Oncofusion

start / graph / interface-hla-b0702-kmt2a-phosphopeptide-747-755

Interfacestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q31-interfaces-structural

KMT2A 747-755 on HLA-B*07:02: a phosphoneoantigen from the half of KMT2A every fusion keeps, with seven crystal structures and a cognate TCR

also called pMLL747-755 · EPRSPSHSM · phosphoneoantigen · TCR27 · 7RZD · 7RZJ

A nine-amino-acid piece of KMT2A, carrying a phosphate, is displayed on the surface of leukaemia cells by a common tissue type and is recognised by a specific T-cell receptor. That piece sits in the part of KMT2A that every leukaemic fusion keeps.

The same thing, for a scientist

pMLL747-755, sequence EPR(pS)PSHSM, is an HLA-B*07:02-restricted acute myeloid leukaemia phosphoneoantigen recognised by a cognate T cell receptor, TCR27, which the authors put forward as an immunotherapy candidate. Seven crystal structures of HLA-B*07:02 with the peptide, the phosphopeptide and synthetic phosphono- and sulfo-analogues were deposited (7RZD, 7RZJ, 7S79, 7S7D, 7S8A, 7S8E, 7S8F); the peptide entity in 7RZD is EPRSPSHSM, which occurs exactly once in the fetched UniProt Q03164 sequence, at residues 747-755, placing the phosphoserine at KMT2A S750. Replacing the P4 phosphoserine with serine or with phosphomimetics leaves the peptide-MHC conformation and the peptide-MHC affinity unchanged but abrogates TCR27-dependent T cell activation - the recognition is of the phosphate itself, non-covalently, by the receptor. Residue 750 lies roughly 350 residues N-terminal to the KMT2A breakpoint cluster region, so this epitope is encoded within the retained KMT2A portion of KMT2A::AFF1 and of every other der(11) fusion.

Go deeper
Why this belongs in an interface layer

Every other entry in this beat describes an interface the fusion uses to do something. This one describes an interface the fusion is seen by. It is a protein-protein interface with structures, a defined epitope, a defined post-translational modification and a cognate receptor, and it is the only such interface in this graph.

The positional fact does the work: KMT2A S750 is inside the retained fragment. Anything that presents this epitope presents it from wild-type KMT2A and from the fusion alike, so the epitope is not fusion-specific - but it is not lost in the fusion either, which is the failure mode that kills most KMT2A-derived epitope ideas.

What makes it a neoantigen, and what that costs

The neoantigenicity is not sequence but phosphorylation state 1. That has two consequences the authors themselves demonstrate. First, the discrimination is at the level of the T cell receptor, not of MHC loading: unphosphorylated and phosphomimetic peptides load equally well and are simply not recognised. Second, it means the target is only as tumour-restricted as the kinase activity that writes S750 - which the paper places in the proline-directed kinase and AKT families generically rather than in a leukaemia-specific enzyme.

The graph's phospho-contacts folder records several KMT2A and AFF1 phosphosites; S750 is not among them. Whether S750 is phosphorylated on the fusion protein has never been measured, and the graph's own cycle-16 audit concluded that no IP-then-phospho-blot has ever been performed on KMT2A::AFF1.

Facts and figures
partner agene-kmt2a
partner bHLA-B*07:02 (P01889) with beta-2-microglobulin (P61769)
motif aKMT2A 747-755, EPRSPSHSM, phosphorylated at S750 (peptide position P4)
motif bHLA-B*07:02 peptide-binding groove
kd statusno peptide-MHC dissociation constant recorded here; the paper's quantitative point is that phospho, non-phospho and phosphomimetic peptides have INDISTINGUISHABLE peptide-MHC affinity, and that the discrimination happens at the T cell receptor
fate in fusionretained - S750 lies roughly 350 residues N-terminal to the KMT2A breakpoint cluster region, so the epitope is encoded within the retained KMT2A portion of every der(11) fusion
fusion specificityNOT fusion-specific: the same epitope is encoded by the wild-type allele. Its selectivity, if any, would have to come from differential S750 phosphorylation, which has not been measured on the fusion or in KMT2A-rearranged cells
receptorTCR27, described by the authors as a candidate for cancer immunotherapy; the TCR-pMHC ternary arrangement is an NMR-informed docking model rather than a crystal structure
druggabilitya TCR- or TCR-mimetic-directed modality rather than a small-molecule one; the structures explicitly show that stable phosphate mimics preserve presentation and abolish recognition
resolution notePDB metadata and entity sequences fetched from data.rcsb.org 2026-09-20; the peptide mapped to UniProt Q03164 fetched the same day, where EPRSPSHSM occurs exactly once at 747-755
interface residues a
[
  "KMT2A E747",
  "KMT2A P748",
  "KMT2A R749",
  "KMT2A S750 (phospho-acceptor, peptide P4)",
  "KMT2A M755"
]
kd measurements
[]
structures
[
  "7RZD",
  "7RZJ",
  "7S79",
  "7S7D",
  "7S8A",
  "7S8E",
  "7S8F"
]
structure details
[
  {
    "pdb": "7RZD",
    "title_fetched": "CRYSTAL STRUCTURE OF HLA-B*07:02 IN COMPLEX WITH MLL(747-755) PEPTIDE",
    "method": "X-ray",
    "resolution_A": 1.82,
    "contents": "HLA-B alpha chain (P01889) + beta-2-microglobulin (P61769) + KMT2A 9-mer EPRSPSHSM",
    "fetched": "2026-09-20"
  },
  {
    "pdb": "7RZJ",
    "title_fetched": "CRYSTAL STRUCTURE OF HLA-B*07:02 IN COMPLEX WITH MLL(747-755) PHOSPHOPEPTIDE",
    "method": "X-ray",
    "resolution_A": null,
    "fetched": "2026-09-20"
  }
]
How this could bite us back
  • Not fusion-specific: the epitope is in the retained fragment and is encoded by the wild-type allele too, so on-target off-tumour recognition of normal cells expressing phospho-S750 KMT2A is the central safety question.
  • Modification-dependent target: selectivity rests on a phosphorylation state written by broadly active proline-directed and AKT-family kinases, so it can be lost by kinase-pathway rewiring without any change to the antigen.
  • Antigen escape: HLA class I downregulation is a standard leukaemic immune-escape route and would remove this target entirely.
  • Mimetic trap: the paper shows that phosphomimetic and phosphono-analogues load onto HLA normally and fail to activate TCR27, so a vaccine built on a hydrolysis-resistant mimic would present the wrong thing.
  • Allele restriction: HLA-B*07:02 restriction limits eligibility to a minority of patients and to populations in which that allele is common.
Still open
  1. Is KMT2A S750 phosphorylated on the KMT2A::AFF1 fusion protein? The graph's cycle-16 audit found that no IP-then-phospho-blot has ever been done on this fusion, so the question is open and answerable.
  2. Is pMLL747-755 detectable by immunopeptidomics on primary KMT2A-rearranged blasts, and at what copy number per cell?
  3. What fraction of infant KMT2A-rearranged ALL patients are HLA-B*07:02 positive? That number bounds the addressable population and appears nowhere in this graph.
  4. Does the epitope's presentation differ between leukaemic blasts and normal haematopoietic cells, given that both express KMT2A?
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    pMLL747-755 (EPR(pS)PSHSM) is an HLA-B*07:02-restricted AML phosphoneoantigen recognised by TCR27; replacing the P4 phosphoserine with serine or phosphomimetics leaves peptide-MHC conformation and affinity unchanged but abrogates TCR27-dependent T cell activation and weakens TCR27-pMHC binding.

