SEPTIN2 (septin-2 / NEDD5) — KMT2A fusion partner in t(2;11)(q37;q23)

No three-dimensional model is available for this protein, so the map above is the whole picture. That is itself informative: AlphaFold does not model chains this long, and the parts that have been crystallised are held under the entries for the individual domains.

• If a coiled-coil-free septin transforms, then designing a drug against 'the partner coiled coil' would leave two of five septin fusions untouched.
• SEPTIN2 is measurably LESS required in KMT2A-rearranged lines than in other leukaemias, so any strategy that assumed the wild-type partner is a co-dependency starts from the wrong sign.

1. SEPTIN2 has no coiled coil and still forms a recurrent KMT2A fusion. Does KMT2A::SEPTIN2 oligomerise, and if so through what surface — the G interface, the 260-270 dimerisation region, or neither?
2. The 260-270 dimerisation-important region lies inside the retained G domain of the fusion. It is the one concretely annotated self-association surface in any KMT2A-partner septin and has never been mutated in a fusion context.

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

   Human septin2 is UniProt Q15019 (SEPT2_HUMAN), primary gene name SEPTIN2, 361 aa, with the domain architecture recorded in data.domains.

   UniProtKB Q15019 (SEPT2_HUMAN) 361 aa, Homo sapiens · UniProt Consortium · UniProtKB 2026
   ▸What this evidence cannot show

   Canonical isoform. Coordinates fetched from the UniProt REST record on 2026-09-19; the PRIMARY gene name was read back rather than relying on a gene_exact match, which for septins is unusually dangerous because the whole family was renamed (SEPT9 -> SEPTIN9 and so on) and every record still carries the retired symbol as a synonym.

   ▸Evidence that argues against this (1)

   • UniProt coiled-coil annotations on septins are prediction-driven and should not be read as assembly modules: crystallography and electron microscopy of the human SEPT2-SEPT6-SEPT7 complex showed the predicted coiled coils are not involved in, and not required for, complex or filament formation, which proceeds through the nucleotide-binding G interface and the N/C extensions. Sirajuddin M et al. 2007
2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   Mammalian septins are GTP-binding proteins that assemble into homo- and hetero-oligomers and filaments through an extended G domain forming a universal bipolar polymer building block, via conserved interactions between adjacent nucleotide-binding sites and the N- and C-terminal extensions, producing an architecture fundamentally different from actin, tubulin and intermediate filaments.

   Structural insight into filament formation by mammalian septins. · Sirajuddin M, Farkasovsky M, Hauer F, et al. · Nature 2007
   PMID 17637674DOI
   ▸What this evidence cannot show

   Structural work on the SEPT2/6/7 complex; extrapolated here to other family members on sequence and architectural homology.

   ▸Evidence that argues against this (1)

   • The fusion context does not obey this architecture: in MLL-SEPT6 the coiled coil is individually required for both self-association and transformation, which native filament assembly says it should not be. Ono R et al. 2005
3. E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.dataset

   In a genome-wide DepMap CRISPR (Chronos) scan, SEPTIN2 had a mean gene effect of -0.199 across 1090 lines, a KMT2A-rearranged group mean of -0.101 versus -0.204 in other leukaemia lines, and ranked 16,960 of 17,728 for KMT2A selectivity (95.7th percentile) — i.e. it is LESS required in KMT2A-rearranged lines than in other leukaemia lines.

   DepMap CRISPR (Chronos) genome-wide KMT2A-selectivity scan of 17,728 protein-coding genes, 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines vs other leukaemia lines (analyses/depmap-kmt2a-selective-scan-full/results.json, all_results; septin/ABI1/MLLT11 rows looked up for this beat) · Broad Institute DepMap; Cellosaurus; Open Targets Platform; scan run by this project · Database and derived analysis 2026
   ▸What this evidence cannot show

   No KMT2A-septin-fusion line exists in the DepMap panel; the KMT2A-rearranged group is 11 lines; Chronos values near zero are noise-dominated. Reported to close a question, not to open one.

   ▸Evidence that argues against this (1)

   • A single-gene CRISPR score cannot detect a requirement buffered by the other twelve septins; the same redundancy that left Sept6-null mice phenotypically normal would produce a null score here. Ono R et al. 2005
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The septin system is highly redundant: Sept6-null mice show no gross abnormality, no cytokinesis change and no spontaneous malignancy, Sept6 loss does not alter MLL-SEPT6-induced disease, and depleting the close homologue Sept11 does not affect Sept6-null cells.

