Oncofusion

start / graph / gene-hmga1

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.

HMGA1 (P17096) - an AT-hook architectural protein that antagonises histone H1 at scaffold-attachment regions, with a childhood B-ALL relapse correlation that is title-only evidence

also called HMGA1 · HMGIY · HMG-I/Y · HMG-I/HMG-Y · P17096 · HMGA1_HUMAN · High mobility group AT-hook protein 1 · High mobility group protein R

HMGA1 grips the narrow groove of A-and-T-rich DNA and loosens the packing that histone H1 imposes there. Those A-and-T-rich stretches are the same kind of sequence that anchors chromosome loops in the region where KMT2A breaks, and HMGA1 levels have been linked to relapse in childhood B-cell leukaemia - though that link is published only as a title.

The same thing, for a scientist

Third and last of the entries the six-member read-queue adjudication committed to. **All seven relevant papers were uncited in this graph.** **Identity.** P17096, `HMGA1_HUMAN`, **107 aa**, *High mobility group protein HMG-I/HMG-Y*; alternative names *High mobility group AT-hook protein 1* and *High mobility group protein R*; the only UniProt gene synonym is **HMGIY**. It binds preferentially the **minor groove of A+T-rich regions** in double-stranded DNA, and is implicated in nucleosome phasing, 3'-end processing of mRNA transcripts, and transcription of genes containing or close to A+T-rich regions. **THE SAR LINK IS REAL, IT IS THIRTY YEARS OLD, AND IT IS REACHABLE ONLY THROUGH THE ALIAS.** The adjudication that commissioned this entry justified it on the A+T-rich preference pointing at **scaffold-attachment regions**, which this graph's break model places in the KMT2A breakpoint cluster. Measured, that whole connection is **three records**. But one of them is exactly on point: Zhao et al., *EMBO J* 1993 — *"SAR-dependent mobilization of histone H1 by HMG-I/Y in vitro: HMG-I/Y is enriched in H1-depleted chromatin"*. Templates containing an upstream scaffold-associated region were **selectively repressed by H1** relative to non-SAR controls, *"due to the nucleation of H1 assembly into flanking DNA brought about by the numerous A-tracts"*, and HMG-I/Y antagonised that repression. **That paper is invisible to `HMGA1[tiab]`** — it predates the nomenclature — and surfaces only under `"HMG-I/Y"`. Third confirmation this cycle that the alias question has no default direction. **It has childhood B-ALL literature, and the two most relevant papers have no deposited abstracts.** `HMGA1 overexpression correlates with relapse in childhood B-lineage acute lymphoblastic leukemia` (Roy 2013, *Leukemia & Lymphoma*) and `STAT3 inhibitor has potent antitumor activity in B-lineage acute lymphoblastic leukemia cells overexpressing the high mobility group A1 (HMGA1)-STAT3 pathway` (Belton 2016, same journal) both return **empty AbstractText** from efetch. So the disease link and the intervention are established **at the level of their titles only** - no cohort size, no effect size, no controls. This entry says so rather than paraphrasing titles as findings. **Effectively never studied in this disease.** The alias union against (MLL OR KMT2A) returns **one** record, and reading it shows it is a 2026 study of chromoanagenesis in myelodysplastic syndromes where MLL appears incidentally. So the true count for KMT2A-rearranged biology is zero. **Dependency, with the caveat that matters.** Chronos **−0.8382** with **90.9% of 1,258 lines** below −0.5 - a harder pan-panel dependency than MEN1 (−0.3616), KMT2A (−0.2058) or DOT1L (−0.1474). Cycle 58 established that a pan-panel mean is the wrong statistic for comparing a broad dependency with a selective one, so this means HMGA1 is **broadly required**, not that it is more important than menin.

