Looking for a second handle on the retained fragment, and not finding one
also called KMT2A 14-3-3 motif · second handle scan · S821 candidate
If the piece of KMT2A that survives the translocation had a second docking site for a drug to grab, the whole strategy would have another option. We looked for the obvious kind of site and did not find a convincing one.
The same thing, for a scientist
The proximity beat found **KMT2A is a 14-3-3 client across two independent studies**. 14-3-3 binding is phosphorylation-gated and sequence-encoded, so if a motif sat N-terminal to the breakpoint cluster region there would be a **second, phosphorylation-gated, non-menin handle on the fragment every fusion retains** — a direct contradiction of this project's cycle-28 finding that menin is very nearly the only retained handle. Scanning UniProt Q03164: **3 canonical motif matches in 3,969 residues against ~1.1 expected by chance, one of them in the retained fragment, and none coinciding with an annotated phosphosite.** The lead is narrowed, not closed.
What was looked for and what was found
Canonical 14-3-3 modes, scanned across the full sequence:
mode 1 R S X pS X P
mode 2 R X X X pS X P
| KMT2A (Q03164) | AFF1 (P51825) | MEN1 (O00255) | |
|---|---|---|---|
| length | 3,969 | 1,210 | 610 |
| annotated phosphosites in UniProt | 25 | 8 | 3 |
| canonical motif matches | 3 | 0 | 0 |
| matches in the retained fragment (≤1400) | 1 | 0 | 0 |
| matches with an annotated phosphosite | 0 | 0 | 0 |
The single retained-fragment candidate is RKQTSAP beginning at residue 817, with the phospho-acceptor at S821 — a textbook mode-2 motif in the right half of the protein. S821 is not annotated as a phosphosite in UniProt.
Three matches against 1.09 expected by chance over a protein this length is not a signal. The motif is six or seven residues of mostly unconstrained sequence; finding a few in four thousand residues is what happens.
Why this does not close the question
The filter is far weaker than it looks. UniProt annotates 25 phosphosites on a 3,969-residue nuclear protein that is heavily regulated; dedicated phosphoproteomics resources carry hundreds for proteins of this class. Requiring a motif to coincide with a UniProt-annotated site is therefore a very low-sensitivity test, and a negative from it is close to uninformative about whether S821 is phosphorylated in a cell.
14-3-3 also binds non-canonically. A substantial share of validated 14-3-3 clients bind through divergent motifs, and the Mol Cell interactome study that makes KMT2A a client did not report a motif for it. A scan for the canonical consensus cannot exclude a real interaction.
So the state of the refutation lead is: KMT2A is a reproduced 14-3-3 client; the interaction is not explained by a canonical annotated motif on the retained fragment; and S821 is the one residue worth a phosphosite lookup and a pulldown. The cheap in-silico test the proximity beat proposed has been run and it does not settle the question. That is a useful outcome — it means the experiment cannot be skipped.
- This does not clear the menin-only picture. It says the canonical-motif route does not explain a reproduced 14-3-3 interaction, which leaves the interaction unexplained rather than absent.
- S821 is a hypothesis for a pulldown, not a target. Nothing here shows it is phosphorylated, that 14-3-3 binds there, or that anything binds the retained fragment besides menin.
- If 14-3-3 binding to KMT2A is chaperone-like, occupying that site would promote aggregation rather than inhibit the fusion. A second handle is not automatically a second opportunity.
- S821 is answered and negative. The live route is now the AFF-family bridge: 14-3-3theta binds AFF1 Ser588, which is retained in the fusion, through a non-canonical motif.
- Read the Mol Cell full text for KMT2A specifically: is it among the clients whose localisation was tested on 14-3-3 loss, and did it coalesce into foci?
- The pulldown is still the answer. 14-3-3 immunoprecipitation from a t(4;11) line, blotted for the fusion, settles in one experiment what no sequence scan can.
- Scan the other retained fragments. Every KMT2A fusion keeps the same N-terminal region, so if there is a second handle it is shared across the whole family - and the same scan run on the 3' partners would say whether the partners contribute one.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.analysis
Scanning UniProt Q03164 for canonical 14-3-3 binding motifs finds 3 matches across 3,969 residues against approximately 1.09 expected by chance, of which exactly one lies in the fragment retained by KMT2A fusions: a mode-2 motif RKQTSAP beginning at residue 817 with its phospho-acceptor at S821. No motif match anywhere in KMT2A coincides with a phosphosite annotated in UniProt, which lists 25 for the protein. AFF1 and MEN1 return zero matches.
