Determination of Protein Monoclonal–Antibody Epitopes by a Combination of Structural Proteomics Methods

Abstract

Structural proteomics techniques are useful for the determination of protein interaction interfaces. Each technique provides orthogonal structural information on the structure and the location of protein interaction sites. Here, we have characterized a monoclonal antibody epitope for a protein antigen by a combination of differential photoreactive surface modification (SM), cross-linking (CL), differential hydrogen–deuterium exchange (HDX), and epitope extraction/excision. We found that experimental data from different approaches agree with each other in determining the epitope of the monoclonal antibody on the protein antigens using the HIV-1 p24–mAb E complex as an illustrative example. A combination of these multiple structural proteomics approaches results in a detailed picture of the interaction of the proteins and increases confidence in the determination of the final structure of the protein interaction interface. Data are available via ProteomeXchange with identifier PXD040902.

Introduction

The determination of protein interaction interfaces is of considerable interest for both academia and the biotech industry.¹ Structural proteomics, which can be defined as a combination of protein chemistry methods with contemporary mass spectrometry, has the potential for elucidation of the structure of the proteins, protein complexes, and protein interaction interfaces.² The antibody–antigen complexes are a specific case of protein–protein interactions, where a protein interaction interface is formed between a portion of the antigenic protein’s surface (the epitope) and the N-terminal face of the Fab or Fv domain portion of the surface of the antibody which contains hypervariable loops (the paratope). Multiple methods, such as limited proteolysis (LP), surface modification, cross-linking, and hydrogen–deuterium exchange, have been proposed and successfully used for the determination of monoclonal-antibody epitopes of protein antigens.³ Each method provides unique experimentally derived structural information about the location of the epitope (see refs (4)−⁷ and references cited therein). Thus, proteolysis of the antigenic protein while still bound to the antibody protein (epitope excision) or antibody binding of fragments of the digested antigenic protein (epitope extraction)—with subsequent determination of the bound fragments by mass spectrometry—can provide information on the stretches of the protein antigen sequences that constitute the epitope.⁴ Protein surface modification (SM) (otherwise called covalent labeling or footprinting) of the free and the antibody-bound protein antigen can reveal differentially modified amino acid residues which are blocked by the interaction and, therefore, are part of the protein interaction interface.⁵ Cross-linking (CL) of the antigen–antibody complex can produce interprotein cross-links, located in the vicinity of the epitope–paratope interface, which can help in determining the mutual orientation of the proteins in the complex.⁶ Hydrogen–deuterium exchange can detect regions of the protein antigen’s sequence where increased protection to the exchange has occurred upon antigen–antibody complex formation as a result of stabilization of the secondary structure close to the interaction interface.⁷

Each method, however, has its own complications and limitations. Thus, epitope excision and extraction methods are influenced by the presence of digestion sites at or in the vicinity of the protein interaction interface. Surface modification and cross-linking can depend on the presence of specific groups amenable to chemical modification at the protein interaction surface of the epitope region. Changes in HDX protection may not necessarily be localized at the epitopes but may reflect allosteric protein structure conformational changes upon antigen–antibody complex formation, for which LP, SM, and CL methods may be less sensitive. Therefore, using multiple approaches for the determination of the protein antigen’s epitopes can complement the results from each method, can resolve uncertainties of each method, and can provide confident identification of the epitope location.^8,9 Here, we present the determination of the epitope on a protein antigen for a monoclonal antibody using limited proteolysis, photoreactive surface modification, cross-linking, and HDX analyses. We have used the interaction of the HIV-1 capsid p24 protein with the mAb E monoclonal antibody as an example of this combined approach.

Materials and Methods

Protein Samples

p24 (0.6 mg/mL, Chiron HIV-1 24-LS Antigen, 4323-800) and mAb E (0.5 mg/mL, Grifols Diagnostic Solutions) samples were dialyzed against a phosphate-buffered saline (PBS, pH 7.4) solution and were stored until use at 4 °C.

