Fragment Screening by Weak Affinity Chromatography - American

Jun 10, 2013 - Michael Mrosek,. ‡. James Murray,. ‡. Roderick E. Hubbard,. ‡,§ and Sten Ohlson*. ,†,⊥. †. Department of Chemistry and Bio...
4 downloads 0 Views 4MB Size
Article pubs.acs.org/ac

Fragment Screening by Weak Affinity Chromatography: Comparison with Established Techniques for Screening against HSP90 Elinor Meiby,† Heather Simmonite,‡ Loic le Strat,‡ Ben Davis,‡ Natalia Matassova,‡ Jonathan D. Moore,‡ Michael Mrosek,‡ James Murray,‡ Roderick E. Hubbard,‡,§ and Sten Ohlson*,†,⊥ †

Department of Chemistry and Biomedical Sciences, Linnaeus University, SE-391 82 Kalmar, Sweden Vernalis, Granta Park, Cambridge CB21 6GB, United Kingdom § YSBL, University of York, Heslington, York, YO10 5DD, United Kingdom ⊥ School of Biological Sciences, Nanyang Technological University, Singapore 637551 ‡

ABSTRACT: The increasing use of fragment-based lead discovery (FBLD) in industry as well as in academia creates a high demand for sensitive and reliable methods to detect the binding of fragments to act as starting points in drug discovery programs. Nuclear magnetic resonance (NMR), surface plasmon resonance (SPR), and X-ray crystallography are well-established methods for fragment finding, and thermal shift and fluorescence polarization (FP) assays are used to a lesser extent. Weak affinity chromatography (WAC) was recently introduced as a new technology for fragment screening. The study presented here compares screening of 111 fragments against the ATPase domain of HSP90 by all of these methods, with isothermal titration calorimetry (ITC) used to confirm the most potent hits. The study demonstrates that WAC is comparable to the established methods of ligandbased NMR and SPR as a hit-id method, with hit correlations of 88% and 83%, respectively. The stability of HSP90 WAC columns was also evaluated and found to give 90% reproducibility even after 207 days of storage. A good correlation was obtained between the various technologies, validating WAC as an effective technology for fragment screening.

O

method enables detection of very weakly binding fragments in a high throughput manner by analysis in fragment mixtures. The aim with this study is to describe a series of experiments which aim to validate WAC as a method for fragment screening, comparing the results with screening by other established methods. The target used for this validation study was the N-terminal domain of the molecular chaperone, HSP90. There have been a number of successful drug discovery campaigns against this protein (reviewed by Biamonte et al.17). A number of these inhibitors were discovered by fragmentbased methods,12,18−23 and this protein has become established as a suitable model system for the development and assessment of fragment discovery methods. In this study, approximately 111 fragments were screened against the target HSP90 by WAC, 1D nuclear magnetic resonance (NMR), surface plasmon resonance (SPR), fluorescence polarization assay (FP), and thermal (Tm) shift assay (TSA). We attempted to obtain crystal structures for 32

ver the past 15 years, there has been increased interest in development and use of fragment-based methods to provide the starting points for discovering small molecule entities that bind to (mostly) protein targets.1−4 The key features of this technology are that a small (typically 500−2000 member) well characterized library5,6 of low molecular weight (typically 110−250 Da) compounds is screened for molecules that bind to a specific site on the target.7 The small size of the fragments increases the likelihood of finding hits;8 the challenge for medicinal chemistry is to progress the low affinity fragments into potent lead molecules.2,7,9−12 One of the distinctive features of fragment-based discovery is the need for suitable screening methods to reliably detect the low affinity hits, typically binding with a KD in the mM to high μM range. These low affinity hits still represent good starting points for hits-to-leads chemistry due to the often high ligand efficiency (LE); for example, a 12 heavy atom fragment binding with a KD of 1 mM has a reasonable LE of 0.34.13 This initial low affinity has led to the development and refinement of a wide variety of biophysical methods to detect such binding.7 Weak affinity chromatography (WAC) was recently introduced as a new technology for fragment screening and has been demonstrated on protease14,15 and kinase targets.16 This © 2013 American Chemical Society

Received: March 8, 2013 Accepted: June 10, 2013 Published: June 10, 2013 6756

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

Figure 1. Compounds referenced in this work.

PMG H10SSGHID4K to aid immobilization for SPR studies.25 Untagged protein was purified solely on the monoQ column. All protein samples were concentrated to approximately 20 mg/mL using ultrafiltration into a final buffer containing 20 mM TRIS pH 7.4 and 0.5 M sodium chloride. A Bradford assay was used to determine the protein concentrations. Fragment Screening by WAC. Preparation of Affinity Capillary Columns. HSP90 was immobilized in situ on three stainless-steel capillary columns (100 × 0.5 mm) packed with spherical silica particles (5 μm in diameter, 300 Å pore size; Kromasil, EKA Chemicals, Bohus, Sweden) which had been silanized into diol-substituted silica according to standard procedures.26 Immobilization was performed on an Agilent 1200 HPLC system (Agilent Technologies, Waldbronn, Germany) by reductive amination of mainly lysine side chains of the proteins and aldehyde groups of the silica surface as described earlier.27 The diol-silica columns were rinsed with isopropanol (>20 column volumes) and water (>20 column volumes). The diol silica columns were oxidized into aldehydesilica by 10 × 40 μL injections of 0.13 g/mL periodic acid with a flow rate of 20 μL/min for 2 min. The flow was stopped for 12 min between injections. The total time for oxidation was 2 h, and the reaction was performed at 12 °C. The columns were rinsed with water and with 0.1 M sodium phosphate buffer pH 7.0 (>20 column volumes). Coupling of tagged HSP90 was performed by 5 × 40 μL (column 1) or 6 × 40 μL (column 2 and column 3) injections of 3.4 mg/mL (column 1), 3.0 mg/ mL (column 2), or 3.1 mg/mL (column 3) HSP90 and 9 mg/ mL sodium cyanoborohydride dissolved in 0.1 M sodium phosphate buffer pH 7.0 and 3 × 40 μL injections of 9 mg/mL

compounds from the test set (31 hits, 1 nonhit). In addition, 27 selected hits were assessed by isothermal titration calorimetry (ITC). Screening results from the different techniques were compared. Furthermore, series of experiments were performed to assess the reproducibility and stability of WAC as a fragment screening methodology.



