AFF4 — the textbook SEC scaffold that KMT2A::AFF1 cells need LESS than other leukemias

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.

• Compensation is the whole story here, and it cuts against us twice. AFF1, AFF2, AFF3 and AFF4 share the CHD oligomerisation module, so a knockout phenotype can be buffered by a paralogue — and in KMT2A::AFF1 the fusion oncoprotein is itself an AFF-family scaffold, which is the most likely reason the knockout is silent. Any AFF4-directed agent would be buffered the same way.
• Essential-function toxicity without a matching efficacy rationale: gnomAD LOEUF 0.166 and pLI 1.0 place AFF4 among the most LoF-constrained genes in this interactome, and HPA finds it in every tissue. A tissue-wide AFF4 inhibitor risks a developmental or broad-tissue phenotype while buying no selective effect in the target genotype.
• The SEC that AFF4 nucleates serves physiologic MLL-dependent transcription as well as oncogenic transcription (PMID 20153263). Blocking the scaffold does not distinguish the two, so on-target normal-haematopoiesis suppression is expected rather than avoidable.
• Delivery and modality: all three AFF4 modules are short segments on a 1,163-residue, largely disordered chain. The ELLBow (301-351) and the P-TEFb-binding N-terminus (2-73) are extended peptide-in-groove interfaces, the hardest PPI class to drug with small molecules; a degrader would remove the whole protein and therefore inherit the full constraint liability.
• Resistance would be trivially easy. A target whose knockout already has no fitness cost in the relevant genotype cannot impose selective pressure — there is nothing for the tumour to evolve around because it was never dependent.

1. Is endogenous AFF4 actually present in the KMT2A::AFF1 complex at stoichiometric occupancy, or is it largely displaced by the AFF1 module the fusion carries? The pan-MLL-chimera proteomics behind the 'shared subunit' claim cannot distinguish these, and no KMT2A::AFF1-specific interactome paper exists that we could verify.
2. Do AFF-family C-terminal homology domains hetero-oligomerise, and specifically does the AFF1 CHD within KMT2A::AFF1 bind the AFF4 CHD? Three independent structures of the isolated AFF4 CHD exist (6K7P, 6KN5, 6R80) and not one CHD-CHD complex. This is a solvable structure and it would decide whether AFF4 is a competitor, a partner or a bystander for this fusion.
3. Why do six of ten KMT2A-rearranged lines score positive on AFF4 knockout while MOLM-14 and MOLM-13 score mildly negative? If that split tracks with fusion partner (AFF1-fused lines needing AFF4 least), it is the paralogue-substitution hypothesis in the data; with n=4 KMT2A::AFF1 lines we cannot test it.
4. Can AFF4's near-maximal human LoF constraint (LOEUF 0.166) be reconciled with its dispensability in 1,208 cancer lines by anything other than a non-proliferative, developmental function? If the answer is no, AFF4 is a toxicity liability with no efficacy rationale.
5. Does AFF4 loss induce compensatory upregulation of AFF1, AFF2 or AFF3 in the knockout lines? A DepMap knockout with no phenotype in a constrained gene is exactly the signature of paralogue buffering, and nobody has measured the transcriptional response.
6. Where is AFF4 in a KMT2A-rearranged nucleus? The SEC scaffold's compartment has never been imaged in any leukaemic cell, and the speckle result that the field assumes carries over was obtained in FRAXE fibroblast biology with AFF1 - the partner that is actually fused - untested.
7. Does AFF4 relocalise under menin inhibition? If the fusion is evicted from chromatin, the SEC it recruits has to go somewhere, and AFF4 is the tractable component to follow.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   AFF4 is not DepMap common-essential: mean Chronos across 1,208 lines is -0.056 with 0.4% of lines depleted, -0.085 across 64 leukemia lines, +0.026 across 10 KMT2A-rearranged lines and +0.024 across 4 KMT2A::AFF1 lines, giving selectivity_vs_leukemia of +0.111 and selectivity_vs_all_lines of +0.082; gnomAD gives pLI 1.0, LOEUF 0.166, 13 observed versus 124.8 expected loss-of-function variants and missense z 3.55; HPA classifies it as low tissue specificity, detected in all tissues, low immune cell specificity.

   DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API; loss-of-function constraint via the gnomAD GraphQL API; normal-tissue expression via the Human Protein Atlas API · Broad Institute DepMap; gnomAD; Human Protein Atlas; Open Targets Platform · Databases 2026
   ▸What this evidence cannot show

   Means were computed by us from per-line API values rather than read from a DepMap release table, and the API does not state the release version. Only 10 KMT2A-rearranged lines (4 of them KMT2A::AFF1) contribute to the selective score, so with a selectivity of +0.111 the effect is inside our own declared noise band of 0.15 and should be read as 'no selective requirement detected', not as 'protective'. Cell lines are not normal tissue, and whole-gene knockout of a three-module scaffold is not the same experiment as inhibiting one of its interfaces. gnomAD LOEUF measures heterozygote selection across human development, which is a different question from cell-autonomous viability.

   ▸Evidence that argues against this (1)

   • gnomAD points the opposite way from DepMap: LOEUF 0.166 with pLI 1.0 is among the tightest constraint in this interactome, so human heterozygous loss of AFF4 is strongly selected against even though homozygous knockout is tolerated by cancer lines in culture. Any AFF4-directed agent therefore carries a developmental/tissue-toxicity question that the DepMap number does not address. gnomAD Consortium 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   AFF4 is a component of the ELL/P-TEFb elongation complex and a shared subunit of MLL chimeras; AFF4 is required for the stability of that complex, and AFF4 knockdown reduces expression of MLL-chimera target genes, linking transcription elongation to leukemia.

   AFF4, a component of the ELL/P-TEFb elongation complex and a shared subunit of MLL chimeras, can link transcription elongation to leukemia · Lin C, Smith ER, Takahashi H, et al. · Molecular Cell 2010
   PMID 20159561DOI
   ▸What this evidence cannot show

   The proteomics is pan-MLL-chimera, not KMT2A::AFF1-specific; we could not verify any KMT2A::AFF1-specific published interactome paper. The functional requirement is established by RNAi knockdown with target-gene expression as the readout over days, not by long-term knockout with a proliferation readout, which is what DepMap measures. Nothing in this work excludes the endogenous AFF4 scaffold being substituted by the AFF1 module carried within the fusion oncoprotein itself.

   ▸Evidence that argues against this (1)

   • DepMap CRISPR knockout across 10 KMT2A-rearranged lines finds no selective requirement for AFF4 (mean Chronos +0.026, selectivity vs leukemia +0.111), with 6 of 10 lines scoring positive. The two results are reconcilable only if AFF4 is required for the magnitude of chimera target-gene transcription without being rate-limiting for proliferation, or if the fusion's own AFF-family scaffold substitutes for it. Broad Institute DepMap; Open Targets Platform 2026
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   A higher-order complex containing AF4-family and ENL-family proteins together with P-TEFb facilitates both oncogenic and physiologic MLL-dependent transcription.

   A higher-order complex containing AF4 and ENL family proteins with P-TEFb facilitates oncogenic and physiologic MLL-dependent transcription · Yokoyama A, Lin M, Naresh A, Kitabayashi I, Cleary ML · Cancer Cell 2010
   PMID 20153263DOI
   ▸What this evidence cannot show

   This establishes that AF4-family and ENL-family proteins co-assemble with P-TEFb in a higher-order complex; it does not establish the stoichiometry of AFF4 versus AFF1 within a KMT2A::AFF1 complex, and it reports no affinity measurement. The statement that the same complex serves physiologic MLL-dependent transcription is itself the toxicity warning: the assembly is not fusion-specific.

