Supporting Information
Ong et al. 10.1073/pnas.0900191106 SI Methods volume of DMSO is then added to 2 mg of heavy HeLaS3 as a 13 13 15 Materials and Reagents. L-arginine- C6 and L-lysine- C6 N2 control. 25 L of 50% of SM-bead is added to both light and were from Sigma Isotec. The cell culture media, Roswell Park heavy lysates in SC experiments. Memorial Institute-1640 (RPMI) deficient in arginine, lysine Affinity enrichments are incubated overnight (Ϸ16 h) on an and methioine, was a custom media preparation from Caisson end-over-end rotator at 4 °C. After incubation, the tubes are ϫ Laboratories. All other L-amino acids were obtained from spun at 1000 g on a benchtop centrifuge to pellet the beads. Sigma. Dialyzed serum was obtained from SAFC-Sigma. Trypsin The supernatant is aspirated, taking care to avoid disturbing the was from Promega and HeLaS3 and H1299 were a kind gift from beads. In BC experiments, beads are combined at the first wash Dr. James Bradner. A PC-12 rat pheochromocytoma cell line for subsequent washing steps. For SC experiments, each tube in stably expressing hErbB4 and GFP was generated with neomycin a set is washed with ModRIPA buffer at least once (twice with selection and 2 independent pcDNA3-neomycin (Invitrogen) high levels of soluble competition) to remove excess soluble constructs. Other cell lines were from ATCC. small molecule competitor. Beads from the 2 tubes were then be Antibodies against the following proteins were used: Anti- combined for later washing steps. In high stringency experi- FKBP1A and anti-MTAP (Abcam); anti-FKBP2 and anti- ments, a wash buffer (high stringency HS) containing ModRIPA FKBP5 R&D Systems); anti-FKBP4 (Bethyl Laboratories); anti- and 0.2% SDS was used to wash beads instead. After the third FKBP8 (US Biological); anti-FKBP10 (listed in the article as and final wash, beads are collected by spinning at 1000 ϫ g and FKBP9/10) (BD Transduction Laboratories). Secondary anti- the wash is aspirated leaving Ϸ20 L of buffer in the tube. bodies were all purchased from Sigma. Anti-GST antibody was from GE Healthcare (Piscataway, NJ). All other reagents and 1D-SDS/PAGE and MS Analysis. Proteins enriched in SILAC affinity chemicals used were of the highest grade available. pull-downs were reduced and alkylated, on bead, in 2 mM DTT and 10 mM iodoacetamide respectively. One part LDS buffer SILAC Media Preparation and Cell Culture Conditions. We followed (Invitrogen) was added to 3 parts sample (including beads) and all standard SILAC media preparation and labeling steps as tubes heated to 70 °C for 10 min. Proteins were resolved on a described in ref. 1. Briefly, 15 mg/L of L-methionine was added 4–12% gradient 1.5 mm thick Bis-Tris gel with Mes running to base media according to standard formulations for standard buffer (Nupage, Invitrogen) and Coomassie stained (Simply RPMI. This base media was divided into 2 and either ‘‘light’’ Blue, Invitrogen). Gel lanes were excised into 6 pieces and then forms of arginine and lysine or ‘‘heavy’’ L-arginine-U-13C6 (87.2 further cut into 1.5 mm cubes. The gel pieces were further mg/L) and L-lysine-13C615N2 (155 mg/L) was added to generate destained in a solution containing 50% EtOH and 50% 50 mM the 2 SILAC labeling mediums. Each medium with the full ammonium bicarbonate, then dehydrated in 100% EtOH before complement of amino acids was sterile filtered through a 0.22 addition of sufficient trypsin (12.5 ng/L) to swell the gel pieces M filter (Milipore). completely. An additional 100 L of 50 mM ammonium bicar- HeLa S3 cells were grown in RPMI labeling media, prepared bonate was added before incubating at 37 °C overnight on a as described above, supplemented with 2 mM L-glutamine, and thermomixer (Eppendorf). Enzymatic digestion was stopped by 5% dialyzed FBS plus antibiotics, in a humidified atmosphere the addition of 100 L of 1% TFA to tubes. A second extraction with 5% CO2 in air. Cells were grown for at least 6 cell divisions with 300 L of 0.1% TFA was combined with the first extract and in labeling media, initially growing in flasks but expanded into the peptides from each gel slice cleaned up on C18 StageTips (2). 2L spinner flasks on a magnetic stir plate to provide larger Peptides were eluted in 50 L of 80% acetonitrile/0.1% TFA and cultures. dried down in a evaporative centrifuge to remove organic solvents. The peptides were then resuspended by vortexing in 7 Biochemical Purification with Small Molecule Affinity Matrices. Sep- L of 0.1% TFA and analyzed by nanoflow-LCMS with an arate cultures of HeLaS3 cells SILAC labeled either with Agilent 1100 with autosampler and a LTQ-Orbitrap. Peptides L-arginine and L-lysine (light) or L-arginine-13C6 and L-lysine- were resolved on a 10 cm column, made in-house by packing a 13C6-15N2 (heavy) are lysed in ice-chilled ModRIPA buffer (low self-pulled 75 m I.D. capillary, 15 m tip (P-2000 laser based stringency buffer LS) containing 1% Nonidet P-40, 0.1% Na puller, Sutter Instruments) column with 3 m Reprosil-C18-AQ deoxycholate, 150 mM NaCl, 1 mM EDTA, 50 mM Tris, pH 7.5, beads (Dr. Maisch GmbH, Ammerbuch-Entringen, Germany) and protease inhibitors (Complete tablets, Roche Applied Sci- with an analytical flowrate of 200 nL/min and a 58 min linear ence, Indianapolis, IN). Lysates are vortexed intermittently while gradient (Ϸ0.57%B/min) from 0.1% formic acid in water to 0.1% chilled on ice for 10 min and clarified by spinning at 14,000 ϫ g. formic acid/90% acetonitrile. The run time was 108 min for a Protein concentrations of light and heavy lysates are estimated single sample, including sample loading and column recondi- with the Protein Assay Dye Reagent Concentrate (Bio-Rad, tioning. Hercules CA) and equalized. The protein concentrations of We used a MS method with a master Orbitrap full scan (60,000 lysates can vary between 1.7 to 2.2 mg/mL, affinity enrichments resolution) and data dependent LTQ MS/MS scans for the top are performed in lysate volumes between 1 mL to 1.4 mL in a 1.5 5 precursors (excluding z ϭ 1) from the Orbitrap scan. Each cycle mL microcentrifuge tube. was Ϸ2 secs long. MS raw files were processed for protein In bead control (BC) experiments, 2 mg of light HeLaS3 lysate identification and quantitation using extractmsn.exe (Thermo), is incubated with 25 L of 50% slurry of EtOH-bead while 2 mg Mascot (Ver. 2.1.03 Matrixscience), and academic software, of heavy lysate is incubated with 25 L of 50% small molecule DTASupercharge and MSQuant (CEBI, open-source http:// affinity beads (SM-bead). This is termed a ‘‘forward’’ label msquant.sourceforge.net). MS/MS peak lists in Mascot Generic experiment. In the ‘‘reverse’’ experiment, the lysates are Format were generated using extractmsn.exe and DTASuper- swapped for each bead type. In a ‘‘forward’’ soluble competitor charge and searched with Mascot using IPI human ver.3.32 experiment, the appropriate amount of small molecule (in (http://ebi.ac.uk) with a concatenated decoy database containing DMSO) is added to 2 mg of light HeLaS3 lysate. An equal randomized sequences from the same database. Common con-
