Current Protocols in Chemical Biology 2009 (Volume 1)

CURRENT PROTOCOLS

in Chemical Biology

cp

Current Protocols in Chemical Biology

Online ISBN: 9780470559277 DOI: 10.1002/9780470559277 Editors & Contributors

EDITORIAL BOARD Adam P. Arkin University of California, Berkeley Berkeley, California Lara Mahal New York University New York, New York Floyd Romesberg The Scripps Research Institute La Jolla, California Kavita Shah Purdue University West Lafayette, Indiana Caroline Shamu Harvard Medical School Boston, Massachusetts Craig Thomas NIH Chemical Genomics Center Rockville, Maryland ASSOCIATE EDITORS Michael Burkart University of California, San Diego San Diego, California John Ellman Yale University New Haven, Connecticut Howard Hang The Rockefeller University New York, New York Hans Luecke National Institute of Diabetes and Digestive and Kidney Diseases, NIH Bethesda, Maryland Andreas Marx Universität Konstanz Konstanz, Germany

Michael Rape University of California, Berkeley Berkeley, California Carsten Schultz EMBL Heidelberg Heidelberg, Germany Oliver Seitz Universität zu Berlin Berlin, Germany Katherine L. Seley-Radtke University of Maryland Baltimore, Maryland Nicky Tolliday Broad Institute of Harvard and MIT Cambridge, Massachusetts Gregory A. Weiss University of California, Irvine Irvine, California CONTRIBUTORS Jasmina J. Allen Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Ruben T. Almaraz The Johns Hopkins University Baltimore, Maryland Rogerio Alves de Almeida University of Manchester Manchester, United Kingdom Alma L. Burlingame Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Christopher T. Campbell National Cancer Institute Frederick, Maryland Jennifer Campbell Harvard Medical School Boston, Massachusetts Yong Chi Fred Hutchinson Cancer Research Center Seattle, Washington Bruce E. Clurman Fred Hutchinson Cancer Research Center Seattle, Washington Benjamin F. Cravatt The Scripps Research Institute La Jolla, California Richard D. Cummings Emory University Atlanta, Georgia Arvin C. Dar

Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Jian Du The Johns Hopkins University Baltimore, Maryland Jeremy R. Duvall The Broad Institute of MIT and Harvard Cambridge, Massachusetts Meng Fang University of Georgia Athens, Georgia Matthew Francis University of California, Berkeley Berkeley, California Jeffrey C. Gildersleeve National Cancer Institute Frederick, Maryland Christian Gloeckner University of Konstanz Konstanz, Germany Jay T. Groves Howard Hughes Medical Institute University of California, San Francisco San Francisco, California and National University of Singapore Singapore And Lawrence Berkeley National Laboratory Berkeley, California Howard C. Hang The Rockefeller University New York, New York Rami N. Hannoush Genentech South San Francisco, California Jamie Heimburg-Molinaro Emory University Atlanta, Georgia Nicholas T. Hertz Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Michal Hocek Academy of Sciences of the Czech Republic Prague, Czech Republic Gregory R. Hoffman Harvard Medical School Boston, Massachusetts Eun Ryoung Jang University of Kentucky Lexington, Kentucky Sean Johnston Harvard Medical School Boston, Massachusetts

Hargun S. Khanna The Johns Hopkins University Baltimore, Maryland Kyung Bo Kim University of Kentucky Lexington, Kentucky Ramon Kranaster University of Konstanz Konstanz, Germany Robert D. Kuchta University of Colorado Boulder, Colorado Maya T. Kunkel University of California at San Diego La Jolla, California Wooin Lee University of Kentucky Lexington, Kentucky Jae-Min Lim University of Georgia Athens, Georgia Wan-Chen Lin Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Hana Macickova-Cahová Academy of Sciences of the Czech Republic Prague, Czech Republic Andrew L. MacKinnon University of California San Francisco San Francisco, California Lisa A. Marcaurelle The Broad Institute of MIT and Harvard Cambridge, Massachusetts Gerard Marriott University of California, Berkeley Berkeley, California Andreas Marx University of Konstanz Konstanz, Germany Nathan J. Moerke Harvard Medical School Boston, Massachusetts Nuzhat Motlekar University of Pennsylvania Philadelphia, Pennsylvania Andrew D. Napper University of Pennsylvania Philadelphia, Pennsylvania and University of Manchester Manchester, United Kingdom and Nemours Center for Childhood Cancer Research Wilmington, Delaware

Alexandra C. Newton University of California at San Diego La Jolla, California Takeaki Ozawa The University of Tokyo and Japan Science and Technology Agency Tokyo, Japan Graham D. Pavitt University of Manchester Manchester, United Kingdom Chutima Petchprayoon University of California, Berkeley Berkeley, California Stewart Rudnicki Harvard Medical School Boston, Massachusetts Kevan M. Shokat Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Sharmila Sivendran University of Pennsylvania Philadelphia, Pennsylvania and GlaxoSmithKline Collegeville, Pennsylvania David F. Smith Emory University Atlanta, Georgia Xuezheng Song Emory University Atlanta, Georgia Anna E. Speers The Scripps Research Institute La Jolla, California Elaine Tan The Johns Hopkins University Baltimore, Maryland Jack Taunton University of California San Francisco San Francisco, California Nicola Tolliday The Broad Institute of MIT and Harvard Cambridge, Massachusetts Sara Triffo Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Yoshio Umezawa Musashino University Tokyo, Japan Milan Vrábel Academy of Sciences of the Czech Republic Prague, Czech Republic

Anita Vrcic The Broad Institute of MIT and Harvard Cambridge, Massachusetts Beatrice T. Wang Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Lance Wells University of Georgia Athens, Georgia Leah S. Witus University of California, Berkeley Berkeley, California Kevin J. Yarema The Johns Hopkins University Baltimore, Maryland Jacob S. Yount The Rockefeller University New York, New York Cheng-Han Yu National University of Singapore Singapore Chao Zhang Howard Hughes Medical Institute University of California, San Francisco San Francisco, California Mingzi M. Zhang The Rockefeller University New York, New York Yalong Zhang National Cancer Institute Frederick, Maryland

Fluorescence Polarization (FP) Assays for Monitoring Peptide-Protein or Nucleic Acid–Protein Binding Nathan J. Moerke1 1

Harvard Medical School, Boston, Massachusetts

ABSTRACT The technique of fluorescence polarization (FP) is based on the observation that when a fluorescently labeled molecule is excited by polarized light, it emits light with a degree of polarization that is inversely proportional to the rate of molecular rotation. This property of fluorescence can be used to measure the interaction of a small labeled ligand with a larger protein and provides a basis for direct and competition binding assays. FP assays are readily adaptable to a high-throughput format, have been used successfully in screens directed against a wide range of targets, and are particularly valuable in screening for inhibitors of protein-protein and protein-nucleic acid interactions when a small binding epitope can be identified for one of the partners. The protocols in this article describe a general procedure for development of FP assays to monitor binding of such a peptide or C 2009 by John oligonucleotide to a protein of interest. Curr. Protoc. Chem. Biol. 1:1-15  Wiley & Sons, Inc. Keywords: fluorescence polarization r FP r peptides r nucleic acids r proteins r high-throughput screening

INTRODUCTION This article describes a general procedure for the development of fluorescence polarization (FP) assays that can detect the binding of a small fluorescently labeled peptide or oligonucleotide to a protein of interest based on the property whereby when a fluorescently labeled molecule is excited by polarized light, it emits light with a degree of polarization that is inversely proportional to the rate of molecular rotation. The basic principle of fluorescence polarization is depicted in Figure 1. When a fluorophore that is covalently attached to a small (typically <1500 Da) ligand, such as a peptide in solution, is excited by polarized light, the emitted light will be largely depolarized. This is due to reorientation of the fluorophore during the lifetime of its excited state (on the order of nanoseconds for most fluorophores) caused by rapid Brownian molecular rotation of the labeled species. If this labeled ligand is bound to a high-molecular-weight protein (typically >10 kDa), the fluorophore reorients to a much smaller degree, due to the significantly reduced rotational speed of the complex. Thus, the emitted light will still be polarized to a significant degree. This property of FP allows it to be used as a technique for measurement of ligand binding, with the observed polarization in a mixture of labeled ligand and receptor being proportional to the fraction of bound ligand (Jameson and Seifried, 1999; Jameson and Croney, 2003). Furthermore, it is straightforward to establish a competition binding assay by measuring the decrease in FP signal produced when an inhibitor of the interaction is added to the mixture of labeled ligand and receptor. FP has a number of key advantages as an assay technology. It is carried out in solution phase, is nonradioactive, does not require any separation of bound and free ligand, and is readily adaptable to low volumes (on the order of 10 μl). This makes it well suited for high-throughput screening applications, and FP assays have been used successfully

Current Protocols in Chemical Biology 1: 1-15, December 2009 Published online December 2009 in Wiley Interscience (www.interscience.wiley.com). DOI: 10.1002/9780470559277.ch090102 C 2009 John Wiley & Sons, Inc. Copyright 

FP Assays for Monitoring Protein Binding

1 Volume 1

depolarized light

polarized light

largely polarized light

excitation of fluorophore at t = 0

emission of fluorophore at t = (excited state lifetime)

Figure 1 Schematic depicting the basic principle of fluorescence polarization. When a small peptide or nucleic acid ligand (dark circle) with a fluorescent label attached (white circle) is excited by polarized light at the excitation wavelength of the fluorophore, the ligand reorients to a significant degree due to molecular tumbling during the excited state lifetime of the fluorophore. This causes the emitted light to be largely depolarized. If the ligand is bound to a protein (gray ellipse), the resulting complex tumbles much slower, and the emitted light retains its polarization.

to study a wide variety of targets including kinases, phosphatases, proteases, G protein– coupled receptors (GPCRs), and nuclear receptors (Owicki, 2000; Burke et al., 2003). An important application of this type of assay is identification of inhibitors of proteinnucleic acid and protein-protein interactions. Even though these interactions can involve extensive interfaces, in many cases there are “hot spots,” small peptide or nucleotide sequences that are disproportionately important for the affinity of the interaction. These features of interfaces can be exploited to design fluorescence polarization probes for competitive binding assays. Novel compounds identified using assays of this type have come to be important not only as tools for basic chemical biology but also as potential drug leads. A significant number of antibiotics and other drugs bind to protein–nucleic acid interfaces (Pommier and Marchand, 2005) and protein-protein interactions have become increasingly recognized as key therapeutic targets (Arkin, 2005).

STRATEGIC PLANNING

FP Assays for Monitoring Protein Binding

Fluorescent Probe Design Fluorescently labeled peptides and oligonucleotides can be synthesized by a wide range of commercial services and academic core facilities. In the context of polarization experiments, these molecules are typically referred to as “tracers” or “probes.” When developing an FP assay to target a protein-peptide or protein–nucleic acid interaction, a number of key strategic considerations are involved in the design of the probe. The first decision is what peptide or oligonucleotide sequence to use. This should correspond to a sequence of one of the binding partners that is known to be necessary for the interaction, and that has a low molecular weight (<1500 Da is typical, although up to 5000 Da can be acceptable if the binding partner is very large). In the best case, a high-resolution crystal or NMR structure of the complex will be available that can guide selection of an appropriate peptide or nucleotide sequence. If this is not the case, site-directed mutagenesis data can be used in defining the probe sequence, and will be useful even if the structure is known. The sequence chosen should be as short as possible (to maximize

2 Volume 1

Current Protocols in Chemical Biology

the molecular weight difference between the probe and its binding partner, and thus the increase in FP) while still having sufficient binding affinity for the protein (generally in the low micromolar range or better). If available, quantitative measurements (using methods such as isothermal titration calorimetry), or rough estimates of the KD values of peptides or oligonucleotides, are valuable in deciding between different sequences. A detailed theoretical analysis of binding interactions in FP assays indicates that the higher the affinity of the fluorescent ligand, the wider the range of inhibitor potencies that can be resolved (Huang, 2003). In addition, the ligand concentration used in the assay must not be significantly higher than twice its KD , to avoid stoichiometric titration of the ligand (which reduces assay sensitivity). Thus, this analysis suggests that the optimal choice is to use a peptide or oligonucleotide with as high an affinity as possible, but not so high an affinity that a concentration lower than twice the KD would be below the lower detection range of the instrument for the fluorophore used. Once the probe sequence is chosen, it is necessary to choose a fluorophore and site of labeling. The key parameters for the fluorophore are the quantum yield and the fluorescence excited state lifetime. The fluorescence lifetime has a strong effect on the dependence of fluorescence polarization on molecular weight (Owicki, 2000). Fluorophores with lifetimes around 4 nsec, such as fluorescein and the rhodamines, provide a good separation in FP between small ligands and proteins of 10 kDa and higher. Fluorescein has traditionally been the most commonly used fluorescent label for FP, and suitable filters for its excitation and emission wavelengths are installed on almost all fluorescence plate readers. A number of alternatives to fluorescein with improved fluorescence properties (such as reduced pH dependence and less photobleaching), such as Alexa 488, are also available (Rusinova et al., 2002). Recently, red-shifted dyes such as Cy5 have been used successfully in FP assays as well (Turek-Etienne et al., 2004). There are a wide variety of different chemistries for covalent linking of a fluorescent label to a peptide or oligonucleotide. A number of fluorophore linking chemistries for both peptides (Brinkley, 1992; Weber et al., 1998; Fischer et al., 2003) and nucleic acids (Proudnikov, 1996) have been described in the literature. These allow one to attach a fluorescent moiety to either end of a peptide or oligonucleotide, or in some cases at an internal position within the molecule. In choosing the site of labeling, the first major consideration is to avoid interference with binding. Thus, for most ligands, the fluorophore will be linked to the peptide N or C terminus (or oligonucleotide 5 or 3 end). The other key factor is to avoid attaching the label at a position that will still be highly mobile even when the ligand is bound to the protein. In what is termed the “propeller effect,” this local mobility of the fluorophore can reduce the fluorescence polarization measured for the bound state. It is desirable to minimize the length of any linker between the peptide or oligonucleotide and the dye, and avoid attaching the dye to a “floppy end” of the ligand that is still highly flexible in complex with the protein. Structural data are thus extremely valuable in deciding where to attach a probe. In some cases, where little information is available, it may be necessary to try multiple labeling sites. The protocols in this article describe a general method for the establishment of a competition-based fluorescence polarization assay to monitor the binding of labeled peptide or oligonucleotide to a protein of interest. This can be divided into three basic procedures: establishment of conditions to measure direct binding of the labeled probe to the protein (Basic Protocol 1), demonstration that an unlabeled form of the ligand can competitively displace the labeled probe (Basic Protocol 2), and validation of the performance of the assay in a high-throughput format by determination of a Z factor (Basic Protocol 3). The protocols described can be generalized to any peptide-protein or nucleotide-protein interaction of interest. As a specific example to illustrate the steps

FP Assays for Monitoring Protein Binding

3 Current Protocols in Chemical Biology

Volume 1

and results of the first two protocols, an FP assay designed to measure binding of an eIF4G peptide to its interaction partner eIF4E is described (Moerke et al., 2007). The protein:protein interaction between eIF4E and eIF4G is necessary for formation of the eIF4F complex, which is involved in the initiation of translation on capped mRNAs (Gingras et al., 1999; von der Haar et al., 2004). Regulation of this complex plays an important role in cancer and other human diseases, and eIF4F is thus a potential therapeutic target (Sonenberg, 2008). BASIC PROTOCOL 1

MEASUREMENT OF DIRECT BINDING OF A FLUORESCENTLY LABELED PEPTIDE OR OLIGNUCLEOTIDE TO A PROTEIN BY FLUORESCENCE POLARIZATION Once a suitable fluorescent probe has been designed and synthesized (see Strategic Planning, above), the first step in assay development is to measure binding to the protein of interest by the increase in FP upon titration of a fixed concentration of the probe with the protein. Although these measurements can be made using a fluorimeter, assuming that the ultimate objective is to develop an assay for high-throughput screening, it is most convenient to use a multiwell plate reader capable of FP measurements. This allows the simultaneous measurement of multiple data points, facilitating the collection of titration curves. A plot of FP versus protein concentration should produce a sigmoidal curve that begins at the baseline polarization for the free ligand (typically ∼50 units of millipolarization, mP; described in more detail in step 5 below, and also in Background Information) and rises to plateau at a maximal polarization value that corresponds to complete binding of all labeled ligand (Fig. 2). The titration should be carried out in a buffer in which the protein is known to be stable. If the KD of the peptide or oligonucleotide that the probe is based on is known, an initial probe concentration close to this value should be chosen for constructing the first binding curve. If desired, subsequent binding curves can be measured at lower probe concentrations until the fluorescence detection limit of the instrument is reached. As long as the FP measurement variability remains constant, it is generally preferable to use as low a concentration of fluorescent probe as possible (as described in the introduction). If the KD is not known, several binding curves over a range of probe concentrations (from 1 nM to 1 μM) should be measured and evaluated. If desired, the binding data obtained can be used for determination of the KD of the labeled probe (Roehrl et al., 2004; see Commentary). The following protocol describes the generation of a direct binding curve for a labeled eIF4G peptide to eIF4E, but can be generalized to any FP system. Note that the KD for eIF4G peptide binding to eIF4E is 3 μM.

Materials Protein stock solution: 40 μM eIF4E (Moerke et al., 2007) in protein dilution buffer (see recipe for buffer) Protein dilution buffer (see recipe) Fluorescently labeled peptide stock solution: 10 μM eIF4E-fluorescein peptide KYTYDELFQLK in protein dilution buffer (see recipe for buffer) Black opaque 384-well microplates (Corning, cat. no. 3820) FP-capable plate reader (e.g., Analyst HT, Molecular Devices) Spreadsheet or graphing software

FP Assays for Monitoring Protein Binding

Prepare triplicate samples for FP measurements 1. Aliquot 180 μl of protein stock solution into a microcentrifuge tube, and make a series of eight two-fold dilutions into protein dilution buffer by sequentially mixing 90 μl of buffer with 90 μl of the previous protein solution in a new microcentrifuge tube.

4 Volume 1

Current Protocols in Chemical Biology

160

Fluorescence polarization (mP)

140 120 100 80 60 40 20 0 0

10

30

20

40

Concentration of elF4E ( M)

160

Fluorescence polarization (mP)

140 120 100 80 60 40 20 0 0.1

1

10

100

Concentration of elF4E ( M)

Figure 2 Direct binding of a fluorescent labeled eIF4G peptide to eIF4E. Fluorescence polarization (in units of mP) is plotted as a function of eIF4E concentration using a linear (top) or logarithmic scale (bottom). The concentration of eIF4G peptide is 1 μM.

It is generally desirable to start with as high a protein concentration as possible to make sure the plateau region of the binding curve is well defined. This, of course, is limited by the protein solubility. A typical starting concentration would be in the range of 40 to 100 μM. This should be achievable for most proteins, although in some cases a lower starting concentration will be necessary. It may be necessary to add more dilution points to adequately define the lower part of the curve, especially if the affinity of the labeled probe is very high.

2. Aliquot 10 μl of labeled peptide stock solution into each of the protein dilution tubes, and also to 90 μl of protein dilution buffer as a no-protein data point, for a final concentration of 1 μM labeled probe. Mix the contents of each tube by pipetting or gentle vortexing. For a final concentration of the labeled probe other than 1 μM, prepare a 10× stock solution in the protein dilution buffer. FP Assays for Monitoring Protein Binding

5 Current Protocols in Chemical Biology

Volume 1

3. Transfer 30 μl of each protein dilution/labeled probe mixture into three wells of a black opaque 384-well plate. Also transfer 30 μl of the no-protein solution and 30 μl of protein dilution buffer (as an assay blank for background fluorescence) into three wells each.

Obtain FP measurements 4. Load the plate into the plate reader, and use the instrument software to determine and set appropriate values for plate dimensions, measurement height, and detector gain. The detector gain will need to be reoptimized whenever the labeled probe concentration is changed, to avoid saturation and ensure adequate sensitivity.

5. Read the plate and obtain measurements of fluorescence polarization for each sample. For most instruments, the polarization is generally multiplied by 1000 and given in units of millipolarization (mP). The total fluorescence intensity polarized parallel (I ) and perpendicular (I⊥ ) to the incident light is also generally part of the readout. The binding between the labeled tracer and the protein typically reaches equilibrium relatively quickly (on the order of 5 to 20 min), so measurements of FP usually can be made within this time range, after addition of reagents to the plate. The stability of the FP measurements after this time should be verified empirically, as in some cases the binding reaction may take longer (on the order of hours) to reach equilibrium.

Analyze data 6. Export the data and plot the average FP measurement of the three wells for each data point as a function of the protein concentration using standard spreadsheet or graphing software. Figure 2 shows representative data for binding of a labeled eIF4G peptide to eIF4E.

7. Optional: Correct the polarization measurements for background contributions to the measured intensity by subtracting the parallel and perpendicular intensity readings, from the blank buffer-only wells, from the intensity readings for each data point. This is important when the labeled probe concentration is low enough that its signal is on a comparable order of magnitude to the background. For fluorescein, this is typically in the low nM range or below. At higher probe concentrations, this step may not be necessary.

8. Optional: Convert the polarization values to anisotropy and fit the data to estimate KD of labeled probe (see Commentary). 9. If necessary, repeat the experiment with a higher starting protein concentration or more data points to define the lower part of the curve. BASIC PROTOCOL 2

FP Assays for Monitoring Protein Binding

FLUORESCENCE POLARIZATION MEASUREMENT OF COMPETITIVE BINDING TO A PROTEIN OF AN UNLABELED PEPTIDE OR OLIGONUCLEOTIDE WITH A FLUORESCENTLY LABELED PROBE Once a fluorescent probe has been designed that binds to a protein of interest, it can be used for a competition binding assay. Such an assay works by measurement of the decrease in FP caused by a ligand that displaces the labeled probe. To establish the specificity of the assay, it is necessary to titrate a mixture of the protein and labeled probe with an unlabeled competitor and demonstrate that the FP decreases to the value observed with the free fluorescent ligand. The “gold standard” for specificity is to titrate with the exact, although unlabeled, peptide or oligonucleotide that was used for the labeled probe. However, other ligands that bind to the same site can be used as well. A concentration of protein should be chosen (based on the binding curve determined in Basic Protocol 1) that produces ∼50% to 80% of the increase in FP between the free ligand and the

6 Volume 1

Current Protocols in Chemical Biology

Fluorescence polarization (mP)

120 100 80 60 40 20 0 0

20

40

60

80

Concentration of unlabeled elF4G peptide ( M)

Fluorescence polarizationn (mP)

120 100 80 60 40 20 0 0.1

1

10

100

Concentration of unlabeled elF4G peptide ( M)

Figure 3 Competitive displacement of a fluorescent labeled eIF4G peptide from eIF4E by an unlabeled eIF4G peptide. Fluorescence polarization (in units of mP) is plotted as a function of unlabeled eIF4G peptide concentration using a linear (top) or logarithmic scale (bottom). The concentration of labeled eIF4G peptide is 1 μM and the concentration of eIF4E is 5 μM.

completely bound state. Higher concentrations of protein will lie in the plateau region of the binding curve and thus be insensitive to the competitor, and lower concentrations will not provide a good assay window for measurement of displacement of the probe. The displacement curve should have a sigmoidal shape, beginning at the bound FP value and decreasing to the FP value of the free labeled probe (Fig. 3). As in Basic Protocol 1, fitting of the binding curve can be used to estimate the KD of the competitor peptide or oligonucleotide (Roehrl et al., 2004; also see Commentary). The following protocol describes the generation of a competition binding curve for an unlabeled eIF4G peptide and eIF4E, but can be generalized to any FP system. An eIF4E concentration of 5 μM is chosen for the competition binding assay based on the binding curve generated in Basic Protocol 1. This concentration gives ∼65% of the increase in FP between free and completely bound states, and thus lies in the middle of the recommended range. As before, a concentration of 1 μM labeled peptide is used. Because of problems

FP Assays for Monitoring Protein Binding

7 Current Protocols in Chemical Biology

Volume 1

with the solubility with the unlabeled form of the first eIF4G peptide, a different peptide that binds to the same site is used as a competitor ligand.

Materials Unlabeled competitor peptide stock solution: 7 mM eIF4G peptide KKQYDREFLLDFQFMPA in DMSO (see recipe) Dimethyl sulfoxide (DMSO) Protein stock solution: 40 μM eIF4E (Moerke et al., 2007) in protein dilution buffer (see recipe for buffer) Fluorescently labeled peptide stock solution: 10 μM eIF4G-fluorescein peptide KYTYDELFQLK in protein dilution buffer (see recipe) Protein dilution buffer (see recipe) Black opaque 384-well microplates (Corning, cat. no. 3820) FP-capable plate reader (e.g., Analyst HT, Molecular Devices) Spreadsheet or graphing software Prepare triplicate samples for FP measurements 1. Aliquot 20 μl of unlabeled peptide stock solution into a microcentrifuge tube, and make a series of eight two-fold dilutions by sequentially mixing 10 μl of DMSO with 10 μl of the previous stock solution in a new microcentrifuge tube. It may be necessary to add more dilution points to adequately define the competition curve.

2. Prepare 1 ml of 5 μM protein/1 μM labeled peptide solution by adding 100 μl of 10 μM labeled peptide stock solution and 125 μl of 40 μM protein stock solution to 775 μl protein dilution buffer in a microcentrifuge tube. Mix well by pipetting or gentle vortexing. 3. Aliquot 100 μl of this solution into a separate microcentrifuge tube corresponding to each peptide dilution prepared in step 1, and into an additional tube for a nocompetitor control. 4. Pipet 1 μl of each peptide dilution into the corresponding 100-μl aliquot of protein/labeled peptide solution. Add 1 μl DMSO to the 100-μl aliquot for the nocompetitor control. Mix all samples well by gentle vortexing or pipetting. Incubate samples for 30 min at room temperature, protected from light. In some cases, the complex between the labeled probe and protein may be kinetically very stable (tight binding), and so longer incubation times may be required for the displacement reaction to reach equilibrium.

5. Add 1 μl of DMSO and 10 μl of labeled peptide stock solution to 90 μl of protein dilution buffer to prepare a probe-only control, to assess the completeness of displacement by the unlabeled peptide. 6. Transfer 30 μl of each sample into each of three wells of a black opaque 384well plate. Also transfer 30 μl of protein dilution buffer into three wells if background correction measurements will be needed (as in the optional step 7, of Basic Protocol 1).

Obtain FP measurements and analyze data 7. Load the plate into the plate reader and use the instrument software to set appropriate values for plate dimensions, measurement height, and gain. FP Assays for Monitoring Protein Binding

8. Read the plate and obtain measurements of fluorescence polarization for each sample.

8 Volume 1

Current Protocols in Chemical Biology

9. Export the data and plot the average FP measurement of the three wells for each data point as a function of the peptide concentration using standard spreadsheet or graphing software, optionally correcting the FP values for background fluorescence as in Basic Protocol 1, step 7. Figure 3 shows representative data for displacement of a labeled eIF4G peptide to eIF4E by an unlabeled eIF4G peptide.

10. Optional: Convert the polarization values to anisotropy, and fit the data to estimate KD of the competitor (see Commentary). 11. If necessary, repeat the experiment with a different protein concentration or more data points.

ADAPTATION OF A COMPETITION FLUORESCENCE POLARIZATION ASSAY TO HIGH-THROUGHPUT SCREENING FORMAT

BASIC PROTOCOL 3

Once a competition binding assay has been established, it can be adapted to highthroughput format for use in screening for small molecules that inhibit the interaction of interest. This involves choosing labeled probe and protein concentrations for the assay and confirming that the assay will perform adequately under the conditions to be used in screening. This will generally involve filling 384-well plates with the assay solution using automated liquid-handling equipment instead of pipetting, and adding controls and library compounds to the plates by means of pin transfer. A standard metric for evaluating assay quality and robustness is the Z factor (Zhang et al., 1999), which is given by Equation 1:

Z ′ = 1−

(3σpos + 3σneg ) μpos − μneg

Equation 1

Here, σ and μ represent the standard deviations and means of the positive and negative controls, respectively. Typically, the Z factor is calculated based on a single plate that contains approximately one-half positive and one-half negative control wells. An assay with a Z value of at least 0.5 is recommended for high-throughput screening, and one with a Z of 0.7 or higher is considered an excellent assay. For a competition-based assay, a positive control should cause a decrease in FP, which is what is expected from a “hit” compound that inhibits the interaction being measured. A negative control should have no significant effect on the FP. Since compound libraries used in high-throughput screening are generally formatted using DMSO as a solvent, the standard negative control is addition (by pin transfer) of a volume of DMSO equivalent to that which will be transferred during the assay (a typical volume would be 100 nl, for a 30-μl assay volume). The standard positive control is the unlabeled peptide or oligonucleotide used for the competition curve in Basic Protocol 2, at a concentration sufficient to completely displace the labeled probe. This should be formulated as a stock in DMSO that will give 1× final concentration when the pin-transfer volume is added to the assay well (so, 300× for transfer of 100 nl for a 30-μl assay volume). A small molecule already known to inhibit the interaction between the probe and the protein may also be used as the positive control.

