SFPQ / PSF (P23246) - the last of 21 fusion-construct partners without an entry, and it is itself the 5' half of an ALL fusion that puts ABL1 in the nucleus

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.

• Nothing here supports giving any drug to anyone. The TKI, navitoclax and S-63845 sensitivities in E3b were measured in cytokine-dependent cell lines ectopically expressing a fusion, and the relapse observation in E3 is a retrospective note on a handful of patients in a case review, not a trial result. SFPQ-ABL1 is also a different disease from this project's index lesion: it is a BCP-ALL kinase fusion, not a KMT2A rearrangement, and no therapeutic inference should cross between them.
• The most citable sentence in this entry - that the 5' partner changes the address and the network of the same kinase - is drawn from one paper's abstract comparing two ectopically expressed constructs. It is a good result and it is not a law. Note in particular that BOTH fusions stayed TKI-sensitive, so the partner did not change the actionable target, and that a rival explanation exists in which the chimaera's missing ABL1 SH2 domain does the work instead of the partner.
• This entry must not be cited as evidence that paraspeckles matter in KMT2A-rearranged leukaemia. `localization-paraspeckles-neat1-kmt2a-r` records zero direct evidence after a dedicated search, and an anti-tag co-IP in HEK293 against an MLL-GAS7 construct does not assay a paraspeckle. The compartment remains unmeasured in this disease.
• The MLL-GAS7 row is typed `association`, not `physical association`, comes from a paper whose abstract never names SFPQ, and has no reported negative control. It is the weakest evidence item in this entry and the entry's opening sentence should not be allowed to borrow confidence from the SFPQ-ABL1 material, which is independent of it.

1. What does the SFPQ moiety contribute to SFPQ-ABL1? The general rule for ABL1 fusions is that the 5' partner supplies a coiled-coil or HLH domain whose oligomerisation is required for kinase activation (PMID 21435002), but that is stated as 'often' and the UniProt fields fetched here return no such domain for SFPQ - only two RRMs, long disordered regions and an RGG stretch. A targeted query for coiled-coil, NOPS and DBHS annotations in P23246, plus the SFPQ exon 9 boundary in residue coordinates, would settle what is actually in the chimaera.
2. Does the ABL1 SH2 deletion, rather than the SFPQ moiety, explain the altered phosphoproteome and weaker proliferation of SFPQ-ABL1? The two explanations are confounded in every experiment cited here, and an SH2-deleted BCR-ABL1 comparator would separate them. This is the same confound structure as the polarity-versus-MBM pair already in this graph.
3. Which figure of PMID 26766589 contains the SFPQ / MLL-GAS7 co-IP, and with what controls? The abstract never names SFPQ, so the row is curator-extracted, and the graph cites this paper ten times without anyone having read its interaction data. The full text would also show whether KHDRBS1, PRMT1 and SFPQ were recovered in one experiment or three.
4. Is SFPQ curated against any other KMT2A fusion construct, or against KMT2A itself? It appears here only against MLL-GAS7, a cytoskeletal-class partner, and a single construct row is the weakest form this enumeration produces.
5. Has SFPQ ever been measured in a KMT2A-rearranged cell at all - expression, localisation or dependency? This entry records evidence from HEK293, from ectopic expression in cytokine-dependent lines, and from BCP-ALL patient cases, and NONE of it is in a KMT2A-rearranged context. DepMap would answer the dependency half cheaply.
6. Does the nuclear localisation of SFPQ-ABL1 involve a paraspeckle or any named body? PMID 35061886 says only 'nuclear compartment'. Given that the SFPQ moiety carries the protein's paraspeckle associations, this is the obvious question and the abstract does not touch it.

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

   SFPQ is UniProt P23246 (SFPQ_HUMAN), Homo sapiens, 'Splicing factor, proline- and glutamine-rich', gene SFPQ, synonym PSF, 707 aa, 76,149 Da. Annotated features: RRM 1 at 297-369, RRM 2 at 371-452, disordered regions 1-273 and 579-707, and three 3-aa R-G-G repeats at 9-27. Subcellular location: Nucleus speckle, Nucleus matrix, Cytoplasm, with the note 'Predominantly in nuclear matrix'.