MATERIALS AND METHODS Chemicals. 111 fragments from Vernalis’ fragment library5,6 were used for screening assays. The fragments were dissolved in DMSO-d6 at a concentration of 200 mM and were further diluted in water to working concentrations. 3-(Trimethylsilyl)propionic-2,2,3,3-d4 acid (TSP-d4), periodic acid, polyethylene glycol (PEG), Bovine serum albumin (BSA), adenosine diphosphate (ADP), sodium cyanoborohydride, and ethanolamine were purchased from Sigma Aldrich (St Louis, USA). Sypro Orange dye was purchased from Invitrogen (Paisley, UK). All other chemicals used in preparation of buffers were of analytical grade. Production of HSP90. HSP90 protein was produced as described previously.24 The N-terminal fragment of HSP90α (residues 9−236) was overexpressed in the E. coli strain BL21 (P Lys S). Three different proteins were used in the work reported here. Tagged protein included a deca-his tag within a 15 amino acid N-terminal extension of MGHHHHHHHHHHSSGH. The tag facilitated purification, which was carried out using a Ni2+ affinity column followed by a monoQ ionexchange column. Multitagged protein was purified in the same way and included a triple His tag with the sequence MH 8 GATGSTAGSGTAGSTGASGASTGGTGATH 8 D4 KS6757

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

sodium cyanoborohydride dissolved in 0.1 M sodium phosphate buffer pH 7.0. The columns were prepared at different times, and conditions such as the number of injections and protein concentrations were slightly varied between the various coupling reactions to obtain the preferred amount of immobilized protein. Injections were performed with a flow rate of 20 μL/min for 2 min, and the flow was stopped for 2 h between injections. The total reaction time was 16 h (column 1) or 18 h (column 2 and column 3), and the column temperature was 4 °C (column 1) or 12 °C (column 2 and column 3). The columns were rinsed with 0.1 M sodium phosphate buffer pH 7.0 (>20 column volumes). The yield of the coupling reaction was approximated indirectly from the absorbance reading at 280 nm of the collected column eluates during immobilization and applied HSP90 sample. Remaining aldehyde-silica groups present on the columns were encapsulated by ethanolamine. This was achieved by 6 × 40 μL injections of ethanolamine (6.15 mg/mL) and sodium cyanoborohydride (9 mg/mL) dissolved in 0.1 M sodium phosphate buffer pH 7.0 at a flow rate of 20 μL/min for 2 min. The flow was stopped for 1 h between injections. The total time for the reaction was 6 h, and it was performed at 12 °C. The columns were finally rinsed with 0.1 sodium phosphate buffer pH 7.0. An ethanolamine reference column was produced as described in earlier work.16 Column Characterization by Frontal Affinity Chromatography (FAC). FAC was performed on column 1 and column 2 essentially as described by Kasai et al.28 An Agilent 1200 HPLC system equipped with a diode array multiple wavelength detector (DAD) was used for the frontal analysis. The mobile phase was PBS for column 1 (10 mM sodium phosphate, 150 mM sodium chloride; pH 7.4) and 20 mM ammonium acetate buffer pH 6.8 for column 2. The flow rate was 18 μL/min, and the column temperature was 22 °C. Fragment 7 (Figure 1) was used for FAC analysis as it had an appropriate retention when measured by zonal affinity analysis. Fragment 7 was dissolved in the mobile phase and injected onto the affinity columns at 100 μL injections with concentrations of 0.025, 0.050, 0.100, 0.200, 0.300, 0.500, and 0.750 mM (column 1) and 0.025, 0.050, 0.100, 0.200, 0.300, and 0.500 mM (column 2) in duplicates. Break-through curves of fragment 7 were monitored by UV at 254 nm (reference wavelength 360 nm). The void time of the column was determined with DMSO (1% in water) by an injection volume of 0.4 μL. DMSO was monitored by UV at 214 nm (reference wavelength of 360 nm). Chromatograms were analyzed by the Agilent ChemStation version B.04.01 chromatography data system, and the front times were assessed from the local maximum of the first derivative of the front curve. The volume for column saturation V′R was calculated by adjusting for the void volume of the column. The number of moles of fragment 7 required for saturation of all active sites present on the column was calculated from V′R and the concentration of fragment 7 for each injection. The VR′ s were plotted against the concentrations of fragment 7 in order to generate a one-site total binding hyperbola by GraphPad Prism 5.0c (San Diego, CA, USA). Here, the contribution from nonspecific binding is defined as a linear relation between analyte concentration and amount bound. By nonlinear regression analysis, the number of binding sites on the column (Btot) and the dissociation constant (KD) of fragment 7 was determined (eq 1).

V R′ ·[ligand] =

Btot ·[ligand] KD + [ligand]

(1)

Zonal Affinity Chromatography Screening of the Fragment Library. Fragment screening was performed on an Agilent 1200 series capillary HPLC system equipped with a diode array multiple wavelength detector (DAD) and a single quadropole mass spectrometry (MS) detector. UV detection was performed at a wavelength of 214 and 254 nm (reference wavelength 360 nm). For mass spectrometry detection, fragments were ionized by electrospray at atmospheric pressure (API-ES) in positive and negative mode alternately. Drying nitrogen gas flow was 7 L/min at 350 °C. The nebulizer pressure was 10 psig. The capillary voltage was 3000 V in positive mode and −2500 V in negative mode. MS signal acquisition was set at selected ion monitoring (SIM) on sample target masses. Both the [M + 1]+ and the [M − 1]− ion was monitored for each analyte. The fragmentor was set to 100 V in positive SIM mode and to 173 V in negative SIM mode. Retention times were based on peak apexes of the extracted ion chromatogram (EIC). Chromatograms were analyzed with the Agilent ChemStation version B.04.01 chromatography data system. Screening was performed with an injection volume of 0.4 μL and a flow rate of 20 μL/min. The column temperature was 22 °C, and each sample was analyzed in duplicates isocratically. During analysis in mixtures, adenosine (0.1 mM) was present as a control in each mixture and the retention time of adenosine was monitored continuously and was used as a measure of the activity of the column. On column 1, fragments 1−12 were screened using both PBS pH 7.4 and 20 mM ammonium acetate pH 6.8 as mobile phases. During screening using PBS as mobile phase, the fragments were analyzed by single injections and detection was performed by UV detection. Each sample was analyzed at a concentration of 0.1 mM (0.05% DMSO) and 0.01 mM (0.005% DMSO). During screening using ammonium acetate buffer as mobile phase, the fragments were analyzed as single injections and in sets of 4. Detection was performed by MS. The sample concentration of each fragment was 0.1 mM, and the DMSO concentration was 0.05% (singles) and 0.2% (mixtures). On column 2, screening was performed of fragments 1−111 in mixtures. Fragments 1−12 were screened in sets of 4 (sample concentration 0.1 mM, 0.2% DMSO) and fragments 13−111 in sets of 16 (0.1 mM concentration of each fragment, 0.8% DMSO) or 17 (0.1 mM concentration of each fragment, 0.85% DMSO). Ammonium acetate buffer (20 mM) pH 6.8 with or without 1 μM of the HSP90 inhibitor compound 1 (Figure 1) was used as mobile phase, and the column temperature was 22 °C. A mixture of fragments 5−8 was also screened on the column with a column temperature of 32 °C and 20 mM ammonium acetate buffer pH 6.8, as mobile phase. Detection was performed by UV detection and MS detection. Column 3 was used for reproducibility and stability studies. A mixture of fragments 97, 98, 105, and 111 and adenosine was analyzed on the column 1, 26, 27, 28, 53, 82, 134, and 207 days after immobilization of HSP90. Analysis of the mixture was performed by 10 injections on each day (6 injections on day 207). Analysis was performed during three consecutive days (26−28 days after immobilization) to study the day-to-day variation in retention times, both using buffer prepared of the same batch as mobile phase and by preparing a fresh batch of 6758

