Special Issue Article Received: 27 May 2014,
Revised: 10 August 2014,
Accepted: 15 August 2014,
Published online in Wiley Online Library: 9 February 2015
(wileyonlinelibrary.com) DOI: 10.1002/jmr.2418
Epitope characterization of an anti-PD-L1 antibody using orthogonal approaches† Gang Hao, John S. Wesolowski, Xuliang Jiang, Scott Lauder‡ and Vanita D. Sood* The binding of programmed death ligand 1 protein (PD-L1) to its receptor programmed death protein 1 (PD-1) mediates immunoevasion in cancer and chronic viral infections, presenting an important target for therapeutic intervention. Several monoclonal antibodies targeting the PD-L1/PD-1 signaling axis are undergoing clinical trials; however, the epitopes of these antibodies have not been described. We have combined orthogonal approaches to localize and characterize the epitope of a monoclonal antibody directed against PD-L1 at good resolution and with high confidence. Limited proteolysis and mass spectrometry were applied to reveal that the epitope resides in the first immunoglobulin domain of PD-L1. Hydrogen–deuterium exchange mass spectrometry (HDX-MS) was used to identify a conformational epitope comprised of discontinuous strands that fold to form a beta sheet in the native structure. This beta sheet presents an epitope surface that significantly overlaps with the PD-1 binding interface, consistent with a desired PD-1 competitive mechanism of action for the antibody. Surface plasmon resonance screening of mutant PD-L1 variants confirmed that the region identified by HDX-MS is critical for the antibody interaction and further defined specific residues contributing to the binding energy. Taken together, the results are consistent with the observed inhibitory activity of the antibody on PD-L1-mediated immune evasion. This is the first report of an epitope for any antibody targeting PD-L1 and demonstrates the power of combining orthogonal epitope mapping techniques. Copyright © 2015 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web site. Keywords: hydrogen deuterium exchange; mass spectrometry; epitope mapping; monoclonal antibody; mutagenesis; PD-L1
INTRODUCTION
J. Mol. Recognit. 2015; 28: 269–276
* Correspondence to: Vanita D. Sood, EMD Serono Research and Development Institute, Inc. 45A Middlesex Turnpike, Billerica, MA, 02144 USA. E-mail: [email protected] †
‡
This article is published in Journal of Molecular Recognition as part of the Special Issue ‘State-of-the-art B-cell Epitope Discovery’, edited by Dr Yasmina Abdiche, Dr Arvind Sivasubramanian, Dr Jerry Slootstra, Dr Darren Flower and Professor Edouard Nice. Current address: Merrimack Pharmaceuticals, 1 Kendall Square, Cambridge, MA 02139 G. Hao, J. S. Wesolowski, X. Jiang, S. Lauder, V. D. Sood EMD Serono Research and Development Institute, Inc., 45A Middlesex Turnpike, Billerica, MA, 02144, USA
Copyright © 2015 John Wiley & Sons, Ltd.
269
Programmed death ligand 1 protein (PD-L1), also known as cluster of differentiation 274 or B7 homolog 1, is a type I transmembrane protein belonging to the B7 family (Dong et al., 1999; Freeman et al., 2000). PD-L1 binds to its receptor, programmed cell death protein 1 (PD-1), which is expressed in activated T cells, B cells, and macrophages (Ishida et al., 1992; Okazaki et al., 2001). The complexation of PD-L1 and PD-1 exerts immunosuppressive effects by inhibiting T cell proliferation and cytokine production of IL-2 and IFN-γ (Freeman et al., 2000; Carter et al., 2002). The upregulation of the PD-L1/PD-1 pathway is implicated in cancer and chronic viral infections, where tumor cells or viruses evade immune surveillance via PD-L1 overexpression (Iwai et al., 2002; Brown et al., 2003; Kao et al., 2011). Inhibition of PD-1 or PD-L1 restores the attenuated immune response and leads to increased antitumor and anti-viral activities (Hirano et al., 2005; Barber et al., 2006; Velu et al., 2009; Bhadra et al., 2011). Because of the high therapeutic potential, several monoclonal antibodies targeting PD-L1/PD-1 are undergoing phase I and II clinical trials for various cancers and showed efficacy for non-small-cell lung cancer, melanoma, and renal-cell cancer Berger et al., 2008; Brahmer et al., 2012; Lipson et al., 2013). A critical step in therapeutic antibody development is to define the region on the antigen that is recognized by the complementarity determining regions, a process termed epitope mapping. The information is required for elucidating the mechanism of the drug and also for regulatory approval (FDA, 1997). Antibody epitopes can be categorized as linear or conformational (Barlow et al., 1986). A linear epitope is typically a single
short peptide sequence, whereas a conformational epitope comprises discontinuous segments brought into proximity upon protein folding. Linear epitopes can be identified by screening a synthetic peptide library spanning the antigen sequence to identify linear epitopes (Geysen et al., 1984). For a conformational epitope that comprises discontinuous segments brought into proximity upon protein folding, structural elucidation of the antigen–antibody complex by X-ray crystallography provides the most detailed information on the binding interface (Amit et al., 1986; Mylvaganam et al., 1998). This approach, however, can be unsuccessful in spite of good efforts. Other methods have been applied in epitope mapping, but each method alone usually does not provide a clear map of the epitope. Alanine scanning mutagenesis is a well-established method, but this approach is labor intensive, and conformational changes or unfolding induced by mutation remote from the contact region
G. HAO ET AL. can confound interpretation of the results (Benjamin et al., 1984). Limited proteolysis monitored by mass spectrometry has also been applied to conformational epitope mapping (Jemmerson and Paterson, 1986). Identification of regions protected from proteolysis by the bound antibody reveals the location of the epitope; however, the resolution of the technique is low because of the sporadic distribution of cleavage sites. Hydrogen–deuterium exchange (HDX) mass spectrometry (MS) has emerged as a powerful tool for studying protein–protein interactions (Hoofnagle et al., 2003; Englander, 2006; Wales and Engen, 2006; Chalmers et al., 2011; Bobst and Kaltashov, 2012; Zhang et al., 2013). In an HDX epitope mapping experiment, the antigen is incubated either alone or in complex with an antibody in deuterium oxide (D2O) to allow the exchange of the backbone amide hydrogen atoms for deuterium. The antigen surface that is excluded from solvent by bound antibody (the epitope) displays a reduced amide exchange rate that can be sensitively detected by MS. The technique has been recently applied to several antibodies against proteins including cytochrome c, thrombin, interleukin-1β, Fas ligand, and cashew food allergen (Baerga-Ortiz et al., 2002; Lu et al., 2005; Coales et al., 2009; Obungu et al., 2009; Zhang et al., 2011). A limitation of the standard HDX-MS method is that it is still challenging to achieve single-residue resolution for the epitope. While electron transfer dissociation has enabled the monitoring of backbone dynamics at single-residue resolution (Huang et al., 2011; Rand et al., 2012), to our knowledge, it has not been applied to the study of antibody–antigen interactions. In our effort to identify the epitope of a specific antibody targeting PD-L1, we attempted to crystallize the antibody/PD-L1 complex, but the complex proved recalcitrant to crystallization. In order to localize and characterize the epitope of anti-PD-L1 at good resolution and with high confidence and to elucidate the mechanism of the antibody, we applied a combination of orthogonal approaches, namely limited proteolysis, HDX-MS, and mutational scanning. Here, we report the first epitope for an antibody against this high-potential therapeutic target.
MATERIALS AND METHODS PD-L1, anti-PD-L1 IgG, and Fab production Anti-PD-L1 IgG was expressed in Chinese hamster ovary cells and purified by protein A affinity chromatography (MabSelect, GE Healthcare). Anti-PD-L1 Fab and the extracellular domain (ECD) of human PD-L1 (residues 19–238 of the open reading frame) were expressed as 6-His-tagged fusion proteins in human embryonic kidney 293 cells. His-tagged proteins were purified by nickel affinity chromatography. All proteins were buffer exchanged with phosphate-buffered saline (PBS) at pH 7.2. The IgG was prepared at a stock concentration of 10 μg/μl, the Fab at 4.34 μg/μl, and the PD-L1 at 1.90 μg/μl. Limited proteolysis
270
PD-L1 (2.5 μl) was mixed with 4.5 μl Fab and 2 μg trypsin (Promega, Madison, WI, USA), diluted to a total volume of 19 μl with PBS, and incubated at 37°C for 2 h. The trypsin digestion was terminated by addition of 1 μl of trifluroacetic acid (final concentration 0.1%) and analyzed using liquid chromatography MS (LC-MS) as described in the succeeding texts. For the control sample, the same amounts of PD-L1 and Fab were digested separately and mixed prior to LC-MS analysis.
wileyonlinelibrary.com/journal/jmr
HDX PD-L1 (2.5 μl) was incubated with or without 4.5 μl Fab at 4°C for 5 min. The preincubated complex or PD-L1 alone was diluted to a final volume of 50 μl with PBS prepared with D2O, then further incubated at 4°C to allow deuterium exchange. The final concentrations in the exchange reaction were 3.65 and 7.94 μM for PD-L1 and Fab, respectively. A molar excess of Fab and concentrations of both components much higher than the KD of the complex (Table 1) were used to ensure saturation binding of PD-L1. Four degrees Celsius was chosen because of precipitation of protein observed upon extended incubation at high concentrations at ambient temperature (data not shown). Aliquots during exchange were taken at 30 s, 2, 10, and 60 min. At each time point, the exchange was quenched using 50 μl of 0.2 M sodium phosphate with 0.2 M tris(2-carboxyethyl)phosphine and 0.4 M guanidine hydrochloride at pH 2.5. Two microliters of pepsin (2 mg/ml) was added to the quenched samples, and pepsin digestion on the chilled mixture was carried out in a thermomixer set to 20°C for 2 min. The digested samples were immediately applied to an LC-MS system and analyzed as described in the succeeding texts. LC-MS The digested samples were analyzed on an LC-MS system comprised of an Acquity UPLC (Waters, Milford, MA, USA) and a microTOF-Q (Bruker, Billerica, MA, USA). The solvent mixture, sample loop, LC column, and the connecting tubing were immersed in a circulating chilled water bath kept at 0°C during the sample run. The peptides were separated on a 1.0 × 50-mm ZORBAX C18 column (Agilent, Palo Alto, CA, USA) at a flow rate of 100 μl/min over a gradient from 8 to 35% solvent B (0.1% Table 1. Summary of PD-L1 mutant binding to anti-PD-L1 Mutation
ΔΔGmut (kcal/mol)
Wild type I54A I54K Y56A Y56K E58A E60A D61A Q66A V68A V68R R113A M115A
— 1.28 0.62 >4 >5 1.90 1.45 >5 0.86 0.02 0.55 1.53 0.97
KD (nM)
T1/2 (°C)
0.55 ± 0.21 0.06 ± 0.09 1.57 ± 0.19 > 1 μM >4 μM 13.58 ± 0.59 6.32 ± 0.44 NB 2.35 ± 0.23 0.57 ± 0.04 1.37 ± 0.05 7.22 ± 0.26 2.79 ± 0.17
60.1 56.6 58.8 59.1 56.9 55.8 55.8 51.2 61.1 61.4 58.2 58.0 51.2
Point mutants of PD-L1 were compared with wild-type PD-L1 antigen for antibody binding. A kinetic SPR study was used to determine the equilibrium dissociation constant (KD) and the change in the Gibbs free energy of binding of mutant relative to wild-type PD-L1 (ΔΔGmut). Mutants at Y56 and D61 had such a low affinity that the KD could not be accurately measured, and the minimum KD and ΔΔGmut are given instead. The temperature midpoint (T1/2) of FMTU is given for the wild-type and mutant proteins. For KD, the mean and standard deviation is given where possible. NB, no binding.