   UniProtKB entry P23246 (SFPQ_HUMAN), fields accession, protein_name, gene_names, length, mass, cc_subcellular_location, ft_domain, ft_region, fetched 2026-09-22
   ▸What this evidence cannot show

   Length, mass, accession, entry name and every coordinate quoted here were read from the JSON response on 2026-09-22, not from recall. The fetch requested only the fields listed, so absence of a coiled-coil or NOPS/DBHS dimerisation annotation in this evidence item is an absence in the QUERY, not in the record - no claim is made that SFPQ lacks such a region. UniProt subcellular assignments are a controlled vocabulary and 'Nucleus speckle' is NOT 'paraspeckle'; the paraspeckle assignment used elsewhere in this graph rests on the DBHS literature and on PMID 28712728, not on this record.

   ▸Evidence that argues against this (1)

   • A UniProt localisation set is a union over experiments, cell types and methods, not a statement about any one cell. 'Nucleus speckle, Nucleus matrix, Cytoplasm' with 'predominantly nuclear matrix' does not tell you where SFPQ is in a B-cell precursor, in a leukaemic blast, or in a cell expressing a KMT2A fusion, and this graph has no imaging of SFPQ in any such cell.
2. E2weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

   IntAct curates SFPQ (P23246, taxid 9606) as an `association` with the fusion construct `human_mll_gas7_fusion_protein` (EBI-1157918), detected by anti-tag co-immunoprecipitation in Homo sapiens HEK293 embryonic kidney cells, from PMID 26766589 (IntAct publication EBI-11512485). In the cycle-55 enumeration of 50 fusion-construct interaction rows spanning five constructs and 21 distinct partners, SFPQ was the only partner with no entry in this graph; its two co-partners on the same construct and the same publication, KHDRBS1 (Q07666) and PRMT1 (Q99873), both had one.

   Targeting Aberrant Epigenetic Networks Mediated by PRMT1 and KDM4C in Acute Myeloid Leukemia. · Cheung N, Fung TK, Zeisig BB, Holmes KB, Rane JK, et al. · Cancer Cell 2016
   PMID 26766589DOI
   ▸What this evidence cannot show

   PMID 26766589's title, journal, year, DOI and first author were verified live by efetch on 2026-09-22, and the abstract text was checked: it NEVER NAMES SFPQ. The interaction is therefore curator-extracted from the paper's figures or supplement, and the experimental context - which figure, which controls, whether it replicated, what the stoichiometry was - has NOT been read here; the full text was not obtained. The IntAct interaction type is `association`, the weaker of the two terms used across these rows, and no negative control is reported in the row. The system is a transfected tagged construct in HEK293 embryonic kidney cells: neither a leukaemia cell nor a B-cell precursor. The construct is MLL-GAS7, NOT the index fusion KMT2A::AFF1. The 50-row / 21-partner denominator inherits its own FLOOR caveat - the construct list was assembled by hand, so further fusion-construct interactors may exist.

   ▸Evidence that argues against this (2)

   • The natural reading - that a paraspeckle protein binds an MLL fusion, so paraspeckles are implicated in KMT2A-rearranged leukaemia - is not supported and is contradicted by what was actually measured. `localization-paraspeckles-neat1-kmt2a-r` records ZERO direct evidence for NEAT1 or paraspeckles in a KMT2A-rearranged context after a dedicated search, and a co-IP in HEK293 against an MLL-GAS7 construct does not assay a paraspeckle at all. This row raises the value of doing that measurement; it does not substitute for it.
   • That SFPQ was the single uncovered partner of 21 is partly an artefact of the coverage rule, which tests whether a `gene-<symbol>` or `protein-<symbol>` file EXISTS rather than whether it says anything. Twenty of twenty-one being 'covered' is a statement about file names, and several of those entries may mention the protein only in passing. The gap this entry closes was real; the 20/21 that flattered the instrument should not be over-read.
3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   SFPQ is itself the 5' partner of a recurrent ABL-class fusion in B-cell precursor ALL. FISH screening of 55 paediatric and 50 adult 'B-other' BCP-ALL cases found ABL-class aberrations (ABL1, ABL2, CSF1R, PDGFRB) in 3 (6%) of the adult and none of the childhood cases; RT-PCR and sequencing confirmed SFPQ-ABL1 in one adult case with t(1;9)(p34;q34). Reviewing all reported cases including that one, only six SFPQ-ABL1-positive BCP ALLs had been described, ALL of them harbouring a fusion between exon 9 of SFPQ and exon 4 of ABL1; the fusion is typically found in adolescents or younger adults without hyperleukocytosis, IKZF1 deletions are recurrent, and the few patients treated with neither a tyrosine kinase inhibitor nor allogeneic stem cell transplantation relapsed.

