Limited Proteolysis Combined with Stable Isotope Labeling Reveals

Article pubs.acs.org/jpr

Limited Proteolysis Combined with Stable Isotope Labeling Reveals Conformational Changes in Protein (Pseudo)kinases upon Binding Small Molecules Michela Di Michele,†,‡,▽ Elisabeth Stes,†,‡,▽ Elien Vandermarliere,†,‡ Rohit Arora,§ Juan Astorga-Wells,∥ Jonathan Vandenbussche,†,‡ Erika van Heerde,⊥ Roman Zubarev,# Pascal Bonnet,§ Joannes T. M. Linders,⊥ Edgar Jacoby,⊥ Dirk Brehmer,⊥ Lennart Martens,†,‡ and Kris Gevaert*,†,‡ †

Department of Medical Protein Research, VIB, A. Baertsoenkaai 3, 9000 Ghent, Belgium Department of Biochemistry, Ghent University, A. Baertsoenkaai 3, 9000 Ghent, Belgium § Institut de Chimie Organique et Analytique (ICOA), UMR 7311 CNRS-Université d’Orléans, Pôle de chimie, Rue de Chartres, 45100 Orléans, France ∥ Biomotif AB, P.O. Box 156, SE-182 12 Danderyd, Sweden ⊥ Oncology Discovery, Janssen Research and Development, A Division of Janssen Pharmaceutica NV, Turnhoutseweg 30, B-2340 Beerse, Belgium # Department of Medical Biochemistry and Biophysics, Karolinska Institutet, Scheelelaberatoriet Scheeles väg 2, SE-171 77 Stockholm, Sweden

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 30, 2015 | http://pubs.acs.org Publication Date (Web): August 25, 2015 | doi: 10.1021/acs.jproteome.5b00282

‡

S Supporting Information *

ABSTRACT: Likely due to conformational rearrangements, small molecule inhibitors may stabilize the active conformation of protein kinases and paradoxically promote tumorigenesis. We combined limited proteolysis with stable isotope labeling MS to monitor protein conformational changes upon binding of small molecules. Applying this method to the human serine/threonine kinase B-Raf, frequently mutated in cancer, we found that binding of ATP or its nonhydrolyzable analogue AMP-PNP, but not ADP, stabilized the structure of both B-RafWT and B-RafV600E. The ATP-competitive type I B-Raf inhibitor vemurafenib and the type II inhibitor sorafenib stabilized the kinase domain (KD) but had distinct effects on the Ras-binding domain. Stabilization of the B-RafWT KD was confirmed by hydrogen/deuterium exchange MS and molecular dynamics simulations. Our results are further supported by cellular assays in which we assessed cell viability and phosphorylation profiles in cells expressing B-RafWT or B-RafV600E in response to vemurafenib or sorafenib. Our data indicate that an overall stabilization of the B-Raf structure by specific inhibitors activates MAPK signaling and increases cell survival, helping to explain clinical treatment failure. We also applied our method to monitor conformational changes upon nucleotide binding of the pseudokinase KSR1, which holds high potential for inhibition in human diseases. KEYWORDS: B-Raf, conformational change, kinase, kinase switch mechanism, KSR1, inhibitor, limited proteolysis, method development, protein structure, small molecule

■

INTRODUCTION

catalytically active kinase domain consisting of a small N-terminal lobe, usually containing β strands, and a bigger C-lobe mainly composed of α-helices, connected by a highly flexible hinge region. ATP binds into the deep cleft between the two lobes, with the adenine ring forming hydrogen bonds with the hinge and the ribose and triphosphate groups lodging in a hydrophilic channel that extends the substrate binding site.2 All kinases have a

Protein kinases play key roles in cellular signaling events that mediate cell metabolism, growth, proliferation, and differentiation. Deregulated kinase activities are manifested in several human diseases, such as immunological, neurological, metabolic, and infectious disorders and most notably cancer.1,2 Kinases are thus attractive targets for inhibition, as witnessed by the increasing but still low number of kinase inhibitors approved by the FDA over the last years.3,4 The 518 human kinases5 have a remarkably similar 3D structure. Their main element is the © XXXX American Chemical Society

