A time-resolved BRG1/RING1B/H2AK119ub CUT&RUN at BENC under menin inhibition - the inv(16) experiment, run on the right disease
also called the BENC swap experiment · menin inhibitor MYC enhancer time course
Treat KMT2A-rearranged leukaemia cells with a menin inhibitor, then at a series of short time points measure which proteins are sitting on the DNA switches that control MYC. If a silencing protein arrives before an activating protein leaves, the drug is working the way it does in a different leukaemia where this has already been shown.
The same thing, for a scientist
A direct transplant of the inv(16) chromatin experiment [E1] into KMT2A-rearranged cells, testing `hyp-menin-inhibition-swaps-brg1-for-ring1b-at-benc`. Design: a human KMT2A-rearranged line (MV4-11 or MOLM-13, KMT2A::AFF1 and KMT2A::MLLT3 respectively) treated with a menin inhibitor at a fixed, fully occupying dose; spike-in-normalised CUT&RUN at 0, 1, 2, 4, 6, 8 and 24 h for **BRG1 (SMARCA4)**, **RING1B (RNF2)**, **H2AK119ub**, **H3K27ac** and at least one PRC1.1-defining subunit (**PCGF1** or **KDM2B**); matched nascent-transcription readout for MYC; and the whole series repeated in PCGF1- and BCOR-knockout derivatives. Positive control locus: a canonical menin-inhibitor-silenced target where H2AK119ub is expected to rise [E3]. Negative control locus: **N-ME**, the +1.4 Mb NOTCH-dependent T-cell MYC enhancer that showed no RING1B recruitment in inv(16) [E1]. **Step zero, which is not a control but the gating experiment: a BRG1 CUT&RUN in the untreated line.** If BRG1 is not at BENC in human KMT2A-rearranged cells, the hypothesis is untestable there and the reason may be the published human-cell SWI/SNF negative [E4].
What each arm decides
Arm 0 - baseline occupancy (gating). BRG1 and BRD4 CUT&RUN at BENC in the untreated line, plus the same in a second line. Decides whether the experiment can be run at all, and simultaneously delivers something the graph does not currently have: a human KMT2A-rearranged BENC occupancy map. This arm is worth running even if everything else is abandoned.
Arm 1 - the swap and its order. The time course. The discriminating quantity is order, not magnitude: RING1B up before BRG1 down supports replacement; the reverse or simultaneous supports vacancy; neither moving while MYC falls supports the differentiation-readout model in mechanism-myc-after-menin-inhibition-direct-or-differentiation-readout. The inv(16) numbers give the window to sample: repressor at 2.5-5 h, activator leaving at 4-6 h 1, so 1/2/4/6/8 h is the informative grid and 24 h is a late anchor.
Arm 2 - is it PRC1.1? PCGF1 or KDM2B CUT&RUN alongside RING1B. This is the step the inv(16) paper omitted, and without it a positive result is a PRC1 result that does not connect to Neville et al. 2 or to this graph's PRC1.1 layer.
Arm 3 - epistasis. PCGF1 and BCOR knockouts. The prediction is that the BENC chromatin response and the MYC fall are both blunted, giving the locus-level version of the published whole-cell resistance result 3. A knockout that abolishes the MYC fall without abolishing the chromatin change would separate them and would be the most interesting possible outcome.
Arm 4 - the combination prediction. Menin inhibitor with and without a BET inhibitor, MYC readout plus BRD4 occupancy at BENC. In inv(16) the interface inhibitor alone was insufficient and JQ1 was required; if BRD4 stays at BENC after the swap, that is the mechanistic reason and it transfers.
Assay choices, and why each one is not the obvious one
CUT&RUN rather than ChIP-seq, because the effect is expected at a small number of defined elements and CUT&RUN's signal-to-noise at low cell input is what makes a patient-sample version conceivable later.
Spike-in normalisation is mandatory, not optional. The DOT1L-inhibitor arm of this biology is a genome-wide H2AK119ub increase 2. Any normalisation that assumes constant total signal will erase a global change and can invert the sign of a local one. This is the single likeliest way for the experiment to produce a confident wrong answer.
Nascent transcription (SLAM-seq or PRO-seq) rather than steady-state RNA for MYC. MYC mRNA has a short half-life but the question is ordering on a 2-6 hour scale, and steady-state RNA at those timepoints is a convolution of synthesis and decay. The published inv(16) work used qRT-PCR, which was adequate for a yes/no and is not adequate for an ordering claim.
H2AK119ub antibody validation against a RNF2-null control, because the mark is a ubiquitin adduct and the antibody literature for it is mixed. A knockout-validated reagent is the difference between this experiment and a plausible artefact.