   SFPQ-ABL1-positive B-cell precursor acute lymphoblastic leukemias. · Biloglav A, et al. · Genes, Chromosomes & Cancer 2021
   PMID 32306475DOI
   ▸What this evidence cannot show

   Abstract-level only; the full text and the case-by-case review table were not read. Every number here is quoted from the abstract as fetched by efetch on 2026-09-22 - 55 paediatric, 50 adult, 3 (6%) adult, 0 childhood, six cases total, SFPQ exon 9 to ABL1 exon 4. 'Six reported cases' is a 2021 literature count and will have moved; it is not a current incidence. The 6% figure is ABL-class aberrations of any kind in ADULT B-other cases, NOT the frequency of SFPQ-ABL1, which was 1 of 105 cases screened. The relapse observation is from a handful of patients in a retrospective review and is not a trial result.

   ▸Evidence that argues against this (2)

   • This is a rare lesion and the numbers are small enough that a uniform junction across six cases can be a reporting artefact as easily as a biological constraint: cases are found by assays designed around known junctions, so an identical exon 9 / exon 4 breakpoint in every reported case may partly reflect how they were detected. The 0-of-55 paediatric result also bounds nothing strongly at that sample size.
   • SFPQ-ABL1 is a B-cell precursor ALL lesion but is NOT a KMT2A-rearranged lesion, and this graph's index disease is infant KMT2A::AFF1 B-ALL. The shared lineage makes the comparison interesting; it does not make SFPQ-ABL1 a model of the index disease, and the two fusions share no component protein.
4. E3bmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   SFPQ-ABL1 and BCR-ABL1 drive B-ALL by measurably different mechanisms despite sharing the ABL1 kinase. SFPQ-ABL1, identified in a child with Ph-like ALL, was sufficient to transform cytokine-dependent cell lines and those cells were sensitive to ABL1-targeting TKIs. In contrast to BCR-ABL1, SFPQ-ABL1 localised to the NUCLEAR compartment and was a WEAKER driver of cellular proliferation. Phosphoproteomics showed up-regulation of cell cycle, DNA replication and spliceosome pathways and down-regulation of signal transduction pathways including ErbB, NF-kB, VEGF and MAPK relative to BCR-ABL1-expressing cells; SFPQ-ABL1 did not activate PI3K/AKT and was associated with phosphorylation of G2/M cell-cycle proteins. It was sensitive to navitoclax and S-63845 and promoted survival by maintaining Mcl-1 and Bcl-xL expression. The authors conclude that the findings 'highlight the role that fusion partners have in mediating the function of ABL1 fusions'.

   SFPQ-ABL1 and BCR-ABL1 use different signaling networks to drive B-cell acute lymphoblastic leukemia. · Brown LM, et al. · Blood Advances 2022
   PMID 35061886DOI
   ▸What this evidence cannot show

   Abstract-level only; the full text, the figures and the phosphoproteomic tables were not read, so effect sizes, the number of replicates, and which comparisons were statistically supported are not established here. The system is ECTOPIC EXPRESSION of each fusion in cytokine-dependent cell lines, not patient cells, so every contrast - localisation, proliferation rate, pathway set - is measured in a construct-overexpression setting and the two fusions may not have been expressed at comparable levels; the abstract does not say. 'Nuclear compartment' is the abstract's phrase; no subnuclear body is specified and none should be inferred. The index case is a single child. Title, journal, year, DOI and first author verified live by efetch 2026-09-22.

