# pMHC-TCR Mechanism Specification & Literature Synthesis
## T-Cell-Targeted Immunotherapy for Eosinophilic Esophagitis

**Version:** 2.0 (real citations retrieved via OpenAlex; epitope burden rebuilt on netMHCIIpan-4.1)
**Status:** Scientific Rationale Document (Pre-IND)

> **Citation policy:** Every reference below was retrieved programmatically from OpenAlex with a verified DOI. Where a specific mechanistic claim originates from a foundational paper I did not retrieve in this session, it is attributed generically ("established in the tolerance literature") rather than to a fabricated citation. Prior version 1.0 contained author/year/page citations that were not retrieved from any database; those are fully retracted and replaced here.

---

## 1. DISEASE IMMUNOLOGY & THERAPEUTIC HYPOTHESIS

### 1.1 EoE is a food-antigen-driven CD4+/Th2 disorder
EoE pathogenesis centers on type-2 inflammation triggered by food (and aero-) allergens, with eosinophil recruitment driven by IL-5/IL-13 and an epithelial-remodeling program. This is established across foundational pathogenesis reviews (Rothenberg 2001, *J Allergy Clin Immunol*, doi:10.1067/mai.2001.120095; Rothenberg 2004 EGID review, doi:10.1016/j.jaci.2003.10.047; Furuta 2007 consensus, doi:10.1053/j.gastro.2007.08.017; Racca 2022 type-2 inflammation review, doi:10.3389/fphys.2021.815842). Experimental models establish an etiologic role for allergen exposure and eosinophils (Mishra 2001, *J Clin Invest*, doi:10.1172/jci10224; Saito Akei 2005 epicutaneous priming, doi:10.1053/j.gastro.2005.06.027), and immunopathology of pediatric EoE documents the T-cell/eosinophil infiltrate (Teitelbaum 2002, doi:10.1053/gast.2002.32998).

The causal role of food antigen is demonstrated by dietary elimination: histologic remission with elimination diets is well documented (Kagalwalla 2006 six-food elimination, doi:10.1016/j.cgh.2006.05.026; Arias 2014 meta-analysis, doi:10.1053/j.gastro.2014.02.006; C. Henderson 2012, doi:10.1016/j.jaci.2012.03.023; Liacouras 2005 cohort, doi:10.1016/s1542-3565(05)00885-2).

Critically for an MHC-II-directed therapy, human **esophageal epithelium expresses MHC class II and can present antigen** (Mulder 2011, *Am J Pathol*, doi:10.1016/j.ajpath.2010.10.027), and epithelial barrier/immune regulation (e.g., DSG1) is disrupted in EoE (Sherrill 2013, *Mucosal Immunol*, doi:10.1038/mi.2013.90).

### 1.2 Therapeutic hypothesis
Selectively silencing or deleting food-allergen-specific CD4+ T cells — via their peptide-MHC-II/TCR interaction — should interrupt the upstream driver of the Th2 cascade, unlike downstream cytokine blockade. Antigen-specific tolerance induction as a therapeutic strategy is reviewed comprehensively by Kenison 2023 (*Nat Rev Immunol*, doi:10.1038/s41577-023-00970-x).

---

## 2. EPITOPE TARGET SPACE (from Task 1, real netMHCIIpan-4.1 predictions)

- 10 allergen proteins (8 food categories) × 5 HLA-DRB1 alleles = **12,140 binding-affinity predictions**.
- **2,287 strong binders** (IC50 < 500 nM); **470 unique binding-core registers** globally (no core shared across allergens → allergen-specific).
- Per-allele unique-core burden varies widely (netMHCIIpan 5-allele: DRB1*01:01 = 363 dominant, DRB1*03:01 = 32 lowest; extended to 13 alleles with cross-validated mhcnuggets, Task 2). This ~11-fold range is why epitope panels are computed **per patient** on their own genotype rather than used to select patients — eligibility is HLA-agnostic (Task 2).
- Highest-load allergens: soy β-conglycinin and glycinin; lowest: peanut Ara h 2, walnut Jug r 1.

This target space defines the peptides that would be loaded into pMHC reagents and used in the T-cell assays (Task 3).