   ▸Evidence that argues against this (1)

   • Complex composition is not requirement. The AEP/SEC papers establish what co-purifies and what a reconstituted complex does; they do not establish that each subunit is rate-limiting in a patient's leukaemia, and this project's own dependency data separates the members by an order of magnitude. Nature 2012
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   AFF4 has five verified PDB entries spanning three structurally independent modules: the N-terminal P-TEFb-binding segment (4IMY, X-ray 2.94 A, CDK9 1-330 + CCNT1 1-264 + AFF4 2-73), the central ELLBow bound to the ELL2 occludin-like domain (5JW9, X-ray 2.00 A, ELL2 519-640 + AFF4 301-351), and the C-terminal homology domain solved three times independently (6K7P X-ray 2.40 A AFF4 902-1163; 6KN5 X-ray 2.20 A AFF4 899-1163; 6R80 X-ray 2.20 A AFF4 899-1163). No structure exists of an AFF4 CHD bound to another AFF-family CHD, and no structure of AFF4 with ELL (as opposed to ELL2) exists.

   RCSB PDB entries 4IMY, 5JW9, 6K7P, 6KN5 and 6R80, with per-entity UniProt accessions, methods, resolutions and modelled residue ranges resolved against the RCSB APIs · RCSB Protein Data Bank · RCSB PDB 2026
   ▸What this evidence cannot show

   Residue ranges are the modelled segments as RCSB annotates them against UniProt, not independently curated domain boundaries. All five are fragment structures: none contains more than one AFF4 module, so the relative arrangement of the P-TEFb-binding N-terminus, the ELLBow and the CHD within one AFF4 chain is unknown. The absence of a CHD-CHD complex structure is an absence of evidence in a searched database, not proof that the interaction does not occur.

   ▸Evidence that argues against this (1)

   • Counting PDB entries measures how tractable a protein was to CRYSTALLISE, which correlates with being small, ordered and well-behaved rather than with being druggable. What decides design value is whether the deposited density covers the site that matters, at a resolution that supports modelling, in the relevant state — and this graph has repeatedly found structures described in the direction that overstates what is resolved. Three structurally independent modules with nothing joining them is the informative shape: AFF4 is largely disordered scaffold, so the resolved fragments are the parts that fold when handed a partner, and the protein's behaviour as a whole is not captured by any of them. RCSB PDB 2026
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   ChEMBL v34 contains almost no chemical matter against AFF4: the single-protein target CHEMBL5724734 has 5 activity rows at pChEMBL >= 6 with no curated mechanism-of-action drug; the best row (CHEMBL5653589, Kd 12.24 nM) is an unnamed kinobead-pulldown compound and the remainder are the BET inhibitor molibresib (CHEMBL1232461), whose apparent nanomolar affinities for AFF1, AFF4, ELL, PAF1, SETD1A and ASH2L all derive from one chemoproteomics document and are super-elongation-complex co-enrichment, not direct binding.

   ChEMBL database version 34, target CHEMBL5724734 (AF4/FMR2 family member 4): target_search, get_bioactivity and get_mechanism queries · EMBL-EBI ChEMBL · ChEMBL 2026
   ▸What this evidence cannot show

   Distinct-compound counts are page-local because of a 25-row API limit, so 'best' is the best within one sampled page rather than a potency-ranked view of the whole target; for a 5-row target this limitation is immaterial. Absence from ChEMBL is not absence from the world, and curation lags the primary literature. The molibresib rows are not wrong as measurements, only as attributions: the compound genuinely enriches AFF4, but through BRD4.