Received: April 1, 2015

A

DOI: 10.1021/acs.jproteome.5b00282 J. Proteome Res. XXXX, XXX, XXX−XXX

Article


Journal of Proteome Research

with and transactivation of other Raf proteins are possibly potentiated, resulting in paradoxical activation of the MAPK pathway and tumorigenesis stimulation.22−26 Indeed, smallmolecule ATP-mimetics, besides (temporarily) blocking the catalytic activity, can stabilize the target kinase in an active conformation by breaking up the autoinhibitory folded state and transactivating other functional subdomains along the whole protein structure (i.e., the switch mechanism).26,27 Other inhibitor-mediated mechanisms have been shown to cause undesired effects, such as induction of binding to KSR proteins that act as scaffolds for the RAF-MEK-ERK complex,28 resulting in ERK signaling activation. This type of kinase activation was first observed for vemurafenib but was later on reported for other protein kinase inhibitors.29−32 KSR1, an important B-Raf partner in the MAPK pathway, contains a kinase domain that lacks the conserved β3 Lys residue of the KD required for its catalytic activity and was thus predicted to be inactive;33 however, several reports highlight pivotal roles for pseudokinases in the regulation of cellular processes.28,34 Recent studies indicated that KSR1 has a potential dual activity, functioning both as a kinase (to directly phosphorylate Raf-1 and ERK) and a scaffold protein with a crucial role in ERK signaling.35,36 Upon stimulation, KSR1 translocates from the cytosol to the plasma membrane, where it forms a complex with Ras, Raf, MEK and 14−3−3 proteins, inducing Ras1 activation and contributing to the regulation of MAPK signaling.37 Moreover, in response to most B-Raf inhibitors, KSR1 binds to B-Raf in a Ras-independent way and competes with C-Raf for the binding of the inhibitor to B-Raf, thus attenuating the paradoxically activating effect of the inhibitors on the ERK pathway.37,38 Additionally, KSR2 was shown to be allosterically activated by B-Raf and to catalyze MEK phosphorylation following a switch from the inactivated to activated conformation.28 Overall, these data highlight a potential important therapeutic implication in targeting the KSR1 scaffold function and the prerequisite to explore which KSR1conformation should be stabilized/destabilized to block function. Given the high clinical relevance of kinases, detailed insights into the switch mechanism should generate important data, instructing the development of novel therapies. Therefore, our primary goal was to set up a method that enables the study of conformational changes and interdomain communication processes in full-length kinases when complexed to small molecules, both nucleotides and inhibitors. Indeed, the currently available and used techniques to resolve protein structures each have their own limitations; some classical methods among these such as X-ray crystallography, nuclear magnetic resonance (NMR), or cryo-electron microscopy have been reviewed by Vandermarliere and colleagues.39 More recently, MS-based methods have been introduced for the characterization of (dynamic) protein conformations. In particular, hydrogen/ deuterium exchange MS (HDX-MS) provides detailed information on protein conformation, as was, for instance, shown for several kinases bound to their inhibitors, for example, ERK2,40,41 MAPK,42 PKA,43 Hck,44 KIT,45 and ABL1.44,46,47 Compared with crystallography and NMR, HDX-MS is more sensitive, requires simpler sample preparation, and can resolve more complex protein mixtures. Nevertheless, it holds disadvantages such as limited spatial resolution, possible back-exchange of hydrogen isotopes, and scrambling that can impair precise measurements. Moreover, the use of buffer systems that are limited to those compatible with subsequent MS (ESI) analysis

