1051 1049 1079 1077 1107 1105 1 2 935 933 963 961 991 989 1019 1017 1047 1045 1075 1073 1103 1101 1131 1129 1159 1157 1187 1185

Double 3 691 719 747 775 803 831 859 887 915 943 3 771 799 827 855 883 911 939 967 995 1023 3 851 879 907 935 963 991 1019 1047 1075 1103 3 931 959 987 1015 1043 1071 1099 1127 1155 1183

Bonds (y) 4 5 689 687 717 715 745 743 773 771 801 799 829 827 857 855 885 883 913 911 941 939 4 5 769 767 797 795 825 823 853 851 881 879 909 907 937 935 965 963 993 991 1021 1019 4 5 849 847 877 875 905 903 933 931 961 959 989 987 1017 1015 1045 1043 1073 1071 1101 1099 4 5 929 927 957 955 985 983 1013 1011 1041 1039 1069 1067 1097 1095 1125 1123 1153 1151 1181 1179

6 685 713 741 769 797 825 853 881 909 937 6 765 793 821 849 877 905 933 961 989 1017 6 845 873 901 929 957 985 1013 1041 1069 1097 6 925 953 981 1009 1037 1065 1093 1121 1149 1177

Note: The values are indicated as the nominal monoisotopic masses expected for M−H− molecular species of PIs (x:y PIs, where x is the total number of carbon atoms in the fatty acyl chains, and y is the total number of double bonds). They were calculated as described (From Marsh, D. (1990) Handbook of Lipid Bilayers, CRC Press, Boca Raton, FL.). Briefly, the molecular weight MW = 331.19 + R, where R is the contribution from the fatty acyl residues. R = 14.027x + q, and q takes into account the degree of unsaturation with q = 28.89 for y = 0 double bond, q = 27.97 for y =1 double bond, q = 25.95, 23.94, 21.92, 19.91, and 17.89 for 2, 3, 4, 5, and 6 double bonds, respectively. Phosphorylation of phosphatidylinositol species increases their molecular weight by 80 mass units, and the masses of PIP, PIP2, and PIP3 species were calculated accordingly. Major molecular species found in mammalian brain are indicated in bold (From Wenk, M.R., Lucast, L., Di Paolo, G.; Romanelli, A.J., Suchy, S.F., Nussbaum, R.L., Cline, G.W., Shulman, G.I., McMurray, W., and De Camilli, P. PI profiling in complex lipid mixtures using electrospray ionization mass spectrometry. Nat. Biotechnol. 2003, 21, 813–817.).

256

Functional Lipidomics

monitoring (MRM). In this mode, transition pairs (e.g., 885/241 for 38:4 PI or 965/321 for 38:4 PIP) are followed in tandem mass operation. This approach not only enhances the sensitivity and selectivity of the measurement but also allows for analysis of isobaric species, (e.g., m/z 861, which is the mass of 36:2 PI and also of 30:0 PIP) (see Table 11.1) (60). A major advantage of ESI-MS is that it circumvents the multistep procedures that are generally required for quantitative analysis of PIs in cell and tissue extracts. Thus, it allows direct mass measurement of PIs in small samples, without the need for radiolabeling. Given that most of the other lipid constituents are also present in the lipid extracts (Figure 11.4), ESI-MS will allow integrating data on PIs with mass spectrometric profiling of other phospholipids. Its major limitation, at least at this stage, is that it does not allow discriminating inositol phosphate isomers of PIP and PIP2. It should be noted, however, that in mammalian membranes, the majority of PIPs exist in the form of PI(4)P and the majority of PIP2s in the form of PI(4,5)P2, at least in resting conditions. Thus, important conclusions can be drawn from the mass levels of PIs despite the nondisclosure of information on head group phosphorylation. Furthermore, inositol phosphates were recently shown to yield isomerspecific fragmentation pattern during CID, which might open possibilities for the discrimination of PI isomers by MS in the future (64). An additional limitation of ESI-MS is that PIP3, one of the most important PI species with key signaling functions, has not be detected so far, using this approach (60). Possible explanations are an additional loss in sensitivity due to enhanced ion suppression by the presence of an additional phosphate and the very low levels of PIP3 in biological membranes, even after appropriate stimulation (Subsection titled The Analytical Problem: LowAbundance, Rapid-Turnover, and Postmortem Degradation). It can be anticipated that LC-MS might help to solve some of these problems (42). Indeed, mobile phases that are similar to the separation of PIs in normal-phase chromatography such as acidic conditions used in TLC (Subsection titled Thin-Layer Chromatography) or phosphate gradients for elution of deacylated head groups on HPLC columns (Subsection titled Receptor Approaches) are an alternative for analytical scale analysis of PIs. However, chemical stability under these conditions and compatibility with MS inlet remain unresolved technical complications (18,65) (see also Subsection titled Receptor Approaches).

MEASUREMENT

OF

PI KINASE

AND

PHOSPHATASE ACTIVITIES

A large number of enzymes implicated in PI metabolism have been identified via genetic and biochemical approaches (66–70). The mechanisms that regulate these enzymes and the potential disruption of these regulatory mechanisms in disease remain an underinvestigated area in the field of PI biology. PIs and their immediate upstream and downstream metabolites (e.g., DAG, PA, etc.; see Figure 11.1) are first-line mediators (and indicators) of cellular signaling. It can thus be hypothesized that not only their acute disruption, but also a slight and chronic imbalance of their relative levels leads to changed cellular functions that may contribute to the onset of pathology (71). As discussed in the preceding text, this view is supported by genetic studies in a variety of organisms, including humans.

Phosphoinositide Profiling in Complex Lipid Mixtures

257

Traditional kinase assays for lipid enzymes involve the use of radiolabeled nucleotides (e.g., gamma-32P-ATP) and pure lipid substrates. The reaction products are then extracted, separated using TLC and HPLC methods described earlier and detected using scintillation counting (26,28,72,73). These approaches are now complemented by a new line of mainly optical assays. These are made possible by the availability of pure and synthetic PI isomers, some of which are water soluble due to the short fatty acyl chains (see also Chapter 9). Free phosphate released by PI phosphatases can readily be detected with a malachite-green-based assay (74) as previously done for investigations of protein phosphatases (75). Fluorescently tagged PI analogs can be used to monitor the activity of inositol phospholipid kinases, although this requires subsequent chromatographic separation of reaction products. The FRET assays described earlier (Subsection titled Fluorescence Techniques) are also powerful new tools for profiling of PI kinase and phosphatase activities with applications in drug screening. An assay based on micellar electrokinetic capillary chromatographic separation of fluorescent lipids for enzymatic profiling was recently developed (76). These novel approaches will greatly facilitate the systematic analysis of the regulation of PI enzymes and will assist in the development of drugs that modulate their activities.

CONCLUSIONS AND OUTLOOK The complexity arising from the heterogeneity of fatty acids found in PIs (and membrane lipids in general) is astonishing, and much remains to be discovered about the biological significance of this characteristic. It is clear that the diversity of the signaling properties of PIs is not limited to those mediated by their head group isomers although these certainly have a major role. Growing evidence indicates critical importance of the chemistry of the hydrophobic portion of the bilayer in cell regulation. For example, the hydrophobic environment of ion channels is thought to regulate their properties. PUFAs that are typically found in high abundance in PIs, are precursors for important inflammatory lipid mediators (see also Chapter 4 of this volume). It will be interesting to learn if PI pools with different fatty acyl composition are segregated in different membrane compartments or membrane microdomains. Over the past 20 years, our knowledge of the biology of PIs has greatly benefited from methodological advances in their analysis, both in living cells and in complex lipid mixtures derived from cell and tissue extracts. A remaining technical challenge will be to integrate the head group isomer information with the fatty acid composition of the various molecular species. PIs (and their immediate upstream and downstream metabolites) are involved in a wide variety of signaling processes, and their levels will reflect in some fashion the physiological (and pathological) state of a cell or tissue. As such, it can be expected that PIs (and their metabolizing enzymes) will become important biomarkers in clinical applications such as diagnostics and prognostics. Furthermore, using modern analytical approaches, the biochemical profiling of PIs in cell and tissue extracts is likely to lead to the discovery of novel PImetabolizing enzymes (and effectors of known enzymes), as well as unknown inositol-containing lipid metabolites such as, for example, head group derivatives

258

Functional Lipidomics

that carry additional functional groups or molecular species with unusual fatty acid compositions.

ACKNOWLEDGMENTS We thank Drs. Gilbert Di Paolo and Guanghou Shui for discussion and critical reading of this manuscript.

ABBREVIATIONS ESI-MS FRET IPs PtdIns PIs DAG PUFA TLC HPLC DEAE CID GC-MS

electrospray ionization mass spectrometry fluorescence resonance energy transfer inositolpolyphosphates phosphatidylinositol phosphoinositides diacylglycerol polyunsaturated fatty acid thin-layer chromatography high-performance liquid chromatography diethylaminoethyl collision-induced dissociation gas chromatography-mass spectrometry

References 1. Cantley, L.C. The phosphoinositide 3-kinase pathway. Science. 2002, 296, 1655–1657. 2. Hammond, G.; Thomas, C. L.; Schiavo, G. Nuclear phosphoinositides and their functions. Curr Top Microbiol Immunol. 2004, 282, 177–206. 3. Wenk, M.R. and De Camilli, P. Inaugural Article: Protein-lipid interactions and phosphoinositide metabolism in membrane traffic: insights from vesicle recycling in nerve terminals. Proc Natl Acad Sci U S A. 2004, 101, 8262–9. 4. Tanaka, C. and Nishizuka, Y. The protein kinase C family for neuronal signaling. Annu Rev Neurosci. 1994, 17, 551–67. 5. Berridge, M.J. Inositol trisphosphate and calcium signaling. Nature. 1993, 361, 315–325. 6. Lemmon, M.A. Phosphoinositide recognition domains. Traffic. 2003, 4, 201–213. 7. De Matteis, M.A. and Godi, A. PI-loting membrane traffic. Nat Cell Biol. 2004, 6, 487–492. 8. Clement, S., Krause, U., Desmedt, F., Tanti, J. F., Behrends, J., Pesesse, X., Sasaki, T., Penninger, J., Doherty, M., Malaisse, W., Dumont, J.E., Le Marchand-Brustel, Y., Erneux, C., Hue, L., and Schurmans, S. The lipid phosphatase SHIP2 controls insulin sensitivity. Nature. 2001, 409, 92–97.

Phosphoinositide Profiling in Complex Lipid Mixtures

259

9. Nussbaum, R.L. and Suchy, S.F. (1998) in Metabolic and Molecular Basis of Inherited Disease., Eds., Scriver, C.R., Beaudet, A.L., Sly, W.S. and Valle, D. (McGraw-Hill, New York). 10. Cremona, O., Di Paolo, G., Wenk, M.R., Luthi, A., Kim, W.T., Takei, K., Daniell, L.; Nemoto, Y., Shears, S.B., Flavell, R.A., McCormick, D.A., and De Camilli, P. Essential role of phosphoinositide metabolism in synaptic vesicle recycling. Cell. 1999, 99, 179–188. 11. Nystuen, A., Legare, M.E., Shultz, L.D. and Frankel, W.N. A null mutation in inositol polyphosphate 4-phosphatase type I causes selective neuronal loss in weeble mutant mice. Neuron. 2001, 32, 203–212. 12. Di Paolo, G., Moskowitz, H. S., Gipson, K., Wenk, M. R., Voronov, S., Obayashi, M.; Flavell, R., Fitzsimonds, R. M., Ryan, T. A., and De Camilli, P. Impaired PI(4,5)P2 synthesis in nerve terminalss. Nature. 2004, 431, 415–422. 13. Soares, J.C. and Mallinger, A.G. Intracellular phosphatidylinositol pathway abnormalities in bipolar disorder patients. Psychopharmacol Bull. 1997, 33, 685–691. 14. Acharya, J.K., Labarca, P., Delgado, R., Jalink, K. and Zuker, C.S. Synaptic defects and compensatory regulation of inositol metabolism in inositol polyphosphate 1phosphatase mutants. Neuron. 1998, 20, 1219–1229. 15. Brumell, J. H. and Grinstein, S. Role of lipid-mediated signal transduction in bacterial internalization. Cell Microbiol. 2003, 5, 287–297. 16. Brigl, M. and Brenner, M.B. CD1: Antigen Presentation and T Cell Function. Annu Rev Immunol. 2004, 22, 817–890. 17. Folch, J. Complete fractionation of brain cephalin:isolation from it of phosphatidyl serine, phosphatidyl ethanolamine, and diphosphoinositide. J Biol Chem. 1949, 177, 497–504. 18. Cronholm, T., Viestam-Rains, M., and Sjovall. Decreassed content of arachidonoyl species of phosphatidylinositol phosphates in pancreas of rats fed on an ethanolcontaining diet. Biochem J. 1992, 287, 925–928. 19. Stephens, L.R., Jackson, T.R., and Hawkins, P.T. Agonist-stimulated synthesis of phosphatidylinositol(3,4,5)-trisphosphate: a new intracellualr signalling system? Biochem Biophys Acta. 1993, 1179, 27–75. 20. Hokin, L.E. and Hokin, M.R. Effects of acetylcholine on the turnover of phosphoryl units in individual phospholipids of pancreas slices and brain cortex slices. Biochim Biophys Acta. 1955, 18, 102–110. 21. Hawthorne, J.N.; Pickard, M.R. Phospholipids in synaptic function. J Neurochem. 1979, 32, 5–14. 22. Mitchell, R.H. Inositol phospholipids and cell surface receptor function. Biochim Biophys Acta. 1975, 415, 81–147. 23. Lee, C.H. and Hajra, A.K. Molecular species of diacylglycerols and phosphoglycerides and the postmortem changes in the molecular species of diacylglycerols in rat brains. J Neurochem. 1991, 56, 370–379. 24. Balla, T. and Varnai, P. Visualizing cellular phosphoinositide pools with GFP-fused protein- modules. Sci STKE. 2002, 2002, L3. 25. Bird, I.M. (1998) in Methods Mol Biol, Vol. 105, pp. 1–9. 26. Wenk, M.R. and De Camilli, P. Assembly of endocytosis-associated proteins on liposomes. Methods Enzymol. 2003, 372, 248–260. 27. Anderson, K.E., Stephens, L., and Hawkins, P.T. (1999) in Signal Transduction: A practical approach, Ed., Milligan, G. Oxford University Press, 283–300. 28. Bird, I.M. (1998) in Methods Mol Biol, Vol. 105, 25–45.

260

Functional Lipidomics

29. Sastry, P.S. (1996) in Manual on Membrane Lipids, Ed. Prasad, R. (Springer-Verlag, Berlin), 167–200. 30. McLaughlin, S., Wang, J., Gambhir, A. and Murray, D. PIP(2) and proteins: interactions, organization, and information flow. Annu Rev Biophys Biomol Struct. 2002, 31, 151–175. 31. Singh, A.K. and Jiang, Y. Quantitative chromatographic analysis of inositol phospholipids and related compounds. J Chromatogr B Biomed Appl. 1995, 671, 255–280. 32. Zhang, X., Loijens, J.C., Boronenkov, I.V., Parker, G.J., Norris, F.A., Chen, J., Thum, O., Prestwich, G.D., Majerus, P.W., and Anderson, R.A. Phosphatidylinositol-4-phosphate 5-kinase isozymes catalyze the synthesis of 3-phosphate-containing phosphatidylinositol signaling molecules. J Biol Chem. 1997, 272, 17756–17761. 33. Drobak, B.K., Brewin, N.J., and Hernandez, L.E. Extraction, separation, and analysis of plant phosphoinositides and complex glycolipids. Methods Mol Biol. 2000, 141, 157–174. 34. Shaw, N. The detection of lipids on thin-layer chromatograms with the periodateSchiff reagents. Biochim Biophys Acta. 1968, 164, 435–436. 35. Siakotos, A.N. Analytical separation of nonlipid water soluble substances and gangliosides from other lipids by dextran gel column chromatography. J Am Oil Chem Soc. 1965, 42, 913–919. 36. Boettcher, C.J.F., Gent, C.M.V., and Fries, C. Anal Chim Acta. 1961, 24, 203–204. 37. Nakamura, T., Hatori, Y., Yamada, K., Ikeda, M., and Yuzuriha, T. A high-performance liquid chromatographic method for the determination of polyphosphoinositides in brain. Anal Biochem. 1989, 179, 127–130. 38. Folch, J., Ascoli, I., Lees, M., Meath, J.A., and Le, B.N. Preparation of lipide extracts from brain tissue. J Biol Chem. 1951, 191, 833–841. 39. Hendrickson, H.S., and Ballou, C.E. Ion exchange chromatography of intact brain phosphoinositides on diethylaminoethyl cellulose by gradient salt elution in a mixed solvent system. J Biol Chem. 1964, 239, 1369–1373. 40. Low, M.G. (1990) in Methods in Inositide Research, Ed., Irvine, R.F., Raven Press, New York, 145–151. 41. Schacht, J. Extraction and purification of polyphosphoinositides. Methods Enzymol. 1981, 72, 626–631. 42. Dragani, L.K., Berrie, C.P., Corda, D., and Rotilio, D. Analysis of glycerophosphoinositol by liquid chromatography-electrospray ionisation tandem mass spectrometry using a beta-cyclodextrin-bonded column. J Chromatogr B Analyt Technol Biomed Life Sci. 2004, 802, 283–289. 43. Serunian, L.A., Auger, K.R., and Cantley, L.C. Identification and quantification of polyphosphoinositides produced in response to platelet-derived growth factor stimulation. Methods Enzymol. 1991, 198, 78–87. 44. Morris, J.B., Hinchliffe, K.A., Ciruela, A., Letcher, A.J., and Irvine, R.F. Thrombin stimulation of platelets causes an increase in phosphatidylinositol 5-phosphate revealed by mass assay. FEBS Lett. 2000, 475, 57–60. 45. Nasuhoglu, C., Feng, S., Mao, J., Yamamoto, M., Yin, H.L., Earnest, S., Barylko, B., Albanesi, J.P., and Hilgemann, D.W. Nonradioactive analysis of phosphatidylinositides and other anionic phospholipids by anion-exchange high-performance liquid chromatography with suppressed conductivity detection. Anal Biochem. 2002, 301, 243–254. 46. Cullen, P.J., Chung, S.K., Chang, Y.T., Dawson, A.P., and Irvine, R.F. Specificity of the purified inositol (1,3,4,5) tetrakisphosphate-binding protein from porcine platelets. FEBS Lett. 1995, 358, 240–242.

Phosphoinositide Profiling in Complex Lipid Mixtures

261

47. Smart, D. Measurement of inositol(1,4,5)trisphosphate using a stereospecific radioreceptor mass assay. Methods Mol Biol. 1999, 114, 175–183. 48. Hamman, B.D., Pollok, B.A., Bennett, T., Allen, J., and Heim, R. Binding of a Pleckstrin homology domain protein to phosphoinositide in membranes: a miniaturized FRET-based assay for drug screening. J Biomol Screen. 2002, 7, 45–55. 49. Gray, A., Olsson, H., Batty, I.H., Priganica, L., and Peter Downes, C. Nonradioactive methods for the assay of phosphoinositide 3-kinases and phosphoinositide phosphatases and selective detection of signaling lipids in cell and tissue extracts. Anal Biochem. 2003, 313, 234–245. 50. Sherman, W.R., Ackermann, K.E., Bateman, R.H., Green, B.N., and Lewis, I. Massanalysed ion kinetic energy spectra and B1E-B2 triple sector mass spectrometric analysis of phosphoinositides by fast atom bombardment. Biomed Mass Spectrom. 1985, 12, 409–413. 51. Jungalwala, F.B., Evans, J.E., and McCluer, R.H. Compositional and molecular species analysis of phospholipids by high performance liquid chromatography coupled with chemical ionization mass spectrometry. J Lipid Res. 1984, 25, 738–749. 52. Kim, H.Y., and Salem, N., Jr. Application of thermospray high-performance liquid chromatography/mass spectrometry for the determination of phospholipids and related compounds. Anal Chem. 1987, 59, 722–726. 53. Smith, P.B., Snyder, A.P., and Harden, C.S. Characterization of bacterial phospholipids by electrospray ionization tandem mass spectrometry. Anal Chem. 1995, 67, 1824–1830. 54. Brugger, B., Erben, G., Sandhoff, R., Wieland, F.T., and Lehmann, W.D. Quantitative analysis of biological membrane lipids at the low picomole level by nano-electrospray ionization tandem mass spectrometry. Proc Natl Acad Sci U S A. 1997, 94, 2339–2344. 55. Ivanova, P.T., Cerda, B.A., Horn, D.M., Cohen, J.S., McLafferty, F.W., and Brown, H.A. Electrospray ionization mass spectrometry analysis of changes in phospholipids in RBL-2H3 mastocytoma cells during degranulation. Proc Natl Acad Sci U S A. 2001, 98, 7152–7157. 56. Koivusalo, M., Haimi, P., Heikinheimo, L., Kostiainen, R., and Somerharju, P. Quantitative determination of phospholipid compositions by ESI-MS: effects of acyl chain length, unsaturation, and lipid concentration on instrument response. J Lipid Res. 2001, 42, 663–672. 57. Kerwin, J.L., Tuininga, A.R., and Ericsson, L.H. Identification of molecular species of glycerophospholipids and sphingomyelin using electrospray mass spectrometry. J Lipid Res. 1994, 35, 1102–14. 58. Hsu, F.F. and Turk, J. Characterization of phosphatidylinositol, phosphatidylinositol4- phosphate, and phosphatidylinositol-4,5-bisphosphate by electrospray ionization tandem mass spectrometry: a mechanistic study. J Am Soc Mass Spectrom. 2000, 11, 986–999. 59. Welti, R., Li, W., Li, M., Sang, Y., Biesiada, H., Zhou, H.E., Rajashekar, C.B., Williams, T.D., and Wang, X. Profiling membrane lipids in plant stress responses. Role of phospholipase D alpha in freezing-induced lipid changes in Arabidopsis. J Biol Chem. 2002, 277, 31994–32002. 60. Wenk, M.R., Lucast, L., Di Paolo, G., Romanelli, A.J., Suchy, S.F., Nussbaum, R.L., Cline, G.W., Shulman, G.I., McMurray, W., and De Camilli, P. Phosphoinositide profiling in complex lipid mixtures using electrospray ionization mass spectrometry. Nat. Biotechnol. 2003, 21, 813–817.

262

Functional Lipidomics

61. Ekroos, K., Chernushevich, I. V., Simons, K., and Shevchenko, A. Quantitative profiling of phospholipids by multiple precursor ion scanning on a hybrid quadrupole time-of-flight mass spectrometer. Anal Chem. 2002, 74, 941–949. 62. Holub, B.J., Kuksis, A., and Thompson, W. Molecular species of mono-, di-, and triphosphoinositides of bovine brain. J Lipid Res. 1970, 11, 558–564. 63. Muller, M., Schiller, J., Petkovic, M., Oehrl, W., Heinze, R., Wetzker, R., Arnold, K., and Arnhold, J. Limits for the detection of (poly-)phosphoinositides by matrixassisted laser desorption and ionization time-of-flight mass spectrometry (MALDITOF MS). Chem Phys Lipids. 2001, 110, 151–164. 64. Hsu, F.F., Turk, J., and Gross, M.L. Structural distinction among inositol phosphate isomers using high-energy and low-energy collisional-activated dissociation tandem mass spectrometry with electrospray ionization. J Mass Spectrom. 2003, 38, 447–457. 65. Gunnarsson, T., Ekblad, L., Karlsson, A., Michelsen, P., Odham, G., and Jergil, B. Separation of polyphosphoinositides using normal-phase high-performance liquid chromatography and evaporative light scattering detection or electrospray mass spectrometry. Anal Biochem. 1997, 254, 293–296. 66. De Camilli, P., Emr, S.D., McPherson, P.S., and Novick, P. Phosphoinositides as regulators in membrane traffic. Science. 1996, 271, 1533–1539. 67. Majerus, P.W., Kisseleva, M.V., and Norris, F.A. The role of phosphatases in inositol signaling reactions. J Biol Chem. 1999, 274, 10669–10672. 68. Odorizzi, G., Babst, M., and Emr, S.D. Phosphoinositide signaling and the regulation of membrane trafficking in yeast. Trends Biochem Sci. 2000, 25, 229–235. 69. Fruman, D.A. and Cantley, L.C. Phosphoinositide 3-kinase in immunological systems. Semin Immunol. 2002, 14, 7–18. 70. Doughman, R.L., Firestone, A.J., and Anderson, R.A. Phosphatidylinositol phosphate kinases put PI4,5P(2) in its place. J Membr Biol. 2003, 194, 77–89. 71. Pendaries, C., Tronchere, H., Plantavid, M., and Payrastre, B. Phosphoinositide signaling disorders in human diseases. FEBS Lett. 2003, 546, 25–31. 72. Serunian, L.A., Haber, M.T., Fukui, T., Kim, J.W., Rhee, S.G., Lowenstein, J.M., and Cantley, L.C. Polyphosphoinositides produced by phosphatidylinositol 3-kinase are poor substrates for phospholipases C from rat liver and bovine brain. J Biol Chem. 1989, 264, 17809–17815. 73. Anderson, R.A., Boronenkov, I.V., Doughman, S.D., Kunz, J., and Loijens, J.C. Phosphatidylinositol phosphate kinases, a multifaceted family of signaling enzymes. J Biol Chem. 1999, 274, 9907–10. 74. Taylor, G.S. and Dixon, J.E. Assaying phosphoinositide phosphatases. Methods Mol Biol. 2004, 284, 217–228. 75. Harder, K.W., Owen, P., Wong, L.K., Aebersold, R., Clark-Lewis, I., and Jirik, F.R. Characterization and kinetic analysis of the intracellular domain of human protein tyrosine phosphatase beta (HPTP beta) using synthetic phosphopeptides. Biochem J. 1994, 298 (Pt. 2), 395–401. 76. Lin, S., Fischl, A. S., Bi, X., Parce, W. Separation of phospholipids in microfluidic chip device: application to high-throughput screening assays for lipid-modifying enzymes. Anal Biochem. 2003, 314, 97–107. 77. Marsh, D. (1990) Handbook of Lipid Bilayers, CRC Press, Boca Raton. 78. Hama, H., Torabinejad, J., Prestwich, G.D., and DeWald, D.B. Measurement and immunofluorescence of cellular phosphoinositides. Methods Mol Biol. 2004, 284, 243–258.

