Oncofusion

start / graph / ana-k1130-and-k1133-acetylation-are-independently-observed-in-deposited-mass-spectrometry-data

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.q165-the-retained-acetylation-site

The PTM-database route runs: K1130 and K1133 acetylation are independently re-observed in deposited mass spectrometry, K1133 in the CBP/p300 acetylome itself - and K1130 is also a ubiquitination site nobody has mentioned

also called PhosphoSitePlus route · KMT2A PTM database query · PXD009839 K1130

Rather than trusting the literature, we asked the raw protein-measurement databases directly: has anyone's machine ever seen this chemical tag on MLL1? Yes - in four separate published experiments, none of which set out to study MLL1 at all. One of them also shows a completely different tag on the same spot, which no paper has ever mentioned.

The same thing, for a scientist

PhosphoSitePlus is Cloudflare-403 to this project (already recorded in `ana-q115-...`), so the query was run instead against the EBI Proteins API `/proteomics-ptm/Q03164` endpoint, which serves PRIDE-reprocessed MS evidence (ECO:0007829) with per-site PRIDE project provenance. It returns 174 distinct site/modification calls on KMT2A. Acetyl-K1130 is seen in PXD009839; acetyl-K1133 in PXD005252 and PXD009839; acetyl-K636 and acetyl-K1235 are also present. The provenance is the informative part: PXD005252 is the CBP/p300 acetylome, which is orthogonal support for the writer assignment the source paper could only make with a C-terminally biased pan-acetyl antibody. And K1130 additionally carries ubiquitinylation (PXD006201, UbiSite), a competing modification on the same lysine that appears nowhere in the KMT2A literature.

Go deeper
The instrument, and why it is not PhosphoSitePlus

PhosphoSitePlus returns HTTP 403 (Cloudflare) to this agent, confirming ana-q115-phosphositeplus-answers-403-to-its-own-homepage-so-no-ptm-absence-can-be-bounded-there from an independent attempt today 4. Rather than record a second failure, the query was moved to:

https://www.ebi.ac.uk/proteins/api/proteomics-ptm/Q03164 — EBI Proteins API, serving PTM calls derived from reprocessed PRIDE submissions, each carrying its PRIDE project accession and, for phosphosites, a localisation probability 1.

This is arguably the better instrument for the question asked. PhosphoSitePlus aggregates and normalises; this endpoint names the deposit, so "has it been seen, and in what experiment" is answerable in one call rather than two.

🟢 The answer, per site
sitemodificationPRIDE depositswhat those deposits are
K636acetylationPXD005252, PXD041171CBP/p300 acetylome; SIRT2/HCMV acetylome
K928acetylationPXD005252, PXD009839, PXD041171all three
K1130acetylationPXD009839HCMV temporal acetylome, human fibroblasts
K1130ubiquitinylationPXD006201UbiSite, ~62,000 ubiquitination sites
K1133acetylationPXD005252, PXD009839CBP/p300 acetylome + HCMV acetylome
K1235acetylationPXD009839HCMV acetylome
K1243acetylationPXD005252, PXD009839, PXD041171all three

12

All four Aguilar-Arnal/Choudhary sites are re-observed. This is a genuine independent replication of the 2009 acetylome call, sixteen years later, on different instruments in different laboratories. The site is not a one-deposit artefact.

🔵 The two findings in here that are new to this graph

1. K1133 is in the CBP/p300 acetylome (PXD005252, Weinert et al., Cell 2018). The source paper could only show that CBP and p300 acetylate MLL1 using a pan-acetyl-lysine antibody that it itself describes as "preferentially detects C-terminal acetylation of MLL1" — so the writer for these two N-terminal residues was strictly unassigned (see ana-no-acetyl-dead-or-acetyl-mimetic-kmt2a-was-ever-put-in-a-methyltransferase-assay). A deposit built specifically around CBP/p300 catalytic inhibition contains acetyl-K1133. That is orthogonal, site-resolved support for CBP/p300 as the writer. ⚠️ Strict limit: presence in the deposit means the peptide was DETECTED in that experiment, not that it was shown to be CBP/p300-DEPENDENT. The Weinert paper's kinetic and inhibitor-responsive subsets would have to be read to claim dependency, and they were not read here.

