HOXA10 - named in 27 entries as part of the HOXA output and never written; a posterior HOX paralogue that is regulated with HOXA9 and tested far less often

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.

• The only intervention that reaches HOXA10 is the shared HOX-PBX interface, and an agent that hits that interface cannot distinguish HOXA10 from HOXA9, from the other posterior paralogues, or from HOX proteins in normal gut, limb, marrow and endometrium.
• HOXA10 is not a dependency in any human leukaemia line [E5], so there is no in vitro potency readout to optimise a HOXA10-directed agent against - the same drug-development problem PBX3 has.
• Paralogue swap is an untested escape route. If a cell can run the posterior HOXA programme on HOXA10 when HOXA9 is removed, then any HOXA9-selective strategy has a resistance mechanism sitting 40 kb away in the same TAD, under the same enhancers.
• This entry's own evidence base is thin and two of its five items are murine myeloid from one laboratory, one of them read only at title level. Treat any HOXA10-specific claim built from it as provisional.

1. What does HOXA10 loss do in a KMT2A-rearranged cell, with HOXA9 loss run alongside it in the same experiment? This is the missing control behind 27 entries' worth of HOXA10 mentions, and nothing found this cycle has done it.
2. Is HOXA10 or HOXA9 the higher-expressed posterior paralogue in infant KMT2A::AFF1 blasts? The graph asserts a HOXA programme without an internal ranking, and which paralogue dominates determines which one a paralogue-selective agent would need to hit.
3. Does the HOXA9/HOXA10 ratio shift under menin inhibition? If the cluster collapses uniformly, the paralogues are one unit; if it shifts, resistance could run through a paralogue swap, which is a resistance mechanism nobody in this field has looked for.
4. Do HOXA9 and HOXA10 occupy the same sites in a KMT2A-rearranged cell? Their hexapeptides differ in sequence between anterior and posterior HOX groups, and HOXA9 uses paralog-specific homeodomain residues for TALE binding - so co-regulation at the RNA level does not imply co-occupancy at the DNA level.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In human MLL-rearranged leukaemia, gene-expression profiling after RNA-interference suppression of HOXA9 shows co-down-regulation of a programme highly expressed in human MLL-AML and murine MLL leukaemia stem cells, including HOXA10, MEIS1, PBX3 and MEF2C; HOXA9 depletion across 17 human AML/ALL lines induces proliferation arrest and apoptosis specifically in the 7 MLL-rearranged lines (P=0.007); and mice transplanted with HOXA9-depleted t(4;11) SEMK2 cells showed significantly lower leukaemia burden.

   HOXA9 is required for survival in human MLL-rearranged acute leukemias · Faber J, Krivtsov AV, Stubbs MC, Wright R, Davis TN, van den Heuvel-Eibrink M, Zwaan CM, Kung AL, Armstrong SA · Blood 113:2375-2385 2009
   PMID 19056693DOI
   ▸What this evidence cannot show

   Abstract read this cycle; full text not retrieved. Human cell lines and primary AML samples; the only B-lineage KMT2A::AFF1 component is the SEMK2 xenograft arm. The HOXA10 result here is CO-REGULATION on HOXA9 knockdown, which cannot distinguish a functional requirement for HOXA10 from shared cluster-level regulation - and the HOXA cluster is known to be co-regulated by shared enhancers and a shared TAD, so co-movement is the null expectation, not evidence.

   ▸Evidence that argues against this (1)

   • In this graph's genome-wide DepMap scan HOXA9 itself ranks 3,649 of 17,728 with a difference of -0.0479, so the knockout data do not reproduce the survival requirement this knockdown reports for HOXA9, let alone establish one for HOXA10. analysis run by this project 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   HoxA10 contains an activation domain, identified in this work, that is necessary for transcription of the gene encoding beta-3 integrin during myeloid differentiation.

   Identification of a HoxA10 activation domain necessary for transcription of the gene encoding beta3 integrin during myeloid differentiation · Bei L, Lu Y, Bellis SL, Zhou W, Horvath E, Eklund EA · Journal of Biological Chemistry 282:16846-16859 2007
   PMID 17439948DOI
   ▸What this evidence cannot show

   Title and bibliographic record fetched from PubMed on 2026-09-20; the abstract was not read in full and the residue boundaries of the activation domain are therefore NOT asserted anywhere in this entry - a domain with unread coordinates is left out under the residue-integer rule rather than guessed. MURINE MYELOID differentiation, not KMT2A-rearranged leukaemia and not B-lineage. Cited for the existence of a mapped HOXA10 activation function, nothing more.

