Oncofusion

start / graph / gene-supt4h1

Proteinweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q5-interactome4

SUPT4H1 (SPT4) — zero papers linking it to KMT2A, and the 138th most KMT2A-selective gene out of 17,728

also called SUPT4H1 · SPT4 · SUPT4H · SPT4H · Transcription elongation factor SPT4

SPT4 is the small half of DSIF, the two-protein clamp on RNA polymerase. Nobody has ever published anything connecting it to KMT2A leukaemia — yet in an unbiased scan of every human gene, KMT2A-rearranged leukaemia cells need it more than other leukaemia cells do, and it ranks in the top 1%.

The same thing, for a scientist

SUPT4H1 (P63272, **117 aa** — one of the smallest proteins in this graph) is the small DSIF subunit. Its entire annotated architecture is a **C4-type zinc finger at aa 16-36** and a **SUPT5H-interaction region at aa 2-40**, which overlap: SPT4 is essentially a zinc finger that grips SPT5 [E1]. Functionally it is defined by the cryo-EM elongation complexes, where DSIF stabilises the paused state and is retained, with its RNA clamp opened by SPT6, in the activated state [E2][E3]. **The literature contains nothing on SPT4 in KMT2A-rearranged leukaemia: `(MLL OR KMT2A) AND (SUPT4H1 OR SPT4 OR SUPT4H)` returns 0 records** [E4]. Against that, the project's genome-wide dependency scan places SUPT4H1 at **rank 138 of 17,728 (0.778th percentile)** for KMT2A-rearranged selectivity, with a difference of **-0.2642** (rearranged -1.1440 against other-leukaemia -0.8798) and a pan-essentiality of **-0.6692** that is materially shallower than its partner SPT5's -1.5066 [E5]. Those two facts together — a top-1% selectivity rank and a literature of zero — are why this entry exists.

What it looks like
Transcription elongation factor SPT4117 aa
SUPT5H (SPT5) interaction regionSUPT5H (SPT5) interaction region · residues 2–40ZnF (C4-type zinc finger)ZnF (C4-type zinc finger) · residues 16–360100
loading AF-P63272-F1…
AlphaFold AF-P63272-F1 ↗
colourshape

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.

very high (90+)confident (70–90)low (50–70)very low (<50)
I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
The number, and how much of it to believe

SUPT4H1's selectivity difference of -0.264 puts it above DOT1L (-0.231, rank 184), above AFF1 (-0.247, rank 160) and above ELL (-0.253, rank 153) in the same scan, and below MEN1 (-0.866, rank 1) and KMT2A (-0.541, rank 11) 5. For a gene with no KMT2A literature at all, that is a striking placement.

Three reasons to be careful before it gets treated as a finding:

  1. Its partner does not agree. SPT5 scores -0.122 at rank 767 5. If DSIF were selectively required, both subunits should move together, and they do not. A complex-level discordance of that size is usually a sign that one number is noise.
  2. Floor effects. SUPT4H1's rearranged mean is -1.14, deep enough that compression is possible, though less severe than SPT5's -1.58.
  3. Eleven lines. The KMT2A-rearranged group is eleven AML-heavy, non-infant lines; a -0.26 difference on eleven versus 43 lines is not a result, it is a lead.

The honest status is 'unexplained and worth one experiment', and the entry says so rather than promoting it.

What SPT4 is, structurally

At 117 residues SPT4 is close to the minimum that can be a functional subunit. UniProt annotates a C4-type zinc finger at aa 16-36 and the SUPT5H-interaction region at aa 2-40 1; the interaction surface and the zinc finger are effectively the same piece of protein. In the cryo-EM complexes the SPT4-SPT5 pair forms a clamp element on Pol II, retained through both the paused and the activated states, with SPT6 opening the DSIF RNA clamp at pause release 23.

One practical consequence, from the only functional depletion study cited here: knocking down either DSIF subunit reduces the protein stability of the other 6. SPT4 and SPT5 are not independently tunable, which matters for interpreting the discordant DepMap scores — a CRISPR knockout of SUPT4H1 is, to a first approximation, a knockout of DSIF, and so is a knockout of SUPT5H, and the two should therefore have behaved alike. That they did not is the anomaly.

Cell type, cell state, localisation and epigenetic readout
AxisDocumentedWhere
Cell typeNone in KMT2A-rearranged disease. Reconstituted complexes; Jurkat T cells for the depletion study.236
Cell stateDSIF depletion impairs growth without causing death6
SubcellularNucleus1
SubnuclearNot established.—
Epigenetic readoutNone. SPT4 has no catalytic activity and no histone mark is attributable to it.1

Every row except the first two is blank, and that is the accurate state of knowledge.

