Planning HDX-MS Epitope Mapping for Large Antigens: Peptide Coverage, Protease Strategy, and Complex Quality

HDX-MS epitope mapping of a large, multidomain antigen is most useful when the experiment can measure the suspected binding region in both the free and antibody-bound states. A high overall sequence-coverage percentage is not enough if a glycosylated domain, disulfide-rich segment, or flexible linker remains unobserved. Equally, a decrease in deuterium uptake identifies a binding-associated change in backbone exchange, not necessarily a direct antibody contact.

Planning therefore starts with three practical questions: which regions can be measured reproducibly, which antibody format maintains a soluble complex, and what additional evidence will be needed to interpret the protected peptides. These questions apply to large soluble antigens, extracellular receptor constructs, viral glycoproteins, and oligomeric assemblies. Protein length increases analytical complexity, but there is no amino-acid threshold at which conventional HDX-MS automatically fails.

Sequence Coverage in Large-Antigen HDX-MS

Construct Boundaries and Domain-Level Coverage

Start with the sequence of the material that will actually enter the instrument. Record mature-chain boundaries, signal-peptide removal, purification tags, linkers, engineered mutations, cleavage products, and known post-translational modifications. Numbering should be traceable to both the submitted construct and the full-length reference sequence. For an oligomer, distinguish the length and mass of one chain from those of the assembled complex.

Map the initial peptide-identification results onto this annotated sequence before selecting an antibody panel. Assess coverage separately for the proposed binding domain, neighboring domains, and regions needed to evaluate a conformational change. If the antibody-binding site is unknown, a gap in any accessible domain may remain consequential even when the whole-protein average appears satisfactory. In a focused HDX-MS epitope mapping study, the feasibility assessment should establish which regions can support a comparison and which require further method development.

A domain construct can be a useful follow-up when full-length coverage is poor, but it is not an interchangeable substitute. Truncation can alter domain packing, glycan presentation, oligomerization, or the epitope itself. Confirm that the shortened construct retains antibody binding and the relevant biochemical properties; interpret its HDX map in the context of the full-length protein rather than transferring the result without qualification.

Peptide Identification and Local Resolution

Use non-deuterated antigen, antibody, and complex samples to establish peptide identities and identify overlapping signals. The search database should contain the actual antigen and antibody sequences where available, the protease, and relevant contaminants. A peptide observed in the antigen-only sample may become difficult to quantify in the complex because antibody-derived peptides co-elute with it. Coverage used for differential analysis must therefore reflect peptides that can be measured in the compared states, not simply every sequence identified during method development. [1]

Overlapping peptides improve localization when their uptake measurements are reliable and mutually consistent. However, shared residues do not make every peptide an independent experiment, and subtracting the uptake of two overlapping peptides does not automatically produce a residue-resolved epitope. Different back-exchange, signal quality, exchange behavior, and uncertainty can complicate that subtraction. Report the smallest region supported by the observed peptide boundaries and data quality.

There is no universal regulatory requirement for 85% coverage or a redundancy factor of 3.5 in every HDX-MS epitope study. Community recommendations emphasize reporting coverage, overlap, deuterium recovery, precision, and the basis for significance. Project-specific acceptance criteria should be agreed before the main study, with explicit treatment of missing regions. A peptide that cannot be measured is an unresolved region, not evidence that an antibody does not bind there. [1]

Figure 1. Local peptide coverage and complex quality are assessed separately when planning large-antigen HDX-MS. A missing region remains unresolved even when other domains are well represented.

Protease Selection and Quench Conditions

Pepsin and Complementary Acid Proteases

Pepsin remains a practical starting point for many HDX-MS workflows. Its performance depends on substrate sequence, residual structure, digestion conditions, and the specific immobilized preparation. Large antigens do not all exhibit the same coverage loss. When a critical region is missing, first distinguish incomplete digestion from poor peptide recovery, chromatographic interference, incorrect identification, and inadequate signal.