Materials Fluorescently labeled peptide or oligonucleotide stock solution (see Basic Protocols 1 and 2) Protein stock solution (see Basic Protocols 1 and 2)

FP Assays for Monitoring Protein Binding

9 Current Protocols in Chemical Biology

Volume 1

Unlabeled competitor peptide or oligonucleotide stock solution (see Basic Protocols 1 and 2) Dimethylsulfoxide (DMSO) Protein dilution buffer (see recipe) Polypropylene 384-well compound storage plates (Thermo Scientific, cat. no. AB-1056) Automated liquid dispenser for multiwall plates (Matrix WellMate, Thermo Scientific) Black opaque 384-well microplates (Corning, cat. no. 3820) Pin transfer apparatus (V&P Scientific, http://www.vp-scientific.com/) FP-capable plate reader (e.g., Analyst HT, Molecular Devices) Spreadsheet or graphing software Prepare plate for Z determination 1. Determine the optimal protein and labeled probe concentrations based on the results of Basic Protocols 1 and 2. 2. Determine the minimum concentration of unlabeled competitor that will completely inhibit binding based on the curve generated in Basic Protocol 2, and prepare 2 ml of a 300× stock in DMSO. 3. Add 10 μl DMSO to one-half of the wells of a polypropylene compound storage plate (negative control) and add 10 μl of the 300× unlabeled competitor stock to the other one-half of the wells (positive control). This will be the source plate for pin transfer. This plate can be reused and should be kept at −20◦ C for long-term storage.

4. Prepare 20 ml of assay solution containing the appropriate concentrations of the protein and labeled probe in protein dilution buffer. This is sufficient to fill a single 384-well plate with extra volume to account for the “dead volume” found in most liquid dispensers.

5. Using the automated liquid dispenser, fill an entire black opaque 384-well plate with the assay solution at 30 μl per well (this will be the assay plate). If the measurements of FP will be corrected for background fluorescence, fill one column with protein dilution buffer alone, instead of assay solution. 6. Using the pin transfer apparatus transfer a volume of 100 nl from each well of the source plate to the assay plate. 7. Incubate the assay plate for 30 min at room temperature, covering with foil or an empty plate to protect it from exposure to light. The length of the incubation can be increased if necessary for complexes of labeled protein and probe that are kinetically very stable (tight binding).

Obtain FP measurements and analyze data 8. Load the plate into the plate reader, and use the instrument software to set appropriate values for plate dimensions, measurement height, and gain. 9. Export the FP measurement data into a standard spreadsheet program, correct FP values for background fluorescence if necessary, and calculate the Z score using Equation 1. FP Assays for Monitoring Protein Binding

10 Volume 1

Current Protocols in Chemical Biology

REAGENTS AND SOLUTIONS Use Milli-Q purified water or equivalent in all recipes and protocol steps.

Protein dilution buffer 50 mM sodium phosphate (mix monobasic and dibasic phosphate in appropriate ratio for pH 6.5) 50 mM KCl Adjust to pH 6.5 with HCl or NaOH Add DTT fresh from 1 M frozen stock, and filter buffer to remove any precipitated matter before using in assays Once DTT is added, store up to 1 week at 4◦ C Fluorescently labeled peptide stock solution Dissolve lyophilized peptide (sequence KYTYDELFQLK with fluorescein conjugated to C-terminal lysine; synthesized by Research Genetics, http://www.researchgenetics.com) in protein dilution buffer (see recipe) to a concentration of 10 μM. Aliquot peptide solution and store protected from light indefinitely at −20◦ C or −70◦ C.

Unlabeled peptide stock solution Dissolve lyophilized peptide (sequence KKQYDREFLLDFQFMPA; synthesized by Tufts University Core Facility) in DMSO to a concentration of 7 mM. Aliquot peptide solution and store indefinitely at −20◦ C or −70◦ C.

COMMENTARY Background Information In a typical instrument that measures fluorescence polarization, light at the proper wavelength for excitation of the fluorophore is selected by a band-pass filter. It then passes through an excitation polarizing filter that causes it to be plane polarized. The polarized light is reflected by a dichroic mirror into the wells of a microtiter plate containing the sample of interest where the fluorophore is excited. Emitted light passes back out of the well through the dichroic mirror and goes sequentially through a polarizing filter that has the same orientation as the excitation polarizing filter (the parallel polarizing filter) and a filter oriented at 90◦ to the excitation filter (the perpendicular polarizing filter). Finally, for both the parallel and perpendicular light, a second band-pass filter selects for the emission wavelength before the light reaches the detector. The fluorescence polarization then is defined by Equation 2:

P=

I − I⊥ I + I⊥

Equation 2

Here I is the intensity of emitted light polarized parallel to the excitation light, and I⊥

is the intensity of emitted light polarized perpendicular to the excitation light. An important property of the polarization that emerges from this equation is that it is independent of the fluorophore concentration. Although this equation assumes that the instrument has equal sensitivity for light in both the perpendicular and parallel orientations, in practice this is not the case. A modified version of the polarization equation includes a correction factor, G, that accounts for differential sensitivity, as in Equation 3.

P=

I − G ∗ I⊥ I + G ∗ I⊥

Equation 3

The G factor is dependent on the specific optical components used to measure fluorescence polarization in an instrument. Determination of G is necessary for the calculation of absolute polarization values, and for the comparison of data between different instruments. However, it is not necessary for analysis of assay and screening data that are all collected on the same instrument, and if this is the case, it can be ignored (i.e., set to 1) if desired. This is due to the fact that variation in the G factor does not significantly affect the

FP Assays for Monitoring Protein Binding

11 Current Protocols in Chemical Biology

Volume 1

relative differences between FP measurements in a binding assay, and does not change the assay window. The details of the procedure for measurement of the G factor will depend on the fluorophore used and the instrument, and instrument documentation should be consulted for this information. A closely related property to polarization is fluorescence anisotropy, which is defined by Equation 4 (without G factor correction):

A=

I − I⊥ I + 2I⊥

=

2P 3− P

Equation 4

For historical reasons most instruments give their output in terms of polarization. However, conversion of data to anisotropy is useful in fitting binding curves and displacement curves to equations describing the multiple binding equilibria between the protein, the labeled probe, and the competitor. This can be used to determine KD values for known ligands and inhibitor compounds, and detailed procedures to do this have been described (Roehrl et al., 2004).

Critical Parameters

FP Assays for Monitoring Protein Binding

Choice of fluorescent label Before beginning the experiments described in this unit, careful consideration must be given to the choice of fluorescent label to use. The initial choice of fluorophore may, of course, be constrained by the capabilities of the facility used for synthesis of the probe. In addition, it is essential that the available fluorescent plate reader have the correct emission filter, excitation filter, and dichroic mirror (if applicable) for the fluorophore being considered. Once the set of possible labels has been narrowed down, another consideration is the excitation and emission wavelengths. Two major sources of interference in screens using FP assays are fluorescence from library compounds and light scattering due to precipitated compounds (scattered light has a very high polarization). Several studies have found that red-shifted fluorophores suffer from significantly less interference from compound fluorescence than dyes with shorter wavelength maxima (Turek-Etienne et al., 2003; Simeonov et al., 2008). Light scattering is also less of a problem for such fluorophores, as the intensity of scattered light decreases with increasing wavelength. Thus, the use of a red-shifted fluorophore may reduce the percentage of false positives and negatives in a screen. This

may be especially advantageous for screens of natural product libraries that have a higher incidence of compound interference (TurekEtienne et al., 2004). Reagent purity As with all fluorescence measurements, the labeled probe should be of the highest available purity to minimize the presence of contaminating fluorescent components. Purification of fluorescently labeled peptide and oligonucleotides is generally performed by HPLC, and the facility may provide HPLC traces and/or mass spectrometry data to confirm purity. The labeled probe should be kept protected from direct light exposure during experiments and in storage. Finally, care must be taken to minimize the presence of precipitated matter in assay solutions that may cause lightscattering interference with FP measurements. Buffers should be filtered, and protein, peptide, or nucleotide solutions should be briefly centrifuged before using. Multiwell plates There are a number of choices for black opaque 384-well plates to be used in these experiments. It is essential that the specific plates used be validated for the plate reader and other automation that will be used. If there is concern about nonspecific interactions of the protein or labeled probe with the surface of wells, it can be worthwhile to start out using plates with a hydrophilic, nonbinding surface (NBS). Also, if the protein used in the assay is difficult or expensive to purify, low-volume assay plates for fluorescence polarization are available (with a typical working volume of 5 to 40 μl) that can minimize protein consumption. Such plates are available from many microplate suppliers, e.g., Corning and Greiner. Fluorescent plate readers differ in the precise methods used in setting plate dimensions, plate measurement height, detector gain, and signal integration time used in FP measurements for a specific plate and fluorophore concentration. Careful attention must be paid to these parameters in order to maximize assay sensitivity.

Troubleshooting An overall flowchart of the protocols described in this unit with appropriate troubleshooting steps at various points is shown in Figure 4. Probe binding There may be a lack of increase in FP in the titration of Basic Protocol 1 due to the fluorescent label interfering with the interaction with

12 Volume 1

Current Protocols in Chemical Biology

choose peptide/nucleotide sequence for FP probe

redesign the FP probe

choose fluorophore and site of labeling

no improvement in binding/competition curves?

generate binding curve of labeled prode to protein of interest yes

acceptable binding curve?

no

generate competition curve by titration with unlabeled probe yes

acceptable competition curve?

no no improvement in Z ?

determine Z factor for assay in highthroughput format yes

acceptable Z factor?

change concentration of labeled probe modify buffer conditions change type of assay plate used

no

adjust protein concentration increase integration time

proceed to screening

Figure 4 Flowchart describing the stages of development of an FP assay for high-throughput screening, with recommended troubleshooting procedures for each stage.

the protein. If another binding assay exists for the protein of interest, this can be used to confirm whether this is the case, and, if so, it will be necessary to redesign the labeled probe. Nonspecific binding Another source of potential error in fluorescence polarization measurements is nonspecific binding of the protein of interest to the labeled probe, possibly due to interactions with the fluorescent label. In Basic Protocol 1, this may cause an abnormal binding curve, possibly with a linear rather than sigmoidal shape. There may be a lack of a clear plateau region, with the FP continuing to increase with increasing protein concentration. Another strong indicator of nonspecific binding is an inability to achieve complete displacement of the labeled probe in Basic Protocol 2. If this type of nonspecific binding is suspected, one troubleshooting step is to reduce the labeled probe concentration as much as possible while still maintaining adequate sensitivity of detection. In addition, nonspecific interactions can often be reduced or eliminated by adding proteins such as BSA or γ globulin and/or detergents such as Tween-20 or Triton X-100 to the buffer. These should be carefully titrated so as to avoid

disrupting the specific interaction between the protein and the labeled probe. If a labeled peptide containing free cysteines is being used, inclusion of reducing agents such as DTT may also be helpful. If these measures fail to eliminate nonspecific binding, it may be necessary to redesign the probe with a different fluorophore or site of labeling. It is also possible to have nonspecific binding of the labeled probe or the protein to the surface of the wells of the plates being used. This can cause there to be no increase, or a smaller-than-expected increase in FP upon titration of the labeled probe with protein in Basic Protocol 1, or also an abnormally high FP for the labeled probe alone. Again, this can potentially be reduced by changing buffer conditions to reduce nonspecific interaction. In addition, the use of NBS plates designed to minimize hydrophobic interactions can often eliminate the problem. Z  factor optimization When determining the Z factor in Basic Protocol 3, the two factors affecting this value are the precision of FP measurements and the size of the assay window, or difference in mP between the free or displaced labeled

FP Assays for Monitoring Protein Binding

13 Current Protocols in Chemical Biology

Volume 1

probe and the bound probe. The assay window is dependent on the concentration of protein used, and so carefully increasing this (without moving into the plateau region of the binding curve) can improve the Z factor. For many plate readers, it is possible to improve the precision of FP measurements by increasing the signal integration time. As the effective probe concentration can be reduced by nonspecific binding to the surface of wells, changing buffer conditions or switching to an NBS plate may also improve the precision of measurements and the Z factor. For these changes in conditions, one should repeat the binding and displacement curves in Basic Protocols 1 and 2 to confirm that the assay is still functioning as before.

Anticipated Results Basic Protocols 1 and 2 will generate one or more FP binding curves for the labeled probe and one or more competition curves for the unlabeled form of the probe (or other appropriate peptide or oligonucleotide). In Basic Protocol 3, these data will be used to establish initial optimal conditions for a high-throughput FP assay, which are tested by determination of the Z factor. Once an adequate Z factor has been achieved (possibly after one or more rounds of optimization of the initial conditions), the assay can be considered to be ready for highthroughput screening.

Time Considerations Once all necessary reagents are in hand, the procedures described above could be completed within a day of work or less, although in practice up to a week is often needed to develop and validate the assay if one or more steps need to be optimized.

Acknowledgement The author would like to acknowledge funding from NIH AI067751 that helped to support this work.

Literature Cited Arkin, M. 2005. Protein-protein interactions and cancer: Small molecules going in for the kill. Curr. Opin. Chem. Biol. 9:317-324. Brinkley, M. 1992. A brief survey of methods for preparing protein conjugates with dyes, haptens, and cross-linking reagents. Bioconjug. Chem. 3:2-13.

FP Assays for Monitoring Protein Binding

Burke, T.J., Loniello, K.R., Beebe, J.A., and Ervin, K.M. 2003. Development and application of fluorescence polarization assays in drug discovery. Comb. Chem. High Throughput Scr. 6:183194.

Fischer, R., Mader, O., Jung, G., and Brock, R. 2003. Extending the applicability of carboxyfluorescein in solid-phase synthesis. Bioconjug. Chem. 14:653-660. Gingras, A.C., Raught, G., and Sonenberg, N. 1999. eIF4 initiation factors: Effectors of mRNA recruitment to ribosomes and regulators of translation. Annu. Rev. Biochem. 68:913-963. Huang, X. 2003. Fluorescence polarization competition assay: The range of resolvable inhibitor potency is limited by the affinity of the fluorescent ligand. J. Biomol. Scr. 8:34-38. Jameson, D.M. and Seifried, S.E. 1999. Quantification of protein-protein interactions using fluorescence polarization. Methods 19:222-233. Jameson, D.M. and Croney, J.C. 2003. Fluorescence polarization: past, present and future. Comb. Chem. High Throughput Scr. 6:167173. Moerke, N.J., Aktas, H., Chen, H., Cantel, S., Reibarkh, M.Y., Fahmy, A., Gross, J.D., Degterev, A., Yuan, J., Chorev, M., Halperin, J.A., and Wagner, G. 2007. Small-molecule inhibition of the interaction between the translation initiation factors eIF4E and eIF4G. Cell 128:257-267. Owicki, J.C. 2000. Fluorescence polarization and anisotropy in high throughput screening: Perspectives and primer. J. Biomol. Scr. 5:297306. Pommier, Y. and Marchand, C. 2005. Interfacial inhibitors of protein-nucleic acid interactions. Curr. Med. Chem. Anticancer Agents 5:421429. Proudnikov, D. and Mirzabekov, A. 1996. Chemical methods of DNA and RNA fluorescent labeling. Nucleic Acids Res. 24:4535-4542. Roehrl, M.H., Wang, J.Y., and Wagner, G. 2004. A general framework for development and data analysis of competitive high-throughput screens for small-molecule inhibitors of protein-protein interactions by fluorescence polarization. Biochemistry 43:16056-16066. Rusinova, E., Tretyachenko-Ladokhina, V., Vele, O.E., Senear, D.F., and Alexander Ross, J.B. 2002. Alexa and Oregon Green dyes as fluorescence anisotropy probes for measuring proteinprotein and protein-nucleic acid interactions. Anal. Biochem. 308:18-25. Simeonov, A., Jadhav, A., Thomas, C.J., Wang, Y., Huang, R., Southall, N.T., Shinn, P., Smith, J., Austin, C.P., Auld, D.S., and Inglese, J. 2008. Fluorescence spectroscopic profiling of compound libraries. J. Med. Chem. 51:23632371. Sonenberg, N. 2008. eIF4E, the mRNA cap-binding protein: From basic discovery to translational research. Biochem. Cell Biol. 86:178-183. Turek-Etienne, T.C., Small, E.C., Soh, S.C., Xin, T.A., Gaitonde, P.V., Barrabee, E.B., Hart, R.F., and Bryant, R.W. 2003. Evaluation of fluorescent compound interference in 4 fluorescence polarization assays: 2 kinases, 1 protease, and 1 phosphatase. J. Biomol. Scr. 8:176-184.

14 Volume 1

Current Protocols in Chemical Biology

Turek-Etienne, T.C., Lei, M., Terracciano, J.S., Langsdorf, E.F., Bryant, R.W., Hart, R.F., and Horan, A.C. 2004. Use of red-shifted dyes in a fluorescence polarization AKT kinase assay for detection of biological activity in natural product extracts. J. Biomol. Scr. 9:52-61. von der Haar, T., Gross, J.D., Wagner, G., and McCarthy, J.E. 2004. The mRNA capbinding protein eIF4E in post-transcriptional gene expression. Nat. Struct. Mol. Biol. 11:503511. Weber, P.J., Bader, J.E., Folkers, G., and BeckSickinger, A.G. 1998. A fast and inexpensive method for N-terminal fluorescein-labeling

of peptides. Bioorg. Med. Chem. Lett. 8:597600. Zhang, J.H., Chung, T.D., and Oldenburg, K.R. 1999. A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Scr. 4:67-73.

Internet Resources http://www.ncgc.nih.gov/guidance/ manual toc.html National Center for Chemical Genomics Assay Guidance Manual. This provides an overview of FP assays in high-throughput screening and a comparison to other assay methods.

FP Assays for Monitoring Protein Binding

15 Current Protocols in Chemical Biology

Volume 1

Spatiotemporal Dynamics of Kinase Signaling Visualized by Targeted Reporters Maya T. Kunkel1 and Alexandra C. Newton1 1

University of California at San Diego, La Jolla, California

ABSTRACT The advent of genetically encoded FRET-based kinase activity reporters has ushered in a new era of signal transduction research. Such reporters allow the direct monitoring of kinase activity in live cells at speciÞc locations, providing unprecedented information on the spatiotemporal dynamics of kinase signaling. SpeciÞcally, FRET-sensitive conformational changes in the reporters following phosphorylation serve as a direct readout of kinase activity. These genetically encoded reporters allow not only temporal resolution of kinase activity, but also spatial resolution: by fusing appropriate targeting sequences, reporters can be positioned at speciÞc subcellular locations. Herein is presented a strategy to generate and target kinase activity reporters to discrete intracellular regions to measure C 2009 by John Wiley kinase signaling in live cells. Curr. Protoc. Chem. Biol. 1:17-28  & Sons, Inc. Keywords: kinase activity reporter r imaging r FRET r localized kinase signaling

INTRODUCTION Kinase signaling is an essential component of many cellular processes. Phosphorylation of target substrate proteins generally modulates their function by altering their activity, localization, or association with other proteins. Thus, measuring kinase activity is integral to understanding cell signaling. The activity of protein kinases is most commonly assessed by immunoblotting with a phospho-speciÞc antibody against the phosphorylated substrate protein. However, this approach has limitations. It is becoming increasingly clear that signaling events often occur at the subcellular level and the spatiotemporal resolution of these events, as measured using conventional methods, is limited by the time resolution of the assay as well as the success of cell fractionation methods. Another common approach to assessing the activation state of a kinase is to immunoblot using phospho-speciÞc antibodies made against sites on the kinase purported to indicate kinase activation. This is also an imperfect measure, as there may be other means of kinase activation, or even inactivating phosphorylations, neither of which are reßected when analyzing a speciÞc site. Genetically encoded reporters overcome many of these problems, allowing visualization of activity in real time at precise intracellular locations. Herein is presented a strategy to generate and target kinase activity reporters to discrete intracellular regions, to measure kinase signaling in live cells (see Basic Protocol). Also included is a Support Protocol to determine the dynamic range of a ßuorescent reporter. A brief discussion of targeting strategies to observe the kinetics, persistence, and level of the kinase response at distinct intracellular regions is provided in the Anticipated Results section of the Commentary.

Kinase Signaling Visualized by Targeted Reporters Current Protocols in Chemical Biology 1: 17-28, December 2009 Published online December 2009 in Wiley Interscience (www.interscience.wiley.com). DOI: 10.1002/9780470559277.ch090106 C 2009 John Wiley & Sons, Inc. Copyright 

17 Volume 1

STRATEGIC PLANNING Reporter design The archetypal model for a FRET-based reporter to measure kinase activity has the modular design indicated in Figure 1. SpeciÞcally, the reporter comprises a phosphorylation sensor [phosphopeptide-binding domain (PBD) and a substrate sequence] that is ßanked by a FRET donor/acceptor pair. In designing reporters, an appropriate consensus substrate sequence for the kinase to be monitored is the foremost decision. To achieve speciÞcity of the reporter, this sequence must be readily phosphorylated by the kinase of interest, yet not readily recognized by other kinases. This sequence is placed adjacent to a phosphoamino acid binding domain which, when the consensus sequence is phosphorylated, undergoes intramolecular complexation resulting in a conformational change of the kinase activity reporter. This change alters the distance and/or orientation between the selected FRET pair that bracket the substrate sequence and phosphoamino acid binding domain, thus resulting in a change in FRET that can be monitored in live cells. Note that some reporters display an increase in FRET upon phosphorylation by the relevant kinase (Fig. 1A; Zhang et al., 2001), whereas others undergo a loss of FRET following phosphorylation (Fig. 1B; Violin et al., 2003). Identifying a suitable consensus sequence There are many approaches one can take in designing a consensus sequence. First, a sequence derived from a prominent substrate of the kinase of interest can be chosen and, if needed, further modiÞed. This approach was used successfully in the development of isozyme-speciÞc kinase activity reporters for protein kinase C (PKC; Kajimoto and

A CFP

PBD

substrate peptide OH

kinase

YFP

CFP

434 nm 434 nm

PBD P

phosphatases

476 nm

YFP

FRET

528 nm

B PBD

substrate peptide OH

CFP YFP 434 nm

P PBD

kinase

phosphatases

CFP

FRET

528 nm

Kinase Signaling Visualized by Targeted Reporters

YFP

434 nm

476 nm

Figure 1 Schematic diagram showing the modular structure of kinase activity reporters. Kinase activity reporters consist of a FRET donor (e.g., CFP), a phosphoamino acid binding domain (PBD), a consensus phosphorylation sequence (substrate peptide), and a FRET acceptor (e.g., YFP). The FRET pair are at a distance and orientation with respect to one another that changes once the substrate peptide is phosphorylated (circle with P) by the kinase; following phosphorylation, the PBD binds to the phosphorylated sequence, resulting in a conformational change that alters FRET between the FRET donor and FRET acceptor. Depending on the reporter construct, phosphorylation can cause an increase (A) or decrease (B) in FRET.

18 Volume 1

Current Protocols in Chemical Biology

Newton, in preparation). The sequence of a known substrate is used, and then a series of amino acid additions, deletions, or substitutions are made to optimize phosphorylation by the kinase of interest and reduce phosphorylation by other kinases. The speciÞcity of the sequence can be assessed in the context of the reporter via live cell imaging (see below) or in vitro (see Troubleshooting). While time consuming (these isozyme-speciÞc PKC reporters underwent over 30 permutations each), this approach can produce a very speciÞc kinase activity reporter. A second approach involves identifying a consensus phosphorylation sequence in silico, and there are a number of databases that have consolidated information about known kinase substrates. KinasePhos (http://kinasephos.mbc.nctu.edu.tw) is one such database that presents a weighted picture of the sequence requirements for phosphorylation by speciÞc kinases. Phospho.ELM (http://phospho.elm.eu.org), PhosphoMotif Þnder (http://www.hprd.org/PhosphoMotif Þnder), Scansite (http://scansite.mit.edu), and PhosphoSite Plus (http://www.phosphosite.org) are other sites that employ different algorithms to create databases of known substrate sequences of kinases. Finally, the most informative approach is an unbiased method using a combinatorial peptide library to determine optimal amino acids at positions N-terminal and C-terminal to the site of phosphorylation (Turk et al., 2006). Here, biotinylated peptides are synthesized in which, for each of the assayed positions surrounding the phosphoacceptor site, a single amino acid is Þxed; the other positions contain a mixture of all amino acids except serine, threonine, or tyrosine, such that the only phosphorylatable residue within the peptide is the phosphoacceptor site. These peptides are incubated in solution with radioactive ATP and the kinase, and then spotted onto a streptavidin membrane. Phosphorylation of the peptide mixtures is then quantiÞed using a phosphor imager. Using this assay, the data yield a thorough picture of the sequence preferences surrounding the phosphorylation site, both favorable and unfavorable, that are recognized by the puriÞed kinase. All of these methods for selecting a consensus substrate sequence sufÞce for kinases that efÞciently phosphorylate their substrates. However, some kinases have additional requirements for phosphorylation of substrates, notably docking motifs. Thus, for example, in the design of the kinase activity reporter for ERK, a docking domain for ERK was included on its C-terminus (Sato et al., 2007). After having settled on a sequence, it is critical to determine if any other kinases may recognize and potentially phosphorylate the sequence. There are a number of databases that predict which kinases will phosphorylate a speciÞc sequence using computational methods. The substrate sequence information can be uploaded into programs such as Scansite, as well as some of the other programs listed above, which will return with a series of potential kinases that are predicted to phosphorylate the input sequence. From here, one can experimentally check the speciÞcity of the kinase activity reporter against these identiÞed kinases.

Choosing a phosphoamino acid binding domain In identifying a suitable substrate sequence to insert into the reporter, one must also consider the sequence requirements for binding of the phosphoamino acid binding domain. The reporter function is based on a conformational change induced following substrate phosphorylation, an event that triggers an intramolecular clamp between the phospho-binding domain and the newly phosphorylated sequence. There are a number of options in choosing such a domain. In the case of a tyrosine kinase activity reporter, Src-homology 2 (SH2) domains or phosphotyrosine binding (PTB) domains serve as the

Kinase Signaling Visualized by Targeted Reporters

19 Current Protocols in Chemical Biology

Volume 1

binding moiety. In the case of serine/threonine kinase activity reporters, one can take advantage of the forkhead-associated (FHA) domains, 14-3-3 domains, and some WW domains. One thing to take into consideration in using these latter domains is whether the phospho-binding domains show a preference for serine or threonine phosphorylation. Furthermore, the sequence requirements for binding for some of these domains have been investigated, and these requirements must be compatible with the consensus substrate sequence used. For example, the FHA2 domain only binds to phosphorylated threonine residues and shows a strong preference for a leucine or isoleucine at the +3 position (Durocher et al., 2000). In all situations, one must decide whether the reporter is to be reversible, allowing one to observe both phosphorylation and dephosphorylation events (following kinase inactivation or inhibition). In order to have reversible signaling, the afÞnity of binding needs be high enough to induce the conformational change necessary, but not so high that phosphatases are unable to access and dephosphorylate the reporter. The optimization of the kinase activity reporter for PKA (A Kinase Activity Reporter, AKAR) from its Þrst to its second permutation highlights these considerations. The original AKAR utilized a 14-3-3 PBD to induce a FRET change upon binding the phosphorylated substrate sequence within AKAR; however, the binding afÞnity of this interaction was so high that phosphatases could not readily reverse the reaction (Zhang et al., 2001). The second version of the PKA reporter, AKAR2, utilized the FHA1 PBD, which displayed one order of magnitude weaker binding than the 14-3-3 used in AKAR, and this new reporter, AKAR2, is now a reversible FRET reporter of PKA activity (Zhang et al., 2005).

