High-throughput peptide–HLA affinity mapping has graduated from academic novelty to become the bottleneck through which all current day T-cell vaccines, cellular therapies or auto-immunity models must pass. Our platform interrogates the HLA groove as a nano-scale calorimeter: each peptide is challenged for its ability to displace a fluorescently labelled reference ligand at full physiological ionic strength and redox potential, thus producing a competitive binding curve that is robust to the optical artefacts that are common to traditional pulldown methods. The use of soluble, truncated HLA heavy chains refolded with recombinant β2-microglobulin effectively removes the effect of membrane proximal residues and restores true solution-phase kinetics. The resultant data set (ranked by half-maximal displacement rather than a single-point intensity value) is then processed by neural-network based algorithms to produce population coverage across all twelve HLA supertypes, reducing what would have previously taken months of serial dilutions to an afternoon of unattended read-outs. The final output given to investigators is a ranked list of epitopes whose potential immunological relevance is no longer inferred from an in-silico score but is instead based on experimentally-derived groove occupancy, offering a regulatory-ready explanation for an otherwise arbitrary selection of peptides to be used in animal challenge or even first-in-human dosing.
Peptide-HLA binding is the final arbiter in whether a T cell will ever "see" its cognate antigen, mediated by an extended network of hydrogen bonds, anchor-pocket complementarity and solvent-mediated contacts that collectively define both affinity and half-life. Single substitutions at primary anchor residues can profoundly increase stability and thus density of complexes on the cell surface, or result in a complete loss of presentation. The TCR scans the peptide backbone and exposed α-helices of the HLA heavy chain simultaneously, so even small changes to peptide conformation can reorganise the topological environment that is presented to the immune system. The HLA–peptide interface is therefore not a static structural motif which tunes the threshold between activation and tolerance, but a dynamic molecular synapse. Understanding the nature of this interaction is therefore critical to determining whether a viral or tumour-derived sequence is likely to induce protective CD8⁺ or CD4⁺ responses, to identifying escape mutations that subvert antigen processing and to designing altered peptide ligands that modulate cytokine polarity or induce clonal anergy. In short, the biological relevance of any epitope is underpinned by the physical interaction between peptide and HLA, and its detailed characterisation is thus critical for rational immunotherapy and vaccine design.
Fig. 1 Models of HLA-II-AAPCs for the expansion of human antigen-specific CD4+ T cells.1,5
To a T cell, the pHLA complex is the elementary information unit for recognition. The TCR interrogates the entire electronic and steric signature of all side chains that are accessible in the complex, not just the surface as a whole, and it converts the chemistry of the peptide into intracellular signals. As the majority of TCRs are in diagonal docking mode, their CDRs make allele-independent contacts with conserved residues on the HLA α-helices as well as contacts with peptide residues that stick out of the groove. This dual read-out places tight restrictions on the peptide's properties: on the one hand, it must be sufficiently flexible to allow the TCR to adapt its conformation, but on the other hand, it must have a unique pattern of solvent exposed atoms to be recognized as non-self. The effect is then fine-tuned by co-receptor binding. CD4 or CD8 engagement stabilizes the TCR-pHLA complex and brings the activation threshold down by several orders of magnitude. This means that small changes, like the addition of a methyl group through a post-translational modification, can mean the difference between full and partial agonism, or even lead to antagonism. The same peptide with two different HLA allotypes can have qualitatively different cytokine signatures because the change in groove geometry reorients the peptide backbone and reshapes the epitope presented to the TCR. So the importance of HLA-peptide interactions for T cell recognition cannot be overstated: knowing it is to understand the grammar of adaptive immunity, and how structural degeneracy at one level of organization gets translated into functional specificity at another, and how certain pHLAs come to dominate immune responses while others are immunologically invisible.