   ▸Evidence that argues against this (1)

   • A transcriptional activation function demonstrated during normal myeloid differentiation is evidence about HOXA10's normal role and may be irrelevant to, or opposed by, what a differentiation-blocked KMT2A-rearranged blast does with the same protein. Faber J et al. 2009
3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   In myeloid cells, HoxA10 activates CDX4 transcription and Cdx4 activates HOXA10 transcription, forming a reciprocal regulatory loop.

   HoxA10 activates CDX4 transcription and Cdx4 activates HOXA10 transcription in myeloid cells · Bei L, Huang W, Wang H, Shah C, Horvath E, Eklund E · Journal of Biological Chemistry 286:19047-19064 2011
   PMID 21471217DOI
   ▸What this evidence cannot show

   Bibliographic record fetched from PubMed on 2026-09-20; the claim is the paper's TITLE, read as a title and not from the abstract or body. Graded moderate for that reason. MYELOID cells, same laboratory as E2, so E2 and E3 are not independent. The loop's relevance to KMT2A-rearranged leukaemia is an inference from CDX genes sitting upstream of the HOXA cluster in haematopoiesis, and that inference is made here and is not made by the paper.

   ▸Evidence that argues against this (1)

   • A claim taken from a title is the weakest citation form this graph permits, and nothing in this entry should be built on it; it is recorded so a later reader knows the loop exists to be checked, not so it can be quoted. Bei L et al. 2011
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review

   Because of functional redundancy amongst HOX proteins, knockdown or knockout studies of individual HOX genes often do not produce an altered phenotype, which is the stated rationale for targeting instead the HOX-PBX interaction through the conserved hexapeptide region - the approach taken by the peptides HXR9 and HTL001.

   Using the Conserved Hexapeptide in HOX Proteins as an Antagonist of HOX/PBX Interactions · Morgan R, Smith C, Pandha H · Methods in Molecular Biology 2889:139-153 2025
   PMID 39745610DOI
   ▸What this evidence cannot show

   A methods review by the group that developed the peptides, so it is both authoritative on the approach and interested in it. Cited only for the redundancy rationale and for the fact that the interface is paralogue-shared, both of which are uncontested. It is a review and is used here as a map, per the standing preference for primary sources.

   ▸Evidence that argues against this (1)

   • The redundancy premise is contradicted for at least one paralogue in this disease: HOXA9 knockdown alone does produce a phenotype, specific to MLL-rearranged human cells, so HOX redundancy is not uniform and an argument built on it should not be applied to HOXA9. Faber J et al. 2009
5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   Fetched live on 2026-09-20, DepMap Chronos gene effect for HOXA10 has mean -0.0628 across 1,255 cell lines and -0.0997 across 57 leukaemia-annotated lines, with 0.4% of lines at or below -0.5 and 0.08% at or below -1.0. HOXA9 in the same query is -0.1727 and -0.2230. Neither posterior HOXA paralogue is a fitness requirement of any human leukaemia line in the panel.

   DepMap CRISPR (Chronos) gene effect for HOXA10 (ENSG00000253293) and HOXA9 (ENSG00000078399), retrieved via the Open Targets Platform GraphQL API · Broad Institute DepMap; Open Targets Platform; query run by this project 2026-09-20 · Database query 2026
   ▸What this evidence cannot show

   A bounded negative with a date, a query and a tool, and bounded only on what a CRISPR fitness screen can see. Paralogue redundancy in the HOXA cluster is the explicit expectation [E4], so a single-gene knockout of HOXA10 is buffered by construction and this number cannot distinguish dispensable from redundant. Immortalised lines, none infant-derived, roughly 21 days, no self-renewal or differentiation readout. A database fact opposable on scope.

   ▸Evidence that argues against this (1)

   • HOXA10 knockout is buffered by at least HOXA9 and probably by the rest of the posterior cluster; the only interpretable experiment would be a combinatorial HOXA9/HOXA10 knockout, which does not exist in DepMap and which no source read this cycle has done in a KMT2A-rearranged cell. Morgan R et al. 2025
6. E_q77_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   HOXA10's entire subcellular record in UniProt P31260 (fetched 2026-09-21) is the single word 'Nucleus' with an EMPTY evidence array - no ECO:0000269, no ECO:0000305, no ECO:0000250, no note and no isoform-scoped statement. It is an unsourced assertion, not an annotation with a provenance.

