Article pubs.acs.org/ac
Affinity Capture of Biotinylated Proteins at Acidic Conditions to Facilitate Hydrogen/Deuterium Exchange Mass Spectrometry Analysis of Multimeric Protein Complexes Pernille Foged Jensen,†,‡ Thomas J. D. Jørgensen,‡ Klaus Koefoed,† Frank Nygaard,†,§ and Jette Wagtberg Sen*,† †
Symphogen A/S, Elektrovej Building 375, DK-2800 Lyngby, Denmark Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark
‡
ABSTRACT: Characterization of conformational and dynamic changes associated with protein interactions can be done by hydrogen/deuterium exchange mass spectrometry (HDX-MS) by comparing the deuterium uptake in the bound and unbound state of the proteins. Investigation of local hydrogen/deuterium exchange in heteromultimeric protein complexes poses a challenge for the method due to the increased complexity of the mixture of peptides originating from all interaction partners in the complex. Previously, interference of peptides from one interaction partner has been removed by immobilizing the intact protein on beads prior to the HDX-MS experiment. However, when studying protein complexes of more than two proteins, immobilization can possibly introduce steric limitations to the interactions. Here, we present a method based on the high affinity biotin−streptavidin interaction that allows selective capture of biotinylated proteins even under the extreme conditions for hydrogen/deuterium exchange quenching i.e. pH 2.5 and 0 °C. This biotin−streptavidin capture strategy allows hydrogen/deuterium exchange to occur in proteins in solution and enables characterization of specific proteins in heteromultimeric protein complexes without interference of peptides originating from other interaction partners in the complex. The biotin−streptavidin strategy has been successfully implemented in a model system with two recombinant monoclonal antibodies that target nonoverlapping epitopes on the human epidermal growth factor receptor (EGFR). We present a workflow for biotinylation and characterization of recombinant antibodies and demonstrate affinity capture of biotinylated antibodies under hydrogen/deuterium exchange quench conditions by the biotin−streptavidin strategy.
T
various periods of deuteration, aliquots are withdrawn and the isotopic exchange of amide hydrogens is slowed down by several orders of magnitude by lowering the pH to 2.5 and the temperature to 0 °C (i.e., quenching conditions).8 To minimize the unavoidable loss of deuterium from labeled backbone amides due to back-exchange with protiated solvents it is pivotal that the quench conditions are maintained in all subsequent steps of the analysis. The half-life for back-exchange at quench conditions varies from minutes to hours depending on the amino acid sequence and thus the following steps must also be completed as quickly as possible.9 For investigation of local deuterium uptake the proteins are digested following quench by use of broad-specific acid-stable proteases, typically pepsin. The pepsin digestion generates peptides with overlapping sequences that are subsequently separated with reversed-phase liquid chromatography (LC) at 0 °C prior to MS analysis.10,11
he analytical method amide hydrogen/deuterium exchange mass spectrometry (HDX-MS) probes protein conformation and dynamics,1,2 and it is increasingly being used to characterize protein−protein and protein−ligand interactions.3−5 The basis of this method is that amide hydrogens that are engaged in stable hydrogen bonds in the protein structure are strongly protected against hydrogen exchange with the solvent, while unstructured or very flexible regions undergo rapid exchange.6,7 These properties are central for the utility of HDX-MS to investigate protein binding interfaces as amide hydrogens located within the interface of a protein complex are protected against exchange due to shielding from the solvent and hydrogen bonding. Investigation of the interactions between multiple proteins by HDX-MS is typically done by determining the deuterium uptake for each protein in the presence and absence of its interaction partner(s). In a typical H-to-D exchange experiment, the protein complex is formed in a protiated solvent and isotopic exchange is initiated by dilution into a deuterated solvent. Importantly, exchange is performed at conditions where the proteins are in their native state, i.e. typically with a pH between 6 and 8 and a temperature of 25−37 °C. After © XXXX American Chemical Society
Received: November 27, 2012 Accepted: March 27, 2013
A
dx.doi.org/10.1021/ac303442y | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
Figure 1. Strategies for removal of antibodies in HDX-MS epitope mapping experiments. (A) Affinity capture strategy based on the biotin− streptavidin interaction. (A1) The heteromultimeric protein complex comprised of two antibodies and a single antigen are in solution during the hydrogen/deuterium exchange. (A2) At quench conditions the antigen is released from the protein complex whereas the biotinylated antibodies are captured by a streptavidin resin. (B) Immobilization strategy. (B1) In the binary antibody−antigen complex the antibody is immobilized to a resin prior to hydrogen/deuterium exchange. (B2) At quench conditions the antigen is released from the antibody.