   ▸Evidence that argues against this (2)

   • The comparison cuts against a simple 'partner sets the address' story as much as for it. SFPQ-ABL1 is a WEAKER proliferative driver that fails to activate PI3K/AKT, so if the partner is doing the work, it is doing less of it: the nuclear address comes with reduced signalling output, not a redirected oncogenic programme of equal strength. A reading in which the partner merely sequesters the kinase away from its cytoplasmic substrates fits these data too, and the abstract does not distinguish the two.
   • Both fusions remained TKI-sensitive, which is the clinically decisive property and is UNCHANGED by the partner. An entry using this paper to argue that the 5' partner determines therapeutic vulnerability would be overstating it: the partner changed the pathway map and the apoptotic dependency (Mcl-1/Bcl-xL, navitoclax, S-63845), but the kinase remained the actionable target in both.
5. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   SFPQ-ABL1 and RCSD1-ABL1 are the ABL1 fusions characterised by a break in intron 4 of ABL1, with the consequence that the chimeric proteins LACK the SH2 domain of ABL1, and both are associated with B-cell ALL. Across ABL1 fusions generally the kinase domain is always retained and the chimaera also carries the N-terminal part of the partner protein, which often includes a coiled-coil or helix-loop-helix domain; these domains allow oligomerisation of the protein, which is required for tyrosine kinase activation, cytoskeletal localisation and neoplastic transformation. At the time of the review six genes other than BCR were known to fuse to ABL1, and NUP214-ABL1 was the second most prevalent ABL1 fusion in malignant haemopathies at 5% of T-cell ALL.

   ABL1 fusion genes in hematological malignancies: a review. · De Braekeleer E, et al. · European Journal of Haematology 2011
   PMID 21435002DOI
   ▸What this evidence cannot show

   A 2011 REVIEW, abstract-level, and the field has moved: 'six other genes' and the ABL-class fusion catalogue are both out of date, and this is cited for the structural logic of the intron-4 break and the general oligomerisation rule, not as a current census. The oligomerisation statement is explicitly the GENERAL pattern across ABL1 fusion partners, qualified in the source by 'often' - it is NOT a claim that SFPQ contributes a coiled-coil or HLH domain, and the UniProt fields fetched for E1 returned no such annotation for SFPQ. What the SFPQ moiety contributes to SFPQ-ABL1 is an open question in this entry, not a recorded fact.

   ▸Evidence that argues against this (1)

   • The SH2 deletion gives a rival explanation for E3b's findings that does not involve the partner's own properties at all. SH2 is a phosphotyrosine-binding module that shapes ABL1 substrate selection and processive phosphorylation, so a chimaera lacking it might show an altered phosphoproteome and weaker proliferative drive for reasons entirely internal to the truncated kinase - no partner-mediated relocalisation required. E3b's nuclear localisation is not explained this way, but the pathway and potency differences may be. Nothing in this graph separates the two explanations.
6. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   SFPQ is a verified component of the HEXIM1-DNA-PK-paraspeckle components-ribonucleoprotein complex (HDP-RNP), a ribonuclear complex built around HEXIM1 and the long non-coding RNA NEAT1 and identified by immunoprecipitation, mass spectrometry and RNA sequencing. The HDP-RNP contains the DNA-PK subunits DNA-PKcs, Ku70 and Ku80 together with the paraspeckle proteins SFPQ, NONO, PSPC1, RBM14 and MATRIN3; binding of HEXIM1 to NEAT1 is required for its assembly, and the complex is required for the innate immune response to foreign DNA through the cGAS-STING-IRF3 pathway.

   HEXIM1 and NEAT1 Long Non-coding RNA Form a Multi-subunit Complex that Regulates DNA-Mediated Innate Immune Response. · Morchikh M, Cribier A, Raffel S, et al. · Molecular Cell 67:387-399.e5 2017
   PMID 28712728DOI
   ▸What this evidence cannot show

   This citation was already stored in `gene-hexim1` (E_q65_loc1) and was RE-FETCHED live on 2026-09-22 rather than copied: PMID, title, journal, year, DOI and first author all confirmed, and the abstract confirmed to name SFPQ, NEAT1 and paraspeckles. The complex was characterised in the context of innate immune signalling, not leukaemia, and nothing in this paper concerns KMT2A, a fusion, or a leukaemic cell. It is cited here to source SFPQ's paraspeckle-protein status, which UniProt's own subcellular-location field does not state.

