PROPOSED: a side-by-side CHD4 co-IP in a t(4;11) B-ALL line and a KMT2A-wildtype B-ALL line, probing one subunit from each NuRD module - the first experiment that would ask whether a KMT2A fusion sees the whole complex
also called the module-stoichiometry co-IP · Q193 experiment
Pull down the NuRD motor protein from a leukaemia cell that carries the KMT2A::AFF1 fusion and from one that does not, run them side by side on the same blot, and ask whether the deacetylase half of the machine comes along in both. Nobody has ever done this in a KMT2A-rearranged cell.
The same thing, for a scientist
Every statement in this beat about what a KMT2A fusion associates with comes from one AP-MS performed in **Jurkat, a T-ALL line**, with a doxycycline-induced Strep-HA transgene, from bulk nuclear lysate, and reported as a top-300-per-bait rank list. The design below replaces all four of those limitations at once for a single focused question: **CHD4 immunoprecipitation from SEM (human t(4;11), KMT2A::AFF1, B-lineage, infant-derived) and from 697 (TCF3::PBX1, KMT2A-wildtype B-lineage), side by side on the same blot, probed for GATAD2A (remodelling module) and MTA2 and HDAC1 (deacetylase module), with quantification as a ratio to recovered CHD4.** It tests the prediction of `hyp-mll-fusions-engage-the-nurd-remodelling-arm-without-its-deacetylase-arm` in the cell type the disease actually occurs in, and it produces the first stoichiometric statement about NuRD in a KMT2A-rearranged cell. It is deliberately ordered **second**: the PXD009338 re-search costs compute and adjudicates the existing data, and should run first.
Design
Cells. SEM (t(4;11), KMT2A::AFF1, human infant B-ALL) and RS4;11 (t(4;11)) as the fusion arm; 697 (TCF3::PBX1) and NALM-6 as KMT2A-wildtype B-lineage comparators. These are the exact lines used by the CRISPR screen that made MBD3 a dependency, so the results join.
Bait. Endogenous CHD4, by antibody. Not a tagged transgene - the Jurkat dataset's bait was induced and tagged, and reproducing that would reproduce its limitation.
Probes. GATAD2A and GATAD2B (remodelling module); MTA2 and HDAC1 (deacetylase module); MBD3 and MBD2 (the paralogue pair, because this graph cannot currently say which form it has been talking about); KMT2A N-terminal (to ask directly whether the fusion is in the CHD4 pulldown at all).
Readout. Band intensity of each probe as a ratio to recovered CHD4 in the same lane, so the comparison is stoichiometric and survives differences in IP efficiency between lines. Both lines on the same blot, three biological replicates.
Controls.
- IgG IP from each line - the essential one, because CHD4 carried 2|2 negative-control spectra in the source dataset and may be sticky.
- Input blots for every probe, so a low co-IP signal from a low-expressing protein is not read as low stoichiometry. This control is what separates 'the module is not in the complex' from 'the subunit is not in the cell', and its absence is how this experiment would fail silently.
- A reciprocal IP of MTA2, asking whether CHD4 comes back. An asymmetric result - CHD4 does not pull MTA2 but MTA2 pulls CHD4 - would indicate a sub-stoichiometric population rather than an absent one, which is a different and more interesting answer than either of the two the hypothesis predicts.
What each outcome means, decided before the data
| result | reading |
|---|---|
| MTA2/CHD4 and HDAC1/CHD4 ratios more than 2-fold lower in SEM than in 697, with equal inputs | supports module-selective association at the fusion; the hypothesis's central prediction |
| ratios within 2-fold between the lines | the module asymmetry seen in the Jurkat AP-MS is not a fusion property; the hypothesis is refuted in the cell type that matters |
| no MTA2 or HDAC1 in either line | NuRD's deacetylase module does not co-IP with CHD4 under these conditions at all - a statement about the assay, and the experiment must be repeated with crosslinking before anything is concluded |
| KMT2A N-terminal band absent from the CHD4 IP in SEM | the premise of the whole beat weakens sharply: the fusion would not be in a CHD4 complex in a t(4;11) cell, and the Jurkat co-purification would look like a tagged-overexpression artefact |
That last row is the one worth flagging. The experiment can undercut the beat rather than extend it, and it is proposed partly for that reason: no published experiment has ever asked whether a KMT2A fusion and CHD4 are in one complex in a t(4;11) cell, and the graph currently assumes they are on the strength of an induced transgene in a T-ALL line.