Epitope Extraction and Excision

The mAb E antibodies were immobilized on M-270 Epoxy Dynabeads (Invitrogen) according to the manufacturer’s protocol. Briefly, 30 μL of beads washed with 1 mL of PBS (pH 7.4) were combined with 20 μL of 0.35 mg/mL mAb E and 40 μL of 0.2 M Na₂HPO₄ and incubated for 2 days at room temperature (23 °C) with end-over-end mixing. The beads were washed with 100 μL of PBS (pH 7.4) three times prior to use. For the epitope extraction experiments, 5 μL of a 0.6 mg/mL p24 protein solution was digested with 3 μL of 1 mg/mL trypsin (Worthington) for 2 h at room temperature. For pepsin digestion, the protein solution was acidified with 0.5 μL of 10% aqueous formic acid (FA) (v/v) and digested with 3 μL of a 1 mg/mL solution of pepsin (Promega) in water for 2 h at room temperature. Digestion was stopped by the addition of 2 μL of 1 M ammonium bicarbonate. Protein digests were combined with immobilized mAb E beads, incubated at room temperature for 10 min, washed with 100 μL of PBS, and then washed with 100 μL of water, and the bound peptides were eluted with 20 μL of 0.1% aqueous FA (v/v). The eluates were centrifuged at 21 000g for 1 min at 23 °C, and 10 μL of each supernatant was transferred to autosampler vial inserts for the subsequent LC-MS analysis, where a 1 μL injection volume was used. For the epitope excision experiments, 5 μL of a 0.6 mg/mL p24 protein solution was combined with 50 μL of immobilized mAb E beads and incubated for 2 h at room temperature with end-over-end mixing, and proteins were digested with 3 μL of a 1 mg/mL trypsin solution for 10 min at room temperature. The beads were processed as above, and 10 μL of eluate was submitted for LC-MS analysis with 1 μL injection volumes being used (Supporting Information Figure S1).

Differential Surface Modification/Cross-Linking

Samples of free p24 or p24 complexed with mAb E were prepared by mixing equivalent amounts of p24 protein (3 μL of a 0.6 mg/mL solution) with 37 μL of PBS or a ∼2:1 molar excess of mAb E protein (37 μL of a 0.5 mg/mL solution), respectively. Solutions were incubated for 2 h at room temperature. Samples were split in half (for SM and CL analyses), and the surface modification SDA (succinimidyl 4,4′-azipentanoate) reagent (Sigma) was added to give a final concentration of 1 mM. Reaction mixtures were irradiated with 366 nm UV light for 10 min at room temperature. The cross-linking reaction was quenched by the addition of ammonium bicarbonate to give a final concentration of 100 mM for 10 min at room temperature. Proteins were digested with 1 μL of a 1 mg/mL trypsin (Promega) solution overnight at 37 °C. Long-distance cross-linking of the p24–mAb E complex was performed at a final DSS (disuccinimidyl suberate)–H₁₂/D₁₂ (Creative Molecules, Inc.) concentration of 1 mM for 10 min at room temperature, and the samples were processed as above.

Differential Hydrogen–Deuterium Exchange

For the intact-protein HDX analyses, 2 μL of free p24 or mAb E-bound p24 protein samples was mixed with 8 μL of H₂O or D₂O (1:4 v/v), incubated for varying time intervals, quenched with 10 μL of 0.2% FA in H₂O (i.e., 1:1 v/v), and 10 μL was injected for LC-MS analysis. LC-MS analysis was performed on a TripleTOF 6600 mass spectrometer (Sciex) interfaced with a Nexera LC system (Shimadzu). Chromatography was performed using short 3-min gradients of 5–95%B (v/v) where mobile phase A was 0.1% FA in water (v/v), and mobile phase B was acetonitrile containing 0.1% FA (v/v). The flow rate was 200 μL/min , and an Agilent C₁₈ 300 Å, 3 μm particle size, 2.1 mm i.d. × 50 mm long column, was used, cooled to 0 °C. MS1 spectra were acquired over the mass range from 100 to 2000 Da. Analyses were performed in triplicate. Spectra were deconvoluted using PeakView and BioToolKit software (Sciex). Deuteration of the p24 protein was determined based on the mass shift from the mass of the nondeuterated p24 protein (Supporting Information Figure S3).

For bottom-up HDX analyses, 2 μL of free p24 or mAb E-bound p24 protein samples was mixed with 8 μL of H₂O or D₂O (1:4 v/v), incubated for varying time intervals (0, 20, 40, 80, and 160 s), quenched with 10 μL of 0.2% FA in H₂O (i.e., 1:1 v/v), supplemented with 2 μL of freshly prepared 1 mg/mL pepsin solution in H₂O, and placed in an autosampler (for ∼30 s at 10 °C), and 10 μL was injected for LC-MS analysis. Peptic peptides from the p24 protein were identified from information-dependent acquisition (IDA) LC-MS/MS analysis of the protein in water. Data were analyzed by Protein Pilot (Sciex) and DXMSMS Match software.¹⁰ Chromatography was performed as described above. Spectra were analyzed by PeakView software (Sciex). Deuteration of the peptides was determined using the MassSpecStudio software package.¹¹