binding interaction and the isotopic labeling. The method is based on affinity capture of biotinylated proteins under HDXMS quench conditions i.e. pH 2.5 and 0 °C using the high affinity biotin−streptavidin interaction (Kd = 10−15 M).22 At quench conditions most protein−protein interactions are disrupted, but here we show that the binding between biotin and streptavidin can be formed rapidly at these harsh conditions. The affinity capture method is illustrated using a ternary protein complex consisting of two recombinant monoclonal antibodies, 992 and 1024, and their target antigen the epidermal growth factor receptor (EGFR). The antibodies constitute the Sym004 antibody mixture, and they bind to EGFR simultaneously.23−25 We demonstrate affinity capture of biotinylated 992 and 1024 from a mixture with EGFR using streptavidin coupled to a resin under quench conditions which allow subsequent LC-MS analysis of EGFR peptides exclusively, without interference from antibody-derived peptides.
Local HDX-MS analysis of protein assemblies with multiple interaction partners is challenging because digestion of all proteins in the complex increase the complexity of the peptide mixtures. The combination of a large number of peptides and short chromatographic gradients result in many coeluting peptides that increase the probability for overlapping isotope patterns that compromise the data analysis even when using ultra high pressure liquid chromatography (UPLC)-based separations.3,12,13 Furthermore, the presence of a large number of peptides from nontarget proteins results in a decreased sequence coverage for the protein of interest. This is a situation which is somewhat similar to the under sampling problem in shotgun proteomics.14 However, unlike proteomics the peak capacity in HDX-MS experiments cannot be increased by longer LC gradients as this will increase the loss of deuterium due to back-exchange to unacceptable levels. The spectral complexity in HDX-MS experiments has been addressed by various approaches, e.g. ion mobility,15 isotopic depletion,16 and using a low D2O content in the exchange solution.17 Another approach to reduce spectral complexity is to immobilize one of the proteins of the complex on a resin before the isotopic exchange reaction is initiated.18−21 This strategy has been used to investigate binary protein complexes for example in epitope mapping experiments where the antigen binds to a covalently immobilized antibody.18 In this way, the antibody and the antigen are easily separated upon quenching and only the antigen is digested by pepsin. However, when investigating a protein with more than one interaction partner the immobilization strategy may prevent the formation of the native multimeric complex due to restricted mobility of the covalently bound proteins. The difference between the affinity capture strategy and the immobilization strategy is illustrated in Figure 1. Here, we present a new strategy to reduce the interference of peptides from nontarget interaction partners in heteromultimeric protein complexes without immobilization during the
■
EXPERIMENTAL SECTION Two recombinant monoclonal chimeric human-mouse IgG1 antibodies 992 and 1024 as well as a recombinant EGFR-Fc fusion homodimer consisting of the extra-cellular domain (ECD) of EGFR and the Fc domain of an IgG1 antibody were produced in CHO cell cultures and formulated in 10 mM citrate and 150 mM NaCl, pH 6 by Symphogen A/S. The theoretical molecular weight of the nonglycosylated protein is approximately 188 kDa. Recombinant ECD EGFR formulated in PBS was purchased from Sino Biological Inc., Beijing, China. The theoretical molecular weight of the nonglycosylated protein is approximately 69 kDa. All other reagents were from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise stated. Biotinylation of Antibodies. The two antibodies 992 and 1024 in 10 mM citrate and 150 mM NaCl, pH 6, were buffer exchanged to PBS, pH 7.2, the protein concentration was B
dx.doi.org/10.1021/ac303442y | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