2. K1130 is a ubiquitination site (PXD006201, UbiSite). No paper in the KMT2A literature says this. Acetylation and ubiquitination are mutually exclusive on the same lysine ε-amine, so a site that carries both is a switch, not a dial — and this graph already holds an entry arguing exactly that acetylation should be read as a cycle rather than a mark. The obvious mechanistic hypothesis writes itself: acetylation at K1130 may protect KMT2A from ubiquitination there, and KMT2A protein stability is a live topic in this disease. Nothing here tests that; it is a hypothesis generated by a database, which is the cheapest possible way to get one.

🔴 What the databases do NOT show, and this is the load-bearing negative

No deposit behind any KMT2A acetylation call is a KMT2A-rearranged cell line. Reading the PRIDE project metadata for every deposit that contributes an acetylation or ubiquitination call at these sites 3:

deposittitlematerial
PXD005252Time-resolved analysis reveals rapid dynamics and broad scope of the CBP/p300 acetylomehuman + mouse; fibroblast, blood cell
PXD009839Orchestrating protein acetylation: a toggle for cellular defense versus virus replication (HCMV)human fibroblast
PXD041171Sirtuin 2 Promotes HCMV Replication by Regulating Cell Cycle Progressionhuman fibroblast
PXD006201Characterisation of protein ubiquitination using UbiSite technologyhuman; blood, larynx SCC line
PXD023218Site-resolved protein turnover (SPOT) profilinghuman epithelial, cervix carcinoma

Two of the five are cytomegalovirus infection studies in fibroblasts. The best-replicated evidence that this site is acetylated in human cells comes from virology.

So the brief's task — would show whether acetylation at these residues has been observed in any MS dataset, including KMT2A-rearranged ones — splits into a yes and a no. Observed in MS: yes, four independent deposits. Observed in a KMT2A-rearranged dataset: no, not in anything the EBI endpoint serves. The one possible exception is the 2009 Choudhary panel, which included MV4-11 and predates ProteomeXchange — see ana-the-founding-acetylome-that-discovered-the-kmt2a-site-was-run-in-a-panel-containing-mv4-11-a-kmt2a-aff1-line.

An instrument caveat that limits every absence built on this endpoint

This endpoint serves reprocessed deposits, not all of PRIDE. A KMT2A-rearranged acetylome could exist in PRIDE and simply not be in the reprocessing pipeline, in which case its sites would be invisible here. This is a floor on coverage, not a census, and a bounded absence from it must say so. The same shape as this project's PSICQUIC-versus-IntAct finding: the query form, not the database, set what could be seen.