Choosing a FRET pair The FRET pair of choice today is still the combination of two modiÞed green ßuorescent proteins (GFPs) that were engineered to ßuoresce in the cyan (CFP) and yellow (YFP) wavelengths. Of the CFP and YFP variants, the authors have experience using ECFP and Citrine as the FRET pairs (Griesbeck et al., 2001; Rizzo et al., 2004). GFP exists as a dimer, and the tendency for the CFP and Citrine variants to dimerize affects the reversibility of the reporter. The introduction of monomeric ßuorescent proteins resulted in a faster response of the kinase activity reporter than had been Þrst observed using the parent CFP and YFP (Dunn et al., 2006). While these are the most popular FRET pairs used today, additional modiÞed ßuorescent proteins have been engineered from the parent GFP as well as from Discosoma sp. red ßuorescent protein (DsRed; Shaner et al., 2005). Note that a monomer version of DsRed, termed mRFP for monomeric Red Fluorescent Protein, has been constructed (DsRed exists as a tetramer; Campbell et al., 2002). Indeed there are reports describing FRET between other ßuorescent proteins, and these could be used in the reporter in place of the CFP and YFP variants (Carlson and Campbell, 2009). More excitingly, the use of dual FRET pairs with noninterfering overlap in their excitation/emission spectra presents the option of imaging two biological processes within the same cell in real time. To begin, however, use of CFP and YFP is recommended (see Table 1); this is the standard FRET pair used to date. Table 1 Filter Sets Used to Monitor CFP, FRET, and YFP Fluorescence

Kinase Signaling Visualized by Targeted Reporters

Fluorochrome

Excitation (nm)

Dichroic mirror (nm)

Emission (nm)

CFP

420/20

450

475/40

FRET

420/20

450

535/25

YFP

495/10

505

535/25

20 Volume 1

Current Protocols in Chemical Biology

Control construct An important and simple control construct is the reporter construct in which the phosphoacceptor site is mutated to an unphosphorylatable residue; i.e., the respective serine, threonine, or tyrosine changed to an alanine. This reporter should not undergo a FRET ratio change following stimulation of the kinase, as the critical residue is mutated and no longer phosphorylatable by the kinase. It also serves as a negative control in experiments where there is concern that the reporter is physically relocalizing within the cell, or where the cell is changing morphology; situations like these can appear as a change in FRET, as the local reporter concentration may change, thereby impacting basal intermolecular FRET. Imaging setup A number of setups can be used to image FRET-based reporters. The protocol described below is the one used in the authors’ laboratory to image a number of distinct kinase activity reporters (Violin et al., 2003; Kunkel et al., 2005; Gallegos et al., 2006). The setup includes a Zeiss Axiovert 200M microscope, a Roper ScientiÞc MicroMAX 512BFT CCD camera, and a Lambda 10-2 Þlterwheel shutter controller from Sutter, and is controlled using Metaßuor software from Molecular Devices. Fluorescent images are acquired at room temperature through a 10% neutral density Þlter and a 40×/1.3 NA oil-immersion objective with the Þlters shown in Table 1. Calibration First, one must calibrate the imaging setup to control against photobleaching. On the authors’ microscope, a series of images (CFP, FRET, YFP) can be acquired every 10 sec without bleaching of the ßuorophores, using the following exposure times: 200 msec for CFP, 200 msec for FRET, 100 msec for YFP. Many parameters are involved, so this must be experimentally determined for each imaging setup. The brightness of the reporter must then be calibrated to select cells in which its expression level is similar to that of endogenous substrates (∼1 μM). This is important because too high a concentration of exogenous reporter could compete with endogenous substrates and perturb normal cell signaling. To crudely estimate the concentration of reporter, the ßuorescence level of a known concentration of puriÞed ßuorescent protein is determined on the microscope. A drop of puriÞed ßuorophore is placed adjacent to a coverslip, which itself is resting on a coverslip (Fig. 2). A third coverslip is placed on top to create a wedge with known length and height. From this, a series of images are captured across the length of the wedge and the ßuorescence intensity is quantiÞed. After plotting this relationship (intensity versus distance along the slide), the concentration of reporter in a given thickness of cell can be estimated. While the ßuorescence of the pure reporter will not equal the ßuorescence in a cell, this method gives an estimate that is within several-fold correct (Violin, 2003).

Figure 2 Schematic depicting a method to calibrate reporter concentration in cells. A drop of puriÞed ßuorophore of known concentration (gray color) is placed in a wedge generated using coverslips. A series of images is acquired across the length of the wedge to correlate the intensity of signal with the height at each point in the wedge. From this, and knowing the estimated thickness of the cell type used, one can estimate the intensity expected for reporter concentration within a cell.

Kinase Signaling Visualized by Targeted Reporters

21 Current Protocols in Chemical Biology

Volume 1

BASIC PROTOCOL

IMAGING KINASE ACTIVITY REPORTERS TO MEASURE KINASE SIGNALING IN LIVE CELLS Materials Cells of interest: e.g., HeLa cells Cell culture medium (e.g., DMEM) DNA encoding kinase activity reporter (see Strategic Planning) Transfection reagent (e.g., FuGENE 6 from Roche) Hanks’ Balanced Salt Solution (HBSS), without Ca2+ and Mg2+ (e.g., Cellgro), but supplemented with 1 mM Ca2+ on the day of the imaging experiment Activator of kinase signaling path: e.g., histamine or phorbol dibutyrate (PdBu) Kinase inhibitor or inhibitor of kinase signaling pathway, as control: e.g., G¨o 6983 (Calbiochem) Control construct: phospho-acceptor mutant (see Strategic Planning) Sterile 35-mm glass-bottom culture dishes (MatTek) Imaging setup (see Strategic Planning) Data acquisition software (e.g., Metaßuor from Molecular Devices and Microsoft Excel) Assay reporter activity 1. Plate adherent cells in 2 ml of medium onto sterile 35-mm glass-bottom imaging dishes. As an example, plate ∼5 × 105 HeLa cells in a 35-mm dish in 2 ml of cell culture medium. A number of reagents exist that can be used to coat imaging dishes to help less adherent cells attach and adopt a more spread-out morphology. These include Þbronectin, polyD-lysine, Matrigel, and collagen, among others, and most are compatible with imaging; however potential background ßuorescence should be assessed for each matrix prior to the experiment.

2. At a time point 12 to 24 hr later, transfect cells with DNA encoding the kinase-activity reporter. For HeLa cells, transfection of 1 μg reporter DNA with FuGENE 6 according to the manufacturer’s protocol transfects a high percentage of cells that can be imaged within the next 24 hr. Coexpression with a kinase will boost the rate and magnitude of the response for initial screening and future speciÞcity checks.

3. At a time point 12 to 24 hr after transfection, aspirate medium from the cells, rinse in HBSS containing 1 mM CaCl2 , and replace with 2 ml HBSS containing 1 mM CaCl2 . 4. Using an imaging setup as described in Strategic Planning, select cells expressing the reporter at levels comparable to those of cellular substrates (approximately 1 μM as determined in the “calibration” section of Strategic Planning) and acquire one series of images. 5. Subtract background levels estimated from areas with no cells or untransfected cells. 6. Begin the experiment, plotting both the FRET ratio (FRET/CFP) and YFP intensity in real time. Acquisition and quantiÞcation of YFP intensity serves as an important control for photobleaching or reporter movement during the experimental protocol. Kinase Signaling Visualized by Targeted Reporters

22 Volume 1

Current Protocols in Chemical Biology

7. After acquiring a stable baseline FRET ratio, add an activator of the kinase signaling pathway. As an example, addition of 10 μM histamine to HeLa cells robustly stimulates activation of a Gq/PLC signal transduction pathway. FRET reporters offer an unambiguous real-time readout of kinase activity following acute treatment of cells with activators or inhibitors of the signaling pathway.

8. As an important experimental control, add a kinase inhibitor, or an inhibitor of the kinase signaling pathway, to return the FRET ratio to baseline levels following a stable response. This control validates that the FRET ratio change observed resulted from activation of the kinase of interest. For example, in validating the PKC reporter (CKAR), 250 nM G¨o 6983 was added to the imaging dish after CKAR attained a maximum FRET change in order to inhibit PKC and monitor reversal of the FRET change due to dephosphorylation of CKAR (Violin et al., 2003). The inhibitor could also be added prior to the experiment to block phosphorylation by the kinase (see Violin et al., 2003).

9. As another important control, image the phospho-acceptor mutant, i.e., the ”control construct” designed as described in Strategic Planning. Only the reporter with the phospho-acceptor site intact within the substrate sequence should undergo a FRET change; this control construct should not display any change in FRET.

10. Upon completion of the experiment, consolidate and graph the data. The baseline FRET ratio varies from one cell to another, so normalize the traces to the baseline FRET ratio from each region of interest. It is recommended to include data from three independent experiments, normalize each trace to the baseline reading, and reference them to the time of ligand addition. The Support Protocol describes how to determine the dynamic range of a ßuorescent reporter, knowledge of which can be important in interpreting FRET data.

DETERMINING THE DYNAMIC RANGE OF A FLUORESCENT REPORTER The change in the FRET ratio from a reporter has a maximal range. Knowing the full range can help in interpreting data during the course of experiments. For example, in experiments where a response is seemingly minute, this may be interpreted as weak activation of the kinase; however, it might also be a reßection of high basal reporter phosphorylation. That is, if basal kinase activity is high, then a portion of the reporter pool will be basally phosphorylated such that further phosphorylation is limited. Knowing the full range of the reporter helps in interpreting the experiments and also helps determine the efÞciency of reporter signaling at different subcellular locations. In Gallegos et al. (2006), this is clearly portrayed in analysis of CKAR at Þve subcellular regions (Fig. 3). The data suggest that the outer membrane mitochondria reporter is not functioning optimally.

SUPPORT PROTOCOL

Additional Materials (also see Basic Protocol) Phosphatase inhibitors that have been determined to inhibit phosphatases that act on the reporter’s substrate sequence, e.g., calyculin A (Calbiochem) 1. In cells expressing the kinase activity reporter, stimulate kinase activity until the maximal FRET ratio change is attained. This is the stimulated kinase activity.

Kinase Signaling Visualized by Targeted Reporters

23 Current Protocols in Chemical Biology

Volume 1

0.25

Average FRET ratio change

0.20

0.15 phosphatase-suppressed PKC activity PdBu-stimulated PKC activity

0.10

basal PKC activity 0.05

0.00 PM

Golgl

Cyto

Mito

Nuc

Figure 3 Quantitation of the basal, stimulated, and phosphatase-suppressed PKC activities at speciÞc regions, showing the range of the C Kinase Activity Reporter (CKAR) at each cellular location. The reporter has a consistent maximal range (20% FRET ratio change) at plasma membrane (PM), Golgi, cytosol, and nucleus, but a reduced range at the outer membrane of mitochondria. (Þgure from Gallegos et al., 2006).

2. Treat the cells that reached their maximal change following kinase activation (from step 1) with phosphatase inhibitors that have been determined to inhibit phosphatases that act on the reporter’s substrate sequence. Blocking phosphatases prevents dephosphorylation and effectively leads to maximal accumulation of phosphorylated reporters. This difference in the FRET ratio is the phosphatase-suppressed activity.

3. In cells expressing the kinase activity reporter, treat the cells with a kinase inhibitor (or inhibitor of the kinase pathway). If basal phosphorylation is present, inhibition of the kinase pathway will allow phosphatases to reverse this phosphorylation, and the FRET ratio will drop to a new level. This difference reßects the basal phosphorylation of the reporter.

4. Calculate the sum of the FRET ratio changes from steps 1, 2, and 3 to reveal the dynamic range of the reporter. In the authors’ experience, reporters utilizing the same PBD—e.g., Akt/B Kinase Activity Reporter (BKAR), C Kinase Activity Reporter (CKAR), and D Kinase Activity Reporter (DKAR), all using an FHA2 domain—have the same maximum range (Kunkel et al., 2005, 2007; Violin et al., 2003).

COMMENTARY Background Information

Kinase Signaling Visualized by Targeted Reporters

As kinase research is becoming increasingly focused on signaling at the subcellular level, assays to determine the spatiotemporal resolution of kinase activity are needed. Assessing local phosphorylation events by immunoblotting or through activity assays is critically dependent on cell fractionation methods

used to isolate localized substrates or kinases. The design of genetically encoded kinase activity reporters was the Þrst step in designing a tool to examine local kinase activity. There are many beneÞts to using genetically encoded reporters. First, a simple transfection of a plasmid is all that is needed to express the reporter in live cells; thus, it is a nondestructive way to

24 Volume 1

Current Protocols in Chemical Biology

introduce the reporter into cells. Second, these reporters can be targeted to subcellular locations through the addition of short targeting sequences (see Anticipated Results for discussion of example targeting experiments). FRET changes observed from targeted reporters reßect phosphorylation events in discrete cellular microenvironments. Using these tools, one can observe the kinetics, persistence, and level of the response at distinct intracellular regions.

Critical Parameters A critical parameter is the ability of the phosphorylated reporter to undergo a FRET change with a sufÞciently high-signal-to-noise ratio. A reporter exhibiting a dynamic range of 20% or higher will yield unambiguous results. Another critical consideration with kinase activity reporters is that the response is speciÞc to the kinase of interest; preventing or reversing the FRET response with a kinase inhibitor is an important control.

Troubleshooting Reporter does not undergo a FRET change Overexpression of the kinase along with the kinase activity reporter should result in the most robust signal. If no FRET change is observed, there are a number of things to test. 1. Verify that the kinase is, in fact, activated under the activating conditions, using immunoblotting or an activity assay. 2. Is the reporter already maximally phosphorylated? If so, addition of kinase inhibitors should cause a change in the FRET ratio. 3. Determine whether the reporter is phosphorylated at the phosphoacceptor site by the kinase by performing an in vitro kinase assay using puriÞed reporter protein and pure kinase. If an antibody to the phosphorylated consensus sequence exists, one can use that to assess phosphorylation at the phosphoacceptor site. Alternatively, the kinase assay will need to include [32 P]ATP such that the amount of radioactivity incorporated into the reporter can be quantiÞed. In both cases, a phosphorylation-resistant reporter (control construct in Strategic Planning) will need to be used as a control. 4. If the reporter is phosphorylated in vitro, verify that phosphorylation induces a change in the relative distance and/or orientation between the FRET pair. This can also be done in vitro by performing a ßuorescence emission scan of the pure reporter protein before and after phosphorylation by the kinase (e.g., Violin et al., 2003; Kunkel et al., 2005).

5. If the reporter is phosphorylated, but does not undergo a FRET change, then one can try techniques listed below to optimize the reporter. Testing reporter speciÞcity The speciÞcity of a kinase reporter is critical to its use. As described in the Strategic Planning section on determining an optimal consensus sequence, computational methods predict kinases most likely to phosphorylate an input sequence. As it is impractical to perform an exhaustive test of every kinase against the reporter, at a minimum, the speciÞcity of the predicted kinases should be examined. To test these, express the kinase activity reporter and image the FRET ratio following stimulation of the cells with an activator of the nonspeciÞc kinase. Should an unwanted kinase efÞciently phosphorylate the reporter, one must adapt by changing the substrate peptide sequence. With knowledge of the substrate speciÞcity of the unwanted kinase, the consensus sequence can be tailored such that it disfavors recognition by the unwanted kinase, yet still maintains efÞcient phosphorylation by the kinase of interest. While this can be done by imaging the FRET ratio changes in cells, the speciÞcity of the reporter can also be assessed by performing a kinase assay in vitro using puriÞed reporter and pure kinase. Maximal FRET ratio change is inconsistent at different cellular compartments The maximal FRET ratio change is likely to be that observed from the reporter in the cytosol. If the maximal change at a subcellular location is less than that observed elsewhere, there may be restricted ability of the reporter to undergo the conformational change when tethered there. One can try adding or altering linker sequences between the reporter and the targeting sequence. In the case of CKAR targeted to mitochondrial outer membrane, it was determined via immunoblotting that some of the reporters were being cleaved, and this affected the apparent FRET ratio change observed (unpub. observ.). If reporter cleavage is observed, mutation of the domain or speciÞc sequence recognized by the proteases will remedy the problem. Optimizing the reporter for maximal FRET ratio change The larger the FRET ratio change, the more sensitive and reliable the experiment can be, so many researchers have tried to optimize their reporters. The Þrst thing to try is to change

Kinase Signaling Visualized by Targeted Reporters

25 Current Protocols in Chemical Biology

Volume 1

the length of the linker sequences between the different modules. The PBD needs to be able to bind the phosphorylated substrate peptide sequence, so there must be some ßexibility between these two domains. Second, FRET depends on the distance and the orientation between the two ßuorophores, so the ßexibility of the linkers can also play a role. As a starting point, linkers including the amino acid sequence GGSGG or GGTGG can be used. Lastly, the ability to create functional, circularly permuted GFPs provides another possible way to potentially increase FRET between FRET pairs. Circularly permutated GFP (and its variants) are proteins in which the amino and carboxyl portions of GFP are joined with a short spacer connecting the original termini (Baird et al., 1999). These proteins thus contain new termini derived from internal sequences to GFP, and they still ßuoresce. Importantly, when inserted into the modular reporter, the position and orientation of the FRET pair is changed, thereby impacting FRET between the reporter pairs. Allen and Zhang (2006) experimentally manipulated all of these parameters and were able to improve their PKA activity reporter, AKAR2, to produce a twofold increase in the FRET ratio change in their improved PKA reporter, AKAR3. This twofold increase is a signiÞcant improvement in signal.

Anticipated Results This protocol describes methods the authors have used to generate and assay targeted kinase activity reporters. With the advent of novel FRET pairs, the possibilities of coimaging one reporter alongside another within the same cell is an exciting future avenue. To do so, one needs the appropriate FRET pairs that have minimal or no overlap in their excitation/emission wavelengths, along with the appropriate Þlter sets to image them, or one could envision a system where two FRET pairs share a common ßuorophore (Ai et al., 2008; Carlson and Campbell, 2009). In addition, the modular design for the reporter presented here can be adapted to read out other cellular functions, and if using the appropriate FRET pairs, one could co-image them in the same cell as a kinase activity reporter. For example, other FRET-based genetically encoded reporters of similar design include reporters of GTPase activity (Hodgson et al., 2008), protease activity (Xu et al., 1998), and Ca2+ levels (Miyawaki et al., 1997). Imaging cellular activities at speciÞc intracellular locations provides unprecedented opportunities for understanding cell signaling. Targeting genetically encoded reporters The ability to poise reporters at speciÞc intracellular locations allows one to image

Table 2 Targeted Sequences Used to Target Genetically Encoded Reporters

Kinase Signaling Visualized by Targeted Reporters

Subcellular region

Targeting sequences

References

Cytosol (nuclear-excluded)

LALKLAGLDI at C-terminus

Gallegos et al. (2006)

Nucleus

PKKRKVEDA at C-terminus

Gallegos et al. (2006); Ananthanarayanan et al. (2005)

Mitochondrial outer membrane

33 N-terminal residues of TOM20 at N-terminus

Sasaki et al. (2003); Gallegos et al. (2006)

Mitochondrial matrix

4 copies of 32 N-terminal residues of cytochrome c oxidase at N-terminus

Palmer et al. (2006)

Golgi

33 amino-terminal residues of eNOS at N-terminus

Sasaki et al. (2003); Gallegos et al. (2006)

ER

MLLPVLLLGLLGAAAD at N-term and KDEL at C-terminus

Palmer et al. (2004)

Plasma membrane

MGCIKSK at N-terminus

Violin et al. (2003); Kunkel et al. (2005); Gallegos et al. (2006)

PDZ domain-containing proteins

PDZ binding motif to C-terminus

Kunkel et al. (2009)

26 Volume 1

Current Protocols in Chemical Biology

cytosol (nuclear-excluded)

nucleus

mitochondria

Golgi

plasma membrane

Figure 4 Targeting kinase activity reporters to subcellular regions. Images of targeted reporters utilizing some the targeting sequences from Table 2.

activity in discrete cellular microenvironments. This tethering to deÞned intracellular compartments is achieved through the addition of short targeting sequences. Listed in Table 2 are examples of targeting sequences that have been successfully used to target genetically encoded FRET reporters to speciÞc intracellular locations. Figure 4 shows some representative examples of subcellular targeting. Using such targeting strategies, one can observe the kinetics, persistence, and level of the response at distinct intracellular regions. An example of the differential dynamics of kinase signaling present at the subcellular level is described in Gallegos et al. (2006). Therein, the PKC activity reporter was targeted to the cytosol, nucleus, Golgi, outer membrane of mitochondria, and plasma membrane, and each subcellular location revealed a distinct signature of kinase activity.

Time Considerations Once the reporter DNA is made, the Basic Protocol to image kinase activity will take 3 days. Plating cells on day 1 will take about 15 min. Transfecting cells on day 2 will take 10 to 30 min, depending on the reagent used. On day 3, imaging each dish takes between 30 and 60 min. It takes 5 to 10 min to prepare the dish on the microscope and to Þnd cells expressing the correct concentration of reporter. It takes another 5 to 10 min to deÞne the regions of interest and to begin the software protocol to acquire a time-lapse series of images. Once the imaging has started, it is important to acquire 5 min of baseline FRET ratio before activating the kinase signaling pathway. Depending on the level of active kinase in the cell, the level of phosphatases, and the mode of signaling (some of the responses we have observed, such as from DKAR in response to histamine, are transient), it will take about 15 min to complete a single imaging protocol. Typically, one images multiple cells per dish

in duplicate along with the necessary control dishes (e.g., phospho-acceptor mutant, kinase inhibitor pretreatment) such that an entire series of dishes would take 3 to 4 hr. Analysis of each experiment using Microsoft Excel (or similar spreadsheet application) takes about 15 min per dish.

Acknowledgement This work is supported by NIH GM 43154 and NIH P01 DK 54441.

Literature Cited Ai, H.W., Hazelwood, K.L., Davidson, M.W., and Campbell, R.E. 2008. Fluorescent protein FRET pairs for ratiometric imaging of dual biosensors. Nat. Methods 5:401-403. Allen, M.D. and Zhang, J. 2006. Subcellular dynamics of protein kinase A activity visualized by FRET-based reporters. Biochem. Biophys. Res. Commun. 348:716-721. Ananthanarayanan, B., Ni, Q., and Zhang, J. 2005. Signal propagation from membrane messengers to nuclear effectors revealed by reporters of phosphoinositide dynamics and Akt activity. Proc. Natl. Acad. Sci. U.S.A. 102:15081-15086. Baird, G.S., Zacharias, D.A., and Tsien, R.Y. 1999. Circular permutation and receptor insertion within green ßuorescent proteins. Proc. Natl. Acad. Sci. U.S.A. 96:11241-11246. Campbell, R.E., Tour, O., Palmer, A.E., Steinbach, P.A., Baird, G.S., Zacharias, D.A., and Tsien, R.Y. 2002. A monomeric red ßuorescent protein. Proc. Natl. Acad. Sci. U.S.A. 99:7877-7882. Carlson, H.J. and Campbell, R.E. 2009. Genetically encoded FRET-based biosensors for multiparameter ßuorescence imaging. Curr. Opin. Biotechnol. 20:19-27. Dunn, T.A., Wang, C.T., Colicos, M.A., Zaccolo, M., DiPilato, L.M., Zhang, J., Tsien, R.Y., and Feller, M.B. 2006. Imaging of cAMP levels and protein kinase A activity reveals that retinal waves drive oscillations in second-messenger cascades. J. Neurosci. 26:12807-12815. Durocher, D., Taylor, I.A., Sarbassova, D., Haire, L.F., Westcott, S.L., Jackson, S.P., Smerdon, S.J., and Yaffe, M.B. 2000. The molecular basis

Kinase Signaling Visualized by Targeted Reporters

27 Current Protocols in Chemical Biology

Volume 1

of FHA domain: Phosphopeptide binding speciÞcity and implications for phospho-dependent signaling mechanisms. Mol. Cell 6:1169-1182.

2006. Ca2+ indicators based on computationally redesigned calmodulin-peptide pairs. Chem. Biol. 13:521-530.

Gallegos, L.L., Kunkel, M.T., and Newton, A.C. 2006. Targeting protein kinase C activity reporter to discrete intracellular regions reveals spatiotemporal differences in agonist-dependent signaling. J. Biol. Chem. 281:30947-30956.

Rizzo, M.A., Springer, G.H., Granada, B., and Piston, D.W. 2004. An improved cyan ßuorescent protein variant useful for FRET. Nat. Biotechnol. 22:445-449.

Griesbeck, O., Baird, G.S., Campbell, R.E., Zacharias, D.A., and Tsien, R.Y. 2001. Reducing the environmental sensitivity of yellow ßuorescent protein: Mechanism and applications. J. Biol. Chem. 276:29188-29194. Hodgson, L., Pertz, O., and Hahn, K.M. 2008. Design and optimization of genetically encoded ßuorescent biosensors: GTPase biosensors. Methods Cell Biol. 85:63-81. Kunkel, M.T., Ni, Q., Tsien, R.Y., Zhang, J., and Newton, A.C. 2005. Spatio-temporal dynamics of protein kinase B/Akt signaling revealed by a genetically encoded ßuorescent reporter. J. Biol. Chem. 280:5581-5587. Kunkel, M.T., Toker, A., Tsien, R.Y., and Newton, A.C. 2007. Calcium-dependent regulation of protein kinase D revealed by a genetically encoded kinase activity reporter. J. Biol. Chem. 282:6733-6742. Kunkel, M.T., Garcia, E.L., Kajimoto, T., Hall, R.A., and Newton, A.C. 2009. The protein scaffold NHERF-1 controls the amplitude and duration of localized protein kinase D activity. J. Biol. Chem. 284:24653-24661. Miyawaki, A., Llopis, J., Heim, R., McCaffery, J.M., Adams, J.A., Ikura, M., and Tsien, R.Y. 1997. Fluorescent indicators for Ca2+ based on green ßuorescent proteins and calmodulin. Nature 388:882-887. Palmer, A.E., Jin, C., Reed, J.C., and Tsien, R.Y. 2004. Bcl-2-mediated alterations in endoplasmic reticulum Ca2+ analyzed with an improved genetically encoded ßuorescent sensor. Proc. Natl. Acad. Sci. U.S.A. 101:17404-17409. Palmer, A.E., Giacomello, M., Kortemme, T., Hires, S.A., Lev-Ram, V., Baker, D., and Tsien, R.Y.

Sasaki, K., Sato, M., and Umezawa, Y. 2003. Fluorescent indicators for Akt/protein kinase B and dynamics of Akt activity visualized in living cells. J. Biol. Chem. 278:30945-30951. Sato, M., Kawai, Y., and Umezawa, Y. 2007. Genetically encoded ßuorescent indicators to visualize protein phosphorylation by extracellular signal-regulated kinase in single living cells. Anal. Chem. 79:2570-2575. Shaner, N.C., Steinbach, P.A., and Tsien, R.Y. 2005. A guide to choosing ßuorescent proteins. Nat. Methods 2:905-909. Turk, B.E., Hutti, J.E., and Cantley, L.C. 2006. Determining protein kinase substrate speciÞcity by parallel solution-phase assay of large numbers of peptide substrates. Nat. Protoc. 1:375-379. Violin, J.D. 2003. Spatiotemporal Dynamics of Protein Kinase C Signaling. Doctoral dissertation. University of California, San Diego, La Jolla, California. Violin, J.D., Zhang, J., Tsien, R.Y., and Newton, A.C. 2003. A genetically encoded ßuorescent reporter reveals oscillatory phosphorylation by protein kinase C. J. Cell. Biol. 161:899-909. Xu, X., Gerard, A.L., Huang, B.C., Anderson, D.C., Payan, D.G., and Luo, Y. 1998. Detection of programmed cell death using ßuorescence energy transfer. Nucleic Acids Res. 26:2034-2035. Zhang, J., Ma, Y., Taylor, S.S., and Tsien, R.Y. 2001. Genetically encoded reporters of protein kinase A activity reveal impact of substrate tethering. Proc. Natl. Acad. Sci. U.S.A. 98:14997-15002. Zhang, J., Hupfeld, C.J., Taylor, S.S., Olefsky, J.M., and Tsien, R.Y. 2005. Insulin disrupts betaadrenergic signalling to protein kinase A in adipocytes. Nature 437:569-573.

Kinase Signaling Visualized by Targeted Reporters

28 Volume 1

Current Protocols in Chemical Biology

Activity-Based Protein ProÞling (ABPP) and Click Chemistry (CC)–ABPP by MudPIT Mass Spectrometry Anna E. Speers1 and Benjamin F. Cravatt1 1

The Skaggs Institute for Chemical Biology and Department of Physiological Chemistry, The Scripps Research Institute, La Jolla, California

ABSTRACT Activity-based protein proÞling (ABPP) is a chemical proteomic method for functional interrogation of enzymes within complex proteomes. This unit presents a protocol for in vitro and in vivo labeling using ABPP and Click Chemistry (CC)-ABPP probes for indepth proÞling using the Multi-dimensional Protein IdentiÞcation Technology (MudPIT) C 2009 by John Wiley & Sons, Inc. analysis platform. Curr. Protoc. Chem. Biol. 1:29-41  Keywords: activity-based protein proÞling r ABPP r click chemistry r mass spectrometry r MudPIT r activity-based probes r biotin r alkyne r azide

INTRODUCTION The Þeld of proteomics aims to characterize and assign function to the tens of thousands of eukaryotic and prokaryotic proteins annotated by genome sequencing efforts. In contrast to global platforms such as 2-D gel electrophoresis (Patton et al., 2002), shotgun LC-MS analysis (Gygi et al., 1999; Washburn et al., 2001), yeast two-hybrid screening (Ito et al., 2002), and protein microarrays (MacBeath, 2002), which analyze proteins based on abundance, activity-based protein proÞling (ABPP; Jessani and Cravatt, 2004; Cravatt et al., 2008) is a chemical proteomic strategy for the analysis of enzyme function within complex biological systems. ABPP utilizes active site–directed chemical probes consisting of two elements: (1) an active site–directed reactive group for binding and covalently labeling a speciÞc subset (or family) of catalytically related enzymes, and (2) a reporter tag (e.g., ßuorophore or biotin) for detection/quantiÞcation and/or enrichment/identiÞcation of labeled enzymes. Because ABPP probes selectively label active enzymes, but not their inactive forms (Jessani et al., 2002; Jessani and Cravatt, 2004), they allow monitoring of changes in enzyme activities resultant from post-translational modiÞcation and/or protein-protein/protein-small molecule interactions that occur without corresponding changes in protein abundance or mRNA expression (Kobe and Kemp, 1999). To date, ABPP probes have been generated for more than a dozen enzyme classes (Evans and Cravatt, 2006; Cravatt et al., 2008; Paulick and Bogyo, 2008). Because most ABPP probes have limited cell permeability due to their bulky reporter tag, an ABPP protocol typically involves homogenization of the proteomic source (e.g., cells, tissue) prior to labeling, which disrupts the native cellular environment, potentially compromising the endogenous activity of certain enzymes. For applications requiring the interrogation of enzyme activities in living cells or organisms, the reporter group is substituted by a small, latent chemical handle (alkyne or azide), which does not impede cell permeability. An orthogonally functionalized reporter tag is then appended to the probe post-homogenization using click chemistry, speciÞcally, the Cu(I)-catalyzed stepwise version of Huisgen’s azide-alkyne cylcloaddition (Kolb and Sharpless, 2003; Speers and Cravatt, 2004).

