The 24 genes that respond to BOTH menin inhibitors in RS4;11 are led by the ATF4 stress programme - CHAC1 -5.11, STC2 -4.40, TRIB3 -3.10 - and CHAC1 and TRIB3 are named zero times in this graph

Twenty-four genes is the intersection of a 61-gene and an 870-gene list, so it is small, and the parent analysis measured the enrichment as 5.79x over chance with 75% directional concordance - meaning roughly a quarter of the overlap is expected to be coincidence, and eight genes did in fact disagree. Individual genes on this list are therefore NOT individually established as menin-dependent; the list as a set is what carries the signal. Several concordant genes sit close to the RNA-seq detection floor (RGS2 1.08, SGK1 1.02, KCNK12 1.08, CASZ1 1.26, IKZF3 1.58), where a large fold change is unstable. No p-values.

• The concordance requirement cuts both ways as evidence. Requiring a gene to cross 1 log2 on two different platforms with different detection characteristics selects for large-effect, well-measured genes AND for genes whose array probe happens to behave, so the 24 are a biased sample of the shared response rather than its strongest members. The parent analysis's Pearson r of 0.184 over all 899 movers says most of the shared axis is not in this list.

Both anchor citations were fetched and verified on 2026-09-22 - PMID 19109178 for CHAC1 and PMID 15775988 (Ohoka N et al., 'TRB3, a novel ER stress-inducible gene, is induced via ATF4-CHOP pathway and is involved in cell death', EMBO J 2005, doi 10.1038/sj.emboj.7600596) for TRIB3 - and a first guess of PMID 19109178's identity from recall was WRONG before fetching (it had been mistyped as 19188445, which is an APE1/NPM1 nucleolar paper). ATF4-target status for SESN2 and STC2 is asserted here at the level of the ISR literature generally and NOT from a fetched primary paper for each, so those two are the weakest members of the identification. Critically, ATF4 itself is not measurable in this comparison: it is not among the concordant genes, and no ATF4 protein or activity measurement exists in either dataset. The identification is of a SIGNATURE, by its targets, not of pathway activity.

• A falling ISR signature has an entirely mundane alternative explanation that this analysis cannot exclude: cells that are dying, differentiating or simply growing more slowly under a drug reduce translation and nutrient demand, and the ISR output falls as a consequence rather than as a mechanism. The ISR is a general readout of cellular stress, so its collapse under an effective anti-leukaemic drug may be downstream of the phenotype rather than part of the drug's action. Distinguishing them needs a timecourse and a growth-matched control, and neither dataset has one.
• The direction is the opposite of the naive expectation. If a drug is killing leukaemia cells one would expect stress-response genes to RISE, and here CHAC1, TRIB3, ATF3 and SESN2 all fall by 1.4-5.1 log2. That inversion is interesting and is also a warning: it does not fit the simplest 'drug stresses cell' model, and any mechanism proposed for it has to explain why ISR output goes DOWN.

Abstract-level; the full text was not read, so what 'regulate ATF4 activity' means mechanistically in that system - transcriptional, post-translational, or via serine availability - is not established here. Ewing sarcoma is a different disease, a different fusion, a different lineage and a mesenchymal rather than haematopoietic context, and EWS-FLI1 has no menin-binding motif analogous to KMT2A's MBM. The convergence is a reason to look, not evidence that the same axis operates in KMT2A::AFF1. Note also a metadata discrepancy recorded rather than silently fixed: this graph's two existing citations of the paper give the year as 2021 while efetch returns 2022, most likely an epub/issue split.

• The direction of the Ewing finding does not obviously match what was measured here. There, ATF4 is HIGH and menin supports serine biosynthesis through it; here, menin inhibition LOWERS ATF4 target output - which is consistent. But in Ewing sarcoma the relevant menin activity is in a fusion that does not bind menin through a KMT2A-like interface, so if both are real the shared node is menin itself rather than the fusion, and that would predict the same ATF4 effect in cells with no fusion at all. Neither dataset contains a KMT2A-wild-type line treated with VTP-50469, so that prediction is untested.

The comparison is within one compound, one dataset, one normalisation batch and one exposure, which is its strength - but the compound is MI-2-2, whose global effect in REH was small anyway (80 probes moving at least 1 log2, against 692 in RS4;11). A null in a line where the drug does little overall is weaker evidence than a null in a line where it does a lot, so this establishes that the signature TRACKS the fusion rather than that the fusion is mechanistically required. The decisive version - VTP-50469 in a KMT2A-wild-type line - does not exist in GSE127506, which contains no wild-type line at all. REH also differs from RS4;11 in fusion, lineage detail, medium and growth rate, so 'wild-type' is not an isogenic control.

• This result REFUTES the prediction this entry derived from the Ewing sarcoma paper. If menin itself were the shared node regulating ATF4, menin inhibition should lower ATF4 output in cells with no KMT2A fusion - and in the one wild-type line available it does not, for any of the 24 genes. Either the axis is fusion-dependent in leukaemia, or MI-2-2 was too weak to show it in REH, and E4's own caveat cannot separate those. The Ewing convergence remains a reason to look and is now one prediction worse off.
• A fusion-tracking signature is exactly what a fusion-selective drug response produces whether or not the ATF4 interpretation is right. Nothing here shows the ISR is the mechanism; it shows that the genes which respond to both compounds are the same genes that respond selectively in fusion lines, which is close to a restatement of the selectivity already established.

Abstract-level; the full text and figures were not read, so the extent of the MLL-rearranged B-ALL work - how many lines, which menin inhibitor, at what dose, and whether ATF4 protein or transcript was measured - is not established here. The paper's primary system IS Ewing sarcoma and the leukaemia result is presented as an extension, so the original entry was not wholly wrong in emphasis, only in the word 'entirely'. Nothing here establishes that the mechanism is the same in the two diseases.

• This paper cuts against this entry's novelty claim more than the correction admits. The entry's framing was that this graph had found an unremarked convergence; a published paper had already reported menin-dependent ATF4 loss in MLL-rearranged B-ALL, and had already identified the post-transcriptional component that cycle 58 then re-derived from expression data. The independent contribution is narrower than stated: the DOT1L comparison, which that paper does not make.
• Its menin->ATF4->serine axis also complicates this project's serine findings rather than simply supporting them. If menin drives ATF4 which drives serine synthesis, then the PHGDH/PSAT1/ASNS collapse this graph attributes to the fusion programme may be a downstream consequence of ATF4 loss - and the DOT1L arm reproducing the ISR collapse while only partly reproducing the PHGDH drop (-3.24 versus -1.52) is unexplained under either reading.