start / graph / ana-q125-losing-pcgf1-reduces-the-silencing-of-the-vacated-sites-without-abolishing-it-in-both-repressive-marks

AnalysismoderateQ125-the-h2ak119ub-arm-of-the-pcgf1-knockout

In a PCGF1-knockout clone the menin inhibitor still evicts and the vacated sites are still silenced, but LESS: H2AK119ub at the most-evicted quartile falls from 1.4450 to 1.1325 and from 1.0151 to 0.8842 in two replicate conventions, and H3K27me3 from 2.0007 to 1.0134 - so prediction 5's silencing clause is directionally right and quantitatively overstated

also called prediction 5 in the knockout · reduced not abolished · the PCGF1 dependence of menin-inhibitor silencing

Knocking out one component of the 'switch-off' machinery makes the switching-off weaker but does not stop it. That is what you would expect when the component has five close relatives that can do part of the same job, and it means a prediction that said the sites would simply NOT be silenced was too strong.

The same thing, for a scientist

**Deposited data, re-analysed.** Same instrument, universe and set as the companion gradient entry. The comparison is the non-targeting-sgRNA control culture `NT` against the PCGF1-knockout clone `C1`, scored on `S_FLAG` - a set defined by a FLAG track in NEITHER genotype - and on Q117's four eviction quartiles of it [E1]. | VTP-50469, most-evicted quartile Q1 | control `NT` | PCGF1-KO `C1` | difference | |---|---:|---:|---:| | H2AK119ub `paired` | 1.4450 | 1.1325 | **-0.3125** | | H2AK119ub `rep2` | 1.0151 | 0.8842 | **-0.1309** | | H3K27me3 (one convention) | 2.0007 | 1.0134 | **-0.9873** | **Both repressive marks are reduced in the knockout, in every convention, at the quartile where the effect exists.** The knockout's own H2AK119ub is still graded (Spearman +1.000 in `paired`, Q1 = 1.1325), so the in-fill is **attenuated, not abolished** [E1]. **That is the expected shape for a PCGF1 knockout read with this antibody and it is stated as a limitation rather than as a result.** NEB 8240S reads H2AK119ub whichever PRC1 deposited it; PCGF1 is one of six PCGF paralogs, and PCGF2/PCGF4-containing canonical PRC1 is untouched by this edit. **A complete loss of the mark was never the right expectation and prediction 5's 'are not silenced' is therefore too strong as written** [E2]. **The aggregate over all 1,228 `S_FLAG` bins gives the UNDECIDED verdict UNDECIDED** - |R_C1 - R_NT| = 0.0813 against a PCGF1-dependent threshold of 0.2089 and a PCGF1-independent threshold of 0.0584 - and it is uninformative for the reason the companion entry gives: the control's own aggregate effect is 1.0383, i.e. 0.0383 from unity, so there is almost nothing there to be dependent ON [E1].

Go deeper

The DOT1L-inhibitor arm, which is the larger and cleaner PRC1 signal

SGC0946 5 uM 72 h against DMSO 72 h is the exposure-matched contrast in this design and it produces a much larger H2AK119ub gain than the menin inhibitor does - Q122 measured 1.4927 at S_FLAG in the F3 culture and this beat reproduces it at 1.4932 (paired) and 1.3855 (rep2) in a third and fourth culture. At the most-evicted quartile the knockout blunts it in both conventions: 1.9183 to 1.5079 (paired, -0.4104) and 1.6978 to 1.5515 (rep2, -0.1463).

At the aggregate the two conventions disagree in SIGN (+0.1895 paired, -0.1443 rep2), which is reported in its own entry and is the reason no aggregate SGC verdict is given here.

What this does to hyp-q124, and it does not overturn it

hyp-q124 holds that H3K79me2 reports the DOT1L arm of a menin inhibitor's action and is blind to its PRC1.1 arm, on the evidence that the mark's drug response is unchanged in this knockout. This entry supplies the other half of that claim for the first time: in the SAME two genotypes and the SAME samples' sibling channels, the PRC1.1 arm IS changed. H2AK119ub and H3K27me3 both move with genotype at Q1 where H3K79me2 did not move at all.

So hyp-q124's asymmetry is supported rather than refuted, and its falsifier F2 is answered in the negative: the knockout's chromatin does respond differently from its control's, so the clone is not inert and cycle 83's ARM-2 nulls are not void for that reason. But F2 as written could not have shown this - see the criterion-informativeness entry.