Current Protocols in Chemical Biology 1: 29-41, December 2009 Published online December 2009 in Wiley Interscience (www.interscience.wiley.com). DOI: 10.1002/9780470559277.ch090138 C 2009 John Wiley & Sons, Inc. Copyright 

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

29 Volume 1

reactive group

biotin

activity-based probe labeling

streptavidin affinity purification

proteome

labeled proteome

MudPIT

site-of-labeling analysis:

labeling site identification

MudPIT

trypsin digest

elute labeled peptides

target identification and quantification

Figure 1 The ABPP-MudPIT method for high-content proteomic analysis of enzyme activities. A proteome is labeled with a biotinylated ABPP probe, and labeled proteins are afÞnity enriched on streptavidin beads. After on-bead trypsin digest, the tryptic peptides are analyzed by MudPIT for identiÞcation and quantiÞcation (e.g., via spectra counting). Box: The probe-labeled peptides can also be eluted from the streptavidin beads for MS analysis of labeling sites.

Following labeling, proteins are enriched using immobilized streptavidin and subjected to on-bead trypsin digestion. As outlined in Figure 1, the resulting peptides are then separated by multidimensional liquid chromatography and analyzed by tandem mass spectrometry (MudPIT), which provides both protein identiÞcations and an estimation of labeled (active) protein abundance using semiquantitative methods such as spectral counting (Jessani et al., 2005; Sieber et al., 2006; Alexander and Cravatt, 2006). The biotinylated active-site peptides still bound to the beads can also be eluted (see Background Information) and analyzed by MudPIT to identify sites of probe labeling (Fig. 1). For standard ABPP using biotinylated probes, follow Basic Protocol 1; for proÞling enzyme activities in living cells or mice using alkyne probes, use the Alternate Protocol. Both of these protocols will generate labeled proteomic samples for enrichment, digestion, and MS analysis according to Basic Protocol 2. Support Protocol 1 details a general method for preparation of membrane and soluble proteomic samples from cultured cells or tissues. Support Protocol 2 provides a detailed method for methanol-chloroform precipitation following the click chemistry (CC) reaction, which can be used if incomplete precipitation of proteins is observed. BASIC PROTOCOL 1

LABELING ENZYMES IN VITRO BY ABPP The following is a standard protocol for in vitro labeling of cell or tissue homogenates using biotinylated ABPP probes. A no-probe control sample should always be prepared for comparison to the experimental sample. Other types of controls, such as competition with a native ligand or a nonbiotinylated probe, may also be useful, especially in cases where the probe is not class speciÞc and may react with a wide variety of enzymes, making speciÞc and nonspeciÞc labeling events more difÞcult to distinguish. NOTE: this protocol speciÞes the use of Tris buffer; however, Ca- and Mg-free D-PBS may be substituted. If using other buffer conditions, ensure compatibility of ABPP probe labeling reaction and enrichment protocol.

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

30 Volume 1

Current Protocols in Chemical Biology

Materials 1 mg protein/ml proteome source (e.g., cell or tissue homogenate; see Support Protocol 1) 100× (0.5 to 2 mM) biotinylated ABPP probe stock in dimethylsulfoxide (DMSO) (for details of synthesis, see Evans and Cravatt, 2006; Cravatt et al., 2008; and Paulick and Bogyo, 2008; store up to several years at –20◦ or –80◦ C) Dimethylsulfoxide (DMSO) 10% (v/v) Triton X-100 (store up to several months at room temperature) 50 mM Tris·Cl, pH 8.0 (store up to several months at 4◦ C) 10% (w/v) SDS in water End-over-end rotator 10DG disposable chromatography columns (BioRad) 15-ml conical centrifuge tubes 90◦ C heating block Label sample 1. Aliquot two 1-ml portions (for experimental and control samples) of 1 mg/ml cell or tissue homogenate (see Support Protocol 1) into microcentrifuge tubes. 2. To the experimental sample, add 10 μl of 100× probe stock, giving a Þnal concentration of 5 to 20 μM. To the control sample, add 10 μl of DMSO. Optimal probe concentration depends on proteome source and probe reactivity. Unless otherwise noted, all subsequent steps are the same for both the experimental and control samples.

3. Vortex samples and incubate for 1 hr at room temperature.

Remove excess probe 4. If samples are membrane fractions, add 100 μl of 10% Triton X-100, yielding a Þnal concentration of 0.9%, and rotate for 1 hr at 4◦ C on an end-over-end rotator. 5. Transfer each sample to a 15-ml conical centrifuge tube and bring volume to 2.5 ml with 50 mM Tris·Cl, pH 8.0. 6. Apply sample to 10DG column (pre-equilibrated with 25 ml of 50 mM Tris·Cl, pH 8.0) and discard ßow-through. Elute proteins with 3.5 ml of 50 mM Tris·Cl, pH 8.0, and collect eluate in 15-ml conical tube. 7. Add 184 μl of 10% SDS to the eluate, giving a Þnal concentration of 0.5%. 8. Heat for 8 min at 90◦ C in a heating block. CAUTION: Loosen caps before heating.

9. Allow samples to cool to room temperature. Proceed directly to Basic Protocol 2 or freeze samples at −20◦ C.

LABELING ENZYMES IN LIVING CELLS OR MICE BY CLICK CHEMISTRY–ACTIVITY-BASED PROTEIN PROFILING (CC-ABPP) This protocol details general methods for the in situ labeling of cells in culture or the in vivo labeling of mice via intraperitoneal (i.p.) injection using a CC-ABPP probe bearing a bio-orthogonal alkyne. Following homogenization (Support Protocol 1), a biotin-azide tag is then appended using CC. Note that (1) a no-probe control sample should always be prepared for comparison to the experimental sample; (2) if labeling enzymes with an azide probe and conjugating to a biotin-alkyne tag, the same protocol may be followed; (3) if labeling cell homogenates using CC probes, label the experimental sample aliquotted in

ALTERNATE PROTOCOL

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

31 Current Protocols in Chemical Biology

Volume 1

step 5 (below) with 5 to 20 μM probe-alkyne for 1 hr (using the same volume of DMSO for the control), and proceed directly to step 6.

Additional Materials (also see Basic Protocol 1) Cells growing in culture in 15-cm plates or laboratory mice 1000× (5 to 25 mM) probe-alkyne stock in DMSO (for details of synthesis, see Evans and Cravatt, 2006; Cravatt et al., 2008; and Paulick and Bogyo, 2008; store up to several years at −20◦ C), for cell culture experiment 1× Dulbecco’s phosphate-buffered saline (D-PBS; Invitrogen; without calcium and magnesium) 1 to 5 mg/ml probe-alkyne in vehicle (see recipe) Vehicle 5 mM biotin-azide (PEG4 carboxamide-6-azidohexanyl biotin; Invitrogen) in DMSO (store up to several years at −20◦ C) 50 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP; Fluka) in H2 O (prepare fresh prior to use) 1.7 mM TBTA (see recipe) 50 mM CuSO4 ·5H2 O in H2 O (store up to several months at room temperature; remake if precipitate forms) Methanol, cold 2.5% (w/v) SDS in Ca- and Mg-free D-PBS (store up to several months at room temperature) Cell scraper Refrigerated centrifuge Animal balance Probe sonicator 60◦ C water bath or heating block Additional reagents and equipment for preparing cell/tissue homogenates (Support Protocol 1) and methanol:chloroform precipitation (Support Protocol 2) NOTE: All protocols using live animals must Þrst be reviewed and approved by an Institutional Animal Care and Use Committee (IACUC) or must conform to governmental regulations regarding the care and use of laboratory animals.

Label proteome In cells 1a. Grow cells to 80% conßuence in appropriate medium (e.g., RPMI-1640 supplemented with 10% fetal bovine serum). Before labeling, rinse cell layer once with medium and then add 5 ml fresh medium. For a typical experiment, use one or two 15-cm plates for each experimental and control sample.

2a. To experimental sample, add 5 μl of 1000× probe-alkyne stock, giving a Þnal concentration of 5 to 25 μM. Swirl plate to mix. To control sample, add 5 μl of DMSO, swirl plate to mix. 3a. After incubating for 1 hr (or longer, as desired, typically not exceeding 24 hr) at 37◦ C, remove medium, rinse cell layer twice with Ca- and Mg-free D-PBS, and harvest cells in ∼5 ml Ca- and Mg-free D-PBS using a cell scraper. Collect cells in a 15-ml conical tube. Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

4a. Centrifuge cells for 5 min at 1000 × g, 4◦ C or room temperature, to pellet. Remove supernatant. Process cells (prepare homogenate; see Support Protocol 1) for biotinazide conjugation (step 5), or store pellet at −80◦ C (stable for months).

32 Volume 1

Current Protocols in Chemical Biology

In mice 1b. Prepare a solution (1 to 5 mg/ml) of probe-alkyne in vehicle as described in Reagents and Solutions. Optimal dosage depends on probe reactivity and bioavailability; a typical range is between 10 and 50 mg probe/kg mouse, corresponding to a 1 to 5 mg/ml solution.

2b. Weigh mice. Administer experimental mouse an intraperitoneal (i.p.) injection of 10 μl/g probe-alkyne in vehicle. Administer control mouse an i.p. injection of 10 μl/g vehicle only. 3b. After 1 hr, sacriÞce mice according to approved animal protocol and harvest tissue. 4b. Snap-freeze tissue on dry ice. Process tissue (prepare homogenate; see Support Protocol 1) for biotin-azide conjugation, or store samples at −80◦ C (stable for months).

Perform biotin-azide conjugation via click chemistry 5. For each experimental and control sample, aliquot two 0.5-ml portions of 2 mg/ml cell/tissue homogenate (prepared in Ca- and Mg-free D-PBS, see Support Protocol 1) into microcentrifuge tubes. The CC reaction works best in Ca- and Mg-free D-PBS. Avoid using buffers containing amines, as they interfere with the Cu(I)-stabilizing TBTA ligand. The CC reaction is less efÞcient with starting reaction volumes greater than ∼0.5 ml.

6. Add the following reagents and vortex sample after each addition: a. b. c. d.

11.3 μl of 5 mM biotin-azide, giving a Þnal concentration of 100 μM. 11.3 μl of 50 mM TCEP, giving a Þnal concentration of 1 mM. 34.0 μl of 1.7 mM TBTA, giving a Þnal concentration of 100 μM. 11.3 μl of 50 mM CuSO4 ·5H2 O, giving a Þnal concentration of 1 mM.

7. Incubate at room temperature for 1 hr, vortexing after 30 min. The majority of proteins will precipitate during this reaction.

Remove excess CC reagents 8. Combine respective experimental and control aliquots, and centrifuge 4 min at 6500 × g, 4◦ C, to pellet protein. Remove supernatant. 9. Add 750 μl cold methanol to the pellet and sonicate for 3 to 4 sec using a probe sonicator (∼30% power level) at 4◦ C to resuspend protein. Centrifuge for 4 min at 6500 × g, 4◦ C, and remove supernatant. Repeat methanol wash in this manner two times. Protein loss during the wash step rarely compromises analysis; however, if a speciÞc protein of interest is not observed during MudPIT analysis, then assess whether or not protein has been lost in the methanol wash by gel-based analysis (Kidd et al., 2001; Patricelli et al., 2001; Speers and Cravatt, 2004). If protein loss is observed, perform a methanol:chloroform precipitation (see Support Protocol 2) instead of the methanol wash described above, and continue with the next step.

10. Add 0.65 ml of 2.5% SDS in Ca- and Mg-free D-PBS to the protein pellet and sonicate three times, each time for 5 sec using the technique described in step 9. At this point, one may store samples at −20◦ C or proceed to next step of the protocol.

11. Heat sample for 5 min at 60◦ C in a water bath or heating block and repeat sonication two to three times, each time for 5 sec, to dissolve pellet. Centrifuge for 4 min at 6500

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

33 Current Protocols in Chemical Biology

Volume 1

× g, room temperature, to pellet any unsolubilized protein, and transfer supernatant to clean 15-ml conical tube. If a pellet larger than ∼20% of original pellet is observed (indicating that a substantial amount of protein has not been resolubilized), repeat heating and sonication steps. If necessary, heat for 2 to 5 min at 90◦ C.

12. Bring volume to 3.5 ml with Ca- and Mg-free D-PBS, yielding a Þnal SDS concentration of ∼0.5%. Freeze sample overnight at −20◦ C or proceed to Basic Protocol 2. SUPPORT PROTOCOL 1

PREPARATION OF CELLS/TISSUE HOMOGENATES The following is a general protocol for preparing membrane and soluble fractions from cells or tissue.

Additional Materials Harvested cells/tissue 1× Dulbecco’s phosphate-buffered saline (D-PBS; Invitrogen; without calcium and magnesium) Protein assay kit: e.g., DC protein assay (BioRad) Razor blade Glass plate or large tissue culture dish Dounce homogenizer Probe sonicator Refrigerated centrifuge Ultracentrifuge Insulin syringe 1a. For preparation of cells: Resuspend cell pellet in ∼4 volumes of Ca- and Mg-free D-PBS on ice and homogenize in a Dounce homogenizer using 10 to 15 strokes; alternatively, sonicate in a probe sonicator (two to four times,10 sec each at ∼30% power) on ice. 1b. For preparation of tissue: Mince tissue with razor blade on ice (e.g., on ice-cold glass plate or large tissue culture dish) and homogenize in a Dounce homogenizer using 15 to 20 strokes in ∼4 volumes of Ca- and Mg-free D-PBS on ice. 2. Centrifuge for 5 min at 1000 × g, 4◦ C, to pellet nuclei and unbroken cells. 3. Transfer supernatant to a clean ultracentrifuge tube and ultracentrifuge for 1 hr at 100,000 × g, 4 ◦ C, to separate membrane (pellet) and soluble (supernatant) fractions. Remove supernatant to clean tube. 4. Hold tube containing the pellet at an angle, and carefully add ∼0.5 ml Ca- and Mg-free D-PBS to the tube without disturbing the pellet. Remove liquid with a pipet, leaving the pellet intact. Resuspend pellet in approximately one-fourth to one-half the original volume with Ca- and Mg-free D-PBS using an insulin syringe. 5. Quantify protein concentration in membrane and soluble fractions and adjust with Ca- and Mg-free D-PBS to the desired concentration (1 to 2 mg/ml). For example, quantiÞcation may be accomplished using the DC protein assay (BioRad) according to the manufacturer’s protocol.

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

6. Divide samples into 1-ml aliquots and store at −80◦ C (stable for 6 to 9 months; avoid freeze-thaw cycles). 7. Prior to ABPP probe labeling (Basic Protocol 1), thaw sample on ice and brießy allow to warm to room temperature.

34 Volume 1

Current Protocols in Chemical Biology

METHANOL/CHLOROFORM PRECIPITATION OF PROTEINS If incomplete protein precipitation is observed following the CC reaction (e.g., if a protein of interest remains in solution or if detergents are used during the CC reaction), this protocol can be used where indicated in the Alternate Protocol.

SUPPORT PROTOCOL 2

Materials Methanol Chloroform 15-ml conical centrifuge tubes Refrigerated centrifuge 1. At the end of the CC reaction (see annotation to step 9 of the Alternate Protocol), combine aliquots for respective samples into 15-ml conical tubes, giving ∼1 ml Þnal volume. 2. Add 4 ml of methanol and 1 ml of chloroform, then vortex the sample. Add 3 ml of water and vortex the sample again. Centrifuge 15 min at 4000 × g, 4◦ C. Remove top and bottom layers, leaving behind the precipitated protein. After centrifugation, protein will precipitate at the interface of the aqueous and organic layers.

3. Transfer protein to microcentrifuge tube by resuspending it in 300 μl methanol and pipetting it into the smaller tube, then wash the 15-ml tube with an additional 300 μl methanol and add this to the microcentrifuge tube. 4. Add 150 μl chloroform and vortex the sample. Add 600 μl of water and vortex the sample again. Centrifuge for 5 min at 9000 × g, room temperature, and remove top and bottom layers, leaving behind the protein at the interface, as in step 2. 5. Add 600 μl of methanol and sonicate for 5 to 10 sec with probe sonicator (∼30% power level) at 4◦ C to resuspend (but not resolubilize) protein. Once fully precipitated, addition of methanol is much less likely to result in protein resolubilization.

6. Centrifuge 5 min at 9000 × g, room temperature, to pellet protein. Remove supernatant. 7. Proceed with Step 10 of the Alternate Protocol (addition of 0.65 ml 2.5% SDS in Ca- and Mg-free D-PBS).

ENRICHMENT AND DIGESTION OF PROBE-LABELED PROTEINS FOR MudPIT ANALYSIS

BASIC PROTOCOL 2

This protocol is used to generate streptavidin-enriched tryptic digests and probe-labeled peptides for MudPIT analysis using the labeled proteomes obtained from Basic/Alternate Protocol.

Materials SDS-solubilized protein sample (Basic Protocol 1 or the Alternate Protocol) 1× Dulbecco’s phosphate-buffered saline (D-PBS; Invitrogen; without calcium and magnesium) Streptavidin beads: 50% (v/v) slurry of Immunopure immobilized streptavidin (Pierce) 1% (w/v) SDS in Ca- and Mg-free D-PBS (store up to several months at room temperature)

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

35 Current Protocols in Chemical Biology

Volume 1

6 M and 2 M urea in Ca- and Mg-free D-PBS (prepare fresh prior to use) 200 mM dithiothreitol (DTT) in H2 O (prepare fresh daily, or store aliquots at −20◦ C for months) 500 mM iodoacetamide (IAA) in H2 O (prepare fresh daily, or store aliquots at −20◦ C for months) 100 mM CaCl2 in H2 O (store at room temperature for months) 0.5 mg/ml sequence-grade modiÞed trypsin (Promega) supplied in resuspension buffer 90% formic acid 50% (v/v) acetonitrile/0.1% trißuoroacetic acid (TFA) in H2 O (prepare fresh prior to use) 5% (v/v) formic acid (FA) in H2 O Empty Micro Bio-Spin Chromatography Columns (BioRad, cat. no. 732-6204) End-over-end rotator Low-adhesion screw-top microcentrifuge tubes (Sarstedt) Gel-loading pipet tips 65◦ C water bath or heat block SpeedVac evaporator Perform streptavidin enrichment 1. To the ∼3.5 ml volume of SDS-solubilized protein from Basic Protocol 1 or the Alternate Protocol, add 5 ml Ca- and Mg-free D-PBS to bring volume to 8.5 ml, yielding a Þnal SDS concentration of 0.2%. 2. To the protein sample from step 1, add a 50-μl aliquot (equal to 100 μl of a 50% slurry) of streptavidin beads that have been prewashed three times in a Micro Bio-Spin Chromatography Column, each time with 1 ml Ca- and Mg-free D-PBS, according to the manufacturer’s instructions for the use of the Bio-Spin Columns. Use a clean razor blade to cut the end off a standard 200-μl pipet tip to generate a wide-bore tip for transferring beads.

3. Rotate samples for 1 to 1.5 hr at room temperature on an end-over end rotator. Centrifuge sample for 2 min at 1400 × g, room temperature, to pellet beads. Remove ∼95% of supernatant. With remaining supernatant, transfer beads to labeled Bio-Spin column. Use a small volume of Ca- and Mg-free D-PBS to transfer any remaining beads to the Bio-Spin column. Make sure to label columns before transferring beads. Beads should not be vortexed, as they may break apart.

4. Wash the beads with the following solutions in the Bio-Spin column according to the manufacturer’s instructions: a. Three times, each time with 1 ml of 1% SDS. b. Three times, each time with 1 ml of 6 M urea. c. Three times, each time with 1 ml Ca- and Mg-free D-PBS. 5. Transfer beads in Ca- and Mg-free D-PBS to a low-adhesion screw-top tube. Screw-top tubes prevent sample loss during overnight incubation at elevated temperature, while low adhesion minimizes sample adhesion.

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

6. Centrifuge 2 min at 1400 × g (or pulse in a microcentrifuge at maximum speed), room temperature, to pellet beads. Remove supernatant with a gel-loading tip. Use a gel-loading tip or other Þne-bore tip to minimize bead loss.

36 Volume 1

Current Protocols in Chemical Biology

Perform on-bead reduction, alkylation, and digestion 7. Resuspend beads in 500 μl of 6 M urea in Ca- and Mg-free D-PBS. Add 25 μl of 200 mM DTT, yielding a Þnal concentration of 10 mM, and heat at 65◦ C in a water bath or heating block for 15 min. 8. Add 25 μl of 500 mM IAA (25 mM Þnal), and rotate for 30 min at room temperature under foil. Centrifuge for 2 min at 1400 × g. Remove supernatant. Wash beads once with ∼1 ml Ca- and Mg-free D-PBS. 9. Add the following reagents to the sample:

200 μl of 2 M urea in Ca- and Mg-free D-PBS 2 μl of 100 mM CaCl2 , (for 1 mM CaCl2 Þnal) 4 μl of 0.5 mg/ml trypsin (2 mg total) in resuspension buffer. 10. Rotate or shake overnight at 37◦ C. To rotate, use incubator with rotisserie, or plug tabletop rotator into an incubator with an internal outlet; if using an enclosed shaker, set on medium speed and lay tube ßat rather than upright.

Elute tryptic peptides for MS analysis 11. Centrifuge samples for 2 min at 1400 × g, room temperature, and transfer beads and supernatant to a Bio-Spin column. Elute solution (containing tryptic peptides) into a clean microcentrifuge tube by centrifuging the Bio-Spin column according to the manufacturer’s instructions. Transfer any remaining solution/beads into the Bio-Spin column with Ca- and Mg-free D-PBS (100 μl) and elute into the same collection tube, giving a total volume of 300 μl. Save the experimental sample beads if following steps 14 to 17 below. 12. To the eluate containing the tryptic peptides, add 17 μl of 90% formic acid, giving a Þnal concentration of 5%. Proceed directly to MS analysis or store tryptic digest samples at −80◦ C (will keep for several months or more). 13. Analyze by MudPIT. For additional details on analysis of tryptic peptides, see Link et al. (2003) and Weerapana et al. (2007).

Elute labeled peptides for MS analysis (optional) Note that it is not necessary to complete steps 14 to 17 unless analysis of probe labeling sites is desired. 14. Wash beads in a Bio-Spin column Þve times, each time with 1 ml Ca- and Mg-free D-PBS, then Þve times, each time with 1 ml water. Transfer beads to a clean, lowadhesion tube in water, spin brießy at maximum speed in a microcentrifuge, and remove supernatant. 15. To elute labeled peptides, add 150 μl of 50% acetonitrile/0.1%TFA to the beads and heat for 2 min at 65◦ C. Transfer beads back to Bio-Spin column, and elute into a clean low-adhesion tube. Add an additional 100 μl of 50% acetonitrile/0.1%TFA to the Bio-Spin column and elute into same collection tube. Dry sample in a SpeedVac evaporator. 16. Resuspend peptides in 100 μl of 5% formic acid and vortex 5 min. Proceed directly to MS analysis or store samples at −80◦ C (stable for up to several months). 17. Analyze by MudPIT. For additional details on analysis of labeled peptides, see Link et al. (2003) and Weerapana et al. (2007).

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

37 Current Protocols in Chemical Biology

Volume 1

REAGENTS AND SOLUTIONS Use Milli-Q-puriÞed water or equivalent for all aqueous solutions used in this unit; for suppliers, see SUPPLIERS APPENDIX.

Probe-alkyne in vehicle (prepare fresh prior to use) Prepare a solution of 18:1 (v/v) saline:castor oil as follows. Prepare a saline solution of 1% (w/v) NaCl in water. Add castor oil (Fisher) to give a ratio of 18:1 (v/v) saline:castor oil. Store at room temperature for days or 4◦ C for weeks. Prepare a stock of probe-alkyne in ethanol as follows. Dissolve probe-alkyne (for details of synthesis see Evans and Cravatt, 2006; Cravatt et al., 2008; and Paulick and Bogyo, 2008) in ethanol at a concentration of X mg/ml, where X = mg/kg × 2 (e.g., if desired dosage is 10 mg/kg, then concentration would be 20 mg/ml; typical dosage is 10 to 50 mg/kg). Assuming a maximum mouse weight of 30 g, prepare 15 μl of ethanol solution per mouse. Prepare a probe-alkyne solution in 18:1:1 saline:castor oil:ethanol as follows. Dilute ethanol solution 20× with 18:1 saline:castor oil, giving a Þnal probe-alkyne concentration of X/20 mg/ml (e.g., 1 mg/ml from a 20 mg/ml ethanol stock) in 18:1:1 saline:castor oil:ethanol. Ten μl of this solution should be injected per gram of mouse.

TBTA, 1.7 mM Prepare a 50× (83.5 mM) stock of tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA; Aldrich) by dissolving 8.85 mg TBTA in 200 μl DMSO (store up to several years at room temperature). Prepare a 1.7 mM TBTA solution by adding 20 μl of 50× stock to a glass vial containing 180 μl of DMSO and mixing thoroughly. Add 800 μl of t-butanol. Thoroughly mix the diluted TBTA before adding the t-butanol, to avoid precipitation. The resulting solution is 1.7 mM in 4:1 t-butanol/DMSO (store at room temperature up to several months; prepare a new solution if crystallization or precipitation is observed). The CC reaction is aided by the presence of t-butanol. This solvent is included in the TBTA stock rather than added to the CC reaction by itself, because it has a melting point of 25.5◦ C, and thus tends to freeze at room temperature. Mixing with DMSO prevents freezing, and including it with the TBTA minimizes the number of solutions that need to be added to the CC reaction.

COMMENTARY Background Information

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

ABPP experiments can be analyzed using a variety of different platforms besides the MudPIT approach described in this unit (Cravatt et al., 2008; Bachovchin et al., 2009). Originally, probe-labeled samples were separated by 1-D SDS-PAGE and visualized by in-gel ßuorescence scanning (for ßuorophoreconjugated proteins) or avidin blotting (for biotinylated proteins) (Liu et al., 1999; Kidd et al., 2001; Patricelli et al., 2001; Greenbaum et al., 2002). For target identiÞcation, biotin-labeled proteins were enriched prior to SDS-PAGE, and bands corresponding to labeled proteins excised for in-gel digestion and MS analysis. Gel-based ABPP offers a robust

and high-throughput platform, capable of analyzing hundreds of proteomes per day, and is still the preferred method for the comparative analysis of many samples in parallel. In contrast, ABPP-MudPIT is signiÞcantly more time-intensive, and is most applicable to the in-depth analysis of dozens, rather than hundreds, of samples. An additional consideration is the amount of sample required, which is 0.5 to 1.0 mg for a MudPIT experiment versus 0.01 to 0.02 mg for gel-based ABPP. As such, ABPP-MudPIT may not be applicable in certain situations of limited protein quantity, such as with clinical biopsy samples. The principal drawback of gel-based analysis is the limited resolution of SDS-PAGE. As

38 Volume 1

Current Protocols in Chemical Biology

such, the MudPIT LC-MS-platform has been adopted for complementary in-depth proÞling of individual proteomic samples, allowing proÞling of 50 to 100 or more enzyme activities per experiment versus ∼10 to 20 enzyme activities per gel lane. ABPP-MudPIT has been implemented for the analysis of a variety of enzyme activities, including serine hydrolases in human breast tumors (Jessani et al., 2005), histone deacetylases (Salisbury and Cravatt, 2007), and metalloproteases (Sieber et al., 2006), the latter targeted using a cocktail of ABPP probes. Kinases (Patricelli et al., 2007) have also been analyzed using a variant of ABPP-MudPIT that only analyzes the probelabeled peptides. Rather than enriching ABPP probe-labeled proteins, the labeled proteome is digested in solution, and the biotinylated peptides are streptavidin-enriched, eluted, and analyzed by LC-MS. In terms of quantitation, both gel-based ABPP and ABPP-MudPIT can be used to provide estimates of active enzyme abundance, using ßuorescence intensity for the former and semi-quantitative methods such as spectral counting for the latter. It should be noted that protein spectral counts in the tryptic dataset (rather than the labeled-peptide dataset) are used to estimate active protein abundance, as labeled peptides typically give too few (<10) spectral counts for reliable estimates of active protein. In contrast to gel-based ABPP, implementation of an MS-based platform allows for direct analysis of site of labeling, which can be helpful for conÞrming the activity-based nature of probe labeling (i.e., modiÞcation of a catalytic residue). Basic Protocol 2 in this unit describes obtaining probe-labeled peptides by denaturing the streptavidin upon heating in a MS-compatible aqueous solution of acetonitrile and trißuoroacetic acid (Okerberg et al., 2005). Alternative methods for elution of labeled peptides involve enzymatic (Weerapana et al., 2007) or chemical (Cravatt et al., 2008) cleavage of tags with labile linkers. These methods have been suggested to provide a cleaner elution, and have the added advantage of removing the bulky tag, a step which should facilitate LC separation and MS ionization (Hansen et al., 2003). However, direct elution of biotinylated peptides can be implemented with a high degree of success. For example, the kinase proÞling experiment mentioned above resulted in the identiÞcation of over a hundred speciÞcally labeled peptides (Patricelli et al., 2007).