---

## 3. FOUR CANDIDATE MECHANISMS

### Mechanism 1 — pMHC-II multimer-induced anergy/tolerance
**Principle:** Multivalent peptide-MHC-II engagement of the TCR **without** costimulation (signal 1 without signal 2/3) drives CD4+ T cells into anergy or deletion. The requirement for a costimulatory/third signal to convert TCR engagement into full activation rather than tolerance is a foundational principle of T-cell biology (Curtsinger 2003 "signal 3", *J Exp Med*, doi:10.1084/jem.20021910). Antigen-specific tolerance-induction platforms — including soluble/multimerized pMHC and nanoparticle-displayed pMHC — are reviewed by Kenison 2023 (doi:10.1038/s41577-023-00970-x).
**Pros:** antigen-specific; protein reagent (fast CMC); no cell lysis; reversible.
**Cons:** durability may require repeat dosing; memory cells may resist.
**Development risk:** LOW — **recommended for Phase 4A proof-of-concept.**

### Mechanism 2 — pMHC-decorated particles delivering a deletion signal
**Principle:** pMHC on nanoparticles/exosomes co-displaying pro-apoptotic or coinhibitory ligands drives TCR-specific deletion. Artificial antigen-presenting constructs for CD4+ T cells are demonstrated (Couture 2019, HLA-class-II artificial APCs, *Front Immunol*, doi:10.3389/fimmu.2019.01081). Coinhibitory-axis biology (PD-1/PD-L1) that such particles could recruit is reviewed by Keir 2008 (*Annu Rev Immunol*, doi:10.1146/annurev.immunol.26.021607.090331).
**Pros:** deletion (not just anergy). **Cons:** bioengineered product, complex CMC; off-target apoptosis risk. **Risk:** MEDIUM.

### Mechanism 3 — Costimulation blockade (CTLA-4 axis) as antigen-nonspecific adjunct
**Principle:** Blocking CD28/B7 costimulation converts antigen encounter into tolerance. The CD28/B7–CTLA-4 axis in autoimmunity is reviewed by Salomon & Bluestone 2001 (*Annu Rev Immunol*, doi:10.1146/annurev.immunol.19.1.225); CTLA4-Ig costimulation blockade has clinical proof-of-concept in T-cell-mediated disease (Abrams 1999, psoriasis, *J Clin Invest*, doi:10.1172/jci5857); CD4+CD25+ regulatory T cells enforce tolerance in part via CTLA-4 (Takahashi 2000, *J Exp Med*, doi:10.1084/jem.192.2.303).
**Pros:** clinically precedented. **Cons:** **not antigen-specific** — global costimulation blockade; best as adjunct, not standalone. **Risk:** MEDIUM.

### Mechanism 4 — Engineered T-cell (CAR / TCR) approaches
**Principle:** Redirected T cells for antigen-specific suppression. Antigen-specific **regulatory** CAR-T (CAR-Treg) suppress in an alloantigen-specific manner (MacDonald 2016, *J Clin Invest*, doi:10.1172/jci82771); Treg cell therapy is reviewed by Romano 2019 (doi:10.3389/fimmu.2019.00043) and Tanaka & Sakaguchi 2016 (doi:10.1038/cr.2016.151). CAR-T killing mechanisms (Benmebarek 2019, doi:10.3390/ijms20061283) and TCR gene-transfer (Johnson 2009, *Blood*, doi:10.1182/blood-2009-03-211714) define the engineering toolkit.
**Pros:** most selective; renewable. **Cons:** gene-therapy regulatory pathway; autologous manufacturing; first-in-class pMHC-directed design. **Risk:** HIGH (deferred to later phase).

---

## 4. MECHANISM DECISION TREE

```
Goal = interrupt food-allergen-specific CD4+ Th2 activation
│
├─ Need fastest, lowest-risk path to IND, antigen-specific?
│   └─► Mechanism 1 (pMHC-II multimer anergy)      ← PHASE 4A (recommended)
│         protein reagent · reversible · epitopes from Task 1 · patients from Task 2
│
├─ Mechanism-1 durability insufficient (memory persists)?
│   ├─► add Mechanism 3 (CTLA-4 costim blockade) as antigen-nonspecific adjunct
│   └─► escalate to Mechanism 2 (pMHC particle + deletion signal)
│
└─ Durable, renewable, single-course intent, regulatory capacity for cell therapy?
    └─► Mechanism 4 (CAR-Treg / engineered T cell)  ← later-phase option
```

**Recommendation:** Phase 4A on **Mechanism 1** — antigen-specificity, a protein-therapeutic CMC path, established signal-1-without-costimulation tolerance biology, and direct reuse of the Task 1 epitope panel + Task 2 HLA stratification + Task 3 anergy readouts. Mechanisms 2–4 are the escalation ladder if durability is inadequate.