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

2% DMSO as a vehicle for the compounds. Compounds were added to 384 well plates and diluted into 5 μL of 20% DMSO/ 80% assay mix, and then, 45 μL of assay mix was added. After a 30 min equilibration at ambient temperature in the dark, fluorescence polarization was measured using a Synergy plate reader (Biotek, Vermont, USA; Excitation 485 nm, Emission 535 nm). For this assay, Z′ values, expressing the quality of the assay33 are typically >0.8. Data was fitted to a 4 parameter logistical model (XLFit version 4, IDBS, Guildford, UK). Characterizing Fragments by ITC. The 27 fragment hits of highest rank identified by WAC were selected for screening by ITC. The measurements were performed using an ITC200 instrument (Microcal, GE Healthcare, New Jersey, USA). The feedback mode was “low” with reference power setting of 4 μCals−1. The cell was stirred at 1000 rpm and thermostatted at 25 °C. All experiments were performed using the dialysis buffer (see below) with 1% DMSO (v/v) and 0.05% Tween 20. All experiments were conducted with 40 μM protein in the cell and 1.5 mM ligand in the syringe. The experiments were conducted with 8 injections, with volumes of 4.75 μL and 190 s spacing. The first “waste” injection of 1.2 μL was discarded in all cases. All data were fitted to a one site model using the provided software. Tagged protein (6 mg/mL) was dialyzed with stirring overnight at 4 °C, in 10 mM HEPES pH 7.4, 150 mM sodium chloride, and 0.5 mM EDTA. Upon recovery from the dialysis cassette, the protein was filtered through a 0.22 μM spin filter. The protein concentration was determined by UV absorbance spectroscopy at 280 nM using an extinction coefficient of 15900 M−1cm−1. The dialysis buffer was degassed; this was then used for subsequent preparation of protein and ligand solutions for the titration experiments. For each experiment, the ligands were freshly prepared from a 200 mM DMSO-d6 stock solution. Fragment Screening by SPR. 109 fragments were screened by SPR as 2 fragments were momentarily out of stock in competition with ADP. SPR measurements were performed on a BIAcore T200 instrument (BIAcore GE Healthcare, New Jersey, USA). All experiments were performed at 20 °C on Series S NTA or CM5 chips according to the supplier’s protocol in HBS-P buffer (10 mM HEPES pH 7.4, 150 mM sodium chloride, 0.05% P-20) supplemented with 0.025 mM EDTA, 10 mM magnesium chloride, and 1% or 5% DMSO. The NTA chip surface was generated using multitagged protein, where introduction of the multiple His tags leads to robust generation of a stable HSP90 surface with no detectable drift of a protein from a chip for at least 30 min (as previously described25,34). First, 0.5 mM Ni2+ was injected into the experimental channel to give a signal of 60 RU. Subsequently, 100 nM multitagged protein was injected over the sensor at 10 μL/min until a saturation level of 1500 RU was recorded of protein stably bound on the surface. A reference surface without immobilized Ni2+ was included on the chip to serve as controls for nonspecific binding and refractive index changes. The sensor surface was regenerated between experiments by 0.35 M EDTA (as advised by manufacturer) with additional injections of 0.1 mg/mL trypsin, 0.5 M imidazole, and 45% DMSO (all for 60 s at a flow rate of 15 μL/min) to eliminate any carry-over of protein and/or analyte. For each concentration in the titration series, the surface was prepared as described above. In some experiments, the protein was further stabilized by covalent amine coupling as advised in the manufacturer’s protocols, both for protein bound to the CM5 chip in 10 mM sodium acetate pH 5.5 or captured via the histidine tag to Ni2+ on an NTA chip. All sample measurements

buffer for mobile phase each day. The column was stored in PBS pH 7.4 at 4 °C between analyses. Data Analysis. The dissociation constant (KD) of each fragment for binding to HSP90 was approximated26,27 according to: KD =

Btot tR′ ,specific·F

(2)

where F is the flow rate during screening and t′R, specific is the adjusted retention time on column 1 using 20 mM ammonium acetate buffer pH 6.8 (t′R, active), corrected for the adjusted retention time on column 1 using ammonium acetate buffer pH 6.8/1 μM compound 1 as mobile phase (t′R, inhibited): tR′ ,specific = tR′ ,active − tR′ ,inhibited

(3)