was set to 30 °C, and agitation during the measurements was 1000 rpm. General Procedure for Affinity Capture of Biotinylated Antibodies at Acidic Conditions Using Streptavidin Agarose. A mixture of proteins containing at least one biotinylated protein at pH 6 was cooled to 0 °C, and pH was lowered to 2.5 by addition of TFA to mimic quenching of amide hydrogen/deuterium exchange. The quenched protein mixture was added to streptavidin agarose (Sigma-Aldrich) in a spin filter and incubated for 1 min at 0 °C. The flow-through was collected following centrifugation at 1500 rcf for 1 min at 0 °C. Capture of Biotinylated Antibodies from Protein Mixture with EGFR Illustrated by SDS-PAGE. A protein mixture of 10 μg EGFR (Sino Biological), 10 μg 992-Biotin, and 10 μg 1024-Biotin (biotinylated with antibody to biotin ratio 1:10) was prepared in 10 mM citrate and 150 mM NaCl, pH 6. The sample was quenched and added to 300 μL streptavidin slurry as described above. Likewise the protein mixture at pH 6 was added to streptavidin as described above without being quenched. The flowthrough of the samples from the streptavidin capture at pH 2.5 and pH 6, and the initial protein mixture were precipitated with 80% acetone for 1 h at −20 °C before being redissolved in LDS and added to a 4−12% Bis Tris Gel run with a MOPS buffer system and a Seeblue plus 2 standard (Invitrogen, Carlsbad, CA, USA). Capture of Biotinylated Antibody Illustrated by Intact Mass LC-MS. A 45 μg portion of biotinylated 1024 (biotinylated with antibody to biotin ratio 1:10) in 10 mM citrate and 150 mM NaCl, pH 6, was quenched and added to 400 μL streptavidin slurry as described above. The starting material and the flow-through were analyzed by intact mass LCMS as described above. LC-MS Analysis of Peptic Digests of Protein Mixture with and without Affinity Capture. A protein mixture of 11 μg EGFR (Sino Biological), 11 μg 992-Biotin, and 11 μg 1024Biotin (biotinylated with antibody to biotin ratio 1:10) was prepared in 10 mM citrate and 150 mM NaCl, pH 6. The sample was quenched and added to 300 μL streptavidin slurry as described above. For LC-MS analysis, the samples were reduced by adding an equal volume of 0.8 M TCEP and 4 M guanidine hydrochloride, pH 2.5, and were injected into a nanoACQUITY UPLC reverse phase system equipped with a 4 mm × 50 mm Omega column (Upchurch Scientific, Oak Harbor, WA, USA) packed with immobilized pepsin (Pierce) in the injection loop where the reduction of disulfides and digestion was done for 5 min. The peptides were desalted for 6 min using a trap column (Agilent Zorbax Bonus-RP microbore C8, 1 mm × 17 mm, 5 μm particle size) with 100% mobile phase A (0.23% FA in water) at a flow rate of 0.3 mL/min and eluted to and separated with a C18 column (Waters Acquity UPLC BEH C18 1.0 mm × 100 mm, 1.7 μm particle size, 130 Å pore size) with a 7 min linear gradient from 8% to 35% mobile phase B (0.23% FA in acetonitrile) at a flow rate of 40 μL/min. The peptides were analyzed by a Waters Synapt G1MS with the following settings m/z range of 300−1500, source temperature 70 °C, desolvation gas temperature 150 °C, capillary voltage 3.5 kV, and sample cone voltage 26 V. Data processing was done in ProteinLynx Global Server version 2.4 (Waters) and Mascot (Matrix Science, London, UK).
adjusted to 2 mg/mL, and the antibodies were biotinylated using the biotin linker Sulfo-NHS-LC-Biotin (Pierce, Rockford, IL, USA) in molar antibody to biotin ratios of 1:2 and 1:10. Biotinylation proceeded for 30 min at room temperature, and biotinylated antibodies were buffer exchanged to 10 mM citrate and 150 mM NaCl, pH 6. Characterization of Biotinylated Antibodies by Intact Mass LC-MS. The mass of the biotinylated antibodies was determined by intact mass LC-MS using an Ultimate 3000 RSLC (Thermo Scientific, Sunnyvale, CA, USA), equipped with a phenyl MassPREP Micro desalting column 2.1 mm × 5 mm, 20 μm particle size, 1000 Å pore size (Waters, Milford, MA, USA) operated at a temperature of 80 °C. Mass analysis was done using a MicrOTOF-Q II (Bruker Daltronik GmbH, Bremen, Germany). Approximately 5 μg of each sample was loaded at 95% mobile phase A (0.1% FA in water) and 5% mobile phase B (0.1% FA in acetonitrile) and eluted by increasing mobile phase B to 90% in 1 min. The mass spectrometer was set to an m/z range of 50−3000, source temperature 180 °C, capillary voltage 4.5 kV, end plate offset −500 V, and in-source collision-induced dissociation energy of 120 eV. Spectra of the multiple charged ions were deconvoluted using the maximum entropy algorithm in the Bruker Data analysis 4.0 software. Characterization of Biotinylated Antibodies by Peptide Mapping LC-MS/MS. Biotinylated and unmodified 992 and 1024 antibodies were buffer exchanged to a 3 M guanidine hydrochloride and 100 mM HEPES, pH 7.6. Disulfide bonds were reduced by addition of 5 nmol DTT per mircogram of protein and incubated at 37 °C for 1 h. Alkylation was done with 12 nmol iodoacetamide per microgram of protein in the dark at room temperature for 15 min, and the reaction was quenched by addition of 3 nmol DTT per microgram of protein. Digestion was performed with trypsin (Promega, Madison, WI, USA) in an enzyme to substrate ratio of 1:20 at 55 °C for 1 h. The peptides were separated by an Ultimate RSLC (Thermo Scientific) equipped with a C18 Acquity UPLC BEH130 column 2.1 mm × 150 mm, 1.7 μm particle