conserved activation loop starting with the sequence DFG, whose conformation is the determinant for a kinase’s activated or inactivated status (DFG-in or -out, respectively).6 Most smallmolecule kinase inhibitors compete with ATP by targeting the ATP binding site in a kinase active DFG-in conformation (type I inhibitors); however, other inhibitors bind and stabilize the inactive DFG-out conformation (type II inhibitors) and were found to be more selective than type I inhibitors, despite some exceptions.7,8 Allosteric inhibitors (type III inhibitors) that bind remote kinase sites and stabilize the inactive conformation without competing with the high intracellular ATP levels were also developed.9 Finally, type IV inhibitors covalently modify and block active kinases.2 Single kinase inhibitor therapy often fails due to the occurrence of drug resistance,10 arisen through target gene alternative splicing, overexpression, epigenetic activation mutations that reduce drug binding or activate alternative signaling pathways.2 This underlines the urgent need for more efficient kinase inhibitors. Previous studies investigated kinase structural changes that are induced upon binding of kinase inhibitors to elucidate their mechanism of action (reviewed in ref 9). Of note here is that the majority of the currently available structural information is generally limited to the catalytic kinase domain,11 whereas data on the overall kinase structure and its interdomain communications are paramount to understand kinase regulation and to develop novel clinically differentiating inhibitors. As an example, a recent study reported that the regulatory SH3 and SH2 domains of Src-family kinases are differentially modulated by ATP-competitive inhibitors through stabilization of different inactive protein conformations.12 This finding highlights the importance of elucidating kinase conformational changes for drug discovery. In the current work we focused on B-Raf, a serine/threonine kinase functioning in the Ras-Raf-MEK-ERK MAPK (mitogenactivated protein kinase) pathway, the first MAPK pathway identified.13 Raf is activated by growth factors or upon hetero- or homodimerization at the plasma membrane14 and in turn activates MEK1 and MEK2 via phosphorylation.15 The latter proteins induce activation of ERK1 and ERK2, which then stimulate the MAPK signaling cascade that regulates cell growth, differentiation, and survival. In humans, three different Raf isoforms originating from three independent genes exist: A-Raf, B-Raf, and C-Raf (or Raf-1), of which B-Raf has the highest kinase activity.16,17 B-Raf is composed of three conserved regions that are characteristic of the Raf kinase family: a Ras-binding domain (RBD), a serine-rich hinge region, and a catalytic kinase domain (KD). Structures are available for both the RBD and KD but not for the full-length protein. B-Raf has a high degree of conformational plasticity, and the KD on its own is dynamic, alternating between two conformational states to propagate signals. In the closed, active conformation, B-Raf is able to catalyze the transfer of a phosphate group from ATP to substrate serines or threonines, while in the open, inactive conformation this transfer is prevented.18,19 B-Raf is the most frequently mutated protein kinase in human cancer (8% of all cancers).20 In malignant melanomas, the V600E mutation, which results in elevated kinase activity, accounts for 80% of the reported B-Raf mutations.21 Because of its frequent occurrence as an oncogene, B-Raf is an obvious target for cancer therapy, and several B-Raf inhibitors are currently in use or in development. Most of these are ATP competitive type I (e.g., vemurafenib) or type II inhibitors (e.g., sorafenib);22 however, because type I inhibitors potentially stabilize the active kinase conformation, dimerization B


Article



incubation at room temperature. Then, kinase detection reagent (Promega) was added to convert ADP to ATP, and the plate was incubated for 1 h at room temperature. Plates were read on a TopCount luminometer (Canberra-Packard, Waverley, U.K.). All experiments were carried out in triplicate.

may prevent the monitoring of proteins in near-native conditions.48 We combined and advanced known technologies, such as limited proteolysis and stable isotope labeling mass spectrometry, to monitor conformational changes in full-length protein (pseudo)kinases. Limited proteolysis was already used to help elucidating protein structural modifications;49−51 however, in the majority of the studies that employed limited proteolysis so far, it is coupled to SDS-PAGE or to MS analysis, thus implying significant limitations in terms of quantitation accuracy and robustness by providing data more at a qualitative than a quantitative level, allowing only the identification of different cleavage sites in different constructs of the protein of interest. We instead combine limited proteolysis with stable isotope labeling MS, providing more detailed information on protein conformational changes. In particular, the conformational plasticity of BRaf and KSR1 was monitored upon binding of nucleotides or inhibitors. Among others, our method allowed us to detect structural changes occurring at several domains over the whole kinase structure upon binding of different types of inhibitors. Together with cell-based assay data, these findings shed new light on the functionally crucial but still poorly understood kinase switch mechanisms.

