Targeted PRM for the KMT2A::AFF1 junction peptide - the experiment that makes every other attribution claim testable
also called junction peptide PRM · fusion-specific parallel reaction monitoring
Chop the proteins from a leukemia cell into pieces and look, with a mass spectrometer, for the one piece that can only come from the fusion protein. We have computed exactly which piece that is and what it weighs, so the experiment is a targeted search rather than a fishing trip.
The same thing, for a scientist
A parallel-reaction-monitoring assay for the junction peptide of KMT2A::AFF1, run on immunoprecipitated material from SEM and RS4;11 and on primary t(4;11) blasts. The transitions are precomputed: **VDFKQTYSNEVHCVEEILK**, [M+H]+ 2281.12, trypsin with one missed cleavage; and orthogonally **FKQTYSNE**, [M+H]+ 1016.47, Glu-C in phosphate buffer with zero missed cleavages. Both are absent from the reviewed human proteome. Heavy-labelled synthetic standards of both peptides give absolute quantification and, critically, make a negative interpretable: spike recovery establishes the limit of detection in the actual matrix, without which a null says nothing. The same run quantifies the **reciprocal** junction peptide MPSQSVEEDCEAENVWEMGGLGILTSVPITPR ([M+H]+ 3474.62, zero missed cleavages), so one experiment gives absolute levels of both fusion products in the same cell for the first time. Adding phospho-forms of the junction peptides to the transition list would, if any were found, be the first phosphosite ever attributed to a KMT2A fusion protein rather than inferred from a breakpoint distribution.
Design
Materials. SEM and RS4;11 (both reciprocal-positive), MV4-11 (reciprocal status unresolved in this graph), a KMT2A-germline B-ALL line as the negative control, and 3-5 cryopreserved primary t(4;11) blasts with breakpoints typed by diagnostic RT-PCR.
Breakpoint-matched targets. The junction is patient-specific. Each sample's diagnostic RT-PCR result is fed into analyses/junction-peptide-detectability/run.py, which regenerates the peptide and mass for that architecture. This is the step that makes the assay generalisable rather than a single-cell-line curiosity, and it is why 81 junctions were modelled rather than one.
Enrichment. KMT2A and its fusions are low-abundance nuclear proteins; unenriched lysate is the wrong starting material. Nuclear extract, then immunoprecipitation with an anti-MLL-N antibody, which pulls the direct fusion and wild-type KMT2A but not the reciprocal product - and separately with an anti-AFF1 antibody, which pulls wild-type AFF1, the direct fusion and the reciprocal product. Comparing the two IPs by junction peptide is itself the readout of which species each antibody actually recovers, which no published experiment has established.
Digestion. Split every sample: trypsin with max missed cleavages = 2 in the search, and Glu-C in phosphate buffer. The two proteases are not redundant here - for KMT2A::MLLT3 the Glu-C peptide is unusable (64 residues, 6,284 Da, from the AF9 poly-Ser tract) and the tryptic one is excellent, while for KMT2A::AFF1 both work.
Controls that decide whether a negative means anything.
- Heavy AQUA standards for every target peptide, spiked before digestion, giving digestion efficiency and matrix limit of detection.
- A wild-type peptide from each parent protein in the same run, to prove the protein was in the sample at all.
- The KMT2A-germline line, to establish the false-positive rate of the transition list.
What each outcome licenses
| result | what it establishes |
|---|---|
| Junction peptide detected, quantified against heavy standard | First sequence-resolved identification of a KMT2A fusion protein. Absolute copy number per cell. Every downstream attribution claim gets a denominator. |
| Detected in anti-AFF1 IP but not anti-MLL-N IP | The reciprocal product, or a fusion the N-terminal antibody does not recover - either way an immediate correction to how IP results in this field are read. |
| Not detected, heavy standard recovered at expected efficiency | A real upper bound on fusion protein abundance. This is the informative negative and it is only available because of the spike. |
| Not detected, heavy standard not recovered | An assay failure, not a result. Without the spike this is indistinguishable from the row above, which is how a null here could be badly misread. |
| Phospho-form detected | The first phosphosite attributed to a fusion protein. Would convert a whole class of inference in this graph into measurement. |
Why this is cheap relative to what it settles
Two synthetic heavy peptides, one IP, one LC-MS/MS run per condition. No animal work, no new cell line, no genome editing. Against that it is the only experiment found in this beat that can, in a single readout, (a) prove a KMT2A fusion protein exists as a species, (b) give its abundance relative to the wild-type parents in the same cell, (c) separate direct from reciprocal product, and (d) put a number on the sensitivity of every antibody-based attribution in the field.
