Peptide Immunogenicity Assessment in Preclinical Research: Distinguishing Aggregation-Related Responses from Sequence-Intrinsic Epitopes
Immunogenicity—the capacity of a compound to provoke an immune response—is among the most consequential and methodologically challenging properties to characterize during preclinical research. For synthetic peptides, the challenge is compounded by the fact that immune responses observed in laboratory assays may originate from the compound's amino acid sequence, from physical states the compound adopts under particular conditions, or from contaminants introduced during synthesis, formulation, or handling. Conflating these sources of signal produces unreliable risk assessments and can either overstate or obscure genuine immunogenic liability.
This article provides a structured framework for critically evaluating preclinical immunogenicity data, with particular attention to the distinction between aggregation-driven immune activation and sequence-intrinsic epitope recognition. It addresses common confounding variables, the appropriate use and limitations of computational prediction tools, species-specific considerations, and the interpretive boundaries of negative immunogenicity findings.
The Core Distinction: Aggregation-Driven Versus Sequence-Intrinsic Immunogenicity
Why Aggregation Matters
Peptide aggregation is one of the most frequently underappreciated sources of immunogenic signal in preclinical assays. When peptide monomers associate into oligomers, protofibrils, or larger particulate structures, they present repetitive epitope arrays that can directly crosslink B-cell receptors, bypassing T-cell help and triggering antibody responses that would not occur with the same sequence in a soluble, monomeric state [1]. This mechanism is well-documented in the context of therapeutic proteins and has been increasingly recognized as relevant to synthetic peptides as well.
The practical consequence is significant: a peptide that appears highly immunogenic in an assay conducted with improperly stored or reconstituted material may show markedly different behaviour when tested under conditions that preserve monomeric structure. Conversely, a peptide that aggregates predictably under physiological conditions may carry inherent immunogenic risk that is not captured when assays are performed with freshly prepared, well-characterized material.
Characterizing Aggregation State Before Immune Testing
Quality control prior to immunogenicity assays should include orthogonal structural characterization. Dynamic light scattering (DLS) can detect particle size distributions and flag aggregation, while analytical ultracentrifugation or size-exclusion chromatography with multi-angle light scattering (SEC-MALS) provides higher-resolution information about oligomeric state [2]. Circular dichroism spectroscopy can indicate secondary structural changes associated with aggregation-prone conformations.
Without this baseline characterization, immune assay data cannot be meaningfully attributed to sequence properties. Research protocols that proceed directly from peptide reconstitution to immune testing without structural confirmation introduce an uncontrolled variable that undermines the interpretive value of any downstream immunogenicity measurement.
Interpreting Core Immune Assays
T-Cell Proliferation Assays
T-cell proliferation assays—typically conducted using peripheral blood mononuclear cells (PBMCs) from human donors or splenocytes from immunized animals—measure the capacity of a peptide to stimulate T-cell expansion, a process that requires antigen presentation via MHC class II molecules. A positive proliferation signal is conventionally interpreted as evidence of T-cell epitope recognition, which is considered a prerequisite for T-cell-dependent antibody responses and a marker of immunogenic potential [1].
However, several confounding factors can produce false-positive proliferation signals. Endotoxin contamination is the most common: lipopolysaccharide (LPS) and related bacterial products are potent polyclonal activators of innate immune cells, which in turn provide non-specific co-stimulatory signals that drive T-cell proliferation independently of antigen-specific recognition [3]. A proliferation response observed with a peptide preparation that has not been tested for endotoxin content is essentially uninterpretable.
False-negative results are equally possible. Peptides that require intracellular processing before MHC loading may not stimulate effectively in assay formats that present intact peptide extracellularly. Donor selection also introduces variability: HLA haplotype diversity means that a given peptide may stimulate T cells from some donors but not others, and small donor panels may fail to capture responses that would be evident in a broader population.
Antibody Binding Assays
Enzyme-linked immunosorbent assays (ELISAs) and related antibody detection methods measure whether immune responses have produced antibodies capable of binding the peptide of interest. In the context of preclinical research, these assays are most informative when used to characterize the nature of immune responses in immunized animals rather than to predict human antibody responses directly.