12

Multiplexed Lipid Arrays of Antiimmunoglobulin M–Induced Changes in the Glycerophospholipid Composition of WEHI231 Cells Stephen B. Milne, Jeffrey S. Forrester, Pavlina T. Ivanova, Michelle D. Armstrong, and H. Alex Brown

CONTENTS Abstract ..................................................................................................................264 Introduction............................................................................................................264 Results....................................................................................................................265 Identification of Lipids by Mass Spectrometry...........................................265 Mathematical Analysis of Mass Spectrometry Data ...................................269 Comprehensive Analysis of Lipid Changes under Stimulation ......269 Lipid Arrays .....................................................................................275 Glycerophospholipid Changes in Basal vs. AIG-Stimulated WEHI-231 Cells...............................................................................276 Discussion ..............................................................................................................276 Methods and Protocols ..........................................................................................282 Cell Extraction and Reconstitution..............................................................282 Mass Spectrometry Analysis of Phospholipid Cell Extracts ......................282 Acknowledgment ...................................................................................................282 References..............................................................................................................283

263

264

Functional Lipidomics

ABSTRACT A goal of the Alliance for Cellular Signaling (AfCS) is to identify the diverse participants that compose an intracellular signaling network. Phospholipids are important participants in transmembrane signaling processes as well as direct mediators of the dynamic aspects of cell membrane structure. Therefore, identifying the contextual changes in membrane lipid composition (in addition to that of genes and proteins) is essential in achieving a comprehensive understanding of signaling networks in cells. Recent advances in high-throughput electrospray ionization mass spectroscopy (ESI-MS), coupled with new computational approaches, have greatly facilitated this goal. One cell type of interest to the AfCS is the splenic B lymphocyte and its experimental surrogate, the WEHI-231 cell (1, 2). Here, we identify more than 200 species of glycerophospholipids from total membrane extracts of WEHI231 cells and qualitatively measure pattern response changes initiated by stimulation of cell surface receptors. In these studies, WEHI cells were treated with antiimmunoglobulin M antibody (AIG) to stimulate the B-cell receptor. The response to AIG stimulation was a conspicuous change in a broad range of phospholipids. An overall temporal trend was observed in which lipid concentration changes were detected by 6 min, pattern changes peaked by 15 min, and by 4 h of stimulation, the cells had largely returned to their prestimulated composition. Statistically significant decreases were observed in many species of phospholipids along with concomitant increases in lysophospholipid concentrations. This study represents the most comprehensive analysis of membrane phospholipid changes in any cell type to date. The procedure described can be applied to any mammalian cell type and provides a basis for the comprehensive study of lipid-signal transduction. Taken together, these changes form unique patterns that will be used to discriminate ligand-stimulated events and to model signaling pathways that lead to developmental and phenotypic changes in cells.

INTRODUCTION Glycerolipids and glycerophospholipids are key molecules in many inter- and intracellular signal transduction pathways. Some of these processes, such as the phospholipase C (PLC)-driven phosphatidylinositol cycle, have been known, since the 1950s, to participate in signal transduction (3). Phosphatidic acid (PA) and diacylglycerol (DAG) participate in signaling pathways initiated by growth factors and Gprotein-coupled receptors (GPCRs) as well. However, the importance of additional lipid classes as cellular signals has only more recently been appreciated. These include the participation of lysophosphatidic acid (LPA) in apoptosis and lysophospholipids as ligands for certain GPCRs involved in cardiac, neuronal, and immunological processes (4–7). Until recently, the detection and identification of low-concentration lipids were quite difficult. Thin-layer chromatography (TLC) was utilized for many decades to separate lipid classes. With the advent of gas chromatography and gas chromatography mass spectrometry (GC-MS), class separation by TLC followed by hydrolysis and derivatization made it possible to identify individual fatty acid species. One of

Multiplexed Lipid Arrays of Antiimmunoglobulin

265

the drawbacks to this method was that large amounts of lipids were normally required. With the introduction of fast atom bombardment mass spectrometry (FABMS), routine analysis of intact phospholipids was possible (8). More recently, ESIMS has greatly simplified the procedures for lipid analysis. The soft ionization process associated with ESI-MS results in decreased molecular ion decomposition and lower detection limits compared to FAB-MS (9,10). A primary goal of these studies has been to identify the phospholipids participating in the cellular signaling pathways downstream of the AIG pathway in WEHI231 cells as well as other ligands of interest to the AfCS. In this way, lipid-signaling components can be integrated into the larger cellular signaling network, appreciated as part of the molecular response elements used by cells to transduce information from the cell surface, and better understood regarding the sometimes mysterious events occurring within cellular bilayers. As such, this report represents the most extensive analysis of cellular lipid content and changes determined to date. The ability to monitor changes in cellular lipid content in parallel with the ability of other AfCS laboratories to determine changes in gene expression, protein modifications, and production of second messengers represents a powerful new approach to understanding the contextual changes that determine cellular responses to complex biological stimuli.

RESULTS IDENTIFICATION

OF

LIPIDS

BY

MASS SPECTROMETRY

Identification of individual glycerophospholipids present in the total lipid extracts (both basal and AIG-stimulated) was accomplished by tandem mass spectrometry (ESI-MS/MS). Resolution and characterization of glycerophospholipids in an unprocessed total lipid extract are based on the predisposition of each lipid class to acquire positive or negative charges under the source energy. A single molecular ion is present with a mass-to-charge ratio (m/z) that refers to the monoisotopic molecular weight. Collision-induced dissociation (CID) of the peaks of interest yielded fragmentation patterns, which were used to unambiguously identify the lipids present at a particular m/z value (Figure 12.1 provides an illustration of this procedure) (11–23). For MS/MS, both positive- and negative-mode ionization were utilized. Traditionally, degree of structural information obtained as a result of this analysis varies by the type of instrumentation used. In negative ionization mode, triple quadrupole instruments tend to yield sn-1 and sn-2 fatty acid residue fragments, whereas ion traps form more lysolipid fragments (19). Positive-ion ESI-MS/MS spectra from ion trap instruments are more likely to create lyso-PC fragmentation products, which reveal the fatty acid composition of the lipid. However, under our triple quadrupole MS experimental conditions, only glycerophospholipid head group information was routinely obtainable from positive-mode fragmentation. Three lipid classes were analyzed in positive ESI mode: phosphatidylcholines (PCs), phosphatidylethanolamines (PEs), and sphingomyelins (SMs). The cholinecontaining species, PCs and SMs, both show a characteristic m/z 184 phosphocholine

266

Functional Lipidomics

ESI/MS (–) mode spectra 885.1

450

650

750 850 m/z

950

1050

1150

818.0

450

0 150

350

550 m/z

Fragmentation and identification of 732 m/z peak

750

(b)

682.1

550

650

750

850

950

1050

1150

m/z

184.1 32:1 PC M+1 732.5

563.8

465.8

Relative abundance

100 90 80 70 60 50 40 30 20 10

550

(a)

747.1

100 90 80 70 60 50 40 30 20 10 0

885.5

282.9 240.7 302.9 419.2 250

350

450

38:4 PI M-1 M-20:4

680.4

Relative abundance

760.5

100 90 80 70 60 50 40 30 20 10 0

M-20:4–163

563.5

Inositolphosphate-H2O 18:0 20:4

732.5

PC headgroup

100 90 80 70 60 50 40 30 20 10 0

Relative abundance

Relative abundance

ESI/MS(+) mode spectra

581.0 550 m/z

650

750

850

Fragmentation and identification of 885 m/z peak

FIGURE 12.1 Fragmentation and identification of lipid species. Individual lipid species from the total cell extract were isolated and fragmented using ESI-MS/MS. Positive-mode analysis was utilized in the identification of three phospholipid classes. Negative-mode analysis was used to assign five classes and to determine fatty acid compositions

head group peak, as well as an [M+H-59]+ peak corresponding to the neutral loss of (CH3)3N. In addition to the diacyl PC compounds, a large number of plasmanyl and plasmenyl phosphocholines were also identified. Together, over 100 cholinecontaining lipids were identified. Fragmentation of PEs exclusively yielded one peak, an [M+H-141]+ ion from the neutral loss of the phosphoethanolamine head group. Again, plasmanyl and plasmenyl lipids were a large proportion of the PE species (over 40) identified. Five lipid classes were detected in negative ESI mode: phosphatidylinositols (PIs), phosphatidylserines (PSs), phosphatidylglycerols (PGs), glycerophosphatidic acids, and PEs. Negative-mode fragmentation of these species yielded a wealth of structural information. In each case, head group fragmentation, lysolipid formation, and fatty acid fragments aided in the lipid identification process. Phosphatidylinositol fragmentation generated a wide variety of product (“daughter”) ions. There were four types of LPA and lysophosphatidylinositols, PA, and five characteristic head group fragments were used in identifying the 27 observed PI and lyso-PI species. Similarly, 33 distinct species of PS and lyso-PS compounds were identified from their phosphatidic (PA) and LPA fragments. A negative-mode-fragmentation library of the PSs is provided as an example in Table 12.1. Fragmentation tables for the remaining phospholipid classes (for both fragmentation modes) are available online at http://www.signaling-gateway.org/reports/v1/DA0011/SupplTables.htm. PC compounds were not identified during the routine negative-mode scans. However, it was determined that these compounds were detectable after the addition of ammonium

34:1 PS

34:0 PS 36:4 PS

36:3 36:2 36:1 36:0 38:6

38:5 PS 38:4 PS

38:3 PS 38:2 PS

760

762 782

784 786 788 790 806

808 810

812 814

PS PS PS PS PS

34:2 PS

758

PS PS PS PS PS

30:0 32:2 32:1 32:0 34:3

Compound

706 730 732 734 756

m/z

725 727

721 723

697 699 701 703 719

675 695

673

671

PA-H 619 643 645 647 669

TABLE 12.1 Fragmentation Table for PSs

417 (18:1), 417 (18:1), (20:3) 419 (18:0), 419 (18:0),

(22:6)

(20:3) (18:1), 419 (18:0) (18:0)

441 (20:3) 443 (20:2)

439 (20:4) 419 (18:0), 439 (20:4), 441

391 (16:0), 441 415 (18:2), 417 417 (18:1), 419 419 (18:0) 391 (16:0), 463

419 (18:0) 439 (20:4)

391 (16:0), 417 (18:1), 419

391 (16:0), 415 (18:2), 417

LPA-H2O-H (14:0), 390 (16:0) (16:1) (16:1), 391 (16:0) (16:0) (16:0), 415 (18:2)

389 (16:1), (18:1) 389 (16:1), (18:0) 391 (16:0), 391 (16:0),

363 389 389 391 391 (16:1) (16:1), 409 (16:0) (16:0) (16:1)

(16:0), (18:2), (18:1), (18:0) (16:0), 481 (22:6)

459 (20:3) 435 (18:1), 437 (18:0) 437 (18:0)

437 (18:0), 459 (20:3) 437 (18:0), 461 (20:2)

435 (18:1) 435 (18:1), 437 (18:0), 457 (20:4)

409 433 435 437 409

437 (18:0) 409 (16:0)

433 (18:2), 435 (18:1), 409 (16:0), 407 (16:1) 435 (18:1), 437 (18:0)

407 407 409 407

LPA-H

Fragmentation Ions Fatty Acid-H 227 (14:0), 255 (16:0) 253 (16:1) 253 (16:1), 255 (16:0) 255 (16:0) 253 (16:1), 255 (16:1), 279 (18:2), 281 (18:1) 253 (16:1), 255 (16:0), 279 (18:2), 281 (18:1) 253 (16:1), 255 (16:0), 281 (18:1), 283 (18:0) 255 (16:0), 283 (18:0) 253 (16:1), 255 (16:0), 303 (20:4), 305 (20:3) 255 (16:0), 305 (20:3) 279 (18:2), 281 (18:1), 283 (18:0) 281 (18:1), 283 (18:0) 283 (18:0) 253 (16:1), 255 (16:0), 327 (22:6), 329 (22:5) 281 (18:1), 303 (20:4) 281 (18:1), 283 (18:0), 303 (20:4), 305 (20:3) 283 (18:0), 305 (20:3) 283 (18:0), 307 (20:2)

153 153

153 153

153 153

153

153

153

153

GP-H2O-H

Multiplexed Lipid Arrays of Antiimmunoglobulin 267

38:1 PS 40:7 PS 40:6 PS

40:5 PS

40:4 PS 16:1 LPS 16:0 LPS 18:1 LPS 18:0 LPS 20:4 LPS 20:3 LPS 20:1 LPS 22:6 LPS 22:5 LPS 22:4 LPS

816 832 834

836

838 494 496 522 524 544 546 550 568 570 572

751 407 409 435 437 457 459 463 481 483 485

749

PA-H 729 745 747

467 (22:4)

463 (22:6)

LPA-H2O-H 419 (18:0) 417 (18:1), 463 (22:6) 417 (18:1), 419 (18:0), 463 (22:6), 465 (22:5) 417 (18:1), 419 (18:0), 465 (22:5), 467 (22:4) 419 (18:0), 467 (22:4) 437 407 409 435 437 457 459 463 481 483 485 (18:0) (16:1) (16:0) (18:1) (18:0) (20:4) (20:3) (20:1) (22:6) (22:5) (22:4)

437 (18:0), 483 (22:5), 485 (22:4)

435 (18:1), 481 (22:6) 435 (18:1), 437 (18:0), 481 (22:6)

LPA-H

Fragmentation Ions Fatty Acid-H 283 (18:0), 309 (20:1) 281 (18:1), 327 (22:6) 279 (18:2), 281 (18:1), 283 (18:0), 327 (22:6), 329 (22:5), 331 (22:4) 281 (18:1), 283 (18:0), 329 (22:5), 331 (22:4) 283 (18:0), 331 (22:4) 253 (16:1) 255 (16:0) 281 (18:1) 283 (18:0) 303 (20:4) 305 (20:3) 309 (20:1) 327 (22:6) 329 (22:5) 331 (22:4)

153 153 153

153 153 153 153 153 153

153

153

GP-H2O-H 153

Note: Using negative-mode ESI-MS/MS, 33 PS and lyso-PS species were identified. The numbers in parentheses following fragment ions (FA:D) refer to the total number of fatty acid carbons (FA) and fatty acid carbon–carbon double bonds (D). GP = glycerophosphate

Compound

m/z

TABLE 12.1 Fragmentation Table for PSs (continued)

268 Functional Lipidomics

Multiplexed Lipid Arrays of Antiimmunoglobulin

269

acetate (15,18,23). Two important categories of signaling lipids were not included in this analysis. DAG was not routinely detected under the optimized conditions for triple quadrupole MS described here; however, DAG species can be detected using a Fourier transform ion cyclotron resonance (FT-ICR) instrument (16). We have also found that DAG can be detected using a triple quadrupole MS but requires formation of a sodium adduct. In the current study, well over 200 glycerophospholipids have been detected and unambiguously identified in WEHI-231 total lipid extracts. A tabular listing of all identified lipids for both positive and negative MS modes is shown in Table 12.2.

MATHEMATICAL ANALYSIS

OF

MASS SPECTROMETRY DATA

Comprehensive Analysis of Lipid Changes under Stimulation The goal of computational analysis is the construction of an array containing the m/z ratios for peaks observed within a mass spectrometry experiment that displays the comprehensive changes in these species between two experimental conditions (e.g., addition of a ligand at a given concentration) over the defined time course. To accomplish this goal, computer algorithms were developed by our group to achieve the following: (1) smooth raw mass spectrometry data to remove extraneous portions, (2) identify peaks within each data set and normalize their signal intensities, (3) create baseline profiles of the transformed signal at each peak identified, (4) statistically compare these baseline conditions with the observed stimulated results, and (5) handle exceptional cases in which assumptions are not validated by the data. All data analysis programs were written in the S-Plus V3.3 for Microsoft Windows environment. These steps are outlined in Figure 12.2. The data analysis begins with the conversion of Xcalibur raw files into text for loading into S-Plus. After the files are loaded, each data set is smoothed using a kernel regression estimator (24). The effect of this smoothing is the removal of shoulders from peaks within the data set, which reduces the overall number of peaks to be analyzed. This smoothed data set is then transformed in the second portion of the analysis. Because the absolute signal intensity at a particular m/z value exhibits a high variability, even between apparent exact replicates, the development of a unitless number was desirable for the comparison of this data. The primary characteristic for this transformation was that it should be a more robust measure of the signal strength at a particular m/z value with respect to the overall pattern observed. Our software is set up for two distinct methods for transforming the data. In the first method, the mean and standard deviation of a function of the observed intensities for the complete spectra are calculated. These statistics are then used to transform the signal intensity at each of the m/z values as: I* = (I − mean)/SD (i.e., the transformed intensity is represented as the number of standard deviations the signal occurs above or below the mean signal strength). Thus, a signal with intensity equal to the mean intensity of the data set would receive a score of zero, and any signal with intensity below the mean would receive a negative score. The second method involves using the rank of the signal in comparison with the other points in the data

496 500 502 506 509 511

m/z 409 421 424 437 440 450 452 454 455 466 467 468 469 478 480 482 483 494

18:1 LPG 18:0 LPG

16:0 LPS 20:4 LPE 20:3 LPE

16:0 LPG 16:1 LPS

18:1 LPE 18:0 LPE

14:0 LPG 16:0 PE std 16:0 PE std

16:1 LPE 16:0 LPE

Negative 16:0 LPA 18:0p LPA 14:0 LPE 18:0 LPA

20:2 LPE

16:0 LPC

16:1 LPC

18:1 LPE 18:0 LPE

16:0 PE std, 14:0 LPC 16:0 PE std

16:1 LPE 16:0 LPE

12:0 LPC

Positive

732 734 736 738 740 742

m/z 700 701 702 703 704 706 712 714 716 718 719 720 721 722 723 724 728 730

TABLE 12.2 Library of Identified Glycerophospholipid Species

PS PE PE PE PE PG

36:1 PEp a/o 36:2 PEe 32:2 PS, 36:0 PEp a/o 36:1 PEe 32:1 PS 32:0 PS 36:5 PE 36:4 PE 36:3 PE 36:2 PE

32:0 PG, 38:5 PA 36:4 PEp 38:4 PA

30:0 34:3 34:2 34:1 34:0 32:1

36:1 PA 34:0PEp a/o 34:1 PEe

Negative

34:4 PCp, 36:5 PE 36:4 PE 36:3 PE, 34:2 PCp a/o 34:3 PCe

32:1 PC, 36:0 PEp a/o 36:1 PEe 32:0 PC

36:4 PEp 32:3 PC, 36:2 PEp a/o 36:3 PEe 32:2 PC, 36:1 PEp a/o 36:2 PEe

32:0 PCe, 34:0 PE

Positive 30:3 PC d 18:1/16:1 SM 30:2 PC d 18:1/16:0 SM 30:1 PC, 34:0 PEp a/o 34:1 PEe 30:0 PC 34:4 PE 34:3 PE 34:2 PE, 32:1 PCp a/o 32:2 PCe 34:1 PE, 32:0 PCp a/o 32:1 PCe

830 832 833 834 835 836

m/z 800 802 804 805 806 807 808 809 810 812 814 816 818 820 822 824 826 828

40:7 34:2 40:6 34:1 40:5

32:2 38:6 32:1 38:5 32:0 38:4 38:3 38:2 38:1

PS PI PS PI PS

PI PS PI PS PI PS PS PS PS

Negative

PC PS, 38:3 PC PC PC PCp a/o 38:0 PCp a/o 40:6 PCp a/o 40:5 PCp a/o 40:4 PCp a/o 40:3 PCp a/o 40:2

PC PCe PCe PCe PCe PCe

40:5 PC

40:6 PC

40:0 PCp a/o 40:1 PCe 40:0 PCe

38:4 38:4 38:2 38:1 40:6 40:5 40:4 40:3 40:2 40:1

38:5 PC

38:6 PC

Positive 38:1 PCp a/o 38:2 PCe 38:0 PCp a/o 38:1 PCe 38:0 PCe

270 Functional Lipidomics

522 524 526 536 544 546 548 550 552 568 569 570 571 572 580 597 599 619 621 636 662 663 664 671 673 674 676 678 686

LPI LPI LPI LPI

18:1 18:0 20:4 20:3

34:2 PA 34:1 PA 32:1 PEe a/o 32:0 PEp 32:0 PEe

30:0 PE

LPS LPI LPS LPI LPS

22:6 16:1 22:5 16:0 22:4

20:1 LPS

18:1 LPS 22:6 LPE, 18:0 LPS 22:5 LPE d 18:1/16:0 ceramide 20:4 LPS 20:3 LPS

LPC LPC LPC LPC LPC LPC

32:1 PEp a/o 32:2 PEe 28:1 PC, 32:1 PEe 32:0 PEe, 28:0 PC 30:2 PCp a/o 30:3 PCe

28:0 PCe, 30:0 PE

26:0 PCe

22:4 LPC 22:0 LPC

22:5 LPC

20:4 20:3 20:2 20:1 20:0 22:6

18:1 LPC 18:0 LPC

744 745 746 747 748 750 752 754 756 758 760 762 764 766 768 769 770 771 772 773 774 775 776 778 780 781 782 784 786 30:0 36:4 36:3 36:2

34:3 34:2 34:1 34:0 38:5 38:4 38:3 36:4 38:2 36:3 38:1 36:2 40:6 36:1 40:5

36:1 34:2 38:6 34:1 38:5 38:4

PI PS PS PS

PS PS PS PS, 38:6 PE PE PE PE PG PE PG PE PG PEp PG PEp a/o 40:6 PEe

PE PG PEp PG PEp PEp PCe PEp a/o 38:6 PEe PC, 38:4 PEp a/o 38:5 PEe PC, 38:3 PEp a/o 38:4 PEe PC PC PC, 34:2 PS PC, 38:0 PEe PE PE, 36:4 PCp PE, 36:3 PCp a/o 36:4 PCe

36:4 PC 36:3 PC 36:2 PC, 40:1 PEp a/o 40:2 PEe

36:0 PCe, 38:0 PE a/o 40:6 PEp 36:6 PC, 40:5 PEp a/o 40:6 PEe 36:5 PC

36:0 PCp a/o 36:1 PCe

36:1 PCp a/o 36:2 PCe

38:3 PE 36:2 PCp a/o 36:3 PCe

34:0 38:5 34:5 34:4 34:3 34:2 34:1 34:0 38:6 38:5 38:4

36:1 PE, 34:0 PCp a/o 34:1 PCe

36:2 PE, 34:1 PCp a/o 34:2 PCe 837 838 840 842 844 846 852 854 855 856 857 858 859 860 861 863 864 865 866 868 870 872 874 880 881 882 883 884 885 38:4 PI

38:5 PI

38:6 PI

36:0 PI

36:2 PI 36:1 PI

36:3 PI

36:4 PI

36:5 PI

34:0 PI 40:4 PS PC PC PC PC PC PCp a/o 42:4 PCe PCp a/o 42:3 PCe

PC PC PC PC PC PCp a/o 44:2 PCe

44:0 PCe

44:0 PCp a/o 44:1 PCe

42:4 42:3 42:2 42:1 42:0 44:1

42:5 PC

42:0 PCe

42:0 PCp a/o 42:1 PCe

42:1 PCp a/o 42:2 PCe

40:4 40:3 40:2 40:1 40:0 42:3 42:2

Multiplexed Lipid Arrays of Antiimmunoglobulin 271

32:0 PE

36:2 PA

690

692 699

Positive 32:2 PE, 30:1 PCp a/o 30:2 PCe 32:1 PE, 30:0 PCp a/o 30:1 PCe 32:0 PE, 30:0 PCe

792 794 796 797 798

790

m/z 788

38:4 PG

40:5 PE

40:6 PE, 36:0 PS

Negative 36:1 PS Positive

38:2 PCp a/o 38:3 PCe

40:6 PE, 38:5 PCp a/o 38:6 PCe 40:5 PE, 38:4 PCp a/o 38:5 PCe 40:4 PE, 38:3 PCp a/o 38:4 PCe

40:0 PEe, 36:0 PC a/o 38:6 PCp

36:1 PC

909 911 913

889

m/z 887

40:6 PI 40:5 PI 40:4 PI

38:2 PI

Negative 38:3 PI

Positive

Note: All of the glycerophospholipids identified by ESI-MS/MS fragmentation in positive and negative modes are summarized. PC and PE compounds with lowercase e or p refer to plasmanyl and plasmenyl (alkyl ether and plasmalogen) subspecies, respectively. When plasmanyl and plasmenyl PE or PC species are separated by and/or (a/o), this indicates that one or both species were detected at that m/z.