Critical Parameters and Troubleshooting If using a new biotin or alkyne probe, or if performing the CC protocol for the Þrst time, it is a good idea to assess probe labeling by gel-based analysis (Kidd et al., 2001; Jessani et al., 2002; Speers and Cravatt, 2004) prior to performing the ABPP-MudPIT protocol. This step will facilitate determination of the optimal probe concentration (maximizing signalto-noise ratio) and, if applicable, ensure that the CC reaction is high yielding. Note that a rhodamine-azide tag is commercially available from Invitrogen for ßuorescence visualization of the CC reaction. If the CC reaction yield appears to be suboptimal, ensure that the TCEP solution is prepared fresh using Fluka TCEP that has been properly stored at 4◦ C; try another lot of the reagent if necessary. The CC reaction works best in a buffer free of amines at pH 7.4 or above, but is tolerant to a wide range of salt and phosphate concentrations. Inclusion of detergents (e.g., SDS) at concentrations greater than 0.1% to 0.5% can also impede the CC reaction and protein precipitation. Thus, keep detergent concentrations at a minimum and perform the methanol-chloroform precipitation described in Support Protocol 2 if inhibition of precipitation is observed. Streptavidin beads can show signiÞcant nonspeciÞc protein binding, so results should always be evaluated with relation to controls. Filter the protein IDs obtained from the trypticdigest experimental sample by comparison to the tryptic-digest control sample; proteins identiÞed in both datasets likely indicate abundant proteins that bind nonspeciÞcally to the streptavidin beads. Filter labeled peptide identiÞcations by comparison to proteins identiÞed in the tryptic digest experimental sample; proteins selectively identiÞed in the labeled peptide dataset (and not in the trypsin digest experimental dataset) should be discarded as false positives.

Anticipated Results The number and type of protein identiÞcations obtained from analyzing tryptic and labeled peptide samples will vary from dozens to hundreds, depending on the proteome source and the type of probe(s) utilized. For example, analysis of human tumor specimens by ABPP-MudPIT using a biotinylated ßuorophosphonate probe that speciÞcally targets serine hydrolases identiÞed over 50 enzymes of that superfamily (Jessani et al., 2005).

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

39 Current Protocols in Chemical Biology

Volume 1

Analysis of labeled peptides obtained from a human cancer cell line homogenate treated with a ATP-mimetic biotinylated kinase probe resulted in identiÞcation of over 100 ATP and other nucleotide-dependent enzymes (Patricelli et al., 2007).

Time Considerations Sample preparation using the described protocols requires 2 days, with the trypsin digest run overnight at the end of day 1. It is suggested that 6 to 12 samples be prepared in parallel. Samples may be stored frozen at the end of Basic Protocol 1, the Alternate Protocol, or Support Protocol 1, or at the penultimate step of Basic Protocol 2, prior to MS analysis. MudPIT analysis of each sample requires 10 to 12 hr, plus 2 to 8 hr (depending on the MS instrument and computing infrastructure) for data analysis. Overall, it should take ∼4 days to complete the full analysis of a tryptic sample, a control tryptic sample, and a labeled peptide sample.

Literature Cited Alexander, J.P., and Cravatt, B.F. 2006. The putative endocannabinoid transport blocker LY2183240 is a potent inhibitor of FAAH and several other brain serine hydrolases. J. Am. Chem. Soc. 128:9699-9704. Bachovchin, D.A., Brown, S.J., Rosen, H., and Cravatt, B.F. 2009. IdentiÞcation of selective inhibitors of uncharacterized enzymes by highthroughput screening with ßuorescent activitybased probes. Nat. Biotechnol. 27:387-394. Cravatt, B.F., Wright, A.T., and Kozarich, J.W. 2008. Activity-based protein proÞling: From enzyme chemistry to proteomic chemistry. Annu. Rev. Biochem. 77:383-414. Evans, M.J. and Cravatt, B.F. 2006. Mechanismbased proÞling of enzyme families. Chem. Rev. 106:3279-3301. Greenbaum, D.C., Baruch, A., Grainger, M., Bozdech, Z., Medzihradszky, K.F., Engel, J., DeRisi, J., Holder, A.A., and Bogyo, M. 2002. A role for the protease falcipain 1 in host cell invasion by the human malaria parasite. Science 298:2002-2006. Gygi, S.P., Rist, B., Gerber, S.A., Turecek, F., Gelb, M.H., and Aebersold, R. 1999. Quantitative analysis of complex protein mixtures using isotope-coded afÞnity tags. Nat. Biotechnol. 17:994-999.

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

Hansen, K.C., Schmitt-Ulms, G., Chalkley, R.J., Hirsch, J., Baldwin, M.A., and Burlingame, A.L. 2003. Mass spectrometric analysis of protein mixtures at low levels using cleavable 13 Cisotope-coded afÞnity tag and multidimensional chromatography. Mol. Cell. Proteomics 2:299314.

Ito, T., Ota, K., Kubota, H., Yamaguchi, Y., Chiba, T., Sakuraba, K., and Yoshida, M. 2002. Roles for the two-hybrid system in exploration of the yeast protein interactome. Mol. Cell. Proteomics 1:561-566. Jessani, N. and Cravatt, B.F. 2004. The development and application of methods for activity-based protein proÞling. Curr. Opin. Chem. Biol. 8:5459. Jessani, N., Liu, Y., Humphrey, M., and Cravatt, B.F. 2002. Enzyme activity proÞles of the secreted and membrane proteome that depict cancer cell invasiveness. Proc. Natl. Acad. Sci. U.S.A. 99:10335-10340. Jessani, N., Niessen, S., Wei, B.Q., Nicolau, M., Humphrey, M., Ji, Y., Han, W., Noh, D.Y., Yates, J.R. 3rd, Jeffrey, S.S., and Cravatt, B.F. 2005. A streamlined platform for high-content functional proteomics of primary human specimens. Nat. Methods 2:691-697. Kidd, D., Liu, Y., and Cravatt, B.F. 2001. ProÞling serine hydrolase activities in complex proteomes. Biochemistry 40:4005-4015. Kobe, B. and Kemp, B.E. 1999. Active site-directed protein regulation. Nature 402:373-376. Kolb, H.C. and Sharpless, K.B. 2003. The growing impact of click chemistry on drug discovery. Drug Discov. Today 8:1128-1137. Link, A.J., Jennings, J.L., and Washburn, M.P. 2003. Analysis of protein composition using multidimensional chromatography and mass spectrometry. Curr. Protoc. Protein Sci. 34:23.1.1-23.1.25. Liu, Y., Patricelli, M.P., and Cravatt, B.F. 1999. Activity-based protein proÞling: The serine hydrolases. Proc. Natl. Acad. Sci. U.S.A. 96:1469414699. MacBeath, G. 2002. Protein microarrays and proteomics. Nat. Genet. 32:526-532. Okerberg, E.S., Wu, J., Zhang, B., Samii, B., Blackford, K., Winn, D.T., Shreder, K.R., Burbaum, J.J., and Patricelli, M.P. 2005. Highresolution functional proteomics by active-site peptide proÞling. Proc. Natl. Acad. Sci. U.S.A. 102:4996-5001. Patricelli, M.P., Giang, D.K., Stamp, L.M., and Burbaum, J.J. 2001. Direct visualization of serine hydrolase activities in complex proteomes using ßuorescent active site-directed probes. Proteomics 1:1067-1071. Patricelli, M.P., Szardenings, A.K., Liyanage, M., Nomanbhoy, T.K., Wu, M., Weissig, H., Aban, A., Chun, D., Tanner, S., and Kozarich, J.W. 2007. Functional interrogation of the kinome using nucleotide acyl phosphates. Biochemistry 46:350-358. Patton, W.F., Schulenberg, B., and Steinberg, T.H. 2002. Two-dimensional gel electrophoresis; better than a poke in the ICAT? Curr. Opin. Biotechnol. 13:321-328. Paulick, M.G. and Bogyo, M. 2008. Application of activity-based probes to the study of enzymes

40 Volume 1

Current Protocols in Chemical Biology

involved in cancer progression. Curr. Opin. Genet. Dev. 18:97-106. Salisbury, C.M. and Cravatt, B.F. 2007. Activitybased probes for proteomic proÞling of histone deacetylase complexes. Proc. Natl. Acad. Sci. U.S.A. 104:1171-1176. Sieber, S.A., Niessen, S., Hoover, H.S., and Cravatt, B.F. 2006. Proteomic proÞling of metalloprotease activities with cocktails of active-site probes. Nat. Chem. Biol. 2:274-281. Speers, A.E. and Cravatt, B.F. 2004. ProÞling enzyme activities in vivo using click chemistry methods. Chem. Biol. 11:535-546. Washburn, M.P., Wolters, D., and Yates, J.R. 3rd. 2001. Large-scale analysis of the yeast proteome by multidimensional protein identiÞcation technology. Nat. Biotechnol. 19:242-247. Weerapana, E., Speers, A.E., and Cravatt, B.F. 2007. Tandem orthogonal proteolysis-activitybased protein proÞling (TOP-ABPP)—a general method for mapping sites of probe modiÞcation in proteomes. Nat. Protoc. 2:14141425.

Click Chemistry Activity-Based Protein ProÞling (CC-ABPP)

41 Current Protocols in Chemical Biology

Volume 1

Overview of Liquid Handling Instrumentation for High-Throughput Screening Applications Stewart Rudnicki1 and Sean Johnston1 1

ICCB-Longwood Screening Facility, Harvard Medical School, Boston, Massachusetts

ABSTRACT Liquid handling in the laboratory has unique challenges speciÞc to the types of research being performed. The devices employed for purposes of performing liquid handling can be broken down into three general categories: bulk reagent dispensers, transfer devices, and plate washers. An overview of these types of liquid handlers, as well as common features and relevance to high-throughput applications, are discussed in this article. Important topics such as sterility, ease of use, cost, and instrument design advantages and disadvantages are also covered. Curr. Protoc. C 2009 by John Wiley & Sons, Inc. Chem. Biol. 1:43-54  Keywords: liquid handling r high throughput screening (HTS) r laboratory automation r plate Þller r plate washer

INTRODUCTION Generally speaking, liquid handling refers to the movement of ßuid from one vessel to another. In the laboratory this can take many forms such as Þlling ßasks with media, transferring liquid from one microplate to another, or even spotting thousands of tiny droplets onto the surface of a slide or agar plate. Instruments used for these purposes are referred to as liquid handlers and are often quite complex, owing to the nature of ßuid dynamics and the difÞculty of accurately transferring small volumes of liquid. The goal of this article is to provide a general overview of the instrumentation currently used in the high-throughput screening (HTS) laboratory for automated liquid handling. The main features of each instrument type, as well as how they affect both liquid handling performance and workßow, are discussed. Common sources of error associated with transfer, as well as instrument design drawbacks, are also covered.

TYPES OF LIQUID HANDLERS There are three main types of automated liquid handlers (divided by application) that are commonly used in the laboratory today: bulk reagent dispensers, transfer devices, and plate washers. Bulk reagent dispensers dispense one or more ßuids quickly into microplates or ßasks and are commonly referred to as plate Þllers. Transfer devices move liquid

from one vessel to another, usually in fairly small quantities, and encompass a variety of technologies such as pipetting, pin-transfer, or acoustic transfer. Plate washers have integrated ßuidics that allow removal (aspiration) of ßuid from microplate wells followed by addition of wash buffers.

Bulk Reagent Dispensers Bulk reagent dispensers are used to dispense large amounts of a single reagent from bulk source (e.g., a ßask of tissue culture cells in suspension) to multiple destination positions. The majority of bulk reagent dispensers are used to Þll microplates, which are used in both basic and drug discovery research. Thus, the destination positions would refer to the different wells of the microplate, and even very large numbers of microplates. The volume range that can be dispensed can vary widely from milliliters (ml) to nanoliters (nl), with the most common volume range being from 1 to 200 μl. Increasingly, there is a trend towards dispensing lower volumes (in the nanoliter range), due in part to the increased use of higher-density microplates such as 1536-well plates (Burnbaum, 1998; Hamilton, 2008). Bulk reagent dispensers use a tubing system that runs from the source bottle or reservoir to one or more tips that are suspended over the plate to be Þlled. Dispensers used for high-throughput screening applications

Current Protocols in Chemical Biology 1: 43-54, December 2009 Published online December 2009 in Wiley Interscience (www.interscience.wiley.com). DOI: 10.1002/9780470559277.ch090151 C 2009 John Wiley & Sons, Inc. Copyright 

Liquid Handling Instrumentation for HTS

43 Volume 1

Figure 1

Overview of Liquid Handling Instrumentation for High Throughput Screening Applications

ThermoFisher Matrix Wellmate peristaltic pump plate filler.

generally use one of three basic technologies to move and dispense liquid. The most common method is the use of a peristaltic pump and ßexible tubing. An example of a plate Þller designed with a peristaltic pump is the ThermoFisher Matrix Wellmate (Fig. 1). The peristaltic pump compresses the tubing using rollers and creates a positivedisplacement motion effect. One advantage of this type of dispensing is that the entire ßuid path (tubing and dispensing manifold) can be contained in a “cartridge” that can be inserted into the dispenser when needed, and then removed for cleaning. Different types of cartridges are available, tailored speciÞcally to the type of dispensing required (standard versus low-volume, special tip coatings, etc.; see http://www.matrixtechcorp.com/automated/ pipetting.aspx?id=11 and http://www.biotek. com/products/liquid handling/microßo select dispenser.html). A signiÞcant advantage of this type of system is that the cartridges can often be autoclaved, thus guaranteeing a completely sterile path from source bottle to destination, which is ideal for cell culture. Additionally, any remaining liquid in the tubing after dispensing can be recovered to the source by running the pump in reverse. One disadvantage is that the cartridge tips are generally incapable of acting independently, thus relegating dispensing options to whole microplate rows or columns rather than individual wells.

The second type of technology used to move ßuid is via a syringe pump (http://www. biotek.com/products/liquid handling/microÞll microplate dispenser.html). In this system, the tubing has check valves (one-way only) on both the inlet and outlet sides of a syringe pump. These valves are oriented so the ßow occurs in one direction only—from the source bottle to the destination position. The syringe pump itself is usually a glass barrel with a motor-driven plunger that makes a seal with the inside of the barrel. By drawing the plunger back and forth inside the barrel, the ßuid is drawn into the chamber and then forced out of it. The check valves ensure that the ßow only moves in the direction of the destination position. In some cases, the entire assembly may be removed and autoclaved, providing complete sterility. An example of a plate Þller designed with a syringe pump is the BioTek μÞll. A couple of disadvantages to syringe pump systems are that there is a larger priming volume (volume to Þll the ßuid path) required due to the necessity of Þlling the syringe pump, and that there is no option to recover this priming volume after dispensing, as the liquid can only move in one direction because of the check valves. A third technology used by bulk reagent dispensers is that of a pressure-driven system (see http://www.matrixtechcorp.com/ automated/pipetting.aspx?id=77 and http:// www.biotek.com/products/liquid handling/

44 Volume 1

Current Protocols in Chemical Biology

nanoquot microplate dispenser.html). Here, the ßuid path is a sealed system, and pressure is introduced into the source bottle to force the liquid through the tubing. The liquid passes through valves at the tip end, which open and close with speciÞed timing to allow a precise amount of liquid to be dispensed. An example of a plate Þller designed with a pressuredriven system is the ThermoFisher MultiDrop Combi nL. Pressure-driven systems generally have the advantage of being able to dispense to any well of a microplate, and are better at dispensing lower (sub-microliter) volumes due to independent valves on each of the tips. They do suffer a couple of disadvantages, however, in that the entire ßuid path is not removable, and thus not autoclavable. In addition, the valves are prone to clogging and must be maintained carefully.

Dead volume One important aspect of bulk reagent dispensers is the priming volume or “dead volume” of the instrument. Priming volume is the amount of liquid required to Þll the ßuid path from source to tips. It is an important aspect of any dispenser, as it dictates the amount of extra reagent required in addition to the amount to be dispensed. Often this volume is nonrecoverable after dispensing and is therefore wasted or “dead” volume. The priming volume is mainly dependent on the length of tubing assembly from the source to the destination, the inner diameter of the tubing, and whether or not the pump itself requires Þlling (e.g., in the case of a syringe pump). It also takes into account the amount that must be dispensed to rid the system of all air bubbles and ensure proper operation. Typical minimum priming volumes are in the range of 8 to 15 ml, with some dispensers having a priming volume as low as 2 to 3 ml, and others much higher (see http:// www.biotek.com/products/liquid handling/ microÞll microplate dispenser.html and http:// www.matrixtechcorp.com/automated/pipetting. aspx?id=77). Clearly, the cost/availability of the reagent being dispensed is an important factor to consider when choosing a plate Þller, and applications such as dispensing antibodies or puriÞed proteins have to be evaluated in terms of the waste inherent in using the dispenser. However, there is an economy of scale when Þlling large quantities of plates; 8 ml of waste is a large percentage of the total volume dispensed when Þlling one microplate, but when Þlling 1000 microplates it becomes much more reasonable.

Volume range The dispense-volume capabilities of bulk reagent dispensers range from 1 nl and up, with Þllers falling into three general ranges: milliliter, microliter, and nanoliter. At the top end of the range are bulk reagent dispensers designed to Þll bottles and ßasks (http://www.essen-instruments.com/pdf/ pipeline brochure.pdf). These Þllers often use a peristaltic or syringe pump along with an electronic control to dispense large quantities of liquid. These devices are frequently used for Þlling tissue culture ßasks with larger volumes of media, hands-free in a sterile environment, and are easily cleaned and maintained. They are not good for dispensing volumes below 100 μl and accuracy is limited to approximately ±2% of dispense volume plus 120 μl. In the middle of the range are bulk reagent dispensers that dispense in the microliter range. These Þllers are typically used for dispensing cells, media, and reagents into microplates. The volume ranges of dispensers in this category vary widely, from as low as 1 μl to several milliliters. As the volume approaches 1 μl, the surface tension of the liquid starts to have a pronounced effect on the accuracy and precision of the volume being dispensed. It is for this reason that the effectiveness of peristaltic and syringe pumps is generally limited to volumes greater than 1 μl. As the volume to be transferred approaches and becomes less than 1 μl, the viscosity, surface tension, and wettability (the afÞnity for a liquid to maintain contact with a solid surface; de Gennes, 1985) of the ßuid being transferred starts having a much greater effect on the accuracy and precision of the instrument. To overcome these issues, manufacturers generally incorporate a pressure/valve– or even a piezoelectric effect–based system (Hsieh et al., 2004) for dispensing. Bulk reagent dispensers in this range are typically more complex pieces of equipment, and are more costly, often an order of magnitude more expensive. Typical volumes for these types of dispensers range from 50 nl to several μl (http://www.biotek.com/products/liquid handling/microÞll microplate dispenser.html). An example of a plate Þller that uses piezoelectric dispensing is the Beckman Coulter PicoRAPTR (http://www.beckman.com/products/ instrument/automatedsolutions/integsystems/ pico raptr.asp). One important caution is that some bulk reagent dispensers are designed to dispense solely in whole-unit increments. For example, one dispenser capable of dispensing from 1 to

Liquid Handling Instrumentation for HTS

45 Current Protocols in Chemical Biology

Volume 1

2000 μl, cannot dispense the fraction 5.5 μl, only either 5 or 6 μl.

The ßuid path The ßuid path from source to destination is important in that it not only determines the priming volume, but also dictates the methods for cleaning and sterilization and factors heavily in “user friendliness.” There are two main categories for bulk reagent dispenser ßuid paths: those that can be removed for sterilization, and those that are integrated with the dispenser. Frequently, those that can be removed are in cartridge form and are made of materials that can be autoclaved for sterilization. For those paths that are not removable, cleaning with detergent solutions or disinfectants such as 70% ethanol or 10% bleach, or a combination of these, can be effective (Hughes et al., 2005). Clogging can be a substantial problem for bulk reagent dispensers that dispense low volumes through narrow ßuidic pathways. Valves and tips can be particularly susceptible to clogging. When using cell or protein solutions that might cause clogs (e.g., that contain clumping cells or precipitates), careful attention should be paid to cleaning the dispenser afterwards. Some dispensers are lined with special low-binding coatings such as PTFE (Teßon) to minimize protein binding along the liquid path.

Well addressability and user interface

Overview of Liquid Handling Instrumentation for High Throughput Screening Applications

One functional aspect of bulk reagent dispensers that varies widely amongst manufacturers is that of microplate well addressability. Pressure/valve dispensers generally offer greater ßexibility in well addressability over peristaltic or syringe pump-driven systems. This is primarily due to design of the dispense manifold. With valve systems, each channel can be independently controlled, whereas in pump systems the manifold is simply an 8-channel branching of the main tubing assembly—the pump system does not control each channel separately. Therefore, the manifold is the smallest individual unit, which translates to a series of 8 wells, usually a column, on a microplate. In contrast, in a pressure/valve system each channel is the smallest individual unit, which translates directly to an individual well on a microplate. What does this mean? It means that a pressure/valve system can dispense a unique volume to each well of a microplate, allowing for such things as plate patterning, serial dilutions, and even random-

ization of values across a plate, something that pump-driven systems simply cannot do. Most dispensers represent a compromise between ease of use and functionality. In most cases, plate Þllers with more user-friendly interfaces have less ßexibility for well addressability. Some dispensers are designed to be operated from a keypad on the unit itself, while others require a computer to program the Þller operation. There are pros and cons to each design. Using a keypad for operator input simpliÞes the design of the dispenser. Simple operations such as specifying which columns are to be Þlled, what volume to use, and whether to forward- or reverse-prime the instrument can be easily input using a keypad. This format reduces the level of user knowledge required to operate the instrument. The main disadvantage to keypad-only dispensers is their more limited ßexibility in dispensing options. For example, most microplate dispensers are capable of specifying which columns to leave empty, but cannot address rows individually. This limitation is usually imposed by the design of the instrument rather than the simple user interface (see above); nevertheless, it would be extremely tricky and tedious to input hundreds of individual well speciÞcations from a keypad input. Other bulk reagent dispensers require a separate computer controller for operation. The advantage of this type of system is the ßexibility that the input method offers. For example, with the ThermoFisher MultiDrop Combi nL plate Þller, the volume to dispense to each well of a 1536-well microplate can be speciÞed quickly and easily using a point-and-click approach or by importing a spreadsheet of values. Differing microplate types can be created and stored in proÞles, as can the speciÞc parameters for many different types of liquids. Disadvantages of this type of system include the requirement for an external computer wired to the plate Þller, which increases the labbench footprint of the dispenser system, and the higher level of user knowledge required to operate the instrument.

Calibration—precision versus accuracy When someone says that a dispenser is “accurate to 1 μl,” what does that mean exactly? To answer this, it Þrst necessary to discuss the difference between precision versus accuracy. Dispensing precision is the ability of a dispenser to repeatedly dispense the same amount of liquid each time it dispenses. How much that volume varies between dispenses is

46 Volume 1

Current Protocols in Chemical Biology

the measure of the dispenser’s precision and is usually represented as coefÞcient of variation (%CV; Taylor et al., 2002; Petersen and Nguyen, 2005): %CV = (σ /μ) × 100 where σ is the standard deviation of the mean and μ is the mean. For most applications, a good dispenser CV is typically less than 10%. CVs below 5% are considered excellent (Harris et al., 2008). Dispensing accuracy is a measure of how closely the dispenser conforms to the true volume requested. For example, if we tell the dispenser to dispense 10 μl to every well of a 384-well plate and it repeatedly dispenses from 8.8 to 9.2 μl in every well, the precision is excellent—with a CV of 0.13%—but the accuracy is not particularly good—averaging 9 μl, which is 10% below the requested volume of 10 μl. Some dispensers offer the ability to calibrate the pump by measuring a series of dispensed volumes of a dye such as tartrazine, and using an absorbance or ßuorescence plate reader to generate a standard curve (Bradshaw et al., 2005). Once generated, an equation can be determined and input into the instrument to offset the dispensed volume by a speciÞed amount to make the dispensing more accurate. Dispensers generally become less and less precise as the volume dispensed decreases. Once CVs top 10% for a given volume dispensed, the dispenser is no longer considered to be “accurate.”

Integration with other lab instruments The primary purpose of a bulk reagent dispenser is to dispense liquid to large numbers of bottles or plates rapidly. This makes it ideally suited for inclusion in automated platforms for HTS. Ease of integration of bulk reagent dispensers with robot arms, plate stackers, and other screening instrumentation is certainly an important factor to consider when purchasing plate Þllers for use in an automated system. One common task performed by bulk reagent dispensers is plating live cells into microplates. Sterile conditions and, often, biocontainment are necessary for such an application, and thus many bulk reagent Þllers are used inside tissue culture hoods. This use favors simple plate Þllers controlled by a keypad, as using a computer to interact with the Þller is more complicated with sterile technique.

Transfer Devices “Transfer device” is a generic term used here to describe a variety of liquid handlers

that are not bulk reagent dispensers and are not plate washers. These liquid handlers employ technologies such as pipetting, pin tools, or acoustic energy to transfer liquid from one location to another. The volumes transferred are generally in the microliter and nanoliter range, and the instruments can be simple stand-alone devices, or part of larger, complex automation platforms capable of running biological assays for days at a time unattended.

Types of transfer devices By far the most common type of transfer device is one that uses pipetting to move liquid from one location to another. This liquidhandling technique transfers liquid by aspirating it into a pipet tip, and then dispensing it to the destination location(s). The aspiration of the liquid into the tip can be accomplished via two different methods: capillary action or volume displacement. Capillary action pipettors use surface tension and the intermolecular forces of the liquid in conjunction with an extremely narrow tube to draw a speciÞc amount of liquid into the pipet. This volume is usually very small—in the nanoliter range— and can be ejected in a non-contact fashion to the desired location afterwards, keeping the pipet tips free from contamination from the destination plate. An example of an automated capillary action pipettor is the Digilab Hummingbird (http://www.digilabglobal. com/HUMMINGBIRD.html). The primary disadvantage to this type of pipetting is that each pipet head can only dispense a single volume of liquid. To dispense a different volume, the head must be changed out for a different one. Also, the pipets must be washed between dispenses, making the process comparatively slow. Finally, the dispenser can only dispense once per aspirate. In contrast, volume displacement pipettors are much more common than their capillary action counterparts mainly due to their inherent ßexibility. Examples of instruments with volume displacement pipettors include the Agilent (formerly Velocity11) Bravo (Fig. 2), the Tecan Freedom EVO, and the Beckman Coulter Biomek FX. In this type of pipetting, a piston (typically a syringe barrel) moves inside a cylinder and displaces a volume of air or other system ßuid such as water (http://www. chem.agilent.com/Library/datasheets/Public/ 5990-3480EN LO.pdf). This displacement changes the pressure inside the cylinder, causing liquid to ßow into or out of an attached pipet depending on the direction of the piston. As a result, volume displacement

Liquid Handling Instrumentation for HTS

47 Current Protocols in Chemical Biology

Volume 1

Figure 2 Agilent (formerly Velocity11) Bravo automated liquid handling platform. Magnifications of the 384and 96-channel pipetting heads are shown in the right panels.