   ▸Evidence that argues against this (1)

   • A ChEMBL activity-row count measures how much medicinal chemistry has been DONE, not how tractable the target is, and the best reported affinity belongs to the most optimised compound in whichever series someone chose to pursue. Zero rows means nobody tried; thousands of rows means a target had a decade of industrial attention. Five rows with no curated mechanism is the profile of a protein nobody has run a campaign against. ChEMBL, EMBL-EBI 2026
6. E_q72_aff4_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   UNIPROT GIVES AFF4 TWO LOCATIONS ON ONE PAPER WHOSE ABSTRACT CONTAINS NO LOCALISATION EXPERIMENT. Fetched from UniProtKB 2026-09-21, Q9UHB7 (AFF4_HUMAN, primary gene AFF4, 1,163 aa) carries two SUBCELLULAR LOCATION values, 'Nucleus' and 'Chromosome', both with a single ECO:0000269 attribution to PMID 12065898 - the 2002 paper that identified MCEF/AFF4 as a P-TEFb-associated protein by affinity purification, cDNA cloning, tissue-expression RT-PCR and immunoprecipitation. No nuclear-localization-signal motif is annotated. No subnuclear value is annotated, in particular no nuclear speckle, although a speckle result for AFF4 exists [E_q72_aff4_loc2].

   UniProtKB entry Q9UHB7 (AFF4_HUMAN), SUBCELLULAR LOCATION comment and feature table · UniProt Consortium · UniProtKB 2026
   ▸What this evidence cannot show

   Abstract of PMID 12065898 read in full; FULL TEXT NOT READ, so a localisation panel inside the paper cannot be excluded and this is explicitly NOT a claim that the annotation is wrong - brief rule 80 says an index that covers abstracts cannot bound a claim about a whole document. What can be said is bounded and still useful: the ECO:0000269 code tells a reader an experiment was read, and a reader who then goes to the abstract to find out WHICH experiment finds none. For a 'Chromosome' assignment - which a downstream tool will read as chromatin association - that gap matters.

   ▸Evidence that argues against this (1)

   • A 2011 paper does report AFF4 localisation directly and UniProt does not cite it: AFF3/LAF4 and AFF4/AF5q31 localise to NUCLEAR SPECKLES and bind RNA with high apparent affinity for G-quadruplex structures. So the database's location set is both under-attributed and incomplete in the same field. Melko M et al. 2011
7. E_q72_aff4_loc2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   IN NON-HAEMATOPOIETIC HUMAN CELLS, AFF4/AF5q31 LOCALISES TO NUCLEAR SPECKLES - the subnuclear stores of pre-mRNA splicing factors - together with AFF3/LAF4 and (previously shown) AFF2/FMR2; all three bind RNA with high apparent affinity for the G-quadruplex structure and modulate in vivo splicing efficiency of a minigene containing a G-quadruplex in an alternatively spliced exon; and OVEREXPRESSION OF AFF2/3/4 INTERFERES WITH THE ORGANISATION AND/OR BIOGENESIS OF NUCLEAR SPECKLES themselves. AFF1 was not among the proteins shown to localise there.

   Functional characterization of the AFF (AF4/FMR2) family of RNA-binding proteins: insights into the molecular pathology of FRAXE intellectual disability. · Melko M, Douguet D, Bensaid M, Zongaro S, Verheggen C, Gecz J, Bardoni B · Hum Mol Genet 2011
   PMID 21330300DOI
   ▸What this evidence cannot show

   Abstract read in full; full text not read. NOT A LEUKAEMIA SYSTEM, not haematopoietic, no KMT2A fusion anywhere in it - the paper's subject is FRAXE intellectual disability (brief rule 73). The localisation is shown in a transfection/overexpression context, and the paper's own finding that overexpressing these proteins DISTORTS speckles is a warning about reading the compartment off an overexpressing cell. Already cited four times in this graph; localization-nuclear-speckles-kmt2a-r is where the reasoning lives and this field should not restate it (rule 82).