Negative 32:1 PE

m/z 688

TABLE 12.2 (continued) Library of Identified Glycerophospholipid Species

272 Functional Lipidomics

Multiplexed Lipid Arrays of Antiimmunoglobulin

273

Smooth data set Transform data Identify peaks Basal

Stimulated

Combine data in tables

Combine data in tables

Generate shewhart chart and test for control Compare stimulated time points with “in-control” basal signals Exception handling Generate array

FIGURE 12.2 Flowchart for the analysis of mass spectrometry data in the creation of lipid arrays.

set as the transformed intensity. This method has proven to be highly robust against the wide changes in signal magnitude observed. These transformed-intensity signals are carried into the next level of analysis. After the data has been smoothed and transformed, the algorithm identifies potential peaks by parsing for low–high–low patterns in the data set. These peaks are collected and concatenated with other data sets from the same condition. During this concatenation, the algorithm averages the locations of peaks identified within different data sets to compensate for m/z measurement error. For example, peaks identified at m/z ratios of 768.4 and 768.6 are placed at 768.5 for the analysis. In the next stage of the analysis, the data from the basal condition are used to create a Shewhart control chart for the means of the transformed data. The object of this construction is the identification of m/z ratios for which the signal remains stable in the basal condition during the course of the experiment so that the information can be pooled for subsequent comparison with the stimulated condition. In the analysis of the WEHI-231 cells, we have n = 4 samples at each time point. The mean of these values is plotted along the time axis, and a set of control limits are calculated for this statistic. The area between these limits represents the expected variability in the mean of four observations of the process, not the individual measurements. The limits are computed from process output, assuming the underlying distribution remains stable. Thus, a kind of running hypothesis test is constructed. An example of this stage of the analysis is shown in the left panel of Figure 12.3.

274

Functional Lipidomics

UCL A B C Mean C B A LCL

Time

FIGURE 12.3 Shewhart control chart. The left panel of the figure represents a control chart for the sample mean of the transformed data at a specific m/z value constructed from the basal case at five time points with four measurements each. The y-axis is unitless. The means of the sets are connected with a solid line. The chart also shows the grand mean as well as the lower and upper control limits, LCL and UCL, respectively. The area between the grand mean and the control limits is divided into zones labeled A, B, and C, as they proceed toward the center of the control chart. These zones represent the 3σ, 2σ, and 1σ distance from the grand mean and are used to test for time-dependent nonrandom patterns. The right panel of the figure shows the data for the stimulated condition, compared against the basal control limits, indicating an out-of-control condition at the third and fourth time points.

A process is said to be in control if it exhibits only random variation, that is, all points (means, in this case) are within the control limits, and no nonrandom patterns are present. Our algorithm uses the zones labeled A, B, and C in the control chart to examine the time series for nonrandom patterns. At all m/z values at which the basal data remain in control over the course of the experiment, it is assumed that the variability in the signal output is appropriately represented by the control limits. These m/z values represent molecules in which metabolic cellular events are negligible, as measured by mass spectrometry, in the nonstimulated condition. An example of this can be seen in the left panel of Figure 12.3, in which the signal denoted by the means of the four measurements is seen to be in control. In this case, these limits are used for comparison with the observed means at this m/z ratio in the stimulated condition, and these results are collected for processing into the lipid arrays. This analysis includes parsing for time points beyond the control limits as well as searching for patterns that can be deduced from the control chart zones. In Figure 12.3, the third and fourth time points in the stimulated condition fail a

Multiplexed Lipid Arrays of Antiimmunoglobulin

275

nonrandom pattern test (two of three consecutive points in zone A or beyond; the fourth time point is beyond the lower control limit, as well) and are flagged by the algorithm as having decreased in the stimulated condition. The final portion of the computational analysis is the handling of exceptions to the discussion in the preceding text. Two possibilities require elucidation here. In the first case, the basal data behave in an out-of-control manner, that is, they contain some nonrandom time-related variation. When the basal condition exhibits out-ofcontrol variation, extending the control limits for comparison with the stimulated condition would be inappropriate. In this instance, a Welch-modified two-sample ttest is performed at each of the time points to determine if differences exist in the means between the two conditions at the given time. Thus, the algorithm performs an alternative statistical test at each time point for every m/z ratio in which the basal signal is found to be out of control. The second possibility involves peaks that appear in different frequencies within the basal and stimulated conditions. In this case, a binomial test is performed, with the null hypothesis that a peak has an equal chance of appearing in either of the two conditions, to determine if the observed difference in the number of occurrences in the two conditions is significant. Lipid Arrays At the conclusion of the analysis, the results are grouped into a comprehensive array containing the m/z values observed as peaks on the vertical and the time points on the horizontal axes. Lipid species that have been identified by CID MS/MS techniques as being present in the sample are assigned their corresponding m/z values. Each m/z and time point combination found to be increasing is scored with a one (1), whereas those decreasing are assigned a negative one (1). Statistically stable combinations are scored with a zero (0). These arrays are color coded to enhance readability and, in many cases, provide a striking display of cellular lipid changes in time after challenge with a biological agonist. An excerpt of a lipid array is shown in Figure 12.4. Lipid 38:4 PI 38:3 PI 38:2 PI 38:1 PI

m/z 885.6 886.6 887.6 888.5 889.5 890.5 891.5 892.5

T1 0 0 0 0 0 0 0 −1

T2 0 0 0 0 0 0 0 1

T3 −1 −1 −1 −1 −1 −1 0 −1

T4 −1 −1 −1 −1 −1 −1 −1 −1

T5 −1 −1 −1 −1 −1 −1 0 −1

FIGURE 12.4 An excerpt from a lipid array in the m/z range of 885.6 to 892.5. Data was collected from WEHI-231 cells challenged with 0.13 αM anti-IgM. Analysis using CID MS/MS determined that this area contained phosphatidylinositol lipids with 38 carbons in several double-bond configurations. The array shows these species decreasing over the time course after the stimulation, as indicated by the negative score.

276

Functional Lipidomics

The number of peak/time point combinations examined in the system can create a significant opportunity for false-positives. This is illustrated by considering that if 1000 different peaks are analyzed over five time points, 5000 chances for a falsepositive are created. If the alpha value is set at 0.05, one would anticipate 250 false indicators on a lipid array of this size occurring by chance alone. This effect can be countered by repeating the entire experiment multiple times and summing the cells from the resulting arrays. Thus, if the experiment is repeated five times, each cell in the summary array will have a score between 5 and 5. Scores occurring toward the extremes (5 and 5) indicate species that are fluctuating under stimulation with high statistical significance. Because random errors are unlikely to occur in the same position, after several repetitions the result is seen to converge to a stable map of lipid changes.

GLYCEROPHOSPHOLIPID CHANGES WEHI-231 CELLS

IN

BASAL

VS.

AIG-STIMULATED

Stimulation of the AfCS WEHI-231 B-cell receptor with 0.13 μM anti-IgM ligand resulted in robust changes in glycerophospholipid concentrations. Lipid arrays were constructed from 10 sets of samples. Each data set contained four exact replicates of paired samples that included a control (basal) and matched ligand-stimulated sample at each of the 5 time points: 1.5, 3, 6, 15, and 240 min. Thus, each array was constructed from 400 samples. The lipid species were identified using both the positive (array 1, supplemental material) and negative (array 2, supplemental material) ESI modes. Excerpts from the positive (a) and negative (b) mode arrays are shown in Figure 12.5. In array one, only a few changes in concentration were observed during the 1.5and 3-min time points. However, highly significant decreases were observed for many PC and PE species at the 6- and 15-min time points, with corresponding increases in several lyso-PC compounds. By the fourth hour, the cells had mostly returned to their prestimulated states. A list of lipids having significant or highly significant changes is summarized in Table 12.3. The temporal trend in array two was shifted towards the later time points. Little movement was observed during the 1.5-, 3-, or 6-min experiments. But clusters of phosphatidylinositols and PSs were observed to decrease highly significantly at the 15-min time point. Corresponding increases in lyso-PI, lyso-PS, and glycerophosphatidic acids were also recorded. A summary of observed changes for the entire array is shown in Table 12.4.

DISCUSSION Antiimmunoglobulin (0.13 μM) stimulation of WEHI-231 cells resulted in a unique pattern of increasing and decreasing levels of glycerophospholipids. A wide variety of phosphatidylinositols, PGs, and PCs decreased in concentration during the 6- and 15-min time points, with corresponding increases in lysophospholipid levels. By the fourth hour, the GPL levels had essentially returned to their prestimulated states.

Multiplexed Lipid Arrays of Antiimmunoglobulin

277

A Lipid 30:3 PC 16:1 SM 30:2 PC 16:0 SM

m/z 699.5 700.6 701.5 702.6 703.6

1.5 min 0 −3 3 0 −5

3 min 0 −1 −1 0 −2

6 min 0 −3 4 2 −6

15 min −2 −5 5 2 −9

240 min 2 −1 2 −1 −5

m/z 854.5 855.5 856.5 857.5 858.5 859.5 860.5 861.5 862.5 863.5 864.5 865.6

1.5 min −2 −1 −3 −1 −2 −2 −2 −2 −3 0 −3 −3

3 min −1 −4 −1 −3 −4 −4 −1 −3 −4 −2 −4 −7

6 min −2 −3 −4 −3 −4 −3 −3 −5 −2 −2 −4 −5

15 min −3 −7 −5 −5 −7 −7 −5 −6 −6 −3 −6 −9

240 min −1 −5 −3 2 1 −4 −4 −5 −4 0 −1 −1

B Lipid 36:5 PI 36:4 PI 36:3 PI 36:2 PI 36:1 PI 36:0 PI

Significance key Signal decreasing −10

−6 −5 −4 Higher significance

Signal increasing 0

4

5 6 Higher significance

10

FIGURE 12.5 Excerpts from positive (A) and negative (B) mode lipid arrays. The first column contains lipid species identified by ESI-MS/MS. The second column indicates the mass-to-charge ratio (m/z) of the observed compounds. The remaining five columns are for the five time points (1.5, 3, 6, 15, and 240 min). The values in these columns represent the significance score, which is the sum of that cell for the ten individual experiments, with positive numbers representing increasing signal and negative values indicating a decreasing signal. Therefore, an array cell containing the number −8 is interpreted to mean the indicated species was observed to decrease in eight of the ten trials and remain stable in the other two, or that the species decreased in nine of the experiments and increased in one at that time point. These scores are color coded by signal frequency with deep blue or red, indicating an absolute score of 6 or more from a possible 10 (shown highly significant by computer simulation). Lighter shades of red and blue indicate significant stimulations (−5 or 5). Greencolored cells, representing a −4 to 4 significance score, indicate statistical stability between basal and stimulated conditions.

Due to detection limitations using triple quadrupole MS, DAGs were not routinely analyzed in this study. Under our current experimental conditions, it is not possible to scan simultaneously for DAGs and the remaining lipid classes. In the future, samples will be split, and changes in DAG levels will be monitored using

Highly Significant

POS

Increasing

Decreasing

Time = 1.5 min Time = 3 min

34:2 PE 40:3 PCp 36:0 PCp 32:1 PCp 40:4 PCe 36:1 PCe 32:2 PCe

36:1 PCp 36:2 PCe 40:2 PCp 40:3 PCe 40:3 PCp 40:4 PCe 38:4 PC

36:2 PE 34:1 PCp 34:2 PCe 28:1 PC 32:1 PEe 36:1 PE 34:0 PCp 34:1 PCe 36:4 PC

38:6 PC

34:2 PC

20:0 LPC 20:1 LPC

40:6 PC

36:1 PC

36:4 PC

34:3 PC

34:0 PCe 38:0 PCe

34:1 PE

30:1 PC 36:2 PC 34:0 PEp 40:1 PEp 34:1 PEe 40:2 PEe

32:2 PC 36:0 PCe 36:2 PE 36:1 PEp 38:0 PE 34:1 PCp 36:2 PEe 40:6 PEp 34:2 PCe

36:1 PE 40:4 PE 38:1 PCp 34:0 PCp 38:3 PCp 38:2 PCe 34:1 PCe 38:4 PCe

34:0 PC 40:5 PCp 36:1 PCp 38:0 PEe 40:6 PCe 36:2 PCe

Time = 15 min 16:0 SM 40:2 PCp 28:1 PC 40:3 PCe 32:1 PEe

Time = 6 min 16:0 SM 36:0 PCp 36:1 PCe

TABLE 12.3 Summary of Glycerophospholipid Changes following AIG Stimulation of WEHI-231 Cells (Positive Mode) Time = 240 min

278 Functional Lipidomics

Significant

Increasing

Decreasing

20:1 LPC

16:0 SM

32:1 PE 30:0 PCp 30:1 PCe

40:2 PCp 36:1 PE 40:3 34:0 PCp 34:1 PCe PCe

36:2 PC 40:1 PEp 40:2 PEe

36:3 PC

34:2 PC

30:1 PC 34:0 PEp 34:1 PEe

40:0 PC

38:1 PC

38:6 PE 36:5 PC

16:1 SM

40:4 PC

24:0 SM

40:0 PC

44:0 PCe

44:1 PCp 42:3 PCp 42:0 PCp 44:2 42:1 42:4 PCe PCe PCp

30:3 PC

20:0 LPC

44:1 PCp 44:2 PCe

16:0 SM

40:6 PC

42:1 PCp 42:2 PCe

36:2 PC 40:1 PCp 40:2 PEe

38:2 PCp 32:1 PE 38:3 PE 30:0 36:2 PCp 38:3 PCp 36:3 PCe PCe 30:1 PCe 40:0 PEe 32:1 PC 40:4 PE 38:3 PCp 36:0 PC 36:0 PEp 38:4 PCe 38:6 36:1 PCp PEe 38:5 PC 30:0 PC 36:3 PC

38:1 PCp 38:2 PCe

38:3 PE 36:2 PCp 36:3 PCe 32:0 PE 30:0 PCe

36:6 PC 36:2 PE 34:1 PCp 40:5 PEp 34:2 PCe 40:6 PEe

40:6 PCp 32:0 PCe 34:1 PC 32:2 PC 36:1 PEp 38:0 PC 34:0 PE 34:2 PS 36:2 PEe

32:1 PE 30:0 PCp 30:1 PCe

Multiplexed Lipid Arrays of Antiimmunoglobulin 279

Significant

Highly Significant

Decreasing

Increasing

Decreasing

NEG

40:6 PS

Time = 1.5 min Time = 3 min 36:0 PI

34:2 PI 36:0 PI 36:2 PI 38:4 PI

Time = 6 min

36:2 PI 36:5 PI

38:2 PS 40:6 PS

38:4 PI

34:2 PI

32:0 PS

36:4 PI

36:4 PG 16:0 LPA

40:7 PS

40:6 PS

40:6 PI

36:4 PS 38:4 PS

36:0 36:2 36:3 36:5 38:6 40:6 20:1 PI PI PI PI PI PI LPS

Time = 240 min 38:6 PI

Time = 15 min 34:2 PI 36:3 PS

TABLE 12.4 Summary of Glycerophospholipid Changes following AIG Stimulation of WEHI-231 Cells (Negative Mode)

280 Functional Lipidomics

Increasing

40:6 PS

22:6 LPS

36:5 PE

34:1 PG 14:0 LPE 16:1 LPE 16:0 LPA 18:0p LPA

38:2 PE

32:1 PG

16:0 LPE

20:4 LPI 34:1 PA

30:0 PA

Multiplexed Lipid Arrays of Antiimmunoglobulin 281

282

Functional Lipidomics

an alternative procedure. Preliminary analysis measuring polyphosphoinositide levels is already underway and will be described in a subsequent report. The current methodology appears to be effective but cannot be run in parallel with other phospholipids. Detection of these species will almost certainly require multiple extractions and or separation by HPLC. Mapping comprehensive lipid changes in time is thought to have many positive applications in cellular biology and cellular signaling. It is believed that data of this type will prove useful in hierarchical clustering schemes as another method in differentiating receptor-mediated cellular events involving lipid second messengers as well as membrane compositional remodeling. It is further believed that these arrays form a fingerprint that can be used to identify specific cellular responses and as such may prove useful in diagnostic profiling and elucidating lipid product–substrate relationships.

METHODS AND PROTOCOLS CELL EXTRACTION

AND

RECONSTITUTION

Phospholipids were extracted using a modified Bligh and Dyer procedure (25). Pellets containing 3 × 106 cells were extracted with 800 μl of 0.1-N HCl: MeOH (1:1) and 400 μl of CHCl3. The samples were vortexed (1 min) and centrifuged (5 min, 18,000 g). The lower phase was then isolated and evaporated (Labconco CentriVap Concentrator, Kansas City, MO), followed by reconstitution with 80 μl MeOH: CHCl3 (9:1). Prior to analysis, 1 μl of NH4OH was added to each sample to ensure protonation. Lipid standards were obtained from Avanti Polar Lipids (Alabaster, AL).

MASS SPECTROMETRY ANALYSIS

OF

PHOSPHOLIPID CELL EXTRACTS

Mass spectral analysis was performed on a Finnigan TSQ Quantum triple quadrupole mass spectrometer (ThermoFinnigan, San Jose, CA) equipped with a Harvard Apparatus syringe pump and an electrospray source. Samples were analyzed at an infusion rate of 10 μl/min in both positive and negative modes over the range of m/z 400 to 1200. Instrument parameters were optimized with 1, 2-dioctanoyl-sn-glycero-3phosphoethanolamine (16:0 PE). Data were collected with the Xcalibur software package (ThermoFinnigan) and analyzed by a software program developed in our research group.

ACKNOWLEDGMENT This chapter originally appeared as an online AfCS Research Report (see www.signaling-gateway.org/reports). It is printed in this volume with the consent of the AfCS Steering Committee.1 Please see policy on authorship available via the online version of this report. 1

To whom scientific correspondence should be addressed.

Multiplexed Lipid Arrays of Antiimmunoglobulin

283

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

Gilman, A.G., Simon, M.I., Bourne, H.R. et al. (2002). Nature 420(6916), 703–706. Sambrano, G.R., Chandy, G., Choi, S. et al. (2002). Nature 420(6916), 708–710. Hokin, L.E. (1985). Annu. Rev. Biochem. 54, 205–235. Hla, T., Lee, M.J., Ancellin, N., Paik, J.H., and Kluk, M.J. (2001). Science 294(5548), 1875–1878. Pagès, C., Simon, M.-F., Valet, P., and Saulnier-Blache, J.S. (2001). Prostaglandins Other Lipid Mediat. 64(1-4), 1–10. Goetzl, E.J. (2001). Prostaglandins Other Lipid Mediat. 64(1-4), 11–20. Fukushima, N. and Chun, J. (2001). Prostaglandins Other Lipid Mediat. 64(1-4), 21–32. Clay, K.L., Wahlin, L., and Murphy, R.C. (1983). Biomed. Mass Spectrom. 10, 489–494. Han, X. and Gross, R.W. (1994). Proc. Natl. Acad. Sci. U.S.A. 91(22), 10635–10639. Kim, H.Y., Wang, T.C., and Ma, Y.C. (1994). Anal. Chem. 66(22), 3977–3982. Kerwin, J.L., Tuininga, A.R., and Ericsson, L.H. (1994). J. Lipid Res. 35(6), 1102–1114. Han, X. and Gross, R.W. (1995). J. Am. Soc. Mass Spectrom. 6, 1202–1210. Han, X. and Gross, R.W. (1996). J. Am. Chem. Soc. 118, 451–457. Brügger, B., Erben, G., Sandhoff, R., Wieland, F.T., and Lehmann, W.D. (1997). Proc. Natl. Acad. Sci. U.S.A. 94(6), 2339–2344. Fridriksson, E.K., Shipkova, P.K., Sheets, E.D., Holowka, D., Baird, B., and McLafferty, F.W. (1999). Biochemistry 38(25), 8056–8063. vanova, P.T., Cerda, B.A., Horn, D.M., Cohen, J.S., McLafferty, F.W., and Brown, H.A. (2001). Proc. Natl. Acad. Sci. U.S.A. 98(13), 7152–7157. Khaselev, N. and Murphy, R.C. (2000). J. Am. Soc. Mass Spectrom. 11(4), 283–291. Murphy, R.C., Fiedler, J., and Hevko, J. (2001). Chem. Rev. 101(2), 479–526. Larsen, A., Uran, S., Jacobsen, P.B., and Skotland, T. (2001). Rapid Commun. Mass Spectrom. 15(24), 2393–2398. Murphy, R.C. (2002). Mass Spectrometry of Phospholipids: Tables of Molecular and Product Ions. Denver, Colo: Illuminati Press. Christie, W.W. (2003). Lipid Analysis: Isolation, Separation, Identification, and Structural Analysis of Lipids. Bridgwater, England: Oily Press. Hsu, F.F. and Turk, J. (2003). J. Am. Soc. Mass Spectrom. 14(4), 352–363. Pulfer, M. and Murphy, R.C. (2003). Mass Spectrom. Rev. 22(5), 332–364. Venables, W.N. and Ripley, B.D. (1994). Modern Applied Statistics with S-Plus. New York: Springer-Verlag. Bligh, E.G. and Dyer, W.J. (1959). Can. J. Biochem. Physiol. 37, 911–917.

13

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes: Shotgun Global Cellular Lipidome Analyses by Electrospray Ionization Mass Spectrometry Using Intrasource Separation Xianlin Han and Richard W. Gross

CONTENTS Abstract ..................................................................................................................286 Introduction............................................................................................................286 Principles of ESI Intrasource Separation Techniques...........................................289 Applications of Shotgun Lipidomics and Implication in Functional Lipidomics....................................................................................................293 Identification of Specific Alterations of Sulfatide in Alzheimer’s Disease .............................................................................................293 Identification of Diabetes-Induced Changes in Specific Lipid Molecular Species ............................................................................298 Conclusion .............................................................................................................300 Acknowledgment ...................................................................................................300 References..............................................................................................................301

285

286

Functional Lipidomics

ABSTRACT The lipidome is the precise complement of chemically distinct covalent lipid entities in a cell type or intact organ. Lipidomics refers to research investigating the biological function, significance, and sequelae of alterations in lipids and protein constituents mediating lipid metabolism, trafficking, or biological function in cells. Lipidomics has been greatly facilitated by recent advances in, and novel applications of, electrospray ionization mass spectrometry (ESI/MS). By employing ESI intrasource separation techniques that we have developed, individual molecular species of most major and many minor lipid classes can be fingerprinted and quantitated directly from biological lipid extracts. In this article, we briefly describe the principles of ESI intrasource separation that we have exploited for lipid analysis based on the inherent electrical propensities of different lipid classes. This strategy allows quantification of lipid molecular species directly from organic extracts of biological samples without the need for chromatographic purification. Examples of the application of this technology to lipid metabolism studies in two prevalent disease states (i.e., Alzheimer’s disease and diabetes) are given. Through appropriate utilization of ESI intrasource separation, the role of lipid alterations underlying disease states and the biochemical mechanisms through which lipids mediate cellular function in health and disease (i.e., functional lipidomics) can be further clarified. It is our hope that this knowledge will help us modify the deleterious consequences of lipidassociated diseases, such as diabetes, atherosclerosis, and obesity, afflicting industrialized countries in epidemic proportions.