LC-MS Analysis

Protein digests were analyzed by nano-LC-MS with data-dependent acquisition (DDA) using an Easy-nLC 1200 System (Thermo Scientific) online coupled to a Q Exactive Plus (Thermo Scientific) mass spectrometer. Peptides were preconcentrated on an Acclaim PepMap 100 C18 precolumn (3 μm particle size, 75 μm inner diameter × 2 cm length) and separated on an Acclaim PepMap 100 C₁₈ main column (2 μm particle size, 75 μm inner diameter × 25 cm length) using a 50 min binary gradient (mobile phase A, 0.1% aqueous FA; mobile phase B, 84% acetonitrile containing 0.1% FA (v/v)) ranging from 3% to 17% B in 30 min and from 17% to 40% B in 20 min (v/v) at a flow rate of 300 nL/min. Full MS scans were acquired from m/z 350 to 1500 at a resolution of 70 000 with an automatic gain control (AGC) target value of 1 × 10⁶ and a maximum injection time of 50 ms. The 15 most intense precursor ions (charge states from +2 to +8) were isolated with a window of m/z 1.2 and fragmented using a normalized higher energy collisional dissociation (HCD) energy of 28 and a dynamic exclusion of 40 s. The MS/MS spectra were acquired at a resolution of 17 500 using an AGC target value of 2 × 10⁴ and a maximum injection time of 64 ms.

LC-MS data were processed using the Proteome Discoverer 2.4 software suite (Thermo Scientific) including the Sequest search engine and the Minora chromatographic feature detection nodes. Data were searched against a custom-generated database containing only the p24 and mAb E amino acid sequences. The following search parameters were used: precursor tolerance 5 ppm; MS/MS tolerance 0.02 Da; allowable variable modification of oxidation (M), SDA–OH (any residue), SDA–NH2 (any residue), and SDA–i (N terminus and K). The fixed value peptide spectral match (PSM) validator setting of max delta Cn was 0.05, and only first-rank PSMs were used for assigning modification sites within peptides. The label-free quantitation (LFQ) workflow of Proteome Discoverer was used to identify residues that were differentially modified between free p24 and the p24–mAb E complex samples (Supporting Information Figure S2). Precursor chromatographic peak areas were used to calculate peptide abundance ratios. Peptides with log₂ ratios exceeding mod (2) (i.e., <0.25 and >4-fold) were considered to be differentially modified. SDA cross-linking data were searched with a Kojak search engine (v.1.6.1)¹² and by the DXMSMS Match program.¹⁰ DSS cross-linking data were searched using the Xlinx node of the Proteome Discoverer with a fixed value FDR setting of 0.05 for target-decoy validation.

Molecular Visualization

Visualization of the experimental results at the molecular level was facilitated by supplementing the crystal structure of p24 (PDB ID: 6AYA) with the missing C-terminal residues 221–232 using PyMOL (v.2.2.3) by Schroedinger LLC.^13,14 The average mass of intact p24 was determined to be 25 571 Da, which agreed with the absence of the N-terminal methionine (Met) residue (as was confirmed by the presence of the Pro-terminated N-terminal peptide). The quantified surface-modified residue side-chain atoms were visualized as spheres, while the DSS-cross-linked residue side chains were visualized as sticks. The quantified surface-modified residue side-chain atoms were colored according to the log₂ ratios of the precursor ion abundance as determined by LFQ analysis using a blue–white–red palette. Minimum and maximum log₂ p24/p24–mAb E ratio values were set to −4 and +4 for blue and red colors, respectively. Colors for ratio values exceeding the range from −4 to +4 were set to blue or red. A homology model of the Fab domain of mAb E was built using the SwissModel server.¹⁵ The L- and H-chain sequences were used separately for sequence homology searches using BLAST.¹⁶ DSS cross-links were visualized as lines connecting NZ atoms of the lysine residue side chains. Differential bottom-up HDX protection values were visualized using a blue–white–red palette with red set as the maximum observed Δ% of the corrected deuteration values of the peptic peptides. The mAb E Fab domains were manually positioned in proximity to the p24 molecule in an orientation which satisfies the SM and CL data.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the data set identifier PXD040902.¹⁷

NOTE: Reviewer account details: Username: reviewer_pxd040902@ebi.ac.uk; Password: 6zXKCj5e.