Overview of Liquid Handling Instrumentation for High Throughput Screening Applications

pipettors can dispense a wide variety of volumes using the same syringe by changing the amount that the piston is moved. This is the same mechanism used by standard laboratory hand-held “pipetmen.” In automated liquid handlers, the speed of the piston can also be controlled, allowing for fast or slow pipetting depending on the requirements of the liquid being transferred. Multi-dispensing (dispensing into more than one vessel after aspirating a single time) becomes possible using this type of pipettor. A further advantage of this type of pipetting is that the syringes themselves may be switched out to accommodate different volumes of ßuid being transferred, accommodating a range from 500 nl to 1 ml (Hughes et al., 2005). A second type of liquid transfer device commonly employed for transferring small volumes is the pin tool (Fig. 3). This device is made up of a set of carefully machined stainless steel pins, with tightly controlled physical dimensions (Dunn and Feygin, 2000). Some pin tools are coated with either a hydrophobic or hydrophilic coating, and some even have tiny slots cut into them at the bottom of each pin. Pin tools work by using properties such as surface tension, wettability, and intermolecular forces of

a liquid to coat the pins with a precise volume (Quintero et al., 2007). The volume transferred is determined primarily by the surface area of each pin, but can be controlled to some degree also by changing the depth to which the pins are submerged in the source liquid, and the speed with which the pins are removed (http://www.vp-scientiÞc. com/pin data cybi well robot.htm#speed-allpins). The pins are then submerged in the destination liquid, which washes off the pins and completes the transfer. There are a few advantages to this type of transfer device, namely that there are no consumables such as pipet tips, and solid particulate matter in the source liquid cannot clog the mechanism of transfer, unlike when using pipettors. Disadvantages include the need to wash pins between transfers, a very limited volume range for each pin size, and the need for strict control of the transfer speciÞcs such as plate type and volume. Furthermore, for greatest accuracy, it is not recommended to pin-transfer solutions into dry (empty) destination wells. Therefore, in general, pin transfer must be performed into destination wells already containing several microliters of cells or buffer. There is also a minimum volume required in the source plate for accurate transfer (typically several microliters

48 Volume 1

Current Protocols in Chemical Biology

Figure 3 V&P Scientific (http://www.vp-scientific.com) floating 0.787-mm diameter 100-nl slotted 384-pin tool.

even if transferring nl amounts), and thus pin tools cannot be used to empty a source well. A third type of liquid transfer device uses acoustic (sound) waves to bounce small droplets of liquid from one microplate to another positioned upside down directly above it. This technology is called acoustic droplet ejection (ADE; Harris et al., 2008). By regulating the amount of energy and the number of pulses, a speciÞc volume of the liquid can be transferred to the destination well. Examples of instruments that use ADE for liquid transfer include the Labcyte Echo and the EDC Biosystems ATS-100 (see http://www.labcyte. com/Echo Liquid Handling Systems/Default. 81.html and http://edcbiosystems.com/bio/ edc products ats100.html). This technique is advantageous in that there are no tips to purchase or pins that need to be washed. Also, acoustic instruments can deliver much smaller volumes of liquid than pipets (as low as 1 nl) into dry wells. There are a few drawbacks, however. The Þrst is that the surface tension of the liquid must be able to hold it in destination wells, upside down, after transfer. This generally means that acoustic transfer will work only for 96-, 384and 1536-well microplates and that buffers containing detergents cannot be transferred by this method. Another drawback is that

the source liquid must be in a microplate with speciÞc well geometry and density requirements (Olechno et al., 2006). Also, this type of device transfers liquid one well at a time, so moving liquid from all wells of one microplate to another is generally slower than for pipetting or pin tools. Finally, transfer devices of this type are complex machines, and tend to be very costly.

Features One clear advantage that transfer devices have over bulk reagent dispensers is that they operate with little or no dead volume. This makes them ideal for transferring precious reagents such as proteins and antibodies. Pin tools tend to have the largest volume requirement, with a speciÞc minimal volume required in the source plate, followed by pipet tips which generally lose volume when liquid clings to the inside and outside of the tip. The priming volume required for acoustic transfer devices is limited to the volume in the source well, with very little waste. The volume range dispensed by transfer devices is generally lower than that of bulk reagent dispensers, and can be as low as 1 nl or as high as 1 ml, depending on the device. Precision and accuracy using these types of devices can be very good.

Liquid Handling Instrumentation for HTS

49 Current Protocols in Chemical Biology

Volume 1

Most transfer devices are designed to transfer liquid from one or more wells on a microplate to another, quickly and efÞciently. To accomplish this, pipetting and pin tool devices offer the option of having many pipets together in an array. These arrays are commonly available in 8-, 12-, 16-, and 24-column format, or 96-, 384-, and 1536-well microplate formats (Fig. 2). The tips themselves can be either Þxed (permanent) or disposable plastic. Disposable tips are used to prevent contamination and carryover between wells, and also for convenience. Fixed tips can be coated with a variety of materials, the most common of which is PTFE (Teßon), that are particularly low-binding. Carryover between transfers is of particular concern when using Þxed tips, and washing is usually necessary to clean tips between transfers (Ouyang et al., 2008). The decision whether to use Þxed versus disposable tips is often dictated by the needs of the experiment being performed. For example, cross-contamination is a much more serious concern in quantitative polymerase chain reaction (qPCR) setup than for general reagent dispensing.

Calibration of liquid handling instruments

Overview of Liquid Handling Instrumentation for High Throughput Screening Applications

Calibration is the process of verifying that the volume requested to be transferred is actually what the device transfers, and is necessary for all liquid handling instrumentation. Since the combined effects of many factors that govern ßuid dynamics are quite pronounced at low volumes, this can be a challenging and timeconsuming process. There are several methods available for calibrating liquid transfer devices, including gravimetric, photometric (absorbance), ßuorescence, and others. In gravimetric analysis, the amount of liquid dispensed is measured by comparing the weight measurements of the vessel being dispensed into, both before and after the addition (Taylor et al., 2002). By comparing the measured weight (mass) to the known density of the solution, the volume transferred can be determined. This method for calibration is often used, as it is the easiest, requiring only a balance, common in most laboratories. It does suffer a few drawbacks, however, in that it cannot determine individual tip precision, only that of all of the tips used at once. It is also directly dependent on the accuracy of the balance being used. For low transfer volumes, this becomes increasingly inaccurate due to factors such as the stability of the balance, its lower

limit of accuracy, and other environmental factors such as evaporation. Photometry is another common method used for calibrating liquid transfer devices. This process employs a spectrophotometer to measure the absorbance of a liquid at a speciÞc wavelength to determine volume using equations based on the Beer-Lambert Law (Bradshaw et al., 2007). An absorbance-based dye (such as tartrazine) is used to generate a standard curve, against which the absorbance of the transferred volume can then be compared. Some methods use more than one absorbance dye and measure the ratio of absorbance values, which allows the measurement to be independent of the path length of the absorbance measurement (Bradshaw et al., 2005). One limitation of this method is that it can only measure volumes that fall within the linear range of the curve that is generated. Depending on the absorbance dye(s) used, this lower limit is usually somewhere between 0.1 and 1.0 μl. When calibration is required for volumes below 1.0 μl, ßuorescent dyes are commonly used due to the linearity of the curve in dilute conditions (Petersen and Nguyen, 2005). Like the photometry method described above, ßuorescent analysis can determine both accuracy and precision of a liquid transfer device. One drawback to using a ßuorescent dye, however, is that of photobleaching—a loss of ßuorescent signal due to extended exposure to excitation light. This can limit the number of times a given standard curve can be measured during the calibration process. Other methods also exist for calibrating liquid transfer devices such as using radioactive isotopes like tritium (3 H), or via acoustic transducers, but they are less common and are beyond the scope of this discussion.

Liquid Handling Workstations Transfer devices are typically part of larger platforms called liquid handling workstations that are designed to automate routine liquid handling tasks such as reagent dispensing, serial dilutions, and microplate replication (Lorenz, 2004). Some workstations may have more than one transfer device on the same workstation, such as an 8-channel and a 96-channel pipettor. These workstations come in many shapes and sizes—from small dedicated machines intended for one primary purpose, to very large platforms with several integrated transfer devices capable of multitasking. When faced with the task of choosing

50 Volume 1

Current Protocols in Chemical Biology

one of these platforms, the following factors should be considered.

Function Perhaps the most important factor when choosing a liquid handler is making sure that the instrument will perform the particular tasks required. By design, automation is very good at performing the same task over and over and can often do so faster, and with more accuracy and precision than if the same thing were done by hand. However, the most common mistake that people make when choosing a liquid handling workstation is thinking that one instrument can “do it all.” While automation is very good at performing the same task over and over, it is not nearly as good at performing many different tasks infrequently. It is true that many platforms offer the ability to perform a diversity of functions; however, setting each one up and verifying that it is operating correctly is usually a time-consuming process. For example, often it is easier to simply perform a serial dilution in a few microplates by hand, rather than set it up on the workstation.

Ease of use and containment Unfortunately, with increased functionality comes increased complexity. Liquid handling workstations that perform one primary task are generally easy to operate, whereas workstations that can perform many tasks usually have a unique programming language called a scripting language, and require extensive training. Frequently, a “superuser” is required to operate and maintain complex workstations. Programming and testing are performed by the superuser, and the programs are run by general lab members. Another factor to consider is how the liquid handling workstation might interact with other pieces of lab equipment. What is the ultimate purpose of the experiment? Is the workstation performing one or more steps of a larger automated process? The ability to integrate one liquid handler with other pieces of equipment and the ease of doing so becomes important in these cases. Sterility and biocontainment are also important issues to consider. Workstations that Þt into conventional-sized biosafety cabinets (hoods) are a convenient solution if biocontainment is required. Larger systems often require a custom solution such as a HEPA-Þltered enclosure built around the workstation.

Cost Inevitably, cost is a major factor in choosing a liquid handling workstation. But cost involves more than just the purchase price of

the system. For instance, service and support contracts typically cost about 10% of the initial purchase price per year, once the warranty runs out. Consumables are also a major contributing factor to the bottom line, particularly disposable pipet tips. Costs of operation should also be factored into the decision—does it require large amounts of electricity, or a separate air pressure or vacuum system to operate? Finally, does it require a dedicated person to operate and maintain the instrument? Has that individual’s salary been taken into account?

Plate Washers Function Plate washers are specialized platforms designed to remove liquid from assay microplates, to rinse the wells with wash buffers, and, in some instruments, also to add Þxative or staining solutions. The coupling of aspiration with dispensing in one unit allows for fast and effective addition and removal of liquid from microplate wells. Originally designed to aggressively “scrub” microplates containing biochemical assays such as the enzymelinked immunosorbent assay (ELISA), the plate washer has more recently proven useful for other applications, such as for Þxing, staining, and washing adherent cell-based assays for imaging, and for processing bead-based assays (Lequin, 2005; Wu et al., 2006). Although it is possible to wash samples with a pipettor or multichannel vacuum manifold in combination with a plate Þller, use of an automated plate washer is much faster and scalable for high-throughput screening. The quick transition from aspiration to dispensing that is possible in an automated plate washer keeps samples from drying out, and allows many plates to be handled in quick succession.

Features The design of automated plate washers can vary, but all platforms possess the same basic components: an aspirator, at least one dispenser, reagent source bottles, waste bottles, a user interface, and the option for a plate stacker to handle many plates. The most common mechanism for volume removal is the aspiration manifold—a series of hollow stainless steel pins that use negative pressure (vacuum) to aspirate liquid. Each pin is lowered into a well and a vacuum force is applied, drawing out liquid and sending it into a waste reservoir. An example of a vacuum aspiration plate washer is the BioTek ELx405 plate washer (Fig. 4; Held and Buehrer, 2004). Depending on the system, the separate purchase of

Liquid Handling Instrumentation for HTS

51 Current Protocols in Chemical Biology

Volume 1

Figure 4 BioTek ELx405 Plate Washer with BioStack Microplate Stacker and reagent source bottles. Inset, top right: close-up of aspirating and dispensing pins.

Overview of Liquid Handling Instrumentation for High Throughput Screening Applications

a vacuum pump may be necessary. Aspiration manifolds typically have anywhere from 8 to 384 channels. Platforms with only a single row or column of channels are known as strip washers. Most current 1536-well microplate washers work in strip format, generally with strips that aspirate/wash a single row or column at a time. The 96-pin layout is popular for high-throughput screening applications because it can be used for whole-plate washing of 96-well plates, as well as for quadrant washing of 384-well plates. Positive pressure has also been used as alternative method for liquid removal. Positive pressure may be employed in the form of an air curtain—a ßat jet of compressed air that is aimed inward and passed across the surface of an inverted assay plate. The volume is forced out of each well and collected in a waste vessel below. This mechanism is non-contact and avoids the issue of pin clogging, but the design makes it nearly impossible to remove any less than the entire well volume. An example of a plate washer that uses positive pressure is the Squirt from MatriCal (http://www.matrical. com/Squirt Microplate Washer Coater.php). Every plate washer comes equipped with a high-volume bulk dispenser. These dispensers are designed with an emphasis on speed and throughput, but they have lower precision and

higher rates of reagent consumption than standard plate Þllers. They are ideal for delivering inexpensive wash buffers and other reagents where variability in the volume dispensed can be tolerated. Media may also be delivered, but the ßuid path is usually not removable and therefore sterility may be difÞcult to ensure. The bulk dispenser usually has the same number of pins as the aspirator array in the same unit—8 to 384 hollow stainless steel pins. Although they are capable of fast and effective wash cycling, this form of dispenser has a few drawbacks. A simultaneous dispense through a large number of channels requires a high ßow rate, which in turn requires wide-bore tubing. As a result, priming volumes are very high. For example, a common priming volume for the BioTek ELx405 (Fig. 4) is 200 ml versus 8 ml for the ThermoFisher Matrix Wellmate plate Þller (Fig. 1). Additionally, one can expect relatively high well-to-well variability when forcing reagent through many channels from a common source. Many newer plate washer models also feature high precision, low-volume dispensers. These dispensers are typically driven by peristaltic or syringe-based pumps and are akin to dedicated plate Þllers. They tend to Þll one column at a time, sacriÞcing speed for accuracy, precision, and lower dead volume. The ßuid

52 Volume 1

Current Protocols in Chemical Biology

path uses smaller priming volumes and may also be removable and autoclavable for sterile applications. An example of a plate washer with a removable low-volume dispenser is the BioTek ELx406. The advent of low-volume dispensers on plate washers has helped to further automate assay processes. For example, antibodies and other costly reagents may now be added in sequence with other wash procedures, without the necessity of manual intervention. Every plate washer also has at least one source bottle for wash buffer or reagents. In some instances, this functionality can be expanded by the addition of a valve module to alternate between several different reagent bottles, usually at the cost of higher priming volumes. The size of the bottle and the length of the ßuid path should be considered carefully with respect to the kinds of reagents that will pass through them. Waste handling is very important. If hazardous substances are to be used on the plate washer, a smaller waste bottle is recommended to prompt frequent cleaning. Otherwise, larger bottles are more convenient for nonhazardous waste. As with all waste systems, a double trap within a secondary containment bin is ideal. This conÞguration provides safety and ensures the longevity of the in-line vacuum pump. The plate washer system may be programmed by way of a keypad or computer interface. Customizable wash proÞles allow a user to Þne tune many variables, such as the number of wash cycles, the aspiration and dispensing rates, aspiration height (and therefore remainder volume), and distance from the well wall. Some platforms also have shaking and soaking steps, which can increase the effectiveness of the wash cycle. Most plate washers may also be integrated with a stacker module. The stacker provides a way to queue up a large number of assay plates (typically 20 to 30), which further increases the speed at which samples can be processed.

Special considerations when using plate washers Plate washers have some limitations, and thus might not be appropriate for some highthroughput assays. Dispense/aspirate volume variability can be problematic for some applications. This is often caused by tip clogging. Staining and wash buffers can include salts and proteins that can crystallize or precipitate and clog the manifold pins if the instrument is not cleaned regularly. Once a pin

becomes clogged, the cleaning process can be intensive. For example, it might be necessary to disassemble an aspiration or dispensing manifold to fully clear the ßuid path. Sonicating baths are available on some plate washer platforms that can be used to unclog blocked channels. Most clogging can be avoided by proactive maintenance and cleaning routines. Between uses, the ßuid path must be rinsed thoroughly with cleaning solutions containing ethanol, methanol, or enzyme “active” detergents (such as Terg-A-Zyme). It can be challenging to adapt plate washer protocols for cell-based assays. A few plates of assay controls should be tested to verify that the wash is effective and not disruptive to the cell layer. To minimize disruption, the dispenser should drop liquid onto the side wall of each well if possible, not directly onto the cell layer. This can be facilitated by the use of angled dispenser pins. Moreover, the dispense ßow rate should be as low as possible while still dispensing the desired volume. Aspirating too close to the cell layer can result in cell damage and cell loss, and can dry out the cells that remain (Parry et al., 1997). It is recommended to leave a residual volume of media/buffer in each well when aspirating, and to increase the number of wash steps as necessary for thorough washing. An easy method to monitor the extent of cell layer disruption or cell loss after washing is to view the well contents with a microscope.

SUMMARY The technology used to perform liquid handling in the laboratory is in a constant state of change, with new features and improvements being made every year. Instruments have become more user-friendly, and precision and accuracy have progressively improved, particularly in the realm of low-volume dispensing. New technologies continually emerge enabling innovative applications for research. Small-volume liquid handling remains an extremely complex and fascinating area. SpeciÞc techniques for mixing solutions in wells after dispensing, ridding pipet tips of bubbles, working with viscous solutions, and multidispensing are topics beyond the scope of this introductory article. However, there many articles available that address these issues (Dunn and Feygin, 2000; Albert et al., 2006; Gu and Deng, 2007; Gurevitch, 2008). For additional information on laboratory liquid handling, often the best sources of information are instrument manufacturers and service technicians.

Liquid Handling Instrumentation for HTS

53 Current Protocols in Chemical Biology

Volume 1

ACKNOWLEDGEMENTS The authors would like to acknowledge funding from NIH AI067751 and NIH AI057159 that helped to support this work.

LITERATURE CITED Albert, K.J., Bradshaw, J.T., Knaide, T.R., and Rogers, A.L. 2006. Verifying liquid-handler performance for complex or nonaqueous reagents: A new approach. J. Assoc. Lab. Automat. 11:172-180. Bradshaw, J.T., Knaide, T.R., Rogers, A., and Curtis, R. 2005. Multichannel veriÞcation system (MVS): A dual-dye ratiometric photometry system for performance veriÞcation of multichannel liquid delivery devices. J. Assoc. Lab. Automat. 10:35-42. Bradshaw, J.T., Curtis, R.H., Knaide, T.R., and Spaulding, B.W. 2007. Determining dilution accuracy in microtiter plate assays using a quantitative dual-wavelength absorbance method. J. Assoc. Lab. Automat. 2007 12:260-266. Burnbaum J.J. 1998. Miniaturization technologies in HTS: How fast, how small, how soon? Drug Disc. Today 3:304-312. de Gennes, P.G. 1985. Wetting: Statics and dynamics. Rev. Mod. Phys. 57:827-863. Dunn, D.A. and Feygin, I. 2000. Challenges and solutions to ultra-high-throughput screening assay miniaturization: Submicroliter ßuid handling. Drug Disc. Today 5:84-91. Gu, H. and Deng, Y. 2007. Dilution effect in multichannel liquid-handling system equipped with Þxed tips: Problems and solutions for bioanalytical sample preparation. J. Assoc. Lab. Automat. 12:355-362. Gurevitch, D. 2008. Case study: Birth of an ALA short course. J. Assoc. Lab. Automat. 13:136144. Hamilton, S.D. 2008. ALA survey on laboratory automation. J. Assoc. Lab. Automat. 14:308-319. Harris, D., Mutz, M., Sonntag, M., Stearns, R., Shieh, J., Pickett, S., Ellson, R., and Olechno, J. 2008. Low nanoliter acoustic transfer of aqueous ßuids with high precision and accuracy of volume transfer and positional placement. J. Assoc. Lab. Automat. 13:97-102. Held, P. and Buehrer, L. 2004. The ELx405 384well microplate washer: Designed to meet the rigors of biomolecular screening. http://www. biotek.com/resources/docs/Lab Automation 2004 Poster.pdf.

Hsieh, H.B., Fitch, J., White, D., Torres, F., Roy, J., Matusiak, R., Krivacic, B., Kowalski, B., Bruce, R., and Elrod, S. 2004. Ultra-high-throughput microarray generation and liquid dispensing using multiple disposable piezoelectric ejectors. J. Biomolec. Screen. 9:85-94. Hughes, S.R., Riedmuller, S.B., Mertens, J.A., Li, X.-L., Bischoff, K.M., Cotta, M.A., Farrelly, P.J. 2005. Development of a liquid handler component for a plasmid-based functional proteomic robotic workcell. J. Assoc. Lab. Automat. 10:287-300. Lequin, R.M. 2005. Enzyme immunoassay (EIA)/enzyme-linked immunosorbent assay (ELISA). Clin. Chem. 51:2415-2418. Lorenz, M.G.O. 2004. Liquid-handling robotic workstations for functional genomics. J. Assoc. Lab. Automat. 9:262-267. Olechno, J., Shieh, J., and Ellson, R. 2006. Improving IC50 results with acoustic droplet ejection. J. Assoc. Lab. Automat. 11:240-246. Ouyang, Z., Federer, S., Porter, G., Kaufmann, C., and Jemal, M. 2008. Strategies to maintain sample integrity using a liquid-Þlled automated liquid-handling system with Þxed pipetting tips. J. Assoc. Lab. Automat. 13:24-32. Parry, J.V., Mortimer, P.P., Friderich, P., and Connell, J.A. 1997. Faulty washers and soiled micropipettors may generate false positive serological results. Clin. Diagn. Virol. 7:173-181. Petersen, J. and Nguyen, J. 2005. Comparison of absorbance and ßuorescence methods for determining liquid dispensing precision. J. Assoc. Lab. Automat. 10:82-87. Quintero, C., Rosenstein, C., Hughes, B., Middleton, R., and Kariv, I. 2007. Quality control procedures for dose-response curve generation using nanoliter dispense technologies. J. Biomolec. Screen. 12:891-899. Taylor, P.B., Ashman, S., Baddeley, S.M., Bartram, S.L., Battle, C.D., Bond, B.C., Clements, Y.M., Gaul, N.J., McAllister, W.E., Mostacero, J.A., Ramon, F., Wilson, J.M., Hertzberg, R.P., Pope, A.J., and Macarron, R. 2002. A standard operating procedure for assessing liquid handler performance in high-throughput screening. J. Biomolec. Screen. 7:554-569. Wu, J.T.Y., Wong, L.S.Y., and Bowlby, E.E. 2006. An automated high-throughput screening enzyme linked immunosorbent assay for Johne’s disease antibodies in bovine serum. J. Assoc. Lab. Automat. 11:323-330.

Overview of Liquid Handling Instrumentation for High Throughput Screening Applications

54 Volume 1

Current Protocols in Chemical Biology

Target IdentiÞcation by Diazirine Photo-Cross-Linking and Click Chemistry Andrew L. MacKinnon1 and Jack Taunton1,2 1

Program in Chemistry and Chemical Biology and Department of Cellular and Molecular Pharmacology, University of California San Francisco, San Francisco, California 2 Howard Hughes Medical Institute, University of California San Francisco, San Francisco, California

ABSTRACT Target identiÞcation is often the rate-determining step in deciphering the mechanism of action of biologically active small molecules. Photo-afÞnity labeling (PAL) represents a useful biochemical strategy for target identiÞcation in complex protein mixtures. This unit describes the use of alkyl diazirine-based photo-afÞnity probes and Cu(I)-catalyzed click chemistry to covalently label and visualize the targets of biologically active small molecules. A general method for afÞnity puriÞcation of probe-modiÞed proteins, useful for identiÞcation of protein targets, is also described. Curr. Protoc. Chem. Biol. 1:55-73 C 2009 by John Wiley & Sons, Inc.  Keywords: photo-afÞnity labeling r diazirine r click chemistry r target identiÞcation r afÞnity puriÞcation

INTRODUCTION Target identiÞcation is often the rate-determining step in deciphering the mechanism of action of biologically active small molecules. Photo-afÞnity labeling (PAL) represents a useful biochemical strategy for target identiÞcation in complex protein mixtures. PAL uses an analog of a biologically active small molecule, known as a photo-afÞnity probe, that bears photo-reactive and reporter functional groups, to identify macromolecular binding partners. The photo-afÞnity probe is designed and synthesized based on SAR (structure-activity relationships) of a parent small molecule having known biological activity. During PAL, the photo-afÞnity probe is incubated with a protein mixture and irradiated with UV light. Irradiation of the photo-reactive group generates a highly reactive chemical species (e.g., carbene, nitrene, or radical) that covalently cross-links the photo-afÞnity probe to its macromolecular binding partner(s) based upon the close proximity of the two constructs. Photo-cross-linked protein targets are then visualized by means of the reporter group (e.g., ßuorophore, biotin, or radioactive label). Covalent bond formation between the probe and targets enables the subsequent puriÞcation and identiÞcation of the targets using techniques such as SDS-PAGE, immunoprecipitation, biotin-streptavidin afÞnity puriÞcation, and mass spectrometry. AfÞnity puriÞcation of protein targets is often difÞcult with non–covalently bound small molecules, especially those with low to moderate binding afÞnity. The challenges are compounded with small molecules that target integral membrane proteins, which often show decreased function after solubilization with detergents, a prerequisite for afÞnity puriÞcation. There are several photoreactive functional groups frequently used in PAL (e.g., benzophenone, trißuoromethyl phenyl diazirine, aryl azide). Like most useful photo-afÞnity groups, the alkyl diazirine (Fig. 1A, I) is activated at a wavelength of light (∼355 nm) that is not damaging to protein(s). However, the alkyl diazirine holds unique advantages.

Current Protocols in Chemical Biology 1: 55-73, December 2009 Published online December 2009 in Wiley Interscience (www.interscience.wiley.com). DOI: 10.1002/9780470559277.ch090167 C 2009 John Wiley & Sons, Inc. Copyright 

Target ID by Crosslinking and Click Chemistry

55 Volume 1

biotin/TAMRA

A

N

1. protein mixture 2. h

N

PAL probe N2

PAL probe R

N

(I)

7

CN 1

O

R

R

H

H

NH

5

N

H N

O

O

NH

O O

N

H N

N

O

NH

O N

O

O OMe

NH

2

O

(IV)

O

N

O O

protein target

O

O

N

N

(III)

(II)

6

O

PAL probe

protein target

R = alkyl

B

N

biotin/TAMRA-azide CuSO4, TCEP, TBTA

PAL probe R

N

4

R 3 N

HUN-7293, 1

N

2

R= 3

Figure 1 (A) Generalized scheme for photo-afÞnity labeling and detection using a diazirine and alkyne-containing photoafÞnity probe (I). UV irradiation of the diazirine generates a carbene intermediate (II) that covalently cross-links to the protein target (III). The adduct is then detected by conjugation with an azide-containing reporter group under click chemistry conditions (IV). (B) Structures of the natural product HUN7293 ( 1), photo-afÞnity probe (2), and the photostable control compound (3).

First, it is compact in size, being nearly isosteric to a methyl group, and is accessed synthetically via an alkyl ketone. This allows installation of the diazirine at positions of a small molecule that would not tolerate larger, aryl-based photoreactive groups. Second, the carbene intermediate formed upon photoactivation of the diazirine (Fig. 1A, II) rapidly inserts into X-H bonds (X = N, S, O), as well as C-H bonds, to form stable covalent insertion products (Brunner, 1993). When not poised for insertion into bonds of the macromolecular target, the alkyl carbene intermediate undergoes rapid quenching by solvent or internal rearrangement to a stable oleÞn side product (Ford et al., 1998). The alkyl diazirine is stable toward acidic and basic conditions and toward ambient light encountered during routine chemical synthesis. Several improved methods for the synthesis of alkyl diazirines starting from alkyl ketone precursors have been recently reported (MacKinnon et al., 2007; Bond et al., 2009). Heterobifunctional amine-reactive alkyl diazirine cross-linkers, as well as alkyl diazirine-containing amino acid analogs, are commercially available (Pierce, Thermo ScientiÞc).