INTRODUCTION The large majority of health problems in industrialized societies are due to the dysfunctional and maladaptive alterations in lipid metabolic flux precipitated by an inappropriate content of fat in consumed food and a sedentary lifestyle [1–3]. These changes in substrate supply and utilization result in a dramatic redistribution in the bulk flow of lipids into and out of critical cells and subcellular storage compartments in mammalian cells. Through the combined effects of excessive fat intake, increased caloric consumption, and lack of exercise, society has unwittingly exposed itself to the largest public health epidemic in human history [4]. Moreover, obesity in humans is almost always accompanied by a number of chemically related disorders including diabetes, atherosclerosis, cardiomyopathy, and hypertension, referred to as the metabolic syndrome [5]. Collectively, these lipid-related diseases are taking a huge toll in human pain, suffering, and productivity losses in modern society. Although it is too early to definitively classify Alzheimer’s disease (AD) as a lipid-associated disease, the connection between lipids and AD has recently received substantial attention from many investigators (see Reference 6 to Reference 14 for recent reviews). The mechanisms underlying AD pathogenesis are still not entirely clear. However, many studies suggest that amyloid beta-protein (Aβ) accumulation and conformational changes likely play a critical role in the development of AD pathology [15]. At present, the only known genetic risk factor for late-onset AD (including both familial and sporadic) is the ε4 allele of apolipoprotein E (apoE)

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

287

[16–19], which is an apolipoprotein that plays an essential role in lipid metabolism and trafficking [20,21]. Lipids fulfill multiple functions in mammalian physiology through their diverse roles as: (1) chemical barriers to diffusion of ions, metabolites, and macromolecules; (2) sources of energy, by harvesting the Gibbs free energy inherent in C–C bonds; (3) provision of an appropriate 2-D matrix for lateral diffusion, leading to productive interactions in supramolecular assemblies that utilize the membrane bilayer as an organizing surface; and (4) endogenous storage reservoirs for the production of lipid second messengers of signal transduction, that communicate information between subcellular compartments (intracellular communication) or between cells (intercellular communication). Remarkably, each of these membrane-mediated functions is subject to alterations induced by the dysfunctional lipid metabolic fluxes present in obese patients and is dramatically amplified in the diabetic state. The technology that accurately measures detailed molecular species alterations, changes in the bulk lipid content, or even more important, alterations in lipid flux between cells or flux between intracellular compartments, which mediate the transition between health and disease, is of vital importance to the American public. In most mammalian cells, phospholipids (a phosphodiester linked to the sn-3 position of glycerides) account for approximately 60 mol% of total lipids; glycolipids/sphingolipids (carbohydrates linked to the sn-3 position of diglycerides or the position 1 of sphingosine moiety) are approximately 10 mol% of total lipids; nonpolar lipids (including triacylglycerols [TAGs] and cholesterol) range from 0.1 to 40 mol% depending on cell type and subcellular compartment. Metabolites (such as nonesterified fatty acids [NEFA], lysolipids, diacylglycerols [DAGs], ceramides, acyl carnitines, and acyl CoA) typically represent less than 5 mol% of total cellular lipids but can accumulate during pathologic conditions and contribute to deleterious pathophysiologic sequelae. The complexity of cellular lipids is not only due to the presence of many different lipid classes categorized by their polar head groups but also due to the presence of different covalent linkages of aliphatic chains to the backbone in many lipids (i.e., subclasses), as well as the presence of different lengths of aliphatic constituents that also differ in their number and locations of double bonds (i.e., molecular species). Therefore, thousands of lipid molecules are present in the lipidome, and the number of mass spectrometric observables increases as the sensitivity of the analytical technology increases. Thus, investigators in lipidomics are currently developing the technology to quantify the thousands of chemically distinct constituents present in the cellular lipidome, determine their subcellular organization (subcellular membrane compartments and microdomains), identify kinetic fluxes into and out of each cellular compartment, and delineate lipid–lipid and lipid–protein interactions, which collectively determine membrane conformational space and dynamics. The first step in global lipidomics is to measure the amount of each distinct chemical entity present in a cell’s lipidome and identify those alterations in chemical constituents that precipitate, or are associated with, phenotypic alterations. As one example, it is easy to imagine that increases in dietary fat consumption alter the lipid molecular species’ composition and physical properties of the membrane bilayer, thereby altering the rate-limiting step of lipid second messenger generation. Such perturbations thus lead

288

Functional Lipidomics

to new steady-state conditions conducive to the propagation and exacerbation of disease. Accordingly, the field requires integrated techniques to determine the chemical identities of all lipids (structural, signaling, energy reserve, and those that organize supramolecular assemblies) to understand the complex interwoven interactions underlying the dysfunctional lipid metabolism and altered metabolic flux present in lipid-related disease states. Previously, we exploited the combined power of high-performance liquid chromatography (HPLC) and mass spectrometry of intact phospholipids to identify plasmalogens as the major phospholipid constituents of the electrically active membranes of canine cardiac myocytes, sarcolemma [22], as well as sarcoplasmic reticulum and mitochondrial membranes [23]. Detailed molecular species analysis by fast atom bombardment mass spectrometry demonstrated that plasmalogens were highly enriched in arachidonic acid at the sn-2 position in these membranes. Because the rate-determining step in the production of biologically active eicosanoids is the release of arachidonic acid from their endogenous phospholipid storage pools, these results identified plasmalogen molecular species as the likely targets of the phospholipases activated during signal transduction in mammalian cells. Thus, these initial mass spectrometry studies redirected thinking on the nature of lipids participating in cellular activation (arachidonic-acid-containing plasmalogens) and the types of phospholipases mediating the release of arachidonic acid in mammalian cells (i.e., intracellular plasmalogen-hydrolyzing phospholipases). Based on these mass spectrometric findings, we focused on the development of methods for plasmalogen synthesis, identified new phospholipase families that hydrolyzed plasmalogen substrates, characterized the molecular dynamics of plasmalogens by both nuclear magnetic resonance and electron paramagnetic resonance spectroscopy, and determined the first solution state structure of phospholipids by truncated driven nuclear Overhauser enhancement analyses [24–30]. Chromatography is a well-established and useful method to separate, concentrate, and enrich many lipid molecular species. However, several issues need to be recognized when significant on-column dwell times are employed and resolution of different species is incomplete. First, HPLC is accompanied by intrapreparative silica-catalyzed hydrolysis of vinyl ether linkages. Second, silica-based surfaces serve as solid-phase supports to catalyze α-hydroxy migration (e.g., 1,2-DAG to 1,3-DAG; or 2-acyl lysolipids to 1-acyl lysolipids) [31]. Third, appropriate methods for quantitation must be employed because internal standards and individual molecular species elute from the HPLC matrix with their own distinct retention times and peak shapes (e.g., differential peak trailing from heterogeneous interactions with the stationary phase) [32–34]. Because hundreds of individual molecular species are present in each chromatographic peak from a given phospholipid class after straightphase HPLC, each possessing different retention times, comparisons with internal standards require consideration of these differential elution times. When reversedphase HPLC is employed for molecular species analysis after lipid classes have been separated with straight-phase or ion exchange HPLC, quantitation of individual molecular species requires special care to avoid concentration-mediated nonlinearities in response factors [34,35]). Thus, although HPLC separation can enrich lowabundance molecular species and eliminate the interactions of different lipid molec-

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

289

ular species, molecular interactions from the same type of molecular species are amplified by HPLC concentration potentially resulting, at least in some cases, in nonlinearity of response factors. This phenomenon has been demonstrated by Delong and colleagues [34] and is described as the chromatographic effect. We have called this lipid–lipid interaction [36,37], and some researchers referred to it as ion suppression. Whatever name utilized, the vast differences in concentration of analytes that elute from HPLC columns must be carefully considered. Due to the pioneering work of Fenn and colleagues, new horizons in lipidomics became feasible [38]. In the late fall of 1991, we, along with others, began ionizing phospholipids from sprayed droplets in a 3 to 5 kV electric field [39–43]. Because of the complex nature of chromatographic interactions stated in the last paragraph, direct-infusion techniques have been developed and employed by us and many of our colleagues [36,39,44–54]. These methods are now commonly referred to as shotgun lipidomics [37]. By using the mass spectrometer ion source as a separation device, the mechanical aspects of lipidome analysis are dramatically simplified, decreasing cost and simultaneously yielding analytic superiority (in most cases) by removing intrapreparative elution artifacts and propagated errors. We wish to specifically point out that although expensive and time-consuming chromatographic separations can be avoided through appropriate use of shotgun lipidomics by analysis directly from extracts of biological samples, chromatography or other enrichment approaches are necessary for successful penetration into the extremely low-abundance regime of the lipidome because of sensitivity considerations alone. In this article, the principles of the intrasource separation technique are briefly discussed. Two specific applications of this technique in AD and diabetes through global lipidome analyses are summarized as examples of the power of shotgun lipidomics. Although lipidomics is still in an early stage of development in comparison to genomics and proteomics, the technology available in lipidomics is already adequate to provide many new insights into the biological mechanisms of multiple disease states. The development of lipidomics will lead to a new level of understanding in lipid-related diseases through the use of mass spectrometry to gain fundamental insights into the biological functions of the cellular lipidome through analysis of disease-related alterations in the lipidome.

PRINCIPLES OF ESI INTRASOURCE SEPARATION TECHNIQUES Advances in ionization technology, in large part fueled by last year’s Nobel laureates John Fenn and Koichi Tanaka [55,56], have now made it possible to identify, quantitate, and profile thousands of lipids directly from crude extracts of biological samples. In one of our early studies [39], we recognized that the presence of a large dipole in phospholipids in conjunction with field-induced charge separation in an electric field could be used to ionize many species of cellular lipids. Through this approach, low-energy ionization of intact pseudomolecular ions was accomplished. By appropriate matrixed adjustments in extract solutions (solution pH or ionic strength), the selective ionization of discrete lipid classes was achieved, rendering

290

Functional Lipidomics

the newly found ability to observe hundreds of molecular species directly from chloroform extracts of biological samples [36,37,39]. Thus, this approach represented a de facto separation method executed within the spectrometer source and has been referred to as intrasource separation [36]. The principles underlying intrasource separation of lipid classes have been described in detail in a recent review [37] and are only briefly discussed here. For any analyte–ion pair, an intrinsically cationic moiety can be readily analyzed by ESI/MS in the positive-ion mode, whereas in an anionic analyte–ion pair, the negative-ion mode may be used. Molecules that do not possess any intrinsic ionizable site can also be ionized in an ESI ion source through formation of adduct ions as long as a sufficient dipole potential is present in a molecule to interact with either a small anion or cation. In other words, the formation of a molecular ion of a compound in an ESI ion source is predictable, and the ionization mechanisms and efficiency of a compound depend largely on its electrical propensities in the infusion solution. Following this principle, cellular lipids can be categorized into four groups based on their electrical propensities and proton affinities. The first group of cellular lipids consists of anionic lipids that contain one or more net negative charge(s) at physiological (neutral) pH, including the classes of cardiolipin, phosphatidylglycerol (PtdGro), phosphatidylinositol (PtdIns), phosphatidylserine (PtdSer), phosphatidic acid (PtdH), sulfatide, and so on. The second group of cellular lipids consists of lipids containing an inducible negative charge under alkaline conditions. Lipid classes in this group commonly include ethanolamine-containing phospholipids (e.g., ethanolamine glycerophospholipid [PE] and lysoPE), NEFA, etc. The third group of cellular lipid classes consists of those that are electrically neutral but contain a large dipole (i.e., zwitterions). Lipid classes in this group include choline-containing lipids (e.g., choline glycerophospholipid [PC], lysoPC, and sphingomyelin [SM]) and glycolipids (e.g., cerebroside). Some classes of nonpolar lipids (e.g., cholesterol and some of its derivatives, TAG, and DAG) possess smaller dipole moments and have to be individually considered (see discussion later). The final group of lipid classes in this categorization scheme consists of the specialized lipid metabolites, such as ceramide, long-chain acylCoAs, and long-chain acylcarnitines, each of which has had a specialized analytical methodology developed specifically, as previously described [57–59]. Based on the electrical propensity of each lipid class, we developed a strategy for lipidome analyses of biological samples without the need for chromatography by judicious matrixed front-end sample preparations [36,37,60]. To date, hundreds of individual lipid molecular species of most of the major lipid classes and multiple minor metabolite groups have been fingerprinted and quantitated. The first step in this strategy is to analyze the chloroform extract of appropriately prepared and extracted biological sample directly at neutral pH by ESI/MS in negative-ion mode, which selects for anionic lipid species, as discussed in the preceding text. Thus, the acquired mass spectrum displays a fingerprint of anionic (e.g., PtdIns, PtdSer, and sulfatide [ST]) lipid molecular ions (Figure 13.1a). These molecular ions can be further characterized through the analyses of head groups and acyl moieties as well as isobaric molecular species by either product-ion, neutral loss, or precursor-ion

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

693.5 (I.S.)

Relative intensity (%)

100

291

N24:1 ST 888.7

50

18:0–22:6 PS 834.7

18:0–20:4 PI 885.7

2OH N24:1 ST 904.8

0 650

700

750

800 m/z

850

900

950

(a) PtdEtn PtdEtn PlsEtn 790.6 766.6 774.6

Relative intensity (%)

100

PlsEtn PlsEtn 746.5 PlsEtn 750.6 726.5

PtdEtn (I.S.) 662.5

50

PlsEtn 700.5 0 650

700

750

800

m/z (b) 766.5 (PC)

Relative intensity (%)

100 I.S. for PC & SM 680.5 50 SM 709.5

GalC 816.6 PC 734.5

I.S. PC GalC 794.6

SM 753.5

GalC 832.7

0 600

650

700

750 m/z

800

850

900

(c)

FIGURE 13.1 Representative ESI mass spectra of a lipid extract of parietal gray matter from a cognitively normal subject. Brain samples were obtained from the brain bank of the Washington University ADRC Neuropathology/Tissue Resource Core, and lipids were extracted by a modified method of Bligh–Dyer [87]. ESI mass spectra of the lipid extract of human parietal gray matter were acquired in the negative-ion mode with LiOH absent (Panel a) and present (Panel b) and in the positive-ion mode with LiOH present (Panel c). All major individual molecular species as indicated were identified using tandem mass spectrometry. (Han, unpublished data).

292

Functional Lipidomics

tandem mass spectrometry [37,60]. By comparing the intensity of each identified ion peak with an anionic lipid internal standard (or standards) and appropriate correction for 13C isotopomer distributions [37,61], quantitation of individual molecular ions can be achieved. Next, a small amount of LiOH is added, and PE, LysoPE, and NEFA molecular species are rendered anionic in solution by deprotonation of either the primary amine or carboxylic acid. Negative-ion ESI results in abundant pseudomolecular ion peaks corresponding to individual ethanolamine-containing molecular species in the mass range between (typically) m/z 600 and 900 for PE (Figure 13.1b), LysoPE between m/z 400 and 600, and NEFA between m/z 200 and 400. These individual molecular species can be quantified by comparison of their ion peak intensities with their selected internal standards and their molecular identities determined by tandem mass spectrometry. As discussed previously [37,60], the degree of overlap of anionic phospholipids with PE molecular species is quite small, and the abundance of anionic phospholipids is low in comparison to the mass of PE species in most cellular lipid extracts. Therefore, the effects of the presence of anionic phospholipid molecular species on the quantitation of PE species can only introduce errors up to 5% in extreme cases. Finally, the dilute chloroform extract at alkaline pH is analyzed in the positiveion mode. Under the experimental conditions employed, molecular classes that contain an endogenous negative charge (e.g., anionic lipids and PE species) are largely prevented from forming positively charged ions during the ESI process. Molecular species of choline-containing phospholipids (i.e., PC, lysoPC, and SM), glycolipids, etc., are readily ionized as lithium adducts under these conditions (Figure 13.1c). These pseudomolecular ions can be further characterized through tandem mass spectrometry by analysis of head groups and acyl moieties as well as isobaric molecular species [37,60]. By comparing the intensity of each identified ion peak with an internal standard (or standards) and appropriately correcting for 13C isotopomer distributions [37,61], quantitation of individual molecular ions can be obtained. TAGs, and even cholesterol esters, can be readily ionized as lithium adducts under identical conditions [37,60–62]. However, quantitation of individual molecular species of these classes having small dipoles is not as straightforward as that for lipid classes possessing large dipoles. An algorithm based on the ionization efficiencies of each individual molecular species of these classes, such as TAG, was developed to correct the effects of acyl chain length and double bond numbers on quantitation [61]. Careful sample preparation is an essential step in the successful performance of global lipid analysis directly from lipid extracts of biological samples [37]. Correction for 13C isotope distributions and utilization of dilute infusion solutions are also important to ensure accurate analyses of lipid molecular species [37]. Under these conditions (dilute infusion solution and low ionic strength), the ionization efficiencies of each individual molecular species within a polar lipid class are essentially identical within experimental error, and the effects of acyl chain physical properties on the quantitative analysis of molecular species possessing large dipoles are negligible. It should be pointed out that the upper limit of concentration, under which the linearity for quantification is still held, is dependent on the solvent utilized in the infusion solution and

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

293

the ion source employed. For example, Koivusalo and colleagues [35] independently demonstrated that ion peak intensities of equimolar mixtures of 14 PC molecular species, including 6 diunsaturated PC species, were essentially identical within experimental error up to 0.5 pmol/μl of each species (i.e., up to 7 pmol/μl of total phospholipids) dissolved in CHCl3/MeOH (1:2; v/v) (Figure 3 in the original reference), if corrections for differential 13C isotopomer abundance among classes are considered. In our work, as an extreme case of differences in analyte physical properties, we demonstrated that, as the relative molar ratio of 16:0-16:0 phosphadylcholine (PtdCho) and 16:0 lysoPtdCho varied, the peak intensity ratio of these two species precisely correlated with the molar ratio of the species in the infusion solutions, as previously reported [39]. Very recently, we have reexamined this relationship in mixtures of 17:0 lysoPtdCho, 14:1-14:1 PtdCho, 16:0-18:1 PtdCho, 16:0-20:4 PtdCho, 15:0-15:0 phosphatidylethanolamine (PtdEtn), and 15:0-15:0 PtdGro using a ThermoFinnigan TSQ Quantum Ultra mass spectrometer. Again, positive-ion ESI/MS demonstrated almost equally intense lithiated molecular ion peaks of lysoPtdCho and PtdChos after correction for 13C isotopomer distributions for an equimolar mixture of aforementioned species without the interference from the presence of PtdEtn and PtdGro species (left panel in Figure 13.2). As the relative molar ratio of PtdCho to 17:0 lysoPtdCho varied, the peak intensity ratios of PtdCho to 17:0 lysoPtdCho precisely correlated with the molar ratio of the species in the infusion solutions (right panel in Figure 13.2). The linear correlation of response factors between anionic molecular species in negative-ion mode was also previously determined [44]. Additionally, we point out that in-source fragmentation, due to either the inherent ion source design or the ionization conditions used, may deleteriously affect quantitative analysis of molecular species if appropriate conditions are not maintained. Finally, we stress that this technique, at its current stage, is only useful if the accuracy of quantitation within 5% is sufficient for the experimental question under consideration. Furthermore, very minor components, particularly those possessing overlap with M + 2 peaks, can only be estimated by primary ion data alone. However, quantitative analyses of these minor molecular species can be improved and refined by employing our newly developed 2-D mass spectrometric technique and multidimensional analyses derived directly therefrom [60].

APPLICATIONS OF SHOTGUN LIPIDOMICS AND IMPLICATION IN FUNCTIONAL LIPIDOMICS IDENTIFICATION OF SPECIFIC ALTERATIONS ALZHEIMER’S DISEASE

OF

SULFATIDE

IN

Sulfatides are a class of specialized glycosphingolipids (i.e., sulfated galactocerebrosides), which differ only in the nature of the aliphatic chain attached to the amino group of the sphingosine backbone. Sulfatides are involved in multiple biological processes, such as cell growth, protein trafficking, signal transduction, cell–cell recognition, neuronal plasticity, and cell morphogenesis (see Reference 63 to Reference 66 for reviews). Sulfatide accumulation is associated with multiple human diseases such as metachromatic leukodystrophy [67]. A mouse model deficient in sulfatide content shows multiple abnormalities, including abnormal axonal function,

Functional Lipidomics

100

17:0 IysoPtdCho 516.4

14:1–14:1 16:0–18:1 16:0–20:4 PtdCho PtdCho PtdCho 680.6 766.6 788.6

Relative intensity (%)

80

60

40 20

0 500

600

700 m/z

800

Log (PtdCho/LysoPtdCho peak intensity ratio)

294

1.0 14:1–14:1 PtdCho 16:0–18:1 PtdCho

0.5

16:0–20:4 PtdCho

0

–0.5

–1.0 –1.0

–1.5

0

0.5

1.0

Log (PtdCho/LysoPtdCho molar ratio)

FIGURE 13.2 Relationship between molar ratio and ion peak intensity ratio of PtdCho species and lysoPtdCho in the presence of PtdEtn and PtdGro. A positive-ion ESI mass spectrum of an equimolar mixture (1 pmol/μl each) of 14:1-14:1 phosphatidylcholine (PtdCho), 16:0-18:1 PtdCho, 16:0-20:4 PtdCho, 17:0 lysophosphatidylcholine (lysoPtdCho), 15:015:0 phosphatidylethanolamine (PtdEtn), and 15:0-15:0 phosphatidylglycerol (PtdGro), in the presence of a small amount of LiOH, shows four lithiated molecular ions with essentially equal intensities after correction for 13C isotopomer distributions relative to 17:0 lysoPtdCho (horizontal lines) (left panel). The correlation between molar ratio and ion peak intensity ratio of each individual PtdCho molecular species and 17:0 lysoPtdCho was linear (slope = 1.00; γ2 = 0.998). Molecular ions corresponding to 15:0-15:0 PtdEtn and 15:0-15:0 PtdGro are separated in the ion source, as described previously [37], and do not show in the spectrum.

dysmyelinosis, and loss of axonal conduction velocity [68–71]. These mice generally die by 3 months of age. By employing functional lipidomics [36,37], we have identified distinct lipid profiles in gray matter and white matter in human brain (Figure 13.3) that served to distinguish these regions [50]. In early work, we [50,72] determined alterations in the brain lipid profiles of subjects with AD at the earliest clinically recognizable stage (i.e., very mild AD) [73]. We found dramatic depletion of sulfatide mass (up to 90%) (Figure 13.4) in gray matter of all examined cerebral regions, including the frontal, parietal, and temporal cortices, from subjects with AD at its earliest clinically recognizable stage [72]. A lower content in sulfatide mass was present in the gray matter of cerebellum than that in cerebral regions. The change in sulfatide mass found in the cerebellum was relatively smaller than changes in cerebral gray matter from subjects with very mild AD. The depletion in mass levels of sulfatide in white matter of subjects with very mild AD relative to controls varied from 35 mol% in cerebellum to 58 mol% in the superior frontal region. Sulfatide content in severe AD subjects has been analyzed previously by using conventional methodologies such as high-performance thin layer chromatography. However, conflicting results were observed [74–76]. One group reported significantly higher sulfatide mass levels in AD subjects than in normal controls [74], but

0 650

750

700 m/z (a)

800

20

0 650

700

750

774.6 (18:0–22.6 PlsEtn)

790.6 (18:0–22:6 PtdEtn)

40

750.6 (18:0–20:4 PlsEtn)

60

726.6 (18:1–18:1 PlsEtn)

766.6 (18:0–20:4 PtdEtn)

80

295

700.6 (16:0–18:1 PlsEtn)

White matter

662.6 (15:0–15:0 PlsEtn, I.S.)

20

100

Relative abundance (%)

40

790.6 (18:0–22.6 PtdEtn)

60

774.6 (18:0–22:6 PlsEtn)

750.6 (18:0–20:4 PlsEtn)

15:0–15:0 PtdEtn 662.6 (I.S.) 726.6 (18:1–18:1 PlsEtn)

80

Gray matter

700.6 (16:0–18:1 PlsEtn)

Relative abundance (%)

100

766.6 (18:0–20:4 PtdEtn)

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

800

m/z (b)

FIGURE 13.3 Representative negative-ion ESI mass spectra of ethanolamine glycerophospholipids from parietal gray and white matter of normal human. Brain samples were obtained from the brain bank of the Washington University ADRC Neuropathology/Tissue Resource Core, and lipids were extracted by a modified method of Bligh–Dyer [87]. Negative-ion ESI mass spectra of lipid extracts of parietal gray matter (Panel a) and white matter (Panel b) of the normal human brain were acquired with LiOH present. Individual molecular species indicated were identified using tandem mass spectrometry and represent the major species. PlsEtn and PtdEtn abbreviate plasmenylethanolamine and phosphatidylethanolamine, respectively. I.S. denotes internal standard. (Han, Unpublished data).

another found 38% lower sulfatide content in late-onset AD cases compared to controls [75]. These differences probably resulted from cross contamination between gray and white matter during sampling, which is inevitable when using conventional methodologies in which several grams of tissue are commonly required. Less than 50 mg of tissue samples were used in our studies by using ESI/MS, and furthermore, the purity of gray matter or white matter was established in situ to minimize cross contamination as previously described [50]. Lipidomic analysis of other classes of lipids demonstrated that there were no mass differences of PC, PtdIns, PtdSer, PtdEtn, cerebroside, SM, and cholesterol present in lipid extracts of both gray and white matter from all examined brain regions between subjects with very mild AD and age-matched cognitively normal controls. For example, negative-ion ESI mass spectra in Figure 13.4 displayed essentially equal intensive ion peaks corresponding to PtdIns and PtdSer molecular species in lipid extracts of parietal gray matter between subjects with very mild AD and controls. Further examination of mass content of these classes of lipids in subjects with more severe AD demonstrated that only modest alterations in these lipids (up to 15% reduction) were found in very severe AD patients. These results suggest that the loss of sulfatide mass content in AD subjects at their earliest clinically recognizable stage is specific, relative to the lipid molecular species of

296

Functional Lipidomics 693.5 (14:0–16:0 PtdGro)

80 Relative intensity (%)

A cognitively normal subject

Internal standard

100

60

N24:1 ST 888.7

18:0–20:4 PtdIns 885.7 18:0–22:6 PtdSer 834.7

40

2OH N24:1 ST 904.8

20

0 650

750

800 m/z (a)

693.5 (14:0–16:0 PtdGro)

850

900

950

A very mild AD subject

Internal standard

100

80 Relative intensity (%)

700

60

40 18:0–22:6 PtdSer 834.7

20

18:0–20:4 PtdIns 885.7

N24:1 ST 888.7 0 650

700

750

800 m/z (b)

850

900

950

FIGURE 13.4 Representative negative-ion ESI mass spectra of lipid extracts from parietal gray matter from a subject who was cognitively normal and a subject with very mild Alzheimer’s-type dementia. Brain samples were obtained from the brain bank of the Washington University ADRC Neuropathology/Tissue Resource Core, and lipids were extracted by a modified method of Bligh–Dyer [87]. Negative-ion ESI mass spectra of lipid extracts of parietal gray matter of cognitively normal human brain (Panel a) and of AD brain (Panel b) were acquired with LiOH absent. All major individual molecular species indicated were identified using tandem mass spectrometry. (Han, unpublished data).