Results and Discussion

The formation of a protein antigen–antibody complex results from complementary interactions of the amino acid side chains of both proteins at the protein interaction interface. Amino acid residues involved in the binding become partially or completely buried at the interface and are excluded from water contacts upon complex formation. Moreover, upon antibody binding, the structure of the surface-exposed interacting region of the protein antigen is stabilized due to restriction of the motion of the amino acid side chains and the stretches of the protein sequences that are involved in the interaction. Each of these changes can be detected by different structural proteomics methods, thus delineating the location of the protein antigen surface (epitope) bound by the antibody. Here, we have applied limited proteolysis, photoreactive surface modification, cross-linking, and HDX analyses for the determination of the p24 epitope of the mAb E monoclonal antibody.

For epitope extraction/excision analyses, the mAb E monoclonal antibody was covalently immobilized on the epoxy-activated magnetic beads and was allowed to interact with either a trypsin or a pepsin digest of the p24 protein (epitope extraction) or with intact p24 protein followed by on-bead digestion of the bound antigen with trypsin (epitope excision). The peptides bound to the antibody were identified by LC-MS analysis. Several tryptic and peptic p24 peptides were detected located at the N terminus and in the middle of the protein sequence (Figure 1).

Figure 1.

Epitope extraction/excision with immobilized mAb E antibodies. (A) Tryptic and peptic peptides found to be bound to the immobilized mAb E antibodies. (B) p24–mAb E complex visualization. p24 peptide sequences found are highlighted in color. The mAb E Fab model is colored in red to blue from the N-to-C terminus for each light and heavy chain.

For differential surface modification analysis, pairs of samples containing free antigen and antigen–antibody complexes were modified with the heterobifunctional amino group- and photo-reactive cross-linking reagent SDA and were digested with trypsin in parallel. The reaction conditions used resulted in an average of 1–2 modifications per protein molecule (as determined by intact protein mass analysis), most likely preventing possible significant conformational changes due to the chemical modification itself. The resulting peptides were analyzed by LC-MS, and modified peptide abundances were compared between samples using the label-free quantitation methodology of the Protein Discoverer software. Photoreactive surface modification data were extracted from the SDA cross-linking data sets as dead-end modifications with the hydrolyzed SDA reagent or as intrapeptide SDA cross-links. Approximately ∼50–80 modified peptides were identified per sample; ∼6 −10 preferentially modified residues in the free-antigen samples suggested involvement in the formation of the epitopes (Figure 2).

Figure 2.

Differential surface modification of free and mAb E-bound p24 with photoreactive reagent SDA. SDA-modified residue side-chain atoms of the p24 protein are shown as spheres colored using a blue–white–red palette according to the observed SDA-modified p24 peptide abundance ratios for free p24 and p24–mAb E complex samples (blue, preferentially modified in the free p24 sample; red, preferentially modified in p24–mAb E sample).

Unfortunately, no short-range SDA interprotein cross-links were detected in these analyses. This was, however, not surprising as the formation of interprotein SDA cross-links requires the presence of lysine or an N-terminal residue in close proximity to the interaction interface, which is rarely the case. When cross-linking was performed on p24–mAb E complexes with the DSS long-distance cross-linking reagent, however, several interprotein p24 mAb E cross-links were found (Figure 3). These cross-links helped to determine the orientation of the antibody molecule toward the protein antigen, indicating the epitope region on the protein antigen’s surface.

Figure 3.

p24–mAb E complex cross-linking with the long-distance cross-linking reagent DSS. The detected interprotein p24–mAb E cross-links are shown as dotted yellow lines.

For differential HDX analysis, samples of free p24 and of p24 complexed with the mAb E antibody were exposed to a D₂O solution for varying time intervals, and the level of the deuteration was measured by mass spectrometry. Intact protein HDX analysis of free and mAb E-bound p24 samples revealed a moderate but measurable increase in protection of ∼15 protons at a 20 s exchange time, indicating the formation of the complex (Figure 4A). We have shown previously that HDX protection at this time point gives a good approximation of the extent of the secondary structure present in the protein molecule.¹⁸ These results were used as justification of the subsequent bottom-up HDX analyses for locating the differentially protected residues on p24 upon interaction with the antibody. Differential bottom-up HDX analysis of the free and mAb E-bound p24 samples revealed several peptides that exhibited increased protection against exchange upon formation of the p24–mAb E complex, suggesting that these peptide residues were probably involved in the binding of the epitope (Figure 4B). The regions which exhibited the largest changes in protection upon mAb E binding involved loops which were flanked by α-helices. This is to be expected upon formation of the protein complex, as protein interaction leads to the stabilization of the structure involved in the formation of the protein interaction interface, including stabilization of the secondary structure elements (peptide bond amide hydrogen bonding), which is subsequently reflected in an increase in HDX protection of the corresponding residues.^19,20

Figure 4.

