AQR (Aquarius) - a spliceosomal RNA helicase whose loss impairs MLL-rearranged leukaemia stem cells

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• AQR is core spliceosomal machinery. Any intervention risks the same therapeutic-window failure that has dogged spliceosome-directed agents, and the abstract offers no evidence of a leukaemia-selective window.
• The long-gene bias is a confound as much as a mechanism: a perturbation that preferentially damages long genes will damage many essential long genes, and 'leukaemia stem cells are more sensitive' may reflect proliferation rate rather than fusion biology.

1. ANSWERED 2026-09-19, unfavourably: AQR ranks 17,654/17,728 in the genome-wide scan, on the less-required side. The live question is now whether that is a floor effect on a strongly pan-essential gene, or a real absence of cell-autonomous selectivity - which an essentiality-matched null would separate.
2. Is the AQR requirement KMT2A-fusion-selective, or does it hold across leukaemia stem cells generally? The abstract does not compare.
3. Does ECAS survive an orthogonal separation from SEC on the same chromatin, or is it a state of SEC rather than a distinct assembly?
4. AQR loss preferentially affects long genes. KMT2A is 3,969 aa and among the longest genes in the relevant programme - is the effect on KMT2A-fusion leukaemia partly a gene-length effect on the fusion's own targets?

1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   Native chromatin proteomics in human cells identifies ECAS, a bipartite assembly linking an AFF4-ELL2-EAF1 elongation module to a U2 snRNP-associated intron binding complex, occupying promoter-proximal chromatin and molecularly distinct from canonical P-TEFb-containing SEC. Acute perturbation of AQR/XAB2-dependent ECAS functions impairs productive gene-body elongation without altering pause release, increases intron retention and preferentially affects long genes. In mouse haematopoietic cells Aqr is highly expressed and its perturbation impairs leukaemia initiation, maintenance and leukaemia stem cell function in MLL-rearranged leukaemia.

   A dual-module ECAS complex couples transcription elongation and RNA processing · Xiang Y, Wang H, Yin Z, et al. (Liang K, senior author) · Molecular Cell, online ahead of print 16 September 2026 2026
   PMID 42748909DOI
   ▸What this evidence cannot show

   Read as abstract only; online ahead of print. The leukaemia phenotype is reported in MOUSE haematopoietic cells and described as 'MLL-rearranged' without the abstract naming the fusion partner, the model, or whether any human KMT2A::AFF1 material was used. 'Profoundly impairs' is the authors' wording and is not quantified in the abstract. Acute perturbation of a spliceosomal RNA helicase has wide consequences, so a leukaemia-selective window is not established by these data.

   ▸Evidence that argues against this (1)

   • The claim that ECAS is distinct from SEC is a proteomics result about a chromatin-associated assembly, and assemblies sharing AFF4 and ELL2 are hard to resolve from one another by purification - this is the same class of inference that produced the 2004-2012 affinity-purification picture this project already treats as provisional. Until an orthogonal method separates ECAS from SEC on the same chromatin, 'distinct complex' and 'distinct state of one complex' are not discriminated. Xiang Y et al. 2026
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProtKB O60306 (AQR_HUMAN), RNA helicase aquarius, primary gene name AQR, 1,485 aa.

   UniProtKB O60306 (AQR_HUMAN), RNA helicase aquarius · UniProt Consortium 2026
   ▸What this evidence cannot show

   Resolved by the coordinator on 2026-09-19 via gene_exact:AQR with reviewed:true, and the PRIMARY gene name read back rather than the query term - this project has twice been caught by a synonym resolving to the wrong protein (TCF3 to TCF7L1) or a stale symbol (WDR61 to SKIC8).

   ▸Evidence that argues against this (1)

   • A UniProt header record gives the accession, the symbol and the length, and nothing about whether the protein does anything in this disease. For a 1,485-residue RNA helicase, the length is the least informative fact available. UniProt Consortium 2026
3. E_depmap_rankmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   DepMap does not support AQR as a KMT2A-rearranged-selective dependency, and points the other way. In the genome-wide scan of 17,728 genes, AQR ranks 17,654 (99.58th percentile) with a difference of +0.244 - i.e. KMT2A-rearranged lines need AQR LESS than other leukaemia lines, not more. Its ECAS partner XAB2 is more extreme still at rank 17,704 (+0.324). For contrast, in the same scan MEN1 ranks 1, KMT2A 11, MEIS1 84, AFF1 160, DOT1L 184 and USP22 226.

   Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
   ▸What this evidence cannot show

   Computed by the coordinator on 2026-09-19 over 11 KMT2A-rearranged and 43 other leukaemia lines. AQR and XAB2 are strongly pan-essential (mean gene effect -1.15 and -2.31), and a difference statistic is least informative where everything dies: a floor effect can compress or invert the sign. This project's own essentiality-confound analysis found essentiality explains only 0.4% of variance ACROSS the scan, but that average does not license reading the extreme tail, which is exactly where AQR sits. None of the eleven rearranged lines is infant-derived.