The adjusted retention times were calculated by subtraction of the retention time of the void marker DMSO from the retention times of the analytes. Fragment Screening by 1D NMR. The 111 fragments were screened by NMR essentially as described in earlier work.21 The fragments were screened in the same mixtures of 16−17 compounds per sample as used for screening by WAC. All measurements were made on a Bruker DRX600 (Bruker BioSpin GmbH, Rheinstetten, Germany) equipped with a cryoprobe and a BACS-120 sample changer. All samples contained 10 μM tagged protein and 500 μM of each fragment, in 50 mM potassium phosphate buffer, pH 7.5, 100 μM TSPd4, 10% D2O. 1H 1D saturation transfer difference spectroscopy (STD),29 water-ligand observed via gradient spectroscopy (waterLOGSY),30 and relaxation filtered spectra31 were acquired on all samples under ICONNMR automation using excitation sculpting to suppress the solvent peak. Competition was determined by addition of compound 2 (Figure 1)21 to a concentration of 50 μM followed by the acquisition of a repeat set of the NMR experiments mentioned above for each sample, in order to confirm specific binding at or near to the active site. Spectra were analyzed using AMIX (Bruker Biospin). Spectra acquired before and after addition of competitor were superimposed, and resonances showing changes were assigned using reference spectra acquired for each compound in the same buffer. In cases where there was potential ambiguity over assignment, individual compounds were screened to identify the binding ligand unambiguously. Compounds were grouped according to whether binding and competition was observed in all three (STD, waterLOGSY, and relaxation filtered spectra) NMR binding experiments (“Class 1” hits), two binding experiments (“Class 2” hits), or a single experiment (“Class 3” hits), reflecting varying degrees of confidence in the experimental data. In addition, the 1D NMR spectrum of the ligand mixture was compared to the composite spectrum formed by superimposing the individual compound reference spectra in order to confirm the presence and chemical integrity of each of the fragment compounds in the mixture. Fluorescence Polarization Assay. The assay was carried out for 106 fragments as 5 fragments were momentarily out of stock essentially as previously described by Howes et al.32 using the fluorescently labeled compound 3 (Figure 1) as a probe. This binds to HSP90 with a KD of 10 nM, and displacement by other compounds can be measured as a change in fluorescence polarization. The assay mixture contained 100 mM Tris-HCl pH 7.4, 20 mM potassium chloride, 6 mM magnesium chloride, 5 μg/mL BSA, 10 nM tagged protein, 10 nM compound 3, and 6759

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

were performed at a flow rate of 35 μL/min. Injection times ranged from 20 to 120 s; dissociations times ranged from 20 to 120 s. Screening of fragments was conducted in triplicate at 250−500 μM in the presence of 100 μM ADP to identify ATPbinding site competitive binders. Under these conditions, the KD for ADP for HSP90 was determined to be 6 μM. The hits (42 fragments) were then followed up in dose response titrations of nine 2-fold diluted experimental points with the top concentration of 500 μM. Data processing was performed using BIAevaluation 1.1 (BIAcore GE Healthcare BioSciencesCorp) or Scrubber2 (BioLogic) software. Sensorgrams were double referenced prior to global fitting of the concentration series to Steady State Affinity or Single Step Kinetic models. Fragment Screening by Thermal Shift Assay. The assay was carried out for 109 fragments as 2 fragments were currently out of stock. Protein and compound solutions were prepared separately and successively pipetted into 96-well plates from 2fold stock solutions to give a final volume of 40 μL per well. Final concentration of tagged protein was 5 μM in a buffer containing 50 mM HEPES pH 7.5, 100 mM sodium chloride, and 5-fold Sypro Orange protein dye (prepared from a 5000fold stock). All compounds were tested at 2 mM final concentration of each compound and 4% final DMSO concentration. The plate was centrifuged at 4000g prior to the assay to ensure that samples were at the bottom of each well. The 96-well plate was heated at 1 °C/min from 25 to 97 °C in a Stratagene Mx3005P thermocycler (Agilent, Santa Clara, USA). A fluorescence reading was taken after a 5 min equilibration time for each temperature point (excitation was at 492 nm and emission readings were taken at 610 nm). Tm values were calculated using nonlinear regression in XLFIT4 (IDBS). Three 4% DMSO control wells and Compound 4 (a potent inhibitor of HSP90 N-terminal domain, used at 0.4 mM; Figure 1)35 were always included on each plate for calibration and cross-plate validation. Each reported value is the average of 2 independent measurements. X-ray Crystallography. Crystallization. Crystallization was performed as described previously.24 Initial conditions for crystallization of both tagged and untagged protein were found using commercial screens. Screening was carried out using the 24 well hanging drop vapor diffusion technique. Optimum conditions for both were found to be in solutions #3 (0.2 M magnesium chloride, 25% polyethylene glycol monomethyl ether approximate molecular weight 2000 (2K MME)) and #6 (0.8 M Na formate, 2K MME) of Clear Strategy Screen number 1 (obtained from Molecular Dimensions, Newmarket, UK) at pH 6.5. The reservoir well contained 500 μL of the precipitant solution, and the hanging drops were formed by mixing 2 μL of the 20 mg/mL protein solution with 2 μL of the reservoir solution. The plates were incubated at 4 °C. Apo crystals appeared within a few hours and were of a suitable size for data collection, in some cases, after overnight growth. The crystals were harvested and soaked overnight in a solution containing 4 μL of the precipitant solution from the reservoir well previously mixed with 0.5 μL of ligand (200 mM stock solution in DMSO) to give a 1.2 mM final ligand concentration. The crystals were then removed into a cryoprotectant solution of this same ligand/reservoir solution containing 25% glycerol for a few seconds and flash frozen in liquid nitrogen. Structure Determination. Protein structure determination was performed as described previously.24 Data were collected at 100 K on an in-house RU-H3R rotating anode generator with

R-Axis IV++ image plate detector and were subsequently processed using D*Trek (Rigaku, The Woodlands, USA). The crystals belong to the space groups I222 or P21212. The structures were solved by molecular replacement using a previously solved HSP90α protein model of the tagged protein (PDB code: 1UY6; PU3 ligand and solvent removed) and the program AMoRe.36 Twenty cycles of rigid-body then restrained refinement were carried out using the refinement program REFMAC537 followed by model building and solvent addition using the molecular graphics program COOT.38 The progress of the refinement was assessed using Rfree and the conventional R factor. Once refinement was completed, the structures were validated using various programs from the CCP4i package.39



RESULTS AND DISCUSSION Weak Affinity Chromatography (WAC). Production and Characterization of HSP90 Columns. Three different HSP90 columns were used in this study: column 1 for initial screening of 12 fragments in various mobile phases and sample concentrations; column 2 for screening of 111 fragments in ammonium acetate; column 3 for reproducibility and stability studies. For each of these columns, HSP90 was immobilized onto diol-silanized silica capillary columns at high coupling amounts and in high yield of immobilization reaction (Table 1). Table 1. Production and Characterization of HSP90 Columnsa column 1 immobilized protein concentration of immobilized protein immobilization yield

column 2

column 3

Column Production 0.64 mg 0.58 mg (20.7 nmol) 0.66 mg (22.9 (23.6 nmol) nmol) 5.1 mg/mL 4.6 mg/mL 5.3 mg/mL

92% 81% Frontal Affinity Chromatography mobile phase during PBS pH 7.4 20 mM ammonium FAC analysis acetate pH 6.8 Btot 14.3 nmol 9.9 nmol ratio of active protein 63% 48% on column KD fragment 7 83 μM 72 μM R2 fit to curve 0.9998 0.9988 Zonal Affinity Chromatography KDb fragment 7 87 μM 77 μM

91% − − − −



A “−” denotes that no data is available. bAn ethanolamine column was used as a reference column for determination of KD for fragment 7 by zonal affinity chromatography for comparison with FAC. a

FAC was used to determine the number of active sites present on column 1 (mobile phase: PBS pH 7.4) and column 2 (mobile phase: 20 mM ammonium acetate buffer pH 6.8). KD values determined by FAC correlated well with KD values obtained from zonal affinity chromatography (Table 1). Overview of Fragment Screening. Figure 2 demonstrates typical chromatograms from analysis of a mixture of fragments by WAC on a HSP90 column, as presented by extracted ion chromatograms (EICs). Fragments with a difference in retention time of 0.5 min or more between noncompetitive and competitive screening had affinities with KDs lower than 2 mM to HSP90 and were regarded as hits. 6760

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

Figure 2. Analysis of a mixture of fragments 46−62 on a HSP90 column by WAC.