Ong et al. www.pnas.org/cgi/content/short/0900191106 1of15 taminants like BSA, trypsin etc. were also added to the database. FKBP10/FKBP9 Antibody Cross-Specificity Determined by Immunopre- Variable modifications used were oxidized methionine, argi- cipitation and MS (IP-MS). In Western blot analysis experiments to 13 13 15 nine- C6, lysine- C6 N2, and carbamidomethyl-cysteine was a validate our IPL ligand SILAC TargetID data, we found the fixed modification. The precursor mass tolerance used in the FKBP10 antibody (BD Transduction Laboratories, Cat. No. search was 15 ppm and fragment mass tolerance was 0.7 Da. 610648) yielded a strong band at Ϸ63 kDa in IPL ligand Proteins with a minimum Mascot score of 66 (at least 1 peptide pull-downs with AP-1480, ProAP-1480, and AP-1497. This was with score Ͼ66) and peptides with score Ͼ20 were quantified by not consistent with our SILAC data for FKBP10, but matched MSQuant. Only proteins with a minimum of 2 quantifiable the compound specificity profile for FKBP9. Because FKBP9 peptides were included in our dataset. The false positive rate for and FKBP10 are very similar in length, 570 aa and 582 aa protein identification is Ͻ1% and Ͻ5% at the peptide level, as respectively and are 58% identical at the amino acid level determined using the decoy database strategy. (BlastP), we hypothesized that the antibody used in our exper- iment was cross-reacting with both proteins. We evaluated this Statistical Analysis of SC Experiments. To model log2 protein ratio with an IP-MS experiment. We incubated 10 g of antibody and values from SC experiments, we adapted the empirical Bayes 50 L of a 50% slurry of Protein G beads with 2 mg of HeLaS3 framework developed by Efron (3) to compute the posterior lysate in our usual pull-down conditions. We used Protein probability that a ratio value arises from the null distribution. G-bead alone with lysate in a separate pull-down as a control. Briefly, by application of Bayes’s theorem, this quantity is Beads were washed, proteins reduced, alkylated, resolved on a Ϸ computed as Pr(Z ϭ 0͉X) ϭ [Pr(X͉Z ϭ 0)Pr(Z ϭ 0)]/Pr(X) where 1D-SDS/PAGE gel and the molecular mass bands between 60 Z is a binary variable taking the value of 1 if the protein is bound to 98 kDa were excised for MS analysis. Our MS analysis by the compound, and X is a measured log2 SILAC ratio. We identified both FKBP9 and FKBP10 in the pull-down with the model Pr(X) using a Gaussian kernel estimator with pareto antibody and strikingly, the number of peptides and sequence distributions fit to 5% of the data at either tail, to avoid coverage for FKBP9 was higher than FKBP10. unreliable estimation in regions of data sparsity. The distribution for Pr(X͉Z ϭ 0)Pr(Z ϭ 0) is then inferred by fitting a Gaussian Surface Plasmon Resonance Experiments. The surface plasmon distribution using only the portion of Pr(X) arising from the resonance assays were conducted on a Biacore S51 instrument central two thirds of the data. using Biacore CM5 sensor chips. Ethanolamine, EDC, NHS, and P-20 surfactant were all obtained from GE Lifesciences. Anti- Annotation of Kinases as ‘‘True Positive’’ Set. We use the human GST antibody (GE LifeSciences) was directly immobilized kinome defined in ref. (4), specifically, the updated Excel through primary amines using standard EDC/NHS chemistry spreadsheet (http://kinase.com/human/kinome/tables/ according to the manufacturer’s instructions. Either MTAP- GST (BPS Biosciences) or GST was captured to generate the KincatHsap.08.02.xls). We note that some of our identified MTAP or reference surfaces. Sensor data were analyzed using targets were not on this list because they were regulatory Scrubber 2 software (BioLogic Software Pty Ltd.) or BiaEvalu- subunits of a kinase (for e.g., PRKAR1A, PRKAR2A and ation (GE LifeSciences). Data were double-reference subtracted PHKB). and corrected for variation in solvent concentration. Binding affinity was calculated using kinetic and equilibrium analyses. Western Blot Analysis Experiments. Western blot analysis experi- Kinetic analysis was performed using a least-squares fit of a ments to validate affinity enrichment were performed in the Langmuir 1:1 binding model. same manner as with SILAC experiments, except that cells grown with normal RPMI-1640 media (Invitrogen) was used. Sensor Chip Preparation. The sensor surface was conditioned using The cells were grown, lysed and-treated identically to the SILAC alternating injections of 10 mM glycine pH 2.2 and 50 mm TargetID experiments as described above. After separation of NaOH. The surface was then activated with 1:1 4 M EDC/1 M SM-bead enriched protein on SDS/PAGE, proteins were trans- NHS for 10 min. Anti-GST antibody was diluted to 30 g/mL in ferred to a nitrocellulose membrane using a semidry blotting 10 mM acetate pH 5.0 and was exposed to the activated surface apparatus (Bio-Rad). The membrane was blocked with 5% skim for 10 min at 5 L/min. The surface was quenched bya7min milk powder in PBS-Tween, washed in 0.1% PBS-T, and then injection of 1M ethanolamine. Between 13,000 and 14,500 incubated with the corresponding antibodies, as indicated in response units (RU) were immobilized for each assay. MTAP- Figs. 5 and 6 and Fig. S5, followed by HRP-conjugated secondary GST and GST were diluted to 75 g/mL and 7.5 g/mL, antibodies. The membranes were subjected to chemiluminescent respectively and captured on the anti-GST antibody surface with detection according to manufacturer’s instructions (ECL, GE a 10 min injection at 5 L/min. Between 1,500 and 1,700 RU of Healthcare). protein were captured for each assay. The running buffer used