SUPT16H (SPT16) - the large FACT subunit: pan-essential, an anti-dependency in KMT2A-rearranged lines, and drugged in MLL-r leukaemia by a compound that works through p53

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.

• SUPT16H is required by essentially every cell line tested (-1.763 across 1,208 lines). Any systemic FACT inhibitor carries general transcription toxicity, and there is no KMT2A-selective window in the data to trade against it.
• CBL0137 is the only FACT-directed agent with MLL-r data and its mechanism runs through p53. The subset of patients with TP53 lesions - those with the worst outcomes - is the subset in which that mechanism should fail.
• Trapping FACT on chromatin is not the same as removing FACT. A programme that used CBL0137 data to justify a FACT degrader would be reasoning from a gain-of-function tool to a loss-of-function hypothesis.
• The interferon response CBL0137 induces is double-edged in a paediatric leukaemia setting: it is both a plausible contributor to efficacy and a plausible source of toxicity, and the published work does not separate them.

1. Does FACT occupancy differ at KMT2A::AFF1 target genes relative to expression-matched non-target genes? A CRISPR viability scan cannot answer this and none of the evidence here does; an SPT16 CUT&RUN in SEM or MV4;11 with an expression-matched control gene set would.
2. Is the ENL-FACT contact preserved when ENL is present as the 3' half of KMT2A::MLLT1? The fusion retains the ENL YEATS domain, so the prediction is directly testable by co-immunoprecipitation in a KMT2A::MLLT1 line.
3. Does CBL0137 efficacy in MLL-r models require p53? The published mechanism says yes; if it does, the compound is not a FACT-pathway argument for this disease at all, and isogenic TP53-null MLL-r lines would settle it.
4. Why is SSRP1 a stronger anti-dependency (+0.2651, rank 17,672) than its obligate partner SPT16 (+0.1165, rank 17,162) when the two function as a heterodimer? Either floor compression differs between them or the subunits are not equally limiting.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   FACT, the activity required for Pol II transcription of chromatin templates in vitro, comprises a human homologue of S. cerevisiae Spt16/Cdc68 and the HMG-1-like protein structure-specific recognition protein 1 (SSRP1); FACT specifically interacts with nucleosomes and H2A/H2B dimers, and its activity is abrogated by covalently crosslinking nucleosomal histones.

   The chromatin-specific transcription elongation factor FACT comprises human SPT16 and SSRP1 proteins · Orphanides G, Wu WH, Lane WS, Hampsey M, Reinberg D · Nature 400(6741):284-8 1999
   PMID 10421373DOI
   ▸What this evidence cannot show

   Abstract fetched from PubMed and read on 2026-09-19; full text not read. Title, authors, journal, volume, pages, PMID and DOI were copied out of that same fetched record. In vitro reconstitution on naked and chromatinised templates with purified factors; no cellular, haematopoietic or leukaemia system, and no KMT2A context whatsoever. The nucleosome-disassembly model stated here is the authors' interpretation of a crosslinking result, not a direct observation of disassembly.

   ▸Evidence that argues against this (1)

   • The in vitro nucleosome-disassembly model from 1999 has been substantially revised: later structural work describes FACT as binding and reorganising the nucleosome rather than simply evicting H2A-H2B, and the modern literature still debates whether FACT's principal role in vivo is disassembly ahead of polymerase or reassembly behind it. Nothing in this entry depends on which model is right, but the 1999 framing should not be quoted as current. Li Y et al. 2023
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The curaxin CBL0137 decreased viability of 12 MLL-rearranged leukaemia cell lines and 3 MLL-r ALL patient-derived xenograft cell populations in vitro with submicromolar IC50s, reduced leukaemia burden in a subcutaneous MV4;11 MLL-r AML model and in 5 MLL-r ALL patient-derived xenograft models, and exerted its anticancer effect by trapping FACT into chromatin, activating the p53 pathway and inducing an interferon response.

   Potent antileukemic activity of curaxin CBL0137 against MLL-rearranged leukemia · Somers K, Kosciolek A, Bongers A, El-Ayoubi A, Karsa M, Mayoh C, Wadham C, Middlemiss S, Neznanov N, Kees UR, Lock RB, Gudkov A, Sutton R, Gurova K, Haber M, Norris MD, Henderson MJ · International Journal of Cancer 146(7):1902-1916 2020
   PMID 31325323DOI
   ▸What this evidence cannot show

   Abstract fetched from PubMed and read on 2026-09-19; full text not read, and all identifiers copied from that fetched record. **The mechanism is p53-dependent and interferon-mediated, not KMT2A-directed**, and the abstract describes no non-MLL-r leukaemia comparator arm, so nothing here establishes MLL-r selectivity. MV4;11 is KMT2A::AFF1 with FLT3-ITD, so a FLT3-driven contribution to sensitivity is not excluded. 'Trapping FACT into chromatin' is a gain-of-occupancy mechanism and is not evidence that FACT function is required.