   Disruption of Sept6, a fusion partner gene of MLL, does not affect ontogeny, leukemogenesis induced by MLL-SEPT6, or phenotype induced by the loss of Sept4. · Ono R, Ihara M, Nakajima H, et al. · Mol Cell Biol 2005
   PMID 16314519DOI
   ▸What this evidence cannot show

   Mouse genetics with gross phenotype as readout; demonstrated for Sept6 and Sept11, extended here to the family as an argument about redundancy rather than a measurement on this particular gene.

   ▸Evidence that argues against this (1)

   • Redundancy is not complete at the expression level in patients: septin transcript changes distinguish MLL-SEPT2 and MLL-SEPT6 leukaemias, which cluster together and apart from other AML subgroups by septin-expression profile, so the family is not interchangeable in vivo. Santos J et al. 2010
5. E_q91_septin2_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProtKB Q15019 (SEPT2_HUMAN, primary gene SEPTIN2, 361 aa, read back live 2026-09-21) carries NINE subcellular location values. Seven are attributed to one ECO:0000269, PubMed:15774761: 'Cytoplasm', 'Cytoplasm, cytoskeleton', 'Cytoplasm, cytoskeleton, spindle', 'Chromosome, centromere, kinetochore', 'Cleavage furrow', 'Midbody' and 'Cytoplasm, cell cortex'. Two more are ciliary: 'Cell projection, cilium, flagellum' (ECO:0000269 PubMed:25588830) and 'Cell projection, cilium membrane' (ECO:0000250). The location note records that in metaphase cells SEPTIN2 is localised within the microtubule spindle and at the metaphase plate IN CLOSE APPOSITION TO the kinetochores of congressed chromosomes; in cytokinesis at the midbody, ingressing cleavage furrow and central spindle; during bacterial infection as a collar next to actin at the pole of invading bacteria; and in epithelial cells colocalising with polyglutamylated tubulin around the trans-Golgi network. NO nuclear or subnuclear value appears anywhere in the record.

   UniProtKB Q15019 SUBCELLULAR LOCATION comments, evidence codes and feature table, fetched live 2026-09-21 (nine location values, seven from one paper, none of them nuclear) · UniProt Consortium · UniProtKB 2026
   ▸What this evidence cannot show

   Seven values from a single 2005 paper. The record is dense but not independent - it is one study's cell-cycle series unpacked into the controlled vocabulary, plus a sperm-annulus study of a different septin.

   ▸Evidence that argues against this (1)

   • One of the nine values is stronger than its source. 'Chromosome, centromere, kinetochore' is a hierarchical term asserting residence at the kinetochore, and the note beneath it in the same record says only 'in close apposition to the kinetochores'; the source paper's functional finding is that septin depletion causes loss of CENP-E FROM kinetochores, which is a statement about a motor protein, not about where septin sits. Anyone filtering this record for chromatin-associated proteins would collect SEPTIN2, and anyone reading the note would not. This is the MAML2 failure of tranche 3 repeated in a different vocabulary branch, and it is the reason this entry writes NOT APPLICABLE for the subnuclear field rather than transcribing the row. Spiliotis ET et al. 2005
6. E_q91_septin2_loc2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The source of seven of those nine values: septins localised TO THE METAPHASE PLATE during mitosis in mammalian cells, and septin depletion resulted in chromosome loss from the metaphase plate, lack of chromosome segregation and spindle elongation, and incomplete cytokinesis upon delayed mitotic exit. These defects correlated with LOSS OF THE MITOTIC MOTOR AND CHECKPOINT REGULATOR CENP-E FROM THE KINETOCHORES of congressing chromosomes, from which the authors propose that mammalian septins form a mitotic scaffold for CENP-E.

   A mitotic septin scaffold required for Mammalian chromosome congression and segregation. · Spiliotis ET, Kinoshita M, Nelson WJ · Science 2005
   PMID 15774761DOI
   ▸What this evidence cannot show

   Mammalian tissue-culture cells, not haematopoietic; the paper is about the mitotic scaffold function, and its chromosome-proximity observations are all in cells with no nuclear envelope. Read from the abstract.