Enhancer identity must be fixed in advance from the published coordinates, and the ambiguity flagged in concept-a-noncoding-element-has-no-symbol-and-three-names applies: BENC modules and Pulikkan's E1-E5 numbering are not aligned in any published table this beat could find, so the amplicons must be defined from a track, not from a module name.
| design | Gated time-course CUT&RUN. Arm 0 (gating): BRG1 and BRD4 CUT&RUN at BENC in two untreated human KMT2A-rearranged lines. Arm 1: menin inhibitor at fully occupying dose, spike-in-normalised CUT&RUN at 0/1/2/4/6/8/24 h for SMARCA4, RNF2, H2AK119ub and H3K27ac. Arm 2: add PCGF1 or KDM2B CUT&RUN at the same timepoints to establish the PRC1 flavour. Arm 3: repeat arms 1-2 in PCGF1-knockout and BCOR-knockout derivatives. Arm 4: menin inhibitor with and without a BET inhibitor, with BRD4 occupancy at BENC and MYC nascent transcription. |
|---|---|
| readout | occupancy of SMARCA4, RNF2, PCGF1/KDM2B, BRD4 and H2AK119ub/H3K27ac at BENC modules, at ME1 and ME2, at a menin-inhibitor-silenced positive-control locus and at the N-ME negative-control enhancer; plus MYC nascent transcription (SLAM-seq or PRO-seq) on the same timepoints |
| controls | N-ME (+1.4 Mb NOTCH-dependent T-cell MYC enhancer) as a non-myeloid negative locus, taken from the inv(16) design; a canonical menin-inhibitor-silenced target as a positive locus so that a null at BENC is interpretable; RNF2-null cells for H2AK119ub antibody validation; spike-in chromatin in every reaction because the DOT1L arm of this biology is genome-wide and ratio normalisation would erase it |
| primary discriminating quantity | the ORDER of the two occupancy changes, not their magnitude - RING1B up before BRG1 down is replacement; simultaneous or reversed is vacancy |
| cost tier | med |
| timeline | arm 0 in 3-4 weeks; arms 1-2 in 3 months; arms 3-4 in 6 months |
| status | planned |
systems
[ "MV4-11 (KMT2A::AFF1, myeloid/monocytic)", "MOLM-13 (KMT2A::MLLT3)", "a B-lineage KMT2A::AFF1 model if one can be found with measurable BRG1 at BENC - the project's index disease has no representation in any supporting result" ]
tests
[ "hyp-menin-inhibition-swaps-brg1-for-ring1b-at-benc" ]
failure modes to pre-register
[ "ratio normalisation instead of spike-in - would erase a genome-wide change and could invert a local one", "steady-state MYC mRNA instead of nascent transcription - cannot support an ordering claim on a 2-6 h scale", "an unvalidated H2AK119ub antibody - the mark is a ubiquitin adduct and reagent quality is mixed", "defining BENC modules by name rather than by coordinate - the module numbering schemes of the two source papers are not aligned" ]
- A negative result at BENC does not exonerate the MYC axis; it relocates it. MYC could be controlled from the promoter, from ME1/ME2, or not directly at all - so a null here must be reported as a null about one element, not about MYC.
- Arm 3's knockouts remove machinery that is separately reported to be required for leukaemia maintenance, so a blunted response could be a fitness artefact rather than an epistasis result; a knockout fitness curve has to be run alongside.
- If the experiment succeeds, the resulting assay would be used to declare menin-inhibitor target engagement in patients, and it would read as silent in precisely the case where MYC is sustained from an alternative enhancer - a false reassurance at the moment of resistance.
- The whole design imports timings and a control locus from a different fusion family and a different drug; if the KMT2A kinetics are slower, a 1-24 h grid could miss the event entirely and return a false negative.
- Which human KMT2A-rearranged line has BRG1 at BENC? Arm 0 answers it and nobody has published it.
- Can a reduced version of arm 1 - two timepoints, three antibodies - be run on a primary patient sample at diagnosis and on-treatment? That is the version that would turn this into a clinical pharmacodynamic assay rather than a cell-line mechanism study.
- Is there a B-lineage KMT2A::AFF1 model in which BENC is active at all? Every supporting result in this design is myeloid, and the project's index disease is lymphoid.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The design being transplanted: in inv(16) AML, ChIP-qPCR time course at MYC enhancer element E3 showed RUNX1 and RING1B occupancy rising from about 2.5 hours and reaching 90% by 5 hours while BRG1 fell between 4 and 6 hours; RING1B was recruited to ME1, ME2 and E3 but not to the N-ME control enhancer; and CRISPR deletion of the RUNX1 site in any one of the three reduced MYC by 40-50% and viability by 60-70%.