The cheaper precursor is cheaper still: re-search already-deposited raw files from SEM and RS4;11 against a fusion-augmented database with two missed cleavages. That costs compute and nothing else, and it has apparently never been done.
Correction: this assay needs a mass per breakpoint, and the one it carried is not SEM's
This experiment was written around a single junction peptide, VDFKQTYSNEVHCVEEILK at MH+ 2281.1169, computed for the 1406/1407 breakpoint that an earlier analysis treated as canonical.
That is not the breakpoint SEM carries. Aligning the published siMA6 siRNA - 8 KMT2A bases plus 13 AFF1 bases - to reconstructed chimeras matches only the 1362/1363 architecture. Run in SEM, an assay targeting 2281.1169 would return a clean negative that means nothing.
| breakpoint | KMT2A tail | junction peptide | missed | MH+ | 2+ m/z |
|---|---|---|---|---|---|
| 1362/1363 (SEM) | ...PKEK | EKQTYSNEVHCVEEILK | 1 | 2048.9957 | 1025.0015 |
| 1406/1407 | ...RVDFK | VDFKQTYSNEVHCVEEILK | 1 | 2281.1169 | 1141.0621 |
| 1444/1445 | ...GHVE | VVCFLCASSGHVEQTYSNEVHCVEEILK | 0 | 3123.4584 | 1562.2328 |
Design consequences. Include all three transitions, or determine the patient's or line's breakpoint first and select the mass. Add carbamidomethyl cysteine (+57.0215 per Cys) for standard preparation. And note that 1444/1445 needs no missed-cleavage allowance, which makes it the cheapest architecture to pilot on.
The deeper limitation stays. Three is the number of breakpoints UniProt annotates, not the number that exist; patients carry junctions none of these matches. A junction assay is patient-specific by nature, and a three-mass panel is a better approximation rather than a general solution.
| design | Targeted parallel reaction monitoring of breakpoint-matched junction peptides in anti-MLL-N and anti-AFF1 immunoprecipitates from t(4;11) cell lines and primary blasts, with heavy AQUA standards. |
|---|---|
| readout | Absolute copies per cell of the direct KMT2A::AFF1 junction peptide and the reciprocal AFF1::KMT2A junction peptide, plus matrix limit of detection from spike recovery. |
| controls | Heavy-labelled synthetic standards spiked pre-digestion; wild-type peptides of both parent proteins in the same run; KMT2A-germline B-ALL line for false-positive rate; two proteases (trypsin 1-2 missed cleavages, Glu-C phosphate buffer) as orthogonal confirmation. |
| cost tier | low |
| timeline | 3-4 months including peptide synthesis |
| cheaper precursor | re-search deposited SEM/RS4;11 proteomics raw files against a fusion-augmented database with 2 missed cleavages - compute only |
precomputed targets — the numbers behind the chart
[
{
"species": "KMT2A::AFF1 direct",
"protease": "trypsin",
"peptide": "VDFKQTYSNEVHCVEEILK",
"MH_plus": 2281.1169,
"missed_cleavages": 1
},
{
"species": "KMT2A::AFF1 direct",
"protease": "Glu-C (phosphate)",
"peptide": "FKQTYSNE",
"MH_plus": 1016.4684,
"missed_cleavages": 0
},
{
"species": "AFF1::KMT2A reciprocal",
"protease": "trypsin",
"peptide": "MPSQSVEEDCEAENVWEMGGLGILTSVPITPR",
"MH_plus": 3474.6225,
"missed_cleavages": 0
},
{
"species": "KMT2A::MLLT3",
"protease": "trypsin",
"peptide": "VDFKDPNR",
"MH_plus": 990.5003,
"missed_cleavages": 1
}
]analyses
[ "ana-junction-peptide-detectability-kmt2a-fusions" ]
closes
[ "absence-no-kmt2a-fusion-junction-peptide-reported-in-any-mass-spectrometry-dataset", "absence-no-phosphosite-mapped-on-any-kmt2a-fusion-protein" ]
target peptides by breakpoint — the numbers behind the chart
{
"UniProt ~1362/1363 (SEM)": {
"peptide": "EKQTYSNEVHCVEEILK",
"missed_cleavages": 1,
"MH_plus": 2048.9957,
"doubly_charged_mz": 1025.0015
},
"UniProt ~1406/1407": {
"peptide": "VDFKQTYSNEVHCVEEILK",
"missed_cleavages": 1,
"MH_plus": 2281.1169,
"doubly_charged_mz": 1141.0621
},
"UniProt ~1444/1445": {
"peptide": "VVCFLCASSGHVEQTYSNEVHCVEEILK",
"missed_cleavages": 0,
"MH_plus": 3123.4584,
"doubly_charged_mz": 1562.2328
}
}- An absolute fusion-protein number will immediately be used as a pharmacodynamic endpoint for menin inhibitors. Menin inhibition is an occupancy mechanism, so fusion protein level may not move at all under an effective drug - and a flat biomarker would be misread as drug failure.