Facts and figures
uniprotP17096
uniprot entry nameHMGA1_HUMAN
gene symbolHMGA1
protein nameHigh mobility group protein HMG-I/HMG-Y
length AA107
domains noteUniProt P17096 annotates no DOMAIN features. The functionally defining elements are the AT-hook motifs, which are described in the literature and named in one of the alternative protein names but are not DOMAIN features on this record, so no coordinates are asserted. Empty array rather than an omitted field, because the viewer requires these to be arrays.
dna binding preferencebinds preferentially the MINOR GROOVE of A+T-rich regions in double-stranded DNA; implicated in nucleosome phasing, 3'-end processing of mRNA transcripts, and transcription regulation of genes containing or in close proximity to A+T-rich regions
localizationUniProt P17096: Nucleus [UNCODED]; Chromosome [UNCODED]. Both assignments carry no experimental evidence code, which is a real weakness for a protein whose entire relevance here is architectural.
subnuclear localizationNONE FINER THAN CHROMOSOME, and even that is uncoded. The functional literature places it in H1-depleted chromatin and at scaffold-attachment regions (PMID 8344261), which is a chromatin-state assignment rather than a compartment. NOT measured in any KMT2A-rearranged cell.
localization shown for fusionfalse
cell typeNo KMT2A-rearranged cell. The B-lineage ALL work is in childhood B-ALL patient material and B-lineage ALL cell lines; the SAR work is in vitro on T7 RNA polymerase templates; the T-cell leukaemia work is mouse and human T-ALL.
cell stateNothing measured across the cell cycle, under differentiation, or under a menin or DOT1L inhibitor. The one state-like statement is chromatin-compositional rather than cellular: HMG-I/Y is ENRICHED IN H1-DEPLETED CHROMATIN (PMID 8344261).
epigenetic mark namedNone. It writes, reads and erases no histone mark. Its chromatin effect is architectural - competing with histone H1 for A+T-rich DNA.
epigenetic readoutNeither writer, reader nor eraser. It is a linker-histone ANTAGONIST, which is a fourth class alongside the writer-degrader recorded in gene-asb7 this cycle.
shown for fusionfalse
absence classnever-measured
uniprot synonyms
[
  "HMGIY"
]
alternative names
[
  "High mobility group AT-hook protein 1",
  "High mobility group protein R"
]
domains
[]
the sar link measured not assumed
{
  "why_it_is_here": "the adjudication that commissioned this entry justified it on the A+T-rich preference pointing at scaffold-attachment regions in the KMT2A breakpoint cluster",
  "how_much_literature_that_is": {
    "(HMGA1 OR HMGIY OR \"HMG-I/Y\") AND (\"scaffold attachment\" OR SAR)": 3,
    "(HMGA1 OR HMGIY OR \"HMG-I/Y\") AND \"AT-hook\"": 149
  },
  "the_one_that_is_on_point": "PMID 8344261, Zhao et al., EMBO J 1993 - SAR-dependent mobilization of histone H1 by HMG-I/Y in vitro; HMG-I/Y is enriched in H1-depleted chromatin. SAR-containing T7 templates were selectively repressed by H1 relative to non-SAR controls, due to nucleation of H1 assembly into flanking DNA brought about by numerous A-tracts.",
  "and_it_is_invisible_to_the_approved_symbol": "the 1993 paper predates HMGA1 nomenclature and surfaces only under the \"HMG-I/Y\" token. Querying HMGA1 alone would have found the reasoning unsupported and dropped it.",
  "honest_status": "the SAR connection is REAL but thin, in vitro, and thirty years old. It is not a measurement at the KMT2A breakpoint cluster and nothing here places HMGA1 at that locus."
}
b lineage all literature
{
  "relapse_correlation": {
    "title": "HMGA1 overexpression correlates with relapse in childhood B-lineage acute lymphoblastic leukemia",
    "pmid": "23472968",
    "venue": "Leukemia & Lymphoma 2013",
    "WARNING": "efetch returns an EMPTY AbstractText for this record. The disease link is established at TITLE LEVEL ONLY - no cohort size, no effect size, no controls, no multivariate adjustment."
  },
  "the_intervention": {
    "title": "STAT3 inhibitor has potent antitumor activity in B-lineage acute lymphoblastic leukemia cells overexpressing the high mobility group A1 (HMGA1)-STAT3 pathway",
    "pmid": "26952843",
    "venue": "Leukemia & Lymphoma 2016",
    "WARNING": "also returns an EMPTY AbstractText. Title-level only, and it touches an intervention - see bite_back."
  },
  "other_haematological_records": [
    "34637548 De Martino 2022, Hematological Oncology - Critical role of the high mobility group A proteins in hematological malignancies (review)",
    "23311322 Di Cello 2013 - Inactivation of the Cdkn2a locus cooperates with HMGA1 to drive T-cell leukemogenesis",
    "17308084 Nagel 2007, Cancer Research - Activation of TLX3 and NKX2-5 in t(5;14)(q35;q32) T-cell ALL by remote 3'-BCL11B enhancers and coregulation by PU.1 and HMGA1"
  ],
  "the_translocation_adjacent_one": "17308084 places HMGA1 as a coregulator of oncogene activation downstream of a translocation-repositioned enhancer in T-ALL - the nearest thing in this literature to a fusion-context role, and it is a different lineage and a different mechanism class"
}
effectively never studied in this disease
{
  "(HMGA1 OR HMGIY OR \"HMG-I/Y\") AND (MLL OR KMT2A)": "1 record",
  "and_reading_it": "PMID 42320760, Wei 2026, Modern Pathology - chromoanagenesis in myelodysplastic syndromes, complex karyotype and TP53 disruption. MLL appears incidentally; it is not an HMGA1-at-KMT2A study.",