Facts and figures

script	analyses/q125-the-h2ak119ub-arm-of-the-pcgf1-knockout/deviation2_quartiles_and_conventions.py
status	run
result summary	H2AK119ub at Q1 falls 1.4450->1.1325 (paired) and 1.0151->0.8842 (rep2) from control to PCGF1-KO; H3K27me3 2.0007->1.0134. Reduced, not abolished: the knockout's own gradient is still Spearman +1.000 in the paired convention.
absence class	measured-not-analysed
absence class note	The 2 x 3 NTvKO H2AK119ub and H3K27me3 designs were deposited in GSE260742 and public from 2025-11-23 and had never been binned. The quantity that remains NEVER-MEASURED is PCGF1 itself in these cells: no protein or transcript measurement of the knockout is deposited for any of the 42 NTvKO samples, so knockout efficiency is inferred from phenotype in all three validations here.
absence class assigned	2026-09-23 (Q125)

inputs

[
  "GSM8123399/401/402 and GSM8123403/405/406 (H2AK119ub paired, C1 and NT)",
  "GSM8123417/418/419 and GSM8123420/421/422 (H2AK119ub rep2)",
  "GSM8123423/424/425 and GSM8123426/427/428 (H3K27me3)"
]

How this could bite us back

If PRC1.1 deposition is part of how a menin inhibitor works, then a PCGF1-, BCOR- or KDM2B-deficient leukaemia is a partial-response genotype whose fusion is still evicted. A pharmacodynamic assay reading fusion occupancy would call that patient fully engaged while the silencing arm is attenuated.
The effect is attenuation and not abolition, which is the harder case to detect clinically: a 20-30% reduction in a chromatin readout will not separate responders from non-responders in a small trial, so the resistance mechanism this entry supports is one that is easy to miss and hard to prove.

Still open

Would a BCOR or KDM2B knockout give a larger effect than PCGF1? PRC1.1 targeting runs through KDM2B's CxxC domain and BCOR is the AHD contact; neither is deposited in this series, so the disjunction in prediction 5 ('PCGF1 or BCOR') remains half-unmeasured.
Is the residual silencing in the knockout canonical PRC1, or incomplete editing? A second clone, a rescue, or a PCGF2/PCGF4 ChIP would separate them and none exists in the deposit.
Does the reduced silencing translate into reduced drug response? No viability, differentiation or transcriptional readout from these exact cultures has been located.

Every source behind this entry (2)