during immobilization and capture was PBS, pH 7.4 with 0.005% MTAP Functional Assay. We used a published method (5) for P-20 surfactant. measuring MTAP activity with slight modifications. Briefly, MTAP’s production of adenine from MTA is monitored after its SPR Binding Assay Parameters. Small-molecule binding assays were conversion to 8-dihydroxyadenine by xanthine oxidase. We performed at 25 °C. The running buffer for the binding assays measured change in absorbance at 305 nm in a 96-well plate was PBS, pH 7.4 with 0.005% P-20 surfactant and 2% DMSO as using a SpectraMax Plus (Molecular Devices, Sunnyvale, CA). a cosolvent. Compounds were diluted from 10 mM stocks in For soluble competition, either a DMSO control, 10ϫ (225 DMSO to the appropriate concentration in buffer with the same nmoles), 50ϫ (1125 nmoles), or 100ϫ (2250 nmoles) of each of solvent concentration as the running buffer (2%). Binding was the IPL ligands was added to Ϸ2 mg or 1 mL of cell lysate and measured at concentrations from 4.8 nM to 20 M in half incubated for 1 hour on an end-over-end rotator at 4 °C. Assays dilutions. Compound was injected for 120 sec followed by 120 sec were done at 37 °C in a total volume of 300 L containing 100 of buffer with no compound. L of cell lysate (containing small molecule or control), 24 L or 0.8 units of xanthine oxidase (Sigma), 176 Lof40mM Preparation of Bait Compounds and General Procedure of Solid-Phase KH2PO4 and10 M MTA. Kinetic Vmax plots were acquired over Immobilization. All chemicals were purchased from Sigma– a 25 min acquisition period and plotted using SpectraMax Aldrich and common solvents were purchased from Fisher software. (J.T.Baker) at HPLC grade unless otherwise noted. Automated
Ong et al. www.pnas.org/cgi/content/short/0900191106 2of15 flash chromatography was performed on Teledyne ISCO Com- The starting acid (137 mg, 0.240 mmol, 1 eq.) and CDI (46.7 biFlash Rf systems. HRMS data were collected on a Bruker mg, 0.288 mmol, 1.2 eq.) were dissolved in anhydrous DMF (2 Daltonics APEXIV 4.7 FT-ICR mass spectrometer. 1H and mL) under nitrogen. The mixture was stirred at room temper- 13C-NMR spectra were collected on a Bruker 300 MHz NMR ature for 15 min to activate the acid. 3-Amino-1-propanol (92.0 spectrometer. L, 90.0 mg, 1.20 mmol, 5.0 eq.) was added to the solution and the reaction was maintained at room temperature for 1 h. The (S)-((R)-3-(3,4-dimethoxyphenyl)-1-(3-(2-(3-hydroxypropylamino)-2- reaction mixture was then concentrated and purified by auto- oxoethoxy)phenyl)propyl) 1-(3,3-dimethyl-2-oxopentanoyl)piperi- mated flash chromatography to afford the pure product (114.6 1 dine-2-carboxylate, AP-1497 alcohol (Fig. S1B). Starting material, mg, 76%). H NMR (300 MHz, CDCl3) ␦ 7.34 (d, 1H, J ϭ 5.8), 2-(3-((R)-3-(3,4-dimethoxyphenyl)-1-((S)-1-(3,3-dimethyl-2- 7.26 (m, 1H), 6.92 (m, 2H), 6.78 (m, 2H), 6.67 (m, 2H), 5.70 (m, oxopentanoyl)piperidine-2-carbonyloxy)propyl)phenoxy)acetic 1H), 4.61 (m, 1H), 4.50 (s, 2H), 3.81 (dd, 6H, J ϭ 4.6, 18.8), 3.62 acid, was prepared with reported synthetic procedure (6). (m, 3H), 3.49 (dt, 5H, J ϭ 5.6, 18.2), 2.92 (d, 1H, J ϭ 1.4), 2.85 The starting acid (103 mg, 0.176 mmol, 1 eq.) and CDI (34.2 (d, 1H, J ϭ 1.2), 2.54 (m, 2H), 2.21 (m, 3H), 2.00 (m, 4H), 1.68 mg , 0.211 mmol, 1.2 eq.) were dissolved in anhydrous DMF (2 (qq, 4H, J ϭ 6.8, 13.3), 1.20 (t, 4H, J ϭ 7.5), 1.13 (s, 1H), 0.93 13 mL) under nitrogen. The mixture was stirred at room temper- (s, 1H), 0.82 (dd, 2H, J ϭ 6.8, 7.6); C NMR (300 MHz, CDCl3) ature for 15 min to activate the acid. 3-Amino-1-propanol (67.3 207.00, 171.42, 168.86, 163.4, 162.52, 148.95, 147.41, 142.02, L, 66.1 mg, 0.880 mmol, 5.0 eq.) was added to the solution and 133.54, 129.85, 120.19, 119.79, 113.48, 113.29, 111.97, 111.5, the reaction was maintained at room temperature for 1 h. The 76.02, 67.21, 59.97, 58.51, 55.95, 47.21, 46.53, 38.04, 36.41, 32.61, reaction mixture was then concentrated and purified by auto- 31.64, 31.11, 29.01, 24.85, 8.74; MS (calcd. 626.3203, ESϩ): mated flash chromatography to afford the pure product (93.2 627.3299 (MϩH)ϩ. 1 mg, 83%). H NMR (300 MHz, CDCl3) ␦ 7.30 (m, 2H), 6.94 (t, 2H, J ϭ 9.7), 6.80 (m, 2H), 6.68 (dd, 2H, J ϭ 3.4, 7.4), 5.76 (dd, 2-(3-((R)-3-(3,4-dimethoxyphenyl)-1-((S)-1-(2-phenylacetyl)pyrroli- 1H, J ϭ 5.5, 7.9), 5.30 (d, 1H, J ϭ 4.9), 4.45 (d, 2H, J ϭ 28.5), dine-2-carbonyloxy)propyl)phenoxy)acetic acid, ProAP-1780, (*) (Fig. 3.81 (dd, 6H, J ϭ 3.5, 19.5), 3.65 (t, 2H, J ϭ 5.3), 3.50 (dd, 2H, S1E). Starting material, (R)-tert-butyl 2-(3-(3-(3,4-dimethoxy- J ϭ 6.1, 12.1), 3.35 (s, 2H), 3.17 (dd, 1H, J ϭ 7.9, 17.9), 2.94 (s, phenyl)-1-hydroxypropyl)phenoxy)acetate, was prepared by the 1H), 2.86 (s, 1H), 2.57 (m, 2H), 2.36 (d, 1H, J ϭ 13.5), 2.23 (dt, reported procedure1. Starting material (S)-1-(2-phenylac- 1H, J ϭ 8.8, 14.3), 2.07 (m, 2H), 1.69 (m, 7H), 1.42 (m, 2H), 1.21 etyl)pyrrolidine-2-carboxylic acid was prepared with commercial (t, 5H, J ϭ 3.6), 1.14 (s, 1H), 0.86 (m, 3H); 13C NMR (300 MHz, compound proline reacting with phenylacetyl chloride as re- CDCl3) 207.79, 169.64, 168.80, 166.61, 162.51, 157.47, 149.00, ported (7). 147.48, 141.96, 133.38, 129.95, 120.20, 119.99, 113.71, 113.47, To alcohol (241 mg, 0.599 mmol, 1.0 eq.), the acid (154 mg, 111.9, 111.52, 76.4, 67.19, 60.33, 56.73, 51.28, 46.68, 44.13, 38.96, 0.659 mmol, 1.1 eq) EDC (137.4 mg, 0.719 mmol and 1.2 eq) and 36.62, 32.56, 31.74, 27.51, 26.41, 24.92, 23.61, 21.09, 8.77; MS DMAP (8.05 mg, 0.066 mmol, 0.11 eq) were dissolved in ϩ (calcd. 