   ▸Evidence that argues against this (1)

   • At physiological expression, low-complexity-domain interactions that control transcription form dynamic hubs WITHOUT detectable phase separation, and are differentially sensitive to hexanediols - so a punctate speckle assignment made in an overexpressing cell is weak evidence about endogenous AFF4, and the paper's own observation that overexpression enlarges and reorganises speckles makes the concern concrete rather than generic. Chong S et al. 2018
8. E_q72_aff4_loc3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   AFF4 IS THE ONLY AFF-FAMILY PROTEIN WITH A MEASURED COMPARTMENT-SWITCHING FUNCTION, AND IT IS IN HIV LATENCY, NOT LEUKAEMIA. Tat's transcriptional activity is oppositely regulated by NPM1-mediated NUCLEOLAR RETENTION and AFF4-INDUCED PHASE SEPARATION IN THE NUCLEOPLASM; PRMT2 methylation of Tat arginine 52 reinforces nucleolar sequestration while counteracting incorporation into SEC droplets, inactivating Tat and promoting latency. So AFF4 is not merely resident in a compartment - it is the component that BUILDS a nucleoplasmic compartment and pulls a cargo out of the nucleolus into it.

   PRMT2 promotes HIV-1 latency by preventing nucleolar exit and phase separation of Tat into the Super Elongation Complex. · Jin J, Bai H, Yan H, Deng T, Li T, Xiao R, Fan L, Bai X, Ning H, Liu Z, Zhang K, Wu X, Liang K, Ma P, Gao X, Hu D · Nat Commun 2023
   PMID 37949879DOI
   ▸What this evidence cannot show

   Abstract read in full; full text not read. HIV-1 LATENCY, not leukaemia, not a KMT2A fusion, and the cargo is a viral protein (brief rule 73) - this is a comparator, and this graph already carries it as localization-nucleolar-retention-versus-sec-droplets-tat-comparator. The phase-separation claim inherits the standing assay problem recorded in localization-condensate-evidence-standards: hexanediol sensitivity and classical FRAP cannot distinguish LLPS from binding to spatially clustered sites, and 1,6-hexanediol itself condenses chromatin in living cells. 'AFF4-induced phase separation' should be read as 'AFF4-dependent nucleoplasmic partitioning', which is what the data support.

   ▸Evidence that argues against this (1)

   • Nothing connects this to AFF4's day job in leukaemia. AFF4 is the SEC scaffold that KMT2A::AFF1 recruits, but no experiment has ever asked whether AFF4 changes compartment, or moves any cargo between compartments, in a KMT2A-rearranged cell. Importing the Tat result would be exactly the partner-and-system-imported inference this project's attribution audits keep finding. Melko M et al. 2011

• ← related toHypothesis — forced-oligomerisation interfaces are the mutation-resistant drug targets the fusion field has been ignoring

  AFF4-CHD/THD structure defines the AFF homo- and AFF1-AFF4 heterodimer interface (PMIDs 31147444, 32128251)

• ← bindsELL2 — domain architecture and residue coordinates

  ELL2 OCEL domain (526-636) binds the AFF4 ELLBow; 2.0 A crystal structure PDB 5JW9

• ← bindsAFF1 / AF4 — domain architecture and residue coordinates

  CHD-mediated AFF1-AFF4 heterodimerisation (shown for the AFF4 CHD, PDB 6R80); untested in the fusion

• ← part ofAFF4 has never been placed in a haematopoietic cell state — every SEC cell-state result is in another tissue

  the bounded cell-state absence for this gene

• ← related toHypothesis — forced-oligomerisation interfaces are the mutation-resistant drug targets the fusion field has been ignoring

  AFF4 CHD dimerisation is a normal SEC function (Guo 2020, PMID 32270036), not a fusion-supplied neomorphic surface, so selectivity would have to come from dependency rather than from the interface itself.

• ← related toKMT2A::ELL (MLL-ELL), t(11;19)(q23;p13.1)

  The scaffold, not the partner, is limiting (Lin 2010, PMID 20159561): AFF4 knockdown reduces KMT2A-chimera target gene expression, so ELL's in vitro elongation activity is not by itself the causal account.