A critical component of rational vaccine design is a priori selection of epitopes that can be presented effectively throughout the genetically heterogeneous human population. As HLA are the most polymorphic genes of the human genome, a peptide that binds with nanomolar affinity to one allotype may not be capable of binding to another. HLA-peptide screening services thus conduct allelic deconvolution, experimentally determine binding capacity for each prevalent HLA variant, and build poly-epitope combinations that can provide population coverage while not compromising on inherent immunogenicity. In addition to binding, peptide stability within the groove is another critical factor that influences the duration that APCs are able to sustain surface expression on their surface and in turn the magnitude and persistence of the resulting T-cell response. Another factor to be considered is conservation of sequence; under immune selection, natural sequence variants may arise. Incorporating conserved regions of the sequence, for which structure is constrained by either viral fitness, or vital enzymatic function, lowers the probability of epitope escape. A final consideration is the processing efficiency of the peptides by the antigen-presenting machinery. It is thus required that they are flanked with appropriate proteasomal cleavage motifs while not containing glycosylation sites or secondary structural motifs that inhibit TAP transport. When these multiple parameters—binding, stability, conservation, and processing—are considered, HLA-peptide screening platforms can be used to build epitope-rich immunogens that induce broad, poly-functional T-cell responses while minimising off-target reactivity against self or microbiota-derived homologues.
The prediction of immunogenicity of a specific peptide within an in vivo immune system environment involves multi-scale computational modelling, which starts with quantum-chemical simulations of anchor residue interactions and culminates with in silico modelling of clonal selection. At the atomic level, machine-learning tools based on large sets of experimental affinities and half-lives are used to predict the likelihood of a peptide to form a stable complex with each HLA allotype, which is then adjusted with a probabilistic model that considers proteasomal cleavage propensity, TAP binding efficiency, and endosomal pH sensitivity to generate an estimate of the mature epitope intracellular concentration. The subsequent layer incorporates TCR recognition likelihood, which is a much more unpredictable parameter as it relies on the inherent germ-line-encoded bias of the TCR repertoire, as well as the degree of central tolerance mechanisms. Structural bioinformatics is applied to model the solvent-accessible surface of the pHLA complex and is matched against a library of TCR footprints, to identify peptides whose predicted topology is different from self-derived references, and therefore have a higher likelihood of cross-reactivity. Population genetic principles are used at the final level to account for allele frequency, linkage disequilibrium and demographic history, to predict the proportion of the population that will express the peptide. The result is a probabilistic score that is used to rank each predicted epitope based on its potential immunogenicity.
Our service offering has been built around a vertically integrated platform linking high-throughput biophysical assessment of peptide–HLA interactions to follow-up immunological validation and, where appropriate, cloud-based analysis. A frozen inventory of properly folded, peptide-binding ready class I and class II heterodimers provides a no-middle-man chain of custody from protein sequence to ranked, functionally validated epitopes. Operated on a 24/7 schedule, the suite of liquid-handling robots, microplate thermal cyclers and high-resolution imagers are interwoven with custom data ingestion layers engineered to stream "as-run" data through feature extraction, automatic unit testing and complex event processing steps before curation by a dedicated scientist. Even though a single client request may span only a few tumor-specific variants or a complete viral proteome, the same core infrastructure is available to exploit: first-pass computation often prunes down candidate sets, high-throughput binding assays measure association and dissociation kinetics, orthogonal T-cell validations can confirm immunogenicity, and customisable interactive visualisations empower immediate inspection of kinetic parameters, allele coverage or potential cross-reactivities. Adherence to standard operating procedures written to established best-practices should mean our work meets users' needs for reproducibility, transparency and data quality; however, the platform itself is fundamentally extensible so that, for example, additional instrumentation, novel HLA alleles or updated sequence-to-structure mapping algorithms can be added at will.
Instead of a single binding read-out, we have devised a matrix of equilibrium, kinetic and stability measurements that, when integrated, predict peptide stability of display at the cell surface. The HLA heavy and light chains are expressed in eukaryotic hosts to ensure native glycosylation, biotinylated under mild conditions that do not sterically occlude the peptide binding groove, and pre-loaded on streptavidin-coated plates whose optical transparency is suited for both FP and TR-FRET. The same physical interaction can thus be interrogated by two independent read-outs, while a custom-built scheduling algorithm randomises plate position, injection order and incubation temperature to reduce positional drift and edge effects that have historically compromised the robustness of high-density screens.