Differential HDX analysis of free and mAb E-bound p24. (A) Intact protein HDX analysis. Deconvoluted intact protein mass spectra of p24 from top to bottom: no HDX, 20 s HDX of free p24, 20 s HDX of mAb E-bound p24. Increase in protection of ∼15 protons is observed upon p24–mAb E complex formation. (B) Bottom-up differential HDX analysis. Representative kinetic plots of p24 peptic peptides HDX from left to right: differentially protected peptide upon complex formation, similarly protected slow-exchanging peptide from the structured region of the protein, and rapidly exchanging peptide from the disordered region of the protein. (C) Bottom-up differential HDX analysis. Differences in deuteration of the peptic p24 peptides between free and mAb E-bound p24 samples. Values are presented in a blue–white–red palette, with red representing higher protection values in the p24–mAb E sample compared to the free p24 sample. Bars represent peptides. Slices of bars from top to bottom correspond to 0, 20, 40, 80, and 160 s HDX time points. (D) Visualization of differences in deuteration between free and mAb E-bound p24 are presented on the three-dimensional structure of p24 using a blue–white–red palette, with red representing higher protection values in the p24–mAb E complex sample compared to the free p24 sample.

In the current case, we did not observe any significant changes in HDX protection to other parts of the p24 molecule, which could be attributed to the absence of significant allosteric conformational changes upon complex formation (Figure 4C).

From a practical standpoint, combining different structural proteomics techniques does not dramatically increase use of the protein material, as SM, CL, and LP proteolysis requires only subtle microgram amounts of protein for the analyses. We found that analyses generally do not require replicates, besides single time point intact protein mass HDX analysis (to estimate the method variability, which in our case was found to be quite low, <2% CV), to reach clear differential conclusions. Standard instrument-specific (Protein Discoverer (Thermo), PeakView with BioToolKit (Sciex), or freely available (MassSpecStudio) software can be used for data analysis. When all of the methods are set up and optimized, complete combined analysis can be completed in a matter of days.

All of the limited proteolysis, differential surface modification, cross-linking, and differential HDX results agreed well with each other. Visualization of the experimental data on the 3D structure of the p24 allowed the unambiguous localization of the epitope at the N-terminal portion of the molecule. Comparison of the results from each method in our case illustrates how certain small pairwise discrepancies between the methods can be resolved taking into consideration results from additional structural proteomics approaches. For example, the epitope excision/extraction approach identified additional peptides from the middle of the protein sequence bound to the antibody. The peptides are located on the opposite surface of the protein and can remain bound to the antibody following limited proteolysis as part of the residual structure of a helical bundle. Those can be confidently ruled out from being in the epitope protein surface by all three additional methods. Similarly, a lack of coverage in the bottom-up HDX analysis of the sequence between helices 3 and 4 did not highlight this loop as a part of the epitope, while other methods clearly indicated it as part of the protein interaction interface. Each method has its own strengths and weaknesses, so combining multiple approaches results in a balanced and confident identification of the protein interaction interface. Having an experimental or predicted 3D structure of the protein antigen along with the Fab fragment of the antibody is certainly helpful in the interpretation of the experimental data and can serve as a basis for synthesizing the conclusions from multiple techniques into final overall boundaries of the determined epitope. We have used this approach for numerous other protein antigen–antibody complexes with successful outcomes, and we have determined that this combined approach can be used for confident and comprehensive epitope mapping of protein antigens.

Conclusions

Limited proteolysis, photoreactive surface modification, cross-linking, and HDX provide orthogonal experimental results for the epitope determination of monoclonal antibodies. This combination of the multiple structural proteomics methods allows unambiguous and confident determination of the protein antigen epitopes for monoclonal antibodies.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jproteome.3c00159.

• Epitope excision/extraction analysis of the p24–mAb E interaction by LC-MS using an Orbitrap mass spectrometer; label free quantitation differential surface modification analysis of the p24 and p24–mAb E samples in Proteome Discoverer; differential intact protein mass LC-MS HDX analysis of free and mAb E-bound p24 (PDF)

Author Contributions

E.V.P.: Conceptualization, methodology, writing-original draft, project administration. E.N. and J.M.W.: Resources, writing-review and editing, supervision, funding acquisition. C.H.B.: Resources, supervision, funding acquisition

The authors declare the following competing financial interest(s): C.H.B. is the CSO of MRM Proteomics, Inc. and the VP of Proteomics at Molecular You. C.H.B. and E.V.P. are the cofounders of Creative Molecules, Inc., which develops novel cross-linkers and provides them at cost to researchers.