  "so": "the true count for KMT2A-rearranged biology is ZERO, established by reading the single hit rather than by counting it",
  "union x leukemia": 24,
  "union x acute lymphoblastic": 6
}
alias split measured
{
  "HMGA1[tiab]": 925,
  "HMGIY[tiab]": 239,
  "\"HMG-I/Y\"[tiab]": 176,
  "\"high mobility group A1\"[tiab]": 218,
  "union of the four ID sets": 1215,
  "HMGA1 intersect HMGIY": 16,
  "HMGA1 intersect \"HMG-I/Y\"": 9,
  "reading": "the approved symbol dominates (925 of 1215) but the overlap with either historic token is under 20 records, so the sets are near-disjoint. The load-bearing SAR paper sits in the HMG-I/Y set only.",
  "registry_significance": "third gene this cycle confirming that the alias question has no default direction - CARM1's approved symbol dominated and still missed a record, NUCKS1's tokens were near-disjoint, and here the approved symbol dominates while the single most relevant mechanistic paper is invisible to it."
}
the paralogue
{
  "HMGA2": "1,995 records against HMGA1's 925 - the better-studied member of the pair",
  "gene-hmga2": "does not exist in this graph",
  "note": "HMGA2 is the family member classically involved in fusion genes in benign mesenchymal tumours. That comparison is not made here because it was not measured; recorded as an open question."
}
depmap
{
  "mean_gene_effect": -0.8382,
  "frac_below_minus_0_5": 0.9094,
  "n_lines": 1258,
  "comparators_same_pull": {
    "MEN1": "-0.3616, 28.9%",
    "KMT2A": "-0.2058, 11.2%",
    "DOT1L": "-0.1474, 12.1%"
  },
  "reading": "broadly required, NOT selective. Cycle 58 established that a pan-panel mean is the wrong statistic for comparing a broad dependency with a selective one, so 'harder than MEN1' is a fact about the statistic and not about importance.",
  "depmap_release": null,
  "rule_83": "the Open Targets depMapEssentiality payload carries no release identifier; figure measured in cycle 58 with all tissue blocks flattened and a payload guard at the panel maximum"
}
How this could bite us back
  • Nothing here is a target nomination, and the dependency data argues actively against one: HMGA1 is required in 90.9% of 1,258 screened lines, which is the profile of a broadly essential gene rather than a selective vulnerability.
  • An intervention appears in this literature and must be read carefully. A STAT3 inhibitor is reported to have antitumour activity in B-lineage ALL cells overexpressing an HMGA1-STAT3 pathway (PMID 26952843) - but that record has NO deposited abstract, so its dose, model, controls and magnitude are all unknown to this entry. It must not be cited as evidence that STAT3 inhibition is indicated in KMT2A-rearranged disease, for which there is no record at all.
  • The mechanistic direction, if it transfers, points the wrong way for therapy. HMG-I/Y displaces histone H1 and opens A+T-rich chromatin; this graph's break model holds that accessible loop-anchored chromatin is what breaks. That makes HMGA1 a candidate fragility RISK factor at the KMT2A locus, and any intervention reasoning would have to contend with raising or lowering breakage in a child.
  • HMGA1 is an oncofetal protein with developmental roles. Nothing in this entry bounds toxicity, and its near-universal requirement across cell lines is the opposite of a therapeutic window.
Still open
  1. Does HMGA1 occupy the KMT2A breakpoint cluster's scaffold-attachment regions in any cell? The mechanism exists in vitro on synthetic templates (PMID 8344261) and has never been measured at this locus - and ENCODE HMGA1 ChIP tracks would answer it cheaply now that the bigWig route is open.
  2. What do the two abstract-less B-ALL papers actually report? Cohort size, effect size and whether the relapse correlation survived adjustment are all unavailable, and the relapse claim is currently title-only.
  3. Is the relapse correlation present in the KMT2A-rearranged subgroup specifically? Childhood B-lineage ALL is mostly not KMT2A-rearranged and no record is scoped to that subgroup.
  4. Why does `gene-hmga2` not exist? HMGA2 has 1,995 records against HMGA1's 925 and is the family member classically involved in fusion genes in benign mesenchymal tumours - which is the analogous-fusion-family dimension the PI asks for, unexamined.
  5. If HMGA1 opens A+T-rich chromatin by displacing H1, does it raise or lower KMT2A fragility? The direction implied by this graph's break model is that it would RAISE it, making HMGA1 a candidate risk modifier rather than a target.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    HMGA1 is UniProt P17096, HMGA1_HUMAN, 107 aa, High mobility group protein HMG-I/HMG-Y, with alternative names High mobility group AT-hook protein 1 and High mobility group protein R and the single gene synonym HMGIY. HMG-I/Y proteins bind preferentially to the minor groove of A+T-rich regions in double-stranded DNA; it is suggested that they function in nucleosome phasing and in the 3'-end processing of mRNA transcripts, and they are involved in the transcription regulation of genes containing, or in close proximity to, A+T-rich regions. Its subcellular location record lists Nucleus and Chromosome, both without an experimental evidence code.