■

Limited Proteolysis and Stable Isotope Labeling Mass Spectrometry

Both the purified B-RafWT and B-RafV600E (60 pmol for limited proteolysis experiments and 360 pmol for complete digestion) were buffer-exchanged to 20 mM triethylammonium bicarbonate (TEAB) buffer pH 8.5, using a NAP-5 column and incubated for 4 h at 30 °C with vehicle 1% DMSO, ADP, ATP, or AMP-PNP (10 nmol) or inhibitors (1 nmol) and in the presence of 1 mM MgCl2 where indicated. Limited proteolysis was performed with trypsin (sequencing-grade-modified trypsin, Promega, Madison, WI) (1:200, enzyme:substrate) in 100 μL at 30 °C for 10 min. The reaction was stopped by removing trypsin through centrifugation over 10 kDa cutoff filters (Nanosep centrifugal device, Pall Life Sciences). The full digested sample was obtained by denaturing the B-Raf protein at 95 °C for 5 min and incubating with trypsin (1:20, enzyme:substrate) in a volume of 500 μL at 37 °C for 4 h. The generated peptide mixtures were labeled for quantification using N-hydroxysuccinimide (NHS) esters of different butyric acid forms: 12C4-butyric acid (light label, limited proteolysis samples) and 13C4-butyric acid (heavy label, full digest).52 In brief, 40 μM NHS-butyrate dissolved in 50% acetonitrile (ACN) was added to the sample and incubated for 1 h at 30 °C, and this step was repeated once. The samples were then incubated with 10 mM glycine for 10 min at 30 °C to quench nonreacted NHS-esters and then with 20 mM hydroxylamine for 5 min at room temperature to revert Oacylation of Ser, Thr, or Tyr residues. 100 μL of light labeled BRaf peptides (limited proteolysis samples) was mixed with 3 μL of heavy labeled peptides (full digested sample), vacuum-dried, and resuspended in 20 μL of loading solvent (2% ACN and 0.1% TFA). These peptide mixtures were then introduced into an LC−MS/MS system, an Ultimate 3000 RSLC nano (Dionex, Amsterdam, The Netherlands) in-line connected to an LTQ Orbitrap Velos (Thermo Fisher Scientific, Bremen, Germany) for analysis. Samples were first loaded on a trapping column (made in-house, 100 μm I.D. × 20 mm, 5 μm C18 Reprosil-HD beads, Dr. Maisch). After back-flushing from the trapping column, the samples were loaded on a reverse-phase column (made in-house, 75 μm I.D. × 150 mm, 5 μm C18 Reprosil-HD beads, Dr. Maisch) with loading solvent and separated with a linear gradient from 2% of 0.1% formic acid (solvent A′) to 55% of 0.1% formic acid in 80% ACN (solvent B′) at a flow rate of 300 nL/min, followed by a wash reaching 100% solvent B′. For KSR1, limited proteolysis and stable isotope labeling conditions were the same as for B-Raf, except for a few modifications. KSR1 (50 pmol for limited proteolysis experiments and 200 pmol for complete digestion) were bufferexchanged to 20 mM TEAB buffer pH 8.5 using a NAP-5 column. Limited proteolysis was carried with trypsin 1:50 (enzyme:substrate) for 10 min, and, following stable isotope labeling, 100 μL of light-labeled KSR1 peptides (limited proteolysis samples) was mixed with 2 μL of heavy labeled peptides (full digested sample) and analyzed by LC−MS/MS. The mass spectrometer was operated in data-dependent mode, automatically switching between MS and MS/MS acquisition for the 10 most abundant peaks in a MS spectrum.

MATERIALS AND METHODS

Protein Expression and Purification

Recombinant human full length B-RafWT and mutant B-RafV600E and KSR1 proteins were provided by Janssen R&D, a division of Janssen Pharmaceutica NV, and purchased from Dr. James Hastie (University of Dundee), respectively. The purity was verified by SDS-PAGE (Supplemental Figure S1). Analysis of the Nucleotide-Bound State and Catalytic Activity of the B-Raf Kinases