   UniProtKB P31260 (HXA10_HUMAN) - SUBCELLULAR LOCATION comment block with evidence codes · UniProt Consortium 2026
   ▸What this evidence cannot show

   A measurement of the database, reported as such, fetched live 2026-09-21 by reading the comment array with evidence attached rather than the flattened field. Absence of an evidence code is not evidence that HOXA10 is not nuclear - it manifestly binds DNA at mapped sites in HLF, DKK1 and GATA1 - it is a statement that the compartment has not been curated from an experiment. Of the sixteen proteins in this tranche, HOXA10 and STAG2 are the only two whose subcellular block carries no evidence code of any kind.

   ▸Evidence that argues against this (1)

   • Functional evidence makes the nuclear call overwhelmingly likely and that is exactly why nobody measured it, which is the failure mode worth naming. The MEIS1 precedent in this graph shows the cost: MEIS1 is a TALE cofactor of HOXA10, its nuclear call was also a prediction, and when someone looked, one isoform turned out to be CYTOPLASMIC. An obvious answer that nobody has checked is not the same as a checked answer. Chen S et al. 2024
7. E_q77_loc2weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

   The only experimental localisation result found for HOXA10 in any system is in ORAL SQUAMOUS CELL CARCINOMA: importin-7 (IPO7) interacts with HOXA10 by GST pull-down and co-IP and promotes its NUCLEAR TRANSLOCATION, with HOXA10 then binding the EIF2AK3 promoter by ChIP and luciferase reporter, driving PERK/ATF4 endoplasmic-reticulum-stress signalling and protective autophagy.

   Importin 7 enhances protective autophagy induced by nuclear translocation of homeobox A10 in oral squamous cell carcinoma. · Chen S, Pan J · Cell Mol Biol (Noisy-le-grand) 2024
   PMID 38650127DOI
   ▸What this evidence cannot show

   Confidence WEAK and deliberately so. The system is oral squamous cell carcinoma, three tissues and one germ layer away from a blast; the journal is low-visibility; the localisation result is one panel inside a paper about autophagy and ER stress, i.e. exactly the supporting-panel case brief rule 80 describes; and no quantification of the nuclear fraction is reported. It is recorded because it is the ONLY experimental handle on HOXA10 nuclear entry that exists, and because the mechanism it names - an importin-beta-family carrier rather than a classical importin-alpha NLS route - is checkable in a blast in a week.

   ▸Evidence that argues against this (1)

   • A single low-visibility paper in an unrelated tumour type is not a basis for a localisation claim, and this entry does not make one. The honest reading is that HOXA10's compartment is unmeasured in every system this project cares about; the IPO7 result is logged as a lead and as a bounded rejection marker so that a future cycle does not re-derive the same empty search. Note also that IPO7 knockdown in that paper reduces proliferation, migration, invasion and tumour growth on its own, so the HOXA10 translocation is only one of several IPO7-dependent effects being reported. UniProt Consortium 2026
8. E_q77_ct1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In MICE, retroviral over-expression of HOXA10 in bone-marrow cells profoundly perturbs MYELOID AND B-LYMPHOID differentiation and leads to acute myeloid leukaemia: reconstituted marrow is virtually devoid of transduced-cell-derived unilineage macrophage and PRE-B-lymphoid progenitors, carries in high frequency a unique megakaryocyte-colony-forming progenitor, and a significant proportion of animals develop transplantable AML at 19-50 weeks latency. The matched control is the decisive part: over-expressing the 3-prime gene HOXB4 in the same system selectively EXPANDS primitive cells WITHOUT altering differentiation.

   Overexpression of HOXA10 in murine hematopoietic cells perturbs both myeloid and lymphoid differentiation and leads to acute myeloid leukemia. · Thorsteinsdottir U, Sauvageau G, Hough MR, Dragowska W, Lansdorp PM, Lawrence HJ, Largman C, Humphries RK · Mol Cell Biol 1997
   PMID 8972230DOI
   ▸What this evidence cannot show

   Murine, retroviral, gain-of-function, 1997 progenitor gating, and no fusion present. It grades what excess HOXA10 can do, not what endogenous or fusion-driven HOXA10 does. The long latency (19-50 weeks) and incomplete penetrance say a second event is required, and the paper does not identify it. Brief rule 73: the system is murine myeloid-onset leukaemia, not human infant B-lineage KMT2A::AFF1, and the B-lineage observation here is a progenitor DEPLETION rather than a transformation.