   ▸Evidence that argues against this (1)

   • The same compound class is reported active in unrelated leukaemias driven by entirely different lesions - JAK2 V617F myeloproliferative disease and BCR::ABL1 chronic myeloid leukaemia - which is what a p53/interferon mechanism predicts and which argues directly against reading CBL0137 activity as evidence for a KMT2A-specific FACT dependency. Pearson S et al. 2022
3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   ENL, the YEATS-domain acetylation reader encoded by MLLT1, occupies the majority of super-enhancers across cancer types; rapid auxin-degron depletion of ENL decommissions those super-enhancers and represses super-enhancer-controlled oncogenes such as MYC; and ENL interacts with the histone chaperone FACT complex and is required for the deposition of FACT over cis-regulatory elements, a process regulated by BRD4.

   Targeting the Epigenetic Reader ENL Inhibits Super-Enhancer-Driven Oncogenic Transcription and Synergizes with BET Inhibition to Suppress Tumor Progression · Chen Y, Ying Y, Ma W, Ma H, Shi L, Gao X, Jia M, Li M, Song X, Kong W, Chen W, Zheng X, Muluh TA, Wang X, Wang M, Shu XS · Cancer Research 84(8):1237-1251 2024
   PMID 38241700DOI
   ▸What this evidence cannot show

   Abstract fetched from PubMed and read on 2026-09-19; full text not read, identifiers copied from the fetched record. **The system is colorectal cancer, not leukaemia, and ENL is wild-type and overexpressed, not fused.** No KMT2A fusion is tested anywhere in this work. The ENL-FACT interaction is asserted at the level of 'interacts' in an abstract; the assay, the direction of the contact and whether it is direct are not established from what was read. Treating this as evidence that KMT2A::MLLT1 or KMT2A::AFF1 recruits FACT would be an unsupported two-step extrapolation.

   ▸Evidence that argues against this (1)

   • The counted absence cuts directly against extrapolating this result: there is no published physical link between FACT and AFF1, AFF4, ELL2 or the super elongation complex, so the ENL-FACT contact is currently a property of ENL outside the fusion context rather than a property of the fusion machinery. NCBI PubMed, queried 2026-09-19 2026
4. E4moderate 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, SUPT16H has a pan-essentiality of -1.763 across 1,208 lines but a KMT2A-rearranged selectivity difference of +0.1165 (KMT2A-rearranged mean -1.8124 against other-leukaemia mean -1.9288), ranking 17,162 of 17,728 (96.807th percentile) - an anti-dependency. SSRP1 is more extreme still at +0.2651, rank 17,672 (99.684th percentile).

   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

   Rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19 in a scripted pass; no number typed from recall. Eleven KMT2A-rearranged against 43 other leukaemia lines, AML-heavy, none infant-derived. **At a pan-essentiality of -1.76 both arms are close to the assay floor, where the difference statistic is compressed and can invert**, so the positive sign should be read as 'no selective requirement' rather than as a real protective effect.

   ▸Evidence that argues against this (1)

   • Floor compression is a genuine alternative explanation rather than a formality here: a gene that kills essentially every line cannot show a selective dependency in this statistic regardless of biology, so a CRISPR knockout viability scan is the wrong instrument for asking whether FACT matters more at fusion target genes. The question it cannot answer is a nascent-transcription question, not a viability question. Broad Institute DepMap / Open Targets Platform; analyses/depmap-interactome5-lineage-split 2026
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   As of 2026-09-19, PubMed returns 1 record for '(MLL OR KMT2A) AND (SUPT16H OR SPT16 OR FACP140 OR FACTP140)', 1 for '(MLL OR KMT2A) AND (SSRP1 OR T160 OR FACT80)', 2 for the combined FACT query, and 1 for '(SUPT16H OR SSRP1 OR "FACT complex") AND (AFF4 OR AFF1 OR ELL2 OR "super elongation complex" OR MLLT1 OR MLLT3)'. Title-checking the combined query's 2 records leaves exactly one relevant paper, the CBL0137 study; the other is a CASC5 gene-expression report. The single SEC-query record is a 2002 PITSLRE kinase paper, not a FACT-SEC result. '(SUPT16H OR SSRP1 OR "FACT complex") AND (DOT1L OR H3K79)' returns 0.

   PubMed E-utilities esearch record counts for the FACT subunits against KMT2A, MLL and the super elongation complex, run for interactome round 5 · NCBI PubMed, queried 2026-09-19 2026
   ▸What this evidence cannot show

   All counts are from the esearch `count` field, queried 2026-09-19. **Every hit in the counted queries was title-checked**, which is why the reported relevant number (1) is lower than the raw count (2). Symbol set searched: SUPT16H, SPT16, FACP140, FACTP140, SSRP1, T160, FACT80, FACT140, and the phrase 'FACT complex'. Not searched as phrases: 'facilitates chromatin transcription', 'Cdc68'. A counted absence is a floor on recall, and the broader query '"FACT complex" AND (leukemia OR leukaemia)' does return 9 records, none of which is about KMT2A on title-check.