A critical interpretive issue is the distinction between antibodies directed against the peptide sequence and antibodies directed against aggregated forms or against carrier proteins used in immunization protocols. Peptides are frequently conjugated to carrier proteins such as keyhole limpet hemocyanin (KLH) to enhance immunogenicity in animal models; antibodies against the carrier can interfere with assay readouts and must be accounted for in data interpretation.
Cytokine Release Assays
Cytokine release assays measure the secretion of immune mediators—including interferon-gamma, interleukin-2, interleukin-4, and others—following peptide stimulation of immune cells. The cytokine profile can provide information about the type of immune response being elicited: Th1-skewed responses (characterized by IFN-γ) versus Th2-skewed responses (characterized by IL-4 and IL-13) have different downstream implications.
As with proliferation assays, endotoxin contamination is a major confound. LPS is a potent inducer of pro-inflammatory cytokines, particularly TNF-α and IL-6, and even low-level contamination can generate cytokine release signals that mimic or obscure antigen-specific responses [3]. Cytokine release assay data should always be accompanied by endotoxin quantification of the test article, typically using the limulus amebocyte lysate (LAL) assay or recombinant factor C-based alternatives.
Process-Related Impurities and Their Immunological Consequences
Endotoxin Contamination
Endotoxin contamination is the single most common technical confound in peptide immunogenicity research. Synthetic peptides produced by solid-phase peptide synthesis (SPPS) are not inherently endotoxin-containing, but contamination can be introduced through reagents, solvents, water, glassware, and handling during purification and formulation. Research-grade peptides that have not been explicitly tested and certified for endotoxin content should be treated as potentially contaminated for the purposes of immunogenicity assay design.
The threshold for biologically relevant endotoxin activity in in vitro assays is substantially lower than the limits applied in clinical contexts. In vitro PBMC-based assays can be sensitive to endotoxin concentrations below 0.1 EU/mL, meaning that peptide preparations that would pass conventional quality control thresholds may still introduce confounding innate immune activation in sensitive assay formats [3].
Host Cell Proteins and Synthesis-Related Impurities
For peptides produced using recombinant or cell-based methods, host cell proteins (HCPs) represent an additional class of potential immunogenic contaminants. HCPs can independently stimulate immune responses or act as adjuvants that amplify responses to co-administered peptides. Even for chemically synthesized peptides, incomplete deprotection, racemization at chiral centers, or the presence of deletion sequences can alter the immunological profile of the preparation relative to the intended compound.
High-performance liquid chromatography (HPLC) purity assessment and mass spectrometric confirmation of molecular identity are minimum quality standards for peptides intended for immunogenicity testing. Purity thresholds of greater than 95% by HPLC are commonly applied in research settings, though the immunological relevance of specific impurity profiles depends on their identity and concentration.
Computational Prediction Tools: Utility and Limitations
HLA-Peptide Binding Prediction
Computational tools for predicting MHC-peptide binding affinity—including NetMHCpan, IEDB analysis tools, and related algorithms—have become standard components of immunogenicity risk assessment workflows [4]. These tools predict the likelihood that peptide fragments will bind to specific HLA alleles with sufficient affinity to be presented to T cells, providing a sequence-level estimate of T-cell epitope content.
The predictive value of these tools is meaningful but bounded. High predicted binding affinity to common HLA alleles is a necessary but not sufficient condition for immunogenicity: peptide-MHC complexes must also be recognized by T-cell receptors present in the repertoire, and central tolerance mechanisms may have eliminated or suppressed T cells reactive to peptides with homology to self-antigens [4]. Conversely, peptides with moderate predicted binding affinity may still elicit responses in individuals with particular HLA haplotypes not well-represented in training datasets.
Cross-Reactivity and Sequence Homology Analysis
Sequence homology analysis—comparing a peptide's sequence against databases of known T-cell epitopes, self-antigens, and pathogen-derived peptides—provides complementary information to binding prediction. Homology to known immunogenic epitopes from pathogens may indicate that pre-existing memory T cells could recognize the research compound, potentially amplifying immune responses. Homology to self-antigens raises questions about tolerance and the potential for autoimmune-relevant cross-reactivity [1].
These analyses should be interpreted cautiously. Sequence similarity at the level of individual amino acid positions does not guarantee functional cross-reactivity, and the structural context of a given sequence within a larger peptide influences both MHC binding and T-cell receptor recognition in ways that linear homology searches do not capture.