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

297

other classes. Thus, substantial sulfatide depletion in the very early stage of AD development may play an important role in AD pathogenesis and may be linked with early events in the pathological process of AD, including neurodegeneration, synapse loss, and synaptic dysfunction. To determine the specificity of altered sulfatide mass levels in AD among other neurodegenerative diseases, we have recently examined alterations in sulfatide content in postmortem brain samples from subjects with non-AD-related Parkinson’s disease (PD) and dementia with Lewy bodies (DLB) [77]. In contrast to AD cases, the sulfatide contents of all examined brain regions in both gray matter and white matter from PD subjects were dramatically elevated compared to cognitively normal controls. For example, the total sulfatide mass increased from 102.6 ± 12.6 in controls to 163.7 ± 22.1 nmol/mg of protein in PD subjects in superior frontal white matter (elevated by 60 mol%). This sulfatide mass level in frontal white matter from PD subjects is 3.5-fold higher than that observed in AD subjects. Examination of lipid alterations in DLB samples demonstrated that mass levels of sulfatide in both gray matter and white matter were similar to those observed in controls. These studies suggest that sulfatide deficiency in very mild AD subjects is specific among examined neurodegenerative diseases. We further hypothesized that the sulfatide content in cerebrospinal fluid (CSF) is altered in parallel to the alterations in brain tissue sulfatide content. By using the ESI/MS technique, we have examined the sulfatide mass levels in the CSF of subjects who have been clinically identified at the very earliest clinical stage of AD (MiniMental State Examination score, 26.4 ± 0.72). We found that there was a marked decline (approximately 40%) in sulfatide mass in the CSF of AD subjects vs. cognitively normal controls, whereas other classes of lipids, such as PtdIns, were not significantly different between the groups [78]. This observation has recently been further validated by the examination of sulfatide content in CSF from a large group of AD patients (unpublished data). These results further support the findings of specific sulfatide alterations in postmortem brain tissues of subjects with very mild AD and suggest that sulfatide mass content in CSF is a potential biomarker for the early diagnosis of AD. These lipidomics studies not only revealed that substantial sulfatide depletion at the very early stage of AD development may be linked with early events in the pathological process of AD, including neurodegeneration, synapse loss, and synaptic dysfunction, but also redirected our research interests in the relationship between sulfatide depletion and AD pathogenesis. Intriguingly, during our course to elucidate the mechanisms underlying the sulfatide deficiency in very early AD, by using the power of lipidomics, we uncovered a novel role of apoE in the CNS: modulation of sulfatide mass content [79]. The role of apoE in sulfatide homeostasis was supported by three experimental observations. First, sulfatides are specifically associated with apoE-containing lipoprotein particles, indicating that sulfatide can be transported by apoE particles to brain interstitial fluid and CSF. Second, apoE modulates sulfatide mass levels in the CNS through the delivery of apoE-containing particles to CSF and the endocytotic recycling of apoE particles containing sulfatide through the lowdensity lipoprotein (LDL) receptor or LDL receptor family members such as LDLreceptor-related proteins. Third, the modulation of sulfatide content in the CNS is

298

Functional Lipidomics

apoE-isoform dependent. This conclusion was supported by the substantially lower sulfatide mass content in human apoE4 transgenic mice than that in human apoE3 transgenic mice. This was further validated by identifying the significantly higher sulfatide mass content in CSF from cognitively normal subjects with one or two alleles of apoE4 in comparison to those individuals homozygous for apoE3. Collectively, these findings not only support the role of apoE in the regulation of sulfatide metabolism in the CNS, but also suggest that apoE may be involved in the sulfatide loss in very mild AD patients. Therefore, apoE [16–19], the LDL receptor, and the LDL-receptor-related protein [80], and also heparan sulfate proteoglycans (HSPG) [81], all of which are linked experimentally with AD, may be interrelated in the abnormal sulfatide metabolism of AD pathogenesis. To date, the reason for the marked sulfatide deficiency in the CNS of Alzheimer’s patients at its earliest clinically recognizable stage is still unresolved, although it may reflect axonal damage or degeneration or abnormal metabolism of apoE-associated lipoproteins, or both. However, the significance of the finding of sulfatide deficiency in very early AD is apparent [82]. First, because sulfatide deficiency is a very early event in AD pathology, the final pathological mechanism for AD must include or explain this biochemical aberration. Most important, this finding suggests that subjects with mild cognitive impairment due to AD already have dramatic biochemical or pathological changes, which ultimately will be critical to target as early as possible in the disease course to have a major therapeutic impact. Second, sulfatide metabolism is associated with apoE, providing new insights into the connection of apoE and the LDL receptor or LDL-receptor-related proteins with AD. Third, sulfatide deficiency in early AD may be a useful biomarker for AD. Therefore, although the implication of these findings in the development of therapeutics for AD treatment is unknown, the determination of CSF sulfatide mass levels, either solely or in combination with other biomarkers, may be useful in differentiating individual subjects with or without the cognitive changes indicative of AD, in predicting cognitive decline, as well as in assessing response to treatment.

IDENTIFICATION OF DIABETES-INDUCED CHANGES MOLECULAR SPECIES

IN

SPECIFIC LIPID

In 1994, we proposed that diabetic cardiomyopathy was predominantly a disease of maladaptive lipid metabolism in response to insulin resistance. Although the prevailing dogma at the time considered diabetic myocardial dysfunction to result predominantly from microvascular ischemic changes, our studies of phospholipases and results from shotgun lipidomics led us to recognize that diabetic cardiomyopathy likely resulted from dysfunctional lipid metabolism [44,83]. Subsequent quantitation of major mass constituents in the myocardial lipidome from control rats, diabetic rats, or diabetic rats treated with insulin, through shotgun lipidomics methods, unambiguously demonstrated alterations in specific molecular species of triglycerides [84]. Moreover, substantial increases in the content of plasmalogen molecular species were also present (Note the increases present at m/z 746.6 and 774.6 as well as their neighboring peaks shown in Figure 13.5.) Other specific lipid alterations induced by diabetes include: (1) the myocardial PtdIns mass increased 46% in

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

650

700

750

800

850

662.6

20 0

790.6

600

700

800

m/z (a)

0

Relative intensity (%)

722.6 693.6

600

650

742.6

40

766.6 774.6

790.6 746.6

714.6

Relative intensity (%)

80

20

693.6

100

60

700

750 m/z (b)

900

m/z (c)

662.6

100

885.6

747.6 723.5

40

800

850

80

885.6

60

747.6

40

662.6

723.5 790.6

20 0

600

700

800

861.6

600

60

834.7

20

774.6

722.6

742.6

746.6

80

861.6

790.6

693.6

40

0

Relative intensity (%)

766.6

714.6

Relative intensity (%)

80 60

693.6

100

834.7

662.6

100

299

900

m/z (d)

FIGURE 13.5 Representative ESI mass spectra of myocardial lipid extracts from control and diabetic rats. Negative-ion ESI mass spectra of myocardial lipid extracts from control (spectra a and c) and diabetic (spectra b and d) rats were acquired with LiOH present (spectra a and b, which show PE molecular species) and absent (spectra c and d, which show anionic phospholipids). Individual molecular species were identified using tandem mass spectrometry. The internal standards were 15:0-15:0 PtdEtn (m/z 662.6) or 14:0-16:0 PtdGro (m/z 693.6). The masses of all ions were rounded to the nearest integer. The clusters of m/z 722.6, 746.6, and 774.6 in spectra a and b correspond to PlsEtn molecular species. (Adapted from Han, X.; Abendschein, D.R., Kelley, J.G., and Gross, R.W. Diabetes-induced changes in specific lipid molecular species in rat myocardium. Biochem. J. 2000, 352(1), 79–89. With permission from Biochemical Society, Copyright 2000).

diabetic rats in comparison to normal controls (Figure 13.5), and (2) a 22% decrease in 18:0-20:4 PtdEtn occurred in the absence of a change in nonesterified AA or total myocardial AA mass (Note the decrease in phosphatidylethanolamine at m/z 766.6 in Figure 13.5.) Insulin treatment allowed the complete recovery of each of the diabetes-induced alterations in phospholipid classes, subclasses, and individual molecular species. In sharp contrast, many alterations in TAG molecular species were not prevented by peripheral insulin treatment after the induction of diabetic state [84]. These results segregate alterations in myocardial lipid metabolism in the diabetic state into changes that either can be remedied easily by routine peripheral insulin treatment or are difficult (perhaps impossible) to recover by this treatment alone. Accordingly, the study of lipidomics has already led to new insights into the diabetic

300

Functional Lipidomics

state and the importance (and potential lipotoxicity) of saturated vs. nonsaturated fatty acids and their differential metabolic fates. Through detailed study of the lipidome, new insights into lipotoxicity and the differential effects of saturated and unsaturated acyl moieties in cell metabolism and TAG storage pools were first demonstrated. The combined presence of increased triglyceride molecular species in conjunction with increased plasmalogen content implicated the peroxisomal–mitochondrial axis as the responsible entity contributing to the pathologic lipid alterations manifest in diabetes. Moreover, lipidomics has influenced the direction of research in many groups by providing new insights into the metabolic dysfunction underlying the diabetic state [85,86]. For example, Finck and colleagues [85] demonstrated that mouse hearts overexpressing peroxisome-proliferators-activated receptor α in a cardiac-myocytespecific fashion (MHC-PPARα) exhibited signatures of diabetic cardiomyopathy (e.g., ventricular hypotrophy), activation of gene markers of pathologic hypertrophic growth, and transgene expression-dependent alteration in systolic ventricular dysfunction. Further studies found that derangements in myocardial energy metabolism typical of the diabetic heart can become maladaptive, leading to cardiomyopathy [85]. Furthermore, cardiomyopathy in diabetic MHC-PPAR mice was accompanied by myocardial long-chain TAG accumulation [86]. Importantly, the cardiomyopathic phenotype was exacerbated in MHC-PPAR mice fed with a diet enriched with TAGs containing long-chain fatty acids, whereas it was reversed by feeding the mice with a diet enriched in medium-chain (mainly C8 and C10) TAGs [86].

CONCLUSION Although lipidomics, as a field, is still in its early stages of development, the power of lipidomics has already been demonstrated by identification of alterations in lipid molecular species in specific disease states. This technology provides analytical penetration that can lead to identification of the biochemical mechanisms underlying multiple disease states. Shotgun lipidomics by ESI mass spectrometry with intrasource separation greatly facilitates the accrual of new information with minimal sample workup. Thus, through application of a systems biology approach, delineation of the markers that identify pathological alterations, diagnostic of disease onset, progression, or severity, can be elucidated. Moreover, evaluation of technology to determine treatment efficacy and response to therapy is a proximal goal of technology development in lipidomics.

ACKNOWLEDGMENT This work was supported by NIH grants PO1HL57278 and RO1HL41250 and NIA grant RO1 AG23168. Dr. Xianlin Han is the 2003/2004 Memory Ride Prize recipient. The authors are grateful to the Washington University Mass Spectrometer Facility Center, which is supported by NIH grants P41-RR00954, P60-DK20579, and P30DK56341, for the use of electrospray ionization mass spectrometer.

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

301

References 1. Friedman, J.M. Obesity in the new millennium. Nature 2000, 404(6778), 632–634. 2. Tuomilehto, J., Lindstrom, J., Eriksson, J.G., Valle, T.T., Hamalainen, H., IlanneParikka, P., Keinanen-Kiukaanniemi, S., Laakso, M., Louheranta, A., Rastas, M., Salminen, V., and Uusitupa, M. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N. Engl. J. Med. 2001, 344(18), 1343–1350. 3. Sinha, R., Fisch, G., Teague, B., Tamborlane, W.V., Banyas, B., Allen, K., Savoye, M., Rieger, V., Taksali, S., Barbetta, G., Sherwin, R.S., and Caprio, S. Prevalence of impaired glucose tolerance among children and adolescents with marked obesity. N. Engl. J. Med. 2002, 346(11), 802–810. 4. Bhargava, S.K., Sachdev, H.S., Fall, C.H., Osmond, C., Lakshmy, R., Barker, D.J., Biswas, S.K., Ramji, S., Prabhakaran, D., and Reddy, K.S. Relation of serial changes in childhood body-mass index to impaired glucose tolerance in young adulthood. N. Engl. J. Med. 2004, 350(9), 865–875. 5. Kenchaiah, S., Evans, J.C., Levy, D., Wilson, P.W., Benjamin, E.J., Larson, M.G., Kannel, W.B., and Vasan, R.S. Obesity and the risk of heart failure. N. Engl. J. Med. 2002, 347(5), 305–313. 6. Chauhan, N.B. Membrane dynamics, cholesterol homeostasis, and Alzheimer’s disease. J. Lipid Res. 2003, 44(11), 2019–2029. 7. Koudinov, A.R. and Koudinova, N.V. Cholesterol, synaptic function and Alzheimer’s disease. Pharmacopsychiatry 2003, 36 Suppl. 2, S107–112. 8. McLaurin, J., Darabie, A.A., and Morrison, M.R. Cholesterol, a modulator of membrane-associated Abeta-fibrillogenesis. Pharmacopsychiatry 2003, 36 Suppl. 2, S130–135. 9. Poirier, J. Apolipoprotein E and cholesterol metabolism in the pathogenesis and treatment of Alzheimer's disease. Trends Mol. Med. 2003, 9(3), 94–101. 10. Puglielli, L., Tanzi, R.E., and Kovacs, D.M. Alzheimer’s disease: the cholesterol connection. Nat. Neurosci. 2003, 6(4), 345–351. 11. Yanagisawa, K. Cholesterol and Abeta aggregation. Pharmacopsychiatry 2003, 36 Suppl. 2, S127–129. 12. Miller, L.J. and Chacko, R. The role of cholesterol and statins in Alzheimer’s disease. Ann. Pharmacother. 2004, 38(1), 91–98. 13. Wolozin, B. Cholesterol and the biology of Alzheimer’s disease. Neuron 2004, 41(1), 7–10. 14. Han, X. Lipid alterations in the earliest clinical stage of Alzheimer’s disease: Implication of the role of lipids in the pathogenesis of Alzheimer's disease. Curr. Alz. Res. 2004, 1, 2005, 2, 65–77. 15. Holtzman, D.M. Abeta conformational change is central to Alzheimer's disease. Neurobiol. Aging 2002, 23(6), 1085–1088. 16. Strittmatter, W.J. and Roses, A.D. Apolipoprotein E and Alzheimer’s disease. Annu. Rev. Neurosci. 1996, 19, 53–77. 17. Cedazo-Minguez, A. and Cowburn, R.F. Apolipoprotein E: a major piece in the Alzheimer’s disease puzzle. J. Cell. Mol. Med. 2001, 5(3), 254–266. 18. Bales, K.R., Dodart, J.C., DeMattos, R.B., Holtzman, D.M., and Paul, S.M. Apolipoprotein E, amyloid, and Alzheimer disease. Mol. Interv. 2002, 2(6), 363–375, 339. 19. Raffai, R.L. and Weisgraber, K.H. Cholesterol: from heart attacks to Alzheimer’s disease. J. Lipid Res. 2003, 44(8), 1423–1430.

302

Functional Lipidomics

20. Mahley, R.W. Apolipoprotein E: cholesterol transport protein with expanding role in cell biology. Science 1988, 240(4852), 622–630. 21. Han, X. The role of apolipoprotein E in lipid metabolism in the central nervous system. Cell. Mol. Life Sci. 2004, 61(15), 1896–1906. 22. Gross, R.W. High plasmalogen and arachidonic acid content of canine myocardial sarcolemma: a fast atom bombardment mass spectroscopic and gas chromatographymass spectroscopic characterization. Biochemistry 1984, 23(1), 158–165. 23. Gross, R.W. Identification of plasmalogen as the major phospholipid constituent of cardiac sarcoplasmic reticulum. Biochemistry 1985, 24(7), 1662–1668. 24. Pak, J.H., Bork, V.P., Norberg, R.E., Creer, M.H., Wolf, R.A., and Gross, R.W. Disparate molecular dynamics of plasmenylcholine and phosphatidylcholine bilayers. Biochemistry 1987, 26(15), 4824–4830. 25. Ford, D.A. and Gross, R.W. Identification of endogenous 1-O-alk-1-enyl-2-acyl-snglycerol in myocardium and its effective utilization by choline phosphotransferase. J. Biol. Chem. 1988, 263(6), 2644–2650. 26. Ford, D.A. and Gross, R.W. Plasmenylethanolamine is the major storage depot for arachidonic acid in rabbit vascular smooth muscle and is rapidly hydrolyzed after angiotensin II stimulation. Proc. Natl. Acad. Sci. U. S. A. 1989, 86(10), 3479–3483. 27. Ford, D.A. and Gross, R.W. Differential metabolism of diradyl glycerol molecular subclasses and molecular species by rabbit brain diglyceride kinase. J. Biol. Chem. 1990, 265(21), 12280–12286. 28. Han, X. and Gross, R.W. Plasmenylcholine and phosphatidylcholine membrane bilayers possess distinct conformational motifs. Biochemistry 1990, 29(20), 4992–4996. 29. Hazen, S.L., Stuppy, R.J. and Gross, R.W. Purification and characterization of canine myocardial cytosolic phospholipase A2. A calcium-independent phospholipase with absolute sn-2 regiospecificity for diradyl glycerophospholipids. J. Biol. Chem. 1990, 265(18), 10622–10630. 30. Han, X., Chen, X., and Gross, R.W. Chemical and magnetic inequivalence of glycerol protons in individual subclasses of choline glycerophospholipids: implications for subclass-specific changes in membrane conformational states. J. Am. Chem. Soc. 1991, 113(19), 7104–7109. 31. Creer, M.H. and Gross, R.W. Separation of isomeric lysophospholipids by reverse phase HPLC. Lipids 1985, 20(12), 922–928. 32. Gross, R.W. and Sobel, B.E. Isocratic high-performance liquid chromatography separation of phosphoglycerides and lysophosphoglycerides. J. Chromatogr. 1980, 197(1), 79–85. 33. Taguchi, R., Hayakawa, J., Takeuchi, Y., and Ishida, M. Two-dimensional analysis of phospholipids by capillary liquid chromatography/electrospray ionization mass spectrometry. J. Mass Spectrom. 2000, 35(8), 953–966. 34. DeLong, C.J., Baker, P.R.S., Samuel, M., Cui, Z., andThomas, M.J. Molecular species composition of rat liver phospholipids by ESI-MS/MS: the effect of chromatography. J. Lipid Res. 2001, 42(12), 1959–1968. 35. Koivusalo, M., Haimi, P., Heikinheimo, L., Kostiainen, R., Somerharju, P. Quantitative determination of phospholipid compositions by ESI-MS: effects of acyl chain length, unsaturation, and lipid concentration on instrument response. J. Lipid Res. 2001, 42(4), 663–672. 36. Han, X. and Gross, R.W. Global analyses of cellular lipidomes directly from crude extracts of biological samples by ESI mass spectrometry: a bridge to lipidomics. J. Lipid Res. 2003, 44(6), 1071–1079.

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

303

37. Han, X. and Gross, R.W. Shotgun lipidomics: Electrospray ionization mass spectrometric analysis and quantitation of the cellular lipidomes directly from crude extracts of biological samples. Mass Spectrom. Rev. 2005, 24, 367–412. 38. Fenn, J.B., Mann, M., Meng, C.K., Wong, S.F., and Whitehouse, C.M. Electrospray ionization for mass spectrometry of large biomolecules. Science 1989, 246(4926), 64–71. 39. Han, X. and Gross, R.W. Electrospray ionization mass spectroscopic analysis of human erythrocyte plasma membrane phospholipids. Proc. Natl. Acad. Sci. U. S. A. 1994, 91(22), 10635–10639. 40. Duffin, K.L., Henion, J.D., and Shieh, J.J. Electrospray and tandem mass spectrometric characterization of acylglycerol mixtures that are dissolved in nonpolar solvents. Anal. Chem. 1991, 63(17), 1781–1788. 41. Weintraub, S.T., Pinckard, R.N., and Hail, M. Electrospray ionization for analysis of platelet-activating factor. Rapid Commun. Mass Spectrom. 1991, 5(7), 309–311. 42. Kerwin, J.L., Tuininga, A.R., and Ericsson, L.H. Identification of molecular species of glycerophospholipids and sphingomyelin using electrospray mass spectrometry. J. Lipid Res. 1994, 35(6), 1102–1114. 43. Kim, H.Y., Wang, T.C., and Ma, Y.C. Liquid chromatography/mass spectrometry of phospholipids using electrospray ionization. Anal. Chem. 1994, 66(22), 3977–3982. 44. Han, X., Gubitosi-Klug, R.A., Collins, B.J., and Gross, R.W. Alterations in individual molecular species of human platelet phospholipids during thrombin stimulation: electrospray ionization mass spectrometry-facilitated identification of the boundary conditions for the magnitude and selectivity of thrombin-induced platelet phospholipid hydrolysis. Biochemistry 1996, 35(18), 5822–5832. 45. Lehmann, W.D., Koester, M., Erben, G., and Keppler, D. Characterization and quantification of rat bile phosphatidylcholine by electrospray-tandem mass spectrometry. Anal. Biochem. 1997, 246(1), 102–110. 46. Schneiter, R., Brugger, B., Sandhoff, R., Zellnig, G., Leber, A., Lampl, M., Athenstaedt, K., Hrastnik, C., Eder, S., Daum, G., Paltauf, F., Wieland, F.T., and Kohlwein, S.D. Electrospray ionization tandem mass spectrometry (ESI-MS/MS) analysis of the lipid molecular species composition of yeast subcellular membranes reveals acyl chain-based sorting/remodeling of distinct molecular species en route to the plasma membrane. J. Cell Biol. 1999, 146(4), 741–754. 47. Brugger, B., Sandhoff, R., Wegehingel, S., Gorgas, K., Malsam, J., Helms, J.B., Lehmann, W.D., Nickel, W., Wieland, F.T. Evidence for segregation of sphingomyelin and cholesterol during formation of COPI-coated vesicles. J. Cell Biol. 2000, 151(3), 507–518. 48. Blom, T.S., Koivusalo, M., Kuismanen, E., Kostiainen, R., Somerharju, P., and Ikonen, E. Mass Spectrometric Analysis Reveals an Increase in Plasma Membrane Polyunsaturated Phospholipid Species upon Cellular Cholesterol Loading. Biochemistry 2001, 40(48), 14635–14644. 49. Duffin, K.L., Obukowicz, M.G., Salsgiver, W.J., Welsch, D.J., Shieh, C., Raz, A., and Needleman, P. Lipid remodeling in mouse liver and plasma resulting from delta6 fatty acid desaturase inhibition. Lipids 2001, 36(11), 1203–1208. 50. Han, X., Holtzman, D.M., and McKeel, D.W., Jr. Plasmalogen deficiency in early Alzheimer's disease subjects and in animal models: molecular characterization using electrospray ionization mass spectrometry. J. Neurochem. 2001, 77(4), 1168–1180. 51. Williams, S.D., Hsu, F.F., and Ford, D.A. Electrospray ionization mass spectrometry analyses of nuclear membrane phospholipid loss after reperfusion of ischemic myocardium. J. Lipid Res. 2000, 41(10), 1585–1595.