   ▸Evidence that argues against this (1)

   • The titratable transgenic complicates every directional reading of this experiment. There, HIGH HOXA10 had no effect on HSC proliferation while INTERMEDIATE HOXA10 gave a 15-fold repopulating increase, and high HOXA10 blocked erythroid and megakaryocyte development - the opposite sign to the megakaryocyte-progenitor accumulation reported here. Retroviral over-expression does not report where on that dose curve it lands, so the two studies may simply be sampling different points, and this graph should not treat them as two independent confirmations of one phenotype. Magnusson M et al. 2007
9. E_q77_ct2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In a MURINE doxycycline-titratable HOXA10 transgenic, the dose-response is NON-MONOTONIC: intermediate HOXA10 induced a 15-fold increase in HSC repopulating capacity after 13 days of in vitro culture, high HOXA10 had NO effect on HSC proliferation, and high HOXA10 BLOCKED erythroid and megakaryocyte development. Downstream, HLF, DKK1, GFI1 and GATA1 change with HOXA10 dose, HOXA10 binding sites are present in HLF, DKK1 and GATA1, and DKK1 and GFI1 are transcriptionally activated by HOXA10.

   HOXA10 is a critical regulator for hematopoietic stem cells and erythroid/megakaryocyte development. · Magnusson M, Brun AC, Miyake N, Larsson J, Ehinger M, Bjornsson JM, Wutz A, Sigvardsson M, Karlsson S · Blood 2007
   PMID 17234739DOI
   ▸What this evidence cannot show

   Murine and gain-of-function again, with the headline repopulation result measured after 13 days of IN VITRO culture rather than natively. Doxycycline titration controls dose better than a retrovirus does but the absolute levels are not related to any physiological or leukaemic reference, so 'intermediate' and 'high' are internal to the system. Target binding is ChIP-level occupancy plus reporter activation, not a locus-resolved requirement. No fusion, no human material.

   ▸Evidence that argues against this (1)

   • The DepMap arm already in this entry is the third instrument and it sees nothing: mean Chronos -0.0628 over 1,255 lines, -0.0997 over 57 leukaemia lines, 0.4% of lines below -0.5. A gene that can 15-fold a repopulation assay when raised is not detectably required when removed from established lines - which is consistent with the paralogue redundancy this entry already argues, and equally consistent with the effect being an over-expression artefact. Per brief rule 84 the selectivity RANK (1,310 of 17,728) is quoted rather than membership of any half of the distribution. Oncofusion knowledge graph (internal analysis) 2026

• ← co-occurs withHOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it

  adjacent posterior HOXA paralogues, co-regulated in the same programme; HOXA10 falls when HOXA9 is suppressed

• co-occurs with →MEIS1 — rate-limiting for MLL leukemia stem cells, a surviving co-dependency, and embryonic-lethal when removed

  co-down-regulated with MEIS1, PBX3 and MEF2C on HOXA9 suppression in human MLL-rearranged leukaemia

• part of →CTCF/cohesin boundaries and the HOXA topological domain — the 3D scaffold the fusion program runs on

  sits inside the HOXA cluster whose co-regulation is the null explanation for HOXA10 moving with HOXA9

• ← related toHow HOXA9 binds DNA: a hexapeptide into a PBX pocket, a 1.9 A structure that is mouse HoxA9 plus human Pbx1 homeodomains, and a second route that makes the motif redundant

  the hexapeptide is paralogue-shared, which is why an interface agent cannot distinguish HOXA9 from HOXA10

• ← refutesLive DepMap on fourteen genes: CDK6's pan-essentiality is a lineage artefact of pooling, and not one gene of the homeodomain cofactor code is a dependency anywhere

  bounded negative: HOXA10 knockout is fitness-neutral; paralogue-buffered by construction

• treated by →The HOX-PBX interface, and its one peptide: HXR9/HTL001 has been tried in eight tumour types and in KMT2A-rearranged leukaemia exactly once

  hit incidentally: the hexapeptide is paralogue-shared, so the agent cannot spare HOXA10

• ← related toSUPERSEDED IN PART, CYCLE 56 - the MI-2-2 response is ~9x fusion-selective (that stands), but 'does not touch HOXA or MEIS1' was a fact about a weak compound: VTP-50469 collapses MEIS1 -5.25 log2 in the same line

  HOXA10 unmoved (+0.06, +0.45) and unmoved in combination (+0.05, +0.49)