Figure 3. Analysis of a mixture of 4 fragments and adenosine on column 2 at a column temperature of 22 and 32 °C, respectively. EICs of individual analytes in SIM positive mode (fragment 5, red; fragment 6, gray; fragment 7, black; fragment 8, blue; adenosine, green). The void is seen as a contaminating peak at m/z = 169 (the same m/z as for fragment 7).

In addition, fragments 21 and 65 interacted with the ethanolamine derivatized silica matrix of the HSP90 affinity

column, resulting in considerable peak broadening during analysis. As a result, no accurate retention times could be 6761

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

Figure 4. Screening of 111 fragments by WAC, NMR, SPR, FP and Tm shift. Selected fragments were also assessed by ITC. Fragments are sorted in decreasing affinity order according to WAC. Hits are denoted as green, nonhits as red, and missing cells indicate that no data is available. For X-ray crystallography, the green shows the fragments for which structures were attempted and determined; see text for more details. Structures previously deposited: fragment 5, 2YEC; 6, 2YEJ; 47, 2YEB; 50, 2YE4; 72, 2YE8; 81, 2YE6; 84, 2YEG; 89, 2YEH.

obtained for fragments 21 or 65 by WAC. The retention time of the internal standard adenosine remained constant throughout the screening campaign, indicating no loss in activity of immobilized HSP90. Thermal Elution. Increasing the column temperature from 22 to 32 °C caused an average reduction of 16−41% in adjusted retention times for a mixture of five compounds (fragments 5− 8 and adenosine). Figure 3 shows EICs for analysis of these analytes at 22 and 32 °C on column 2. Increasing the

temperature resulted in this case in lower affinities of ligands due to binding being enthalpy driven. This suggests that, in such cases, thermal elution can be used to elute high affinity binders (well below μM in KD) as long as the activity of the target protein can be maintained. Thermal elution increases the affinity range of analytes that can be screened on the WAC column. However, there is still a risk that analytes of very high affinities to the target are not being eluted from the column within a reasonable time. 6762

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

Table 2. Correlation between Various Fragment Screening Methods WAC NMR SPR ITC X-ray FP

WAC

NMR

SPR

ITC



88% (n = 103) −

83% (n = 107) 90% (n = 102) −

52% (n = 27) 63% (n = 24) 73% (n = 26) −

X-ray 83% 93% 76% 57%

(n = (n = (n = (n = −

30) 28) 29) 23)

FP

Tm shift

66% (n = 109) 74% (n = 104) 73% (n = 109) 63% (n = 27) 73% (n = 30) −

71% (n = 107) 74% (n = 102) 73% (n = 109) 58% (n = 26) 66% (n = 29) 73% (n = 109)

demonstrated a measurable retention by WAC, representing an 81% eventual success rate in crystallography. This compares well with NMR where, for the six failures in crystallography, one was identified as a nonbinder, two were classed as ambiguous, two as “Class 1” hits, and one as a “Class 2” hit, representing a minimal success rate of 84%. Of these 32 fragments, 21 were identified as hits by SPR (note that 2 were not determined), three failing to give crystal structures, representing an 87% success rate. However, 6 of the fragments which gave crystal structures were not detected as binders by SPR. TSA demonstrated a 72% success rate in predicting crystallographic success; however, it failed to identify 7 fragments that gave a crystal structure. The FP assay performs rather well if a threshold of 15% inhibition is used as the hit boundary (9.9 mM KI) with an 81% success rate. Interestingly, two compounds identified as ambiguous in the NMR screen were shown to be good binders by all the other methods, including ITC, but failed to generate a crystal structure despite numerous attempts. It is possible that some of these hits bind cooperatively to other sites on the protein which are not accessible in the crystal system; alternately, there could be significant conformational change of the structure upon binding. The one fragment that is a WAC hit, but a nonbinder by NMR, does not appear to bind by either TSA or SPR. It is clear that all the screening methods give similar success rates in establishing which compounds should be tried in crystallography. TSA performs marginally worse than the other methods; however, under these conditions, the false negative rate is the lowest for NMR and WAC. Comparison of Techniques for Fragment Screening. Figure 4 shows screening results from WAC in comparison with NMR, SPR, thermal shift analysis, and FP. The figure also includes the results of ITC measurements on the highest affinity (by WAC) fragments and a record of which fragment crystal structures were attempted and determined. Detection limits were set or estimated from the screening conditions as: 4 mM (NMR), 2 mM (WAC), 4 mM (SPR), and 9.9 mM (FP) in KD values and 0.6 °C in temperature shift for thermal (Tm) shift analysis. The fragments are arranged according to rank from fragment screening by WAC with the fragments of highest affinity on top. Only “Class 1” binders by NMR are denoted as hits for comparison with other technologies, whereas for the sake of this analysis, “Class 2” and “Class 3” binders by NMR are denoted as nonhits. Out of the 111 fragments that were screened by NMR, 7 produced ambiguous results where spectral overlap from the cocktail of 17 fragments did not provide a clear resonance for that fragment. Also, fragments that did not give sufficient saturation in the SPR experiments (partial binders) are denoted as nonhits. Table 2 shows the correlation between the different methods of screening. For completeness, this includes correlations between all techniques reported here; however, the discussion focuses on comparison of WAC with the other techniques and