640.3360, ESϩ): 641.3446 (MϩH) . anhydrous CH2Cl2 (10.0 mL) at room temperature under nitro- gen. The reaction was stirred for 2.5 h. The crude reaction (S)-((R)-3-(3,4-dimethoxyphenyl)-1-(3-(2-(3-hydroxypropylamino)-2- mixture was extracted with H2O, dried over MgSO4, and con- oxoethoxy)phenyl)propyl) 1-(2-phenylacetyl)piperidine-2-carboxy- centrated before being purified by automated chromatography late, AP-1780 alcohol (Fig. S1C). Starting material, 2-(3-((R)-3-(3,4- to afford the tert-butyl-protected intermediate as colorless oil. dimethoxyphenyl)-1-((S)-1-(2-phenylacetyl)piperidine-2- The intermediate compound is directly dissolved in CH2Cl2 (9 carbonyloxy)propyl)phenoxy)acetic acid, was prepared with the mL) at 0 °C before TFA (3.0 mL) was added. The reaction was reported synthetic procedure (6). stirred for 30 min at 0 °C and1hatroom temperature. The crude The starting acid (202 mg, 0.352 mmol, 1 eq.) and CDI (68.4 product was obtained by concentrating the reaction mixture mg, 0.422 mmol, 1.2 eq.) were dissolved in anhydrous DMF (4 under vacuum, which was further purified by automated flash mL) under nitrogen. The mixture was stirred at room temper- chromatography to afford the pure product (254.7 mg, 76% 2 1 ature for 15 min to activate the acid. 3-Amino-1-propanol (134.0 steps). H NMR (300 MHz, CDCl3) ␦ 10.63 (s, 1H), 7.12 (d, 6H, L , 132 mg, 1.758 mmol, 5.0 eq.) was added to the solution and J ϭ 9.7), 6.72 (dt, 7H, J ϭ 23.8, 32.4), 5.63 (s, 1H), 4.55 (s, 1H), the reaction was maintained at room temperature for 1 h. The 4.39 (s, 2H), 3.74 (s, 6H), 3.62 (s, 2H), 3.45 (d, 3H, J ϭ 33.0), 2.43 13 reaction mixture was then concentrated and purified by auto- (s, 2H), 1.96 (d, 7H, J ϭ 56.1); C NMR (300 MHz, CDCl3) mated flash chromatography to afford the pure product (161.1 171.25, 170.83, 157.96, 158.12, 148.94, 147.37, 141.88, 133.98, 1 mg, 72%). H NMR (300 MHz, CDCl3) ␦ 7.47 (s, 1H), 7.24 (m, 133.71, 129.59, 128.96, 128.68, 126.93, 120.27, 119.53, 114.51, 7H), 6.85 (t, 2H, J ϭ 6.4), 6.77 (d, 2H, J ϭ 5.6), 6.65 (t, 3H, J ϭ 112.16, 112.02, 111.53, 77.6, 65.19, 59.35, 55.98, 47.56, 41.69, 10.2), 5.70 (dd, 1H, J ϭ 5.1, 8.2), 5.45 (s, 1H), 4.44 (d, 2H, J ϭ 38.07, 31.19, 29.36, 24.74; MS (calcd. 561.2363, ESϩ): 562.2495 4.0), 3.83 (d, 7H, J ϭ 1.4), 3.74 (d, 3H, J ϭ 12.1), 3.66 (s, 1H), (MϩH)ϩ. 3.56 (dd, 2H, J ϭ 4.7, 10.3), 3.43 (m, 3H), 3.17 (d, 1H, J ϭ 10.4), 2.91 (s, 1H), 2.84 (s, 1H), 2.55 (dd, 3H, J ϭ 5.8, 14.5), 2.29 (d, 1H, (S)-((R)-3-(3,4-dimethoxyphenyl)-1-(3-(2-(3-hydroxypropylamino)-2- J ϭ 13.2), 2.16 (d, 1H, J ϭ 5.7), 2.02 (s, 2H), 1.66 (d, 6H, J ϭ 7.4), oxoethoxy)phenyl)propyl) 1-(2-phenylacetyl)pyrrolidine-2-carboxy- 13 1.25 (dd, 3H, J ϭ 8.0, 15.1); C NMR (300 MHz, CDCl3) 171.18, late, ProAP-1780 alcohol (Fig. S1F). The starting carboxylic acid (83 169.76, 168.62, 162.53, 149.00, 147.47, 142.19, 134.72, 133.46, mg, 0.148 mmol, 1 eq.) and CDI (28.8 mg, 0.177 mmol, 1.2 eq.) 129.84, 128.67, 126.77, 120.22, 119.76, 113.36, 113.22, 111.92, were dissolved in anhydrous DMF (2.0 mL) under nitrogen. The 111.57, 75.97, 67.17, 60.12, 55.96, 52.16, 44.01, 40.95, 38.09, 36.89, mixture was stirred at room temperature for 15 min to activate 31.44, 31.29, 26.68, 25.06, 20.76; MS (calcd. 632.3098, ESϩ): the acid. 3-Amino-1-propanol (56.5 L, 0.739 mmol, 5.0 eq.) was 633.3164 (MϩH)ϩ. added to the solution and the reaction was maintained at room temperature for 1 h. The reaction mixture was then concentrated (S)-((R)-3-(3,4-dimethoxyphenyl)-1-(3-(2-(3-hydroxypropylamino)-2- and purified by automated flash chromatography to afford the 1 oxoethoxy)phenyl)propyl) 1-(3,3-dimethyl-2-oxopentanoyl)pyrroli- pure product (85.3 mg, 93%). H NMR (300 MHz, CDCl3) ␦ 7.99 dine-2-carboxylate, ProAP-1497 alcohol (Fig. S1D). Starting material, (s, 1H), 7.59 (s, 1H), 7.25 (m, 6H), 6.96 (s, 1H), 6.87 (d, 1H, J ϭ 2-(3-((R)-3-(3,4-dimethoxyphenyl)-1-((S)-1-(3,3-dimethyl-2- 7.8), 6.77 (t, 2H, J ϭ 6.6), 6.68 (d, 2H, J ϭ 5.3), 5.72 (m, 1H), 4.62 oxopentanoyl)pyrrolidine-2-carbonyloxy)propyl)phenoxy)acetic (m, 1H), 4.46 (s, 2H), 4.25 (m, 1H), 3.84 (s, 6H), 3.76 (m, 1H), acid, was prepared with the reported synthetic procedure (6). 3.69 (s, 2H), 3.64 (s, 1H), 3.45 (m, 6H), 3.29 (s, 1H), 2.93 (s, 2H),
Ong et al. www.pnas.org/cgi/content/short/0900191106 3of15 2.86 (s, 2H), 2.55 (m, 2H), 2.19 (m, 3H), 1.96 (m, 5H), 1.61 (s, 99.31, 84.44, 84.40, 64.88, 46.90, 44.61, 41.86, 23.64; MS (calcd. 13 ϩ 2H); C NMR (300 MHz, CDCl3) 171.61, 170.00, 168.75, 162.52, 496.1747, ESϩ): 497.1820 (MϩH) . 148.98, 147.43, 142.22, 134.24, 133.63, 129.65, 128.95, 128.61, 126.85, 120.21, 119.51, 113.44, 112.88, 111.99, 111.53, 77.56, 3-(1-(3-hydroxypropyl)-1H-indol-3-yl)-4-(1-methyl-1H-indol-3-yl)-1H- 67.15, 60.14, 59.13, 55.97, 47.43, 41.85, 39.89, 36.97, 31.25, 29.38, pyrrole-2,5-dione, RO31–7549 alcohol derivative (Fig. S1K). The prep- 24.83, 21.2; MS (calcd. 618.2941, ESϩ): 619.3025 (MϩH)ϩ. aration of RO31–7549’s alcohol derivative follows a procedure reported in ref. 8. 2-(4-(4-(4-fluorophenyl)-5-(pyridin-4-yl)-1H-imidazol-2-yl)phenoxy)-N- (3-hydroxypropyl)acetamide, SB202190 alcohol. Starting material General Procedure for Immobilization of Bait Molecules. SB202190 (4-(4-(4-fluorophenyl)-5-(pyridin-4-yl)-1H-imidazol- Fig. S1L. Bio-Rad Affi Gel 102, an agarose-based gel with a 2-yl)phenol) was purchased from Sigma (cat# S7067). The 6-atom hydrophilic amino-terminated arm. starting compound was subjected to 3 steps of chemical reactions Solid-phase (beads) preparation. The solid-phase beads used in small without purification at intermediate steps (Step1 and 2) molecule immobilization and affinity chromatography is Affigel 102 (Bio-Rad) with a loading level of 12 mol/mL suspension Fig. S1G. Step 1: SB202190 (25.0 mg, 0.075 mmol, 1 eq.) was (Fig. S1). The bead suspension (1.0 mL) was transferred to a 2.0 dissolved in anhydrous THF (0.5 mL) and K2CO3 (209 mg, 1.509 mL Eppendorf tube and washed with H2O(3ϫ 1.5 mL) and mmol, 20 eq.) was added at room temperature. The solution DMF (3 ϫ 1.5 mL). The beads were then suspended in anhy- turned deep red upon K2CO3 addition before tert-butyl 2-bro- drous DMF (0.5 mL). moacetate (13.4 L, 0.091 mmol, 1.2 eq.) was added. The reaction was stirred at room temperature for 48 h and the crude Fig. S1M. Example of bait molecule activation (RO-31-7549). reaction mixture was filtered and concentrated. The crude 2) Bait molecule activation (Fig. S2): The bait molecule (10 intermediate compound tert-butyl 2-(4-(4-(4-fluorophenyl)-5- mol) was dissolved in 200 L of anhydrous acetonitrile (or (pyridin-4-yl)-1H-imidazol-2-yl)phenoxy)acetate obtained was DMF). DSC (7.69 mg, 30 mol, 3 eq.) was dissolved in 400 L used directly for the next step without further purification. of anhydrous acetonitrile (or DMF) and was added to the bait molecule solution before TEA (5.55 L, 4.05 mg, 40 mol, 4 eq.) Fig. S1H. Step 2: The crude tert-butyl 2-(4-(4-(4-fluorophenyl)- was added. The reaction solution was stirred at 50 °C for1hand 5-(pyridin-4-yl)-1H-imidazol-2-yl)phenoxy)acetate intermedi- the activation efficiency was monitored by LC-MS. ate was dissolved in CH2Cl2 (1.0 mL) and TFA (0.3 mL) was added. The reaction was stirred at room temperature for 1 h. The Fig. S1N. Example