304

Functional Lipidomics

52. Ekroos, K., Chernushevich, I.V., Simons, K., and Shevchenko, A. Quantitative profiling of phospholipids by multiple precursor ion scanning on a hybrid quadrupole time-of-flight mass spectrometer. Anal. Chem. 2002, 74(5), 941–949. 53. Welti, R., Li, W., Li, M., Sang, Y., Biesiada, H., Zhou, H.-E., Rajashekar, C.B., Williams, T.D., and Wang, X. Profiling membrane lipids in plant stress responses. Role of phospholipase Da in freezing-induced lipid changes in Arabidopsis. J. Biol. Chem. 2002, 277(35), 31994–32002. 54. Esch, S.W., Williams, T.D., Biswas, S., Chakrabarty, A., and Levine, S.M. Sphingolipid profile in the CNS of the twitcher (globoid cell leukodystrophy) mouse: A lipidomics approach. Cell. Mol. Biol. 2003, 49(5), 779–787. 55. Fenn, J.B. Electrospray wings for molecular elephants (Nobel lecture). Angew. Chem. Int. Ed. Engl. 2003, 42(33), 3871–3894. 56. Tanaka, K. The origin of macromolecule ionization by laser irradiation (Nobel lecture). Angew. Chem. Int. Ed. Engl. 2003, 42(33), 3860–3870. 57. Han, X. Characterization and direct quantitation of ceramide molecular species from lipid extracts of biological samples by electrospray ionization tandem mass spectrometry. Anal. Biochem. 2002, 302(2), 199–212. 58. Ford, D.A., Han, X., Horner, C.C., and Gross, R.W. Accumulation of unsaturated acylcarnitine molecular species during acute myocardial ischemia: metabolic compartmentalization of products of fatty acyl chain elongation in the acylcarnitine pool. Biochemistry 1996, 35(24), 7903–7909. 59. Kalderon, B., Sheena, V., Shachrur, S., Hertz, R., and Bar-Tana, J. Modulation by nutrients and drugs of liver acyl-CoAs analyzed by mass spectrometry. J. Lipid Res. 2002, 43(7), 1125–1132. 60. Han, X., Yang, J., Cheng, H., Ye, H., and Gross, R.W. Towards fingerprinting cellular lipidomes directly from biological samples by two-dimensional electrospray ionization mass spectrometry. Anal. Biochem. 2004, 330(2), 317–331. 61. Han, X. and Gross, R.W. Quantitative analysis and molecular species fingerprinting of triacylglyceride molecular species directly from lipid extracts of biological samples by electrospray ionization tandem mass spectrometry. Anal. Biochem. 2001, 295(1), 88–100. 62. Duffin, K., Obukowicz, M., Raz, A., and Shieh, J.J. Electrospray/tandem mass spectrometry for quantitative analysis of lipid remodeling in essential fatty acid deficient mice. Anal. Biochem. 2000, 279(2), 179–188. 63. Vos, J.P., Lopes-Cardozo, M., and Gadella, B.M. Metabolic and functional aspects of sulfogalactolipids. Biochim. Biophys. Acta 1994, 1211(2), 125–149. 64. Ishizuka, I. Chemistry and functional distribution of sulfoglycolipids. Prog. Lipid Res. 1997, 36(4), 245–319. 65. Ikami, T., Ishida, H., and Kiso, M. Synthesis and biological activity of glycolipids, with a focus on gangliosides and sulfatide analogs. Methods Enzymol. 2000, 311, 547–568. 66. Marcus, J. and Popko, B. Galactolipids are molecular determinants of myelin development and axo-glial organization. Biochim. Biophys. Acta 2002, 1573(3), 406–413. 67. von Figura, K.; Gieselmann, V.; Jaeken, J. Metachromatic leukodystrophy: Lysosomal disorders. in The Metabolic and Molecular Bases of Inherited Diseases, 8th ed., Sachdev, H.S., Beaudet, A.L., Sly, W.S., Valle, D., Eds., McGraw-Hill: New York, 2001; 3695–3724. 68. Bosio, A., Binczek, E., and Stoffel, W. Functional breakdown of the lipid bilayer of the myelin membrane in central and peripheral nervous system by disrupted galactocerebroside synthesis. Proc. Natl. Acad. Sci. U. S. A. 1996, 93(23), 13280–13285.

Specific Lipid Alterations in Alzheimer’s Disease and Diabetes

305

69. Coetzee, T., Fujita, N., Dupree, J., Shi, R., Blight, A., Suzuki, K., and Popko, B. Myelination in the absence of galactocerebroside and sulfatide: normal structure with abnormal function and regional instability. Cell 1996, 86(2), 209–219. 70. Bosio, A., Bussow, H., Adam, J., and Stoffel, W. Galactosphingolipids and axonoglial interaction in myelin of the central nervous system. Cell Tissue Res. 1998, 292(2), 199–210. 71. Coetzee, T., Dupree, J.L., and Popko, B. Demyelination and altered expression of myelin-associated glycoprotein isoforms in the central nervous system of galactolipiddeficient mice. J. Neurosci. Res. 1998, 54(5), 613–622. 72. Han, X., Holtzman, D.M., McKeel, D.W., Jr., Kelley, J., and Morris, J.C. Substantial sulfatide deficiency and ceramide elevation in very early Alzheimer's disease: potential role in disease pathogenesis. J. Neurochem. 2002, 82(4), 809–818. 73. Morris, J.C. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 1993, 43(11), 2412–2414. 74. Majocha, R.E., Jungalwala, F.B., Rodenrys, A., and Marotta, C.A. Monoclonal antibody to embryonic CNS antigen A2B5 provides evidence for the involvement of membrane components at sites of Alzheimer degeneration and detects sulfatides as well as gangliosides. J. Neurochem. 1989, 53(3), 953–961. 75. Svennerholm, L. and Gottfries, C.G. Membrane lipids, selectively diminished in Alzheimer brains, suggest synapse loss as a primary event in early-onset form (type I) and demyelination in late-onset form (type II). J. Neurochem. 1994, 62(3), 1039–1047. 76. Gottfries, C.G.,Karlsson, I., and Svennerholm, L. Membrane components separate early-onset Alzheimer’s disease from senile dementia of the Alzheimer type. Int. Psychogeriatr. 1996, 8(3), 365–372. 77. Cheng, H., Xu, J., McKeel, D.W., Jr., and Han, X. Specificity and potential mechanism of sulfatide deficiency in Alzheimer's disease: An electrospray ionization mass spectrometric study. Cell. Mol. Biol. 2003, 49(5), 809–818. 78. Han, X., Fagan, A.M., Cheng, H., Morris, J.C., Xiong, C., and Holtzman, D.M. Cerebrospinal fluid sulfatide is decreased in subjects with incipient dementia. Ann. Neurol. 2003, 54(1), 115–119. 79. Han, X., Cheng, H., Fryer, J.D., Fagan, A.M., and Holtzman, D.M. Novel role for apolipoprotein E in the central nervous system. Modulation of sulfatide content. J. Biol. Chem. 2003, 278(10), 8043–8051. 80. Van Uden, E., Kang, D.E., Koo, E.H., and Masliah, E. LDL receptor-related protein (LRP) in Alzheimer's disease: towards a unified theory of pathogenesis. Microsc. Res. Tech. 2000, 50(4), 268–272. 81. Mulder, M. and Terwel, D. Possible link between lipid metabolism and cerebral amyloid angiopathy in Alzheimer's disease: A role for high-density lipoproteins? Haemostasis 1998, 28(3-4), 174–194. 82. Irizarry, M.C. A turn of the sulfatide in Alzheimer’s disease. Ann. Neurol. 2003, 54(1), 7–8. 83. Hazen, S.L., Hall, C.R., Ford, D.A., and Gross, R.W. Isolation of a human myocardial cytosolic phospholipase A2 isoform. Fast atom bombardment mass spectroscopic and reverse-phase high pressure liquid chromatography identification of choline and ethanolamine glycerophospholipid substrates. J. Clin. Invest. 1993, 91(6), 2513–2522. 84. Han, X., Abendschein, D.R., Kelley, J.G., and Gross, R.W. Diabetes-induced changes in specific lipid molecular species in rat myocardium. Biochem. J. 2000, 352(1), 79–89.

306

Functional Lipidomics

85. Finck, B.N., Lehman, J.J., Leone, T.C., Welch, M.J., Bennett, M.J., Kovacs, A., Han, X., Gross, R.W., Kozak, R., Lopaschuk, G.D., and Kelly, D.P. The cardiac phenotype induced by PPARalpha overexpression mimics that caused by diabetes mellitus. J. Clin. Invest. 2002, 109(1), 121–130. 86. Finck, B.N., Han, X., Courtois, M., Aimond, F., Nerbonne, J.M., Kovacs, A., Gross, R.W., and Kelly, D.P. A critical role for PPARalpha-mediated lipotoxicity in the pathogenesis of diabetic cardiomyopathy: modulation by dietary fat content. Proc. Natl. Acad. Sci. U. S. A. 2003, 100(3), 1226–1231. 87. Bligh, E.G. and Dyer, W.J. A rapid method of total lipid extraction and purification. Can. J. Biochem. Physiol. 1959, 37, 911–917.

14

High-Throughput Lipid Profiling to Identify and Characterize Genes Involved in Lipid Metabolism, Signaling, and Stress Response Ruth Welti, Jyoti Shah, Steven LeVine, Wynn Esch, Todd Williams, and Xuemin Wang

CONTENTS Introduction............................................................................................................308 The Platforms to Profile Lipid Molecular Species ...............................................308 Profiling Processes .......................................................................................308 Sample Preparation ......................................................................................309 Quantification and Internal Standards .........................................................310 Data Handling ..............................................................................................310 Species Profiled............................................................................................312 Functional Characterization of Genes Involved in Lipid Metabolism and Signaling................................................................................................312 Phospholipases .............................................................................................312 Desaturases and DHAP Reductase ..............................................................314 Identification of Genes Involved in Lipid Signaling ............................................316 Genes Involved in Plant Defense Responses ..............................................316 Analysis of Lipid Changes in Pathological Conditions........................................317 Sphingolipids in Globoid Cell Leukodystrophy..........................................317 Lipid Changes in Plant Disease...................................................................318 Perspectives............................................................................................................319 Potential........................................................................................................319 Challenges ....................................................................................................319 References..............................................................................................................320

307

308

Functional Lipidomics

INTRODUCTION The Kansas Lipidomics Research Center (KLRC) comprises the authors of this chapter and other scientists, who are studying complex lipids and the metabolism of complex lipids in plants, animals, fungi, and eubacteria with a focus on membrane lipids. We are utilizing mass-spectrometry-based lipid profiling to identify and characterize gene products involved in lipid metabolism, in lipid signaling, and in response to stress. Several members of our group work with the model plant Arabidopsis thaliana and other organisms in which genetically altered strains are readily available; thus, a major goal of our group is to determine gene-product function by comparing lipid profiles of wild-type and genetically altered organisms. Another research thrust is to identify lipid changes that occur in response to physiological manipulations or stress. An additional aim is to make lipid-profiling technology available to the international community of scientists through the KLRC Analytical Laboratory, which is a fee-for-service facility. The KLRC group has focused on implementing analytical strategies suitable for determining the lipid profiles of relatively large numbers of samples. Although we have utilized chromatography for some specific analyses, our primary approach is to introduce unfractionated extracts directly into a mass spectrometer, utilizing electrospray ionization tandem mass spectrometry (ESI-MS/MS) in the precursor and neutral loss scanning modes to identify and quantify lipid species. This directinfusion, high-throughput strategy provides head group and mass information that generally allows speciation of polar lipids to the level of number of carbon atoms and number of double bonds present in the acyl chains.

THE PLATFORMS TO PROFILE LIPID MOLECULAR SPECIES PROFILING PROCESSES ESI-MS/MS precursor and neutral loss scanning methods for phosphatidylcholine (PC), sphingomyelin, phosphatidylethanolamine (PE), phosphatidylinositol (PI), phosphatidylserine (PS), phosphatidylglycerol (PG), and phosphatidic acid (PA) are based on the work of Brügger and coworkers [1] (Table 14.1). To these analyses, we have added scans for the plant lipids, monogalactosyldiacylglycerol (MGDG), digalactosyldiacylglycerol (DGDG), and sulfoquinovosyldiacylglycerol (SQDG) (Table 14.1) [2,3]. Our strategy for plant lipid analysis involves dividing each sample into two aliquots (Table 14.1). Precursor and neutral loss scans of these unfractionated extracts produce a series of spectra, with each spectrum revealing a set of lipid species containing a common head group fragment. A lipid profile is generated from a series of precursor and neutral loss spectra obtained sequentially as the sample is continuously infused (Table 14.1). Mammalian sphingolipids are analyzed by LC-ESI-MS/MS in the positive ion mode, with upstream separation of crude lipid extracts by reverse-phase HPLC using 2.1-mm C18 columns [4]. Precursor-ion experiments permit selective exploratory profiling of brain ceramides and galactosylceramides with d18:1 and d18:2 long-

High-Throughput Lipid Profiling to Identify and Characterize Genes

309

TABLE 14.1 Precursor and Neutral Loss Scans for Analysis of Lipid Species from Plants Scan Mode + + — — — — + +

Collision Energy (V) 40 28 57 58 34 57 50 84

Fragment Detected Prec 184+ NL 141 Prec 153– Prec 241– NL 87 Prec 225– Prec 243+ Prec 243+

Classes Analyzed LysoPC/PC as [M+H]+ LysoPE/PE as [M+H]+ LysoPG/PG/PA as [M−H]– PI as [M−H]– PS as [M−H]– SQDG as [M−H]– MGDG as [M+Na]+ DGDG as [M+Na]+

Time (min) 3 2 4 2 3 2 5 5

Note: The first six scans are performed on an aliquot of extract dissolved in chloroform/methanol/water (300:665:35) containing 10.5-mM ammonium acetate, whereas the MGDG and DGDG scans are performed on a second aliquot of extract that is dissolved in chloroform/methanol/water (300:665:35) containing 1.75-mM sodium acetate. The collision energies that are indicated are for an Applied Biosystems API 4000 ESI-MS/MS, using nitrogen gas in the collision cell. Spectra are acquired sequentially by scanning in the listed modes for the indicated time period.

chain bases because these lipids have characteristic product ions at m/z 264.3 and 262.3, respectively [5,6,7]. Once precursor-ion screening of lipid components in a tissue extract has been achieved for an experimental sample, follow-up multiple reaction monitoring (MRM) experiments allow semiquantitative characterization of targeted sphingolipids, often with a tenfold or greater gain in sensitivity. While retaining a reverse-phase column separation method, introduction of a solution of lithium hydroxide into the column eluate upstream of the ESI source enhances ionization and allows us to obtain predictable fragments [8].

SAMPLE PREPARATION The Bligh–Dyer method [9] is utilized for extraction of lipids from animal, fungal, and bacterial tissues. Because plant tissues contain particularly active phospholipase D, the first step with plants is extraction at 75˚C with isopropanol, a secondary alcohol that is not a substrate for phospholipase D’s transphosphatidylation activity. This 15-min treatment is followed by repeated extractions in chloroform/methanol (2:1) [2]. Butylated hydroxytoluene is added to the solvents to decrease the possibility of free-radical-mediated oxidation reactions. The solvent is evaporated, and the combined extracts are dissolved in chloroform. The direct-infusion solvent is chloroform/methanol/water (300:665:35) containing either 10.5-mM ammonium acetate or 1.75-mM sodium acetate (Table 14.1).

310

Functional Lipidomics

2

PG

* 0.6

PE

0.4 0.2 0.8 0.6 0.4 0.2 0.0

* PC *

*

32: 0 32: 1 34: 0 34: 1 34: 2 34: 3 34: 4 36: 1 36: 2 36: 3 36: 4 36: 5 36: 6 38: 2 38: 3 38: 4 38: 5 38: 6 40: 2 40: 3 42: 2 42: 3 42: 4

Lipid amount (nmol)

1

Lipid species (total carbons: total double bonds)

FIGURE 14.1 The ability of ESI-MS/MS lipid profiling to detect small changes in lipid molecular species. The lipid species in 25 μl of an Arabidopsis extract from leaves were analyzed and, to demonstrate the ability of the lipid-profiling methodology to detect changes, small, known amounts of specific lipid species were added to the extract. The additions of the lipid species indicated by the asterisks varied over a 15-fold range (with bars for each species representing no addition, 1x addition, 2x addition, 5x addition, and 15x addition). Specifically, the red bars (first series of bars) represent the composition of the extract (n = 5). The green bars (second series of bars) represent extract with 0.022-nmol 32:0 PC, 0.022nmol 36:2 PC, 0.012-nmol 36:2 PE, and 0.012-nmol 32:0 PG added (n = 5). The blue bars (third series of bars) represent extract with 0.044-nmol 32:0 PC, 0.044-nmol 36:2 PC, 0.024nmol 36:2 PE, and 0.024-nmol 32:0 PG added (n = 5). The aqua bars (fourth series of bars) represent extract with 0.11-nmol 32:0 PC, 0.11-nmol 36:2 PC, 0.06-nmol 36:2 PE, and 0.06nmol 32:0 PG added (n = 5). The lavender bars (fifth series of bars) represent extract with 0.33-nmol 32:0 PC, 0.33-nmol 36:2 PC, 0.18-nmol 36:2 PE, and 0.18-nmol 32:0 PG added (n = 5). Error bars are standard deviation.

QUANTIFICATION

AND INTERNAL

STANDARDS

Quantification of direct-infusion samples is in comparison to two internal standards of each head group class (Table 14.2) [1,2]. This method provides good precision and allows detection of small changes in lipid composition (Figure 14.1). In the LCESI-MS/MS analysis of sphingolipids, data are normalized to the detected signal of added d18:1/8:0 ceramide or d18:1/8:0 galactosylceramide. This approach, employing a single internal standard, allows precise comparisons of the relative amounts of lipids in different sets of tissues, but does not provide as accurate quantification as obtained with two internal standards.

DATA HANDLING Mass spectral data are exported into spreadsheet (Microsoft Excel) files. The lists of detected masses and peak areas are automatically sorted against lists of the masses

High-Throughput Lipid Profiling to Identify and Characterize Genes

311

TABLE 14.2 Internal Standards Utilized for Lipid Profiling by the KLRC Group Lipid Class

Low Mass Standard

Internal Standards for Direct-Infusion Analysis Phosphatidylcholine Di 14:0 PCa Phosphatidylethanolamine Di 14:0 PEa Phosphatidylglycerol Di 14:0 PGa Phosphatidylinositol 16:0–18:0 PIc Phosphatidic acid Di 14:0 PAa Phosphatidylserine Di 14:0 PSa Lysophosphatidylcholine 13:0 LysoPCa Lysophosphatidylethanolamine 14:0 LysoPEa Lysophosphatidylglycerol 14:0 LysoPGa Sulfoquinovosyldiacylglycerol 16:0–18:0 SQDGf Monogalactosyldiacylglycerol 16:0–18:0 MGDGg Digalactosyldiacylglycerol 16:0–18:0 MGDGg

Ceramide

High Mass Standard of (Plant) Lipids Di 24:1 PCa Di 24:1 PEb Di 24:1 PGb Di 18:0 PIc Di-phytanoyl (20:0d) PA Di-phytanoyl (20:0d) PS 19:0 LysoPCa 18:0 LysoPEae 18:0 LysoPGae 18:0–18:0 SQDGf 18:0–18:0 MGDGg 18:0–18:0 MGDGg

Internal Standards for LC-MS/MS Analysis of Sphingolipids d18:1/8:0 Cera

Hexosylceramide

d18:1/8:0 GalCera

a

Available from Avanti Polar Lipids, Inc., Alabaster, AL Prepared by transphosphatidylation [33] of Di 24:1 PC purchased from Avanti Polar Lipids, Inc. cHydrogenated plant PI, purchased from Avanti Polar Lipids, Inc. dPhytanoyl is a saturated, branched chain that has the same chemical formula as 20:0. e18:0 lysoPE and 18:0 lysoPG are present in very minor amounts in natural mixtures. Using these compounds as internal standards does, of course, make it impossible to measure the amounts of these species in the sample. However, if the amount of the internal standard is in large excess to the naturally occurring amount of that species, the error in measurement of other species is negligible. fHydrogenated species of Arabidopsis SQDG (prepared by J. Zhao, E. Baughman, and R. Welti). gHydrogenated species of Arabidopsis MGDG and DGDG (prepared by J. Zhao, E. Baughman, C. Buseman, and R. Welti; also available from from Matreya, Inc., Pleasant Gap, PA). b

of lipid species, the data are automatically entered into spreadsheets that perform isotopic overlap corrections, and the corrected signal values are utilized in comparison to the internal standard signals and amounts to calculate the amount of each lipid species. We are in the process of developing a web-based, publicly accessible data storage and retrieval system for lipid profile data.

312

Functional Lipidomics

TABLE 14.3 Plant Lipid Species Included in KLRC’s Direct-Infusion LipidProfiling Strategy PG 32:1 PG 32:0 PG 34:4 PG 34:3 PG 34:2 PG 34:1 PG 34:0 PA 34:6 PA 34:4 PA 34:3 PA 34:2 PA 34:1 PA 36:6 PA 36:5 PA 36:4 PA 36:3 PA 36:2 PE 34:4 PE 34:3 PE 34:2 PE 34:1 PE 36:6 PE 36:5 PE 36:4 PE 36:3 PE 36:2 PE 36:1

PE 38:6 PE 38:5 PE 38:4 PE 38:3 PE 38:2 PE 40:3 PE 40:2 PE 42:4 PE 42:3 PE 42:2 PI 34:4 PI 34:3 PI 34:2 PI 34:1 PI 36:6 PI 36:5 PI 36:4 PI 36:3 PI 36:2 PI 36:1 PS 34:4 PS 34:3 PS 34:2 PS 34:1 PS 36:6 PS 36:5

PS 36:4 PS 36:3 PS 36:2 PS 36:1 PS 38:6 PS 38:5 PS 38:4 PS 38:3 PS 38:2 PS 38:1 PS 40:4 PS 40:3 PS 40:2 PS 40:1 PS 42:4 PS 42:3 PS 42:2 PS 42:1 PS 44:3 PS 44:2 PC 32:0 PC 34:4 PC 34:3 PC 34:2 PC 34:1 PC 36:6

PC 36:5 PC 36:4 PC 36:3 PC 36:2 PC 36:1 PC 38:6 PC 38:5 PC 38:4 PC 38:3 PC 38:2 PC 40:5 PC 40:4 PC 40:3 PC 40:2 LysoPE 16:1 LysoPE 16:0 LysoPE 18:3 LysoPE 18:2 LysoPE 18:1 LysoPC 16:1 LysoPC 16:0 LysoPC 18:3 LysoPC 18:2 LysoPC 18:1 LysoPC 18:0 lysoPG 16:1

lysoPG 16:0 lysoPG 18:3 lysoPG 18:2 lysoPG 18:1 MGDG 34:6 MGDG 34:5 MGDG 34:4 MGDG 34:3 MGDG 34:2 MGDG 34:1 MGDG 36:6 MGDG 36:5 MGDG 36:4 MGDG 36:3 MGDG 36:2 MGDG 36:1 MGDG 38:6 MGDG 38:5 MGDG 38:4 MGDG 38:3 DGDG 34:6 DGDG 34:5 DGDG 34:4 DGDG 34:3 DGDG 34:2 DGDG 34:1

DGDG 36:6 DGDG 36:5 DGDG 36:4 DGDG 36:3 DGDG 36:2 DGDG 36:1 DGDG 38:6 DGDG 38:5 DGDG 38:4 DGDG 38:3 SQDG 32:3 SQDG 32:2 SQDG 32:1 SQDG 32:0 SQDG 34:6 SQDG 34:5 SQDG 34:4 SQDG 34:3 SQDG 34:2 SQDG 34:1 SQDG 36:6 SQDG 36:5 SQDG 36:4 SQDG 36:3 SQDG 36:2 SQDG 36:1

SPECIES PROFILED Currently, our direct-infusion protocols support the profiling of 157 plant lipid species (Table 14.2) and 136 yeast lipid species; we are currently testing datahandling templates for 330 animal lipid species. LC-ESI-MS/MS of sphingolipids has focused on 28 species of hydroxylated and nonhydroxylated ceramide and galactosylceramide species.

FUNCTIONAL CHARACTERIZATION OF GENES INVOLVED IN LIPID METABOLISM AND SIGNALING PHOSPHOLIPASES Phospholipases are involved in many cellular processes through their roles in lipid metabolism, signal transduction, membrane remodeling, and/or membrane degrada-

High-Throughput Lipid Profiling to Identify and Characterize Genes

313

tion. These enzymes are grouped into four major classes, phospholipase D (PLD), phospholipase C (PLC), phospholipase A2 (PLA2), and phospholipase A1 (PLA1), according to the site of hydrolysis on phospholipids [10,11]. Lipid profiling provides a powerful approach to help understand the function and regulation of different in the cell. For example, it can be utilized to identify lipid metabolic pathways involved in stress responses. The analysis of lipid molecular species during cold acclimation and freezing has revealed many changes associated with the temperature stress [2] (Figure 14.2). These include losses of PC, PE, PG, and MGDG and increases in PA, lysoPC, and lysoPE. The production of lysoPC and lysoPE suggests activation of PLA activity, whereas an increase in PA at the expense of PC, PE, and PG suggests a hydrolysis of the head group by PLD. Moreover, the profiling of PA molecular species provides insight that is not readily obtainable by traditional lipid analysis. One PA species, 34:6PA, was increased during freezing, but because this acyl chain combination is not present in PC, PE, or PG (nor in other phospholipids classes not shown), their hydrolysis did not result in the freezing-associated increase in 34:6PA. However, 34:6 is the major lipid species of MGDG and is also present in small amounts in DGDG. Thus, the observation of 34:6PA formation has allowed us to suggest the existence of a pathway from galactolipids to PA, perhaps by hydrolysis of MGDG to diacylglycerol, followed by phosphorylation. Of various phospholipases in plants, PLD, which produces PA and a free head group, is the most prevalent phospholipase in plant tissues. The plant PLD family comprises multiple enzymes with distinguishable biochemical, structural, and molecular properties. [12,13]. Arabidopsis has 12 PLD genes that are grouped into six types, PLDα (3 genes), β (2 genes), γ (3 genes), δ, ε, and ζ (2 genes). The different characterized types of PLDs display different requirements for Ca2+, PI 4,5-bisphosphate (PIP2), substrate vesicle composition, and/or free fatty acids [13]. These properties suggest that individual PLDs can be activated differently in the cell and thus may have unique functions. The unique functions have been documented for some PLDs from studies involving genetic manipulation of specific PLDs. For example, PLDα1 plays a role in plant freezing injury [2], whereas PLDδ increases plant freezing tolerance [14]. PLDα1 is involved in reactive-oxygen production [15], whereas PLDδ plays a role in a cell’s response to hydrogen peroxide [16]. Lipid profiling, in combination with genetic manipulation, has provided a uniquely effective approach to determining the roles of particular PLD genes and enzymes in stress responses. For instance, the lipid profiles of leaves from wild-type Arabidopsis showed a large loss of molecular species of PC, PE, and PG during freezing treatment, whereas PI and DGDG were unaffected. Levels of PA were increased several fold. Deficiency in PLDα reduced the amount of PA formed during freezing by about 50%, indicating that PLDα contributes about 50% of the PLD hydrolytic activity [2] (Figure 14.2). By comparison, unlike PLDα, a loss of PLD activity does not result in substantial decrease in lipid hydrolysis, but an increase in PLDα expression produces selective PA species during freezing stress [14]. These results indicate that PLDα and PLDα are involved in plant response to freezing via different mechanisms. In addition, comparative lipid profiling of mutant and wild-type tissue also allowed identification of the in vivo substrate of PLDα [2] (Figure 14.2). In vitro,

314

Functional Lipidomics

A

ENDOPLASMIC RETICULUM G3P

PI; PE; PS, PG PC

PA

LPA

CYTOPLASM DHAP

G3P Sfd 1

LPA act 1

PA PG

DAG

DAG

MGDG

SQDG MGDG

PLASTID

DGDG fad4 fad 5

B 16:0 18:0

SQDG

DGDG fad7 fad8

fad6 16:1

16:2

16:3

18:1

18:2

18:3

ssi2

FIGURE 14.2 Freezing-induced changes in lipid molecular species in Arabidopsis as revealed by ESI-MS/MS. The black bars represent cold-acclimated plants, and the hatched bars represent plants subjected to 8ºC. Values are means ± SD (n = 4 or 5). An L indicates that the value is lower than that of the cold-acclimated plants of the same genotype; P < 0.05. An H indicates that the value is higher than that of cold-acclimated plants of the same genotype; P < 0.05. A. Phospholipids and galactolipids of wild-type plants. B. Phospholipids and galactolipids of PLDα-deficient plants. Welti, R., Li, W., Li, M., Sang, Y., Biesiada, H., Zhou, H-E., Rajashekar, C.B., Williams, T.D., and Wang. X. Profiling membrane lipids in plant stress responses: role of phospholipase Dα in freezing-induced lipid changes in Arabidopsis. J. Biol. Chem. 2002, 277, 31994-32002. (Reproduced by permission, obtained through the Copyright Clearance Center, of The American Society for Biochemistry & Molecular Biology.)