Facts and figures
scriptanalyses/q165-the-retained-acetylation-site/ptm-retained-or-lost.py
statusrun
result summary174 MS-evidenced PTM calls on KMT2A. All four Aguilar-Arnal sites re-observed. K1133 present in the CBP/p300 acetylome. K1130 also ubiquitinylated. No contributing deposit is a KMT2A-rearranged line.
endpointhttps://www.ebi.ac.uk/proteins/api/proteomics-ptm/Q03164
phosphositeplus statusHTTP 403 (Cloudflare) - instrument unavailable, absence cannot be bounded there
n sites total174
n acetylation21
n phosphorylation130
n ubiquitinylation19
n sumoylation4
kmt2a rearranged deposit foundfalse
measurements
contributing deposits per…0.000contributing deposits performed in a KMT2A-rearranged cell line: 0PRIDE deposits reporting …1.000PRIDE deposits reporting acetyl-K1130: 1PRIDE deposits reporting …2.000PRIDE deposits reporting acetyl-K1133: 2
inputs
[
  "EBI Proteins API /proteomics-ptm/Q03164 (183,891 bytes, HTTP 200, 2026-09-24)",
  "UniProt Q03164.fasta",
  "PRIDE Archive v3 project metadata for 6 accessions"
]
sites of interest
{
  "K636": {
    "acetylation": [
      "PXD005252",
      "PXD041171"
    ]
  },
  "K928": {
    "acetylation": [
      "PXD005252",
      "PXD009839",
      "PXD041171"
    ]
  },
  "K1130": {
    "acetylation": [
      "PXD009839"
    ],
    "ubiquitinylation": [
      "PXD006201"
    ]
  },
  "K1133": {
    "acetylation": [
      "PXD005252",
      "PXD009839"
    ]
  },
  "K1235": {
    "acetylation": [
      "PXD009839"
    ]
  },
  "K1243": {
    "acetylation": [
      "PXD005252",
      "PXD009839",
      "PXD041171"
    ]
  }
}
measurements — the numbers behind the chart
[
  {
    "quantity": "PRIDE deposits reporting acetyl-K1133",
    "subject": "Q03164 K1133 acetylation",
    "value": 2,
    "units": "deposits",
    "n": 21,
    "universe": "21 acetylation site-calls on Q03164 in the EBI reprocessed-PRIDE set",
    "null": null,
    "note": "PXD005252 (CBP/p300 acetylome) and PXD009839 (HCMV acetylome)"
  },
  {
    "quantity": "PRIDE deposits reporting acetyl-K1130",
    "subject": "Q03164 K1130 acetylation",
    "value": 1,
    "units": "deposits",
    "n": 21,
    "universe": "21 acetylation site-calls on Q03164 in the EBI reprocessed-PRIDE set",
    "null": null,
    "note": "PXD009839 only - the weakest of the four Aguilar-Arnal sites by deposit count"
  },
  {
    "quantity": "contributing deposits performed in a KMT2A-rearranged cell line",
    "subject": "the 5 deposits behind the K636/K1130/K1133/K1235 calls",
    "value": 0,
    "units": "deposits",
    "n": 5,
    "universe": "PXD005252, PXD009839, PXD041171, PXD006201, PXD023218",
    "null": null,
    "note": "rule of three: true rate below 60% at n=5, which is a weak bound - this is a small-n zero and should not be quoted as a strong absence"
  }
]
How this could bite us back
  • If the acetyl/ubiquitin competition at K1130 is real, an intervention that raises acetylation there would also stabilise wild-type KMT2A - and wild-type KMT2A is the residual allele this disease depends on in ways this graph documents elsewhere. Raising the level of the fusion's wild-type partner is not obviously the direction anyone wants.
  • The best-replicated evidence for this site being acetylated at all is from cytomegalovirus-infected fibroblasts. Building a leukaemia programme on a site whose deepest data come from a virology model imports every confound of that model.
Still open
  1. Is acetyl-K1133 CBP/p300-DEPENDENT in PXD005252, or merely detected there? Weinert et al. quantified acetylation under CBP/p300 catalytic inhibition; their supplementary tables would say whether this site drops under A-485. That single lookup would assign a writer to the residue, which the source paper could not do.
  2. Does acetylation at K1130 compete with ubiquitination at K1130? The two calls come from different deposits and nothing links them, but they are chemically exclusive on the same amine, and this is the first suggestion in this graph that KMT2A carries an acetyl/ubiquitin switch in retained sequence.
  3. Does any acetylome or ubiquitinome deposit in PRIDE use a KMT2A-rearranged line at all, reprocessed or not? A direct PRIDE search on cell line rather than on protein would answer this, and it was not run here.
  4. Why do two of the five deposits come from cytomegalovirus biology? If HCMV infection drives this site, that is a confound in the best-replicated evidence for it - and it may also be a free perturbation experiment nobody has read as one.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The EBI Proteins API proteomics-ptm endpoint returns 174 distinct site/modification calls on UniProt Q03164, comprising 21 acetylation, 130 phosphorylation, 19 ubiquitinylation and 4 SUMOylation sites, each with PRIDE project provenance.