Compounds of very high affinity can therefore not be analyzed isocratically by WAC. Their overall influence on the WAC column is however negligible as the amount of immobilized protein on the column is much higher than the amount of sample in each injection, and hence, only a very small portion of the immobilized protein would be affected. Effects of Various Mobile Phases and Sample Concentrations during Fragment Screening. Fragments 1−12 were analyzed by WAC on column 1 in both PBS pH 7.4 and 20 mM ammonium acetate pH 6.8. The retention times obtained using ammonium acetate buffer as the mobile phase were about 20% shorter than for PBS, which is probably due to the lower ionic strength and pH of the ammonium acetate buffer. However, retention times in the two mobile phases correlated well (R2 = 0.99782; data not shown). Fragments 1−12 were also analyzed on column 1 as single injections and in mixtures using ammonium acetate buffer as the mobile phase. There was no difference in retention times during analysis in mixtures compared to single injections. This implies that there was no competition between the fragments in the mixtures, which is expected at the relatively low sample concentrations (0.1 mM) used in the analysis. In addition, fragments 1−12 were screened on column 1 as single injections with a sample concentration of 0.1 mM and 0.01 mM using PBS as mobile phase. There was no difference in the retention times. A sample concentration of 0.1 mM is therefore considered to be low enough to give a linear isotherm, due to sample dilution upon injection onto the column, and eq 2 gives accurate KD values from WAC. Stability and Reproducibility of HSP90 Columns. The precision of WAC as a screening method was also evaluated by multiple (10 per day) injections of a mixture of 4 fragments and adenosine on column 3. The retention times of individual compounds varied between 1.8 and 15 min. The coefficients of variation (CVs) were as low as 1.1% within the same day. For the three consecutive days (25−27 days), the CV values were 4.8% from day-to-day if a fresh buffer was prepared each day and 3.3% if the same batch of buffer was used each day. The stability of column 3, during storage in PBS pH 7.4 at 4 °C, was evaluated, and the remaining activity was determined from the retention times of adenosine and two selected fragments that were identified as fragment hits. The remaining activity was averaged at 90% (ranging between 83 and 98% for the 3 selected compounds) after 207 days of storage and in total about 100 injections and 50 h of operation. Furthermore, a mixture of fragments 96−111 was screened on column 2 and column 3 using ammonium acetate buffer as the mobile phase. The resulting retention times of the two different columns correlated well (R2 = 0.99322; data not shown), demonstrating the reproducibility of WAC columns. X-ray Crystallography. 32 selected fragments were soaked into apo-HSP90 crystals. Six of these failed to give crystal structures after multiple attempts; 5 only provided structures after between 2 to 4 attempts. All of those attempts 6763

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

Figure 5. Measured KD vaues by (A) WAC (40 analytes), (B) SPR (21 analytes), and (C) ITC (14 analytes).

not between the other techniques. The correlation of screening data between WAC and NMR and between NMR and SPR was excellent at 88% (n = 103) and 90% (n = 102), respectively. When considering all three techniques, the correlation was 82% (n = 101). When comparing WAC with Tm shift, a reasonable correlation was seen (71%; n = 107). In Tm shift assays, weakly binding fragments are frequently not able to stabilize the protein sufficiently to produce a shift in protein melting temperature. Thus, fragment screening by the Tm shift assay may result in a number of false negatives (as observed by others40). For example, as can be seen in Figure 4, fragments 4, 35, 55, and 111 were considered as hits by WAC, NMR, and SPR but not by the Tm shift. However, there are also a number of fragments that were a hit by the Tm shift assay but not by any other techniques and should be regarded as false positives. Finally, the FP assay recorded the largest number of false negatives across the techniques, even allowing for the 9.9 mM threshold set for the sensitivity of the assay. The fragments of highest affinity as identified by WAC were also assessed by ITC, and there is a good correlation between the methods, although WAC is a more sensitive screening technology for more weakly binding fragments. Figure 4 also records for which of the fragments a crystal structure has been attempted and determined. Overall, there is a high chance of success in obtaining a crystal structure for the highest affinity hits from WAC (correlation between WAC and X-ray crystallography of 83% (n = 30)). Crystal structures are

recorded for some of the weaker binding fragments that were a hit in WAC but not seen by NMR; this is probably due to difference in the concentration of fragment used. For some fragments, especially for those with lower affinities, the screening techniques showed variable results. For example, for fragment 105, an ambiguous signal was obtained by NMR and no X-ray crystal structure could be successfully produced, while all other screening methods identified this fragment as a good hit. The KD of fragment 105 was determined to be 50, 155, and 56 μM by WAC, SPR, and ITC, respectively. Further, WAC identified fragment 20 as a reasonably good hit (KD = 393 μM) whereas it was not identified as a hit by either SPR, Tm shift, ITC, or FP and the affinity could not be determined by NMR or X-ray crystallography. Moreover, fragment 74 had a KD of 435 μM according to WAC, but binding could not be definitively detected by any other technique (although identified as a “Class 3” hit by NMR). Out of the 42 fragments that were identified as hits by the first competitive screen by SPR, KD values could be measured for 21 fragments by dose response titrations. KD values for fragments, as measured by WAC, SPR, and ITC, are presented in Figure 5. Here, the 40 fragments that were identified as hits by WAC are included. Further, out of the 27 fragments that were selected for characterization by ITC, dose response titrations enabled determination of KD values for 14, which are also included for this comparison. The correlation between KD values as measured by WAC and SPR was poor (R2 = 0.016, n = 6764

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry

Article

21), as well as between WAC and ITC (R2 = 0.38, n = 14). However, the KDs between ITC and SPR correlated well (R2 = 0.91, n = 13). The buffers for SPR and ITC were similar, while that used for WAC was substantially different. The buffer used for WAC did not contain magnesium, although this should not influence fragment binding. The affinity of weakly binding fragments by SPR may also be less reliable as the top concentration used for response titrations by SPR was 250 μM, and hence, KD values above 250 μM were determined by extrapolation. SPR and ITC are routinely relied upon for affinity determination and show a close correlation in the KD values obtained here. The difference in values obtained by WAC could be due to differences in buffer composition, differently modified HSP90, and supports or (less likely) minor variations in temperature. However, although there is not a good correlation in KD values, out of the 15 highest ranked fragments by WAC, 14 gave acceptable titration data by ITC. Fragment 49 was the exception, as it did not give good data in ITC but was confirmed as a hit by NMR, SPR, and the Tm shift assay. Each of the techniques was operated under optimal conditions for that particular method, rather than attempting to use the same or highly similar conditions for all assays. The differences in results between the various techniques may therefore have more than one explanation. For example, the presence or absence of DMSO during analysis may influence the screening result. Most experiments are performed in 1−5% DMSO in most screening assays, whereas in WAC binding fragments interact with the protein in a DMSO-free environment. This is due to the fact that target-binding fragments are retarded on the affinity column while DMSO is eluted in the void volume. The risk of precipitation of fragments of poor solubility in WAC as a result of the separation from DMSO is kept to a minimum by low sample concentrations (and further dilution of the fragments when traveling on the column). Furthermore, for these experiments, the molecular ions were detected by MS for all fragments, indicating that substantial precipitation had not occurred. Exposure of the protein to DMSO, even at low concentrations, has been shown to influence the properties of some proteins that were tested.41 Different competitors were used for the different screening methods used in this study. Inclusion of inhibitors is important to identify nonspecific binding and to minimize the occurrence of false positives. Therefore, if the inhibitors bind to the active site in different ways, this may influence which fragments are blocked, depending on whether the inhibitor and fragment utilize the same parts of the active site for binding or not. For SPR and WAC, the target was immobilized prior to fragment screening whereas for all other assays the protein was free in solution. For WAC, this was done by attachment directly to the column; for SPR, it was via an N-terminal double-his tag. Immobilization of a target protein can be both an advantage and a disadvantage. Immobilization may change the binding behavior of the protein due to, for instance, conformational changes of the protein, which impairs its activity. Furthermore, nonspecific binding can occur to both the support and linkages with the target. However, it is also clear that immobilization can stabilize the protein, as was seen with WAC in this study. The results presented in this paper demonstrate that WAC provides a useful alternative to other methods for fragment screening. Protein consumption for a screening campaign is relatively low (about 1 mg), much lower than for, e.g., NMR and ITC. A very distinctive feature is that the WAC assay is