- If the assay shows fusion protein persisting under menin inhibitor while HOXA output falls, that is evidence for displacement rather than degradation, and it removes one of the arguments for combination strategies aimed at fusion turnover.
- Reporting copies per cell invites comparison across patients, but the assay's efficiency depends on the breakpoint-specific peptide's ionisation, which differs between architectures. Cross-patient comparison without per-peptide standards would manufacture a spurious biological difference.
- What is the actual copy number of a KMT2A fusion protein per blast? Nobody knows, and every stoichiometric argument in this field - occupancy, complex composition, condensate formation - assumes an answer.
- Does the anti-MLL-N immunoprecipitate contain the direct fusion at all, or is it dominated by wild-type KMT2A? The junction peptide is the only way to find out, and the answer determines how every published MLL-N IP in this disease should be read.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
The transitions for this assay are precomputed and proteome-unique: VDFKQTYSNEVHCVEEILK at [M+H]+ 2281.12 (trypsin, 1 missed cleavage) and FKQTYSNE at [M+H]+ 1016.47 (Glu-C, 0 missed cleavages) for KMT2A::AFF1.
In-silico junction-peptide detectability analysis for KMT2A fusions (this graph) · Oncofusion Knowledge Graph 2026What this evidence cannot show
The peptide sequences depend on the breakpoint. The deposited RS4;11 architecture (KMT2A 1-1406 | AFF1 347-1210) is one breakpoint among many; the script must be re-run per patient. Uniqueness was tested against the reviewed human proteome, not against the unreviewed proteome or common contaminants.
Evidence that argues against this (1)
- KMT2A breakpoints are distributed across a multi-kilobase cluster region rather than concentrated at one site, so any fixed transition list will be wrong for a large fraction of patients and a multiplexed assay would need many targets. Meyer C et al. 2023
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The KMT2A::AFF1 fusion protein is present at lower abundance than wild-type AFF1 in t(4;11) cell lines, which sets the sensitivity requirement for this assay.
AF4 encodes a ubiquitous protein that in both native and MLL-AF4 fusion types localizes to subnuclear compartments. · Li Q, Frestedt JL, Kersey JH · Blood 1998What this evidence cannot show
The statement is from densitometry of western blots with a polyclonal anti-AF4 antibody in 1998 - a relative, not absolute, comparison, and one whose antibody affinity for the two species is not established. It is nonetheless the only quantitative statement of fusion-versus-wild-type abundance this beat found for any KMT2A fusion.
Evidence that argues against this (1)
- The same paper reports that the wild-type AF4 antibody detects the fusion at the expected 240 kDa in all t(4;11) lines - so an abundance disadvantage has not prevented antibody-based detection, and the sensitivity bar for mass spectrometry may be lower than feared. Li Q et al. 1998
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Both standard t(4;11) cell lines, RS4;11 and SEMK2-M1, express the reciprocal AF4-MLL transcript as well as MLL-AF4, so both are suitable material for measuring the two junction peptides in the same sample.
t(4;11) leukemias display addiction to MLL-AF4 but not to AF4-MLL. · Kumar AR, Yao Q, Li Q, et al. · Leuk Res 2011What this evidence cannot show
An RT-PCR-level statement about transcript made as a premise rather than tested, and SEMK2-M1 is a subline so it does not strictly settle SEM. MV4-11's reciprocal status is not addressed anywhere this beat could find.
Evidence that argues against this (1)
- Because both standard lines are reciprocal-positive, there is no reciprocal-negative comparator line in routine use, so the assay cannot be validated against a true protein-level negative from the same disease - only against a KMT2A-germline line, which differs in many other ways. Agraz-Doblas A et al. 2019
transitions precomputed by the in-silico digest
the experiment that converts this absence of testing into a result
adding phospho-forms of the junction peptides to the transition list would give the first attributed fusion phosphosite
absolute copies per cell of the fusion protein, never measured
quantifies direct and reciprocal products in the same sample
the proposed fix is a PTM-aware extension of that PRM assay
the two divide the problem: the knock-in validates methods in a line, the peptide reaches patient material
the cheaper experiment available today while no junction antibody exists
the assay carried one mass; SEM needs 2048.9957, not 2281.1169
a PRM assay on the 2281.1169 mass would return an uninterpretable negative in SEM