How Human Metabolites May Influence Multiple Sclerosis Through HLA Binding

Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous system in which immune-mediated damage to myelin contributes to neurological disability. Among the genetic factors associated with MS, the human leukocyte antigen (HLA) region has the strongest influence, with HLA-DRB1*15:01 representing the major susceptibility allele in many populations. Intriguingly, the closely related allele DRB1*15:03 differs from DRB1*15:01 at only one amino-acid position, yet appears not to confer the same level of MS susceptibility. Because HLA class II molecules present peptide antigens to CD4+ T cells, structural differences in their peptide-binding grooves could influence which molecular complexes the immune system encounters. The study by Misra and colleagues therefore examined a provocative possibility: rather than peptides acting alone, naturally occurring human metabolites might enter HLA binding pockets and modify antigen presentation. This idea connects genetics, metabolism, and adaptive immunity in a potentially important model of autoimmune disease.

Why Small Metabolites Might Matter to Antigen Presentation
HLA class II molecules contain a peptide-binding groove organized into several pockets, conventionally described from P1 to P9, that accommodate side chains from bound antigenic peptides. In MS, one particularly important antigen is myelin basic protein (MBP). An immunodominant MBP region binds DRB115:01, with valine at MBP position 89 acting as a P1 anchor and phenylalanine at position 92 serving as a P4 anchor. The P4 pocket of DRB115:01 is relatively large and predominantly hydrophobic, whereas its P1 pocket is smaller. Previous research had already shown that small exogenous molecules can occupy HLA-DR pockets and alter peptide loading or peptide conformation, sometimes provoking abnormal immune responses. Misra and colleagues extended this concept to endogenous metabolites—small molecules generated by human metabolism or influenced by the microbiome. They hypothesized that these molecules might transiently occupy HLA pockets, alter peptide positioning, and consequently modify how T cells recognize self-antigens.

Screening Tens of Thousands of Metabolites in Silico
To test this hypothesis, the researchers used a structure-based virtual-screening strategy. They began with the X-ray crystal structure of DRB1*15:01 bound to an MBP peptide, resolved at 2.6 Å, and computationally generated a DRB1*15:03 model by changing residue 30 from tyrosine to histidine—the single amino-acid difference highlighted between the two alleles. They then screened the complete version 3.6 of the Human Metabolome Database, which contained 41,993 metabolites from endogenous, exogenous, food-derived, microbial, toxic, and drug-related sources. Independent docking analyses were conducted across the HLA peptide-binding pockets, with candidate interactions evaluated through energetic and structural parameters including electrostatic interactions, van der Waals forces, solvation components, molecular contacts, and alternative conformations. The workflow illustrated in Figure 4 on page 8 summarizes this pipeline, moving from HLA structure and metabolite selection through computational docking and analysis.

The P4 Pocket Emerges as the Critical Structural Site
The most striking structural result centered on the P4 pocket. Structural comparison showed that phenylalanine at MBP position 92 could coexist with metabolite binding in the P4 pocket of DRB115:01, whereas valine at MBP position 89 effectively prevented comparable metabolite accommodation in the smaller P1 pocket. The molecular surface shown in Figure 1 on page 3 visually emphasizes the large hydrophobic character of the P4 region. The authors attributed this geometry largely to alanine at DRβ71, which creates space capable of accommodating aromatic groups, while valine at DRβ86 constrains the P1 pocket. Importantly, metabolites appeared to bind more strongly to the P4 pocket of DRB115:01 than to that of DRB115:03. The authors proposed that the residue at position 30 may contribute: DRB115:01 contains neutral tyrosine, whereas DRB1*15:03 contains positively charged histidine. Thus, a single allelic difference may influence the energetic environment experienced by small molecules.

Which Metabolites Were Predicted to Bind?
When the researchers ranked the top docking results in the P4 pocket, they identified metabolites unique to each allele as well as many shared between them. Among the top 200 candidates, 78 metabolites were reported as individually binding DRB115:01, 81 as individually binding DRB115:03, and 79 as common to both, with overall docking scores suggesting stronger interactions with DRB115:01. The heat map in Figure 2 on page 4 visualizes differences in energy scores across 157 metabolites. Particularly interesting were molecules already associated in previous studies with MS or neurological biology. Pathway analysis highlighted L-tryptophan, glutamic acid, D-sphingosine, sphingosine-1-phosphate, cysteinyl-glycine, and NAD+, among others, as molecules linked to MS or neurodegenerative phenotypes. Table 1 also identifies myo-inositol hexakisphosphate as a high-scoring example, with a reported docking energy of −85.92 kcal/mol for DRB115:01 compared with −55.03 kcal/mol for DRB1*15:03.

A Possible Mechanism: Metabolite-Induced Peptide Register Shifts
How could metabolite binding translate into altered immunity? The authors proposed a mechanistic model based on a peptide register shift. HLA class II molecules possess an open-ended peptide-binding groove, allowing peptides longer than the core nine-residue binding segment to extend outside the groove. If a metabolite transiently occupies a pocket such as P4, it could potentially alter the position of the peptide within the groove, shifting which peptide residues contact the HLA molecule and which residues face the T-cell receptor. Figure 3 on page 7 illustrates several possible MBP binding registers and shows how moving the peptide could change the pattern of HLA-contact and T-cell-receptor-contact residues. The authors also suggest an alternative mechanism: transient occupation of the P4 pocket might stabilize a peptide-receptive form of HLA and thereby accelerate antigen loading. Either mechanism could, in principle, alter T-cell recognition of myelin-derived peptides and influence autoimmune responsiveness.

A Promising Hypothesis—But One That Still Requires Experimental Proof
The study provides a compelling conceptual bridge between HLA genetics, metabolism, antigen presentation, and autoimmune susceptibility. If endogenous metabolites truly modify peptide presentation by disease-associated HLA molecules, metabolomic differences between individuals could potentially influence immune responses even among people carrying similar genetic risk alleles. The authors suggest that this framework might eventually support biomarker discovery, structure-guided immunomodulatory drug design, and investigations of other HLA-associated autoimmune diseases. However, the most important limitation is equally clear: the predicted DRB1–metabolite complexes were not experimentally tested in the laboratory. The study therefore demonstrates computationally plausible molecular interactions rather than proving that these metabolites alter T-cell activity or cause MS in vivo. The authors explicitly identify experimental validation of metabolite effects on antigen presentation and T-cell proliferation as an essential next step. Even with this limitation, the work expands the conventional view of autoimmunity by suggesting that the molecular environment surrounding HLA molecules—not genetics or peptides alone—may help shape immune recognition.

Disclaimer: This blog post is based on the provided research article and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.

References:
Misra, M.K., Damotte, V. & Hollenbach, J.A. Structure-based selection of human metabolite binding P4 pocket of DRB1*15:01 and DRB1*15:03, with implications for multiple sclerosis. Genes Immun 20, 46–55 (2019). https://doi.org/10.1038/s41435-017-0009-5