Facts and figures
uniprotP63272
gene symbolSUPT4H1
protein nameTranscription elongation factor SPT4
length AA117
alphafold IDAF-P63272-F1
accession resolutionUniProt REST query gene_exact:SUPT4H1 AND organism_id:9606 AND reviewed:true, run 2026-09-19; one reviewed hit; primary gene name read back as SUPT4H1 and length 117 confirmed from the same fetched record.
complexDSIF (DRB sensitivity-inducing factor), with SUPT5H
classificationNOT a documented interactor of KMT2A::AFF1; an unexplained dependency signal with zero supporting literature
classification reason0 PubMed records co-occurring with MLL or KMT2A across three symbol spellings; the only KMT2A-linked observation is a dependency rank, which is a lead and not a finding
interaction evidence classcryo-EM of reconstituted Pol II complexes; CRISPRi in an HIV latency model; unbiased DepMap scan
shown for fusionnone
epigenetic readoutnone - no catalytic activity toward histones
aliases retired or alternate
[
  "SPT4",
  "SUPT4H",
  "SPT4H"
]
complex partners
[
  "gene-supt5h"
]
domains
[
  {
    "name": "SUPT5H (SPT5) interaction region",
    "start": 2,
    "end": 40,
    "source": "UniProt P63272, fetched 2026-09-19"
  },
  {
    "name": "ZnF (C4-type zinc finger)",
    "start": 16,
    "end": 36,
    "source": "UniProt P63272, fetched 2026-09-19"
  }
]
localization
{
  "subcellular": "nucleus",
  "subnuclear": "not established",
  "chromatin_targeting": "travels with Pol II as part of DSIF"
}
essentiality
{
  "depmap_scan": "analyses/depmap-kmt2a-selective-scan-full",
  "pan_essentiality_all_lines": -0.6692,
  "n_all_lines": 1208,
  "mean_kmt2a_rearranged": -1.144,
  "mean_other_leukemia": -0.8798,
  "kmt2a_selectivity_difference": -0.2642,
  "rank": 138,
  "n_scanned": 17728,
  "percentile": 0.778,
  "interpretation": "top 1% for KMT2A selectivity, ahead of DOT1L, AFF1 and ELL - but discordant with its obligate partner SPT5 (rank 767), which is the main reason to treat it as a lead rather than a result"
}
enumeration record counts
[
  {
    "query": "(MLL OR KMT2A) AND (SUPT4H1 OR SPT4 OR SUPT4H)",
    "records": 0,
    "date": "2026-09-19"
  }
]
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -1.144,
  "chronos_mean_other_leukaemia": -0.8798,
  "difference": -0.2642,
  "pan_essentiality_all_lines": -0.6692,
  "selectivity_rank": 138,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
How this could bite us back
  • Pan-essentiality of -0.67 across 1,208 lines means any DSIF-directed agent has a narrow window at best, and the differential here is a quarter of MEN1's.
  • Because the two DSIF subunits destabilise each other, targeting SPT4 is targeting DSIF, so the apparent subunit selectivity would not survive into a drug.
  • Acting on a dependency rank with zero supporting literature is exactly the failure mode this graph tries to avoid. This entry is filed as a lead and should not be promoted without the permutation test and a direct knockdown.
  • SPT4 is 117 residues with essentially no surface that is not a protein-protein interface. It has no enzymatic pocket and would be a hard small-molecule target even if the biology held.
Still open
  1. Is the SUPT4H1 selectivity signal real? A permutation test on this gene, and a check of whether it survives lineage conditioning and line-normalisation, would cost nothing and is the obvious next step.
  2. Why do SPT4 and SPT5 score so differently when they are an obligate, mutually stabilising pair? If the answer is guide quality or copy number rather than biology, this entry's central observation dissolves.
  3. Does SPT4 have a DSIF-independent function? That would be the only mechanistic route to a genuine subunit-specific dependency, and none is documented.
  4. Would SUPT4H1 knockdown in a KMT2A::AFF1 B-ALL line reduce proliferation more than in a matched non-rearranged line? This is the direct test and it has never been run.
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB P63272 (SPT4H_HUMAN, Transcription elongation factor SPT4) is 117 aa and annotates a C4-type zinc finger at aa 16-36 and a region at aa 2-40 that interacts with SUPT5H.

    UniProtKB P63272 (SPT4H_HUMAN), Transcription elongation factor SPT4 · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
    What this evidence cannot show

    Record fetched 2026-09-19; primary gene name read back as SUPT4H1 and length 117 confirmed from the same fetched record. Both annotated features are curator aggregations; no residue-level contact map is asserted.

    Evidence that argues against this (1)
    • A 117-residue protein annotated with a zinc finger and an interaction region that overlap almost completely is not well described by a domain diagram. The cryo-EM structures show SPT4-SPT5 as a clamp element on the polymerase, a description a linear domain map cannot convey, and any figure built from these coordinates will be close to uninformative. Vos SM et al. 2018
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Cryo-EM of the paused transcription elongation complex Pol II-DSIF-NELF at 3.2 A shows paused Pol II stabilised by DSIF and NELF, with NELF binding the polymerase funnel, contacting the trigger loop and restraining the polymerase mobility required for pause release, and with flexible NELF tentacles able to contact DSIF and exiting RNA.