To verify the nucleotide status of the B-Raf and KSR1 protein, 10 μM ATP and ADP standard, and 10 μg of B-Raf protein with or without preincubation for 30 min at 30 °C with 1 nmol ATP in 10 mM Tris-HCl, we analyzed pH 8 and 10 mM MgCl2 by RPHPLC. In the case of sample preincubation with ATP, excess ATP was removed by dialysis. Samples were then heated for 5 min at 90 °C to detach bound nucleotides, which were collected by centrifugation over a 10 kDa cutoff filter (Nanosep centrifugal device, Pall Life Sciences, Fontenay-sous-Bois, France). The resulting samples and the nucleotide standards were vacuumdried, dissolved in 8 mM tetrabutylammonium hydrogen sulfate (solvent A) and separated over a RP-HPLC column (2.1 mm internal diameter (I.D.) × 150 mm (length) 300SB-C18 column, Zorbax, Agilent, Waldbronn, Germany) using an Agilent 1100 Series HPLC system. The gradient used for separation was: 0 → 7 min, 0 → 0.5% solvent B; 7 → 37 min, 0.5 → 23% solvent B; 37 → 38 min, 23 → 30% solvent B; 38 → 39 min, 30 → 0.5% solvent B; 39 → 49 min 30, 0.5% solvent B, where solvent B was 8 mM tetrabutylammonium hydrogen sulfate in methanol. Using Agilent’s electronic flow controller, a constant flow of 80 μL/ min was applied. Absorbance was monitored at 254 nm. Enzyme kinetic reactions were performed with the ADP-Glo Kinase Assay (Promega, Madison, WI) according to the manufacturer’s instructions. In brief, B-RafWT or B-RafV600E (30 pmol) were preincubated with inhibitors (1 nmol) for 30 min at 30 °C in 20 mM TEAB, pH 8 containing 10 mM MgCl2 before the addition of 10 nmol ATP for 5 min, 10 min, 1 h, or 4 h. Reactions were stopped, and residual ATP was depleted by adding the ADP-Glo reagent (Promega), followed by a 40 min C


Article



solution (5 M urea, 100 mM TCEP, and 0.5% TFA) and flashfreezing in liquid nitrogen. A B-Raf solution deuterated for 3 days was used as a fully deuterated sample for back-exchange correction. Samples were analyzed in a semiautomated HDX-MS system (Biomotif AB, Stockholm, Sweden) in which manually injected samples were automatically digested, cleaned, and separated at 1 °C. Deuterated samples were digested using an in-house packed immobilized pepsin column (2.1 mm I.D. × 30 mm from ACE HPLC columns packed with pepsin-agarose from porcine gastric mucosa obtained from Sigma-Aldrich, St. Louis, MO) by a 1 min stop-flow protocol, followed by an online desalting step using a 2 mm I.D. × 10 mm length C-18 precolumn (ACE HPLC Columns, Aberdeen, U.K.) using 0.05% TFA at 300 μL/min for 3 min. Peptic peptides were then separated by a 18 min 2−40% linear gradient of ACN in 0.3% formic acid using a C18 Chromolith FastGradient 2 mm I.D. × 50 mm column. An Orbitrap XL mass spectrometer (Thermo Fisher Scientific) operated at 60 000 resolution at m/z 400 was used for analysis. Five LC−MS/MS runs were carried out to identify peptic peptides. The Mascot software (Matrix Science) was used to search a database composed of the B-Raf sequence using the following parameters: variable modifications: oxidation of methionine, enzyme setting was “none”, and peptide and fragment mass tolerances of ±5 ppm and ±0.6 Da, respectively. Peptides with Mascot ion scores higher than 20 were further selected for HDX kinetic studies. In addition, each selected peptide was further validated by inspecting the MS/MS spectrum. The HDExaminer software (Sierra Analytics, Modesto, CA) was used to process all HDX-MS data.