Complementary acid proteases can change the peptide map rather than simply add more peptides. Nepenthesin-based digestion is supported by experimental studies, but nepenthesin I, nepenthesin II, natural extracts, and immobilized enzyme preparations should not be treated as identical reagents. Kadek and colleagues demonstrated that immobilization and quench additives affect nepenthesin-1 performance. Yang and colleagues characterized recombinant nepenthesin II, including its broader cleavage behavior and chemical stability. [2,3]

In a systematic comparison involving 11 proteins, Zheng and colleagues found distinct cleavage preferences for immobilized porcine pepsin and nepenthesin II, including useful cleavage at basic residues with nepenthesin II. This supports testing an alternative enzyme when the relevant sequence is poorly represented; it does not establish a guaranteed coverage increase for every large antigen. [4]

Compare alternative digestions separately before deciding whether to combine results. Parallel enzyme workflows can preserve a useful long peptide in one map and provide shorter overlapping peptides in another. Sequential digestion may also be evaluated, but additional cleavage can produce fragments that are too short, poorly retained, or difficult to assign uniquely. Protease type XIII and other acid-active enzymes are additional candidates when a suitable preparation and validated workflow are available; none is mandatory solely because the antigen is large.

Quench Buffer, Disulfide Reduction, and Peptide Recovery

Quenching commonly uses acidic conditions near pH 2.5 and low temperature to slow backbone exchange. This is distinct from the labeling reaction, which is conducted under the conditions selected for the biological comparison. Cooling the post-quench workflow does not mean that all proteins should be labeled at 0°C. The actual labeling temperature, pH, deuterium fraction, and timing must be matched between states and reported. [1]

TCEP and a denaturant such as guanidine hydrochloride may improve access to disulfide-rich regions after labeling, but reduction is not automatically rapid or complete under cold acidic conditions. Screen the final mixed-sample concentrations, contact time, and compatibility with the selected enzyme and LC system. Distinguish a concentrated quench stock from its concentration after mixing with the labeled sample. A literature formulation is a starting point for evaluation, not a universal prescription for an antigen of a particular size.

Increasing denaturant or reductant can change protease activity, substrate solubility, peptide trapping, and downstream recovery. A condition that produces more identifications may still be unsuitable if it increases carryover or reduces deuterium retention. Evaluate candidate buffers using the same digestion, trapping, and separation conditions planned for the comparative experiment. The balance between unfolding, reduction, recovery, and deuterium retention is more informative than the highest additive concentration. [1,2]

Glycosylation and Chromatographic Interference

Glycosylation can distribute a peptide signal across multiple glycoforms and change proteolysis, retention, and identification. Annotate glycosylated regions in the coverage map and determine whether missing data concern the proposed epitope. A glycopeptide that is identified in a non-deuterated run may still be unsuitable for reliable uptake measurement because its isotopic envelope overlaps another species.

Deglycosylation is an option, not a routine requirement. Removing glycans before labeling can alter folding, accessibility, or antibody affinity; such an experiment describes the treated antigen. If it is used, confirm structural integrity and binding, and identify which glycans were removed. Enzymatic treatment after labeling must instead be compatible with label preservation and sample workup. PNGase F should not be assumed to function as an ordinary online protease at cold acidic quench conditions.

For crowded chromatograms, optimize the load, separation, and extraction of individual peptide envelopes rather than assuming that the shortest gradient gives the best result. Longer separation may improve measurement of a critical region while increasing back-exchange. Higher resolving power can help separate some interferences, but it cannot recover an ion that was suppressed before detection. Compare these trade-offs experimentally within the HDX-MS workflow.

Figure 2. Candidate digestion and quench conditions are compared using usable peptide coverage, recovery, interference, and deuterium retention. Parallel protease maps need not be implemented as sequential digestion columns.

Antibody Format, Aggregation, and Binding Occupancy

Intact IgG, Fab, and Complex Stability

Intact IgG is not intrinsically unsuitable for large-antigen HDX-MS. Whether it forms heterogeneous assemblies depends on antigen valency, epitope accessibility, stoichiometry, concentration, and solution conditions. Cross-linking becomes a concern when the binding geometry permits extended networks; a large molecular mass alone does not prove that precipitation will occur. Inspect both antigen alone and the complex over the concentration and incubation ranges used during labeling.

A monovalent Fab can reduce avidity-mediated cross-linking and the amount of antibody-derived peptide background. However, changing antibody format may alter apparent affinity or complex stability. Confirm the Fab's purity, binding competence, and behavior under the intended conditions instead of assuming that format conversion guarantees a monodisperse complex.

Fragment preparation must also produce the intended valency. Papain digestion or recombinant expression can provide Fab with appropriate purification and verification. IdeS cleavage normally produces F(ab′)₂, which remains bivalent; removing Fc is not equivalent to generating a monovalent Fab. Any additional hinge-reduction or processing step must be evaluated for structural and binding effects. [7]

Analytical SEC is useful for detecting higher-mass species, dissociation, and sample changes. SEC-MALS can add molecular-mass information, while DLS can screen for large particles. These techniques are complementary: SEC can dilute a weak complex during separation, and a DLS distribution does not identify each molecular species. Neither an isolated SEC peak nor a low scattering polydispersity value alone establishes antibody occupancy during the HDX labeling reaction.