PLDα can act on PC, PE, and PG [17], but in vivo, only the decrease in PC levels was greater in wild-type than in PLD1-deficient plants, indicating that PC is PLD1’s main in vivo substrate. Consistent with this conclusion, the molecular species of the PA formed during freezing were similar to those of PC, except 34:6PA, which is most likely to come from a different reaction as discussed earlier.

DESATURASES

AND

DHAP REDUCTASE

Plastid and the endoplasmic reticulum (ER) are the major sites of plant complex lipid synthesis. However, de novo fatty acid synthesis occurs solely in the plastid, resulting in the synthesis of 16:0 (palmitic acid) and 18:0 (stearic acid) bound to acyl carrier protein (ACP). 18:0-ACP is further converted to 18:1-ACP by stearoylACP desaturases, primarily the enzyme encoded by the FAB2 (also known as SSI2) gene. These acyl chains are either incorporated into complex lipids via the prokaryotic pathway in the plastid, or they are exported as CoA-thioesters into the cytoplasm and incorporated in complex lipids via the eukaryotic pathway. An abbreviated scheme of lipid biosynthesis in Arabidopsis is shown in Figure 14.3. Desaturation

High-Throughput Lipid Profiling to Identify and Characterize Genes

3

A

B

PI

PI

315

2 1 12 4

Lipid/dry weight (nmol/mg)

HH PG

L

L L

L

8 L L

4 PE 3 2 1

L

L

L

L

L

L L

L

L

L L

L L

L

L

H

2 H

L

L L

H

H H

H

L L

L LL

L L L H

H

L

L

L

L L

L

L

L PC

PA

1

L PE

L

8 PC 6 4 2 3

HH

H PG

PA H HH

H MGDG

60 MGDG

H H

HHH

L

40 20 30 DGDG

DGDG

20

32: 0 32: 1 34: 0 34: 1 34: 2 34: 3 34: 4 34: 5 34: 6 36: 1 36: 2 36: 3 36: 4 36: 5 36: 6 38: 2 38: 3 38: 4 38: 5 38: 6 40: 2 40: 3 42: 3 42: 4

H 32: 0 32: 1 34: 0 34: 1 34: 2 34: 3 34: 4 34: 5 34: 6 36: 1 36: 2 36: 3 36: 4 36: 5 36: 6 38: 2 38: 3 38: 4 38: 5 38: 6 40: 2 40: 3 42: 2 42: 3 42: 4

10

Lipid molecular species (total acyl carbons: double bonds)

FIGURE 14.3 Abbreviated scheme of glycerolipid synthesis in leaves of Arabidopsis. (a) Glycerolipid synthesis in endoplasmic reticulum (eukaryotic pathway) and plastid (prokaryotic pathway). DHAP, dihyroxyacetonephosphate; DAG, diacylglycerol; G3P, glycerol-3phosphate; LPA, lysophosphatidic acid; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PG, phosphatidylglycerol; DGDG,digalactosyldiacylglycerol; MGDG, monogalactosyldiacylglycerol; SQDG, sulfoquinovosyldiacylglycerol. (b) Desaturation of 16C and 18C acyl chains in plastid. In A and B, the blue broken lines indicate the steps affected in the various mutants.

of 16:0 and 18:1 species, which is catalyzed by membrane-bound desaturases, occurs in complex lipids. Desaturation of 18:1 to yield complex lipids with 18:2 and 18:3 occurs in the plastid and the ER. In the plastids, the FAD6-encoded desaturase catalyzes the conversion of 16:1 and 18:1 to 16:2 and 18:2, respectively, whereas the FAD7- and FAD8-encoded desaturases catalyze the conversion of 16:2 and 18:2 to 16:3 and 18:3, respectively. We have shown that a loss-of-function mutation in the SSI2 (FAB2) gene results in an overall reduction in the complex lipid content in leaves of the ssi2 mutant plant [18]. This reduction in the lipid content in the ssi2 mutant plant is essentially due to lowered levels of lipids that are synthesized and localized in the plastids. However, considerable differences in the composition of extraplastidic lipids were also observed in the ssi2 mutant plant. For example, ESI-MS/MS analysis showed that the ssi2 mutant contains higher proportion of PC, PE, and PI, containing either a 36:2 or a 36:3 acyl combination. In contrast, the level of 34:2-PC, -PE, and -PI

316

Functional Lipidomics

were lower in the ssi2 mutant plant than the wild-type plant. The ssi2 mutant plant is a dwarf and spontaneously develops lesions containing dead cells. In addition, as described in the following text, multiple plant defense mechanisms are affected in the ssi2 mutant plant. The ssi2-conferred dwarfing and lesion development is suppressed by secondsite mutations in other genes involved in plastid lipid biosynthesis, suggesting that the ssi2-conferred dwarfing and lesion development are due to changes in plastid integrity/function and/or altered lipid-mediated signal transduction. Mutations in the SFD1 (also called GLY1) and FAD6 (also called SFD4) genes suppress the ssi2 phenotypes [18,19]. Presence of the sfd1 allele restored the lipid content and the 18:1 mol% in ssi2-containing plants, whereas the sfd4-allele-bearing plants contain very high levels of 18:1 and lowered content of polyunsaturated 18C acyl group containing lipids. In addition, presence of the sfd1 and sfd4 mutant alleles also results in lowered levels of 16:3-containing lipids. The increases in 18:1 mol% in these plants, or alternatively, the lowered content of 16:3-containing lipids, could account for the suppression of ssi2-conferred phenotypes. SFD1 exhibits extensive homology to bacterial and animal dihydroxyacetone phosphate (DHAP) reductases (also known as glycerol-3-phosphate [G3P] dehydrogenase) [19]. We have shown that SFD1 complements the DHAP reductase deficiency of the Escherichia coli gpsA mutant [19]. DHAP reductases catalyze the interconversion of DHAP and G3P. In plants, the bulk of G3P, which provides the glycerol backbone of glycerolipids, is synthesized from DHAP. The major impact of the sfd1 mutation is on the composition of plastid-synthesized and localized galactolipids [19], suggesting that SFD1 is required for plastid lipid biosynthesis. Arabidopsis has at least three other genes that encode putative proteins with homology to DHAP reductases. The At5g40610 gene encodes a protein that localizes to the plastids and complements the E. coli gpsA mutant. Similar to the sfd1 mutation, a T-DNA insertion in the At5g40610 gene affects plastid lipid composition, in particular, that of the galactolipid, 34:6MGDG (K. Krothapalli, M. Roth, R. Welti, and J. Shah, unpublished), suggesting an overlap in the biochemical function of SFD1 and the At5g40610-encoded protein.

IDENTIFICATION OF GENES INVOLVED IN LIPID SIGNALING GENES INVOLVED

IN

PLANT DEFENSE RESPONSES

The ssi2 mutant was identified in a screen for mutants with altered defense responses [20]. Resistance to multiple pathogens and insects is altered in the ssi2 mutant plant. For example, the ssi2 mutant exhibits enhanced resistance to the bacterial pathogen, Pseudomonas syringae (the oomycete), Peronospora parasitica (cucumber mosaic virus [CMV]), and the generalist aphid, Myzus persicae (green peach aphid) [20,21, V. Pegadaraju and J. Shah, unpublished). The heightened resistance of the ssi2 mutant plant to P. syringae and P. parasitica is primarily due to the constitutive activation of salicylic acid (SA) signaling [18,20]. In contrast to the biotrophic pathogens, the ssi2 mutant exhibits enhanced susceptibility to the necrotrophic fungus, Botrytis

High-Throughput Lipid Profiling to Identify and Characterize Genes

317

cinerea [18,22]. The ssi2-conferred susceptibility to B. cinerea correlates with the inability of the ssi2 mutant to activate jasmonic acid (JA)-dependent defense pathways; especially those associated with the expression of the defensin gene, PDF1.2. Resistance to B. cinerea is restored in ssi2 plants, in which PDF1.2 is constitutively expressed due to presence of the eds5 and pad4 mutant alleles (A. Nandi and J. Shah, unpublished). Simultaneously, the ssi2-conferred resistance to Pseudomonas syringae is compromised in the ssi2 eds5 and ssi2 pad4 double mutant plants. SAand JA-signaling pathways are known to antagonize each other and thus fine tune the plant defense responses. We are currently testing the hypothesis that an SSI2generated/derived lipid molecule is involved in modulating cross talk in plant defense signaling. The sfd1 mutant compromises the ssi2-conferred resistance to biotrophic pathogens. Furthermore, the sfd1 mutant plant is compromised in the activation of systemic acquired resistance, an inducible defense mechanism that confers resistance to a broad spectrum of pathogens [19]. Thus, our studies with ssi2 and sfd1 implicate an important role for lipids in plant defense signaling, especially those containing unsaturated fatty acids. SFD1 and SFD4 provide excellent examples of how metabolic profiling can expedite the identification and cloning of genes affecting biochemical processes. Lipid profiling of the ssi2 sfd1 and ssi2 sfd4 mutant plants suggested that the SFD1 and SFD4 genes were involved in plastid lipid biosynthesis [18]. In case of the sfd4 mutant allele, we could predict that the mutation affects activity of a ω6-desaturase that is involved in the desaturation of 16:1 and 18:1 to 16:2 and 18:2, respectively, in plastid-localized galactolipids. The FAD6-encoded desaturase was predicted to be the enzyme whose activity was altered in the sfd4 mutant. Genetic mapping placed the sfd4 mutation in the vicinity of the FAD6 locus. Furthermore, like the sfd4 mutant, a previously identified fad6 mutation suppressed the ssi2-conferred phenotypes. Finally, sequencing of the FAD6 locus from the sfd4 mutant plant confirmed that the sfd4 phenotype was due to a mutation in the FAD6 gene. Similarly, a survey of approximately 300 genes in the chromosomal location to which SFD1 mapped identified a gene (At2g40690) that was predicted to encode a putative DHAP reductase as the most likely candidate involved in lipid biosynthesis. Glycerol application could overcome the sfd1 defect, strongly suggesting that sfd1 contained a mutation in At2g40690. Sequencing of the At2g40690 locus from the sfd1-1 and sfd1-2 mutant plants confirmed that both contained mutations in At2g40690. Finally, genetic complementation and studies of the SFD1 gene in E. coli confirmed the identity of SFD1.

ANALYSIS OF LIPID CHANGES IN PATHOLOGICAL CONDITIONS SPHINGOLIPIDS

IN

GLOBOID CELL LEUKODYSTROPHY

Globoid cell leukodystrophy, which is also known as Krabbe disease, is an autosomal recessive disease due to mutations in the lysosomal enzyme galactosylceramidase. This enzyme removes galactose from galactosylceramide (fatty acid + sphingosine + galactose) and psychosine (sphingosine + galactose). Galactosylceramide is a major lipid constituent of myelin in both the central and peripheral nervous systems, whereas

318

Functional Lipidomics

psychosine is usually produced in trace amounts and is not thought to serve a productive physiological role. During disease, the myelin-forming cells for both the central and peripheral nervous systems, oligodendrocytes and Schwann cells, respectively, show pathological features due to the inability to digest galactosylceramide and/or psychosine. Globoid cells, which are multinucleated macrophages filled with lipids (e.g., galactosylceramide and psychosine), accumulate in the central nervous system. The overall concentration of galactosylceramide is thought to decrease from normal to diseased states [23–25] probably due to the loss of myelin, but more importantly, the distribution of this glycolipid is thought to shift from myelin and myelin-forming cells to globoid cells in the central nervous system. The concentration of psychosine in the central nervous system increases from normal to diseased states [24,36], suggesting that it accumulates in myelin-forming cells prior to their demise, and this accumulation of psychosine has been hypothesized to result in toxicity [27]. Earlier studies have characterized the changes in the sphingolipid profile between normal and diseased states using classic lipid methodologies [28,29], but more recently, we have reexamined this matter utilizing mass spectrometry due to the enhanced capacity to analyze lipid species [4]. Using an authentic animal model of globoid cell leukodystrophy, called twitcher mice, we found that there was an overall shift from long-chain fatty acids in galactosylceramide to shorter chains in diseased brains; this shift had not been noted previously in this disease [28,29], but it was observed in other inherited diseases that affect myelin [30,31]. This finding suggests that there is some decrease in the length of the carbon chain, perhaps after the galactosylceramides move from myelin to globoid cell, and/or that there is a reduction in chain elongation in sick oligodendrocytes. This shift to shorter carbon chains was also observed for ceramide in disease animals. Of particular interest was a sharp decline in the level of sphingosine1-phosphate, which is thought to be an important molecule in signaling pathways. We suggest that sphingosine-1-phosphate is enriched within oligodendrocytes, and the loss of oligodendrocytes in this disease leads to an overall decline in sphingosine-1phsophate levels in the central nervous system. It is possible that in future studies, sphingosine-1-phosphate levels may become a marker of myelination and/or demyelination. In human globoid cell leukodystrophy, the clinical course can range from infantonset and early death to adult-onset and survival with disability. In ongoing studies, we are examining the lipid profile of long-surviving twitcher mice [32] compared to short-surviving animals in the hope of gaining potential insights as to whether lipids can influence the disease course. In summary, mass spectrometry has revealed changes in sphingolipid species in globoid cell leukodystrophy that have not been noted in earlier studies using older methodologies. A more complete profile of all lipid species by mass spectrometry could reveal additional information regarding the pathogenesis of this disease at the molecular level.

LIPID CHANGES

IN

PLANT DISEASE

We have observed changes in leaf lipid composition in wild-type Arabidopsis plants as early as 6 h post inoculation with P. syringae strains. At this early stage, pathogen growth is minimal, and disease symptoms are not evident. We are currently studying

High-Throughput Lipid Profiling to Identify and Characterize Genes

319

host lipid compositional changes in response to virulent and avirulent pathogens. When completed, our studies on lipid compositional changes in Arabidopsis challenged with pathogen will provide a large data set, which, along with data on lipid compositional changes in mutants exhibiting altered defense responses, will aid in identifying lipid changes of importance to plant resistance/susceptibility to pathogen and adaptation to biotic stress.

PERSPECTIVES POTENTIAL Lipid profiling, particularly the direct-infusion strategy, provides a large amount of information in a minimal amount of time. The ESI-MS/MS platform has proven to be inclusive for most lipid molecular species. This analysis is quantitative and requires only simple sample preparation and small samples. This strategy is a particularly good fit for identifying lipid changes that are due to mutations and/or physiological treatments of biological samples. As described in this chapter, this approach can be employed to characterize (1) lipid metabolic pathways involved in stress responses and disease processes, (2) roles of specific genes/enzymes in stress responses, and (3) lipid species that serve as substrates and products of specific enzymes. In addition, the lipid-profiling information can be used for gene discovery by helping identify genes in mutants. Combination of the profiling strategy with genetic and physiological manipulations has the potential to generate unprecedented information about lipid complexity and its regulation and roles in cellular metabolism and function.

CHALLENGES Lipid profiling is an emerging technology, and much is needed to be overcome to achieve true comprehensiveness and robustness of the analysis. The direct infusion of lipid extracts into ESI-MS/MS presents a straightforward platform for highthroughput analysis. But the dynamic range of detection may limit the one-injection analysis of trace lipid species. It is likely that more than one platform, such as GCMS and/or coupling with liquid chromatography for trace compounds and oxylipins, will be needed for full-profiling lipid species. However, it should be possible to extend the ESI-MS/MS platform so that routine lipid profiling will cover additional lipids, such as sphingolipids, phosphoinositides, N-acyl PEs, and free fatty acids. At the present stage, ESI-MS/MS-based lipid profiling has been adapted to highthroughput analysis, and it is a targeted strategy. Thus, such analysis would likely not detect the presence of lipids in untargeted classes. Further development is needed to uncover new lipid classes that can be targeted for analysis. In addition, the short ESI-MS/MS time, coupled with an autosampler at the front end, generates huge amounts of mass spectral data. Data handling is often the slowest step in metabolic profiling. Development of better informatics will be critical to the full realization of the potential and power of lipid profiling.

320

Functional Lipidomics

ACKNOWLEGMENTS Grant support from National Science Foundation (MCB 0110979 and MCB0455318) and support of the Kansas Lipidomics Research Center Analytical library from National Science Foundation’s EPSCoR program, under grant EPS-0236913 with matching support from the State of Kansas through Kansas Technology Enterprise Corporation and Kansas State University are gratefully acknowledged, as is Core Facility Support from NIH grant P20 RR016475 from the INBRE program of the National Center for Research Resources. Jyoti Shah and Xuemin Wang would also like to acknowledge grand support (Agreement# 2002-3519-11655 and 200535308-05253, respectively) from the Unites States Department of Agriculture.

References 1. Brügger, B., Erben, G., Sandhoff, R., Wieland, F.T., and Lehmann, W.D. Quantitative analysis of biological membrane lipids at the low picomole level by nano-electrospray ionization tandem mass spectrometry. Proc. Natl. Acad. Sci. USA 1997, 94, 2339–2344. 2. Welti, R., Li, W., Li, M., Sang, Y., Biesiada, H., Zhou, H-E., Rajashekar, C.B., Williams, T.D., and Wang. X. Profiling membrane lipids in plant stress responses: role of phospholipase Dα in freezing-induced lipid changes in Arabidopsis. J. Biol. Chem. 2002, 277, 31994–32002. 3. Welti, R., Wang, X., and Williams, T.D. Electrospray ionization tandem mass spectrometry scan modes for plant chloroplast lipids. Anal. Biochem. 2003, 314, 149–152. 4. Esch, S.W., Williams, T.D., Biswas, S., Chakrabarty, A., and LeVine, S.M. Sphingolipid profile in the CNS of the Twitcher (globoid cell leukodystrophy) mouse: A lipidomics approach. Cell. Mol. Biol. 2003, 49, 779–787. 5. Leibisch, G., Drobnik, W., Reil, M., Trümbach, B., Arnecke, R., Olgenmöller, B., Roscher, A., and Schmitz, G. Quantitative measurement of different ceramide species from crude cellular extracts by electrospray ionization tandem mass spectrometry (ESI-MS/MS). J. Lipid Res. 1999, 40, 1539–1546. 6. Whitfield, P.D., Sharp, P.C., Taylor, R., and Meikle, P. Quantification of galactosylsphingosine in the twitcher mouse using electrospray ionization-tandem mass spectrometry. J. Lipid Res. 2001, 42, 2092–2095. 7. Colsch, B., Afonso, C., Popa, I., Portoukalian, J., Fournier, F., Tabet, J.-C., and Baumann, N. Characterization of the ceramide moieties of sphingoglycolipids from mouse brain by ESI-MS/MS: identification of ceramides containing sphingadienine. J. Lipid Res. 2004, 45, 281–286. 8. Hsu, F.-F. and Turk, J. Structural determination of sphingomyelin by tandem mass spectrometry with electrospray ionization. J. Am. Soc. Mass Spec. 2000, 11, 437–449. 9. Bligh, E.G. and Dyer, W.J. A rapid method of total lipid extraction and purification. Can. J. Biochem. Physiol. 1959, 37, 911–917. 10. Wang, X. Plant phospholipases. Annu. Rev. Plant Physiol. Plant Mol. Biol. 2001, 52, 211–231. 11. Wang, X. Lipid signaling. Curr. Opin. Plant Biol. 2004, 7, 329–336. 12. Qin, C. and Wang, X. The Arabidopsis phospholipase D family. Characterization of a calcium-independent and phosphatidylcholine-selective PLD zeta 1 with distinct regulatory domains. Plant Physiol. 2002, 128, 1057–1068.

High-Throughput Lipid Profiling to Identify and Characterize Genes

321

13. Wang, X. Phospholipase D in hormonal and stress signaling. Curr. Opin. Plant Biol. 2002, 5, 408–414. 14. Li, W., Li, M., Zhang, W., Welti, R., and Wang X. The plasma membrane-bound phospholipase D enhances freezing tolerance in Arabidopsis. Nature Biotech. 2004, 22, 427–433. 15. Sang, Y., Cui, D., and Wang, X. Phospholipase D and phosphatidic acid-mediated generation of superoxide in Arabidopsis. Plant Physiol. 2001, 126, 1449–1458. 16. Zhang, W., Wang, C., Qin, C., Wood, T., Olafsdottir, G., Welti, R., and Wang, X. The oleate-stimulated phospholipase D, PLDδ, and phosphatidic acid decrease H2O2induced cell death in Arabidopsis. Plant Cell 2003, 15, 2285–2295. 17. Pappan, K., Austin-Brown, S., Chapman, K.D., and Wang, X. Substrate selectivities and lipid modulation of plant phospholipase D, -, and -. Arch. Biochem. Biophys. 1998, 353, 131–140. 18. Nandi, A., Krothapalli, K., Buseman, C., Li, M., Welti, R., Enyedi, A., and Shah, J. The Arabidopsis thaliana sfd mutants affect plastidic lipid composition and suppress dwarfing, cell death and the enhanced disease resistance phenotypes resulting from the deficiency of a fatty acid desaturase. Plant Cell 2003, 15, 2383–2398. 19. Nandi, A., Welti, R., and Shah, J. The Arabidopsis thaliana dihydroxyacetone phosphate reductases gene suppressor of fatty acid desaturase deficiency1 is required for glycerolipid metabolism and for the activation of systemic acquired resistance. Plant Cell 2004, 16, 465ñ477. 20. Shah, J., Kachroo, P.K., Nandi, A., and Klessig, D.F. A recessive mutation in the Arabidopsis SSI2 gene confers SA- and NPR1-independent expression of PR genes and resistance against bacterial and oomycete pathogens. Plant J. 2001, 25, 563–574. 21. Sekine, K.T., Nandi, A., Ishihara, T., Hase, S., Ikegami, M., Shah, J., and Takahashi, H. Enhanced resistance to cucumber mosaic virus in the Arabidopsis thaliana ssi2 mutant is mediated via an SA-independent mechanism. Mol. Plant-Microbe Interaction 2004, 17, 623–632. 22. Kachroo, P, Shanklin, J., Shah, J., Whittle, E.J., Klessig, D.F. A fatty acid desaturase modulates the activation of defense signaling pathways in plants. Proc. Natl. Acad. Sci. USA 2001, 98, 9448–9453. 23. Vanier, M.T., and Svennerholm, L. Chemical pathology of Krabbe’s disease. III. Ceramide-hexosides and gangliosides of brain. Acta Paediatr. Scand. 1975, 64, 641–648. 24. Svennerholm, L., Vanier, M.T., and Mansson, J.E. Krabbe disease: a galactosylsphingosine (psychosine) lipidosis. J. Lipid Res. 1980, 21, 53–64. 25. gisu, H., Shimomura, K., Kishimoto, Y., and Suzuki, K. Lipids of developing brain of twitcher mouse. An authentic murine model of human Krabbe disease. Brain 1983, 106, 405–417. 26. Igisu, H. and Suzuki, K. Progressive accumulation of toxic metabolite in a genetic leukodystrophy. Science 1984, 224, 753–755. 27. Miyatake, T. and Suzuki, K. Globoid cell leukodystrophy: additional deficiency of psychosine galactosidase. Biochem. Biophys. Res. Commun. 1972, 48, 539–543. 28. Eto, Y., Suzuki, K., and Suzuki, K. Globoid cell leukodystrophy (Krabbe’s disease): isolation of myelin with normal glycolipid composition. J. Lipid Res. 1970, 11, 473–479. 29. Vanier, M.T. and Svennerholm, L. Chemical pathology of Krabbe’s disease. II. Fatty acid composition of cerebrosides, sulfatides and sphingomyelins in brain. Acta Paediatr. Scand. 1974, 63, 501–506.