In the LTQ-Orbitrap Velos, full-scan MS spectra were acquired in the Orbitrap at a target value of 1E6 with a resolution of 60 000. The 10 most intense ions were then isolated for fragmentation in the linear ion trap, with a dynamic exclusion of 20 s. Peptides were fragmented after filling the ion trap at a target value of 1 × 104 ion counts. From the MS/MS data in each run, Mascot Generic Files (mgf files) were created using the Mascot Distiller software (version 2.3.2.0, Matrix Science, www. matrixscience.com/Distiller.html). These mgf files were then searched with the Mascot search engine (MatrixScience, www. matrixscience.com) using the Mascot Daemon interface (version 2.3, Matrix Science). Spectra were searched against an in-house database including the human B-Raf wild-type and mutant (V600E) sequences in a Spodoptera f rugiperda Sf9 insect cell background (UniProtKB/Swiss-Prot protein database, release of 15/02/2012 containing 488 sequence entries) for B-Raf peptides and the human UniProtKB/Swiss-Prot database (release of 16/ 04/2014 containing 20 340 sequence entries) for KSR1 peptides. Variable modifications were set to methionine oxidation and pyro-glutamate formation of N-terminal glutamine, while fixed modifications were on 12C4 or 13C4 butyrate on lysines and peptide N-termini, reflecting the peptide labeling strategy used. The mass tolerance on precursor ions was set to ±10 ppm (with Mascot’s C13 option set to 1) and on fragment ions to ±0.5 Da. The peptide charge was set to 2+ and 3+, and the instrument setting was put on ESI-TRAP. The enzyme setting was trypsin, allowing for one missed cleavage, and cleavage was also allowed when arginine or lysine are followed by proline. Only peptides that were ranked first and scored above the threshold score, set at 99% confidence, were withheld. Identified peptides were then quantified using the Mascot Distiller Toolbox version 2.3.2.0 (MatrixScience) in the precursor mode. Abundance ratios are calculated from the area underlying the light and heavy isotopic envelope of the corresponding peptide and are normalized to the control (with vehicle DMSO, without small molecule or inhibitor treatment). Statistical analysis of the differences between the samples treated with small molecules or inhibitors and the control were assessed by Student’s two tailed t test, after transformation of the peptide abundance ratios to log 2 values. All data were processed and managed by ms_lims,53 and data analysis was performed using R (http://www.R-project.org) embedded in Knime (https://www.knime.org/).

Preparation of B-Raf Kinase Domains and Molecular Dynamics Simulations

The starting B-RafWT KD model in its active conformation was prepared using the homology modeling function in MOE2012 (MOE, Chemical Computing Group). The sequence for the BRafWT KD was obtained from its UniProt entry (B-RAF_HUMAN, P15056, isoform 1). The crystal structure of the B-RafWT KD in active state (3C4C.A) was used as template for the BRafWT KD model, and the cocrystallized ligand in the template binding site was included as environment for induced fit for the generation of the final model. The model was refined using the Amber12EHT force field with Born solvation, and hydrogens were added using the Protonate3D function. The binding modes of ADP, ATP, AMP-PNP, and vemurafenib molecules were obtained by superposition of B-RafWT onto the crystal structures 1HCK, 4GVA, 3LMG and 3OG7, respectively. The magnesium ion in the PDB structure 1HCK was placed in the active site of the B-RafWT complexes with ATP, ADP, and AMP-PNP. Molecular dynamics (MD) simulations were carried out on BRafWT complex models using the Amber12 suite54 with Amber12sb force field. The parameters for the ligands were prepared using the antechamber utility55 of AmberTools12. The topology files for the B-RafWT complexes were generated, and all the systems were then solvated using TIP3P solvent box of 10 Å radii and neutralized by adding counterions. The solvated systems were then subjected to a two-step minimization. In the first step, positional restraints on the solute molecules were applied, and only the solvent and ion molecules were allowed to minimize for 2500 steps of the steepest descent method, followed by 2500 steps of the conjugate gradient method. In the second step, the entire system was subjected to minimization for 2500 steps using the steepest descent method, followed by 2500 steps

Hydrogen−Deuterium Exchange Mass Spectrometry (HDX-MS)

An aliquot of 500 μL of B-Raf (0.66 mg/mL in 50 mM Tris-HCl pH 7.5 with 0.1 mM EGTA, 270 mM sucrose, 150 mM NaCl, 0.03% Brij-35, 0.1% β-mercaptoethanol, 0.2 mM PMSF, and 1 mM benzamidine) was loaded on a Sephadex G-25 column (Illustra NAP-5 column, GE Healthcare, Little Chalfont, U.K.) using 5 mM DTT in 50 mM Tris-HCl pH 7.5 (with 5 mM MgCl2 for the samples incubated with nucleotide) as equilibration and elution buffer. To compare amide HDX kinetics between B-Raf and B-Raf/ligand complexes, we mixed 4 μL of the purified protein (23.6 pmol) and 1.7 μL of each ligand (17 nmol for ATP and AMP-PNP and 170 pmol for vemurafenib and sorafenib) with 17 μL of deuterated buffer with the same ionic composition as the protein/ligand sample and incubated for 4 h at 30 °C. ATP was dissolved in 50 mM Tris-HCl pH 7.5, AMP-PNP in water, and vemurafenib and sorafenib in 1% DMSO. Deuteration was allowed at 4 °C for 1, 4, or 60 min, as indicated, in triplicate. Each reaction was stopped by the addition of 23 μL of quenching D