Target ID by Crosslinking and Click Chemistry

Cu(I)-catalyzed click chemistry is an important method for bioconjugation of probelabeled proteins with reporter groups (Best, 2009). During the click reaction, Cu(I) catalyzes a highly selective 1,3 dipolar cycloaddition reaction between a terminal alkyne group and an azide group to form a stable triazole product (Fig. 1A, III, IV). The terminal alkyne is typically present in the small-molecule probe, while the azide is present in a ßuorescent or biotinylated reporter group. Alternatively, the azide can be incorporated

56 Volume 1

Current Protocols in Chemical Biology

into the probe and the alkyne incorporated into the reporter. However, this arrangement has been shown to produce higher background labeling of proteins (Speers and Cravatt, 2004). Following covalent labeling of protein targets via a (latent) chemically reactive moiety within the probe, probe-modiÞed proteins are conjugated to the azide-bearing reporter under click chemistry conditions (Fig. 1A, IV). The reporter group is thus introduced after covalent bond formation between the probe and target protein. This approach thereby avoids directly introducing a bulky reporter into the small-molecule probe, which could perturb the interaction between probe and target. The terminal alkyne (or azide) is extremely compact and therefore minimally perturbs the structure of the small molecule, while providing the chemical functionality necessary for detection and afÞnity puriÞcation of targets. A variety of azide and alkyne reporters designed for use in bioconjugate click reactions have been described (Speers and Cravatt, 2004), and many are commercially available (Invitrogen). The Basic Protocol presented below describes a method for photo-afÞnity labeling and detection of photo-cross-linked proteins in complex protein mixtures. The method requires a photo-afÞnity probe that bears both an alkyl diazirine photoreactive group and a terminal alkyne group. The scope and limitations of the method, as well as essential controls, parameters, and variables, are discussed. Key design strategies that lead to the synthesis of photo-afÞnity probe 2 (labeled in Fig. 1B), as well as a summary of pros and cons of commonly used photoreactive groups, are also discussed. The Basic Protocol describes the method applied on an analytical scale, followed by a Support Protocol that describes scale-up of the reactions, post–click chemistry work-up, and afÞnity puriÞcation of labeled proteins using monomeric avidin agarose or antibodies directed against carboxy-tetramethylrhodamine (TAMRA). The afÞnity-puriÞcation protocol is useful for purifying and identifying unknown protein targets of biologically active small molecules.

STRATEGIC PLANNING Design and synthesis of a photo-afÞnity probe can be one of the major challenges of applying PAL to small-molecule target identiÞcation. Structure-activity-relationships (SAR) of the parent molecule typically guide the choice and placement of the photoreactive or reporter groups within the parent scaffold. For example, in designing photo-afÞnity probe 2 (Fig. 1B), a detailed SAR study of the HUN-7293 scaffold (Chen et al., 2002) revealed that the N-methoxytryptophan side chain at position 5 (Fig. 1B) could be replaced with a smaller, phenylalanine side chain without signiÞcantly altering its biological activity. While this suggested that a phenyl azide at this position might also preserve biological activity, photoactivation of the phenyl azide requires a wavelength of light (∼260 nm) that is damaging to protein structures. The SAR study also suggested that a larger aromatic photoreactive group at this position, such as benzophenone, would signiÞcantly reduce biological activity. To preserve biological activity, we therefore sought a compact, hydrophobic photoreactive group that could be placed into one of the many hydrophobic alkyl side chains of the molecule (Fig. 1B). The diazirine represented a suitable choice. Being nearly isosteric with a methyl group, the diazirine was intended to replace a terminal methyl group of a leucine residue in HUN-7293. To accomplish this, we synthesized a diazirine-containing leucine analog, known as photo-leucine (Suchanek et al., 2005), starting from an alkyl ketone precursor, and used this precursor in the total synthesis of photo-afÞnity probe 2 (MacKinnon et al., 2007). We also required a method a detect photo-cross-linked proteins. The SAR indicated a tolerance for smaller side chains at position 1. We therefore installed a propargyl group at this position to enable detection of photo-cross-linked proteins using click chemistry. In parallel, we also synthesized a photostable control compound (labeled 3 in Fig. 1B) that was used in control experiments for distinguishing background from speciÞc

Target ID by Crosslinking and Click Chemistry

57 Current Protocols in Chemical Biology

Volume 1

Table 1 Comparison of Commonly Used Photoreactive Groups

Photoreactive group BeneÞts

Downsides

Benzophenone

Photoactivation at ∼350 nm is reversible, leading to high cross-linking yields with proteins. Selective for insertion into C-H σ bonds over bulk solvent (Dorm´an and Prestwich, 1994). Chemically stable.

Large size. Reported to selectively react with methionine residues in proteins leading to inaccurate determination of probe-binding sites (Wittelsberger et al., 2006).

Trißuoromethyl phenyl diazirine

Generates a highly reactive carbene intermediate upon photoactivation at ∼350 nm. Photoinsertion of the carbene into proteins can proceed in high (>70%) yield (Brunner, 1993).

Relatively large size. Insertion products may be reversible under some conditions (Platz et al., 1991). Can undergo UV light-induced rearrangement to electrophilic diazo isomer (Brunner, 1993), leading to nonspeciÞc labeling. Challenging to synthesize.

Alkyl diazirine

Small size. Generates highly reactive carbene intermediate upon photoactivation at ∼350 nm. Good yield of insertion into protein targets (∼25%, MacKinnon et al., 2007). Synthesized directly from the ketone precursor.

May undergo UV light-induced rearrangement to electrophilic linear diazo isomer (Brunner, 1993), leading to nonspeciÞc labeling. Intramolecular rearrangement of the alkyl carbene intermediate may compete with intermolecular insertion into proteins (Ford et al., 1998).

Phenyl azide

The singlet nitrene intermediate formed on photoactivation is highly reactive. Photoactivation of nitro-substituted aryl azides occurs at ∼340 nm and is therefore not damaging to protein. Perßuoro phenyl azides react primarily via the singlet nitrene intermediate (Brunner, 1993). Easily synthesized.

Unsubstituted phenyl azides require activation at short wavelengths (∼260 nm) that are damaging to protein. In nonperßuorinated phenyl azides, the singlet nitrene intermediate is prone to ring-expansion to a long-lived electrophilic species (Brunner, 1993), resulting in nonspeciÞc labeling. Phenyl azide is chemically less stable than other photoreactive groups.

photo-cross-links to protein targets (discussed in Critical Parameters). Compounds 2 and 3 were found to maintain nanomolar potency in biological assays, indistinguishable from the natural product 1 (all labeled in bold in Fig. 1B). Ideally, SAR-guided design of photo-afÞnity probes should involve replacing elements found in the parent molecule with photoreactive groups having similar chemical properties. Several types of photoreactive groups that differ in size, hydrophobicity, and ease of chemical synthesis are commonly used (see Table 1). Due to intrinsic differences in chemical and photophysical properties between these groups, it is difÞcult to predict a priori which one will be best suited for a speciÞc PAL application. In some cases, it may be possible to test different photoreactive groups in the same position of a probe, or the same photoreactive group at different positions within the probe. In all cases, it is important to evaluate the biological activity of photo-afÞnity compounds. A brief comparison of beneÞts and downsides of commonly used photoreactive groups is presented in Table 1. For more detailed descriptions of these photoreactive groups and their use in PAL, see Brunner (1993), Dorm´an and Prestwich (1994), Dorm´an (2000), and Sadakane and Hatanaka (2006).

Target ID by Crosslinking and Click Chemistry

Another important consideration in planning a PAL experiment is obtaining a photostable competitor compound to be used in a critical control experiment to distinguish background from speciÞc photo-cross-links to protein targets (discussed in Critical Parameters). The competitor is often the parent compound or other competitive antagonist. Considerable time and effort may be required to synthesize the photostable competitor.

58 Volume 1

Current Protocols in Chemical Biology

DIAZIRINE PHOTOACTIVATION AND Cu(I)-CATALYZED CLICK CHEMISTRY FOR COVALENT LABELING AND DETECTION OF PROTEIN TARGETS

BASIC PROTOCOL

This Basic Protocol describes the use of diazirine- and alkyne-containing photo-afÞnity probes for detection of protein targets in complex protein mixtures. Following diazirine photoactivation to covalently modify macromolecular binding partners, Cu(I)-catalyzed click chemistry is used to install a ßuorescent or biotin reporter group for visualizing probe-modiÞed proteins. While the method is described using photo-afÞnity probe 2 (Fig. 1B) in a crude preparation of endoplasmic reticulum (ER) microsomes, it should serve as a general experimental guide for other PAL experiments. Critical experimental variables and essential controls are discussed.

Materials Endoplasmic reticulum (ER) microsomes (∼1 mg/ml total protein) or other soluble or membrane protein lysate containing the unknown macromolecular target, in PBS (see recipe for PBS) 0.8 mM stock solution of photostable competitor compound (labeled 3 in Fig. 1) Dimethylsulfoxide (DMSO) 20 μM stock solution of photo-afÞnity probe (labeled 2 in Fig. 1) in DMSO 10% (w/v) sodium dodecyl sulfate (SDS) in H2 O 5 mM TAMRA-azide (labeled 4 in Fig. 2) or biotin-azide (labeled 5 in Fig. 2), synthesized by published methods (Speers and Cravatt, 2004; Weerapana et al., 2007); similar reagents are available commercially from Invitrogen, e.g., PEG4 carboxamide-6-azidohexanyl biotin (Fig. 2) 1.7 mM TBTA in 80% t-butanol/20% DMSO (see recipe) 50 mM CuSO4 in H2 O 50 mM Tris(2-carboxyethyl)phosphine (TCEP) in H2 O, adjusted to pH ∼7 with 1 M NaOH; prepare immediately before use 6× Laemmli sample buffer (see recipe) Fluorescent molecular weight markers (Pierce) 96-well plate or other open, shallow container 1000 W Hg(Xe) lamp (Oriel Instruments, model 66923) with band-pass Þlter for irradiation at ∼355 nm (Oriel Instruments, cat. no. 59810) and a Þlter to absorb heat (Oriel Instruments, cat. no. 59044); http://www.oriel.com/ 0.5-ml polypropylene microcentrifuge tubes Typhoon 9400 phosphor imager (Amersham) Additional reagents and equipment for SDS-PAGE (e.g., Gallagher, 2006) and immunoblotting (western blotting ; e.g., Gallagher et al., 2008) Set up samples 1. In 0.5-ml polypropylene tubes, prepare Þve samples (labeled A to E), each containing 19 μl of ER microsomes at a total protein concentration of ∼1 mg/ml in PBS. Sample A is the experimental sample Sample B is the competition control, Sample C is the negative PAL probe control Sample D is the negative UV-irradiation control Sample E is the negative click chemistry control. Other protein lysates (e.g., cytosolic proteins, subcellular fractions, crude plasma membrane fractions, or whole-cell lysates) containing the unknown protein target of the small molecule can also be tested. Prepare the lysate at a concentration of between 0.5 and 10 mg/ml total protein in a buffer compatible with the click chemistry (see Critical Parameters).

Target ID by Crosslinking and Click Chemistry

59 Current Protocols in Chemical Biology

Volume 1

N

O

N

COOH H N

H S

TAMRA-azide (4)

O NH

O

H

HN

N H

N3

TEV recognition sequence

O

OH

H 2N O N H

H N O

O N H

H N

OH H N

O N H

O

O O N H

O O

H N

O

H N

N H

O

NH 2

O

O N H

O

H N O

O N H

H N

O NH 2

O

OH N3

biotin-azide (5) O HN H

NH H

H N

S O

O O 4

N H

N3 6

PEG4 carboxamide-6-azidohexanyl biotin

Figure 2 Structures of TAMRA-azide (4) and biotin-azide (5) used in the protocols in this unit, and the structure of a commercially available biotin-azide reagent (Invitrogen, cat. no. B10184).

2. To sample B, add 0.5 μl of a 0.8 mM stock solution in DMSO of the photo-stable competitor compound (3) for a Þnal concentration of 20 μM. Add 0.5 μl of DMSO to samples A, C, D, and E and incubate all reactions for 15 min at 0◦ C. Preincubation of sample B with a large excess of the photo-stable competitor (3) should presaturate relevant protein targets. This controls for nonspeciÞc photo-cross-linking during irradiation and is used to distinguish speciÞcally photo-cross-linked proteins from nonspeciÞc background. Typically a 10- to 100-fold molar excess of competitor is used for this control. Optimal preincubation times and temperatures may vary depending on the kinetics of small-molecule binding to the protein target(s).

Introduce photo-afÞnity probe and perform cross-linking 3. To samples A, B, D, and E add 0.5 μl of a 20 μM stock solution in DMSO of the photo-afÞnity probe (2) for a Þnal concentration of 500 nM. To sample C add 0.5 μl of DMSO. Incubate the reactions for an additional 15 min at 0◦ C. The optimal concentration of the photo-afÞnity probe (2) and the optimal time and temperature of the incubation may depend on the particular system under study. The PAL probe is typically tested at between 0.1 and 10 μM. The concentration of DMSO in the Þnal reaction should be kept as low as possible.

4. Transfer each reaction to a well of a 96-well plate or other shallow container. A shallow dish serves to maximize the surface area of the liquid for good exposure during the irradiation step. Target ID by Crosslinking and Click Chemistry

60 Volume 1

Current Protocols in Chemical Biology

5. Position the reactions ∼6 cm from the source of a 1000-W Hg(Xe) lamp equipped with a band-pass Þlter for irradiation at ∼350 nm. Irradiate samples A, B, C, and E for 1 min. Keep sample D protected from irradiation with aluminum foil. The samples can be irradiated simultaneously, provided they all lie within the boundaries of the incident light. Alternatively, samples can be irradiated sequentially. Irradiation of the diazirine releases N2(g) and generates a carbene intermediate that covalently cross-links the photo-afÞnity probe to the protein target. The half-life of the diazirine (λmax ∼355 nm), and thus the optimal irradiation time, depends on the wavelength of irradiation, the wattage of the UV light source, and the distance between the sample and the source (i.e., the power per unit area). A 1000-W Hg(Xe) lamp provides an intense source of radiation and requires short (≤ 1 min) irradiation times. A lower-intensity longwavelength UV lamp that emits at ∼365 nm is sufÞcient for diazirine photoactivation, but usually requires longer irradiation times (∼5 to 10 min for a 100-W lamp).

6. Transfer 19.5 μl each from samples A to E into new, labeled 0.5-ml polypropylene tubes for the click reaction.

Perform click chemistry 7. Add 2.5 μl of 10% SDS to each reaction and mix by vortexing. Addition of SDS denatures proteins and exposes the terminal alkyne to the click reagents.

8. Add 0.5 μl of 5 mM TAMRA-azide (labeled 4 in Fig. 2) or 5 mM biotin-azide (labeled 5 in Fig. 2) to each reaction. Other reporter-azide reagents are equally suitable and are commercially available (e.g., PEG4 carboxamide-6-azidohexanyl biotin, Invitrogen; shown in Fig. 2). Biotin-azide (5) has a TEV protease recognition sequence positioned between the biotin and azide groups (Fig. 2). This feature can be useful for proteolytically cleaving biotinylated proteins or peptides from a streptavidin afÞnity matrix using TEV protease (Weerapana et al., 2007).

9. Prepare a master mix of the catalyst immediately before use by combining:

1.5 volumes 1.7 mM TBTA in 80% t-butanol/20% DMSO 0.5 volumes 50 mM CuSO4 0.5 volumes 50 mM TCEP Vortex to mix. The catalyst master mix should have a faint blue color and be heterogeneous.

Mix again, then add 2.5 μl of catalyst mix to samples A, B, C, and D and mix by vortexing. 10. Prepare mock catalyst mix as described in step 9, but substitute deionized water for the CuSO4 . Add 2.5 μl of mock catalyst mix to sample E and mix by vortexing. Without CuSO4 in the mock catalyst, the click reaction should not proceed. Sample E therefore serves as a negative click chemistry control.

11. Incubate the reactions 30 min at 32◦ C. Incubation for 1 hr at room temperature is also sufÞcient for labeling in the click reaction. Following the incubation, reactions can be diluted ∼10-fold in afÞnity puriÞcation buffer (see recipe) to reduce the concentration of SDS, and immunoprecipitated using speciÞc antibodies directed against candidate target proteins. Target ID by Crosslinking and Click Chemistry

61 Current Protocols in Chemical Biology

Volume 1

Resolve proteins by electrophoresis 12. After the incubation, add 5 μl of 6× Laemmli sample buffer to each tube and mix. It is not necessary to heat the samples after addition of sample buffer. Some proteins, including hydrophobic membrane proteins, irreversibly aggregate upon heating in the presence of the click reagents.

13. Resolve 12.5 μl of samples A to E and 5 to 10 μl of ßuorescent or broad molecular weight markers by SDS-PAGE (Gallagher, 2006). Run the gel until the dye front has completely exited the gel. Running the dye front (which also contains the click chemistry reagents) off the gel ensures less carryover of free TAMRA-azide (labeled 4 in Fig. 2) into the imaging step (step 15). Snap-freeze the remaining samples in liquid N2 and store at −80◦ C. Storage of samples at −20◦ C leads to an increase in nonspeciÞc background labeling due to the click chemistry reagents. The stored samples are stable for at least 1 week.

14. Wash the gel three times, each time for 10 min, with deionized water. The gel is thoroughly washed with several changes of deionized water to help remove residual traces of free TAMRA-azide from the gel.

Scan gels/perform immunoblotting 15a. If the click reaction was performed with TAMRA-azide (labeled 4 in Fig. 2): Scan the gel using a Typhoon ßuorescent gel scanner (excitation wavelength 532 nm, emission wavelength 580 nm). 15b. If the click reaction was performed with biotin-azide (labeled 5 in Fig. 2) or other biotin-azide: Transfer proteins to nitrocellulose or PVDF membrane with a western blot transfer apparatus (Gallagher et al., 2008) and probe for biotinylated proteins using streptavidin-HRP. SUPPORT PROTOCOL

AFFINITY PURIFICATION OF PROBE-MODIFIED PROTEINS In this Support Protocol, probe-modiÞed proteins labeled with TAMRA-azide (labeled 4 in Fig. 2) or biotin-azide (labeled 5 in Fig. 2) under click chemistry conditions are afÞnity puriÞed. This is accomplished using monomeric avidin agarose (Pierce) for biotin-labeled proteins, or antibodies directed against TAMRA (Invitrogen) for TAMRA-labeled proteins. The use of monomeric avidin agarose or anti-TAMRA antibodies enables mild elution of labeled proteins. The protocol can be followed after optimizing the photolabeling and click reaction steps described in the Basic Protocol. The Support Protocol is useful for ultimately identifying the protein target(s) using techniques such as Edman sequencing or mass spectrometry.

Additional Materials (also see Basic Protocol)

Target ID by Crosslinking and Click Chemistry

Protein mixture labeled with photo-afÞnity probe (Basic Protocol, steps 1 to 5) Liquid N2 Acetone cooled to −20◦ C 1% SDS in PBS (see recipe for PBS) AfÞnity puriÞcation buffer (see recipe) Protein A–Sepharose beads (GE Healthcare) Anti-TAMRA antibody (Invitrogen, cat. no. A6397) Monomeric avidin–agarose beads (Pierce) Wash buffer (see recipe) Elution buffer (see recipe)

62 Volume 1

Current Protocols in Chemical Biology

Polyallomer 1.5-ml microcentrifuge tubes (Beckman-Coulter) Benchtop ultracentrifuge Sonicating water bath Refrigerated microcentrifuge Rotating tube mixer Perform photo-activation and click chemistry 1. Complete steps 1 to 5 of the Basic Protocol on a 25× to 50× scale (e.g., 0.7 ml of 1 mg/ml total protein). The complete set of control experiments described in the Basic Protocol are not necessary here if they were previously performed in analytical-scale experiments. During scale-up, a larger vessel, such as a well of a 12- or 24-well tissue culture dish, can be used during the irradiation step. For proper irradiation, it is important that the entire sample lie within the bounds of the incident light (a circle ∼6 cm in diameter using the lamp setup described here). A 1-min irradiation of a 0.7-ml of sample in a well of a 24-well dish, as used here, is sufÞcient for complete photoactivation of the diazirine. For sample volumes signiÞcantly larger than 0.7 ml, longer irradiation times, coincident sample mixing, or irradiation in batches may be required. A small aliquot of the mixture can be removed and analyzed by SDS-PAGE following click chemistry to determine the extent of cross-linking.

2. Following irradiation, transfer the mixture containing the ER microsomes to a 1.5ml polyallomer microcentrifuge tube and sediment the microsomes in a benchtop ultracentrifuge 10 min at 50,000 × g, 4◦ C. Sedimentation concentrates the microsomes and allows the click reaction to be conducted on a smaller volume. For other types of protein mixtures that cannot be concentrated by sedimentation, other protein-precipitation methods can be tested in pilot experiments, or the mixture can be directly subjected to the click chemistry.

3. Aspirate the supernatant and resuspend the microsomes in 97.5 μl of PBS. 4. Follow steps 7 to 11 of the Basic Protocol, scaling up the click reagent volumes according to the volume of the resuspended microsome pellet. For 97.5 μl of resuspended microsomes, add 12.5 μl 10% SDS, 2.5 μl TAMRA-azide (labeled 4 in Fig. 2) or biotin-azide (labeled 5 in Fig. 2), and 12.5 μl of the catalyst mix (see steps 9 and 10 in the Basic Protocol).

Precipitate and redissolve proteins 5. Following the click reaction, remove a 5-μl aliquot, add 6× Laemmli sample buffer (see step 12 of Basic Protocol), snap-freeze in liquid nitrogen, and store the sample at −80◦ C. The saved sample should be stable for at least one week at −80◦ C. This saved aliquot represents the “input” into the afÞnity puriÞcation. At this stage, detection of speciÞc photo-cross-linked proteins can be determined by SDS-PAGE (step 13 of the Basic Protocol). An aliquot of the sample can also be diluted ∼10-fold in afÞnity puriÞcation buffer (see recipe) to reduce the concentration of SDS, and immunoprecipitated using speciÞc antibodies directed against candidate target proteins.

6. To the remaining sample (120 μl), add 0.5 ml of acetone, cooled to −20◦ C, for a Þnal concentration of ∼80% (v/v). Vortex brießy and place at −20◦ C for 30 min. A white precipitate should form, which contains precipitated proteins. Cold acetone precipitation removes the large molar excess of free TAMRA-azide (4) or biotin-azide (5) that would interfere with the binding to the TAMRA-antibody or monomeric avidin beads,

Target ID by Crosslinking and Click Chemistry

63 Current Protocols in Chemical Biology

Volume 1

respectively. The azide label is soluble in acetone while the proteins are not, permitting their separation. Following addition of cold acetone, the sample can also be stored overnight at −20◦ C

7. Sediment the precipitated protein in a by microcentrifuging 10 min at 20,000 × g, 4◦ C. A white pellet containing precipitated protein should be observed in the bottom of the tube.

8. Aspirate the supernatant and add 0.5 ml of cold acetone. 9. Use a sonicating water bath to break up and disperse the precipitated protein pellet in the acetone. Avoid heating the sample during this process by sonicating only for brief periods.

10. Return the sample to −20◦ C for 10 min. Longer incubation times at −20◦ C may improve the recovery of precipitated protein.

11. Repeat steps 7 to 11 of this protocol two more times. It is essential to remove all traces of free TAMRA-azide (4) or biotin-azide (5) for maximum yield of afÞnity-puriÞed proteins.

12. Aspirate the supernatant and air-dry the pellet brießy for ∼10 min at room temperature. Air drying removes residual acetone from the protein pellet. Do not over-dry the pellet as it will become difÞcult to resolubilize.

13. Add 50 μl of 1% SDS in PBS to the pellet and gently dislodge the pellet from the side of the tube by vortexing and/or sonication. Do not use a pipet tip to dislodge the pellet, as the pellet may stick to the tip. The SDS aids in resolubilizing the protein pellet.

14. Once the pellet has completely dissolved, dilute the sample with 0.5 ml of afÞnity puriÞcation buffer. The SDS must be diluted to ≤0.1% for efÞcient puriÞcation in steps 16 to 19. The NP-40 detergent helps “mask” the SDS in mixed micelles.

15. Remove a 20-μl aliquot of the mixture, add 4 μl of 6× Laemmli sample buffer for a Þnal concentration of 1×, snap-freeze in liquid nitrogen, and store at −80◦ C. The sample should be stable for at least 1 week at −80◦ C. This sample can be used to evaluate the efÞciency of the precipitation and resolubilization (steps 6 to 14) by comparison to an aliquot of the “input” click reaction (saved in step 5).

Perform afÞnity chromatography Follow steps 16a to 17a and 24a for TAMRA-labeled proteins; follow steps 16b to 17b and 24b for biotin-labeled proteins. For TAMRA-labeled proteins 16a. Equilibrate 30 μl of protein A–Sepharose in afÞnity puriÞcation buffer and prepare a 50% slurry in the same buffer.

Target ID by Crosslinking and Click Chemistry

17a. Add 30 μl of the protein A–Sepharose slurry and 5 μl of anti-TAMRA antibody (directly as received from manufacturer) to the resolubilized protein pellet.

64 Volume 1

Current Protocols in Chemical Biology

For biotin-labeled proteins 16b. Prepare the monomeric avidin beads according to the manufacturer’s direction, equilibrate with afÞnity puriÞcation buffer, and prepare a 50% slurry in afÞnity puriÞcation buffer. 17b. Add 50 μl of the monomeric avidin agarose slurry to the resolubilized protein pellet. 18. Incubate samples (from step 17a or 17b) on a rotating tube mixer for 3 hr at 4◦ C. Samples can also be incubated overnight at 4◦ C.

19. Sediment the agarose beads for 1 min at 10,000 × g, 4◦ C, in a microcentrifuge and remove the supernatant. 20. Save a 20-μl aliquot of the supernatant, add 4 μl of 6× Laemmli sample buffer for a Þnal concentration of 1×, snap-freeze in liquid nitrogen, and store at −80◦ C. The saved sample should be stable for at least 1 week at −80◦ C. The saved aliquot of the supernatant can be used to evaluate the efÞciency of the afÞnity puriÞcation step by comparison to an equal aliquot of the resolubilized pellet (saved in step 15).

21. Add 1 ml of afÞnity puriÞcation buffer to the sedimented agarose beads and rotate on a rotating tube mixer for 10 min at 4◦ C. Longer mixing times may be more effective at removing nonspeciÞcally bound proteins from the agarose resin.

22. Repeat steps 19 to 21 (the supernatants from the washes can be discarded). 23. Repeat steps 19 to 21 two more times, but replace the afÞnity puriÞcation buffer with wash buffer.

Elute and resolve proteins 24a. For TAMRA-labeled proteins: After the Þnal wash, elute bound proteins with 50 μl of 1× Laemmli sample buffer for 20 min at room temperature. 24b. For biotin-labeled proteins: After the Þnal wash, elute bound proteins with 50 μl of elution buffer for 20 min at room temperature, or as described by the manufacturer. The NP-40 detergent in the elution buffer is included to help maintain proteins in solution after elution. The detergent may not be required when eluting soluble proteins.

25. Resolve equivalent amounts of the “input” sample (saved in step 5, 4.5 μl), the resolubilized protein pellet (saved in step 15, 20 μl), the post-afÞnity puriÞed supernatant (saved in step 20, 20 μl), and eluent (saved in step 24, 1.8 μl) by SDS-PAGE (Gallagher, 2006). Scan the gel for ßuorescence (step 15 of the Basic Protocol) or transfer proteins to a nitrocellulose or PVDF membrane and probe with streptavidin-HRP (Gallagher et al., 2008). Comparison of the signal intensity of the “input” and resolubilized pellet samples indicates the efÞciency of the acetone precipitation and resolubilization steps (steps 6 to 15). Comparison of the signal intensity of the resolubilized pellet with post-afÞnity puriÞed supernatant indicates the efÞciency of the pull down (steps 16 to 19). Comparison of the signal intensity of the resolubilized pellet with the eluent indicates the recovery of labeled proteins (steps 21 to 24).

Target ID by Crosslinking and Click Chemistry

65 Current Protocols in Chemical Biology

Volume 1

REAGENTS AND SOLUTIONS Use deionized, distilled water in all recipes and protocol steps.

AfÞnity puriÞcation buffer 50 mM HEPES, pH 7.4 100 mM NaCl 1% (v/v) NP-40 or Triton X-100 Store up to 1 month at 4◦ C Elution buffer Phosphate-buffered saline (PBS; see recipe) containing: 2 mM D-biotin 1% (v/v) NP-40 or Triton X-100 Store up to 1 month at 4◦ C Laemmli sample buffer, 6× 12% (w/v) sodium dodecyl sulfate (SDS) 60% (v/v) glycerol 375 mM Tris·Cl, pH 8.0 (see recipe) 0.015% (w/v) bromphenol blue 30% (v/v) 2-mercaptoethanol Store up to 1 year at −20◦ C Phosphate-buffered saline (PBS) 137 mM NaCl 10 mM Na2 HPO4 2.7 mM KCl Store up to 1 year at 4◦ C TBTA in 80% t-butanol/20% DMSO Solid Tris(benzyltriazolylmethyl)amine (TBTA) is commercially available (Anaspec) or can be synthesized by published methods (Chan et al., 2004). The working stock is prepared by mixing one volume 8.5 mM TBTA stock in DMSO with four volumes t-butanol. This solution is stable for months when stored at −20◦ C.