   ▸Evidence that argues against this (1)

   • Even granting the scaffold model, this is the wrong kind of evidence to license a chromatin-localisation claim, and the reason generalises past septins. A protein required to keep CENP-E AT kinetochores need not be at kinetochores itself - it can act on the spindle side, on transport, or on the geometry of the plate, and the paper's own framing is a SCAFFOLD, i.e. a structure adjacent to the things it holds. This graph has had to make the same distinction before for cooperativity (see gene-spi1's nuclear-matrix near-miss in this tranche): functional dependence is not co-residence. Kremer BE et al. 2007
7. E_q91_septin2_loc3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   Counted absence and a measured symbol ratio, 2026-09-22, querytranslation read back on every line. Total literature: SEPT2[tiab] = 143 (retired symbol), SEPTIN2[tiab] = 71 (current symbol). Against the disease: SEPT2[tiab] AND MLL[tiab] = 10, SEPTIN2[tiab] AND MLL[tiab] = 0, SEPT2[tiab] AND KMT2A[tiab] = 0. Nuclear pool: SEPT2[tiab] AND (nuclear OR nucleus)[tiab] = 5, all five titles read, none reporting nuclear SEPT2; SEPTIN2[tiab] AND (nuclear OR nucleus)[tiab] = 6, same result. The fusion-name form "MLL-SEPT6"[tiab] OR "MLL-SEPT2"[tiab] = 14.

   Retired-symbol ratio and counted absence for SEPT2/SEPTIN2 against MLL and KMT2A, plus a nuclear-pool screen with all titles read, querytranslation recorded, run 2026-09-22 · NCBI E-utilities esearch (db=pubmed), run by Q91 · PubMed E-utilities 2026
   ▸What this evidence cannot show

   Bounded by titles and abstracts (rule 80). Three name forms were run - current symbol, retired symbol and fusion name - and NEDD5, DIFF6 and KIAA0158 were not. SEPT* was deliberately NOT used as a wildcard because it matches September, which this project has already been burned by.

   ▸Evidence that argues against this (1)

   • The 10-versus-0 split is a live reproduction of the table in this project's standing brief, which recorded (MLL OR KMT2A) AND SEPT2 = 10 and AND SEPTIN2 = 0 in cycle 40. Two years of indexing later the numbers have not converged at all, which refutes the natural assumption that a symbol change propagates into the index over time - PubMed does not retrospectively re-tag abstracts, so a renamed gene's older literature stays permanently unreachable from its new symbol. Any counted absence in this graph written against a renamed gene is therefore not merely stale, it will never self-correct. Cerveira N et al. 2006
8. E_q91_septin2_loc4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The nearest thing to a nuclear role for SEPTIN2, and the protein that moves is not SEPTIN2: knockdown of SEPT2, SEPT6 and SEPT7 causes stress fibres to disintegrate and cells to lose polarity, and that phenotype is induced by NUCLEAR ACCUMULATION OF THE ADAPTOR PROTEIN NCK, which is carried into the nucleus by SOCS7. SOCS7 possesses nuclear import and export signals and interacts with septins and with NCK through distinct domains. DNA damage induces actin and septin rearrangement and rapid nuclear accumulation of NCK and SOCS7, and the septin-SOCS7-NCK axis intersects the canonical DNA-damage cascade downstream of ATM/ATR and is essential for p53 Ser15 phosphorylation.

   Septins regulate actin organization and cell-cycle arrest through nuclear accumulation of NCK mediated by SOCS7. · Kremer BE, Adang LA, Macara IG · Cell 2007
   PMID 17803907DOI
   ▸What this evidence cannot show

   The nuclear import and export signals belong to SOCS7, not to SEPTIN2, and the paper does not report SEPT2 entering the nucleus. Cited here to close off a plausible misreading and because the DNA-damage connection is relevant to this graph's break-repair beat.

   ▸Evidence that argues against this (1)

   • This is the paper most likely to be mis-cited as evidence of nuclear septin, and it argues the opposite: septins act by SEQUESTERING a partner in the cytoplasm, so their functional reach into the nucleus is a consequence of their staying out of it. The parallel with gene-sumo1 in this tranche is exact and worth naming - in both cases the retrievable 'nuclear' literature for a protein is really the nuclear literature of something it binds, and in both cases transcribing it would have produced a confident wrong compartment. UniProt Consortium 2026
9. E_q91_septin2_loc5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Cell type and abundance in the index lesion, from primary patient material: both SEPT2 and MLL are DOWN-REGULATED in myeloid neoplasias carrying the MLL-SEPT2 fusion gene, and a survey of the other 13 known septin genes across 67 cases of myeloid neoplasia - including three patients with MLL-SEPT2, four with MLL-SEPT6 and three with MLL-SEPT9 - found statistically significant down-regulation relative to normal controls. The disease context for KMT2A::SEPTIN2 specifically is de novo and THERAPY-RELATED acute myeloid leukaemia and therapy-related myelodysplastic syndrome with t(2;11)(q37;q23).