• ← related toKMT2A::AFDN (MLL-AF6), t(6;11)(q27;q23)

  Dimerisation and SEC recruitment describe different partner sets (Lin 2010, PMID 20159561): KMT2A chimeras co-purify ELL/P-TEFb/AFF4 and AFF4 knockdown reduces target-gene expression, so neither account has been shown to cover both.

• ← bindsELL2 — a SEC subunit that almost no cell needs, next to a paralogue almost every cell needs

  OCEL domain binds the AFF4 scaffold; crystal structure 5JW9 at 2.0 A

• ← recruitsELL3 — the third ELL paralogue, which marks enhancers and recruits SEC rather than sitting inside it

  ELL3 enhancer occupancy is required for SEC recruitment after differentiation signals

• ← part ofEAF1 — the interactor with an actual subnuclear phenotype: an MLL fusion evicts it from Cajal bodies, and the bodies themselves fall apart

  AFF4-ELL2-EAF1 is the elongation module of ECAS

• ← bindsThe AFF-family N-terminus on cyclin T1: Kd 10 nM, a 2.94 A crystal, and every number in it measured on AFF4 rather than on AFF1

  AFF4 2-73 on P-TEFb, Kd 10 nM

• ← bindsThe ELLbow: AFF4 301-351 wrapped over the ELL2 occludin domain at 2.0 A, Kd 86 nM - and no structure of ELL itself, the partner in t(11;19)

  AFF4 ELLbow 301-351, Kd 86 nM

• ← bindsAFF1-AFF4 heterodimerisation through the C-terminal homology domain: the one SEC interface KMT2A::AFF1 certainly keeps, defined by a 2.2 A structure of the wrong subunit

  AFF4 CHD 899-1163, alpha5/alpha6 dimerisation interface

• ← bindsDAXX - the strongest of five selective dependencies among 254 unwritten interactors but it FAILS Bonferroni correction (p=0.0128, rank 4 of 254), and not one paper connects it to this disease

  one of four baits recovering DAXX, all on one platform

• part of →Mechanism class 1 — chromatin and elongation hijack (KMT2A-r, NUP98-r)

  the textbook scaffold, which these cells need less of than expected

• ← part ofAQR (Aquarius) - a spliceosomal RNA helicase whose loss impairs MLL-rearranged leukaemia stem cells

  AQR sits in the same ECAS assembly as the AFF4-ELL2-EAF1 elongation module; if ECAS is distinct from SEC, AFF4 membership alone stops being evidence that SEC is the relevant complex.

• ← related toThe case that the KMT2A-fusion condensate is an artefact — what 1,6-hexanediol, overexpressed tags and in-vitro droplets actually prove

  the only AFF4-adjacent condensate report rests on aliphatic alcohol, which inactivates CDK9

• ← related toThe strongest published case that condensates are load-bearing in KMT2A-fusion leukemia — and it is AKAP95, not AFF1

  AFF4 fluidises SEC droplets; AFF1 is the untested paralogue in t(4;11)

• ← related toHypothesis: KMT2A::AFF1 does not require phase separation — recruitment biochemistry is sufficient, and the condensate model does not transfer from NUP98/ENL

  Counter-evidence (Guo 2020, PMID 32270036): AFF4 fluidises ENL-initiated SEC droplets and MLL-ENL fusion boosts SEC phase separation. The entry's 'zero reports of AFF-family condensates' premise is withdrawn.

• ← related toPlain-English: what transcription elongation is, and why a traffic light explains this leukemia

  Open dispute inside the elongation model (Guo 2020, PMID 32270036): SEC may concentrate P-TEFb by phase separation rather than acting as a purely stoichiometric complex; AFF4 fluidises ENL-initiated droplets.