performed at low concentrations of fragments. This is in contrast to most other fragment screening methods which rely on higher fragment concentrations to be able to measure weaker affinities. This puts less demand on solubility of the fragment libraries, and analysis in fragment mixtures is expected to be less of a problem since the concentration of each individual fragment is too low to compete for the active site with other components of the same mixture. Another useful feature is that relatively large numbers of fragments can be screened in mixtures, resulting in relatively high-throughput: 1000 fragments can be screened in 67 h when analyzed in mixtures of 15 fragments for 1 h. The affinity range of WAC can be adjusted by the amount of protein immobilized onto the column: the higher the protein load, the weaker is the binding of fragments that can be detected. Although for a particular column, the affinity range of analytes is narrower than for NMR; for instance, thermal elution can be used to broaden the range of affinities that can be screened using the same column. Finally, since WAC is based on separation, samples do not need to be as pure as for other methods, such as for SPR and FP.



CONCLUSIONS We have compared the performance of WAC with established methods for screening of more than 100 fragments for binding to the protein HSP90. The results show that, for hit identification, WAC correlates well with NMR and SPR, the most reliable and widely used of current screening methods, and there is consistency in the results obtained from ITC and crystal structure determination. There is less correlation with Tm shift analysis; however, this technique is recognized to be rather unreliable for screening of weak binding fragments.40 This paper has demonstrated a number of strengths for WAC compared to other fragment screening technologies. The column can be very stable, with the HSP90 columns reported here giving reproducible results when used repeatedly over a period of months. Linking the WAC column to MS allows simultaneous screening of complex mixtures, which increases throughput. The method uses LC/MS equipment that is available in most medicinal chemistry laboratories and does not require a large investment in new, expensive equipment. There are other attractive features of WAC that have not been explored in the work reported here. As it relies on a separation step, it should be possible to analyze components in natural extracts and reaction mixtures; in addition, it should be possible to select stereoisomers. This study illustrates the benefits of using several orthogonal techniques for fragment screening campaigns. HSP90 is a particularly robust target that performs well in measurements by most biophysical techniques; other targets show more variability when screened by different methods, and it is important to cross-validate by as many techniques as possible. We believe that WAC (affinity LC/MS) shows great potential to be a standard and high throughput technology for fragment screening especially in the early or primary evaluation of fragment libraries.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +46 480 446262. Notes

The authors declare no competing financial interest. 6765

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766

Analytical Chemistry



Article

Müller, A.; Varasi, M.; Whittaker, M.; Yarnold, C. J. ChemMedChem 2010, 5, 1697−1700. (23) Murray, C. W.; Carr, M. G.; Callaghan, O.; Chessari, G.; Congreve, M.; Cowan, S.; Coyle, J. E.; Downham, R.; Figueroa, E.; Frederickson, M.; Graham, B.; McMenamin, R.; O’Brien, M. A.; Patel, S.; Phillips, T. R.; Williams, G.; Woodhead, A. J.; Woolford, A. J. J. Med. Chem. 2010, 53, 5942−5955. (24) Wright, L.; Barril, X.; Dymock, B.; Sheridan, L.; Surgenor, A.; Beswick, M.; Drysdale, M.; Collier, A.; Massey, A.; Davies, N.; Fink, A.; Fromont, C.; Aherne, W.; Boxall, K.; Sharp, S.; Workman, P.; Hubbard, R. E. Chem. Biol. 2004, 11, 775−785. (25) Fischer, M.; Leech, A. P.; Hubbard, R. E. Anal. Chem. 2011, 83, 1800−1807. (26) Strandh, M.; Andersson, H. S.; Ohlson, S. Methods Mol. Biol. 2000, 147, 7−23. (27) Bergström, M.; Liu, S.; Kiick, K. L.; Ohlson, S. Chem. Biol. Drug Des. 2009, 73, 132−141. (28) Kasai, K.; Oda, Y.; Nishikata, M.; Ishii, S. J. Chromatogr. 1986, 376, 33−47. (29) Mayer, M.; Meyer, B. Angew. Chem., Int. Ed. Engl. 1999, 38, 1784−1788. (30) Dalvit, C.; Pevarello, P.; Tato, M.; Veronesi, M.; Vulpetti, A.; Sundstrom, M. J. Biomol. NMR 2000, 18, 65−68. (31) Hajduk, P. J.; Olejniczak, E. T.; Fesik, S. W. J. Am. Chem. Soc. 1997, 119, 12257−12261. (32) Howes, R.; Barril, X.; Dymock, B. W.; Grant, K.; Northfield, C. J.; Robertson, A. G.; Surgenor, A.; Wayne, J. L.; James, K.; Matthews, T.; Cheung, K. M.; McDonald, E.; Workman, P.; Rysdale, M. J. Anal. Biochem. 2006, 350, 202−213. (33) Zhang, J.-H. J. Biomol. Screening 1999, 4, 67−73. (34) Potter, A.; Ray, S.; Gueritz, L.; Nunns, C.; Bryant, C.; Scrace, S.; Matassova, N.; Baker, L.; Dokurno, P.; DA, R.; Surgenor, A.; Davis, B.; Murray, J.; Richardson, C.; Moore, J. Bioorg. Med. Chem. Lett. 2010, 20, 586−590. (35) Dymock, B. W.; Barril, X.; Brough, P. A.; Cansfield, J. E.; Massey, A.; McDonald, E.; Hubbard, R. E.; Surgenor, A.; Roughley, S. D.; Webb, P.; Workman, P.; Wright, L.; Drysdale, M. J. J. Med. Chem. 2005, 48, 4212−4215. (36) Navaza, J. Acta Crystallogr. 1994, 50, 157−163. (37) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Acta Crystallogr. 1997, D53, 240−255. (38) Emsley, P.; Cowtan, K. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2004, 60, 2126−2132. (39) Collaborative Computational Project, N. 4. Acta Crystallogr., Sect. D: Biol. Crystallogr. 1994, 50, 760−763. (40) Schulz, M. N.; Landström, J.; Hubbard, R. E. Anal. Biochem. 2012, 433, 43−47. (41) Tjernberg, A.; Markova, N.; Griffiths, W. J.; Hallén, D. J. Biomol. Screening 2006, 11, 131−137.