KMT2A 14-3-3 motif scan: does the retained fragment carry a phosphorylation-gated handle? · Oncofusion knowledge graph (internal analysis) 2026What this evidence cannot show
Run 2026-09-20 against the live UniProt REST record. The retained-fragment boundary is taken as residue 1400, which is a conventional approximation of where the breakpoint cluster region falls in protein terms rather than a measured boundary — every match is reported with its residue so a reader can redraw the line. Chance expectation is a back-of-envelope calculation assuming uniform amino-acid composition, which real sequence does not have.
Evidence that argues against this (2)
- **A negative from this filter is close to uninformative.** UniProt annotates only 25 phosphosites on a large, heavily regulated nuclear protein, so requiring a motif to coincide with an annotated site has very low sensitivity. S821 not being annotated is weak evidence that it is not phosphorylated, and no dedicated phosphoproteomics resource was queried. UniProt Consortium 2026
- 14-3-3 proteins bind many validated clients through non-canonical motifs, and the study that makes KMT2A a client mapped interactomes rather than motifs — it reports no motif for KMT2A at all. A consensus scan cannot exclude an interaction that does not use the consensus. Molecular Cell 2023
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The 14-3-3 study that motivates this scan mapped the interactomes of all human 14-3-3 paralogs and found that loss of 14-3-3 binding causes a large fraction of clients to coalesce into discrete foci in a client-specific manner, indicating a chaperone-like function; engrafting 14-3-3 binding motifs onto non-clients suppressed their aggregation or phase separation.
A central chaperone-like role for 14-3-3 proteins in human cells · Molecular Cell 2023What this evidence cannot show
Abstract read on 2026-09-20; full text not read, so whether KMT2A is among the clients whose localisation was tested, and what happened to it if so, is not recorded here. The claim that KMT2A is recovered by six of seven paralogs comes from the Q20 beat's reading of the curated deposition rather than from this abstract.
Evidence that argues against this (1)
- A chaperone-like function that prevents client aggregation is not the kind of handle a drug discovery programme wants. If 14-3-3 binding to KMT2A is about solubility rather than about targeting or activity, then occupying the site would promote aggregation rather than inhibit the fusion - the opposite of a therapeutic handle, and arguably a liability. Molecular Cell 2023
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
S821 is not phosphorylated in any published dataset reachable here, so the single retained-fragment 14-3-3 motif candidate this scan produced is dead. The EBI Proteins API proteomics-ptm endpoint reports 130 observed phosphosites for KMT2A - 5.2 times UniProt's 25 - and S821 is absent, as are T820 and the acceptors of both other canonical motif matches. The nearest observed sites are S750 and S897.
EBI Proteins API, proteomics-ptm endpoint for UniProt Q03164 (KMT2A) · EMBL-EBI / PRIDE 2026What this evidence cannot show
Queried by the Q22 beat on 2026-09-19. **The negative has content because the same query recovers four sites whose answers were known in advance** - S516 (ATR), S136 and S142 (CK2/LEDGF) and T2724/S2726 (CK2/taspase1). Bounded: no constituent dataset is a KMT2A-rearranged leukaemia phosphoproteome, PhosphoSitePlus returned HTTP 403 and iPTMnet HTTP 503, so this is 130 sites from one aggregator rather than the whole record.
Evidence that argues against this (1)
- A site absent from aggregated phosphoproteomics may simply be condition-specific: these datasets are overwhelmingly from unsynchronised cultured cells, and a signalling-gated site could require a stimulus none of them applied. The 14-3-3 interaction is reproduced in two interactome datasets and remains unexplained by any motif. EMBL-EBI / PRIDE 2026
the KMT2A sites that motif scan relies on all number consistently with Q03164
both bear on where on KMT2A a functional phosphosite could be
- supports →KMT2A / MLL1 — the gene that breaks
closes the S821 candidate the scan left open: absent from 130 PRIDE phosphosites
a demonstrated 14-3-3 contact in this system that a canonical-motif scan cannot see