   Expression pattern of the septin gene family in acute myeloid leukemias with and without MLL-SEPT fusion genes. · Santos J, Cerveira N, Bizarro S, Ribeiro FR et al. · Leuk Res 2010
   PMID 19748670DOI
   ▸What this evidence cannot show

   Transcript-level, small numbers (three MLL-SEPT2 patients), and myeloid throughout - nothing here is B-lineage or KMT2A::AFF1 (rule 73). Down-regulation of the wild-type allele in a fusion-bearing cell is a level, not a compartment.

   ▸Evidence that argues against this (1)

   • The down-regulation cuts against a model this entry's structural section leans on. If the fusion works by contributing a septin oligomerisation surface, one would expect the septin pathway to matter in the fusion-bearing cell - yet both partners' wild-type transcripts go DOWN, and Sept6 disruption in mouse does not affect ontogeny or leukaemogenesis (this entry's E4). An alternative reading of the whole septin-partner family is that the septin contributes a passive dimerisation module and its native biology is irrelevant, which would also explain why a septin with NO coiled coil, SEPTIN2, forms a recurrent fusion at all. Ono R et al. 2005
10. E_q91_septin2_loc6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The architecture that makes the fusion's compartment an open question: MLL fusion partners fall into four categories - nuclear proteins, cytoplasmic proteins, histone acetyltransferases and SEPTINS - and five septin genes (SEPT2, SEPT5, SEPT6, SEPT9, SEPT11) have been identified as MLL fusion partners, giving rise to chimeric proteins in which the N terminus of MLL is fused IN FRAME TO ALMOST THE ENTIRE OPEN READING FRAME of the septin partner gene, from heterogeneous breaks in distinct introns of both genes.

    MLL-SEPTIN gene fusions in hematological malignancies. · Cerveira N, Bizarro S, Teixeira MR · Biol Chem 2011
    PMID 21714766DOI
    ▸What this evidence cannot show

    A review, cited as a map of the partner architecture rather than as primary evidence (the quality bar permits this use). It reports no localisation experiment on any MLL-septin fusion protein, which is itself the finding.

    ▸Evidence that argues against this (1)

    • The four-category taxonomy this review uses - nuclear partners, cytoplasmic partners, HATs, septins - quietly assumes that a partner's native compartment survives fusion, and the whole point of this tranche is that this is untested for septins. It is testable and cheap: a GFP-tagged KMT2A::SEPTIN2 in any cell line, co-stained for a septin filament marker, would settle whether the chimera is nuclear like its KMT2A half, filamentous like its septin half, or something else. Nobody has done it for any of the five septin fusions in the twenty years since the first was cloned, which is why this entry's shown_for_fusion is a hard NO rather than an inference from the partner. van Binsbergen E et al. 2007

• ← fuses withKMT2A / MLL1 — the gene that breaks

  t(2;11)(q37;q23)

• part of →KMT2A::SEPTIN2 (MLL-SEPT2), t(2;11)(q37;q23) — the therapy-related septin

  G domain retained; no coiled coil exists to retain

• binds →SEPTIN6 (septin-6) — cytoskeletal GTPase, KMT2A fusion partner in t(X;11)(q24;q23)

  native SEPT2-SEPT6-SEPT7 heterotrimer

• ← related toSEPTIN5 (septin-5 / CDCrel-1) — KMT2A fusion partner in t(11;22)(q23;q11.2)

  both group-II septins and both KMT2A partners

• ← related toHow a cytoskeletal partner reaches chromatin at all — the KMT2A N-terminus drags it in, and the traffic runs both ways

  predicted sequestration target as an endogenous hetero-oligomer partner

• related to →Cytoplasmic localisation in KMT2A fusions: the SEPTIN case has an image, the GPHN case has none, and neither partner wins the compartment

  SEPTIN2 subnuclear localisation is NOT APPLICABLE - a cytoplasmic GTPase with no interphase nuclear pool in 143 SEPT2 or 71 SEPTIN2 records; UniProt's 'Chromosome, centromere, kinetochore' row is mitotic AND over-specifies its source, which says 'in close apposition to' (Q91 2026-09-22)

• related to →A partner's 'normal' compartment is usually read off a different tissue, and for at least two KMT2A partners that reading is wrong in blood cells

  Worked example of a partner's baseline compartment being miscalled BY A DATABASE rather than by a reader: SEPTIN2's kinetochore term is stronger than the sentence it came from (PMID 15774761) (Q91 2026-09-22)