---

## 5. PRECLINICAL VALIDATION (aligned to Mechanism 1)

1. **Ex vivo human (primary):** recruit all-comers EoE patients (no HLA selection); for each, expand allergen-specific CD4+ T cells (Task 3) and treat with **patient-matched** pMHC-II multimers loaded with that individual's top predicted cores; read TCR-signaling block (pERK/pAkt), proliferation loss on rechallenge, anergy transcriptome, viability (anergy vs deletion), and reversibility (IL-2 rescue). Success: ≥3-fold SI suppression, with suppression tracking each patient's individualized predicted epitope burden.
2. **In vivo (proof-of-concept):** HLA-DR-transgenic or humanized mouse with an allergen/OVA-substituted epitope; established experimental-EoE models exist (Mishra 2001; Saito Akei 2005). Endpoints: esophageal/GI eosinophilia, allergen-specific Th2 cytokines, preserved unrelated immunity (specificity), safety labs.
3. **Biomarker bridge:** the Task 3 ex vivo anergy panel is the pharmacodynamic readout carried into the clinical protocol.

---

## 6. REGULATORY FRAMING

Mechanism 1 is a recombinant protein (pMHC-II multimer) → biologics pathway (351(a)). Key pre-IND topics: GMP pMHC production and multimer assembly, potency assay (anergy induction), repeat-dose GLP toxicology, immunotoxicology (protective-immunity sparing via epitope selectivity), PK/PD (anergy-marker kinetics), and the per-patient personalized-epitope platform with HLA-agnostic eligibility (Task 2). Illustrative timeline to Phase 4A start ≈ 30 months.

---

## 7. REFERENCES (all retrieved via OpenAlex, verified DOIs)

**EoE pathogenesis / diet:** Rothenberg 2001 (10.1067/mai.2001.120095); Rothenberg 2004 (10.1016/j.jaci.2003.10.047); Furuta 2007 (10.1053/j.gastro.2007.08.017); Racca 2022 (10.3389/fphys.2021.815842); Mishra 2001 (10.1172/jci10224); Saito Akei 2005 (10.1053/j.gastro.2005.06.027); Teitelbaum 2002 (10.1053/gast.2002.32998); Kagalwalla 2006 (10.1016/j.cgh.2006.05.026); Arias 2014 (10.1053/j.gastro.2014.02.006); C. Henderson 2012 (10.1016/j.jaci.2012.03.023); Liacouras 2005 (10.1016/s1542-3565(05)00885-2).
**Epithelial antigen presentation / barrier:** Mulder 2011 (10.1016/j.ajpath.2010.10.027); Sherrill 2013 (10.1038/mi.2013.90).
**Tolerance / costimulation / Treg:** Kenison 2023 (10.1038/s41577-023-00970-x); Curtsinger 2003 (10.1084/jem.20021910); Keir 2008 (10.1146/annurev.immunol.26.021607.090331); Salomon 2001 (10.1146/annurev.immunol.19.1.225); Abrams 1999 (10.1172/jci5857); Takahashi 2000 (10.1084/jem.192.2.303).
**Engineered T cells / APC:** MacDonald 2016 (10.1172/jci82771); Romano 2019 (10.3389/fimmu.2019.00043); Tanaka 2016 (10.1038/cr.2016.151); Benmebarek 2019 (10.3390/ijms20061283); Johnson 2009 (10.1182/blood-2009-03-211714); Couture 2019 (10.3389/fimmu.2019.01081).

---

**Corrections in v2.0:** (i) all fabricated citations removed, replaced with OpenAlex-retrieved DOI-verified references; (ii) epitope burden updated to real netMHCIIpan (2,287 strong binders, 470 unique cores, DRB1*01:01 dominant / *03:01 lowest — reversed from retracted v1); (iii) claims not traceable to a retrieved paper are attributed generically rather than to invented sources.