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
EPITOPE OF AN ANTI-PD-L1 ANTIBODY formic acid in acetonitrile) in 10 min. The same digestion conditions and the LC-MS step were applied to a mixture of fully deuterated peptide standards (Genscript, Piscataway, NJ, USA) to determine the overall deuterium recovery of the method. The average deuterium recovery was about 70% for the standards. The mass spectra were collected in automatic MS/MS mode from 200 to 2000 amu to identify sequence coverage of the nondeuterated sample. Mascot (Matrix Science, Boston, MA, USA) was used to search against the PD-L1 sequence. A mass accuracy of 0.1 Da was used for the precursor ions and 0.2 Da for the product ions. The score cutoff used to identify positive hits was 20. To measure the deuterium content of the exchanged samples, the mass spectra were recorded in MS mode. The raw MS data were processed by HDExaminer (Sierra Analytics, Modesto, CA, USA) to calculate the deuterium incorporation. An extracted ion chromatogram was generated for each identified peptide, and the deuterium incorporation was calculated based on the increase in centroid mass. Mutant PD-L1 production Plasmids encoding mutant PD-L1 were constructed using QuikChange mutagenesis (Agilent Technologies) and expressed and purified as described for wild-type PD-L1 in the preceding texts. The mutant proteins were analyzed by analytical size exclusion chromatography (SEC) on a Bio SEC-3 column (4.6 × 300 mm, 300 Å, 3 μm; Agilent) and compared with wild type; all mutants were eluted at the same volume as the wild type. The thermal stability of the wild-type and mutant proteins were measured by fluorescence monitored thermal unfolding (FMTU) on a quantitative PCR instrument (Applied Biosystems). Protein (95 μl) at 0.2 mg/ml was mixed with 5 μl of SYPRO Orange (5000× concentrate, Life Technologies), and fluorescence at 570 nM was monitored as a function of temperature. The data were fit to the Boltzmann sigmoidal equation to obtain the temperature at half-maximal fluorescence.
Surface plasmon resonance Kinetic binding experiments were performed on a Biacore 4000 (GE Healthcare). Purified goat antihuman IgG Fc (Jackson ImmunoResearch Laboratories) was immobilized on a CM5 chip (GE Healthcare) using amine coupling chemistry. Anti-PD-L1 IgG was captured by the antihuman IgG. Binding experiments were carried out in 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-buffered saline (20 mM HEPES, 150 mM NaCl, 3.4 mM EDTA, and 0.05% P20 surfactant) at 25°C. Kinetic data were collected at six different concentrations of wild-type or mutant PD-L1 analyte: 100, 50, 25, 12.5, 6.3, and 0 nM. For the two hot spot residues (Y56 and D61), data were collected at 1 μM to 62.5 nM. The analyte was allowed to associate for 3 min followed by a dissociation step in buffer for 10 min at 30-μl/min flow rate. The surface was regenerated between each concentration of analyte, and fresh antibody was captured. The data were fit to a 1:1 Langmuir binding model with the BIA evaluation software (GE Healthcare). Kinetic rate constants were determined from the fits of the association and dissociation phases, and the KD was derived from the ratio of these constants.
RESULTS AND DISCUSSION We sought to determine the epitope of anti-PD-L1 to gain insight into its mechanism of action using two complementary approaches in parallel: HDX-MS and site-directed mutagenesis. We initially attempted to cocrystallize the complex of human PD-L1 and the antigen binding domain of antiPD-L1. While both individual components could be crystallized, the complex did not crystallize. Using HDX-MS and mutagenesis, however, we were able to determine the epitope to residue-level resolution.
J. Mol. Recognit. 2015; 28: 269–276
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jmr
271
Figure 1. The epitope for anti-PD-L1 is located on the immunoglobulin domain I of human PD-L1, as revealed by limited proteolysis and mass spectrometry. (A) Total ion chromatograms of PD-L1 subjected to limited proteolysis with trypsin, unbound control (upper panel), or in complex with the Fab (lower panel). (B) Peptides generated from PD-L1 by limited proteolysis shown on the sequence of the ECD. Bold with dashed underscore: peptides generated in PD-L1 only in the absence of Fab. Underscored: peptides generated both in the presence and absence of Fab binding.
G. HAO ET AL. Limited proteolysis Human PD-L1 ECD contains two immunoglobulin-like domains (Lin et al., 2008). Limited proteolysis was used to rapidly establish an approximate location of the epitope. PD-L1 was digested with trypsin under nondenaturing condition in the presence or absence of anti-PD-L1 Fab. As shown in Figure 1A, several peaks present in the control sample without Fab disappeared in the complex sample, corresponding to the peptides protected from proteolytic cleavage by Fab binding. Importantly, the protected region was located in the first immunoglobulin domain of PD-L1 (Figure 1B), where PD-1 engages with PD-L1 (Lin et al., 2008). Binding in a region that would block receptor–ligand interaction is consistent with an immunostimulatory mechanism of action for the antibody. While the entire length of the antigen was scanned in the subsequent HDX experiment, the information here was helpful for optimizing the peptide digestion conditions as described in the next section.
Optimization of pepsin digestion of PD-L1 To improve the resolution of the epitope from domain to peptidelevel resolution using HDX, it was necessary to optimize the digestion conditions to achieve good sequence coverage of the antigen. In particular, generation of overlapping peptides for each region is helpful as each peptide can be separately analyzed for HDX protection, increasing the confidence in observed exchange rates and potentially increasing the resolution of the epitope to a smaller region. Initially, we used full-length IgG in complex with PD-L1 for
epitope mapping by HDX. However, digestion of the complex created numerous peptide fragments from the IgG that complicated the data analysis of the antigen fragments. While one could immobilize the IgG and remove it before digestion, we instead simplified the process by using the Fab portion of the antibody in place of fulllength IgG. Figure 2A shows the total ion chromatograms for the pepsin digestion of PD-L1 alone and in complex with Fab, which were similar. The absence of peaks corresponding to peptides from the Fab suggests that it remained largely intact during digestion, consistent with the published observations on Fab resistance to proteolysis (Wilson et al., 2002), allowing us to analyze peptides from the antigen without removing Fab prior to digestion. We found that PD-L1 was efficiently digested in immunoglobulin domain I, which generated numerous overlapping peptides (Figure 2B). Interestingly, despite the known lack of specificity of pepsin, domain II was not covered by as many overlapping peptides as domain I but rather was covered by a few longer peptides. However, because limited proteolysis showed that domain I likely contained the epitope (refer to preceding texts), coverage of domain II by fewer overlapping peptides was not a concern for identifying the epitope. Sequence coverage (95%) was achieved by MS/MS analysis and is shown schematically in Figure 2B. One region in domain I absent from the MS/MS analysis (residues 33–39) is known to be modified by N-linked glycosylation and is likely to be protected from proteolytic digestion (Rehm, 2006). Although antibodies do not often recognize carbohydrates, there are several examples of carbohydrate– antibody interactions, and it was a priori a possibility that if the
272
Figure 2. Epitope mapping of anti-PD-L1 by HDX mass spectrometry. (A) Total ion chromatograms of PD-L1 peptic digestion with and without Fab. (B) Sequence coverage of human PD-L1 in peptic digestion is shown by boxes under the sequence. Peptides showing reduced exchange rate upon Fab binding are annotated with black boxes. (C) Example deuterium uptake plots for segment 54–66 and 112–122, which contain 11 and 9 potentially exchangeable backbone amide protons, respectively, not including the terminal amides that exchange at a much faster timescale (Bai et al., 1993). Blue diamonds, PD-L1 alone; brown squares, PD-L1 in complex with Fab. Refer to Supplemental Figure S1 for all plots. (D) Example mass spectra showing isotope distribution of exchanged peptides in PD-L1 alone (top panel), PD-L1 in complex with Fab (middle panel), and control with no heavy water (lower panel). (E) The crystal structure (Lin et al., 2008) of human PD-L1 immunoglobulin domain I rendered as a molecular surface (blue). The two epitope segments with strong and modest protection are in magenta and light pink, respectively. The PD-1 ligand is rendered in red cartoons. This and subsequent structural figures were generated using PyMol (http://www.pymol.org/) and the coordinates were taken from 3BIK (Lin et al., 2008).
wileyonlinelibrary.com/journal/jmr
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
EPITOPE OF AN ANTI-PD-L1 ANTIBODY epitope were located in this region, we would be unable to identify it. However, in contrast to the high affinity and high specificity of our antibody, the known carbohydrate–antibody interactions are of low affinity and low specificity (Manimala et al., 2007), so it seemed unlikely that the epitope would be located in a glycosylated peptide. Nevertheless, to exclude the possibility of a carbohydrate epitope and to identify the epitope we performed HDX-MS using the optimized pepsin digestion conditions. HDX
J. Mol. Recognit. 2015; 28: 269–276
Mutational scanning of PD-L1 binding to antibody To further corroborate the results of HDX-MS, to obtain a finer, residue-level mapping of the epitope, and to obtain information on the energetic hot spots of binding within the epitope, we applied an orthogonal technique to complement the HDX-MS data. We selected solvent-exposed residues within and around the epitope region for mutation. Alanine mutants were generated and were evaluated for the binding affinity to the antibody using SPR. In some cases, mutations to nonalanine mutants were also tested to obtain additional information. Prior to the SPR analysis, the mutants were analyzed using analytical SEC and FMTU to ensure that the mutation did not lead to significant conformational change that could confound interpretation of binding results. All epitope mutants displayed SEC retention time comparable with the wild type (Supplemental Figure S2) and a half-maximal temperature of FMTU similar to native or at most 10°C lower than native, as well as similar fluorescence at the folded baseline (Figure 3 and Table 1). These two sets of observations indicate that below 40°C, the native structure was not significantly perturbed by the mutations and allowed straightforward interpretation of subsequent binding experiments that were performed at room temperature. A total of 48 PD-L1 mutants were tested for binding to anti-PD-L1 using SPR. The analysis identified a number of residues that display reduced binding affinity upon substitution to alanine, as well as two residues that slightly decreased affinity upon mutation to larger residues while maintaining or increasing affinity upon alanine
Figure 3. Fluorescence monitored thermal melt analysis of PD-L1 mutant proteins. Wild-type and mutant PD-L1 proteins were heated in the presence of a fluorophore to monitor exposure of the hydrophobic core upon denaturation of the protein. Most mutants displayed similar denaturation profiles to the wild type, while D61A and M115A denatured at somewhat lower temperature. All mutant and wild-type proteins had a similar folded baseline at room temperature. Colors correspond to cartoon in Figure 5.