   ▸Evidence that argues against this (1)

   • The alias FACT is an ordinary English word and 'T160' is a phosphosite label used across the kinase literature, so a naive alias sweep here inflates rather than deflates - the opposite hazard to the septin case in the standing brief. The counts above were kept deliberately narrow and title-checked for that reason, which also makes them conservative on recall: a paper discussing 'facilitates chromatin transcription' without a gene symbol would be missed. Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   On the lineage-split dependency run for this beat, SUPT16H shows no KMT2A-rearranged requirement in either lineage half: KMT2A-rearranged myeloid raw mean gene effect -1.958 (z -0.11 in other-leukaemia SDs), KMT2A-rearranged B-lineage raw mean -1.639 (z +0.86), against an other-leukaemia mean of -1.921 (SD 0.327) and a non-leukaemia mean of -1.753. The NPM1c line OCI-AML3 sits at raw -2.053 (z -0.40) and U-937 at z -0.13. SSRP1 behaves the same way: myeloid raw -1.233 (z +0.54), B-lineage raw -1.118 (z +0.90), other leukaemia -1.404.

   Interactome round 5 panel, lineage-split DepMap dependency (FACT, NSL, MSL, COMPASS remainder, Mediator) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-interactome5-lineage-split 2026
   ▸What this evidence cannot show

   Rows read out of analyses/depmap-interactome5-lineage-split/results.json on 2026-09-19. Six myeloid and four B-lineage index lines contributed for this gene, so a single line moves a mean substantially. The split is by lineage LABEL, not by expression signature. **Raw effects are quoted beside every z deliberately**: at these effect sizes both arms are lethal and the z is a difference between two lethal effects, which is precisely the regime in which a z-score misleads.

   ▸Evidence that argues against this (1)

   • The same run flags CXXC1 and MED1 as myeloid-identity artefacts, which shows the split can separate lineage effects when they exist. FACT is not such an artefact - it is simply uniformly essential - but with six and four index lines this run can exclude only a large KMT2A-specific requirement, not a modest one. Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProtKB Q9Y5B9 (SP16H_HUMAN, FACT complex subunit SPT16) is 1,047 aa with synonyms FACT140 and FACTP140, is annotated to nucleus and chromosome, has 17 PDB cross-references, and carries exactly one annotated structural feature with residue boundaries: a coiled coil at residues 432-507. No named domain is given coordinates.

   UniProtKB Q9Y5B9 (SP16H_HUMAN), FACT complex subunit SPT16 · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
   ▸What this evidence cannot show

   Record fetched 2026-09-19 via the UniProt REST search with gene_exact:SUPT16H AND organism_id:9606 AND reviewed:true; **primary gene name read back as SUPT16H and length 1,047 confirmed from the same fetched record**, per the standing accession rule. A single reviewed hit was returned, so no synonym-collision disambiguation was needed. Absence of annotated domain boundaries reflects curation state, not absence of structure - 17 PDB entries exist.

   ▸Evidence that argues against this (1)

   • Because UniProt assigns no boundaries to the aminopeptidase-like, dimerisation, middle and acidic C-terminal regions that the structural literature names, this entry contributes only a coiled-coil lane to the domain viewer. Filling the gap from a review schematic would be exactly the kind of invented coordinate the join rules forbid. Marciano G et al. 2016

• binds →SSRP1 - the small FACT subunit, and the only protein in this beat that binds a confirmed KMT2A-fusion interactor directly (LEDGF/p75, via HMG box to PWWP)

  SPT16 and SSRP1 form the obligate FACT heterodimer.

• part of →FACT (SPT16-SSRP1) - the chromatin arm of transcription elongation, and a clean negative result for the KMT2A::AFF1 interactome

  Large subunit of the FACT histone chaperone.

• binds →MLLT1 / ENL — the reader that beat its own paralogue, with a degrader already in hand

  ENL interacts with FACT and is required for FACT deposition over cis-regulatory elements - shown in colorectal cancer, not leukaemia.

• related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  No published physical link: FACT-versus-SEC/AFF1 queries return zero relevant records and the DepMap difference is an anti-dependency.

• ← bindsSSRP1 - the small FACT subunit, and the only protein in this beat that binds a confirmed KMT2A-fusion interactor directly (LEDGF/p75, via HMG box to PWWP)

  Obligate FACT heterodimer partner.

• ← part ofFACT (SPT16-SSRP1) - the chromatin arm of transcription elongation, and a clean negative result for the KMT2A::AFF1 interactome

  SPT16 is the large subunit.

• ← related toA null that costs one command, run first rather than afterwards - and SSRP1 is genuinely LESS required in KMT2A-rearranged lines, while the rest of the chaperone layer is not much of anything

  no significant selectivity, unlike its partner