Article


shown in several studies61−66 but was so far not applied for the study of small molecule-induced conformational changes in kinases. Moreover, unlike previous studies based on limited proteolysis in which the digestion is ended by adding formic acid, EDTA, SDS loading buffer, or protease inhibitors, we terminated the proteolysis by physically removing the protease via centrifuge filtration, thus immediately stopping the proteolytic activity and making the peptides available for further MS analysis. Another novel aspect of this technology is the combination with stable isotopic labeling MS, which allows an accurate quantification of the proteolytic peptides obtained from the protein bound to different small molecules depending on its conformation and protease cleavage site accessibility. More in particular, limited proteolysis was combined with stable isotope labeling of the peptide mixtures using NHS-esters of isotopic variants of butyric acid. We incorporated light (12C4) butyrate at the α- and ε-amines of peptides derived from the limited proteolysis experiment, whereas heavy (13C4) butyrate was used to label the peptides resulting from full proteolysis (Figure 1). In a mixed population of both peptide samples, each label introduces a fixed mass difference of 4 Da between differently labeled peptides, thus allowing us to calculate differences in ion peak intensities in a straightforward way in MS spectra. As we will illustrate in the next sections, our

using the conjugate gradient method. The systems were then heated from 0 to 300 K using Langevin thermostat, at a constant volume, for 1000 ps with a time-step of 2 fs. A weak restraint of 5 kcal/mol/Å2 was applied on the solute during this run. The solute restraints were then removed, and the systems were equilibrated over a period of 1000 ps with a time-step of 2 fs under constant pressure conditions to relax the solvent density. Finally, the production runs were carried out using the NPT ensemble for 100 ns with a nonbonded cutoff of 8 Å and at time step of 2 fs at 300 K. During the equilibration and production steps, bonds involving hydrogens were constrained using the SHAKE algorithm. Long-range electrostatic interactions were treated using the Particle Mesh Ewald (PME) method.56 All of the simulations were carried out using the PMEMD module54 of Amber12 suite. The analyses were performed using the Bio3D package.57 Downloaded by UNIV OF NEBRASKA-LINCOLN on August 30, 2015 | http://pubs.acs.org Publication Date (Web): August 25, 2015 | doi: 10.1021/acs.jproteome.5b00282

Cell Viability

The human colon cancer cell line SW620 and the malignant melanoma A375 cell line were obtained from ATCC, seeded at 4000 and 1000 per well (96 well format), respectively, and treated with compounds on the next day for 96 h. The compounds’ impact on cell viability was analyzed via the MTT assay.58 Immunoblotting

SW620 and A375 cells were seeded at 750 000 and 300 000 per well (six-well format) and 48 h later (at ∼75% confluence) treated with increasing compound concentrations either for 1 or 4 h. Cells were lysed in 10 mM Tris-HCl pH 7.6, 1% SDS, and 20 mM Na3VO4, and 10 to 25 μg of protein material from each sample was separated by SDS-PAGE and transferred to PVDF membranes. Membranes were incubated with the indicated primary antibodies and analyzed by the Odyssey detection system (Li-Cor, Lincoln, NE). Used primary antibodies for MEK, phospho-MEK, and phospho-ERK were purchased from Cell Signaling (Danvers, MA), B-Raf from Santa Cruz (Santa Cruz, CA), and actin from Sigma-Aldrich (Copenhagen, Denmark). Corresponding secondary antibodies were from Rockland Immunochemicals (Gilbertsville, PA) and Invitrogen (Carlsbad, CA).

■

RESULTS AND DISCUSSION

Limited Proteolysis Approach That Monitors Conformational Changes in Kinases upon Small Molecule Binding

We coupled limited proteolysis with stable isotope labeling and mass spectrometry analysis to characterize the kinase substructures that are linked to conformational changes upon binding small molecules. In a conventional proteomics experiment, protein digestion should be as complete as possible; therefore, proteins are denatured and digested overnight with a rather high enzyme to substrate ratio.59 In a limited proteolysis experiment, the protease is added to a nondenatured protein sample in a low enzyme to substrate ratio, and after a short incubation period the protease is removed or inactivated, thus avoiding full digestion. As the native conformation of substrates is attacked by the protease, only sites located at the protein surface get cleaved.60 As a result, limited proteolysis allows the identification of surface-exposed residues and flexible regions within a protein. The concept that a protein bound to a small molecule (or folded) is more protected from proteolysis compared with a drug-free (or unfolded) protein has been