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jmr
273
The HDX experiment was carried out for PD-L1 alone or in complex with Fab for 0.5, 2, 10, and 60 min. The exchange was quenched and subjected to proteolytic digestion and LC-MS analysis (Figure 2A). The deuterium incorporation was calculated based on the centroid mass shift of the exchanged sample relative to the nonexchanged control, and an example mass spectrum is shown in Figure 2C. Deuterium incorporation plotted as a function of time revealed that most peptides showed no difference in exchange rate between PD-L1 alone or in complex with Fab (Supplemental Figure S1). The exceptions to this were two regions consisting of residues 53–66, which displayed strong protection in the presence of Fab (>2-Da mass shift, Figure 2D left panel) and residues 112–122, which displayed modest protection (>0.5-Da mass shift, Figure 2D right panel). The results of the HDX epitope mapping are summarized in Figure 2B. In the case of the peptide comprising residues 112–122, a more precise definition of the subregion protected by Fab was precluded because of the lack of availability of overlapping peptides. However, the small number of differentially exchanging hydrogen atoms in peptide 112–122 is consistent with only a subregion of this peptide being protected by bound Fab (refer to mutational analysis, in the succeeding texts). In contrast, for the region consisting of residues 53–66, shorter peptides comprised of its N and C-terminal regions also showed strong protection from deuterium exchange, providing independent corroboration of the protection by the Fab in this region (Supplemental Figure S1). These two segments were assigned as the putative epitope regions on PD-L1. Inspection of the crystal structure of human PD-L1 ( Lin et al., 2008) indicates that although the two peptides are far apart in the primary sequence, they are proximal in the three-dimensional structure, constituting a single patch on the surface of the antigen (Figure 2E). Furthermore, this region significantly overlaps with the binding footprint with PD-1 (Figure 2E, and Lin et al., 2008), consistent with an inhibitory PD-1 competitive mechanism of activity of the anti-PD-L1 monoclonal antibody. The mechanism for the reduced exchange rate from Fab binding could be because of solvent exclusion, stabilization of PD-L1 by Fab binding, or a combination of both these mechanisms. From other studies comparing the epitope determined from HDX and X-ray crystallography (Gerhardt et al., 2009; Malito et al., 2013), it seems that both mechanisms may play a role and that epitopes tend to colocalize with observed areas of protection from HDX. The exchange behavior also provides insights into to the dynamics underlying the observed exchange rate. There are two types of kinetics for HD exchange. In the EX1 regime, the chemical exchange rate is much greater than the protein unfolding rate or, in the case of a complex, the rate of dissociation. In the EX1 regime, the exchange profile thus displays a bimodal distribution (Brock, 2012; Percy et al., 2012; Jaswal, 2013). In the EX2 regime, the amide is exposed transiently to the solvent upon rapid
breathing of the local environment, which is faster than the intrinsic chemical exchange rate. The exchange profile for the EX2 behavior is generally Gaussian, as observed here for the PD-L1 peptides (Figure 2C), supporting an EX2 mechanism featuring rapid breathing and local unfolding, which is reduced upon Fab binding. In the context of native protein–protein interactions, the observed protection from exchange is consistent with binding and stabilizing the epitope, as well as excluding solvent from the identified epitope regions (Brock, 2012; Percy et al., 2012; Jaswal, 2013).
G. HAO ET AL.
Figure 4. SPR analysis of PD-L1 mutant protein binding to antibody. Example sensorgrams for selected PD-L1 mutants and wild type are shown. Anti-PD-L1 antibody was captured and wild-type or mutant antigen was bound to the captured antibody. The antigen was titrated down from 100 nM, except for the hot spot mutations (Y56A, Y56K, and D61) that were titrated from 1 μM down to increase the poor binding signal. Even at 1 μM, no detectable binding could be observed for D61A. Y56A displayed variable but low-affinity binding in independent experiments, and Y56K did not display enough binding to accurately measure KD. All other wild-type and mutant proteins could be fit to the Langmuir equation and the KD derived from the kinetic constants.
substitution (Table 1 and Figure 4). The equilibrium dissociation constant (KD) and the change in binding free energy compared with wild-type PD-L1 (ΔΔG) for these mutants are summarized in Table 1. Two residues residing on the strongly protected segment identified by HDX-MS, Y56 and D61, strongly or completely disrupt the binding upon mutation to alanine (Figure 4) and thus represent the binding hot spots of the epitope. This suggests that the hot spots Y56 and D61 engage in significant hydrogen bonding and/or van der Waals interactions with the antibody. Several other residues on the same segment, as well as two residues (R113 and M115) residing on the modestly protected segment, also display a measurable although more modest contribution to the binding energy (Figure 4 and Table 1). It is also interesting to note that residues I54 and V68, which are located on the periphery of the protected region, slightly destabilize the binding when mutated to lysine or arginine but have no effect or even stabilize the binding when mutated to alanine. Taken together, the alanine and other mutations at these two residues further support the localization of the epitope by HDX-MS and the binding hot spots by demonstrating that introducing long charged side chains close to the putative epitope can reduce antibody binding, presumably by steric hindrance or charge repulsion. Identification of the epitope of anti-PD-L1 antibody by orthogonal techniques
274
Figure 5 summarizes the results by mapping the epitope residues that contribute to the antibody binding energy and are
wileyonlinelibrary.com/journal/jmr
Figure 5. Summary of epitope mapping by HDX and mutational scanning. The two epitope segments with strong and modest protection identified by HDX are in magenta and light pink, respectively. Epitope residues important for binding affinity, as determined by mutational scanning, are mapped on the crystal structure of domain I of human PD-L1 (Lin et al., 2008). Residues are colored according to the energetic contribution. Red, >3 kcal/mol; orange, >0.7 kcal/mol; green, destabilizes binding only upon nonalanine mutation.
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
EPITOPE OF AN ANTI-PD-L1 ANTIBODY protected from HD exchange on the crystal structure of PD-L1. The mutational analysis, which probes the role of side chains in binding, identified several residues that contribute energetically to antibody binding. These residues are all located in the same structurally contiguous region identified by HDX, consistent with the antibody engaging with the side chains in this region of the antigen and excluding solvent from the entire surface patch. Binding of the Fab may also stabilize PD-L1 in this region, decreasing the rate of “breathing” and reducing exchange by this additional mechanism. The epitope for the antibody overlaps significantly with the binding site of PD-1, consistent with the inhibitory activity of the antibody. Collectively, our data demonstrate that the combination of HDX-MS and site-directed
mutagenesis provides a viable approach for epitope mapping of antigen–antibody interactions. The workflow can be generically applied to other systems where a crystal structure of the interaction complex is not attainable and was applied here to elucidate for the first time an epitope for an antibody targeting PD-L1.
Acknowledgements We thank Steven Degon for Fab purification, Syngene International Ltd. for purification of mutant antigen protein, Paul Towler for Fab crystallization, and Roland Keller for critical review of the manuscript.
REFERENCES
J. Mol. Recognit. 2015; 28: 269–276
Englander SW. 2006. Hydrogen exchange and mass spectrometry: A historical perspective. J. Am. Soc. Mass Spectrom. 17: 1481–1489. FDA. 1997. Points to Consider in the Manufacture and Testing of Monoclonal Antibody Products for Human Use. http://www.fda.gov/downloads/ biologicsbloodvaccines/guidancecomplianceregulatoryinformation/ otherrecommendationsformanufacturers/ucm153182.pdf: U. S. Department of Health and Human Services Food and Drug Administration Center for Biologics Evaluation and Research. Freeman GJ, Long AJ, Iwai Y, Bourque K, Chernova T, Nishimura H, Fitz LJ, Malenkovich N, Okazaki T, Byrne MC, Horton HF, Fouser L, Carter L, Ling V, Bowman MR, Carreno BM, Collins M, Wood CR, Honjo T. 2000. Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member leads to negative regulation of lymphocyte activation. J. Exp. Med. 192: 1027–1034. Gerhardt S, Abbott WM, Hargreaves D, Pauptit RA, Davies RA, Needham MR, Langham C, Barker W, Aziz A, Snow MJ, Dawson S, Welsh F, Wilkinson T, Vaugan T, Beste G, Bishop S, Popovic B, Rees G, Sleeman M, Tuske SJ, Coales SJ, Hamuro Y, Russell C. 2009. Structure of IL-17A in complex with a potent, fully human neutralizing antibody. J. Mol. Biol. 394: 905–21. Geysen HM, Meloen RH, Barteling SJ. 1984. Use of peptide synthesis to probe viral antigens for epitopes to a resolution of a single amino acid. Proc. Natl. Acad. Sci. USA 81: 3998–4002. Hirano F, Kaneko K, Tamura H, Dong H, Wang S, Ichikawa M, Rietz C, Flies DB, Lau JS, Zhu G, Tamada K, Chen L. 2005. Blockade of B7-H1 and PD-1 by monoclonal antibodies potentiates cancer therapeutic immunity. Cancer Res. 65: 1089–1096. Hoofnagle AN, Resing KA, Ahn NG. 2003. Protein analysis by hydrogen exchange mass spectrometry. Annu. Rev. Biophys. Biomol. Struct. 32: 1–25. Huang RY, Garai K, Frieden C, Gross ML. 2011. Hydrogen/deuterium exchange and electron-transfer dissociation mass spectrometry determine the interface and dynamics of apolipoprotein E oligomerization. Biochemistry 50: 9273–82. Ishida Y, Agata Y, Shibahara K, Honjo T. 1992. Induced expression of PD-1, a novel member of the immunoglobulin gene superfamily, upon programmed cell death. EMBO J. 11: 3887–3895. Iwai Y, Ishida M, Tanaka Y, Okazaki T, Honjo T, Minato N. 2002. Involvement of PD-L1 on tumor cells in the escape from host immune system and tumor immunotherapy by PD-L1 blockade. Proc. Natl. Acad. Sci. USA 99: 12293–12297. Jaswal SS. 2013. Biological insights from hydrogen exchange mass spectrometry. Biochim. Biophys. Acta 1834: 1188–1201. Jemmerson R, Paterson Y. 1986. Mapping epitopes on a protein antigen by the proteolysis of antigen-antibody complexes. Science 232: 1001–1004. Kao C, Oestreich KJ, Paley MA, Crawford A, Angelosanto JM, Ali MA, Intlekofer AM, Boss JM, Reiner SL, Weinmann AS, Wherry EJ. 2011. Transcription factor T-bet represses expression of the inhibitory receptor PD-1 and sustains virus-specific CD8+ T cell responses during chronic infection. Nat. Immunol. 12: 663–671. Lin DY, Tanaka Y, Iwasaki M, Gittis AG, Su HP, Mikami B, Okazaki T, Honjo T, Minato N, Garboczi DN. 2008. The PD-1/PD-L1 complex resembles the antigen-binding Fv domains of antibodies and T cell receptors. Proc. Natl. Acad. Sci. USA 105: 3011–3016.