    What this evidence cannot show

    Fetched 2026-09-24, 183,891 bytes, HTTP 200. Counts are of distinct (modification, absolute position) pairs after mapping peptide-relative positions to absolute coordinates via each feature's begin offset; a site seen in several peptides is counted once. The endpoint carries only ECO:0007829 large-scale MS evidence and deliberately does NOT include curated low-throughput sites from UniProt's own feature table, so this is a picture of what mass spectrometers have seen, not of what is known.

    Evidence that argues against this (1)
    • Presence of a site in a reprocessed deposit is an identification, not a validated site. Acetyl-lysine identifications in particular suffer from trimethyl-lysine isobaric confusion (both nominally +42 Da; acetyl 42.0106, trimethyl 42.0470) and require high mass accuracy to separate. Localisation probability is served for phosphosites here but NOT for acetylation, so per-site localisation confidence for K1130/K1133 is unavailable from this endpoint.
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Acetylation at Q03164 K1130 is reported in PRIDE project PXD009839; at K1133 in PXD005252 and PXD009839; K1130 also carries ubiquitinylation in PXD006201.

    What this evidence cannot show

    The per-site dataset assignment is read out of the endpoint's dbReferences and has not been checked against the individual PRIDE submissions' own result files. If a deposit's reprocessing assigned a peptide to the wrong protein - KMT2A shares its N-terminal AT-hook architecture with no close paralogue, but the peptide APPIKPIKPVTR is short and proline-rich - the call would be wrong. The K1130 ubiquitination call rests on a SINGLE deposit and should be treated as a lead, not a fact.

    Evidence that argues against this (1)
    • K1130 acetylation rests on ONE deposit (PXD009839), which is weaker than K1133 (two) or K1243 (three). If the single-deposit call is a misassignment, the specific residue the source paper's antibody is named for would be the least well supported of the four.
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    None of the PRIDE deposits contributing acetylation or ubiquitination calls at K636/K1130/K1133/K1235 was performed in a KMT2A-rearranged cell line; two of them are human cytomegalovirus infection studies in fibroblasts.

    What this evidence cannot show

    Judged from each project's declared organism/organism-part/cell-type fields and its abstract, not from its sample-level metadata files. A project whose declared cell type is 'Blood cell' or 'Permanent cell line cell' could in principle include a KMT2A-rearranged line among many; PXD005252 in particular declares 'Blood cell' without naming it. This absence is therefore at the project-declaration level and is NOT a peptide-level audit of every submission.

    Evidence that argues against this (1)
    • The EBI proteomics-ptm endpoint serves REPROCESSED deposits only. A KMT2A-rearranged acetylome could sit in PRIDE unreprocessed and be invisible here, which would make this a property of the pipeline rather than of the field. No attempt was made to search PRIDE directly for acetylome submissions in KMT2A-rearranged lines.
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    PhosphoSitePlus returns HTTP 403 to this project, independently reproducing a failure this graph already recorded.

    What this evidence cannot show

    A Cloudflare interstitial, not a server error - PhosphoSitePlus works fine in a browser. This says nothing about the database's contents and everything about automated access from this host.

    Evidence that argues against this (1)
    • A 403 is not a neutral failure here, and treating it as one would understate the damage. PhosphoSitePlus records the CELL LINE behind each site, and its acetylation curation incorporates the Choudhary 2009 dataset. It is therefore the single most likely source to resolve the one question this beat could not close - whether the KMT2A acetyl peptides came from MV4-11. The instrument that failed is the instrument that mattered, so the 'no KMT2A-rearranged deposit' conclusion in this entry is weaker than the EBI numbers make it look.
Connected to