Species-Specific Considerations in Preclinical Models
Rodent Models
Murine and rat models are widely used in immunogenicity research because of their genetic tractability and the availability of immunological reagents. However, the murine immune system differs from the human system in ways that limit direct translation of immunogenicity findings. The MHC repertoire in inbred mouse strains is far less diverse than human HLA diversity, meaning that a peptide's immunogenicity in a particular mouse strain reflects binding to a narrow set of MHC molecules that may not be representative of human population-level responses [6].
Transgenic mouse models expressing human HLA alleles have been developed to partially address this limitation, and preclinical data from these models may have somewhat greater relevance to human immunogenicity prediction. Nevertheless, differences in T-cell repertoire development, regulatory T-cell function, and innate immune signalling between mice and humans mean that animal model data should be interpreted as hypothesis-generating rather than predictive [6].
Non-Human Primate Models
Non-human primate (NHP) models, particularly cynomolgus macaques, are considered more translationally relevant for immunogenicity assessment due to greater similarity in immune system architecture and MHC diversity relative to rodents. Animal studies in NHPs have shown that immune responses to therapeutic peptides can more closely approximate patterns observed in human clinical studies [6]. However, NHP studies are resource-intensive, and the practical constraints of sample size limit statistical power.
Importantly, even NHP immunogenicity data does not reliably predict human responses. Species differences in regulatory T-cell populations, gut microbiome composition, and prior pathogen exposure history all influence immune reactivity in ways that are difficult to control across species.
Interpreting Negative Immunogenicity Data
The absence of an immune response in preclinical assays is frequently misinterpreted as evidence of immunological safety. This interpretation requires careful qualification. A negative result in a T-cell proliferation assay conducted with a small donor panel, using a single peptide concentration, over a fixed incubation period, reflects only the conditions of that specific experiment [1].
Negative preclinical immunogenicity data does not establish that a compound is non-immunogenic in humans. It establishes only that the compound did not elicit a detectable response under the tested conditions. Assay sensitivity, donor HLA coverage, peptide concentration range, and assay duration all influence the probability of detecting a response that exists. Reporting negative immunogenicity data should include explicit description of assay parameters and their known limitations.
Conversely, a well-designed negative result—obtained with endotoxin-controlled, structurally characterized material, tested across a diverse donor panel covering common HLA supertypes, using validated assay formats—carries substantially more informational weight than a negative result from a single, poorly characterized experiment.
Quality Control Metrics That Minimize Immunogenicity Artifacts
Minimizing the contribution of artifacts to immunogenicity data requires systematic quality control at multiple stages. Before immune testing, peptide preparations should be confirmed for molecular identity by mass spectrometry, assessed for purity by HPLC, tested for endotoxin content using a validated LAL or equivalent assay, and characterized for aggregation state by at least one biophysical method. Sterility testing is appropriate for preparations intended for cell-based assays.
Storage conditions should be documented and standardized, as freeze-thaw cycling and prolonged storage at suboptimal temperatures can induce aggregation in peptides that are stable under ideal conditions. Reconstitution protocols—solvent composition, concentration, temperature, and agitation—should be specified and followed consistently across experiments to ensure that structural characterization data reflects the material actually used in assays.
These quality control measures do not eliminate all sources of variability in immunogenicity data, but they substantially reduce the contribution of preventable artifacts and improve the interpretive value of experimental findings.
Conclusion
Preclinical immunogenicity assessment is most valuable when it is approached as a structured risk characterization exercise rather than a binary determination of immunogenic or non-immunogenic status. The distinction between aggregation-driven immune activation and sequence-intrinsic epitope recognition is foundational to this exercise, and it cannot be made without adequate structural characterization of test materials. Confounding variables—endotoxin contamination, process-related impurities, species-specific immune architecture, and assay design limitations—must be explicitly addressed in study design and acknowledged in data interpretation.
Computational prediction tools and in vitro assays provide complementary but individually incomplete pictures of immunogenic potential. Animal model data, while informative, carries inherent translational uncertainty that should be reflected in how findings are framed. Negative data, properly contextualized, is informative; negative data from poorly controlled experiments is not. Rigorous application of these principles produces immunogenicity assessments that accurately characterize what is known, what remains uncertain, and what additional investigation would be required to reduce that uncertainty.