Tris·Cl [tris(hydroxymethyl)aminomethane], 1 M Dissolve 121 g Tris base in 800 ml H2 O Adjust to desired pH with concentrated HCl Mix and add H2 O to 1 liter Approximately 70 ml of HCl is needed to achieve a pH 7.4 solution, and approximately 42 ml for a solution that is pH 8.0. IMPORTANT NOTE: The pH of Tris buffers changes signiÞcantly with temperature, decreasing approximately 0.028 pH units per 1◦ C. Tris-buffered solutions should be adjusted to the desired pH at the temperature at which they will be used. Because the pKa of Tris is 8.08, Tris should not be used as a buffer below pH ∼7.2 or above pH ∼9.0.

Wash buffer

Target ID by Crosslinking and Click Chemistry

50 mM HEPES, pH 7.4 500 mM NaCl 1% (v/v) NP-40 or Triton X-100 Store up to 1 month at 4◦ C

66 Volume 1

Current Protocols in Chemical Biology

COMMENTARY Background Information Identifying the target of a biologically active small molecule is a major step toward understanding its underlying mechanism of action. A traditional biochemical method for small-molecule target identiÞcation employs afÞnity chromatography of the target followed by identiÞcation by mass spectrometry or Edman degradation (Harding et al., 1989; Taunton et al., 1996; Ding et al., 2004). In this method, a complex protein mixture is passed over a resin matrix that has been covalently modiÞed with the small molecule of interest. The afÞnity matrix is stringently washed, and speciÞcally bound proteins are eluted, resolved by SDS-PAGE, and identiÞed. The success of this approach requires that the target and small molecule have a sufÞciently strong binding afÞnity (typically in the nM range) to survive the extensive washing steps required to reduce nonspeciÞc binding of proteins to the afÞnity matrix. In a recent variation of this technique, less stringent washing conditions coupled with highly sensitive quantitative mass spectrometry were used to identify speciÞc protein targets of inhibitors with micromolar afÞnity (Ong et al., 2009). The approach works best with soluble protein targets, since integral membrane proteins require detergent solubilization prior to chromatography, which often prevents binding to the immobilized small-molecule. PAL has several features that distinguish it from the traditional afÞnity chromatography approach. First, since photoactivation is performed under native conditions, PAL provides the opportunity for detection and identiÞcation of integral membrane protein targets (Colca et al., 2004; Saghatelian et al., 2004), an important class of proteins targeted by a large number of small-molecule drugs. PAL can also be used to characterize and map the ligand binding sites of known integral membrane proteins or other targets that lack high-resolution structural information (Al-Mawsawi et al., 2006; Xi et al., 2006). Furthermore, since PAL establishes a stable, covalent bond between the small-molecule probe and the target, the targets of even moderately potent smallmolecules can, in principle, be identiÞed. Several different click reaction conditions have been described in the literature. In the version described here, Cu(II)SO4 serves as the precursor to the Cu(I) species that catalyzes triazole formation between the terminal alkyne and azide. Tris-carboxyethyl

phosphine (TCEP) presumably reduces Cu(II) to Cu(I) in situ during preparation of the catalyst mix (Basic Protocol step 9). Tris(benzyltriazolylmethyl) amine (TBTA) is a polytriazole ligand that stabilizes the Cu(I) ion and enhances its catalytic activity in solution (Chan et al., 2004). Highly pure Cu(I)Br (99.999%) (Dieterich et al., 2007) or Cu(I) trißate (Strable et al., 2008) have also been used to effect the click reaction. However, we prefer in situ generation of Cu(I), since Cu(I)Br is sparingly soluble and aqueous Cu(I) solutions are prone to oxidation by dissolved oxygen. Tagging of probe-modiÞed proteins with reporter groups by click chemistry requires the presence of only a small alkyne or azide group in the probe. This avoids introducing a bulky reporter directly into the small-molecule scaffold. While there are examples of successful target identiÞcation using chemically reactive probes that have been directly modiÞed with a biotin reporter (Sin et al., 1997; Kwok et al., 2001), such bulky groups can perturb the interaction between the probe and protein targets. This point is exempliÞed in a study of compounds 6, 7, and 8 (labeled in bold in Fig. 3), where the rhodamine reporter is conjugated directly to the natural product scaffold via a triazole linkage and variable-length alkyl spacer arm. Photo-cross-linking to the target, Sec61α, in ER microsomes, was only ∼5% to 10% as efÞcient using 6 compared to photocross-linking followed by click chemistry using 2. Compound 7 cross-linked even less efÞciently than 6, and speciÞc photo-cross-links to Sec61α were undetectable using compound 8 (data not shown). The reduced photo-crosslinking yield presumably reßects a reduction in binding afÞnity after conjugating the molecule with a bulky rhodamine reporter. Introduction of the alkyne group preserved the nanomolar potency of the compound, while providing the chemical functionality needed to detect photocross-linked proteins in a second, click chemistry step. Introduction of a radiolabel into the smallmolecule probe is a widely used approach to detect probe-modiÞed proteins. While radiolabels are small in size, extremely sensitive, and offer a high ratio of signal to noise, radiolabeled probes can be costly to synthesize and radioactive materials require special handling and dedicated equipment. Click chemistry provides a nonradioactive alternative which is also highly sensitive, and has the added advantage of coincidently installing a chemical handle

Target ID by Crosslinking and Click Chemistry

67 Current Protocols in Chemical Biology

Volume 1

N

N

O O HN

OH

O

N

O

O

N HN

N

OH

N

N

O N N N

O O

O O

O

O

O

O

N

N O O

NH

NH

O

O O

NH

O N

O

O

H N

NH

O N

N

O

O

H N

N

O N

N

N

6

N

7 N

O O OH

HN

N

O N

N N

O O N O O

NH

NH

O O N

O

H N

N

O

N

N

8 Figure 3 Structures of compounds 6, 7, and 8, which have a ßuorescent reporter group (TAMRA) directly incorporated into the natural product scaffold.

Target ID by Crosslinking and Click Chemistry

68 Volume 1

Current Protocols in Chemical Biology

1% ) nt (

%)

ata

t (1

er n

e lle

elu

en t

(12

%)

sup

dp

id in t-av

po s

20

lize

act io n

(1% )

10

ubi

5

clic k re

3 ( M) h 200 119 108 86

sol

B

2

r e-

A

150 100 75 50

*

37

53

*

25 20 15

27

10 TAMRA fluorescence

Coomassie Strep-HRP

Figure 4 (A) Photo-cross-linking in ER microsomes with 2 (see Fig. 1) followed by click chemistry with TAMRA-azide (4; see Fig. 2) as described in the Basic Protocol. Sec61α is marked with an asterisk (Þgure adapted with permission from MacKinnon et al., 2007). (B) Photo-cross-linking in ER microsomes with 2 followed by click chemistry with biotinazide (5) and afÞnity puriÞcation using monomeric avidin as described in the Support Protocol. Samples representing the click reaction (lane 1), the resolubilized protein pellet following acetone precipitation (lane 2), the post-monomeric avidin supernatant (lane 3), and the eluent (lane 4) were resolved by SDS-PAGE, transferred to nitrocellulose, and probed for biotinylated proteins with streptavidin-HRP (Strep-HRP). Percentages indicate the fraction of the total sample that was loaded in each lane. The position of Sec61α is marked with an asterisk. The biotinylated protein at ∼21 kDa is a background band.

(biotin or TAMRA) that can be used to afÞnity purify and identify probe-modiÞed proteins. In some cases, this method can target the precise site of probe modiÞcation at the amino acid level (Adam et al., 2004; Speers and Cravatt, 2005; Weerapana et al., 2007). AfÞnity puriÞcation of probe-modiÞed targets is an essential step in target identiÞcation. PuriÞcation of biotinylated molecules with matrix-immobilized tetrameric streptavidin is a widely used technique that takes advantage of the extremely tight interaction between biotin and streptavidin (Kd ∼10−15 M). While this tight interaction permits stringent washing conditions resulting in low background, elution of speciÞcally bound material requires harsh conditions, typically boiling in SDS-PAGE sample buffer. Such conditions may not be suitable for some proteins, and SDS present in sample buffer is not compatible with many downstream applications including liquid chromatography/mass spectrometry (LC/MS). Several novel cleavable biotin reagents have been described (Verhelst et al., 2007; Weerapana et al., 2007) and others are commercially available (e.g., Pierce, cat.

nos. 21331 and 21442). These reagents allow elution of streptavidin-bound material chemically or enzymatically without disrupting the strong biotin-streptavidin interaction. However, many of these cleavable reagents still suffer from low elution efÞciencies. For example, biotin-azide (labeled 5 in Fig. 2), used in the present protocol, has a TEV protease recognition sequence positioned between the biotin and azide groups (Fig. 2). This feature is designed to permit elution of biotinylated proteins by incubation with TEV protease. However, in our case, we were unable to efÞciently elute bound Sec61α by incubation with TEV protease, possibly due to steric occlusion of the protease recognition sequence. To circumvent problems associated with tetrameric streptavidin and cleavable biotinazide reagents, the Support Protocol utilizes monomeric avidin (Pierce) for afÞnity puriÞcation of biotinylated targets (Fig. 4B). Monomeric avidin has a lower afÞnity for biotin (Kd ∼10−8 M), permitting elution of bound material with 2 mM biotin in PBS, a condition more suitable for diverse downstream applications. This unit also presents a

Target ID by Crosslinking and Click Chemistry

69 Current Protocols in Chemical Biology

Volume 1

mild capture and elution method for TAMRAlabeled proteins utilizing anti-TAMRA antibodies (Invitrogen) and protein A–Sepharose. AfÞnity puriÞcation with anti-TAMRA antibodies has the advantage that, following afÞnity puriÞcation, puriÞed protein targets can be proteolytically digested and probe-labeled peptide fragments visualized by ßuorescence detection (Adam et al., 2004; Okerberg et al., 2005). AfÞnity puriÞcation with monomeric avidin or anti-TAMRA antibodies both provide ∼25% yield of labeled proteins.

Critical Parameters

Target ID by Crosslinking and Click Chemistry

NonspeciÞc photo-cross-linking to highly abundant or “sticky” proteins in a complex protein mixture can be problematic in PAL. High background can obscure detection of speciÞc cross-links to less abundant proteins. Detection of speciÞc cross-links therefore depends on the relative abundance of the protein target (i.e., the amount of target protein per total protein). Subcellular or biochemical fractionation that enriches the sample for a putative target can be employed to improve the ratio of speciÞc signal-to-background noise. For example, while we could not detect speciÞc photo-cross-links to Sec61α in crude mammalian cell extract (data not shown), robust cross-linking was observed in puriÞed ER microsomes (Fig. 4). Even in the presence of high background, valuable information on speciÞc photo-cross-links to targets can be determined by immunoprecipitation against candidate proteins (Kukar et al., 2008). Extremely low-abundance targets may be difÞcult or impossible to detect, even in fractionated lysates. In such cases, it may be possible to detect these targets following an afÞnity puriÞcation step after the click reaction. Ultimately, detection of targets in a crude mixture of proteins depends on a favorable conßuence of variables, including the relative abundance of the target, the photo-cross-linking speciÞcity and yield, and the click chemistry yield (discussed below). Several controls are essential for distinguishing background from speciÞc photocross-linking. First, it is important to demonstrate that labeling of a putative target depends on the presence of both the photo-afÞnity probe and on UV irradiation. When labeling is observed in the absence of the probe or UV light, it may indicate background labeling due to the click reaction (discussed below). Secondly, it is important to conduct a competition experiment to control for the speciÞcity

of photo-cross-linking. This is done by incubating the protein sample with a large excess of a photostable competitor compound prior to UV irradiation. Cross-links to speciÞcally labeled proteins are dose-dependently competed by the photostable compound, whereas background cross-links are weakly competed or not competed at all (Fig. 4A). Background labeling that is independent of the photo-afÞnity probe or UV irradiation is due to nonspeciÞc labeling during the click reaction. The level of background appears to be strongly dependent on the total protein concentration, with lower concentrations of total protein yielding less background. The optimal concentration of total protein that provides the best ratio of speciÞc signal to background noise should be determined empirically. Reducing the concentration of the TAMRA-azide (4) or biotin-azide (5) in the reaction (we have gone as low as ∼25 μM) can also help reduce nonspeciÞc background labeling. We have found that background increases when click reactions are stored at −20◦ C, even after addition of Laemmli sample buffer. Quickfreezing samples in liquid N2 and storing at −80◦ C prevents this. A low concentration (0.1% to 1%) of SDS in the click reaction also reduces nonspeciÞc background labeling. TAMRA-azide (4) or other commercially available ßuorescent-azides (e.g., Invitrogen, cat. no. A10270 and T10182) used during the click reaction can trail though the gel lanes when reactions are resolved by SDSPAGE. Trailing ßuorophore contributes to background ßuorescence in the gel and reduces the sensitivity of in-gel ßuorescent scanning. To mitigate this problem, the dye front containing the ßuorophore should be run completely off the bottom of the gel during electrophoresis. Removal of excess free TAMRAazide by gel Þltration, dialysis, or protein precipitation (e.g., acetone precipitation as used in this protocol) prior to SDS-PAGE can also signiÞcantly reduce the background due to trailing ßuorophore. Gels should be thoroughly washed with several changes of deionized water prior to in-gel ßuorescence scanning, to remove traces of residual ßuorophore. The click reaction tolerates a fairly broad range of salt, buffer, and detergent concentrations, as well as a broad range of pH and temperatures. Metal chelators such as EDTA or EGTA should be avoided during preparation of the protein lysate, as these sequester the Cu(II) ions required for the reaction. We have also found (unpub. observ.) that 2-mercaptoethanol

70 Volume 1

Current Protocols in Chemical Biology

Table 2 Troubleshooting Guide for Target IdentiÞcation by Diazirine Photo-Cross-Linking and Click Chemistry

Problem

Cause

Solution

No signal observed on gel/blot following the click reaction

Incorrect wavelength for diazirine photoactivation; insufÞcient time for irradiation

Check the wavelength settings on the lamp; perform an irradiation time course

Concentration of the photo-afÞnity probe is too low

Titrate the photo-afÞnity probe into a Þxed amount of protein lysate and conduct PAL and click reactions

The photo-afÞnity probe does not ConÞrm that the photo-afÞnity probe is biologically bind the target of the parent molecule active

High background observed on gel/blot following the click reaction

Labeled proteins are not depleted during afÞnity puriÞcation

The target’s relative abundance is too low

Enrich the sample for the target by biochemical or subcellular fractionation; afÞnity purify following the click reaction

The click reaction failed

Prepare new reagent stocks; use freshly prepared TCEP solution

Non-speciÞc photo-cross-linking

Reduce the concentration of the photo-afÞnity probe or the concentration of total protein; enrich the sample for putative targets prior to PAL

NonspeciÞc background due to click reaction

Reduce the total protein concentration; reduce the concentration of the azide used during the click reaction; store reactions at −80◦ C

NonspeciÞc and speciÞc bands overlap on SDS-PAGE

Change SDS-PAGE conditions; for example, test different acrylamide concentrations, different buffer systems (e.g., Tris-Tricine), or 2-D gel electrophoresis

Contaminating free TAMRA-azide (4) or biotin-azide (5)

Repeat the acetone-precipitation steps to remove contaminating TAMRA-azide or biotin-azide and repeat the afÞnity puriÞcation

(2-ME) and dithiothreitol (DTT) inhibit the reaction at fairly low concentrations (∼100 μM). Labeling of some probe-modiÞed proteins requires or is improved by the presence of a low concentration of SDS (0.1% to 1%) or other detergent including deoxy-BigChaps, TX-100, NP-40, or sodium cholate.

Troubleshooting A troubleshooting guide is presented in Table 2.

Anticipated Results We recently utilized alkyl diazirine photoactivation and click chemistry methods to identify the molecular target of “cotransins,” a class of cyclic heptadepsipeptides derived from the fungal natural product HUN-7293 (labeled 1 in Fig. 1B). Cotransins inhibit cotranslational translocation of nascent proteins across the endoplasmic reticulum (ER) membrane in a signal-sequence-dependent manner (Garrison et al., 2005). Inhibition occurs at the level of insertion of the nascent protein into an

ER membrane–embedded multiprotein complex, termed the translocon, which recognizes signal sequences and forms a pore through which substrate proteins traverse (Osborne et al., 2005). Utilizing an alkyl diazirine–based photo-afÞnity probe that bears an alkyne handle (labeled 2 in Fig. 1B), we identiÞed an integral membrane protein subunit of the translocon complex, Sec61α, as the molecular target of cotransins (MacKinnon et al., 2007). Figure 4A shows a gel (adapted from MacKinnon et al., 2007) resulting from following the Basic Protocol in crude ER microsomes using photo-afÞnity probe 2 and click chemistry with TAMRA-azide (labeled 4 in Fig. 2). Three proteins were labeled in the presence of the PAL probe (lane 1), including one major band at ∼45 kDa (marked with an asterisk). Labeling of the major band was dependent on UV light (lane 5) and the PAL probe 2 (Lane 6), and was dose-dependently competed by preincubation with the photostable competitor 3 (Lanes 2 to 4), indicating speciÞc photo-cross-linking to this protein. Labeling of

Target ID by Crosslinking and Click Chemistry

71 Current Protocols in Chemical Biology

Volume 1

two other proteins at ∼60 kDa and ∼40 kDa also depended on the PAL probe and UV light, but was not competed by excess 3, indicating nonspeciÞc (i.e., nonsaturable) photo-crosslinking to these proteins. Coomassie staining indicated equal loading of protein across all samples. The ∼45 kDa protein, previously identiÞed as Sec61α (MacKinnon et al., 2007), is present at about 1% of total ER proteins and represented the major labeled protein. However, Sec61α did not represent a major Coomassie-stainable band, indicating the speciÞcity of the reaction and the ability to detect a protein target of moderate abundance in a complex mixture of ER proteins. Labeling that was independent of 2 and UV light was background due to the click reaction (Lanes 5 and 6). Figure 4B shows afÞnity puriÞcation of biotinylated proteins using monomeric avidin following photo-cross-linking with 2 and click chemistry with biotin-azide (labeled 5 in Fig. 2), as described in the Support Protocol. The position of Sec61α is marked with an asterisk. Comparison of equivalent aliquots of the starting click reaction (1% of total reaction, lane 1) with the resolubilized protein pellet (1% of total sample, lane 2), showed ∼50% protein recovery following the acetoneprecipitation protocol. Comparison of equivalent aliquots of the resolubilized protein pellet (lane 2) with the post-monomeric avidin supernatant (1% of total sample, lane 3), showed quantitative depletion of biotinylated proteins from the sample using monomeric avidin. Recovery of biotinylated proteins by mild elution with 2 mM biotin proceeded in ∼25% yield, as determined by comparison of lanes 2 and 4.

Literature Cited

Time Considerations

Ding, S., Wu, T.Y.H., Brinker, A., Peters, E.C., Hur, W., Gray, N.S., and Schultz, P.G. 2004. Synthetic small molecules that control stem cell fate. Proc. Natl. Acad. Sci U.S.A. 100:856-861.

A signiÞcant investment of time and resources is required for the design and synthesis of a photo-afÞnity probe that retains potent biological activity. Additional time may be required to synthesize a photostable competitor. Once a suitable probe is in hand, it can be rapidly tested in the photo-cross-linking and click reactions in <4 hr when using TAMRAazide (4) in the click reaction. The time required for optimization of the photo-crosslinking and click chemistry will vary, but may be completed in <1 week. AfÞnity puriÞcation and analysis of samples takes 1 to 2 days.

Acknowledgements Target ID by Crosslinking and Click Chemistry

This work was supported by the NIH (GM81644) and the Howard Hughes Medical Institute.

Adam, G.C., Burbaum, J., Kozarich, J.W., Patricelli, M.P., and Cravatt, B.F. 2004. Mapping enzyme active sites in complex proteomes. J. Am. Chem. Soc. 126:1363-1368. Al-Mawsawi, L.Q., Fikkert, V., Dayam, R., Witvrouw, M., Burke, T.R. Jr., Borchers, C.H., and Neamati, N. 2006. Discovery of a smallmolecule HIV-1 integrase inhibitor-binding site. Proc. Natl. Acad. Sci. U.S.A. 103:10080-10085. Best, M.D. 2009. Click chemistry and bioorthogonal reactions: Unprecedented selectivity in the labeling of biological molecules. Biochemistry 48:6571-6584. Bond, M.R., Zhang, H., Vu, P.D., and Kohler, J.J. 2009. Photocrosslinking of glycoconjugates using metabolically incorporated diazirinecontaining sugars. Nat. Protoc. 4:1044-1063. Brunner, J. 1993. New photolabeling and crosslinking methods. Annu. Rev. Biochem. 62:483-514. Chan, T.R., Hilgraf, R., Sharpless, K.B., and Fokin, V.V. 2004. Polytriazoles as copper(I)-stabilizing ligands in catalysis. Org. Lett. 6:2853-2855. Chen, Y., Bilban, M., Foster, C.A., and Boger, D.L. 2002. Solution-phase parallel synthesis of a pharmacophore library of HUN-7293 analogues: A general chemical mutagenesis approach to deÞning structure-function properties of naturally occurring cyclic (depsi)peptides. J. Am. Chem. Soc. 124:5431-5440. Colca, J.R., McDonald, W.G., Waldon, D.J., Leone, J.W., Lull, J.M., Bannow, C.A., Lund, E.T., and Mathews, W.R. 2004. IdentiÞcation of a novel mitochondrial protein (“mitoNEET”) cross-linked speciÞcally by a thiazolidinedione photoprobe. Am. J. Physiol. Endocrinol. Metab. 286:E252-E260. Dieterich, D.C., Lee, J.J., Link, A.J., Graumann, J., Tirrell, D.A., and Schuman, E.M. 2007. Labeling, detection and identiÞcation of newly synthesized proteomes with bioorthogonal noncanonical amino-acid tagging. Nat. Protoc. 2:532-540.

Dorm´an, G. 2000. PhotoafÞnity labeling in biological signal transduction. Top. Curr. Chem. 211:169-225. Dorm´an, G. and Prestwich G.D. 1994. Benzophenone photophores in biochemistry. Biochemistry 33:5661-5673. Ford, F., Yuzawa, T., Platz, M.S., Matzinger, S., and Fulscher, M. 1998. Rearrangement of dimethylcarbene to propene: Study by laser ßash photolysis and ab initio molecular orbital theory. J. Am. Chem. Soc. 120:4430-4438. Gallagher, S. 2006. One-dimensional SDS gel electrophoresis of proteins. Curr. Protoc. Mol. Biol. 75:10.2A.1-10.2A.37. Gallagher, S., Winston, S.E., Fuller, S.A., and Hurrell, J.G.R. 2008. Immunoblotting and

72 Volume 1

Current Protocols in Chemical Biology

immunodetection. Curr. Protoc. Mol. Biol. 83:10.8.1-10.8.28.

teases. Proc. Natl. Acad. Sci. U.S.A. 101:10001005.

Garrison, J.L., Kunkel, E.J., Hegde, R.S., Taunton, J. 2005. A substrate-speciÞc inhibitor of protein translocation into the endoplasmic reticulum. Nature 436:285-289.

Sin, N., Meng, L., Wang, M.Q.W., Wen, J.J., Bornmann, W.G., and Crews, C.M. 1997. The anti-angiogenic agent fumagillin covalently binds and inhibits the methionine aminopeptidase, MetAP-2. Proc. Natl. Acad. Sci. U.S.A. 94:6099-6103.

Harding, M.W., Galat, A., Uehling, D.E., and Schreiber, S.L. 1989. A receptor for the immuno-suppressant FK506 is a cis-trans peptidyl-prolyl isomerase. Nature 341:758760.

Speers, A.E. and Cravatt, B.F. 2004. ProÞling enzyme activities in vivo using click chemistry methods. Chem. Biol. 11:535-546.

Kukar, T.L., Ladd, T.B., Bann, M.A., Fraering, P.C., Narlawar, R., Maharvi, G.M., Healy, B., Chapman, R., Welzel, A.T., Price, R.W., Moore, B., Rangachari, V., Cusack, B., Eriksen, J., Jansen-West, K., Verbeeck, C., Yager, D., Eckman, C., Ye, W., Sagi, S., Cottrell, B.A., Torpey, J., Rosenberry, T.L., Fauq, A., Wolfe, M.S., Schmidt, B., Walsh, D.M., Koo, E.H., and Golde, T.E. 2008. Substrate-targeting γsecretase modulators. Nature 453:925-930.

Speers, A.E. and Cravatt, B.F. 2005. A tandem orthogonal proteolysis strategy for highcontent chemical proteomics. J. Am. Chem. Soc. 127:10018-10019.

Kwok, B.H.B., Koh, B., Ndubuisi, M.I., Elofsson, M., and Crews, C.M. 2001. The antiinßammatory natural product parthenolide from the medicinal herb Feverfew directly binds to and inhibits IκB kinase. Chem. Biol. 8:759-766.

Suchanek, M., Radzikowska, A., and Thiele, C. 2005. Photo-leucine and photo-methionine allow identiÞcation of protein-protein interactions in living cells. Nat. Methods 2:261-268.

MacKinnon, A.L., Garrison, J.L., Hegde, R.S., and Taunton, J. 2007. Photo-leucine incorporation reveals the target of a cyclodepsipeptide inhibitor of cotranslational translocation. J. Am. Chem. Soc. 129:14560-14561. Okerberg, E.S., Wu, J., Zhang, B., Samii, B., Blackford, K., Winn, D.T., Shreder, K.R., Burbaum, J.J., and Patricelli, M.P. 2005. Highresolution functional proteomics by active-site peptide proÞling. Proc. Natl. Acad. Sci. U.S.A. 102:4996-5001. Ong, S., Schenone, M., Margolin, A.A., Li, X., Do, K., Doud, M.K., Mani, D.R., Kuai, L., Wang, X., Wood, J.L., Tolliday, N.J., Koehler, A.N., Marcaurelle, L.A., Golub, T.R., Gould, R.J., Schreiber, S.L., and Carr, S.A. 2009. Identifying the proteins to which small-molecule probes and drugs bind in cells. Proc. Natl. Acad. Sci. U.S.A. 106:4617-4622. Osborne, A.R., Rapoport, T.A., and van den Berg, B. 2005. Protein translocation by the Sec61/SecY channel. Annu. Rev. Cell Dev. Biol. 21:529-550. Platz, M., Admasu, A.S., Kwiatkowski, S., Crocker, P.J., Imai, N., and Watt, D.S. 1991. Photolysis of 3-aryl-3-(trißuoromethyl) diazirines: A caveat regarding their use in photoafÞnity probes. Bioconjug. Chem. 2:337-341. Sadakane, Y. and Hatanaka, Y. 2006. Photochemical Þshing approaches for identifying target proteins and elucidating the structure of a ligandbinding region using carbene-generating photoreactive probes. Anal. Sci. 22:209-218. Saghatelian, A., Jessani, N., Joseph, A., Humphrey, M., and Cravatt, B.F. 2004. Activity-based probes for the proteomic proÞling of metalopro-

Strable, E., Prasuhn, D.E. Jr., Udit, A.K., Brown, S., Link, A.J., Ngo, J.T., Lander, G., Quispe, J., Potter, C.S., Carragher, B., Tirrell, D.A., and Finn, M.G. 2008. Unnatural amino acid incorporation into virus-like particles. Bioconjug. Chem. 19:866-875.

Taunton, J., Hassig, C.A., and Schreiber, S.L. 1996. A mammalian histone deacetylase related to the yeast transcriptional regulator Rpd3p. Science 272:408-411. Verhelst, S.H.L., Fonovic, M., and Bogyo, M. 2007. A mild chemically cleavable linker system for functional proteomic applications. Angew. Chem. Int. Ed. 46:1284-1286. Weerapana, E., Speers, A.E., and Cravatt, B.F. 2007. Tandem orthogonal proteolysis-activitybased protein proÞling (TOP-ABPP)—A general method for mapping sites of probe modiÞcation in proteomes. Nat. Protoc. 2:1414-1425. Wittelsberger, A., Thomas, B.E., Mierke, D.F., and Rosenblatt, M. 2006. Methionine acts as a “magnet” in photoafÞnity crosslinking experiments. FEBS Lett. 580:1872-1876. Xi, J., Liu, R., Rossi, M.J., Yang, J., Loll, P.J., Dailey, W.P., and Eckenhoff, R.G. 2006. Photoactive analogues of the haloether anesthetics provide high-resolution features from lowafÞnity interactions. ACS Chem. Biol. 1:377384.

Key References Best, M.D. 2009. See above. A recent review of bio-orthogonal click chemistry methods. Brunner, 1993. See above. An excellent introduction to the structure and chemistry of photoreactive groups and their use in photoafÞnity labeling in biological systems. Colca et al., 2004. See above. An excellent example of PAL for identifying a novel integral membrane target of a therapeutically relevant small molecule.

Target ID by Crosslinking and Click Chemistry

73 Current Protocols in Chemical Biology

Volume 1