# EoE Food-Antigen Immunology & the Dual pMHC Rationale

## 1. Food antigens drive EoE — the evidence

EoE is a food-antigen-driven, chronic immune-mediated disease of the esophagus (PubMed;
Hirano & Furuta, JAMA 2021, doi:10.1001/jama.2021.14920). The causal role of dietary antigen is
established by dietary-elimination therapeutics:
- **Elemental (amino-acid) formula** — removing all intact food protein — induces remission in >90%
  of children, the strongest single piece of causal evidence (Spergel, Curr Opin Allergy Clin
  Immunol 2010, doi:10.1097/ACI.0b013e328338cbab).
- **Empiric elimination diets** (six-food / milk-egg-wheat-soy) induce remission in ~50–74%, and
  reintroduction reproducibly triggers relapse.
- Milk is the most common single trigger; wheat, egg, soy follow.

Mechanistically EoE is a **type-2 (Th2), non-IgE-dominant** food hypersensitivity: a breakdown of
oral tolerance, biased Th2 response, and an impaired epithelial barrier (Hill & Spergel, JACI 2018,
doi:10.1016/j.jaci.2018.05.001; Rothenberg, JCI 2019, doi:10.1172/JCI124604). Food antigen crosses a
defective barrier, is presented to CD4+ T cells, and drives IL-4/IL-5/IL-13 — recruiting eosinophils
and mast cells and producing eotaxin-3/CCL26. This is why dupilumab (anti-IL-4Rα) works, and why the
disease is fundamentally a **CD4/MHC-II** process.

## 2. The user's immunology point is correct

MHC-I presents to CD8+ T cells; MHC-II presents to CD4+ T cells. EoE is CD4/Th2-driven, so the
disease-relevant antigen-presentation arm is **MHC class II**. My Step-7 analysis scored MHC-I
(CD8) peptides — a real but secondary axis. This review adds the disease-appropriate MHC-II arm.

Critically, esophageal epithelial cells can act as **non-professional APCs presenting food antigen on
MHC class II** in EoE (Mulder et al., Am J Pathol 2011, doi:10.1016/j.ajpath.2010.10.027) — a
mechanism placing the epithelium directly in the CD4/Th2 priming loop, not just as a barrier.

## 3. What our own data show (reconciling the two arms)

Single-cell APM scoring (GSE218607, this project) localizes the two arms differently:
- **MHC-II / CD4 arm:** dominated by professional APCs — myeloid cells carry by far the highest
  MHC-II (score ~3.1), constitutively; epithelial MHC-II is low and not induced in our snapshot.
  The CD4 priming machinery is present via dendritic/myeloid cells (and, per the literature, inducible
  epithelium under sustained IFN-γ).
- **MHC-I / CD8 arm:** this is where the disease-associated CHANGE sits — epithelium acquires HLA-I
  (suprabasal Δ+0.82, basal Δ+0.55), IFN-γ-driven (r=0.94). A cytotoxic/IFN-γ CD8 component is a
  documented but less-studied feature of EoE epithelium.

Food-antigen presentation load (predicted): class-II strong binders vastly outnumber class-I
(milk β-lactoglobulin 212 vs 28; egg ovomucoid 91 vs 13) — consistent with a CD4-dominant response
to these antigens.

## 4. Rationale for pursuing BOTH pMHC-I and pMHC-II

**Yes — but they serve different purposes and are not co-equal.**

**pMHC-II (CD4/Th2) — primary, disease-central.** This is the arm that matches EoE's core biology.
Therapeutic angles: (a) tolerance-based approaches (antigen-specific immunotherapy / peptide
tolerization to the dominant DR-presented food epitopes — e.g. the β-lactoglobulin cores nominated
here); (b) blocking the CD4-priming step (the epithelial/myeloid MHC-II–TCR interface); (c)
biomarker use — patient-specific dominant food epitopes could rationalize targeted elimination
rather than empiric six-food removal. This is where an antigen-directed (as opposed to
cytokine-blocking) strategy would most plausibly modify disease.

**pMHC-I (CD8) — secondary, mechanistically interesting, exploratory.** Justified by the IFN-γ-driven
epithelial HLA-I induction we found, which points to a cytotoxic/IFN-γ CD8 component in the injured
epithelium. This is a genuine mechanistic thread (and a novelty angle), but it is downstream/parallel
to the CD4 driver and lacks the causal, therapy-validated status of the Th2/CD4 axis. Worth
characterizing; not the lead antigen-directed strategy.

**Net recommendation:** if pursuing an antigen-directed (pMHC/TCR) programme, **lead with class-II /
CD4** — it is where EoE's causal food-antigen biology lives and where tolerance induction could be
transformative. Treat class-I / CD8 as a mechanistic sub-study attached to the IFN-γ-epithelial-injury
finding, not as an independent therapeutic pillar. Both are legitimate; their priority is not equal.

## 5. Impact on the Phase-3 deliverables (review)

- **Step-7 pMHC analysis** was class-I-only — now augmented with the class-II arm (pmhc_ii_candidates.csv,
  pmhc_dual_axis.png, pmhc_dual_load.csv) and the cell-type localization of each arm.
- **The three protein-design leads (SIGLEC6, IL1RL1/ST2, CCL26) are unaffected** — they are
  effector/receptor targets downstream of antigen presentation, and their rationale (validated
  effectors, dupilumab differentiation) stands independent of the CD4-vs-CD8 question.
- **Companion-Dx and endotype work unaffected.**
- **New caveat carried forward:** the epithelial-antigen-presentation "pMHC entry point" in the
  Phase-3 report is, on our data, a CD8/MHC-I (IFN-γ) signal at the epithelium plus a CD4/MHC-II
  signal at professional APCs — the therapeutic antigen-directed opportunity is class-II-led.

## References (PubMed; DOIs)
- JAMA 2021 — doi:10.1001/jama.2021.14920 (EoE review)
- Curr Opin Allergy Clin Immunol 2010 — doi:10.1097/ACI.0b013e328338cbab (food allergy & EoE; elimination-diet remission rates)
- JCI 2019 — doi:10.1172/JCI124604 (mechanisms of GI allergic disorders)
- JACI 2018 — doi:10.1016/j.jaci.2018.05.001 (allergic components of EoE)
- Am J Pathol 2011 — doi:10.1016/j.ajpath.2010.10.027 (epithelial MHC-II antigen presentation in EoE)
- Gastroenterology 2017 — doi:10.1053/j.gastro.2017.05.066 (diagnosis & treatment)

Source: PubMed.