E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In murine MLL-AF9, at the most-evicted quartile of the FLAG-defined top-0.1% set, 500 nM VTP-50469 48 h against DMSO 72 h gives H2AK119ub 1.4450 in the non-targeting control against 1.1325 in the PCGF1-knockout clone in the paired convention (GSM8123403/406 vs GSM8123399/402) and 1.0151 against 0.8842 in rep2 (GSM8123420/422 vs GSM8123417/419), and H3K27me3 2.0007 against 1.0134 (GSM8123426/428 vs GSM8123423/425); the aggregate over all 1,228 bins gives |R_C1 - R_NT| = 0.0813, an UNDECIDED verdict against thresholds of 0.0584 and 0.2089.
Q125 re-analysis of GSE260742: the PCGF1-knockout genotype comparison at the eviction quartiles. analyses/q125-the-h2ak119ub-arm-of-the-pcgf1-knockout/run.py, deviation1_hg38_container.py, deviation2_quartiles_and_conventions.py, deviation3_matched_null.py and results.json, run 2026-09-23 · Iqbal Lab oncofusion knowledge graph 2026
What this evidence cannot show
NO SPIKE-IN NORMALISATION IS POSSIBLE. Q117 verified zero hits for the entire spike-in vocabulary across all 419,824 characters of GSE260742's 158 sample records and no scale factor is deposited, so every cross-condition number here is a WITHIN-SAMPLE SHARE, which cancels the unknown per-sample scale factor exactly and IS BLIND TO A UNIFORM CHANGE BY CONSTRUCTION. Nothing here measures any mark's abundance; everything measures its distribution. EXPOSURE ASYMMETRY: DMSO 72 h and SGC0946 5 uM 72 h are matched, VTP-50469 500 nM is 48 h, so every VTP ratio is a LOWER BOUND on the 72-hour effect - identical in both genotypes, so the genotype DIFFERENCE is unaffected. REAGENTS: H2AK119ub is Ubiquityl-Histone H2A (Lys119) (D27C4) XP Rabbit mAb, NEB 8240S, and it CANNOT distinguish PRC1.1 from canonical PRC1 - RING1A/B is the ligase in both and PCGF1 is one of six PCGF paralogs - nor deposition from loss of a deubiquitinase. The H3K27me3 and H3K4me3 reagents are NOT RESOLVED and are recorded unresolved rather than guessed; `antibod` occurs zero times in the 158 GEO records. S_FLAG is defined by DYKDDDDK Tag (D6W5B) Rabbit mAb, NEB 14793S, on a RETROVIRALLY OVER-EXPRESSED tagged MLLAF9 construct, so it is the over-expressed molecule's sites and not an endogenous fusion's. THE KNOCKOUT is one CRISPR guide, one clone C1, no second clone, no rescue and NO deposited PCGF1 protein or transcript measurement. SYSTEM: murine MLL-AF9 (KMT2A::MLLT3) myeloid blood cells, NOT the infant B-lineage KMT2A::AFF1 disease this graph is about. THE QUARTILE-LEVEL GENOTYPE DIFFERENCE IS DERIVED, NOT PRE-REGISTERED AS A SCORED CONTRAST: the per-quartile ratios in both genotypes were pre-registered (criterion G1, 'scored in both genotypes'), but applying the 0.2089 / 0.0584 thresholds to their DIFFERENCE is POST HOC and is labelled so in the analysis. On that post-hoc application the paired convention (-0.3125) would be PCGF1-DEPENDENT and rep2 (-0.1309) UNDECIDED. H3K27me3 has one convention and no spread.
Evidence that argues against this (2)
- Egan B et al. 2016
- NCBI Gene Expression Omnibus 2026
E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
PCGF1 is one of six PCGF paralogs and defines the non-canonical PRC1.1 complex; the H2AK119ub ligase RING1A/RING1B is shared with canonical PRC1, so an antibody against the mark cannot attribute it to PRC1.1 and losing PCGF1 is not expected to abolish the mark.
Targeting Menin disrupts the KMT2A/B and polycomb balance to paradoxically activate bivalent genes. · Sparbier CE, Gillespie A, Gomez J, et al. · Nature Cell Biology 2023
PMID 36635503 DOI
What this evidence cannot show
Cited for the complex architecture, which is background in that paper rather than its result. Its own result - that menin inhibition disrupts the KMT2A/B-polycomb balance and paradoxically ACTIVATES bivalent genes - is the competing account of what a menin inhibitor does to polycomb and is engaged in the counter-evidence.
Evidence that argues against this (1)
- Sparbier CE et al. 2023

Connected to

refutes →Menin inhibition removes the fusion from its strongest sites without relocating it to a new programme, and PRC1.1 fills in only where the eviction is complete - so the drug's chromatin action is subtraction plus local substitution, not redistribution
prediction 5's 'the vacated sites are not silenced' is too strong: silencing is reduced, not abolished
supports →H3K79me2 reports only the DOT1L arm of a menin inhibitor's action and is blind to its PRC1.1 arm - so the mark's response is unchanged in a PCGF1 knockout even though PCGF1 loss is a published resistance mechanism, and a cell resistant through PRC1.1 loss would look fully drug-engaged by this readout
supplies the missing half: the PRC1.1 arm DOES change with genotype where H3K79me2 did not
depends on →PCGF1 — the PRC1.1 scaffold whose LOSS causes menin-inhibitor resistance, so it is a restraint on the leukaemia rather than a dependency of it
the knocked-out subunit
related to →H3K27me3 - the repressive mark this graph argued about for seventy-seven cycles without ever writing an entry for it
second silencing readout, first binning of this arm
supports →A menin inhibitor does not only take something away - it causes PRC1.1 to put something on, and the drug needs the cell's own silencing machinery to work
the mark itself, graded by eviction depth and reduced in a PCGF1 knockout
related to →GSE260742's PCGF1-knockout samples carry no deposited measurement of PCGF1 - no protein, no transcript, no sequencing of the edited locus - and the knockout rests on one guide and one clone with no rescue, so every null result from that arm is a null in an unverified genotype
the absence this entry works around with three phenotype validations
← supportsQ124's criterion H2 was written to be able to void Q124's own conclusion, and it cannot - because it is scored on a contrast in which the control genotype shows no effect to lose. Its raw arithmetic FIRES at 0.0813 against a 0.05 threshold, which reads as 'the knockout is a knockout' and is not evidence of anything
the three validations that did work, because their control effects are far from unity