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jmr
275
Amit AG, Mariuzza RA, Phillips SE, Poljak RJ. 1986. Three-dimensional structure of an antigen antibody complex at 2.8 A resolution. Science 233: 747–753. Baerga-Ortiz A, Hughes CA, Mandell JG, Komives EA. 2002. Epitope mapping of a monoclonal antibody against human thrombin by H/Dexchange mass spectrometry reveals selection of a diverse sequence in a highly conserved protein. Protein Sci. 11: 1300–1308. Bai Y, Milne JS, Mayne L, Englander SW. 1993. Primary structure effects on peptide group hydrogen exchange. Struct. Funct. Genet. 17: 75. Barber DL, Wherry EJ, Masopust D, Zhu B, Allison JP, Sharpe AH, Freeman GJ, Ahmed R. 2006. Restoring function in exhausted CD8 T cells during chronic viral infection. Nature 439: 682–687. Barlow DJ, Edwards MS, Thornton JM. 1986. Continuous and discontinuous protein antigenic determinants. Nature 322: 747–748. Benjamin DC, Berzofsky JA, East IJ, Gurd FR, Hannum C, Leach SJ, Margoliash E, Michael JG, Miller A, Prager EM. 1984. The antigenic structure of proteins: a reappraisal. Annu. Rev. Immunol. 2: 67–101. Berger R, Rotem-Yehudar R, Slama G, Landes S, Kneller A, Leiba M, KorenMichowitz M, Shimoni A, Nagler A. 2008. Phase I safety and pharmacokinetic study of CT-011, a humanized antibody interacting with PD-1, in patients with advanced hematologic malignancies. Clin. Cancer Res. 14: 3044–3051. Bhadra R, Gigley JP, Weiss LM, Khan IA. 2011. Control of Toxoplasma reactivation by rescue of dysfunctional CD8+ T-cell response via PD1-PDL-1 blockade. Proc. Natl. Acad. Sci. USA 108: 9196–9201. Bobst CE, Kaltashov IA. 2012. Localizing flexible regions in proteins using hydrogen deuterium exchange mass spectrometry. Methods Mol. Biol. 896: 375–85. Brahmer JR, Tykodi SS, Chow LQ, Hwu WJ, Topalian SL, Hwu P, Drake CG, Camacho LH, Kauh J, Odunsi K, Pitot HC, Hamid O, Bhatia S, Martins R, Eaton K, Chen S, Salay TM, Alaparthy S, Grosso JF, Korman AJ, Parker SM, Agrawal S, Goldberg SM, Pardoll DM, Gupta A, Wigginton JM. 2012. Safety and activity of anti-PD-L1 antibody in patients with advanced cancer. N. Engl. J. Med. 366: 2455–2465. Brock A. 2012. Fragmentation hydrogen exchange mass spectrometry: a review of methodology and applications. Protein Expr. Purif. 84: 19–37. Brown JA, Dorfman DM, Ma FR, Sullivan EL, Munoz O, Wood CR, Greenfield EA, Freeman GJ. 2003. Blockade of programmed death1 ligands on dendritic cells enhances T cell activation and cytokine production. J. Immunol. 170: 1257–1266. Carter L, Fouser LA, Jussif J, Fitz L, Deng B, Wood CR, Collins M, Honjo T, Freeman GJ, Carreno BM. 2002. PD-1:PD-L inhibitory pathway affects both CD4(+) and CD8(+) T cells and is overcome by IL-2. Eur. J. Immunol. 32: 634–643. Chalmers MJ, Busby SA, Pascal BD, West GM, Griffin PR. 2011. Differential hydrogen/deuterium exchange mass spectrometry analysis of protein-ligand interactions. Expert Rev. Proteomics 8: 43–59. Coales SJ, Tuske SJ, Tomasso JC, Hamuro Y. 2009. Epitope mapping by amide hydrogen/deuterium exchange coupled with immobilization of antibody, on-line proteolysis, liquid chromatography and mass spectrometry. Rapid Commun. Mass Spectrom. 23: 639–647. Dong H, Zhu G, Tamada K, Chen L. 1999. B7-H1, a third member of the B7 family, co-stimulates T-cell proliferation and interleukin-10 secretion. Nat. Med. 5: 1365–1369.
G. HAO ET AL. Lipson EJ, Sharfman WH, Drake CG, Wollner I, Taube JM, Anders RA, Xu H, Yao S, Pons A, Chen L, Pardoll DM, Brahmer JR, Topalian SL. 2013. Durable cancer regression off-treatment and effective reinduction therapy with an anti-PD-1 antibody. Clin. Cancer Res. 19: 462–468. Lu J, Witcher DR, White MA, Wang X, Huang L, Rathnachalam R, Beals JM, Kuhstoss S. 2005. IL-1beta epitope mapping using site-directed mutagenesis and hydrogen-deuterium exchange mass spectrometry analysis. Biochemistry 44: 11106–11114. Malito E, Faleri A, Lo Surdo P, Veggi D, Maruggi G, Grassi E, Cartocci E, Bertoldi I, Genovese A, Santini L, Romagnoli G, Borgogni E, Brier S, Lo Passo C, Domina M, Castellino F, Felici F, van der Veen S, Johnson S, Lea SM, Tang CM, Pizza M, Savino S, Norais N, Rappuoli R, Bottomley MJ, Masignani V. 2013. Defining a protective epitope on factor H binding protein, a key meningococcal virulence factor and vaccine antigen. Proc. Natl. Acad. Sci. USA 110: 3304–3309. Manimala JC, Roach TA, Li Z, Gildersleeve JC. 2007. High-throughput carbohydrate microarray profiling of 27 antibodies demonstrates widespread specificity problems. Glycobiology 17: 17C-23C. Mylvaganam SE, Paterson Y, Getzoff ED. 1998. Structural basis for the binding of an anti-cytochrome c antibody to its antigen: crystal structures of FabE8-cytochrome c complex to 1.8 A resolution and FabE8 to 2.26 A resolution. J. Mol. Biol. 281: 301–322. Obungu VH, Gelfanova V, Rathnachalam R, Bailey A, Sloan-Lancaster J, Huang L. 2009. Determination of the mechanism of action of antiFasL antibody by epitope mapping and homology modeling. Biochemistry 48: 7251–7260. Okazaki T, Maeda A, Nishimura H, Kurosaki T, Honjo T. 2001. PD-1 immunoreceptor inhibits B cell receptor-mediated signaling by recruiting src homology 2-domain-containing tyrosine phosphatase 2 to phosphotyrosine. Proc. Natl. Acad. Sci. USA 98: 13866–13871.
Percy AJ, Rey M, Burns KM, Schriemer DC. 2012. Probing protein interactions with hydrogen/deuterium exchange and mass spectrometry-a review. Anal. Chim. Acta 721: 7–21. Rand KD, Pringle SD, Morris M, Brown JM. 2012. Site-specific analysis of gas-phase hydrogen/deuterium exchange of peptides and proteins by electron transfer dissociation. Anal. Chem. 84: 1931–40. Rehm H. 2006. Glycoproteins. In Protein biochemistry and proteomics, Rehm H (Ed.) 4th ed. Access Online via Elsevier: Elsevier Science: New York, NY. Velu V, Titanji K, Zhu B, Husain S, Pladevega A, Lai L, Vanderford TH, Chennareddi L, Silvestri G, Freeman GJ, Ahmed R, Amara RR. 2009. Enhancing SIV-specific immunity in vivo by PD-1 blockade. Nature 458: 206–210. Wales TE, Engen JR. 2006. Hydrogen exchange mass spectrometry for the analysis of protein dynamics. Mass Spectrom. Rev. 25: 158–170. Wilson DS, Wu J, Peluso P, Nock S. 2002. Improved method for pepsinolysis of mouse IgG(1) molecules to F(ab’)(2) fragments. J. Immunol. Methods 260: 29–36. Zhang Q, Willison LN, Tripathi P, Sathe SK, Roux KH, Emmett MR, Blakney GT, Zhang HM, Marshall AG. 2011. Epitope mapping of a 95 kDa antigen in complex with antibody by solution-phase amide backbone hydrogen/deuterium exchange monitored by Fourier transform ion cyclotron resonance mass spectrometry. Anal. Chem. 83: 7129–7136. Zhang H, Cui W, Gross ML. 2013. Mass spectrometry for the biophysical characterization of therapeutic monoclonal antibodies. FEBS Lett. 5793: 00871–00875.
SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher’s web site.