Occupancy After Dilution into Labeling Buffer

Calculate binding conditions using the concentrations after D₂O addition. A complex prepared at high concentration may dissociate when it is diluted for labeling. The outcome depends on affinity, binding-site stoichiometry, and equilibration or dissociation kinetics; a fixed 1.5:1 or 2:1 antibody-to-antigen ratio cannot by itself guarantee 90% or 95% occupancy.

For illustration, consider a simple 1:1 interaction with final antigen and Fab concentrations of 1 and 2 µM. At a K_D of 1 µM, equilibrium antigen occupancy is approximately 59%, despite the twofold Fab excess. At a K_D of 0.01 µM under the same concentration conditions, it is approximately 99%. These are mass-action calculations, not experimental performance claims. Multivalent or heterogeneous systems require a more appropriate binding model.

Use available antibody binding-kinetics measurements to plan the concentration range and interpret dissociation risk. Measurements in H₂O buffer are useful inputs, but should not be assumed to describe every deuterated labeling condition without checking compatibility. Increasing Fab concentration can improve occupancy while also adding peptide interference, sample demand, and aggregation risk. A small concentration series can identify a workable compromise.

Reduced occupancy can diminish the measured difference between free and bound states, but a uniform linear correction is not always valid. Resolved isotope populations, binding kinetics, and exchange regimes can make the observed envelope more complex than a single diluted centroid shift. Retain and inspect the underlying spectra when broadening or bimodality appears.

HDX Labeling Times, Controls, and Replicates

Select labeling times that capture measurable exchange in the regions of interest. An illustrative screening series might include 10 s, 1 min, 10 min, 1 h, and several hours, but the appropriate range depends on the protein and the question. Rapidly exchanging segments may require shorter times, while stable domains may require longer ones. Do not force every antigen into an identical time course or extend an incubation beyond the period over which the sample remains stable. [1]

For each state, maintain the same labeling-buffer composition, temperature, deuterium fraction, handling, and quench timing. Record concentrations before and after mixing. If a stock or formulation changes during optimization, repeat the relevant baseline measurements rather than combining uptake values collected under different solution conditions.

• Non-deuterated samples: establish peptide identity, retention, and undeuterated mass for antigen, antibody, and complex.
• Free-antigen and bound-antigen measurements: provide the matched comparison using common, interpretable peptides.
• Antibody-only measurements: identify potential peptide interferences; they do not justify automatic subtraction of overlapping signals.
• Maximally labeled or system-suitability controls: characterize deuterium recovery and back-exchange using a documented preparation. If complete labeling cannot be established for the target, explain the limitation and the role of the model control.
• Blanks and carryover checks: test whether the complex, hydrophobic fragments, or protease-derived peptides persist into later injections.

Technical replication should involve independently initiated labeling reactions. Repeated LC-MS injections of the same quenched reaction assess a different component of variability and do not replace reaction replicates. The community recommendations call for at least three independent labeling reactions at a minimum of one time point to estimate error, with replication across additional conditions and times selected according to the study. Randomize or interleave free and bound measurements to reduce confounding by instrument drift. [1]

Back-exchange should be characterized for the actual workflow. A mean percentage can conceal poorly retained deuterium in a critical peptide, so inspect the distribution and relevant local measurements. Similarly, a single global precision value should not conceal unresolved envelopes or poor repeatability in the suspected binding region.

HDX-MS Data Analysis and Epitope Interpretation

Differential Uptake and Statistical Significance

Define the direction and units of the reported difference. If ΔD is calculated as uptake in the bound state minus uptake in the free state, negative values indicate protection. An absolute difference in daltons and a percentage difference in exchange are different quantities; a threshold of 0.5 Da must not be labeled as 0.5%.

Choose the statistical analysis and minimum meaningful difference using the experiment's precision, replication, peptide behavior, and study objective. A combined effect-size and significance display can help prioritize changes, but there is no universal 0.5 Da, p < 0.01 rule that proves an epitope for every system. Document treatment of multiple comparisons, related time points, missing values, and overlapping peptides where relevant. Avoid counting several correlated observations as independent confirmation.