    Structure of paused transcription complex Pol II-DSIF-NELF · Vos SM, Farnung L, Urlaub H, Cramer P · Nature 560(7720):601-606 2018
    PMID 30135580DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. Reconstituted complex with Sus scrofa Pol II and human factors, no chromatin, no leukaemia and no MLL fusion. The abstract describes DSIF as a whole and does not separate SPT4's contribution from SPT5's.

    Evidence that argues against this (1)
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In the activated elongation complex Pol II-DSIF-PAF-SPT6 at 3.1 A, PAF displaces NELF from the Pol II funnel for pause release, P-TEFb phosphorylates the Pol II CTD linker, and SPT6 binds the phosphorylated linker and opens the RNA clamp formed by DSIF; DSIF is retained in both the paused and activated states.

    Structure of activated transcription complex Pol II-DSIF-PAF-SPT6 · Vos SM, Farnung L, Boehning M, Wigge C, Linden A, Urlaub H, Cramer P · Nature 560(7720):607-612 2018
    PMID 30135578DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. Reconstituted and chromatin-free. 'DSIF is retained in both states' is read from the comparison the authors draw between the two structures rather than from a time-resolved experiment.

    Evidence that argues against this (1)
    • Because DSIF is present in both states, no DSIF measurement can by itself report whether a gene is paused or released - which is precisely what a KMT2A-fusion experiment would need it to do. NELF, which is removed at pause release, is the informative factor, and NELF has even less KMT2A literature than DSIF. Vos SM et al. 2018
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    As of 2026-09-19, PubMed returns 0 records for '(MLL OR KMT2A) AND (SUPT4H1 OR SPT4 OR SUPT4H)'. SPT4 is entirely unstudied in KMT2A-rearranged leukaemia.

    What this evidence cannot show

    Symbol set searched: SUPT4H1, SPT4, SUPT4H. Not searched: SPT4H, and any paper that names only 'DSIF' without naming SPT4 - although the companion DSIF query in gene-supt5h#E4 returned 1 record, so the combined coverage is better than this query alone. A zero is a floor, not a proof of absence.

    Evidence that argues against this (1)
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    In the project's genome-wide KMT2A-rearranged selectivity scan of 17,728 genes, SUPT4H1 ranks 138 (0.778th percentile) with a difference of -0.2642 (KMT2A-rearranged mean -1.1440 against other-leukaemia mean -0.8798) and a pan-essentiality of -0.6692 across 1,208 lines. In the same scan its DSIF partner SUPT5H ranks 767 (-0.1215, pan-essentiality -1.5066), DOT1L ranks 184 (-0.2310), AFF1 ranks 160 (-0.2472), ELL ranks 153 (-0.2533), KMT2A ranks 11 (-0.5411) and MEN1 ranks 1 (-0.8656).

    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

    All rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19 in one pass; no number typed from recall. Eleven KMT2A-rearranged against 43 other leukaemia lines, none infant-derived, AML-heavy. A rearranged mean of -1.14 is deep enough for partial floor compression. The scan has not been permutation-tested for this gene.

    Evidence that argues against this (1)
    • The strongest argument against taking this rank seriously comes from inside the same dataset: SPT4 and SPT5 are an obligate pair whose knockdown destabilises each other, so their selectivity scores should track, and they differ by a factor of two in difference and by 629 rank positions. Either the pairing is less obligate in CRISPR knockout than in CRISPRi knockdown, or one of the two numbers is noise - and there is no way to tell which from this data alone. Krasnopolsky S et al. 2021
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    CRISPR interference knockdown of Spt4, Spt5 or the whole DSIF complex inhibited transcription, decreased promoter occupancy of Pol II and Cdk9 and to a lesser extent AFF4, and targeting the expression of one DSIF subunit reduced the protein stability of its counterpart partner; depletion impaired cell growth but did not cause cell death.

    CRISPRi-mediated depletion of Spt4 and Spt5 reveals a role for DSIF in the control of HIV latency · Krasnopolsky S, Novikov A, Kuzmina A, Taube R · Biochimica et Biophysica Acta - Gene Regulatory Mechanisms 1864(1):194656 2021
    PMID 33333262DOI
    What this evidence cannot show

    Abstract fetched and read on 2026-09-19; full text not read. Jurkat cells and HIV-1 Tat transactivation; not leukaemia, not a KMT2A fusion. CRISPRi knockdown is not CRISPR knockout, and the mutual-destabilisation result may not transfer to the knockout regime DepMap uses.

    Evidence that argues against this (1)
Connected to