of immobilization of activated bait molecules reaction mixture was concentrated and the crude product 2-(4- (RO-31-7549). (4-(4-fluorophenyl)-5-(pyridin-4-yl)-1H-imidazol-2-yl)phenoxy) Bait molecule immobilization (Fig. S3). Different bait compounds give acetic acid was used directly in the final step without further different activation efficiencies, therefore the volume of activa- purification. tion solution added to the beads was calculated individually. For example, when the LC-MS indicated 85% of bait molecules was Fig. S1I. Step 3: The crude 2-(4-(4-(4-fluorophenyl)-5-(pyridin- activated, for 6.25% loading level, 53.4 L of activation solution 4-yl)-1H-imidazol-2-yl)phenoxy)acetic acid intermediate was (contains 0.75 mol of activated bait molecule) was added to the dissolved in anhydrous DMF (2.0 mL) and CDI (25.0 mg, 0.154 beads suspension; for 12.5% and 25.0% loading levels, 106.8 L mmol, 2.05 eq.) was added to activate the acid at room temper- and 213.6 L were added to the beads respectively. After adding ature. After 15 min, 3-amino-1-propanol (48.2 mg, 0.642 mmol, the activation solution, the suspension was vortexed at room 8.5 eq.) was added to the solution and the reaction was stirred temperature for 1 h and the depletion of free activated bait at room temperature for 2 h. The crude reaction mixture was molecule was monitored by LC-MS. concentrated and purified by automated flash chromatography Work-up. After the immobilization, the vials were centrifuged, the to afford the pure product (23.8 mg, 71%, 3 steps). 1H NMR (300 supernatant was removed and the beads were washed with DMF MHz, CD3OD) ␦ 8.45 (d, 2H, J ϭ 5.7), 8.05 (s, 2H), 7.99 (m, 2H), (3 ϫ 2 mL)and H2O(3ϫ 2 mL). The beads were subsequently 7.55 (m, 4H), 7.22 (t, 2H, J ϭ 8.8), 7.14 (d, 2H, J ϭ 8.9), 4.92 (s, suspended in 1ϫ PBS (0.8 mL) and stored at 4 °C before use. 13H), 4.60 (s, 2H), 3.63 (m, 8H), 3.41 (t, 2H, J ϭ 6.8), 3.30 (m, 9H), 2.89 (s, 1H), 1.76 (m, 9H), 1.30 (s, 1H); 13C NMR (300 MHz, SI Materials CDCl3) 170.87, 167.38, 163.93, 162.75, 160.13, 149.83, 149.16, SPR Binding Assays. 132.21, 132.10, 129.18, 128.75, 124.27, 117.09, 116.8, 116.33, FKBP1A and MTAP to Immunophilin ligands. 68.34; MS (calcd. 446.1754, ESϩ): 447.1845 (MϩH)ϩ. The binding of Immunophilin (IPL) ligands to FKBP1A was measured by surface plasmon resonance. The observed response .(؉)-K252a alcohol derivative, (Fig. S1J). The starting material (ϩ)- rates were too fast to be analyzed in kinetic terms (ka, kd) K252a was purchased from BioMol International L.P. (catalog#: Affinity constants (KD) were derived from equilibrium binding EI152). (ϩ)-K252a (25.0 mg, 0.043 mmol, 1.0 eq.) and lipase measurements. The values range from 26 nM to 44 M (Fig. 5A). acrylic resin (66 mg) from Candida antarctica (Sigma cat# L4777) were mixed in ethanolamine (4.0 mL, 1545 eq.) at 40 °C Sensor Preparation. The FKBP12 surface plasmon resonance under nitrogen. The reaction was stirred overnight, filtered, assay was performed on a Biacore S51 instrument using Biacore washed with MeOH, concentrated and taken up with saturated CM5 sensor chips. GST capture kit, ethanolamine, N-ethyl-NЈ- NH4Cl, and then extracted with EtOAc. The organic layer was (3-dimethylaminopropyl)carbodiimide (EDC), N-hydroxysuc- dried over Na2SO4, concentrated and purified by automated cinimide (NHS) and P-20 surfactant were all obtained from flash chromatography to afford the pure product (19.5 mg, Biacore Inc. Immunophilin igands were synthesized following 1 1 92%). H NMR (300 MHz, THF-d8) H NMR (300 MHz, THF) the general procedure outlined by Holt. The surface of the ␦ 8.46 (d, 1H, J ϭ 7.9), 7.11 (m, 2H), 6.78 (d, 1H, J ϭ 8.2), 6.53 sensor chip was conditioned using alternating 1 min injections (t, 2H, J ϭ 7.7), 6.42 (t, 1H, J ϭ 7.4), 6.33 (t, 1H, J ϭ 7.5), 6.17 (90 L/min flow rate) of 10 mM glycine pH 2.2 and 50 mM (dd, 1H, J ϭ 4.8, 7.3), 4.64 (s, 4H), 4.08 (q, 2H, J ϭ 17.2), 2.81 NaOH (repeated 3 times). The surface carboxyl groups were (d, 2H, J ϭ 5.0), 2.55 (m, 2H), 2.38 (s, 1H), 1.35 (s, 3H); 13C NMR activated with 1:1 0.4M EDC/0.1M NHS. A 25 g/mL solution (300 MHz, CDCl3) 171.19, 170.90, 139.57, 136.43, 131.50, 128.09, of anti-GST inl acetate buffer pH 5.5 was flowed for 10 min at 125.59, 124.32, 123.47, 122.44, 119.79, 115.86, 113.85, 106.8, a rate of 5 L/min over spots 1 and 2. The remaining NHS-ester
Ong et al. www.pnas.org/cgi/content/short/0900191106 4of15 groups on the sensor surface were quenched witha7min bead specific; Cluster 2: 269 SM-bead specific). We analyzed injection of 1 M ethanolamine. Recombinant GST (10 g/mL un-enriched whole cell lysates and compared protein abundances solution in PBS-P pH 7.4) was captured on spot 1 followed by in these samples to those identified in our pull-down experi- FKBP12-GST fusion protein (50 g/mL solution in PBS-P pH ments. Proteins in clusters 1 and 2 were significantly different 7.4) on Spot 2. Table 1 contains information on sensor prepa- from whole cell lysates (Fig. 3B, Kolmogorov–Smirnov test, P Ϫ ration for each day’s assay. value: Ͻ 2.2 ϫ 10 16), whereas the cluster of target proteins had a K-S P value of 0.0026, also indicating that clusters 1 and 2 were Data Analysis. All data were analyzed using T100 Evaluation and mostly dominated by abundant proteins. Indeed, over half of the Scrubber software. Data were double reference subtracted. The cluster 3 target proteins were not identified at all in the analysis steady state affinity constant (K ) for each ligand was derived of whole cell lysate, likely because they were not among the top D Ϸ from a plot of Req against concentration. The plot was then fitted 2000 proteins in abundance. We performed the same analysis using the whole set of proteins identified in the SC experiment to a general steady state model. Table 2 contains KD values derived from equilibrium binding measurements. (Fig. S3C) and find the majority of proteins identified in SC experiments are abundant proteins that were also identified in Accompanying Text for Fig. S2: Interpretation of Bead Control Data. unenriched lysate. In analysis of SILAC data, a common practice is to normalize all quantified protein ratios to account for a systematic error made Accompanying Text for Fig. S4. We compared the list of targets when combining protein lysates. This normalization is often identified in SC and BC experiments