276 wileyonlinelibrary.com/journal/jmr
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
Revised: 10 August 2014,
Accepted: 15 August 2014,
Published online in Wiley Online Library: 9 February 2015
(wileyonlinelibrary.com) DOI: 10.1002/jmr.2418
Epitope characterization of an anti-PD-L1 antibody using orthogonal approaches† Gang Hao, John S. Wesolowski, Xuliang Jiang, Scott Lauder‡ and Vanita D. Sood* The binding of programmed death ligand 1 protein (PD-L1) to its receptor programmed death protein 1 (PD-1) mediates immunoevasion in cancer and chronic viral infections, presenting an important target for therapeutic intervention. Several monoclonal antibodies targeting the PD-L1/PD-1 signaling axis are undergoing clinical trials; however, the epitopes of these antibodies have not been described. We have combined orthogonal approaches to localize and characterize the epitope of a monoclonal antibody directed against PD-L1 at good resolution and with high confidence. Limited proteolysis and mass spectrometry were applied to reveal that the epitope resides in the first immunoglobulin domain of PD-L1. Hydrogen–deuterium exchange mass spectrometry (HDX-MS) was used to identify a conformational epitope comprised of discontinuous strands that fold to form a beta sheet in the native structure. This beta sheet presents an epitope surface that significantly overlaps with the PD-1 binding interface, consistent with a desired PD-1 competitive mechanism of action for the antibody. Surface plasmon resonance screening of mutant PD-L1 variants confirmed that the region identified by HDX-MS is critical for the antibody interaction and further defined specific residues contributing to the binding energy. Taken together, the results are consistent with the observed inhibitory activity of the antibody on PD-L1-mediated immune evasion. This is the first report of an epitope for any antibody targeting PD-L1 and demonstrates the power of combining orthogonal epitope mapping techniques. Copyright © 2015 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web site. Keywords: hydrogen deuterium exchange; mass spectrometry; epitope mapping; monoclonal antibody; mutagenesis; PD-L1
INTRODUCTION
J. Mol. Recognit. 2015; 28: 269–276
* Correspondence to: Vanita D. Sood, EMD Serono Research and Development Institute, Inc. 45A Middlesex Turnpike, Billerica, MA, 02144 USA. E-mail: [email protected] †
‡
This article is published in Journal of Molecular Recognition as part of the Special Issue ‘State-of-the-art B-cell Epitope Discovery’, edited by Dr Yasmina Abdiche, Dr Arvind Sivasubramanian, Dr Jerry Slootstra, Dr Darren Flower and Professor Edouard Nice. Current address: Merrimack Pharmaceuticals, 1 Kendall Square, Cambridge, MA 02139 G. Hao, J. S. Wesolowski, X. Jiang, S. Lauder, V. D. Sood EMD Serono Research and Development Institute, Inc., 45A Middlesex Turnpike, Billerica, MA, 02144, USA
Copyright © 2015 John Wiley & Sons, Ltd.
269
Programmed death ligand 1 protein (PD-L1), also known as cluster of differentiation 274 or B7 homolog 1, is a type I transmembrane protein belonging to the B7 family (Dong et al., 1999; Freeman et al., 2000). PD-L1 binds to its receptor, programmed cell death protein 1 (PD-1), which is expressed in activated T cells, B cells, and macrophages (Ishida et al., 1992; Okazaki et al., 2001). The complexation of PD-L1 and PD-1 exerts immunosuppressive effects by inhibiting T cell proliferation and cytokine production of IL-2 and IFN-γ (Freeman et al., 2000; Carter et al., 2002). The upregulation of the PD-L1/PD-1 pathway is implicated in cancer and chronic viral infections, where tumor cells or viruses evade immune surveillance via PD-L1 overexpression (Iwai et al., 2002; Brown et al., 2003; Kao et al., 2011). Inhibition of PD-1 or PD-L1 restores the attenuated immune response and leads to increased antitumor and anti-viral activities (Hirano et al., 2005; Barber et al., 2006; Velu et al., 2009; Bhadra et al., 2011). Because of the high therapeutic potential, several monoclonal antibodies targeting PD-L1/PD-1 are undergoing phase I and II clinical trials for various cancers and showed efficacy for non-small-cell lung cancer, melanoma, and renal-cell cancer Berger et al., 2008; Brahmer et al., 2012; Lipson et al., 2013). A critical step in therapeutic antibody development is to define the region on the antigen that is recognized by the complementarity determining regions, a process termed epitope mapping. The information is required for elucidating the mechanism of the drug and also for regulatory approval (FDA, 1997). Antibody epitopes can be categorized as linear or conformational (Barlow et al., 1986). A linear epitope is typically a single
short peptide sequence, whereas a conformational epitope comprises discontinuous segments brought into proximity upon protein folding. Linear epitopes can be identified by screening a synthetic peptide library spanning the antigen sequence to identify linear epitopes (Geysen et al., 1984). For a conformational epitope that comprises discontinuous segments brought into proximity upon protein folding, structural elucidation of the antigen–antibody complex by X-ray crystallography provides the most detailed information on the binding interface (Amit et al., 1986; Mylvaganam et al., 1998). This approach, however, can be unsuccessful in spite of good efforts. Other methods have been applied in epitope mapping, but each method alone usually does not provide a clear map of the epitope. Alanine scanning mutagenesis is a well-established method, but this approach is labor intensive, and conformational changes or unfolding induced by mutation remote from the contact region
G. HAO ET AL. can confound interpretation of the results (Benjamin et al., 1984). Limited proteolysis monitored by mass spectrometry has also been applied to conformational epitope mapping (Jemmerson and Paterson, 1986). Identification of regions protected from proteolysis by the bound antibody reveals the location of the epitope; however, the resolution of the technique is low because of the sporadic distribution of cleavage sites. Hydrogen–deuterium exchange (HDX) mass spectrometry (MS) has emerged as a powerful tool for studying protein–protein interactions (Hoofnagle et al., 2003; Englander, 2006; Wales and Engen, 2006; Chalmers et al., 2011; Bobst and Kaltashov, 2012; Zhang et al., 2013). In an HDX epitope mapping experiment, the antigen is incubated either alone or in complex with an antibody in deuterium oxide (D2O) to allow the exchange of the backbone amide hydrogen atoms for deuterium. The antigen surface that is excluded from solvent by bound antibody (the epitope) displays a reduced amide exchange rate that can be sensitively detected by MS. The technique has been recently applied to several antibodies against proteins including cytochrome c, thrombin, interleukin-1β, Fas ligand, and cashew food allergen (Baerga-Ortiz et al., 2002; Lu et al., 2005; Coales et al., 2009; Obungu et al., 2009; Zhang et al., 2011). A limitation of the standard HDX-MS method is that it is still challenging to achieve single-residue resolution for the epitope. While electron transfer dissociation has enabled the monitoring of backbone dynamics at single-residue resolution (Huang et al., 2011; Rand et al., 2012), to our knowledge, it has not been applied to the study of antibody–antigen interactions. In our effort to identify the epitope of a specific antibody targeting PD-L1, we attempted to crystallize the antibody/PD-L1 complex, but the complex proved recalcitrant to crystallization. In order to localize and characterize the epitope of anti-PD-L1 at good resolution and with high confidence and to elucidate the mechanism of the antibody, we applied a combination of orthogonal approaches, namely limited proteolysis, HDX-MS, and mutational scanning. Here, we report the first epitope for an antibody against this high-potential therapeutic target.
MATERIALS AND METHODS PD-L1, anti-PD-L1 IgG, and Fab production Anti-PD-L1 IgG was expressed in Chinese hamster ovary cells and purified by protein A affinity chromatography (MabSelect, GE Healthcare). Anti-PD-L1 Fab and the extracellular domain (ECD) of human PD-L1 (residues 19–238 of the open reading frame) were expressed as 6-His-tagged fusion proteins in human embryonic kidney 293 cells. His-tagged proteins were purified by nickel affinity chromatography. All proteins were buffer exchanged with phosphate-buffered saline (PBS) at pH 7.2. The IgG was prepared at a stock concentration of 10 μg/μl, the Fab at 4.34 μg/μl, and the PD-L1 at 1.90 μg/μl. Limited proteolysis
270
PD-L1 (2.5 μl) was mixed with 4.5 μl Fab and 2 μg trypsin (Promega, Madison, WI, USA), diluted to a total volume of 19 μl with PBS, and incubated at 37°C for 2 h. The trypsin digestion was terminated by addition of 1 μl of trifluroacetic acid (final concentration 0.1%) and analyzed using liquid chromatography MS (LC-MS) as described in the succeeding texts. For the control sample, the same amounts of PD-L1 and Fab were digested separately and mixed prior to LC-MS analysis.
wileyonlinelibrary.com/journal/jmr
HDX PD-L1 (2.5 μl) was incubated with or without 4.5 μl Fab at 4°C for 5 min. The preincubated complex or PD-L1 alone was diluted to a final volume of 50 μl with PBS prepared with D2O, then further incubated at 4°C to allow deuterium exchange. The final concentrations in the exchange reaction were 3.65 and 7.94 μM for PD-L1 and Fab, respectively. A molar excess of Fab and concentrations of both components much higher than the KD of the complex (Table 1) were used to ensure saturation binding of PD-L1. Four degrees Celsius was chosen because of precipitation of protein observed upon extended incubation at high concentrations at ambient temperature (data not shown). Aliquots during exchange were taken at 30 s, 2, 10, and 60 min. At each time point, the exchange was quenched using 50 μl of 0.2 M sodium phosphate with 0.2 M tris(2-carboxyethyl)phosphine and 0.4 M guanidine hydrochloride at pH 2.5. Two microliters of pepsin (2 mg/ml) was added to the quenched samples, and pepsin digestion on the chilled mixture was carried out in a thermomixer set to 20°C for 2 min. The digested samples were immediately applied to an LC-MS system and analyzed as described in the succeeding texts. LC-MS The digested samples were analyzed on an LC-MS system comprised of an Acquity UPLC (Waters, Milford, MA, USA) and a microTOF-Q (Bruker, Billerica, MA, USA). The solvent mixture, sample loop, LC column, and the connecting tubing were immersed in a circulating chilled water bath kept at 0°C during the sample run. The peptides were separated on a 1.0 × 50-mm ZORBAX C18 column (Agilent, Palo Alto, CA, USA) at a flow rate of 100 μl/min over a gradient from 8 to 35% solvent B (0.1% Table 1. Summary of PD-L1 mutant binding to anti-PD-L1 Mutation
ΔΔGmut (kcal/mol)
Wild type I54A I54K Y56A Y56K E58A E60A D61A Q66A V68A V68R R113A M115A
— 1.28 0.62 >4 >5 1.90 1.45 >5 0.86 0.02 0.55 1.53 0.97
KD (nM)
T1/2 (°C)
0.55 ± 0.21 0.06 ± 0.09 1.57 ± 0.19 > 1 μM >4 μM 13.58 ± 0.59 6.32 ± 0.44 NB 2.35 ± 0.23 0.57 ± 0.04 1.37 ± 0.05 7.22 ± 0.26 2.79 ± 0.17
60.1 56.6 58.8 59.1 56.9 55.8 55.8 51.2 61.1 61.4 58.2 58.0 51.2
Point mutants of PD-L1 were compared with wild-type PD-L1 antigen for antibody binding. A kinetic SPR study was used to determine the equilibrium dissociation constant (KD) and the change in the Gibbs free energy of binding of mutant relative to wild-type PD-L1 (ΔΔGmut). Mutants at Y56 and D61 had such a low affinity that the KD could not be accurately measured, and the minimum KD and ΔΔGmut are given instead. The temperature midpoint (T1/2) of FMTU is given for the wild-type and mutant proteins. For KD, the mean and standard deviation is given where possible. NB, no binding.