    Molecular mechanism of phosphopeptide neoantigen immunogenicity. · Patskovsky Y, Natarajan A, Patskovska L, et al. · Nat Commun 2023
    PMID 37353482DOI
    What this evidence cannot show

    The work is structural and biophysical with a T cell activation readout; it is not a clinical or in vivo efficacy study, and the abstract describes TCR27 as a candidate rather than a validated therapeutic. The ternary TCR-pMHC complex is a docking model informed by NMR, not a crystal structure. Nothing in the paper concerns KMT2A fusion proteins specifically. Full text read from Europe PMC fullTextXML (195,776 characters).

    Evidence that argues against this (1)
    • Because the epitope lies in the retained KMT2A fragment, it is presented by wild-type KMT2A as well as by any fusion, so selectivity depends entirely on differential S750 phosphorylation between leukaemic and normal cells - which this paper does not measure. Patskovsky Y et al. 2023
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PDB 7RZD is a 1.82 A crystal of HLA-B*07:02 with beta-2-microglobulin and a nine-residue KMT2A peptide EPRSPSHSM, which maps to Q03164 747-755; 7RZJ is the corresponding phosphopeptide structure.

    PDB 7RZD: CRYSTAL STRUCTURE OF HLA-B*07:02 IN COMPLEX WITH MLL(747-755) PEPTIDE · Patskovsky Y, Natarajan A, Patskovska L, et al. · RCSB Protein Data Bank 2023
    What this evidence cannot show

    Titles, methods, resolutions and all three polymer entity sequences fetched from data.rcsb.org 2026-09-20; the residue span is this cycle's alignment of the fetched peptide sequence to the UniProt Q03164 sequence fetched the same day, where EPRSPSHSM occurs exactly once. The RCSB entity is labelled 'MLL cleavage product N320 peptide', which is a correct description of which taspase1 product contains it but is not a KMT2A domain name.

    Evidence that argues against this (1)
    • A peptide-MHC crystal establishes that the epitope can be presented, not that it is presented at meaningful density on primary leukaemic blasts; that is a mass-spectrometry immunopeptidomics question the structures cannot answer. Patskovsky Y et al. 2023
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Seven deposits cover this epitope on HLA-B*07:02, including synthetic phosphono- and sulfo-analogues of the peptide.

    PDB 7RZJ: CRYSTAL STRUCTURE OF HLA-B*07:02 IN COMPLEX WITH MLL(747-755) PHOSPHOPEPTIDE · Patskovsky Y, Natarajan A, Patskovska L, et al. · RCSB Protein Data Bank 2023
    What this evidence cannot show

    Fetched from data.rcsb.org 2026-09-20 by listing entries containing Q03164; the seven are 7RZD, 7RZJ, 7S79, 7S7D, 7S8A, 7S8E and 7S8F, each with the same 384-residue deposited polymer content. The analogue structures test a chemical hypothesis about the phosphate and are not themselves therapeutic candidates.

    Evidence that argues against this (1)
    • The analogues that preserve conformation and MHC affinity are precisely the ones that fail to activate TCR27, so a synthetic-epitope vaccine strategy built on a stable phosphate mimic is directly contraindicated by these structures. Patskovsky Y et al. 2023
Connected to