    What this evidence cannot show

    Accession, entry name, length and synonyms were read from the JSON of an exact-gene query asserted to return one reviewed record. The FUNCTION comment is hedged in UniProt's own words ('it is suggested that'), and both localisation assignments are UNCODED - no ECO:0000269 - which is a genuine weakness given that this protein's entire relevance to this graph is architectural. No AT-hook coordinates exist as DOMAIN features, so the motif that defines the protein cannot be mapped against anything.

    Evidence that argues against this (1)
    • A 107-residue protein that binds a sequence composition rather than a motif, has no coded localisation and no annotated domains is close to the weakest possible object for a locus-specific claim. A+T-rich DNA is abundant genome-wide, so 'HMGA1 binds A+T-rich regions and the breakpoint cluster contains A+T-rich regions' is a statement about base composition and not evidence of a relationship.
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The scaffold-attachment-region connection is real, in vitro, and thirty years old. An experimental assay searching for proteins capable of antagonising histone H1-mediated general repression of transcription found that T7 RNA polymerase templates containing an upstream scaffold-associated region were highly selectively repressed by H1 relative to non-SAR control templates, due to the nucleation of H1 assembly into flanking DNA brought about by the numerous A-tracts, and that HMG-I/Y mobilises histone H1 in a SAR-dependent manner and is enriched in H1-depleted chromatin. Measured against the alias union, the whole scaffold-attachment literature for this gene is three records, while the AT-hook literature is 149.