Review candidate changes against isotope envelopes, charge states, retention, signal intensity, and the non-deuterated map. A protected peptide surrounded by unmeasured regions is not automatically false, and a cluster of protected peptides is not automatically a direct interface. Overlap strengthens localization only when the contributing measurements are valid and the interpretation accounts for the protein structure.

Direct Binding and Allosteric Protection

HDX protection can reflect direct binding, a strengthened hydrogen-bond network, reduced local unfolding, domain rearrangement, or changes in oligomerization. Increased uptake can likewise accompany altered dynamics without marking an antibody contact. Conventional bottom-up HDX-MS usually provides a peptide-level comparison of exchange, not an atom-by-atom contact map.

Time dependence is informative, but early protection does not prove a direct contact and delayed protection does not prove allostery. In a pre-equilibrated complex, an allosteric change can already exist before the first labeling time point. Conversely, a directly contacted peptide may exchange slowly and show its largest difference only later. The two mechanisms cannot be assigned from a simple early-versus-late curve rule.

Deng and colleagues investigated a specialized millisecond workflow in which antibody binding and labeling began together, attenuating some allosteric contributions in a myoglobin model. That experimental design differs from a conventional pre-incubated complex measured over seconds to hours. Its findings should not be converted into a universal rule that a standard 10-second measurement identifies direct epitopes. [5]

Map significant changes, non-significant measurements, and missing coverage as distinct categories on the sequence and any suitable structure. A predicted structure can help inspect spatial relationships, but model confidence, domain orientation, and unresolved segments must remain visible. A colored surface generated from HDX data is a visualization of the measurements, not independent validation.

Mutagenesis and Orthogonal Structural Measurements

Choose follow-up experiments according to the unresolved question. Site-directed mutagenesis with binding analysis can test whether a candidate region contributes to the interaction, but loss of binding must be separated from loss of folding, expression, or assembly. Include structural or functional checks of the mutant and, where practical, binding by an antibody recognizing an unaffected region.

A published factor VIII study illustrates the value of combining approaches. HDX identified KM33-associated protection in two C1-domain regions; the investigators then used a glycosylation-site variant, binding measurements, and functional assays to examine the role of that region. This is a target-specific example of HDX-guided follow-up, not a general coverage benchmark or a demonstration that any protected residue is a direct energetic hotspot. [6]

Cross-linking mass spectrometry can add residue-pair proximity restraints, while electron microscopy or crystallography can address overall binding geometry or detailed structure when feasible. Cross-linker chemistry determines which residues are sampled; DSSO and DSBU are commonly amine-reactive and should not be described as interchangeable tools for directly linking every acidic side chain. Cross-link absence does not exclude a contact. Interpret restraints using the reagent, site-assignment confidence, and structural context rather than a universal distance cutoff.

These complementary measurements should address a specific ambiguity: whether distant protected regions form one discontinuous surface, whether a mutation destabilizes the antigen, or whether the antibody changes assembly state. Adding more techniques without defining that ambiguity can increase sample use without resolving the central interpretation.

Figure 3. Differential HDX identifies binding-associated protection. Uptake curves and structural mapping guide interpretation, while orthogonal experiments address whether protected regions contribute directly to the binding interface.

Feasibility Testing Before a Large-Antigen HDX-MS Study

A pilot should answer whether the intended comparison is measurable before the complete antibody panel and time course are committed. The most useful acceptance criteria are specific to the target and the proposed conclusion. The following checks connect each development task to a decision about proceeding.

Build the sample budget from the actual design: free antigen, antibody-only development runs, complex preparations, candidate digestion conditions, labeling times, independent reactions, controls, blanks, and material lost during buffer exchange. A small non-deuterated development injection is not representative of the total sample requirement. When several antibodies share one antigen lot, preserve enough material to repeat the common baseline and investigate a failed complex.

The final comparison should identify the exact construct, labeling conditions, common peptide set, local coverage gaps, uptake curves, variability, significance criteria, and the structural interpretation of each protected region. Distinguish an unmeasured segment from one with no detectable difference within the sampled time range. This is particularly important for a large antigen, where an apparently clear global map can conceal a decisive local gap.

For project assessment, provide the antigen and antibody sequences, construct boundaries, formulation, available binding measurements, known modifications, sample quantity, and the question the epitope map needs to resolve. These inputs support a focused pilot and a realistic choice between full-length mapping, qualified domain studies, and orthogonal confirmation.