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
EPITOPE OF AN ANTI-PD-L1 ANTIBODY formic acid in acetonitrile) in 10 min. The same digestion conditions and the LC-MS step were applied to a mixture of fully deuterated peptide standards (Genscript, Piscataway, NJ, USA) to determine the overall deuterium recovery of the method. The average deuterium recovery was about 70% for the standards. The mass spectra were collected in automatic MS/MS mode from 200 to 2000 amu to identify sequence coverage of the nondeuterated sample. Mascot (Matrix Science, Boston, MA, USA) was used to search against the PD-L1 sequence. A mass accuracy of 0.1 Da was used for the precursor ions and 0.2 Da for the product ions. The score cutoff used to identify positive hits was 20. To measure the deuterium content of the exchanged samples, the mass spectra were recorded in MS mode. The raw MS data were processed by HDExaminer (Sierra Analytics, Modesto, CA, USA) to calculate the deuterium incorporation. An extracted ion chromatogram was generated for each identified peptide, and the deuterium incorporation was calculated based on the increase in centroid mass. Mutant PD-L1 production Plasmids encoding mutant PD-L1 were constructed using QuikChange mutagenesis (Agilent Technologies) and expressed and purified as described for wild-type PD-L1 in the preceding texts. The mutant proteins were analyzed by analytical size exclusion chromatography (SEC) on a Bio SEC-3 column (4.6 × 300 mm, 300 Å, 3 μm; Agilent) and compared with wild type; all mutants were eluted at the same volume as the wild type. The thermal stability of the wild-type and mutant proteins were measured by fluorescence monitored thermal unfolding (FMTU) on a quantitative PCR instrument (Applied Biosystems). Protein (95 μl) at 0.2 mg/ml was mixed with 5 μl of SYPRO Orange (5000× concentrate, Life Technologies), and fluorescence at 570 nM was monitored as a function of temperature. The data were fit to the Boltzmann sigmoidal equation to obtain the temperature at half-maximal fluorescence.
Surface plasmon resonance Kinetic binding experiments were performed on a Biacore 4000 (GE Healthcare). Purified goat antihuman IgG Fc (Jackson ImmunoResearch Laboratories) was immobilized on a CM5 chip (GE Healthcare) using amine coupling chemistry. Anti-PD-L1 IgG was captured by the antihuman IgG. Binding experiments were carried out in 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-buffered saline (20 mM HEPES, 150 mM NaCl, 3.4 mM EDTA, and 0.05% P20 surfactant) at 25°C. Kinetic data were collected at six different concentrations of wild-type or mutant PD-L1 analyte: 100, 50, 25, 12.5, 6.3, and 0 nM. For the two hot spot residues (Y56 and D61), data were collected at 1 μM to 62.5 nM. The analyte was allowed to associate for 3 min followed by a dissociation step in buffer for 10 min at 30-μl/min flow rate. The surface was regenerated between each concentration of analyte, and fresh antibody was captured. The data were fit to a 1:1 Langmuir binding model with the BIA evaluation software (GE Healthcare). Kinetic rate constants were determined from the fits of the association and dissociation phases, and the KD was derived from the ratio of these constants.
RESULTS AND DISCUSSION We sought to determine the epitope of anti-PD-L1 to gain insight into its mechanism of action using two complementary approaches in parallel: HDX-MS and site-directed mutagenesis. We initially attempted to cocrystallize the complex of human PD-L1 and the antigen binding domain of antiPD-L1. While both individual components could be crystallized, the complex did not crystallize. Using HDX-MS and mutagenesis, however, we were able to determine the epitope to residue-level resolution.
J. Mol. Recognit. 2015; 28: 269–276
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jmr
271
Figure 1. The epitope for anti-PD-L1 is located on the immunoglobulin domain I of human PD-L1, as revealed by limited proteolysis and mass spectrometry. (A) Total ion chromatograms of PD-L1 subjected to limited proteolysis with trypsin, unbound control (upper panel), or in complex with the Fab (lower panel). (B) Peptides generated from PD-L1 by limited proteolysis shown on the sequence of the ECD. Bold with dashed underscore: peptides generated in PD-L1 only in the absence of Fab. Underscored: peptides generated both in the presence and absence of Fab binding.
G. HAO ET AL. Limited proteolysis Human PD-L1 ECD contains two immunoglobulin-like domains (Lin et al., 2008). Limited proteolysis was used to rapidly establish an approximate location of the epitope. PD-L1 was digested with trypsin under nondenaturing condition in the presence or absence of anti-PD-L1 Fab. As shown in Figure 1A, several peaks present in the control sample without Fab disappeared in the complex sample, corresponding to the peptides protected from proteolytic cleavage by Fab binding. Importantly, the protected region was located in the first immunoglobulin domain of PD-L1 (Figure 1B), where PD-1 engages with PD-L1 (Lin et al., 2008). Binding in a region that would block receptor–ligand interaction is consistent with an immunostimulatory mechanism of action for the antibody. While the entire length of the antigen was scanned in the subsequent HDX experiment, the information here was helpful for optimizing the peptide digestion conditions as described in the next section.
Optimization of pepsin digestion of PD-L1 To improve the resolution of the epitope from domain to peptidelevel resolution using HDX, it was necessary to optimize the digestion conditions to achieve good sequence coverage of the antigen. In particular, generation of overlapping peptides for each region is helpful as each peptide can be separately analyzed for HDX protection, increasing the confidence in observed exchange rates and potentially increasing the resolution of the epitope to a smaller region. Initially, we used full-length IgG in complex with PD-L1 for
epitope mapping by HDX. However, digestion of the complex created numerous peptide fragments from the IgG that complicated the data analysis of the antigen fragments. While one could immobilize the IgG and remove it before digestion, we instead simplified the process by using the Fab portion of the antibody in place of fulllength IgG. Figure 2A shows the total ion chromatograms for the pepsin digestion of PD-L1 alone and in complex with Fab, which were similar. The absence of peaks corresponding to peptides from the Fab suggests that it remained largely intact during digestion, consistent with the published observations on Fab resistance to proteolysis (Wilson et al., 2002), allowing us to analyze peptides from the antigen without removing Fab prior to digestion. We found that PD-L1 was efficiently digested in immunoglobulin domain I, which generated numerous overlapping peptides (Figure 2B). Interestingly, despite the known lack of specificity of pepsin, domain II was not covered by as many overlapping peptides as domain I but rather was covered by a few longer peptides. However, because limited proteolysis showed that domain I likely contained the epitope (refer to preceding texts), coverage of domain II by fewer overlapping peptides was not a concern for identifying the epitope. Sequence coverage (95%) was achieved by MS/MS analysis and is shown schematically in Figure 2B. One region in domain I absent from the MS/MS analysis (residues 33–39) is known to be modified by N-linked glycosylation and is likely to be protected from proteolytic digestion (Rehm, 2006). Although antibodies do not often recognize carbohydrates, there are several examples of carbohydrate– antibody interactions, and it was a priori a possibility that if the
272
Figure 2. Epitope mapping of anti-PD-L1 by HDX mass spectrometry. (A) Total ion chromatograms of PD-L1 peptic digestion with and without Fab. (B) Sequence coverage of human PD-L1 in peptic digestion is shown by boxes under the sequence. Peptides showing reduced exchange rate upon Fab binding are annotated with black boxes. (C) Example deuterium uptake plots for segment 54–66 and 112–122, which contain 11 and 9 potentially exchangeable backbone amide protons, respectively, not including the terminal amides that exchange at a much faster timescale (Bai et al., 1993). Blue diamonds, PD-L1 alone; brown squares, PD-L1 in complex with Fab. Refer to Supplemental Figure S1 for all plots. (D) Example mass spectra showing isotope distribution of exchanged peptides in PD-L1 alone (top panel), PD-L1 in complex with Fab (middle panel), and control with no heavy water (lower panel). (E) The crystal structure (Lin et al., 2008) of human PD-L1 immunoglobulin domain I rendered as a molecular surface (blue). The two epitope segments with strong and modest protection are in magenta and light pink, respectively. The PD-1 ligand is rendered in red cartoons. This and subsequent structural figures were generated using PyMol (http://www.pymol.org/) and the coordinates were taken from 3BIK (Lin et al., 2008).
wileyonlinelibrary.com/journal/jmr
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
EPITOPE OF AN ANTI-PD-L1 ANTIBODY epitope were located in this region, we would be unable to identify it. However, in contrast to the high affinity and high specificity of our antibody, the known carbohydrate–antibody interactions are of low affinity and low specificity (Manimala et al., 2007), so it seemed unlikely that the epitope would be located in a glycosylated peptide. Nevertheless, to exclude the possibility of a carbohydrate epitope and to identify the epitope we performed HDX-MS using the optimized pepsin digestion conditions. HDX
J. Mol. Recognit. 2015; 28: 269–276
Mutational scanning of PD-L1 binding to antibody To further corroborate the results of HDX-MS, to obtain a finer, residue-level mapping of the epitope, and to obtain information on the energetic hot spots of binding within the epitope, we applied an orthogonal technique to complement the HDX-MS data. We selected solvent-exposed residues within and around the epitope region for mutation. Alanine mutants were generated and were evaluated for the binding affinity to the antibody using SPR. In some cases, mutations to nonalanine mutants were also tested to obtain additional information. Prior to the SPR analysis, the mutants were analyzed using analytical SEC and FMTU to ensure that the mutation did not lead to significant conformational change that could confound interpretation of binding results. All epitope mutants displayed SEC retention time comparable with the wild type (Supplemental Figure S2) and a half-maximal temperature of FMTU similar to native or at most 10°C lower than native, as well as similar fluorescence at the folded baseline (Figure 3 and Table 1). These two sets of observations indicate that below 40°C, the native structure was not significantly perturbed by the mutations and allowed straightforward interpretation of subsequent binding experiments that were performed at room temperature. A total of 48 PD-L1 mutants were tested for binding to anti-PD-L1 using SPR. The analysis identified a number of residues that display reduced binding affinity upon substitution to alanine, as well as two residues that slightly decreased affinity upon mutation to larger residues while maintaining or increasing affinity upon alanine
Figure 3. Fluorescence monitored thermal melt analysis of PD-L1 mutant proteins. Wild-type and mutant PD-L1 proteins were heated in the presence of a fluorophore to monitor exposure of the hydrophobic core upon denaturation of the protein. Most mutants displayed similar denaturation profiles to the wild type, while D61A and M115A denatured at somewhat lower temperature. All mutant and wild-type proteins had a similar folded baseline at room temperature. Colors correspond to cartoon in Figure 5.