ACKNOWLEDGMENTS We thank Allan Surgenor and Lisa Baker for crystallography and Neil Whitehead for protein production. Prof. Roland Isaksson is gratefully acknowledged for reviewing the manuscript.



REFERENCES

(1) Shuker, S. B.; Hajduk, P. J.; Meadows, R. P.; Fesik, S. W. Science 1996, 274, 1531−1534. (2) Congreve, M.; Chessari, G.; Tisi, D.; Woodhead, A. J. J. Med. Chem. 2008, 51, 3661−3680. (3) Erlanson, D. A. Curr. Opin. Biotechnol. 2006, 17, 643−652. (4) Schulz, M.; Hubbard, R. Curr. Opin. Pharmacol. 2009, 9, 615− 621. (5) Baurin, N.; Aboul-Ela, F.; Barril, X.; Davis, B.; Drysdale, M.; Dymock, B.; Finch, H.; Fromont, C.; Richardson, C.; Simmonite, H.; Hubbard, R. E. J. Chem. Inf. Comput. Sci. 2004, 44, 2157−2166. (6) Chen, I.-J.; Hubbard, R. E. J. Comput.-Aided Mol. Des. 2009, 23, 603−620. (7) Hubbard, R. E.; Murray, J. B. Methods Enzymol. 2011, 493, 509− 531. (8) Hann, M. M.; Leach, A. R.; Harper, G. J. Chem. Inf. Comput. Sci. 2001, 41, 856−864. (9) Lanter, J.; Zhang, X.; Sui, Z. Methods Enzymol. 2011, 493, 421− 445. (10) Maurer, T. Methods Enzymol. 2011, 493, 469−485. (11) Orita, M.; Ohno, K.; Warizaya, M.; Amano, Y.; Niimi, T. Methods Enzymol. 2011, 493, 383−419. (12) Roughley, S.; Wright, L.; Brough, P.; Massey, A.; Hubbard, R. E. Top. Curr. Chem. 2011, 317, 61−82. (13) Hopkins, A. L.; Groom, C. R.; Alex, A. Drug Discovery Today 2004, 9, 430−431. (14) Duong-Thi, M.-D.; Meiby, E.; Bergström, M.; Fex, T.; Isaksson, R.; Ohlson, S. Anal. Biochem. 2011, 414, 138−146. (15) Duong-Thi, M.-D.; Bergström, M.; Fex, T.; Isaksson, R.; Ohlson, S. J. Biomol. Screening 2012, 18, 160−171. (16) Meiby, E.; Knapp, S.; Elkins, J.; Ohlson, S. Anal. Bioanal. Chem. 2012, 404, 2417−2425. (17) Biamonte, M. A.; Van de Water, R.; Arndt, J. W.; Scannevin, R. H.; Perret, D.; Lee, W.-C. J. Med. Chem. 2010, 53, 3−17. (18) Barril, X.; Brough, P.; Drysdale, M.; Hubbard, R. E.; Massey, A.; Surgenor, A.; Wright, L. Bioorg. Med. Chem. Lett. 2005, 15, 5187− 5191. (19) Huth, J. R.; Park, C.; Petros, A. M.; Kunzer, A. R.; Wendt, M. D.; Wang, X.; Lynch, C. L.; Mack, J. C.; Swift, K. M.; Judge, R. A.; Chen, J.; Richardson, P. L.; Jin, S.; Tahir, S. K.; Matayoshi, E. D.; Dorwin, S. A.; Ladror, U. S.; Severin, J. M.; Walter, K. A.; Bartley, D. M.; Fesik, S. W.; Elmore, S. W.; Hajduk, P. J. Chem. Biol. Drug Des. 2007, 70, 1−12. (20) Brough, P. A.; Aherne, W.; Barril, X.; Borgognoni, J.; Boxall, K.; Cansfield, J. E.; Cheung, K.-M. J.; Collins, I.; Davies, N. G. M.; Drysdale, M. J.; Dymock, B.; Eccles, S. A.; Finch, H.; Fink, A.; Hayes, A.; Howes, R.; Hubbard, R. E.; James, K.; Jordan, A. M.; Lockie, A.; Martins, V.; Massey, A.; Matthews, T. P.; McDonald, E.; Northfield, C. J.; Pearl, L. H.; Prodromou, C.; Ray, S.; Raynaud, F. I.; Roughley, S. D.; Sharp, S. Y.; Surgenor, A.; Walmsley, D. L.; Webb, P.; Wood, M.; Workman, P.; Wright, L. J. Med. Chem. 2008, 51, 196−218. (21) Brough, P. A.; Barril, X.; Borgognoni, J.; Chene, P.; Davies, N. G. M.; Davis, B.; Drysdale, M. J.; Dymock, B.; Eccles, S. a; GarciaEcheverria, C.; Fromont, C.; Hayes, A.; Hubbard, R. E.; Jordan, A. M.; Jensen, M. R.; Massey, A.; Merrett, A.; Padfield, A.; Parsons, R.; Radimerski, T.; Raynaud, F. I.; Robertson, A.; Roughley, S. D.; Schoepfer, J.; Simmonite, H.; Sharp, S. Y.; Surgenor, A.; Valenti, M.; Walls, S.; Webb, P.; Wood, M.; Workman, P.; Wright, L. J. Med. Chem. 2009, 52, 4794−4809. (22) Barker, J. J.; Barker, O.; Courtney, S. M.; Gardiner, M.; Hesterkamp, T.; Ichihara, O.; Mather, O.; Montalbetti, C. A. G. N.; 6766

dx.doi.org/10.1021/ac400715t | Anal. Chem. 2013, 85, 6756−6766