   ▸Evidence that argues against this (1)

   • The assay cannot see the phenotype the ECAS paper reports. That result is leukaemia INITIATION, MAINTENANCE and leukaemia stem cell function in mouse haematopoietic cells in vivo; DepMap measures proliferation of established human cell lines in culture, where self-renewal is not a readout and no stem compartment exists. A gene required for LSC function and dispensable for bulk proliferation would look exactly like this. So the scan weakens the case for AQR as a cell-autonomous KMT2A-r-selective vulnerability without touching the claim actually made. Xiang Y et al. 2026
4. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   AQR's speckle address is functional, not merely positional: RNAi depletion of the nuclear speckle proteins RNPS1, SRm160 or IBP160 - IBP160 being AQR under its former name - causes the MALAT-1 noncoding RNA to diffuse out of nuclear speckles into the nucleoplasm, so these proteins contribute to retaining MALAT-1 in the compartment. This is the observation behind UniProt O60306's uncoded note that AQR 'localizes to speckle-like regions of the nucleoplasm'.

   Identification of cis- and trans-acting factors involved in the localization of MALAT-1 noncoding RNA to nuclear speckles. · Miyagawa R, Tano K, Mizuno R, Nakamura Y, Ijiri K, Rakwal R, et al. · RNA 2012
   PMID 22355166DOI
   ▸What this evidence cannot show

   AQR appears here under the retired name IBP160 and is one of three proteins tested, so the result is about a shared speckle function rather than an AQR-specific address; the paper's subject is MALAT-1, not AQR. HeLa-class human cells, RNAi, no haematopoietic system and no fusion. Note the search hazard: 'speckle*' is 40% echocardiography and related imaging noise in PubMed (data/symbol-collisions.json), so this literature has to be reached with the quoted phrase.

   ▸Evidence that argues against this (1)

   • A speckle-only answer would put AQR in the wrong place for everything this graph uses it for. Native chromatin proteomics puts AQR in ECAS, an assembly that OCCUPIES PROMOTER-PROXIMAL CHROMATIN and joins a U2 snRNP-associated intron binding complex to an AFF4-ELL2-EAF1 elongation module - and is explicitly reported as molecularly DISTINCT from the canonical P-TEFb-containing SEC. Nuclear speckles are where splicing factors are stored and assembled; promoter-proximal chromatin is where AQR's measured phenotype occurs, since acute perturbation impairs gene-body elongation without altering pause release. The two addresses are not in conflict, but the speckle one is the less informative of the two for a fusion graph, and an entry recording only it would obscure that AQR sits beside AFF4. Xiang Y et al. 2026

• ← related toNELFE (RDBP) — the RNA-binding subunit of the pausing brake, and the alias that inflated a record count fifty-fold before it was caught

  two RNA-engaging promoter-proximal assemblies; proposed as an experiment, NOT a reported connection - the ECAS paper says AQR loss does not alter pause release

• ← part ofEAF1 — the interactor with an actual subnuclear phenotype: an MLL fusion evicts it from Cajal bodies, and the bodies themselves fall apart

  ECAS: an AFF4-ELL2-EAF1 elongation module joined to the AQR/XAB2 intron binding complex, distinct from P-TEFb-containing SEC

• ← supportsThe comparator arm was never screened for pathway-equivalent lesions — and it made us understate our own result

  AQR stays positive after cleaning (+0.261), so the negative reported against it is not an artefact of the contaminated comparator.

• part of →ELL2 — a SEC subunit that almost no cell needs, next to a paralogue almost every cell needs

  AQR and ELL2 are both subunits of ECAS, a promoter-proximal assembly reported as molecularly distinct from canonical P-TEFb-containing SEC (PMID 42748909).

• part of →AFF4 — the textbook SEC scaffold that KMT2A::AFF1 cells need LESS than other leukemias

  AQR sits in the same ECAS assembly as the AFF4-ELL2-EAF1 elongation module; if ECAS is distinct from SEC, AFF4 membership alone stops being evidence that SEC is the relevant complex.

• related to →Mechanism class 1 — chromatin and elongation hijack (KMT2A-r, NUP98-r)

  Acts on productive gene-body elongation WITHOUT altering pause release, unlike the pause-release lever this mechanism is built around, and preferentially affects long genes.

• related to →49 of 91 proteins in this graph have never had a drug-like molecule bound to them

  A new dependency whose tractability is untested: AQR is core spliceosomal machinery, the class where the therapeutic window usually fails.

• related to →Nuclear speckles — the compartment everyone assumes KMT2A::AFF1 uses, on evidence that skips AFF1 itself

  AQR/IBP160 is a nuclear speckle protein whose depletion releases MALAT-1 into the nucleoplasm (PMID 22355166); UniProt's 'speckle-like regions' note rests on this (Q87 2026-09-21)

• related to →Transcription factories and Pol II clusters - no KMT2A::AFF1 imaging, condensate evidence for SEC and MLL-ENL, and 3D-FISH that actively contradicts the factory story for t(4;11)

  AQR occupies promoter-proximal chromatin as part of ECAS, joining a U2 snRNP intron binding complex to an AFF4-ELL2-EAF1 elongation module and distinct from canonical SEC (PMID 42748909)

• ← co-occurs withZRANB2 - the only bait in any published proximity experiment that reaches AFF1, AFF4, ELL, MLLT1, CCNT1, KMT2A and DOT1L at once

  Both are RNA-processing proteins assigned to nuclear speckles; AQR additionally occupies promoter-proximal chromatin in an AFF4-ELL2 assembly, an address never looked for in ZRANB2