size, 130 Å pore size (Waters) and mass analysis was done on a MicrOTOF-Q II (Bruker). Approximately 140 pmol of each sample was loaded at 98% mobile phase A (0.1% TFA in water) and 2% mobile phase B (0.1% TFA in acetonitrile) and eluted by increasing mobile phase B to 40% in 40 min at a flow rate of 0.2 mL/min. The mass spectrometer acquired MS and MS/MS in a data dependent mode in an m/z range of 50−3000, source temperature 200 °C, capillary voltage 4.5 kV, and end plate offset −500 V. Data processing was done in Bruker Dataanalysis 4.0 and Biotools 3.2 with Sequence Editor software where the biotin linker with a mass of 339.5 Da was selected as an optional modification on lysine. Biolayer Interferometry Based EGFR Binding Assay. The binding of biotinylated and unmodified 992 and 1024 to EGFR was determined by a biolayer interferometry based assay using Octet QK384 (Fortebio, Menlo Park, CA, USA). Protein G biosensors (Fortebio) were coated with the EGFR-Fc fusion homodimer diluted in PBS with 0.01% BSA, 0.002% Tween-20, and 0.005% sodium azide to a concentration of 20 μg/mL. Free Protein G binding sites were blocked with an unspecific antibody at a concentration of 200 μg/mL. The EGFR coated sensors were immersed in 20 μg/mL antibody solutions of either biotinylated or unmodified 992 and 1024 for 300 s followed by 1200 s dissociation in PBS with 0.01% BSA, 0.002% Tween-20, and 0.005% sodium azide. The temperature C
dx.doi.org/10.1021/ac303442y | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Article
Figure 2. Optimizing biotinylation of antibodies. Intact mass LC-MS analysis of (A) 992 and (B) 1024 without biotin and biotinylated with a molar antibody to biotin linker ratio of 1:2 and 1:10. The intact antibodies contain one glycosylation site on each heavy chain in the Fc region and the main glycoforms observed are Man5/Man5 (+2434 Da), G0F-GlcNAc/G0F (+2687 Da), G0F/G0F (+2891 Da), G0F/G1F (+3053 Da), G1F/G1F (+3215 Da), and G0F/G2F(+3215 Da).32
■
RESULTS AND DISCUSSION Biotinylation and Characterization of Antibodies. To enable selective capture of the two antibodies 992 and 1024 by the biotin−streptavidin strategy, the antibodies must be labeled with biotin prior to the HDX-MS experiments. Protein biotinylation can be done chemically by use of reactive biotin derivatives that target specific functional groups, e.g. amines (lysine and N-terminus), sulfhydryls (cysteine), carboxyls (Cterminus, aspartic acid, and glutamic acid), imidazoles (histidine), phenols (tyrosine), and aldehydes (carbohydrates).22 Alternatively, site-specific biotinylation can be performed enzymatically in vivo or in vitro using the biotin ligase (BirA) that conjugates biotin to a specific lysine in 15residue sequence tags that can be inserted in the protein N- or C-terminus or in loop regions.26−28 The optimal biotinylation procedure will vary with different protein systems. It is essential that the biotin labeling chemistry does not impact the protein− protein interactions and thus the biotinylation procedure should be carefully selected and evaluated. Here, the antibodies were labeled with a N-hydroxysulfosuccinimide ester of biotin that reacts with the free amino groups in side chains of lysine residues and the protein N-terminus. For antibodies, the antigen binding specificity is determined by the six complementary determining regions (CDRs) located in the variable domains in the heavy and light chains.29 To minimize the probability of biotinylating residues in the CDRs, the labeling was optimized so the antibodies were not extensively biotinylated while the proportion of the unmodified antibodies was as low as possible. Figure 2 shows the deconvoluted mass spectra of the antibodies 992 and 1024 without biotin and biotinylated with a molar antibody to biotin ratio of 1:2 and 1:10. Of the 88 lysines in 992 and 90 lysines in 1024 up to 3 and 6 biotins were attached when biotinylation was done with a molar antibody to biotin ratio of 1:2 and 1:10, respectively.
Table 1 shows the relative distribution of the antibody subpopulations containing different numbers of conjugated Table 1. Relative Distribution of the Biotin Subpopulations of 992 and 1024 Biotinylated with a Molar Antibody to Biotin Linker Ratio of 1:2 and 1:10a number of biotins/IgG molecule
992Biotin 1:2
992-Biotin 1:10
1024Biotin 1:2
1024-Biotin 1:10
0 1 2 3 4 5 6
47% 36% 14% 3% 0% 0% 0%
1% 19% 21% 25% 19% 11% 4%
46% 36% 15% 3% 0% 0% 0%
1% 12% 22% 26% 21% 13% 5%
a The relative distribution between the different subpopulations of the biotinylated antibodies was determined from the peak intensity of one population divided by the sum of the total peak intensity. 1024-Biotin 1:10 is also detected with seven biotins with