and found that both done by determining dividing each protein’s ratio by the median experimental formats were effective at enriching known targets. ratio for the group of proteins that do not exhibit a significant We observed some proteins with high small molecule (SM)- change in SILAC ratios. However, in the case of BC data (Fig. specific SILAC ratios in BC experiments but ratios close to 1 in SC experiments, and hence, were nonspecific (for e.g., VDAC2, S2A), this is not appropriate as ratio distributions shift consid- GSK3B, NQO2 at 10ϫ SC, Fig. S4A). Using an orthogonal erably depending on the amount of compound loaded. The approach with recombinant GST-fusion proteins and Western distributions of protein ratios itself is dramatically different in blot analysis, we found that the GST-western experiments were 25% loaded beads (Fig. S2C). We analyzed lysates mixed one- in complete agreement with SILAC ratio data (Fig. S4B). to-one without any affinity enrichment to determine the mixing Ϯ Proteins identified in the BC experiment were indeed enriched ratio empirically and found this to be very close to 1 (1.0 0.15) by small molecule affinity matrices, but soluble competition was in our experiments. This determination, in addition to gel most effective with the primary target PRKCD and not other visualization, indicates that the effect of compound bead loading proteins, and depended on the levels of soluble competitor used. in each BC experiment on SILAC ratio distributions was in fact In applying a range of soluble competitor amounts in other SC an accurate representation of protein binding specificities. How- experiments, we made the general observation that our ability to ever, this makes data interpretation much less direct. We found identify targets improved with increasing levels of soluble com- k-means clustering of data from multiple bead loadings to be petition (Fig. 4). useful for this purpose (Fig. S3) Accompanying Text for Dataset S1. Dataset S1 contains multiple Accompanying Text for Fig. S3. We compared datasets generated worksheets that summarize soluble competitor (SC) experiments from 3 loading levels (6, 12 and 25%) of the kinase inhibitor for kinase inhibitors and the IPL series. The k252a and Ro-31- Ro-31-7549 with k-means clustering (Euclidean distance, num- 7549 tables are summarized across multiple experiments to ber of clusters: 3 through 7, only 3-cluster data shown), moni- combine FDR values for each protein into a single summary toring membership of proteins within the modeled distributions statistic. This was calculated as: 10͚[(w.f.) ϫ log(local FDR)], with the across the different load levels (Fig. S3A). Known targets like total of weighting factors (w.f.) equaling 1. W.f. for different PRKCD, PRKCA, GSK3B, ADK, CAMK2D, and NQO2 experiments were chosen heuristically and we provide 2 alternate among others, dominated a cluster of 42 proteins, characterized sets of w.f. as examples. We chose w.f. that would raise the by consistently high SILAC ratios irrespective of compound contribution of higher SC levels or remove the influence of loading levels, whereas 2 clusters comprising the majority of the experiments that may have been affected by compound precip- proteins had ratios indicating increased binding to either EtOH- itation (for e.g., Ro-31-7549 at 100ϫ SC). The choice of w.f. does beads or small molecule-loaded beads (Cluster 1: 132 EtOH- not dramatically change the list of targets.
1. Ong SE, Mann M (2006) A practical recipe for stable isotope labeling by amino acids in 6. Keenan T, et al. (1998) Synthesis and activity of bivalent FKBP12 ligands for the cell culture (SILAC). Nat Protoc 1(6):2650–2660. regulated dimerization of proteins. Bioorg Med Chem 6(8):1309–1335. 2. Rappsilber J, Mann M, Ishihama Y (2007) Protocol for micro-purification, enrichment, 7. Boto A, De Leon Y, Gallardo JA, Hernandez R (2005) Synthesis of alkaloid analogues pre-fractionation and storage of peptides for proteomics using StageTips. Nat Protoc from alpha-amino acids by one-pot radical decarboxylation/alkylation. (Translated 2(8):1896–1906. from English) Eur J Org Chem (16):3461–3468 (in English). 3. Efron B, Tibshirani R (2002) Empirical bayes methods and false discovery rates for 8. Faul MM, Winneroski LL, Krumrich CA (1998) A new, efficient method for the synthesis microarrays. Genet Epidemiol 23(1):70–86. of bisindolylmaleimides. (Translated from English) J Org Chem 63(17):6053–6058 (in 4. Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S (2002) The protein kinase English). complement of the human genome. Science 298(5600):1912–1934. 5. Christopher SA, Diegelman P, Porter CW, Kruger WD (2002) Methylthioadenosine phosphorylase, a gene frequently codeleted with p16(cdkN2a/ARF), acts as a tumor suppressor in a breast cancer cell line. Cancer Res 62(22):6639–6644.
Ong et al. www.pnas.org/cgi/content/short/0900191106 5of15 O O A O N N O O O O H 2 N O O O N R O O R O N R OH H activation O immobilization Affigel-102
H H N MeO O N O O H N N OH O MeO O N O O O O HN O N OH N N N N H Me O N R Me F O n=0,1 HO NH OH HO Immunophilin Ligands * SB-202190 RO-31-7549 K252a
B C MeO MeO MeO MeO OH HN OH MeO O MeO O OH HN OH MeO O CDI, DMF MeO O CDI, DMF O O O O O O O O N N N 3-amino-1-propanol N 3-amino-1-propanol O O O O O O O O O O
D E MeO MeO MeO OH OH MeO O MeO N OH NH OH DMAP, CH Cl O O MeO O MeO O O 2 2 CDI, DMF MeO O OtBu O N O O O O O EDC O N 3-amino-1-propanol N HO O O O O O O O
MeO MeO F H G N OH N OH N MeO O MeO O H O H N O CDI, DMF O O tert-butyl 2-bromoacetate O N O O N N OH N O 3-amino-1-propanol O N O O K 2CO3 in THF F F
H N I OH N N N H OH HN H O N H OH H N TFA in CH2Cl2 N 3-amino-1-propanol N O O O O O O O O N N N CDI, DMF N
F F F F
H H N J N O O K L O N O lipase acrylic resin OH HN H N O 2 N N N N O N N O ethanolamine, 40 ºC O H H N MeO2C OH HO O OH
M N O O O O N N O O H 2 N O O N O N N O O O O N H O N OH O O O O HN HN O O O HN HN O O N O N O N O N HN O O N N O N activation O H O
Fig. S1. Generation of affinity matrices for small-molecule pull-down experiments. (A) A general scheme of the coupling reaction of a small molecule containing a functionalizable handle (ϪOH in this case) to a solid support (Affigel-102) is shown. The chemical structures of immunophilin ligands and kinase inhibitors used are shown. (B–N) Chemical reaction schemes to match SI Methods.