   ▸Evidence that argues against this (1)

   • Membership of an immunoprecipitated ribonucleoprotein is not evidence of residence in a paraspeckle BODY, and the paper's own naming - 'paraspeckle components' - describes the proteins' provenance rather than the complex's location. Calling SFPQ a paraspeckle protein on this basis is standard usage but is one inferential step from what was measured, and UniProt independently assigns it to nuclear speckles and the nuclear matrix.

• ← related toAsking IntAct for its own fusion constructs found a sixth, exposed a duplicate record that would have inflated the count by three, and failed its own control 4 of 5 - so the floor is lower, not closed

  the duplicate MLL-GAS7 interactor record carries the same SFPQ row under the same interaction ac, so it adds no evidence for SFPQ

• ← analogous toNUMA1 / NuMA (Q14980) - a nuclear-matrix protein that becomes the spindle pole, and the 5' partner of a variant APL fusion whose coiled-coil is required for oncogenicity

  third interactome-layer protein in this graph that is also a fusion partner in its own right, after SFPQ-ABL1 and NPL-DHX9 - and the first whose fusion has structure-function data and animal models rather than a case report

• supports →IntAct holds 51 interaction records whose participant IS a KMT2A fusion across 6 constructs, our coverage instrument dropped every one of them silently, and closing the single gap they contained took the partner list to 21 of 21

  SFPQ was the one gap of 21 fusion-construct partners; this entry closes it

• binds →KMT2A::GAS7 (MLL-GAS7), t(11;17)(q23;p13) — three patients, and the best mouse model in the partner class

  anti-tag co-IP with the MLL-GAS7 construct in HEK293, IntAct interaction type 'association' (the weaker of the two terms in this enumeration), PMID 26766589

• co-occurs with →PRMT1 — an MLL-fusion cofactor whose first-in-human inhibitor was terminated for lack of efficacy and toxicity

  recovered on the same MLL-GAS7 construct in the same publication; PRMT1 had an entry, SFPQ did not

• co-occurs with →KHDRBS1 (Sam68) — an RNA-binding protein that holds the MLL-AF4 message DOWN, a class no dropout screen can find

  the third partner of the same MLL-GAS7 co-IP rows (Q07666)

• related to →Paraspeckles and NEAT1 — no KMT2A evidence at all, but a fusion oncoprotein in another leukemia uses NEAT1 as a differentiation brake

  first physical link in this graph between a paraspeckle/DBHS protein and a KMT2A fusion construct -- explicitly NOT a refutation: the compartment was never assayed

• part of →HEXIM1 - the brake pad, the only part of the reservoir anyone has deliberately raised in a KMT2A-rearranged cell, and the result was that the leukaemia grew less

  SFPQ is a component of the HEXIM1-NEAT1 HDP-RNP with NONO, PSPC1, RBM14 and MATRIN3 (PMID 28712728)

• fuses with →ABL1 — domain architecture and residue coordinates (with the 1a/1b numbering trap)

  SFPQ exon 9 to ABL1 exon 4 in every reported case; the intron-4 break leaves the chimaera without the ABL1 SH2 domain

• co-occurs with →IKZF1 - the defining co-lesion of every other high-risk B-ALL, and this disease does not use it until after treatment

  IKZF1 deletions are recurrent in SFPQ-ABL1-positive BCP-ALL

• analogous to →A ten-residue export signal is the 5' partner's whole contribution - and the field disagrees about which direction it pushes

  both are 5-prime partners that set the fusion address -- but in OPPOSITE directions: SFPQ-ABL1 is nuclear, CALM-AF10 requires nuclear export

• ← analogous toDHX9 / RNA helicase A (Q08211) - the other TDRD3 effector, and it is itself the 3' partner of a leukaemia fusion

  both are interactome-layer RNA-binding proteins that are ALSO fusion partners in their own right - SFPQ-ABL1 in B-ALL, NPL-DHX9 in T-LGLL (PMID 35600388); DHX9 is the second instance of the shape in this graph