Fig. 2 Approaches to interrogating differences between killer-cell immunoglobulin-like receptor (KIR)–human leukocyte antigen I (HLA-I) interactions.2,5
As in vitro binding does not ensure that a given peptide will be processed, presented and ultimately recognized by a polyclonal T-cell repertoire, each high-throughput experiment is encompassed by a validation pyramid. Its first layer is based on staining peripheral blood mononuclear cells from healthy donors whose HLA alleles match the predicted restrictors using stabilised peptide-exchange tetramers. Quantification of this output has three levels: the prevalence and phenotype of tetramer-positive CD8+ or CD4 populations (measured by 18-colour flow cytometry that captures simultaneously activation markers, memory subset distribution and exhaustion signatures); cell culture of magnetically enriched tetramer-positive cells with peptide-pulsed autologous dendritic cells, with or without cytokine polarization cocktails, followed by multiparametric intracellular staining that maps the granularity of effector functions like degranulation, cytokine synergy and chemokine receptor expression; and thirdly a transition to miniaturised tumour spheroid or viral replication models in which peptide-specific T-cell lines are co-incubated with cognate target cells. Impedance-based killing assays and single-cell RNA sequencing provide a kinetic read-out of cytolytic potency while simultaneously revealing the transcriptional programmes preceding target demise. Peptides with similar binding affinity but no predicted immunogenicity are interleaved at each step to estimate background noise, and positive-control peptides derived from well-characterised viral epitopes serve as internal calibrators. Only candidates that reproducibly elicit multifunctional, cytolytic or helper responses in at least two independent donor cohorts are elevated to the final recommendation list.
Raw affinity scores are not informative unless benchmarked against genomic, transcriptomic and immunogenetic context. As such, we have integrated a data analysis layer to ingest multi-omic information: A containerised workflow manager first maps exome or RNA-seq reads against the latest human reference genome, calls somatic mutations and encodes each variant as all possible nine- to twenty-five-mer peptides. These sequences are then evaluated in parallel by an ensemble of predictors that fuse classical machine-learning models with structure-based energy calculations, and the resulting vectors are integrated with any experimental binding data in a Bayesian framework that re-calibrates prior distributions using empirical data. Interactive heat-maps visualise how each peptide is ranked across alleles, and promiscuous binders that might provide broad population coverage can be immediately identified. The pipeline can also screen shortlisted peptides against the human proteome and environment-associated microbe databases to assess viral mimicry or auto-immune risk upon request from the client, and highlight sequences that surpass a user-defined identity threshold. Cloud-hosted notebooks make the codebase accessible, and version-controlled images ensure that every analysis remains reproducible months or years later. Finally, export modules can be used to generate standardised output files to adhere to the format required by clinical trial databases, regulatory submission templates and public repositories, so that the entire epitope discovery chain (from initial screen to final nomination) remains traceable, shareable and ready for downstream translational use.
We deliberately built our epitope-discovery platform as a general-purpose research tool, rather than as a pipeline targeting a particular disease indication. The fusion of high-resolution HLA ligand mapping, functional T-cell read-outs and cloud-based analytics allows the platform to be applied to onco-immunology, auto-immunity or infectious disease research questions without the need to re-engineer underlying core components. Experiment teams are provided with end-to-end solutions that start at any desired point of entry – bulk or single-cell RNA-seq reads, peptide libraries or patient cells – and end with validated sets of epitopes and an interpretive dashboard. Thus, the same workflow can be used by a tumour immunologist looking for neo-antigens, a rheumatologist interested in self-peptides that break tolerance, or a virologist studying viral antigens that might elicit cross-reactive memory. Traceable quality metrics, version-controlled software containers and regulatory-grade documentation are baked into each project from the start, so that all downstream translational goals – whether a personalised vaccine, a T-cell diagnostic or a mechanistic publication – are built on reproducible data.
The journey from antigenic source to T-cell activator is a series of bottlenecks: proteolytic release, TAP translocation, HLA loading, surface stability and finally TCR recognition, all of which can eliminate a potential epitope with single residue precision. We tackle the problem in parallel orthogonal steps. First, the sequence space is pruned by in silico predictors trained on mass spectrometry eluted ligandomes to peptides whose anchor motif is compatible with the groove chemistry of the restricting allele, these shortlists are synthesised as crude arrays and then interrogated by differential scanning fluorimetry to generate thermal stability signatures which map onto intracellular half-life. Peptides whose melting curves surpass allele-specific thresholds are subjected to multiplexed binding assays where association and dissociation curves are assayed in real time, providing kinetic confirmation of stable complex formation. However, biophysical stability is only half of the immunogenic story, the peptide also needs to present solvent exposed residues that carry a non-self signal. We thus leverage our in-house T-cell repositories - expanded from healthy donor blood or patient sample - whose TCR repertoire has been pre-characterised by single-cell sequencing. Peptide pulsed APCs are interrogated by tetramer staining, cytokine secretion profiling and impedance based cytolysis, allowing us to rapidly triage peptides that elicit multifunctional, high avidity responses from those that simply bind. The final output is a ranked catalogue where every epitope is annotated with binding affinity, thermal stability, processing likelihood and experimentally validated T-cell reactivity, providing researchers with a prioritised tool-kit for vaccine design or adoptive cell engineering.