• ← regulatesThe experiment that refutes cycle 55's headline was already public: VTP-50469 collapses MEIS1 32-fold in the same line where MI-2-2 left it untouched - and HOXA9 was never actually measured by either dataset

  HOXA10 -0.88 at day 2 and -1.46 at day 7 under VTP-50469 in RS4;11

• ← regulatesNUTM1 (NUT) - a testis-restricted, fully disordered protein that writes no mark and instead switches ON a writer, and the 3' partner of the KMT2A-wild-type infant B-ALL with the opposite prognosis

  HOXA10 up with HOXA9 in the ACIN1-NUTM1 BLUEPRINT case and in the ACIN1::NUTM1-inducible model

• ← regulatesACIN1::NUTM1 - the commonest NUTM1 fusion in infants, cryptic to cytogenetics, and the only one with a human cell line

  raised with HOXA9 in both the BLUEPRINT patient and the inducible model

• ← related toUnder EPZ5676 the canonical KMT2A-fusion targets look like they lose H3K79me2 far more than the genome (p=0.017 and p=0.0004), and against a baseline-matched null the entire effect vanishes (p=0.34 and p=0.22) - eight of the ten are simply top-decile genes

  baseline decile 4 in RS4;11, retains 5.78% - one of only two canonical targets not in the top deciles

• ← related toIRX1 — Iroquois homeobox 1, the gene that names the HOXA-negative half of infant t(4;11) ALL

  the HOXA9/HOXA10 pair is the axis partner of IRX1

• ← supportsHOXA is occupied as a 107 kb cluster; IRX1 is occupied as one gene, with both IRX clusters cold

  HOXA10 rises with HOXA9 as part of the coordinated cluster

• ← related toHOXA occupancy is the partner-invariant: occupied in 17 of 18 non-AFF1 KMT2A-r samples, 2.94x to 29.1x
• ← related toThe HOXA/IRX1 mutual exclusivity is reproduced at the transcript level in 48 KMT2A::AFF1 infant ALL cases (Fisher two-sided p < 1e-6) and is ABSENT in KMT2A::MLLT1 and in KMT2A-germline controls

  HOXA10 is the readout because GPL570 has no single-gene HOXA9 probe

• ← related toThree of seventeen KMT2A-GERMLINE infant ALL cases are IRX1-high, one at 10.37 - so a high IRX1 in an infant leukaemia is not by itself evidence of anything fusion-related

  HOXA10 is ON in 7 of 17 fusion-free infant ALL cases, so HOXA-high is not diagnostic of a KMT2A rearrangement

• ← co-occurs withHOXA7 - the one HOXA paralogue besides HOXA9 with a published loss-of-function requirement for MLL-fusion transformation, and the requirement was shown in vitro only

  both reduced together with HOXA9 when MLL-AF4 is depleted

• related to →HOXA9 has its own nuclear bodies, and in monocytic AML they move into the eIF4E body - a subnuclear switch measured in patient marrow and never genotyped

  the paralogue comparison that has never been run: HOXA9 has a subnuclear entry, HOXA10 has no subnuclear measurement at all

• related to →MEIS1's 'nucleus' is a profile prediction, not a measurement - and a homeodomain-less isoform is cytoplasmic and does the opposite job

  same unsourced-nucleus shape, and MEIS1 is HOXA10's own TALE cofactor - where someone looked, an isoform turned out cytoplasmic

• expressed in →Bone marrow haematopoietic stem cell — the postnatal comparator, and why the same fusion behaves differently here

  murine titratable transgenic: intermediate HOXA10 gives a 15-fold HSC repopulating increase, high HOXA10 gives none

• expressed in →Megakaryocyte-erythroid progenitor (MEP) — the branch KMT2A fusions almost never take, which is why it was a usable control

  high HOXA10 blocks erythroid and megakaryocyte development, while retroviral over-expression accumulates a megakaryocyte-colony-forming progenitor

• related to →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia

  lineage-selective block in a fusion-free mouse: pre-B and macrophage progenitors depleted, HOXB4 in the same system expands without blocking

• ← regulatesSET::NUP214 (TAF-Ibeta-NUP214, SET-CAN), cryptic del(9)(q34.11q34.13) - a HOXA-driving fusion of a histone chaperone to a nucleoporin, with no KMT2A and no menin-binding motif

  binds the HOXA10 promoter by ChIP in LOUCY; cooperates with MLL on the HOXA10 reporter