GATA2 outlier activation in KMT2A::USP2 B-ALL - the only transcriptional output ever measured for this fusion
also called KMT2A-USP2 GATA2 activation · GATA2 trans-activation by KMT2A-USP2
In a large set of childhood leukaemias, two patients had wildly abnormal levels of a stem-cell gene called GATA2, and both turned out to carry the same rare KMT2A::USP2 fusion - the only two in the whole cohort. No other KMT2A fusion did this, which is the single piece of evidence that this fusion switches on a distinctive gene.
The same thing, for a scientist
Wang et al. screened RNA-seq from 629 B-ALL patients in St Jude Cloud for aberrant GATA2 expression and found 13 outlier cases. Two of the 13 carried KMT2A::USP2, and those two were the only KMT2A::USP2 cases in the entire cohort - a 2/2 hit rate with no other KMT2A-rearranged patient showing GATA2 outlier expression despite high wild-type USP2 expression in some of them, which the authors take as evidence that the chimeric protein as a whole, rather than USP2 dosage, drives GATA2 transcription [E1]. Downstream, GATA2 activation associated with a myeloid-like differential expression signature and with three tumour-associated targets: BAG3, EPOR and KLF1 [E1]. Separately, a distinct mechanism of GATA2 activation was documented in the same study - cis activation by a somatic focal deletion disrupting a topologically-associating-domain boundary near GATA2 in one patient [E1]. **This is the entire published transcriptional characterisation of KMT2A::USP2.** There is no ChIP-seq, no CUT&RUN, no histone-mark profile and no HOXA/MEIS1 measurement for this fusion, so a two-patient expression correlation is doing all the mechanistic work.
What was actually observed, and what it is not1 deeper
Observed: in 629 B-ALL transcriptomes, 13 GATA2 expression outliers; 2 of them KMT2A::USP2-positive; those 2 were the cohort's only KMT2A::USP2 cases; other KMT2A-rearranged patients, including some with high wild-type USP2 expression, did not show GATA2 outlier expression 1.
Not observed: any binding of the fusion protein at the GATA2 locus; any chromatin mark; any functional requirement for GATA2 in a KMT2A::USP2 cell; any replication in a second cohort. The inference from 'these two patients have both features' to 'the fusion activates GATA2' is a correlation with n = 2 and no mechanism attached.
What makes it worth an entry rather than a footnote is the specificity structure: 2/2 of the genotype, 0 of the other KMT2A-r patients, and the wild-type-USP2-high patients acting as an internal control that separates the fusion from USP2 dosage. That is about as much as two patients can carry.
Two routes to the same output, in the same paper
The same study found GATA2 activated in cis in a different patient, by a somatic focal deletion that disrupted the boundaries of a topologically associating domain near GATA2 1. So aberrant GATA2 expression in B-ALL is convergent: reachable by enhancer hijacking through TAD disruption and, apparently, by a KMT2A fusion acting in trans. Convergence on one output from two unrelated lesion types is the kind of pattern that usually means the output is selected for, which raises the stakes on testing whether GATA2 is actually required.
Why GATA2 is a plausible output for this fusion specifically
GATA2 is restricted in normal haematopoiesis to HSCs, progenitors, early erythroid and megakaryocytic cells; it is not a B-lineage transcription factor 1. Its aberrant presence in a B-ALL blast therefore imposes a partly myeloid transcriptional programme, and the paper reports exactly that - differentially expressed genes on GATA2 activation were significantly myeloid-like 1.
That aligns with an otherwise unexplained clinical feature of this genotype: 4 of 17 KMT2A::USP2 patients in the founding series were B/myeloid mixed-phenotype acute leukaemia 2, a far higher MPAL fraction than KMT2A-rearranged B-ALL generally. A fusion that installs a myeloid transcription factor into a B-progenitor is a candidate explanation for a mixed phenotype. This connection is drawn here for the first time and is not made in either source paper; it is a hypothesis-grade inference and is flagged as such.
What would have to be true
For this mechanism to be real rather than a coincidence of two patients:
- More KMT2A::USP2 cases must show GATA2 outlier expression. The claim is falsifiable with 5-10 additional transcriptomes, which exist in the DCAL and St Jude collections.
- The fusion protein must be detectable at the GATA2 locus or at its enhancers. Retention of CXXC in the fusion gives it CpG-island targeting, so this is checkable.
- GATA2 knockdown in a KMT2A::USP2 cell must have a phenotype - which cannot currently be done, because no such cell exists.