    PMID 8344261
    What this evidence cannot show

    Abstract-level, in vitro, on T7 RNA polymerase templates - not chromatin in a cell, and certainly not the KMT2A locus. A 1993 result with no replication known to this project, and no DOI was returned by efetch so it is cited by PMID and title. 'Enriched in H1-depleted chromatin' is a compositional observation, not a localisation measurement. This establishes a mechanism HMGA1 CAN perform at a SAR in vitro; it establishes nothing about whether it does so at the KMT2A breakpoint cluster, where no measurement exists.

    Evidence that argues against this (2)
    • Three records is close to no literature, and a thirty-year-old in vitro result on synthetic templates is the kind of evidence this project has repeatedly downgraded elsewhere. The honest reading is that the SAR reasoning which commissioned this entry survives only in the weakest sense - it is not unsupported, but it is not a foundation either, and the adjudication presented it with more confidence than three in vitro records justify.
    • The direction is also not obviously the one this graph would want. HMG-I/Y ANTAGONISES H1-mediated repression at SARs, i.e. it opens A+T-rich chromatin. This graph's break model holds that accessible, transcribed, loop-anchored chromatin is what breaks - so if HMGA1 acted at the KMT2A SARs it would be predicted to increase fragility, which makes it a candidate risk factor rather than a therapeutic opportunity.
  3. E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    HMGA1 has childhood B-lineage ALL literature, and the two most relevant records carry no deposited abstract. PubMed efetch returns an empty AbstractText for both 'HMGA1 overexpression correlates with relapse in childhood B-lineage acute lymphoblastic leukemia' (Roy et al., Leukemia & Lymphoma 2013) and 'STAT3 inhibitor has potent antitumor activity in B-lineage acute lymphoblastic leukemia cells overexpressing the high mobility group A1 (HMGA1)-STAT3 pathway' (Belton et al., Leukemia & Lymphoma 2016). Related records that do exist include a review of the high mobility group A proteins in haematological malignancies, a report that Cdkn2a inactivation cooperates with HMGA1 to drive T-cell leukemogenesis, and a study in which PU.1 and HMGA1 coregulate TLX3 and NKX2-5 activation by remote 3'-BCL11B enhancers in t(5;14)(q35;q32) T-cell ALL. Against the alias union, (MLL OR KMT2A) returns one record, which on reading is a 2026 study of chromoanagenesis in myelodysplastic syndromes where MLL appears incidentally.

    PMID 23472968DOI
    What this evidence cannot show

    THE EVIDENCE IS THE TITLE. Both load-bearing records return an empty AbstractText from efetch, so no cohort size, effect size, control group, multivariate adjustment or subtype breakdown is available, and neither full text was retrieved. 'Childhood B-lineage ALL' does not mean KMT2A-rearranged: the KMT2A-rearranged subgroup is a small minority of childhood B-ALL and no record here is scoped to it. The T-ALL records are a different lineage. Confidence is weak by design and should stay weak until a full text is read.

    Evidence that argues against this (2)
    • HMGA1 is a broadly overexpressed oncofetal protein reported in many carcinomas and haematological malignancies, so 'overexpression correlates with relapse' is a common and weakly specific finding - the kind of association that appears for most proliferation-linked genes and rarely survives multivariate adjustment. With no abstract, there is no way to know whether this one did.
    • The pan-panel dependency cuts against a disease-specific reading. HMGA1 is required in 90.9% of 1,258 screened lines, so a correlation between its expression and relapse is at least as consistent with proliferation rate as with anything specific to B-lineage ALL biology, let alone to the KMT2A-rearranged subgroup.
Connected to