| design | Endogenous CHD4 immunoprecipitation from SEM and RS4;11 (t(4;11), KMT2A::AFF1) and from 697 (TCF3::PBX1) and NALM-6 (KMT2A-wildtype), all B-lineage, run side by side on the same blot in three biological replicates. Probes: GATAD2A, GATAD2B (remodelling module); MTA2, HDAC1 (deacetylase module); MBD2, MBD3 (the paralogue pair); KMT2A N-terminal (is the fusion in the complex at all). Plus a reciprocal MTA2 IP probed for CHD4. |
|---|---|
| readout | Band intensity of each probe expressed as a RATIO to recovered CHD4 in the same lane, compared between the fusion-positive and fusion-negative arms. Pre-specified threshold: a more than 2-fold lower deacetylase-module ratio in the t(4;11) arm, with equal inputs, supports module-selective association. |
| controls | IgG IP from every line (essential - CHD4 carried 2|2 negative-control spectra in the source AP-MS and may be sticky). INPUT BLOTS FOR EVERY PROBE, which is the control that separates 'the module is not in the complex' from 'the subunit is not in the cell' and whose absence is how this experiment would fail silently. Reciprocal MTA2 IP, because an asymmetric result indicates a sub-stoichiometric rather than an absent population - a third answer the hypothesis does not predict. |
| cost tier | low |
| timeline | Weeks. Standard antibodies exist for CHD4, HDAC1, MTA2, GATAD2A and the MBDs; nothing here requires a new reagent, a knock-in or an animal. |
| order of operations | RUN SECOND. ana-q193-planned-re-search-of-pxd009338-for-the-nurd-deacetylase-module costs compute only, adjudicates the existing dataset, and would change what this experiment is worth doing - if the re-search shows the deacetylase module was merely rank-truncated, the module-selectivity question is answered and this experiment becomes a stoichiometry measurement rather than a test. |
| what it would NOT settle | Occupancy. This is a solution-phase co-IP and says nothing about whether any of these proteins is at a KMT2A::AFF1 target gene; a scaffold-directed ChIP is a separate experiment and is named as an open question in gene-sin3a and gene-gatad2a. |
| falsifier it serves | The second falsifier of hyp-mll-fusions-engage-the-nurd-remodelling-arm-without-its-deacetylase-arm: MTA2 and HDAC1 recovered at a CHD4-normalised ratio within 2-fold between SEM and 697, on the same blot, would refute module-selective association in the disease-relevant lineage. |
| relationship to existing entries | The wet-work test of hyp-mll-fusions-engage-the-nurd-remodelling-arm-without-its-deacetylase-arm, ordered after ana-q193-planned-re-search-of-pxd009338-for-the-nurd-deacetylase-module. It uses the cell lines from gene-mbd3's and gene-arid4b's CRISPR evidence so the binding and dependency results would join. It does not duplicate any existing experiment node in this corpus: no CHD4 or NuRD co-IP in a KMT2A-rearranged cell is proposed or recorded anywhere in the graph. |
what it would settle
[ "Whether a KMT2A fusion is in a CHD4 complex in a t(4;11) B-lineage cell at all - which no published experiment has asked, and which the whole beat currently assumes from a tagged transgene in a T-ALL line.", "Whether NuRD's deacetylase module travels with CHD4 at different stoichiometry in fusion-positive and fusion-negative B-ALL.", "Which MBD paralogue the complex in these cells contains - the question that decides whether this graph's NuRD discussion and its MBD3 dependency concern the same complex." ]
measurements
[
{
"quantity": "prespecified_effect_threshold",
"subject": "deacetylase-module-to-CHD4 band ratio, t(4;11) arm versus KMT2A-wildtype arm",
"value": 2,
"units": "fold difference",
"n": 3,
"universe": "three biological replicates, four B-lineage cell lines, same-blot comparison, inputs matched",
"null": 1,
"note": "threshold set before the experiment; a difference below 2-fold is scored as refuting module-selective association rather than as a weak positive"
}
]- A negative result here is ambiguous in a way the outcome table cannot fully fix: a transient association that does not survive lysis looks identical to no association. Whoever runs this should decide in advance whether a crosslinked repeat is part of the experiment or a separate one.
- The experiment could undercut the beat rather than extend it. If the KMT2A N-terminal band is absent from the CHD4 IP in SEM, then the co-purification this graph has built several entries around is a property of an induced transgene in a T-ALL line, and those entries need re-reading rather than extending.
- Four cell lines is a statement about four cell lines. Any ratio difference found here needs a panel or a primary sample before it becomes a statement about the disease.