Ong et al. www.pnas.org/cgi/content/short/0900191106 6of15 AB C
EtOH AP AP1497 6% EtOH AP AP1497 12% AP1497 25% bead 1497 bead 1497 EtOH AP MW 6% 6% 12% 12% bead 1497 MW 25% 25%
188 188 98 98 62 62 49 49 38 38 28 28 17 14 17 14 6 6
D Ro-31-7549 bead loading 6,12,25% Immunophilin Ligand Series Low Stringency Low Stringency
Ro 31-7549 6% Ro 31-7549 12.5% Ro 31-7549 25% AP1497 Pro-AP1497 AP1780 Pro-AP1780
High Stringency High Stringency
Ro 31-7549 6% Ro 31-7549 12.5% Ro 31-7549 25% AP1497 Pro-AP1497 AP1780 Pro-AP1780
Fig. S2. (A–C) Gel visualization, SILAC ratio distribution plots for Immunophilin ligand AP1480 at 3 compound load levels (6%, 12%, 25%). For each load level, unmixed SILAC pull-downs were resolved on a 4–12% gradient Bis-Tris gel with each bead type and lysate loaded in a separate gel lane. A parallel set of pull-downs was mixed and loaded on the gel for quantitative MS analysis. Log2 SILAC ratios from that set is plotted as a histogram (Upper) or plotted with SILAC ratios sorted in descending order (Lower). (D) Modeled distributions of SILAC ratios for bead control experiments for different molecules. The different ratios distributions are accurate representations of the binding characteristics of the affinity matrices (see also Fig. S2 A–C). The variability depends on various parameters such as the small molecule used, loading density of small molecules on bead, and stringency of wash buffer used. High stringency washing reduced background interactions to a degree. Target proteins were largely unaffected by high stringency washing, comparing LS and HS experiments is therefore a useful method for identifying SM-specific targets. However, very abundant proteins that bound to SM-beads were still identified with high SILAC ratios, thereby still requiring cross-experiment comparisons to avoid this bias (see Fig. S3). Because log2 ratio values for BC experiments in general displayed visually separable populations, we modeled these data using mixtures of t-distributions and selected the optimal number of components using the Bayesian information criterion (BIC). We note that the data were not well modeled by Gaussian distributions. Indeed, BIC analysis indicates that the data are best described by at least2 t-distributions. The plots display the number of components corresponding to the lowest BIC score.
Ong et al. www.pnas.org/cgi/content/short/0900191106 7of15 mro.Seicba oadaudn rtiswsosre ihwa fnt tH rS-edbnes(lses1ad2 - value p k-s 2, and 1 (Clusters binders Kolm SM-bead by or lysate EtOH- cell affinity whole weak unfractionated with from observed abundance protein was ranked proteins to abundant reference toward in bias clusters) Specific (k-means Smirnov. specificities binding bead 3 of enrichment bnatpoen rmlsts oee,teS xeietldsg losietfiaino pcfi agt eas bnatpoen aeratios have proteins abundant because targets specific of identification allows design experimental SC the one-to-one. However, lysates. from proteins abundant rtisi lse hwmc ekrsgicne( au .06 n h ecnaeo rtisietfidi fnt uldwsbtudtce nw in undetected but pull-downs affinity in identified ( proteins 3. of cluster percentage the in and highest 0.0026) also value (p is significance lysate weaker cell much show 3 cluster in proteins -en.Konadpttv agt eefudecuieyi lse ,caatrzdb ihSLCrto cosal3la ees ( levels. load 3 all across ratios SILAC high by characterized 3, cluster in exclusively found were targets putative and Known k-means. n tal. et Ong S3. Fig. -en lseigadpl-onercmn fpoen.( proteins. of enrichment pull-down and clustering k-means www.pnas.org/cgi/content/short/0900191106 BC A Protein XICs Ranked by Ro-31-7549 BCk-meansclustersmappedtoHeLaS3wholecelllysates C abundant Least abundant Most h apn fpoen nwoecl yae oteRO25 the to lysates cell whole in proteins of mapping The ) log2 SILACratio -5 0 5 132 EtOHbeadbinders Cluster 1: k-means clusteringofRo-31-7549BCdataacrossthreeloadinglevels 22 22 225 12 6 25 12 6 25 12 6 in wholelysate not identified 29 of132(22%) Ro Loading Level .0e1 .e1 3.083e-10 <2.2e-16 1.503e-11 lse lse Cluster3 2 Cluster Cluster 1 RankEtOH 2087 1 in wholelysate not identified 88 of269(33%) RankSM 2087 1 log2 SILACratio A
Cpoenrto rm3laiglvl fR-174 eecutrdit lseswith clusters 3 into clustered were Ro-31-7549 of levels loading 3 from ratios protein BC ) -5 0 5 in wholelysate not identified 22 of42(52%) 269 SMbeadbinders Cluster 2: RankTarget Ro Loading Level 2087 1
log2 SILACratio Protein XICs Ranked by -5 0 5 42 Ro-31-7549targets Cluster 3: toHeLaS3wholecelllysates Ro-31-7549 SCproteinsmapped abundant Least abundant Most Ro Loading Level 10 ϫ Cdtstas hw htteS aae contains dataset SC the that shows also dataset SC Ro25_10x_SC RankRo25 2087 1 whole lysate 218 notidentifiedin 618 identified, < 2.2e-16 B vlaigaffinity Evaluating ) Ͻ 2 ϫ 10 Ϫ 16 .Target ). ogorov– ls to close 8of15 hole A Soluble Comp. SILAC ratios Bead Ctrl 0.1x 1x
PRKCD >50 4.5 15 GSK3B 10 1.1 1.4 NQO2 45 0.97 0.96 VDAC2 7.2 0.94 0.92
B Direct input Pull-downs
RO Sol. comp - - 0.1x 0.5x 1x
PRKCD
GSK3B
NQO2
VDAC2
GST
Fig. S4. Validation of SILAC pull-down data with recombinant proteins. (A) Table of SILAC ratios for proteins identified as specific protein interactors to the kinase inhibitor Ro-31-7549. (B) Validation of SC data. Equimolar amounts of purified GST-fusion proteins were incubated overnight with Ro-31-7549 with or without soluble competition. This experiment shows that although many proteins may be identified as specific binders in BC experiments (Table in Fig. 4), some of these (GSK3B, NQO2, VDAC2) show diminished selectivity toward soluble forms of the small molecule. The negative controls FKBP12 and GST-alone showno binding to the solid phase matrix.