322

Functional Lipidomics

30. Norton, W.T. and Poduslo, S.E., Biochemical studies of metachromatic leukodystrophy in three siblings. Acta Neuropathol. (Berl.) 1982, 57, 188–196. 31. Taketomi, T., Hara, A., Kitazawa, N., Takada, K., and Nakamura, H. An adult case of adrenoleukodystrophy with features of olivo-ponto-cerebellar atrophy: II. Lipid biochemical studies. Jpn. J. Exp. Med. 1987, 57, 59–70. 32. Biswas, S., Biesiada, H., Williams, T.D., and LeVine, S.M. Delayed clinical and pathological signs in twitcher (globoid cell leukodystrophy) mice on a C57BL/6 x CAST/Ei background. Neurobiol. Dis. 2002, 10, 344–357. 33. Comfurius, P., and Zwaal, R.F. The enzymatic synthesis of phosphatidylserine and purification by CM-cellulose column chromatography. Biochim. Biophys. Acta. 1977, 488, 36–42.

Index A Acyl, 10, 135, 203, 315 N-, 59, 69, 154, 319 BZDC-, 193 carrier protein, 314 chain, 105, 172, 190, 193, 202, 219, 226, 244–245, 254–255, 257, 292, 308, 313–315 CoA, 80, 90, 287 composition, 11, 257 fatty, 9, 34, 37, 105, 149, 255, 257 group, 31, 59, 316 labeled, 193 linked, 192, 255 lysolipid, 288 modified, 193–194, 203 moieties, 290, 292, 300 substitute, 31, 34, 37 substructure, 51 tethered, 193 transfer, 105 Affinity, 23, 62, 65, 67, 81, 86, 102, 104, 150, 163, 180, 183, 193, 197, 202–203, 219– 220, 223–224, 228–229, 234, 244, 249 based, 218–219, 234 captured protein, 230 chromatography, 221, 223, 231–233, 249 interaction, 224 isolated, 234 isolation, 221, 226, 234 lipid, 220 LPA-, 109 matrices, 193, 229, 234 matrix, 231 probes, 191, 194, 221, 223 proton, 23 resin, 192 separation, 235 Akt, 150–151, 190, 195, 211, 213, 216, 219, 228, 229, 230, 234 Alzheimer’s Disease, 5, 89, 285–286 Anandamide, 57, 59, 63, 66, 69, 79 Arabidopsis, 308, 310, 311, 313–316, 318, 319

Arachidonic acid, 7, 29, 30, 58–60, 62–64, 68, 79–82, 84–89, 91–92, 154, 244–245, 288 2-Arachidonoylglycerol, 59 Assay(s), 14, 25, 65, 112, 115, 191, 197–200, 202–203, 212, 219–220, 225, 230–231, 233, 243, 250–251, 257 amplified luminescent proximity homogenous (ALPHA), 197 binding, 202 cell-based, 189, 197, 200, 234 enzymatic, 222 enzyme, 249 fluorescence polarization (FP), 192, 197–198 FRET, 257 HTS (high throughput screening), 198 in vitro, 189, 197 inositol phosphate, 200 kinase, 202, 257 lipid, 189, 197, 200 malachite-green-based, 257 PI 3-K, 197–199 radioactive, 198 receptor-based, 244

B Bilayer, 9–10, 64, 220, 225, 244, 257 lipid, 193–194 membrane, 66, 150, 287 Bioinformatics, 3–4, 6, 92, 219 Biomarker, 126, 257, 297–298 1,2-Bistetradecanoyl glycerophosphoglycerol, 40 Bruton’s tyrosine kinase 1 (Btk1), 190, see also Btk1

C Calcimycin, 51–52 Cancer, 5, 83, 87, 89, 115–116, 127, 190, 212, 218, 245 colorectal, 214 oral, 87 ovarian, 101–102, 104–109, 111–116, 133, 136–137 uterine, 193

323

324

Functional Lipidomics

Cannabinoid(s), 60, 68 receptor(s), 57, 59, 60, 61–64, 66–67 receptor-active, 62 receptor-inactive, 67–68 Ceramide(s), 11, 14, 41, 43, 58, 79, 149–154, 156–157, 161–162, 167, 172, 174, 176, 181, 184–185, 271, 290, 310–311, 318 N-acyl, 154 di-hydro, 161 kinase, 149, 151, 153 moiety, 171 nonhydroxylated, 312 -1-phosphate (-1-P), 11, 149 150, 153, see also Ceramide-1-phosphate phosphoethanolamines, 161, 164 phosphoinositols, 161 synthase, 149 Ceramide-1-phosphate, 11, 149, 150, 153 Cholesterol, 7, 12, 45–46, 105, 127, 287, 290, 295 biosynthetic pathway, 12, 45 catabolism, 12 esters, 295 Cholesteryl oleate, 46 Cholic acid, 47–48 Chromatography, 22, 92, 133, 183, 244, 288–290, 308 affinity, 221, 249 cartridge, 248 gas, 9, 92, 126, 129, 135, 247, 258, 264 gel filtration, 223 ion exchange, 249 lipid-affinity, 231–233 liquid, 11, 17, 92–93, 125–126, 129, 133, 159, 202, 219, 243, 249, 258, 288, 319 mass spectrometry (LC/MS), 2, 21 normal-phase, 159, 167, 182, 256 reverse-phase, 159, 166, 178, 180, 182 size exclusion, 231 strong anion exchange, 249 thin-layer, 125–126, 129, 162, 243, 248, 256, 258, 264, 294 Collision induced decomposition, 27 Collision induced dissociation, 170, 251, 258, 265 COX-1, 80, 82 COX-2, 80, 82, 87 Cyclooxygenase(s), 61, 79, 81–82, 88–89, 91

lipase, 59, 62 kinase, 245 DEAE, 249, 258 Derivatization, 28, 92, 136, 247, 249, 264 Diabetes, 9, 83, 89, 190, 285, 289, 298–300 Diacylglycerol, 8, 105, 152, 154, 192, 194, 212, 258, 264, 313, 315, see also DAG 1,2-Dihexadecanoyl glycerol, 33 1,2-Dioleoyl-glycerophosphoinositol, 42 Direct infusion, 289, 309–312, 319 Displacement, 243, 247, 250–251

D

F

DAG (Diacylglycerol), 7, 9–11, 33, 62, 149, 151, 154, 192, 194, 200, 216, 244–245, 256, 258, 264, 269, 277–288, 290, 314–315, see also Diacylglycerol

Fatty acid(s), 1–2, 5–11, 17, 23, 29, 31, 34–35, 41, 81, 127, 129, 134, 135, 142, 154, 157, 160–161, 174–175, 185–186, 204, 244–245, 247–248, 251–255,

E Eicosanoid(s), 1, 2, 5–8, 10, 28, 58, 62, 79–81, 83, 87, 89–93, 157, 244, 288 biosynthesis, 8, 81 biosynthetic, 8, 92 cascade, 8 lipidomics, 79 metabolism, 79, 90 metabolites, 91 production, 5, 81, 153 receptors, 88 synthesis, 81 Electrospray, 21–22, 24, 28 droplets, 21, 23 ionization, 9, 11, 20, 92, 126, 129, 135, 187, 244, 251–252, 255, 258, 282, 285–286, 300, 308 mass spectra, 17, 28 needle, 21 process, 21 source, 282 tube, 132 Endoplasmic reticulum (ER), 157, 302, 314–315 5,6-Epoxycholesterol, 45–47 Extraction, 129, 131, 134, 137, 159, 183–184, 197, 243, 246, 247–248, 253, 263, 282, 309 lipid, 125, 129–130, 135, 137, 139, 141 methods, 131 organic, 4 solid-phase, 125, 134 solvent, 131

Index

325

257–258, 264–268, 287, 300, 313, 317–319 acyl chain, 172 amide(s), 7, 60, 67–68 biosynthetic, 12 chains, 135–136, 138, 140, 180 derivatives, 135 hydrolysis, 135 hydroperoxide, 34 metabolites, 8 polyunsaturated (PUFA), 35, 244, 258 side chains, 135 synthesis, 314 unsaturated, 317 Fluorescence, 115, 199, 243, 247, 250, 257 polarization (FP), 192, 197 quenching, 250 resonance energy transfer (FRET), 193, 250, 258 total internal reflectance (TIRF), 197

G Gas chromatography, see chromatography Galactolipid(s), 313–314, 316–317 Ganglioside(s), 11–12, 149, 162, 168–169, 173–175, 185 Genomics, 2-4, 6, 147–148, 217–218, 289 Glucosylceramide(s), 11, 149, 168, 172, 184 Glycerolipid(s), 17, 31, 152–154, 157, 264, 316 bilayer, 9 biosynthesis, 10 synthesis, 315 Glycerophospholipid(s), 7–8, 10, 17, 34, 80, 263–265, 269–270, 272, 276, 278, 280, 290, 295 Glycolipid(s), 2, 6, 10, 13, 127, 157, 287, 290, 292, 318 Glycosphingolipid(s), 1, 2, 6–7, 11–12, 149–150, 161–162, 167–168, 173, 293

H Head group, 10–11, 160–161, 171–172, 174, 181, 183, 191–194, 226, 244, 247, 250–252, 256, 266, 308, 310, 313 carbohydrate, 181 derivatives, 257 inositol, 189, 223, 243–245, 248, 252, 254 isomers, 248, 249, 257 glycerophospholipid, 265 PC, 266

phosphoethanolamine, 181, 185, 266 phosphor, 172 phosphoryl choline, 185 polar, 34 sugar, 171–172 1-Hexadecanoyl-2-arachidonoylglycerophosphocholine, 24 1-Hexadecanoyl glycerol, 33 1-Hexadecanoyl-2-oleoylglycerophosphoethanolamine, 36 1-O-Hexadec-1’-enyl-2-docosahexaenoylglycerophosphoethanolamine, 36 High performance liquid chromatography (HPLC), 17, 20–22, 25, 27–29, 92, 125–126, 129, 133–134, 141, 162–167, 171, 177–182, 184, 243, 247, 249, 256–258, 282, 288–289, 308 HPLC, 17, 20–22, 25, 27–29, 92, 125–126, 129, 133–134, 141, 162–163, 171, 177–179, 182, 184, 247, 249, 257–258, 282, 288 column(s), 164–166, 180, 256, 289, MS/MS, 167, 179, 181, 184 reverse phase, 308 Hydroxyeicosatetraenoic acid(s), 29, 30, 80, 87 4-Hydroxysphinganine, 161

I Immobilized, 189, 202–204, 220 Inositol(s), 212–213, 245, 248, 251, 253, 257 Myo-, 248, 249 head group(s), 189, 223, 243–245, 248, 250, 252–254 lipids, 212, 245 (1,3,4,5)P4 (IP4), 250 1-phosphate, 212 4-phosphate, 245 phosphate(s), 200, 228, 250 phospholipids, 244, 246, 251, 253, 257 polyphosphate(s), 245, 250 ring, 192, 212–213, 253 triphosphate, 105, 214 Inositolpolyphosphates (IPs), 245, 258 Intrasource separation, 285–286, 289–290, 300 Invasion, 8, 103, 106, 113, 246 Isoprenoid(s), 1–2, 4, 6–7, 12–13, 17, 47, 49 13C Isotopomer, 292–294

326

Functional Lipidomics

K Kinase(s), 89, 150–151, 189, 197–198, 200, 213–215, 217, 220, 225–227, 229–234, 257 Bruton’s tyrosine (Btk1), 190, 228 ceramide, 149, 151, 153 DAG, 245 Erk, 114 inositol phospholipids, 257 integrin-linked (ILK), 234 lipid, 194, 199, 212, 243 mitogen-activated protein, 58, 60, 81 myristoylated-alanine-rich (MARKs), 193 P21-activated (Pak2), 227, 234 phosphoinositide (PIP-Ks), 190, 232–233, 245 phosphoinositide-dependent protein (PDK1), 190, 211, 219, 228–229 phosphotidylinositol 3- (PI 3-K), 10, 60, 105, 195, 198–199, 212 PI, 226, 243, 246, 251, 256–257 protein,105, 151, 198, 202, 212, 227, 230, 234 A, 105, 152 B, 60, 190 C, 62, 66, 83, 105, 150–152, 219, 227, 229 receptor tyrosine (RTK), 213 Rho, 195 ribosomal S6 (RSK), 219 sphingosine, 149–150, 152–153 sphingosine 1-, 58 Tec, 213, 234 tyrosine, 189, 190, 214, 227 Kdo2-lipid A, 51, 53

L Lactosylceramide(s), 11, 149, 162, 167, 172, 184 Leukotriene(s), 7, 80, 84–86, 92 leukotriene B4 (LTB4), 29, 32 leukotriene C4 (LTC4), 84–85 leukotriene D4 (LTD4), 86 leukotriene E4 (LTE4), 86 Ligand(s), 192–193, 199–200, 212, 214, 216, 220, 223–224, 230, 235, 249, 264–265, 269, 276 anti-IgM, 276 of cannabinoid receptors, 65 cytoplasmic, 65 for cytoplasmic proteins, 244 GPCR, 200 for liver X, 12 PIP-affinity, 232 PIPn (PIP), 191, 231, 232

receptor activation, 234 receptor binding, 213, 215 SS-PIP3, 234 Lipid arrays, 10, 203, 263, 273–277 Lipidomics, 1–3, 5–8, 10, 13–14, 102, 107, 147–148, 155–157, 202–205, 286–287, 289, 297, 299–300, 308, 320 eicosanoid, 79 shotgun, 285, 289, 293, 298, 300 structural, 1, 6, 13, 14 Lipoxygenase(s), 61, 79–81, 84, 86–88 5- (5-LO), 61, 84–85 12- (12-LO), 86, 91 15- (15-LO), 62, 84, 86–87 Liquid chromatography, see Chromatography LPA (lysophosphatidic acid), 58, 101–109, 111–116, 127–133, 136–138, 245, 264, 266–268, 270, 280–281, 314–315 LPC (lysophosphatidylcholine), 105, 127, 129, 131–136, 138, 140, 270–271, 278–279 Lysophosphatidic acid(s) (LPA), 58, 101–102, 127, 245, 264, 315 Lysophosphatidylinositol(s) (LPI), 131, 266 Lysophosphatidylcholine (LPC), 105, 127, 294, 311 Lysophospholipids (lyso-PLs), 7–8, 79, 104, 106, 125–126, 157, 264, 276

M Mass spectrometry, 2, 18, 51, 53, 92, 127, 148, 163, 202, 204, 218, 243, 244, 249, 253, 263, 269, 273–274, 282, 288–289, 308, 318 electrospray ionization (ESI), 53, 92, 126, 129, 244, 252, 255, 258, 285–286, 300 fast atom bombardment (FAB), 251, 265, 288 gas chromatography, 258, 264 liquid chromatography, 21, 92 liquid secondary ionization, 163 SELDI, 223 tandem, 11, 20, 27–29, 43, 52, 92, 159–161, 163, 165, 202, 251–253, 265, 291–292, 295–296, 299, 308 time of flight, 18, 135 Metastasis, 101, 105–106, 112–113 Microarray, 146, 155, 204, 222 Motility, 103–104, 112, 115, 213, 216, 232 Multiple reaction monitoring (MRM), 11, 27, 160, 309

Index

327

N Neovascularization, 103, 106 Neutral loss scan(s), 28, 159, 165, 170, 177, 185, 308–309

O 1-Octadecanoyl-2-arachidonoyl glycerophosphate, 35 1-Octadecanoyl-2-oleoyl-glycerophosphoserine, 39 Overlay, 203, 220, 233, 234

P PDK1, 190, 213, 216, 219, 228, 230, 234 PH domain(s), 190–191, 194, 200, 203, 211, 219, 222–224, 227–230, 234, 251 Phorbol 12-myristate-13-acetate, 50 Phosphatase(s), 150–151, 189–190, 194, 199–200, 215, 217, 229, 232–233, 243, 245–246 activities, 243, 246, 251, 256–257 inositol polyphosphate, 245 lipid, 10 3-phosphatase, 190 phosphatidic acid (PAP), 245 PI, 257 PI 5-, 245 PIP, 226 PIPn 5- (SHIP2), 190 polyphosphoinositide, 245 protein, 150–152, 156, 257 S-1-P, 149 type 1 Inositol 4-, 245 Phosphatidylcholine(s) (PC, PtdCho), 9, 35, 58, 105, 131, 202, 246, 265, 294, 308, 311, 315 Phosphatidylethanolamine(s) (PE, PtdEtn), 9, 34, 59, 131, 194, 202, 246, 265, 293–294, 299, 308, 311, 315 Phosphatidylinositol(s), 9, 25, 41, 154, 190, 212, 243, 151, 253–255, 258, 264, 266, 275–276, 290, 308, 311, 315 bisphosphate, 62, 81, 244–245 3-kinase, 60, 105, 190 phosphate, 244, 226 Phosphatidylinositol 3-kinase, 60, 105, 190, see also PI 3-K Phosphatidylserine(s) (PS, PtdSer), 9, 37, 131, 246, 251, 266, 290, 308, 311

Phosphoinositide(s), 59, 62, 190–193, 195–204, 211–213, 216, 218–219, 221, 226, 228–230, 232, 234–235, 243, 258, 319 analogs, 204, 221 binding, 191, 202, 211, 215, 226–231 biotylated, 231 kinase, 232 3-kinase, 212 metabolism, 254 4,5-phosphate, 58 phosphorylated, 196, 212–213, 232 probes, 189, 191, 194, 197, 199 profiling, 243 signaling, 189, 204, 211, 214, 216, 230, 233–234 signal transduction, 211–212, 214, 216 Phospholipase(s), 81, 189, 199, 288, 298, 307, 312–313 A, 245 A1, 105, 313 A2, 7, 58, 62, 80–81, 150–151, 244, 313 C, 59, 84, 105, 200, 244–245, 264, 313 D, 59, 66, 228, 245, 309, 313–314 Phytoceramide, 161, 162 Phytosphingosine, 166, 179 Phytosphingosine-1-Phosphate, 166, 179 PI 3-K, 190–191, 197–199, 202, 205 PI 3-kinase, 195, 198–199 Plasmalogen(s), 7, 9, 35–36, 272, 288, 298, 300 Polyketide(s), 17, 51, 52 Precursor ion scan, 159, 165, 175, 177, 254 Prostaglandin(s), 29, 61–62, 67, 80, 82–84, 87, 89, 91–92 A, 91 D (PGD), 82–84 E (PGE), 32, 82–83, 162, ethanolamines, 61 F (PGF), 62, 82–84, 91 G (PGG), 82 glyceryl esters, 61 H (PGH), 82, 87 I (PGI), 82 J, 91 pathway, 82 receptors, 61 synthases, 61 transporters, 84 Proteolipidomics, 202, 204 Proteomics, 2–4, 6, 135, 148, 155, 211, 216–219, 221, 224, 226, 230, 289 chemical, 211–212, 226, 230 functional, 211, 235 lipid, 191, 202–203 signaling-, 231–232

328

Functional Lipidomics

PTEN (phosphatase and tensin homologue deleted on chromosome ten), 190, 199, 215, 229, 232, 245

Q Quadrupole, 11–12, 17–18, 20, 27–29, 51, 92, 167, 169–170, 173–175, 186, 253, 262, 265, 269, 277, 282 Qualitative, 9–10, 53, 126, 134, 138, 218, 220–221, 223–225, 250 Quantitative, 25, 92, 136, 186, 218, 224–225, 248, 250, 253, 319 analysis (analyses), 4, 10–11, 27, 53, 126, 132, 134–136, 138–139, 166, 249, 256, 293 assays, 53 data, 160, 163–164, 176–177, 183, 185 determination, 134, 139, 220 information, 176, 186 measure, 89 method, 134, 222 tools, 9 Quantitation, 6, 11, 79–80, 92, 159, 165, 170, 175–177, 180, 182–185, 288, 292–293, 298

R Radioactive, 92, 197–198, 248 Radiolabeling, 8, 248, 250, 256 Rb (Retinoblastoma), 150, 151 Receptor(s), 11–12, 61–63, 65–67, 79, 83–85, 89, 102–104, 107, 116, 150–151, 191, 200–201, 212–216, 223, 227, 234, 243–244, 250, 256, 282 acetylcholine, 63 B-cell, 264, 276 cannabinoid, 57, 59–63, 67–68 CB1, 63–64, 68 CB2, 63, 67, 68 CD1, 246 cell-surface, 9–10, 104, 190, 264 cysLT, 81 Edg, 150 eicosanoid, 88 EP1, 83, EP2, 83 EP3, 83 EP4, 83 Fc, 215 FP, 84 G-protein, 189

G-protein coupled, 9, 58, 83–85, 90, 102, 104, 111, 200, 214, 264 Gq-coupled, 64 heptahelical, 200 IP, 83, 247 IP3, 250 isoprostane, 89 lipoxin-A4-specific, 87, EGF, 227 liver X, 12 5-LO product, 85 low-density lipoprotein (LDL), 297–298 LPA, 101, 102, 104, 105, 107, 109, 114–116 LPA3, 107, 109 LPA-pathway, 116 membrane, 89, 150 muscarinic, 64, 66 nicotinic, 63, 66 noncannabinoid, 62 nuclear, 12, 203 PGE2, 89 PPAR, 61–62, 83, 300 prostaglandin, 61 prostanoid, 62 S1P, 151 S1P1, 58 T-cell antigen, 214 thromboxane, 89 transmembrane, 79, 83, 156 TRPV1, 65, 67 tyrosine kinase, 190, 213–214 tyrosine kinase growth factor, 189 Retinoblastoma (Rb), 150, see also Rb RNAi, 111, 116, 153

S S1P, see Sphingosine-1-Phosphate Saccharolipids, 17, 51 Signaling, 10–13, 65, 83, 102, 104–105, 111, 114, 127, 135, 150, 152–153, 160, 190, 211–213, 216, 218, 221, 227, 232, 234, 244–246, 256–257, 264, 282, 288, 307, 312, 316–317 autocrine, 103 cascade(s), 226–227 cell, 150, 152, 191, 246 complexes, 109, 191, 226, 229 downstream, 213–214, 227 enzymes, 10, 214 events, 105–106, 114, 190 G-protein, 151 insulin, 104, 190–191 lipid(s), 10–11, 152, 204, 212, 216, 265, 269, 307–308, 316

Index lipid-mediated, 154 mechanisms, 79, 93 molecules, 104, 126–127, 131, 151–152, 191, 213, 244 pathways, 11–12, 14–15, 84, 150–151, 183, 189–191, 195–196, 202, 212, 218, 264–265, 317–318 phosphoinositide (PIP), 189, 204, 211, 214–216, 219, 226, 230, 233 PI 3-K, 202, 212, 227, 230 process, 233, 257, 264 protein(s), 213, 226–227, 229, 233 receptor, 191, 214, 227 target(s), 230, 233 transduction, 127, 131 Signal transduction, 9, 105, 107, 116, 131, 150, 204, 211–212, 215–216, 235, 264, 287–288, 293, 312, 316 lipid-, 264 pathways, 216, 264, phosphoinositide, 211–212, 214, 216 Signalosome, 190 SPC (Sphingosylphosphorylcholine), 127, 131, 133 Sphingosine-1-Phosphate, 11, 41, 44, 57, 58, 66, 79, 127, 149, 161, 166, 179, 318 Sphinganine, 149, 160–161, 166, 179 Sphingomyelin, 7, 41, 45, 149–151, 154, 157, 290, 308 Sphingomyelin synthase, 149, 152–153 Sphingomyelinase(s), 14, 150–152 Sphingosine, 9, 11, 41, 58, 149–154, 157, 161–162, 166, 179, 317 backbone, 11, 293 ceramide, 41 derived compounds, 10 3-ketodihydro, 149 kinase, 149, 152–153 moiety, 287 N-octadecenoyl, 43 phosphate, 7, 152 Sphingosylphosphorylcholine (SPC), 127, see also SPC

329 SPR (Surface Plasmon Resonance), 192, 194, 203, 224 Sterol(s), 1–3, 6–7, 12, 17, 45, 127, 134, 157 Sulfatide(s), 149, 168, 197, 285, 290, 293–295, 297–298 Surface Plasmon Resonance (SPR), 192, 224, see also SPR

T Target(s), 58, 62, 66–68, 82, 102, 104, 107, 109, 115–116, 147, 151–153, 156, 190, 197, 218, 220, 223–245, 230, 288, 298, cells, 77 downstream, 224, 230 drug, 212 enzymes, 214 gas, 170, 174 genes, 103 gene therapy, 103 intracellular, 66 lipid, 22, 27 molecular, 51, 58–59, 61, 63, 67, 102, 193, 204 protein(s), 223, 235, 207, 212, 222 signaling, 230, 233 therapeutic, 81, 103, 109, 219 validation 214 Taurocholic acid, 45, 48 Tocopherol succinate, 47, 49 Triacylglycerol, 31, 154 1,2,3-Trihexadecanoyl glycerol, 33

V Vanilloid, 61–62, 65–66, 68

Y Yeast, 8, 103, 150, 155–156, 191, 196, 219, 222, 253, 312