- Should this be done with crosslinking? A transient CHD4-core association may not survive a standard lysate co-IP at all, in which case a negative result would be uninterpretable - and the Drosophila work says the association IS transient.
- Is a fusion-specific antibody available for KMT2A::AFF1, or does the KMT2A N-terminal probe detect wild-type KMT2A too? If only the N-terminus can be probed, the 'is the fusion in the complex' arm is confounded by the wild-type allele and needs a tagged knock-in instead.
- Would size-exclusion fractionation before the IP add more than it costs? It would say which NuRD species the fusion travels with rather than only which subunits co-IP, and that is closer to the question the hypothesis actually asks.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
All existing co-purification evidence placing any NuRD subunit near a KMT2A fusion comes from a single affinity purification in Jurkat, a T-ALL line, using doxycycline-induced Strep-HA-tagged single-copy fusion transgenes and bulk nuclear lysate, reported as the top ~300 preys per bait; in that experiment CHD4 is the only NuRD subunit in the MLL-AF4 column, at p = 0.0116 with 7|9 spectra against a non-zero negative control of 2|2.
MLL-fusion-driven leukemia requires SETD2 to safeguard genomic integrity. · Skucha A, Ebner J, Schmoellerl J, Roth M, Eder T, Cesar-Razquin A, Stukalov A, Vittori S, Muhar M, Lu B, Aichinger M, Jude J, Mueller AC, Gyorffy B, Vakoc CR, Valent P, Bennett KL, Zuber J, Superti-Furga G, Grebien F · Nature Communications 2018What this evidence cannot show
The CHD4 numbers are from Supplementary Table 5, not the body text - hence support_location 'supplement' - and the expected_absent_term records that 'NuRD' does not appear in the paper's body at all, a measurement made by an earlier beat of this project and not re-run by me. The claim that this is the ONLY such evidence rests on my bounded literature search returning zero KMT2A-fusion records for MTA2, GATAD2A and SIN3A and one for MBD3 (a dependency screen), which bounds reporting rather than measurement.
Evidence that argues against this (1)
- A single-source premise is a reason for caution about the experiment's motivation as well as about the literature. If the Jurkat co-purification is itself an artefact of tagged overexpression, this experiment's most likely outcome is the fourth row of the outcome table - no fusion in the CHD4 IP - in which case it refutes the premise rather than testing the module question, and the beat should say so plainly rather than treating a negative as a technical failure. This repository 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
SEM and ALL-PO are KMT2A::AFF1-positive human B-ALL lines and 697 (TCF3::PBX1) and NALM-6 are the KMT2A-wildtype B-lineage comparators used in the CRISPR screen that identified MBD3 and ARID4B as KMT2A-rearranged-selective vulnerabilities, so a co-IP run in the same lines joins directly to the only dependency evidence in this beat.
CRISPR-Cas9 Library Screening Identifies Novel Molecular Vulnerabilities in KMT2A-Rearranged Acute Lymphoblastic Leukemia. · Schneider P, Wander P, Arentsen-Peters STCJM, Vrenken KS, Rockx-Brouwer D, Adriaanse FRS, Hoeve V, Paassen I, Drost J, Pieters R, Stam RW · International Journal of Molecular Sciences 2023What this evidence cannot show
The line identities were read from the PMC full text on 2026-09-24, because the MCP PubMed tool's rendering of the abstract deletes italicised gene symbols (the raw PubMed record keeps them). RS4;11 is NOT from this paper - it is added here as a second t(4;11) line on general grounds, and note that this corpus's symbol registry records MV4-11/MV4;11/MV411 as a three-token naming hazard, which applies to RS4;11 and RS4-11 as well. Cell-line identity should be STR-verified before use; this graph has no independent confirmation of any of these lines' provenance.
Evidence that argues against this (1)
- Two KMT2A::AFF1 lines and two wildtype lines is a four-line comparison, which is underpowered for any claim about fusion-driven biology: line-to-line variation in NuRD subunit expression alone could produce a 2-fold ratio difference. The input-blot control limits that but does not remove it, and a result resting on SEM versus 697 alone would be a statement about two cell lines. DepMap via this repository 2026
side-by-side CHD4 co-IP in t(4;11) versus KMT2A-wildtype B-ALL, module stoichiometry as the readout
endogenous CHD4 is the bait; the experiment also asks whether the fusion is in a CHD4 complex in a t(4;11) cell at all
MTA2 is the deacetylase-module probe