Ong et al. www.pnas.org/cgi/content/short/0900191106 9of15 k252a k252a k252a k252a SB202190 SB202190 Ro31-7549 Ro31-7549 Gene symbol A B pc12 BC h1299 BC HeLa BC HeLa SC HeLa BC HeLa SC HeLa BC HeLa SC AAK1 1 ADK AURKA BMP2K 0.9 CAMK2A CAM2KB CAMK2D 0.8 CAMK2G CDC2 CDC2L5 0.7 CDC42BPB CDK2 CDK4 CDK5 0.6 CDK6 CDK7 CDK9 0.5 CLK1 CLK4 CHEK1
Precision 0.4 CIT CRKRS CSK 0.3 CSNK2A1 CSNK2A2 CSNK1D 0.2 DAPK2 DEK EIF2AK2 0.1 SC EIF2AK4 FER BC FRAP 0 GAK 0 5 10 15 20 25 30 35 40 GSK3A GSK3B HUNK Number of Kinases IRAK4 LIMK1 MAP2K1 MAP2K2 MAP2K3 MAP2K4 MAP2K5 MAP2K6 MAP3K9 MAP3K11 MAP3K15 MAP4K4 MAPK1 MAPK3 MAPK8 MAPK9 MAPK10 MAPK14 MAPKAPK2 MARK1 MARK2 MARK3 MAST1 MINK1 MST4/RP6-213H19.1 MYLK PAK1 PAK4 PCTK1 PCTK2 PDPK1 PDXK PHKG2 PKN1 PKN2 PRKAA1 PRKACA PRKACB PRKAG1 PRKAR1 PRKAR2A PRKAR2B PRKCA PRKCD PRKCI PRKD1 PRKD2 PRKD3 PRKDC PRKRA PTK2 RIOK2 RIPK2 RIPK4 ROCK2 RPS6KA1 RPS6KA3 SKP2 SLK SRPK1 STK10 STK17A STK3 STK4 TBK1 TGFBR1 TGFBR3 TP53RK TTK ULK1 ULK3 VRK2 YES1 Total, Specific 40 42 47 56 9 8 44 21
>2 peps, but not significant >2 peps, significant not detected
Fig. S5. (A) Accuracy of kinase identification for K252a 100ϫ SC and BC experiments. Proteins are sorted by their average log2 ratio (high to low) across replicate experiments. Protein kinases (4) are used as a true positive set and at each rank the number of inferred kinases is plotted against the precision, which corresponds to the percentage of kinases among all inferred targets at a given threshold. (B) Kinases identified from kinase inhibitor experiments. Kinases identified in BC and SC experiments with k252a, Ro 31–7549, and SB202190 are summarized. Red indicates kinase was identified with Ͼ2 peptides, and with a significant SILAC ratio. Gray indicates protein was identified with Ͼ2 peptides but lack of specificity for the small molecule.
Ong et al. www.pnas.org/cgi/content/short/0900191106 10 of 15 A
MTAP Ratio in Bead Control Low/High Stringencies
80 Low Stringency 70 High Stringency
60
50
40
30
SILAC H/L ratio H/L SILAC 20
10
0
AP1497 AP1780 Pro-AP1497 Pro-AP1780
B
MTAP ratio and intensity plot in SC
Billions 20 10x H/L ratio 25 18 100x H/L Ratio 16 Total intensity 10x 20 Total intensity 100x 14 12 15 10 8 10 6 SILAC H /L ratio 4 5 Total Protein XIC 2 0 0
AP1497 AP1780 ProAP-1497 ProAP-1780
Fig. S6. MTAP is a protein binder to the immunophilin ligand series. (A) MTAP ratios determined from bead control experiments with the IPL series. MTAP is identified as a specific binder to the IPL ligands in both low and high stringency bead control experiments. (B) MTAP ratios in SC experiments overlaid with plots of total extracted ion intensities (XIC) for the protein. Although the SILAC ratio is largest in the SC experiment with AP1780, the total protein XIC indicates that the Pro-AP1780 affinity matrix enriched the most MTAP across the IPL series, suggesting that the levels of soluble competitor used in this experiment (max: 100ϫ) was insufficient to compete with protein bound to the solid phase.
Ong et al. www.pnas.org/cgi/content/short/0900191106 11 of 15 RU RU RU RU FKBP12 : AP1497 Sensogram FKBP12 : AP1780 Sensogram FKBP12 : Pro-AP1497 Sensogram 25 FKBP12 : Pro-AP1780 Sensogram 25 25 25
20 20 20 20 10.0 μ M 5.0 μ M 1000 nM 2.50 μ M 15 15 333 nM 15 60.0 μ M 15 μ M 120 μM 111 nM 30.0μ M 1.25 15.0 μM μ 10 60.0 μ M 10 37.0 nM 10 10 0.625 M 7.50 μ M 30.0 μ M 0.313 μM Response baseline) (0 = μ baseline) (0 = Response μ 12.3 nM Response baseline) (0 = 3.75 μM 5 15.0 M 5 baseline)Response (0 = 5 5 0.156 μM 7.50 μM 4.12 nM 1.88 μM 3.75μ μ M 1.88 μμM 0 0 0 0
-5 -5 -5 -100 10 20 30 40 50 60 70 -20 0 20 40 60 80 100 120 s -5 -10 0 10 20 30 40 50 60 70 -100 10 20 30 40 50 60 70 s s Time (0 = Sample start) Time (0 = Sample start) Time (0 = Sample start) Time (0 = Sample start)
Steady State Affinity AP1497 Steady State Affinity Pro-AP1497 Steady State Affinity Pro-AP1780 RU Steady State Affinity AP1780 μ μ RU KD = 43.8 M 18KD = 25.7 nM RU K μ M RU KD = 0.800 M 14 16 D = 5.17 20 12 14 14 16 10 12 10 8 10 12 6
8 Response Response 6 Response 8 4 Response 6 2 2 4 4 0 2 0 2e-5 4e-5 6e-5 8e-5 1e-4 1.2e-4 0 2e-7 4e-7 6e-7 8e-7 1e-6 M 0 1e-5 2e-5 3e-5 4e-5 5e-5 6e-5 7e-5 0 2e-6 4e-6 6e-6 8e-6 1e-5 M M Concentration M Concentration Concentration Concentration
MTAP-GST : Pro-AP1780 Sensogram MTAP-GST : AP1780 Sensogram MTAP-GST : AP1497 Sensogram MTAP-GST : Pro-AP1497 Sensogram
Steady-state affinity AP1780 Steady-state affinity AP1497 Steady-state affinity Pro-AP1497 Steady-state affinity Pro-AP1780
Fig. S7. SPR binding curves for FKBP1A and MTAP with immunophilin ligands.
Ong et al. www.pnas.org/cgi/content/short/0900191106 12 of 15 Table S1. Biacore Steady State Affinity Data FKBP12
Date Sensor Chip FKBP-ligand Steady-state KD
5/21/07 1162163–2 AP1497 33.9 nM AP1497* 25.7 nM AP1780 14.7 M AP1780* 5.17 M Pro-AP1497 566 nM 6/01/07 1162163–15 AP1780* 6.70 M Pro-AP1497 800 nM Pro-AP1780 43.8 M
*Ligands from different batch
Ong et al. www.pnas.org/cgi/content/short/0900191106 13 of 15 Table S2. SPR data of MTAP-GST with Immunophilin ligand series
Ϫ1 Ϫ1 Ϫ1 IPL ligand KD equil KD kinetic ka,M ⅐s kd,s
Pro-AP1780 18.0 Ϯ 2nM 12.4 Ϯ 1nM 4.29 ϫ 106 0.053 AP1780 1.55 Ϯ 2nM 2.43 Ϯ 4nM 1.86 ϫ 107 0.045 AP1497 2.82 Ϯ 3M 1.9 Ϯ 1M 50217.0 0.09315 ProAP1497 8.54 Ϯ 7M 20.0 Ϯ 9M 3395.9 0.07208
Ong et al. www.pnas.org/cgi/content/short/0900191106 14 of 15 Table S3. List of tables in Dataset S1 Name of table SC level # experiments # replicates Comments
K252a Multiple Seven SC levels 2 per SC level Summarized Ro-31–7549 Multiple 2 bead loadings, five SC levels 2 per experiment Summarized SB202190* Multiple T2 SC levels 10ϫ , 100ϫ 3 pooled per SC AP149710ϫ (HoltHolt) 10ϫ 12 AP1497100ϫ (HoltHolt) 100ϫ 12 ProAP149710ϫ (HoltPro) 10ϫ 12 ProAP1497100ϫ (HoltPro) 100ϫ 12 AP178010ϫ (HoltBz) 10ϫ 12 AP1780100ϫ (HoltBz) 100ϫ 12 ProAP178010ϫ (HoltDmg) 10ϫ 12 ProAP1780100ϫ (HoltDmg) 100ϫ 12
Other Supporting Information Files
Dataset S1 (XLS)
Ong et al. www.pnas.org/cgi/content/short/0900191106 15 of 15