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jmr
273
The HDX experiment was carried out for PD-L1 alone or in complex with Fab for 0.5, 2, 10, and 60 min. The exchange was quenched and subjected to proteolytic digestion and LC-MS analysis (Figure 2A). The deuterium incorporation was calculated based on the centroid mass shift of the exchanged sample relative to the nonexchanged control, and an example mass spectrum is shown in Figure 2C. Deuterium incorporation plotted as a function of time revealed that most peptides showed no difference in exchange rate between PD-L1 alone or in complex with Fab (Supplemental Figure S1). The exceptions to this were two regions consisting of residues 53–66, which displayed strong protection in the presence of Fab (>2-Da mass shift, Figure 2D left panel) and residues 112–122, which displayed modest protection (>0.5-Da mass shift, Figure 2D right panel). The results of the HDX epitope mapping are summarized in Figure 2B. In the case of the peptide comprising residues 112–122, a more precise definition of the subregion protected by Fab was precluded because of the lack of availability of overlapping peptides. However, the small number of differentially exchanging hydrogen atoms in peptide 112–122 is consistent with only a subregion of this peptide being protected by bound Fab (refer to mutational analysis, in the succeeding texts). In contrast, for the region consisting of residues 53–66, shorter peptides comprised of its N and C-terminal regions also showed strong protection from deuterium exchange, providing independent corroboration of the protection by the Fab in this region (Supplemental Figure S1). These two segments were assigned as the putative epitope regions on PD-L1. Inspection of the crystal structure of human PD-L1 ( Lin et al., 2008) indicates that although the two peptides are far apart in the primary sequence, they are proximal in the three-dimensional structure, constituting a single patch on the surface of the antigen (Figure 2E). Furthermore, this region significantly overlaps with the binding footprint with PD-1 (Figure 2E, and Lin et al., 2008), consistent with an inhibitory PD-1 competitive mechanism of activity of the anti-PD-L1 monoclonal antibody. The mechanism for the reduced exchange rate from Fab binding could be because of solvent exclusion, stabilization of PD-L1 by Fab binding, or a combination of both these mechanisms. From other studies comparing the epitope determined from HDX and X-ray crystallography (Gerhardt et al., 2009; Malito et al., 2013), it seems that both mechanisms may play a role and that epitopes tend to colocalize with observed areas of protection from HDX. The exchange behavior also provides insights into to the dynamics underlying the observed exchange rate. There are two types of kinetics for HD exchange. In the EX1 regime, the chemical exchange rate is much greater than the protein unfolding rate or, in the case of a complex, the rate of dissociation. In the EX1 regime, the exchange profile thus displays a bimodal distribution (Brock, 2012; Percy et al., 2012; Jaswal, 2013). In the EX2 regime, the amide is exposed transiently to the solvent upon rapid
breathing of the local environment, which is faster than the intrinsic chemical exchange rate. The exchange profile for the EX2 behavior is generally Gaussian, as observed here for the PD-L1 peptides (Figure 2C), supporting an EX2 mechanism featuring rapid breathing and local unfolding, which is reduced upon Fab binding. In the context of native protein–protein interactions, the observed protection from exchange is consistent with binding and stabilizing the epitope, as well as excluding solvent from the identified epitope regions (Brock, 2012; Percy et al., 2012; Jaswal, 2013).
G. HAO ET AL.
Figure 4. SPR analysis of PD-L1 mutant protein binding to antibody. Example sensorgrams for selected PD-L1 mutants and wild type are shown. Anti-PD-L1 antibody was captured and wild-type or mutant antigen was bound to the captured antibody. The antigen was titrated down from 100 nM, except for the hot spot mutations (Y56A, Y56K, and D61) that were titrated from 1 μM down to increase the poor binding signal. Even at 1 μM, no detectable binding could be observed for D61A. Y56A displayed variable but low-affinity binding in independent experiments, and Y56K did not display enough binding to accurately measure KD. All other wild-type and mutant proteins could be fit to the Langmuir equation and the KD derived from the kinetic constants.
substitution (Table 1 and Figure 4). The equilibrium dissociation constant (KD) and the change in binding free energy compared with wild-type PD-L1 (ΔΔG) for these mutants are summarized in Table 1. Two residues residing on the strongly protected segment identified by HDX-MS, Y56 and D61, strongly or completely disrupt the binding upon mutation to alanine (Figure 4) and thus represent the binding hot spots of the epitope. This suggests that the hot spots Y56 and D61 engage in significant hydrogen bonding and/or van der Waals interactions with the antibody. Several other residues on the same segment, as well as two residues (R113 and M115) residing on the modestly protected segment, also display a measurable although more modest contribution to the binding energy (Figure 4 and Table 1). It is also interesting to note that residues I54 and V68, which are located on the periphery of the protected region, slightly destabilize the binding when mutated to lysine or arginine but have no effect or even stabilize the binding when mutated to alanine. Taken together, the alanine and other mutations at these two residues further support the localization of the epitope by HDX-MS and the binding hot spots by demonstrating that introducing long charged side chains close to the putative epitope can reduce antibody binding, presumably by steric hindrance or charge repulsion. Identification of the epitope of anti-PD-L1 antibody by orthogonal techniques
274
Figure 5 summarizes the results by mapping the epitope residues that contribute to the antibody binding energy and are
wileyonlinelibrary.com/journal/jmr
Figure 5. Summary of epitope mapping by HDX and mutational scanning. The two epitope segments with strong and modest protection identified by HDX are in magenta and light pink, respectively. Epitope residues important for binding affinity, as determined by mutational scanning, are mapped on the crystal structure of domain I of human PD-L1 (Lin et al., 2008). Residues are colored according to the energetic contribution. Red, >3 kcal/mol; orange, >0.7 kcal/mol; green, destabilizes binding only upon nonalanine mutation.
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
EPITOPE OF AN ANTI-PD-L1 ANTIBODY protected from HD exchange on the crystal structure of PD-L1. The mutational analysis, which probes the role of side chains in binding, identified several residues that contribute energetically to antibody binding. These residues are all located in the same structurally contiguous region identified by HDX, consistent with the antibody engaging with the side chains in this region of the antigen and excluding solvent from the entire surface patch. Binding of the Fab may also stabilize PD-L1 in this region, decreasing the rate of “breathing” and reducing exchange by this additional mechanism. The epitope for the antibody overlaps significantly with the binding site of PD-1, consistent with the inhibitory activity of the antibody. Collectively, our data demonstrate that the combination of HDX-MS and site-directed
mutagenesis provides a viable approach for epitope mapping of antigen–antibody interactions. The workflow can be generically applied to other systems where a crystal structure of the interaction complex is not attainable and was applied here to elucidate for the first time an epitope for an antibody targeting PD-L1.
Acknowledgements We thank Steven Degon for Fab purification, Syngene International Ltd. for purification of mutant antigen protein, Paul Towler for Fab crystallization, and Roland Keller for critical review of the manuscript.
REFERENCES
J. Mol. Recognit. 2015; 28: 269–276
Englander SW. 2006. Hydrogen exchange and mass spectrometry: A historical perspective. J. Am. Soc. Mass Spectrom. 17: 1481–1489. FDA. 1997. Points to Consider in the Manufacture and Testing of Monoclonal Antibody Products for Human Use. http://www.fda.gov/downloads/ biologicsbloodvaccines/guidancecomplianceregulatoryinformation/ otherrecommendationsformanufacturers/ucm153182.pdf: U. S. Department of Health and Human Services Food and Drug Administration Center for Biologics Evaluation and Research. Freeman GJ, Long AJ, Iwai Y, Bourque K, Chernova T, Nishimura H, Fitz LJ, Malenkovich N, Okazaki T, Byrne MC, Horton HF, Fouser L, Carter L, Ling V, Bowman MR, Carreno BM, Collins M, Wood CR, Honjo T. 2000. Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member leads to negative regulation of lymphocyte activation. J. Exp. Med. 192: 1027–1034. Gerhardt S, Abbott WM, Hargreaves D, Pauptit RA, Davies RA, Needham MR, Langham C, Barker W, Aziz A, Snow MJ, Dawson S, Welsh F, Wilkinson T, Vaugan T, Beste G, Bishop S, Popovic B, Rees G, Sleeman M, Tuske SJ, Coales SJ, Hamuro Y, Russell C. 2009. Structure of IL-17A in complex with a potent, fully human neutralizing antibody. J. Mol. Biol. 394: 905–21. Geysen HM, Meloen RH, Barteling SJ. 1984. Use of peptide synthesis to probe viral antigens for epitopes to a resolution of a single amino acid. Proc. Natl. Acad. Sci. USA 81: 3998–4002. Hirano F, Kaneko K, Tamura H, Dong H, Wang S, Ichikawa M, Rietz C, Flies DB, Lau JS, Zhu G, Tamada K, Chen L. 2005. Blockade of B7-H1 and PD-1 by monoclonal antibodies potentiates cancer therapeutic immunity. Cancer Res. 65: 1089–1096. Hoofnagle AN, Resing KA, Ahn NG. 2003. Protein analysis by hydrogen exchange mass spectrometry. Annu. Rev. Biophys. Biomol. Struct. 32: 1–25. Huang RY, Garai K, Frieden C, Gross ML. 2011. Hydrogen/deuterium exchange and electron-transfer dissociation mass spectrometry determine the interface and dynamics of apolipoprotein E oligomerization. Biochemistry 50: 9273–82. Ishida Y, Agata Y, Shibahara K, Honjo T. 1992. Induced expression of PD-1, a novel member of the immunoglobulin gene superfamily, upon programmed cell death. EMBO J. 11: 3887–3895. Iwai Y, Ishida M, Tanaka Y, Okazaki T, Honjo T, Minato N. 2002. Involvement of PD-L1 on tumor cells in the escape from host immune system and tumor immunotherapy by PD-L1 blockade. Proc. Natl. Acad. Sci. USA 99: 12293–12297. Jaswal SS. 2013. Biological insights from hydrogen exchange mass spectrometry. Biochim. Biophys. Acta 1834: 1188–1201. Jemmerson R, Paterson Y. 1986. Mapping epitopes on a protein antigen by the proteolysis of antigen-antibody complexes. Science 232: 1001–1004. Kao C, Oestreich KJ, Paley MA, Crawford A, Angelosanto JM, Ali MA, Intlekofer AM, Boss JM, Reiner SL, Weinmann AS, Wherry EJ. 2011. Transcription factor T-bet represses expression of the inhibitory receptor PD-1 and sustains virus-specific CD8+ T cell responses during chronic infection. Nat. Immunol. 12: 663–671. Lin DY, Tanaka Y, Iwasaki M, Gittis AG, Su HP, Mikami B, Okazaki T, Honjo T, Minato N, Garboczi DN. 2008. The PD-1/PD-L1 complex resembles the antigen-binding Fv domains of antibodies and T cell receptors. Proc. Natl. Acad. Sci. USA 105: 3011–3016.
