Oncofusion

start / graph / ana-kmt2a-s821-phosphosite-verdict

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q22-phospho-contacts

S821 is not phosphorylated in any dataset we could query: the 14-3-3 motif candidate is dead as stated

also called S821 · RKQTSAP 817 · 14-3-3 candidate verdict

One spot on the KMT2A protein looked like it might be a second docking site, but only if the cell actually puts a phosphate there. We checked a phosphate database five times deeper than the one used before, and the spot is empty.

The same thing, for a scientist

The cycle-52 scan flagged **RKQTSAP at residue 817, phospho-acceptor S821**, as the one canonical 14-3-3 motif in the KMT2A fragment that fusions retain, and closed by noting that UniProt's 25-site table was too insensitive to decide it. We queried the PRIDE/PTMeXchange re-analysis served by the EBI Proteins API, which reports **130 distinct phosphosites on Q03164, 5.2x UniProt's table**. **S821 is absent. So is T820, the alternative acceptor.** The nearest observed phosphosites are S750 and S897, 71 and 76 residues away. The same query recovers **S516** (the ATR site of a 2010 Nature paper) and **T2724/S2726** (the two residues a 2019 Genes Dev paper mutates to block CK2-dependent taspase1 cleavage), so the resource demonstrably has the sensitivity to find functional sites of this class. **Verdict: S821 as a 14-3-3 docking site is not supported and should not be carried forward as a live candidate.**

Go deeper
The query, and the positive controls that make its negative mean something
GET https://www.ebi.ac.uk/proteins/api/proteomics-ptm/Q03164

returns 188 peptide features carrying 174 distinct modified sites: 130 phosphorylation, with acetylation, ubiquitinylation and SUMOylation making up the rest. Every site carries its PRIDE project accession and a localisation probability.

residuein this set?why it is a control
S516yes (PXD012174, loc. prob. 0.82)the ATR site; MLL S516A fails to rescue the S-phase checkpoint 2
T2724yesmutated to Ala in the CK2 peptide assay that blocks taspase1 cleavage 3
S2726yesthe second Ala in the same MLL-AA peptide 3
S136 / S142yesthe CK2/LEDGF IBM sites
T820no-
S821nothe candidate

Four independently characterised functional phosphosites on this protein are recovered. A resource that finds those and does not find S821 is giving a negative with content.

The brief's standing warning was checked: this is not a clean round negative from a broken query. The query returns 130 positives, including every site whose answer was known in advance.

What would still overturn it

A site phosphorylated only transiently, only in one cell type, or only on a stimulus need not appear in a pooled re-analysis of public shotgun runs. None of the PRIDE projects behind this table is a KMT2A-rearranged leukaemia phosphoproteome, and none enriched for KMT2A. The honest form of the verdict is therefore: S821 is not phosphorylated in any of the public datasets re-processed by PTMeXchange, and no dedicated experiment has looked.

What would overturn it, concretely:

  • A phospho-S821 peptide in a KMT2A immunoprecipitation from a t(4;11) line, at any stoichiometry.
  • 14-3-3 co-immunoprecipitation of KMT2A that is lost on S821A. That would make the residue matter regardless of what a database says.
  • A phosphatase-inhibitor or genotoxic-stress condition under which the site appears.
And the candidate was weak on its own terms

The motif scan was reproduced independently here against the live Q03164 sequence and gives the same three matches: RKQTSAP at 817 (acceptor S821), RVHTSTP at 2809 (acceptor S2813) and RSPTVP at 2937 (acceptor T2940). None of the three acceptors is in the 130-site phosphoproteomics set. So the negative is not specific to S821; the whole canonical-motif route on KMT2A comes back empty at the phosphosite step, which is what a chance-level motif hit rate predicts.

One near-miss worth recording so nobody rediscovers it as a finding: S2938 is phosphorylated and sits at position 2 of the RSPTVP match, not at the acceptor position. A scan that scored 'motif overlaps a phosphosite' without checking which residue carries the phosphate would have called that a hit.