CBFbeta-SMMHC Inhibition Triggers Apoptosis by Disrupting MYC Chromatin Dynamics in Acute Myeloid Leukemia · Pulikkan JA, Hegde M, Ahmad HM, Belaghzal H, Illendula A, Yu J, O'Hagan K, Ou J, Muller-Tidow C, Wolfe SA, Zhu LJ, Dekker J, Bushweller JH, Castilla LH · Cell 2018What this evidence cannot show
Read in full from PMC6211564 on 2026-09-20. ChIP-qPCR at named amplicons rather than genome-wide, in one line (ME-1), with H3K27me3 rather than H2AK119ub as the repressive mark and no PRC1 flavour identified. The timings are for a different drug in a different disease and are used here to set a sampling grid, not to predict the answer.
Evidence that argues against this (1)
- The mechanism was complete and the single agent did not clear disease; JQ1 was required. An experiment that reproduces the chromatin result in KMT2A-rearranged cells would therefore not by itself support a single-agent menin strategy, which is why arm 4 is part of the design rather than a follow-up. Rathert P et al. 2015
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin inhibition induces PRC1.1-dependent H2AK119ub deposition at a subset of MLL-fusion targets, whereas DOT1L inhibition produces a genome-wide increase in H2AK119ub.
DOT1L provides transcriptional memory through PRC1.1 antagonism. · Neville D, Ferguson DT, Heikamp EB, Lai Z, Magor GW, Lam C, et al. · Nat Cell Biol 2026What this evidence cannot show
Abstract-level. The genome-wide DOT1L result is what makes spike-in normalisation mandatory for this design; the subset-restricted menin result is what makes a locus-anchored assay necessary rather than a bulk blot. The subset is not enumerated, so the positive-control locus has to be taken from a different paper.
Evidence that argues against this (1)
- Zhou et al. describe H2AK119ub as a pre-existing bivalent feature of the non-canonical menin targets rather than a drug-induced deposition. If the mark is already present before treatment at the relevant loci, the experiment's expected signal is a change in occupancy of the complex rather than a gain of mark, and the design must be able to see both. Zhou X et al. 2024
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Loss of PRC1.1 subunits such as PCGF1 or BCOR confers menin-inhibitor resistance by epigenetic reactivation of non-canonical menin targets including MYC, and MYC inhibition resensitises PRC1.1-deficient cells; the targets are bivalently co-occupied by active menin and repressive H2AK119ub.
Epigenetic regulation of noncanonical menin targets modulates menin inhibitor response in acute myeloid leukemia. · Zhou X, Zhang L, Aryal S, Veasey V, Tajik A, Restelli C, et al. · Blood 2024What this evidence cannot show
Whole-cell epistasis with no locus resolution at MYC regulatory elements, in human and mouse KMT2A-rearranged AML models with no patient confirmation. It establishes that MYC is under PRC1.1 control in this setting and supplies arm 3's knockouts; it does not say where on the MYC locus that control is exerted, which is the gap this experiment addresses.
Evidence that argues against this (1)
- Rossi et al. show PRC1 activity and H2AK119ub are REQUIRED for the leukaemogenic potential of MLL fusions in mice, the opposite therapeutic sign. PCGF1 and BCOR knockouts may therefore be poorly tolerated in some backgrounds, and a failure of arm 3 could reflect fitness cost rather than mechanism. Rossi A et al. 2016
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In human MLL-rearranged cells, SMARCA4, SMARCD2 and DPF2 are not required for the expression of c-MYC-regulated genes, in explicit contrast to previous results obtained in mouse cells.
SWI/SNF Subunits SMARCA4, SMARCD2 and DPF2 Collaborate in MLL-Rearranged Leukaemia Maintenance. · Cruickshank VA, Sroczynska P, Sankar A, Miyagi S, Rundsten CF, Johansen JV, et al. · PLoS One 2015What this evidence cannot show
A gene-signature readout, not an occupancy measurement; the paper reports no enhancer ChIP. It is the reason arm 0 exists and it does not by itself predict that BRG1 is absent from BENC in human cells - only that its loss does not move a MYC signature there.
Evidence that argues against this (1)
- Shi et al. measured Brg1 occupancy at the element directly and showed it is required for transcription-factor occupancy and looping. Occupancy and signature dependency are different quantities; arm 0 measures the one Cruickshank et al. did not. Shi J et al. 2013
gated time-course CUT&RUN; the discriminating quantity is the order of the two occupancy changes
the assay reads the mark directly rather than substituting H3K27me3, which is what the inv(16) study did
arm 2 and arm 3: does a PRC1.1-defining subunit arrive, and does its loss blunt the response
arm 3 is the locus-level version of the published whole-cell PRC1.1 resistance epistasis
why that experiment's PRC1 arm must be genetic rather than pharmacological