Figure 1. Limited proteolysis and stable isotope-labeling mass spectrometry analysis method for monitoring protein conformational changes upon small-molecule binding. Kinases were incubated with vehicle 1% DMSO, ADP, ATP, or AMP-PNP (10 nmol) or inhibitors (1 nmol vemurafenib and sorafenib) for 4 h at 30 °C. Subsequently kinases are partially digested in their native conformation, and the released peptides are labeled using the light form of N-hydroxysuccinimide butyrate (L). Here peptides obtained upon full digestion are labeled using the corresponding heavy form (H). The differential stable isotopelabeled peptides are then mixed in a given ratio (L:H of 100:3 for B-Raf and 100:2 for KSR1 analysis) and analyzed by LC−MS/MS. The ratio between the ion signals of the light and heavy peptides allows the quantification of peptide ion signals. More details are described in the Materials and Methods session. E


Article



Figure 2. Conformational changes in B-Raf kinases induced by small-molecule binding by limited proteolysis coupled to stable isotope labeling MS. Cartoon representation of the RBD and KD of B-RafWT and B-RafV600E to illustrate the changes in abundance of peptides obtained in the limited proteolysis experiments of B-RafWT complexed with small molecules (incubation for 4 h at 30 °C with 10 nmol ADP, ATP, or AMP-PNP) or its inhibitors (1 nmol vemurafenib or sorafenib) by using Pepshell.83 The PDB entries 3NY5 and 3OG784 were used to visualize the RBD and the KD, respectively. Only the detected peptides that showed a statistically significant difference versus the control samples according to Student’s t test (p < 0.05) are indicated and colored as follows. Upper panel: brown: SPQKPIVR (aa 151−158), blue: VFLPNK (aa 159−164), green: TVVPAR (aa 167− 172). Lower panel: pink: DSSDDWEIPDGQITVGQR (aa 444−462), dark green: IGSGFGTVYK (aa 462−473), red: WHGDVAVK (aa 476−483), light blue: LIDIAR (aa 553−558), yellow: QTAQGMDYLHAK (aa 559−570), purple: IGDFGLATVK (aa 592−601), orange: GYLSPDLSK (aa 672− 680).

kinases were active, as both B-RafWT and B-RafV600E converted ATP to ADP. As expected, only an ATP peak was observed for the pseudokinase KSR1 upon preincubation with ATP. In parallel, the kinase activity was tested using a luminescent ADP detection assay (Supplemental Figure S2B). Here again, both the wild-type and mutant B-Raf kinases were active in the presence of ATP, while their activity was inhibited by the inhibitors assayed in this study. These results thus confirm the proper enzymatic activity of the proteins analyzed. We first tested different proteolysis conditions (the amount of starting material, the type of digestion enzyme, the protein/ enzyme ratio, the digestion time, and temperature) and found the clearest indications for B-Raf conformational differences induced by small-molecule binding upon digestion of the B-Raf/ ligand complexes with trypsin at a 1:200 trypsin:B-Raf ratio for 10 min at 30 °C (data not shown) and therefore used these conditions for all further experiments. Note that these limited proteolysis conditions had to be adapted for KSR1 (see Materials and Methods), likely due to the different structural organization and folding of this protein. Thus, optimization of proteolysis conditions is recommended for individual proteins. In fact, this optimization of parameters can be conveniently assessed by examining protein patterns of limited proteolysis via SDS-PAGE.

approach allowed us to monitor conformational changes in kinases induced by binding small molecules, after calculating the ratio of the signal intensities from peptides of the actual sample (light labeled peptides) over the full digested reference sample (heavy labeled). Importantly, this quantification step does not require any further purification, as is needed in other approaches,63,67 meaning that labeled peptides can be directly analyzed and therefore, sample losses and artifacts are minimal. Furthermore, compared with other techniques that explore protein conformational changes, another advantage of our approach is the relatively low amount of protein that is required (

Limited Proteolysis Combined with Stable Isotope Labeling Reveals

Recommend Documents