Facts and figures
scriptanalyses/kmt2a-phosphosite-lookup/run.py
statusrun
result summary130 phosphosites in PRIDE/PTMeXchange vs 25 in UniProt; S821 and T820 absent from both; all three canonical 14-3-3 motif acceptors (S821, S2813, T2940) absent from both; positive controls S516, S136, S142, T2724, S2726 all recovered.
verdictS821 not supported as a phosphosite; candidate closed pending a dedicated experiment
inputs
[
  "https://www.ebi.ac.uk/proteins/api/proteomics-ptm/Q03164",
  "https://rest.uniprot.org/uniprotkb/Q03164.json?fields=ft_mod_res"
]
resources queried
[
  "EBI Proteins API proteomics-PTM (PRIDE/PTMeXchange)",
  "UniProtKB ft_mod_res",
  "PhosphoSitePlus (HTTP 403, not readable)",
  "iPTMnet REST (HTTP 503 on two attempts)"
]
Still open
  1. Does any KMT2A-rearranged phosphoproteome contain a KMT2A peptide covering residues 815-825 at all? A site cannot be called absent in a region the instrument never sampled.
  2. PhosphoSitePlus and iPTMnet were both unreachable from this session. Someone with access should check S821 in their curated low-throughput sets before the verdict is treated as final.
  3. If 14-3-3 really does contact KMT2A, and not through a canonical motif on KMT2A, is the contact indirect - bridged by an AFF-family protein whose 14-3-3 site is known?
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    A PRIDE/PTMeXchange re-analysis served by the EMBL-EBI Proteins API reports 130 distinct phosphorylation sites on KMT2A (Q03164) across 188 peptide features, and neither S821 nor T820 is among them; the nearest observed phosphosites are S750 and S897. The same set contains S516, S136, S142, T2724 and S2726, all of which have published functional characterisation.

    What this evidence cannot show

    Fetched and parsed 2026-09-19; counts are from that response and will move as PTMeXchange reprocesses. This is a re-analysis of public shotgun proteomics, so coverage is biased towards abundant, tryptically tractable, non-conditional sites; 130 sites on a 3,969-residue protein is deep by database standards and shallow by biological ones. No constituent dataset is a KMT2A-rearranged leukaemia phosphoproteome. Absence here is absence of observation in those runs, not absence in biology.

    Evidence that argues against this (2)
    • There exist KMT2A-rearranged AML phosphoproteomes that this resource does not incorporate: a 74-patient integrative study defines two KMT2A-rearranged groups by phosphorylation signature and reports differential DOT1L phosphorylation between them. A site absent from pooled public re-analysis can be present in disease-specific data, and that data was not queried residue-by-residue here. Casado P et al. 2023
    • PhosphoSitePlus, the resource the dispatch named first, could not be queried: both its search endpoint and its protein pages returned HTTP 403 to this session on 2026-09-19, and iPTMnet returned HTTP 503 on two attempts. Their curated low-throughput sets are not subsets of PRIDE re-analysis, so a site they carry could be missing here. The verdict is bounded by that gap. Cell Signaling Technology 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MLL is phosphorylated at serine 516 by ATR in response to genotoxic stress in S phase, which disrupts its interaction with, and hence its degradation by, the SCF(Skp2) E3 ligase; reconstitution of Mll-null mouse embryonic fibroblasts with wild-type but not S516A MLL rescues the S-phase checkpoint defect.

    PMID 20818375DOI
    What this evidence cannot show

    Abstract read 2026-09-19; full text not retrieved. The rescue is in mouse embryonic fibroblasts. Used here as a positive control for the phosphoproteomics resource, which independently observes S516 in human data.

    Evidence that argues against this (1)
    • That paper's own reading of MLL fusions is that they act as dominant negatives abrogating ATR-mediated phosphorylation of wild-type MLL - the S516 axis is presented as something the fusion disables in trans, not something the fusion carries as a usable handle. A site retained in the fusion's sequence is not the same as the fusion using it. Liu H et al. 2010
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    A casein kinase II in vitro phosphorylation assay used an MLL wild-type peptide GVDD/GTESDTSVTAT and an MLL-AA peptide GVDD/GAEADTSVTAT; the wild-type peptide was dose-dependently phosphorylated by recombinant CKII and the AA peptide was not, identifying the two substituted residues as CK2 acceptors adjacent to taspase1 cleavage site CS2.

    PMID 30573454DOI
    What this evidence cannot show

    Full text read via Europe PMC fullTextXML (PMC6317322) on 2026-09-19. Mapping the peptide to Q03164 places GVDD at 2719-2722 and the two substituted residues at T2724 and S2726; that mapping is ours, done by string search against the canonical sequence, because the paper states the residues as a peptide rather than by number. Used here as a positive control.

    Evidence that argues against this (1)
    • An in vitro assay with recombinant CKII and a synthetic peptide shows that CK2 can phosphorylate the residues, not that it does on full-length MLL in a cell; the paper's cellular evidence is inhibitor- and knockdown-based rather than site-resolved. Zhao Z et al. 2019
Connected to