Step 3 is the binding constraint on this entire entry and on the fusion's entry as a whole.
| cohort | 629 B-ALL patients with definitive subtype information, RNA-seq from St Jude Cloud |
|---|---|
| gata2 outliers | 13 |
| kmt2a usp2 among outliers | 2 |
| kmt2a usp2 in cohort | 2 |
| hit rate | 2/2 |
| internal control | other KMT2A-rearranged patients, including some with high wild-type USP2 expression, did not show GATA2 outlier expression |
| signature | differentially expressed genes on GATA2 activation were significantly myeloid-like |
| second route | cis activation by a somatic focal deletion disrupting TAD boundaries near GATA2 in one (different) patient |
| novel inference flagged | The link from GATA2-driven myeloid-like signature to the 4/17 B/myeloid MPAL fraction is drawn in this entry and appears in neither source paper. Treat as hypothesis-grade. |
downstream targets reported
[ "BAG3", "EPOR", "KLF1" ]
what is missing
[ "no ChIP-seq or CUT&RUN placing the fusion at GATA2", "no histone-mark profile of any kind for this fusion", "no HOXA9/MEIS1 measurement for this fusion", "no GATA2 knockdown, because no KMT2A::USP2 cell line exists", "no replication cohort" ]
- If GATA2 rather than HOXA9 is the fusion's output, then menin inhibition - whose proximal effect is loss of the HOXA/MEIS1 programme - may not touch this genotype even though the fusion retains the menin-binding motifs. That is a concrete route by which an approved drug could fail in patients selected for it by rearrangement status alone.
- BAG3 is anti-apoptotic and EPOR is a growth-factor receptor. If GATA2 activation is real in this genotype, both are candidate resistance nodes against apoptosis-directed therapy such as venetoclax, which is one of the few agents reported in a KMT2A::USP2 patient.
- Two patients is an unstable foundation. Any therapeutic reasoning built on this entry should be treated as a reason to sequence more patients, not as a reason to choose a drug.
- Do additional KMT2A::USP2 transcriptomes show GATA2 outlier expression? Five to ten more cases would move this from anecdote to finding, and those transcriptomes plausibly already exist in DCAL and St Jude collections.
- Is the fusion protein present at the GATA2 locus? The fusion retains CXXC, so CpG-island targeting is available to it and the experiment is a standard CUT&RUN once a model exists.
- Were the two GATA2-activated patients the MPAL cases, the B-ALL cases, or a mix? Nobody has cross-tabulated GATA2 status against immunophenotype in this genotype, and doing so is a reanalysis, not a new experiment.
- Does KMT2A::USP2 induce HOXA9 and MEIS1 at all? The canonical KMT2A-r programme has never been measured in this fusion, so the graph cannot say whether it belongs to that programme or sits outside it.
- E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
Of 629 B-ALL transcriptomes from St Jude Cloud, 13 showed aberrant GATA2 activation; two of these carried KMT2A::USP2 and were the only two KMT2A::USP2 cases in the cohort, while other KMT2A-rearranged patients with high wild-type USP2 expression showed no GATA2 outlier expression; GATA2 activation associated with a myeloid-like signature and with BAG3, EPOR and KLF1, and was separately produced in cis in one patient by a focal deletion disrupting a TAD boundary near GATA2.
Aberrant GATA2 Activation in Pediatric B-Cell Acute Lymphoblastic Leukemia. · Wang H, Cui B, Sun H, Zhang F, Rao J, Wang R, Zhao S, Shen S, Liu Y · Frontiers in Pediatrics 2022What this evidence cannot show
Two patients carry the entire KMT2A::USP2 claim. Retrospective reanalysis of a public expression resource with no functional validation, no ChIP, no knockdown and no second cohort. 'Outlier expression' is a statistical definition applied within one dataset and is sensitive to how the threshold is set. The 2/2 hit rate is arithmetically impressive and epidemiologically fragile: with two cases, one misclassified fusion call would halve it. Abstract and the relevant full-text passages read in this session.
Evidence that argues against this (1)
- Wild-type USP2 is expressed at high level in some other KMT2A-rearranged patients in the same cohort without GATA2 activation, which the authors use as an internal control - but this cuts both ways, since it also shows GATA2 outlier status is not a general consequence of USP2 protein abundance and therefore cannot be explained by the deubiquitinase activity the rest of this beat's mechanism hypothesis rests on. The GATA2 route and the UCH-rescue route are not obviously the same story. Meyer C et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Four of 17 KMT2A::USP2 patients in the founding series had B/myeloid mixed-phenotype acute leukaemia, a mixed-lineage fraction that a myeloid transcription factor installed in a B progenitor could plausibly explain.
Human MLL/KMT2A gene exhibits a second breakpoint cluster region for recurrent MLL-USP2 fusions. · Meyer C, Lopes BA, Caye-Eude A, Cavé H, Arfeuille C, et al., Marschalek R · Leukemia 2019What this evidence cannot show
The MPAL fraction is 4 of 17 and no comparison against a matched KMT2A-r B-ALL group is offered, so 'higher than usual' is an impression rather than a tested difference. The link to GATA2 is made in this entry, not in either source, and no patient has been shown to have both MPAL phenotype and GATA2 activation. Full text read (PMC6756029).
Evidence that argues against this (1)
- Mixed-phenotype presentation is a recognised feature of KMT2A-rearranged infant leukaemia generally, independent of partner, so a 4-of-17 MPAL fraction may be entirely ordinary for the age group and require no fusion-specific explanation at all. The graph holds disease-mpal for that background. Blackburn PR et al. 2020
the only published transcriptional output for this fusion: GATA2 outlier expression in 2 of 2 cases in a 629-patient cohort
a myeloid-like signature installed in a B progenitor is a candidate explanation for the 4/17 B/myeloid MPAL fraction - inference drawn in this graph, not in either source
gene-level entry for the factor that mechanism entry is about; source re-fetched rather than copied