Copyright © 2015 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jmr
275
Amit AG, Mariuzza RA, Phillips SE, Poljak RJ. 1986. Three-dimensional structure of an antigen antibody complex at 2.8 A resolution. Science 233: 747–753. Baerga-Ortiz A, Hughes CA, Mandell JG, Komives EA. 2002. Epitope mapping of a monoclonal antibody against human thrombin by H/Dexchange mass spectrometry reveals selection of a diverse sequence in a highly conserved protein. Protein Sci. 11: 1300–1308. Bai Y, Milne JS, Mayne L, Englander SW. 1993. Primary structure effects on peptide group hydrogen exchange. Struct. Funct. Genet. 17: 75. Barber DL, Wherry EJ, Masopust D, Zhu B, Allison JP, Sharpe AH, Freeman GJ, Ahmed R. 2006. Restoring function in exhausted CD8 T cells during chronic viral infection. Nature 439: 682–687. Barlow DJ, Edwards MS, Thornton JM. 1986. Continuous and discontinuous protein antigenic determinants. Nature 322: 747–748. Benjamin DC, Berzofsky JA, East IJ, Gurd FR, Hannum C, Leach SJ, Margoliash E, Michael JG, Miller A, Prager EM. 1984. The antigenic structure of proteins: a reappraisal. Annu. Rev. Immunol. 2: 67–101. Berger R, Rotem-Yehudar R, Slama G, Landes S, Kneller A, Leiba M, KorenMichowitz M, Shimoni A, Nagler A. 2008. Phase I safety and pharmacokinetic study of CT-011, a humanized antibody interacting with PD-1, in patients with advanced hematologic malignancies. Clin. Cancer Res. 14: 3044–3051. Bhadra R, Gigley JP, Weiss LM, Khan IA. 2011. Control of Toxoplasma reactivation by rescue of dysfunctional CD8+ T-cell response via PD1-PDL-1 blockade. Proc. Natl. Acad. Sci. USA 108: 9196–9201. Bobst CE, Kaltashov IA. 2012. Localizing flexible regions in proteins using hydrogen deuterium exchange mass spectrometry. Methods Mol. Biol. 896: 375–85. Brahmer JR, Tykodi SS, Chow LQ, Hwu WJ, Topalian SL, Hwu P, Drake CG, Camacho LH, Kauh J, Odunsi K, Pitot HC, Hamid O, Bhatia S, Martins R, Eaton K, Chen S, Salay TM, Alaparthy S, Grosso JF, Korman AJ, Parker SM, Agrawal S, Goldberg SM, Pardoll DM, Gupta A, Wigginton JM. 2012. Safety and activity of anti-PD-L1 antibody in patients with advanced cancer. N. Engl. J. Med. 366: 2455–2465. Brock A. 2012. Fragmentation hydrogen exchange mass spectrometry: a review of methodology and applications. Protein Expr. Purif. 84: 19–37. Brown JA, Dorfman DM, Ma FR, Sullivan EL, Munoz O, Wood CR, Greenfield EA, Freeman GJ. 2003. Blockade of programmed death1 ligands on dendritic cells enhances T cell activation and cytokine production. J. Immunol. 170: 1257–1266. Carter L, Fouser LA, Jussif J, Fitz L, Deng B, Wood CR, Collins M, Honjo T, Freeman GJ, Carreno BM. 2002. PD-1:PD-L inhibitory pathway affects both CD4(+) and CD8(+) T cells and is overcome by IL-2. Eur. J. Immunol. 32: 634–643. Chalmers MJ, Busby SA, Pascal BD, West GM, Griffin PR. 2011. Differential hydrogen/deuterium exchange mass spectrometry analysis of protein-ligand interactions. Expert Rev. Proteomics 8: 43–59. Coales SJ, Tuske SJ, Tomasso JC, Hamuro Y. 2009. Epitope mapping by amide hydrogen/deuterium exchange coupled with immobilization of antibody, on-line proteolysis, liquid chromatography and mass spectrometry. Rapid Commun. Mass Spectrom. 23: 639–647. Dong H, Zhu G, Tamada K, Chen L. 1999. B7-H1, a third member of the B7 family, co-stimulates T-cell proliferation and interleukin-10 secretion. Nat. Med. 5: 1365–1369.
G. HAO ET AL. Lipson EJ, Sharfman WH, Drake CG, Wollner I, Taube JM, Anders RA, Xu H, Yao S, Pons A, Chen L, Pardoll DM, Brahmer JR, Topalian SL. 2013. Durable cancer regression off-treatment and effective reinduction therapy with an anti-PD-1 antibody. Clin. Cancer Res. 19: 462–468. Lu J, Witcher DR, White MA, Wang X, Huang L, Rathnachalam R, Beals JM, Kuhstoss S. 2005. IL-1beta epitope mapping using site-directed mutagenesis and hydrogen-deuterium exchange mass spectrometry analysis. Biochemistry 44: 11106–11114. Malito E, Faleri A, Lo Surdo P, Veggi D, Maruggi G, Grassi E, Cartocci E, Bertoldi I, Genovese A, Santini L, Romagnoli G, Borgogni E, Brier S, Lo Passo C, Domina M, Castellino F, Felici F, van der Veen S, Johnson S, Lea SM, Tang CM, Pizza M, Savino S, Norais N, Rappuoli R, Bottomley MJ, Masignani V. 2013. Defining a protective epitope on factor H binding protein, a key meningococcal virulence factor and vaccine antigen. Proc. Natl. Acad. Sci. USA 110: 3304–3309. Manimala JC, Roach TA, Li Z, Gildersleeve JC. 2007. High-throughput carbohydrate microarray profiling of 27 antibodies demonstrates widespread specificity problems. Glycobiology 17: 17C-23C. Mylvaganam SE, Paterson Y, Getzoff ED. 1998. Structural basis for the binding of an anti-cytochrome c antibody to its antigen: crystal structures of FabE8-cytochrome c complex to 1.8 A resolution and FabE8 to 2.26 A resolution. J. Mol. Biol. 281: 301–322. Obungu VH, Gelfanova V, Rathnachalam R, Bailey A, Sloan-Lancaster J, Huang L. 2009. Determination of the mechanism of action of antiFasL antibody by epitope mapping and homology modeling. Biochemistry 48: 7251–7260. Okazaki T, Maeda A, Nishimura H, Kurosaki T, Honjo T. 2001. PD-1 immunoreceptor inhibits B cell receptor-mediated signaling by recruiting src homology 2-domain-containing tyrosine phosphatase 2 to phosphotyrosine. Proc. Natl. Acad. Sci. USA 98: 13866–13871.
Percy AJ, Rey M, Burns KM, Schriemer DC. 2012. Probing protein interactions with hydrogen/deuterium exchange and mass spectrometry-a review. Anal. Chim. Acta 721: 7–21. Rand KD, Pringle SD, Morris M, Brown JM. 2012. Site-specific analysis of gas-phase hydrogen/deuterium exchange of peptides and proteins by electron transfer dissociation. Anal. Chem. 84: 1931–40. Rehm H. 2006. Glycoproteins. In Protein biochemistry and proteomics, Rehm H (Ed.) 4th ed. Access Online via Elsevier: Elsevier Science: New York, NY. Velu V, Titanji K, Zhu B, Husain S, Pladevega A, Lai L, Vanderford TH, Chennareddi L, Silvestri G, Freeman GJ, Ahmed R, Amara RR. 2009. Enhancing SIV-specific immunity in vivo by PD-1 blockade. Nature 458: 206–210. Wales TE, Engen JR. 2006. Hydrogen exchange mass spectrometry for the analysis of protein dynamics. Mass Spectrom. Rev. 25: 158–170. Wilson DS, Wu J, Peluso P, Nock S. 2002. Improved method for pepsinolysis of mouse IgG(1) molecules to F(ab’)(2) fragments. J. Immunol. Methods 260: 29–36. Zhang Q, Willison LN, Tripathi P, Sathe SK, Roux KH, Emmett MR, Blakney GT, Zhang HM, Marshall AG. 2011. Epitope mapping of a 95 kDa antigen in complex with antibody by solution-phase amide backbone hydrogen/deuterium exchange monitored by Fourier transform ion cyclotron resonance mass spectrometry. Anal. Chem. 83: 7129–7136. Zhang H, Cui W, Gross ML. 2013. Mass spectrometry for the biophysical characterization of therapeutic monoclonal antibodies. FEBS Lett. 5793: 00871–00875.
SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher’s web site.
276 wileyonlinelibrary.com/journal/jmr
Copyright © 2015 John Wiley & Sons, Ltd.
J. Mol. Recognit. 2015; 28: 269–276