Personalised cancer vaccines benefit from the intuitive hypothesis that mutations expressed exclusively in tumours will present neo-epitopes that are absent from the thymic selection environment and hence will escape central tolerance. Translating this hypothesis to the clinic, however, requires a scaleable but patient-tailored discovery pipeline that typically starts with next-generation sequencing of tumour-normal matched DNA, continues with RNA-seq validation of transcript abundance and results in a curated list of neo-epitopes with experimentally validated presentation likelihood and immunogenicity. Our pipeline automatically annotates non-synonymous single-nucleotide variants, indels and gene fusions to all possible nine- to twenty-five-mer frames and tests these peptides through the biophysical triage outlined above. Special consideration is given to clonal mutations that are present in all cancer cells rather than sub-clonal passenger mutations, since epitopes derived from these mutations are less likely to be subject to immune escape under selective pressure of vaccination. Neo-epitopes that pass binding and T-cell validation are selected for poly-epitope strings with order and spacing optimised to avoid junctional neo-junctions that could create unintended dominant epitopes. The strings are then synthesized as messenger RNA or long-peptide constructs and put through small-scale GMP-compatible production runs, including endotoxin testing, aggregation profiling and HPLC-based purity certification. All along this process, longitudinal blood draws from the patient are taken to capture baseline snapshots of the TCR repertoire, which can then be re-sequenced after vaccination to quantify clonal expansion, cytokine polarization, and acquisition of exhaustion markers. The resulting information closes the feedback loop and allows for iterative improvement of epitope selection for subsequent vaccine boosting or combinatorial use with checkpoint blockade.
Autoreactive epitope discovery reverses the ontological logic of oncology projects: whereas the starting point there is to identify tumor-exclusive mutations, the search for self-reactive T cells is instead predicated on finding self-peptides that have, in some way, managed to evade tolerance checkpoints that would otherwise have repressed them. In autoimmune studies, we start by single-cell sorting of clonally expanded T cells that have been isolated from the target tissue or from more accessible secondary sites such as cerebrospinal fluid, synovial fluid or skin lesions. We reconstruct and express the TCR α-β pair of the expanded clone in a reporter cell line that has a nuclear factor of activated T-cells (NFAT) fluorescent readout. We screen combinatorial peptide libraries presented by the patient's own HLA allotype and identify mimotopes of the recombinant TCR by their ability to trigger the NFAT response; the peptide recognition motif deduced from these mimotopes is used to query the human proteome for homologous self-sequences. Candidate peptides are synthesized and then tested for their ability to activate the original ex vivo T cell clone, now using physiologically processed antigen from the parent protein as presented by autologous antigen-presenting cells. Only peptides that are able to support activation under these natural processing conditions are accepted as bona fide autoimmune epitopes.
HLA-peptide interactions are central to immune recognition, making screening services essential for vaccine and immunotherapy research. Our HLA binding assays identify strong peptide candidates that can elicit T-cell responses, supporting epitope discovery and immune profiling. With advanced high-throughput screening, we provide accurate results that shorten research timelines and reduce risk in translational studies. These services enable researchers to identify, validate, and prioritize immunogenic peptides that have the highest potential for therapeutic or vaccine development applications.
Products & Services:
Advantages:
HLA binding is a cornerstone of immunology research. Partner with us to accelerate your discovery process with validated peptide screening services that ensure strong, reliable results.
1. What is HLA binding screening?
Testing peptides for their ability to bind HLA molecules.
2. Why is it important?
It predicts immune recognition and vaccine efficacy.
Can it be customized?
Yes, we provide disease-specific screening.
4. What assays confirm results?
Functional T-cell assays and cytokine testing.
References