• ← related toAFF1 / AF4 — the 3' half of the fusion, and a target with almost no structural footing

  Counter-evidence (Xiang 2026, PMID 42748909): AFF4-ELL2-EAF1 occurs in ECAS, reported as molecularly distinct from canonical P-TEFb-containing SEC. Presence in a 2004-2012 affinity purification underdetermines which assembly is meant, so the AFF1-versus-AFF4 DepMap asymmetry may be between assemblies rather than paralogues.

• ← bindsKMT2A::MLLT1 engages P-TEFb through one 66-residue surface: the ENL AHD, residues 494-559, and L550E breaks it

  AF5q31/AFF4 is the member mapped to ENL 494-559 in the transformation assay

• ← regulatesENL's disordered region and condensates, restricted to what actually bears on KMT2A::MLLT1 - which is one figure panel

  AFF4 fluidises ENL/P-TEFb droplets; shRNA or KL-1 makes them irregular and non-coalescing

• ← part ofKMT2A::AFF1 has no mapped interface to DOT1L: it lacks both structurally defined recruitment modules, and the one complex-level route claimed for it is contested by reconstitution

  the AFF1/AFF4 SEC is the complex-level route claimed for DOT1L, and is contested

• ← related toThe leukaemia that changed lineage in eleven days and changed back when the drug stopped: KMT2A::AFF4 infant ALL on blinatumomab

  KMT2A::AFF4 fusion from an atypical ins(5;11); no KMT2A::AFF4 fusion node exists in this graph

• ← regulatesP-TEFb phosphorylates AFF4 and rearranges the complex the fusion hijacks

  P-TEFb phosphorylates AFF4, modulating SEC assembly and AFF1/AFF4 dimerisation

• ← related toAFF1's modern proximity neighbourhood is four RNA-binding proteins - NPM1, RBM15, SBDS, ZRANB2 - and it barely overlaps AFF4's

  AFF4's four-bait neighbourhood shares only ZRANB2 with AFF1's

• ← co-occurs withZRANB2 - the only bait in any published proximity experiment that reaches AFF1, AFF4, ELL, MLLT1, CCNT1, KMT2A and DOT1L at once

  AFF4 recovered in the same neighbourhood - the only bait reaching both scaffolds

• ← co-occurs withGPKOW - a G-patch spliceosomal protein whose BioID neighbourhood contains AFF4, ELL, ENL, CDK9, KMT2A and DOT1L

  AFF4 recovered as prey

• ← co-occurs withPPIL4 - a spliceosomal cyclophilin whose BioID neighbourhood is the widest overlap with this graph's subject matter of any bait

  AFF4 recovered as prey

• ← co-occurs withHNRNPC - the nascent-RNA packaging protein whose BioID neighbourhood contains AFF4, ELL, ENL, CDK9, cyclin T1 and KMT2A

  AFF4 recovered as prey

• ← related toAFF1- and AFF4-containing super elongation complexes occupy physically distinct neighbourhoods in living cells, and KMT2A::AFF1 inherits the AFF1 one

  AFF4 scaffold neighbourhood

• ← depends onHEXIM1 - 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

  HEXIM1 induction after P-TEFb release requires the SEC subunits AFF4 and ELL2

• ← depends onThe fusion begins at AFF1 residue 347 and the cyclin T1-binding sequence ends at 308, so KMT2A::AFF1 is the one AFF-family protein that cannot enter the reservoir

  in-trans access to P-TEFb through the AFF-family CHD heterodimer

• ← co-occurs withRSBN1 / KDM9 (Q5VWQ0) - an H4K20me2/me3 demethylase in the AFF4 proximity record, and the only read-queue partner that is a chromatin enzyme at all

  two BioPlex rows (28514442, 33961781) - one screening programme seen twice; RSBN1 is the BAIT and AFF4 the prey in both; HEK293T; IntAct MI-score 0.53

• ← co-occurs withPES1 / pescadillo (O00541) - the missing PeBoW subunit of this graph's nucleolar block, pan-essential in 96.9% of DepMap lines, and the one myeloid experiment on it shows proliferation WITHOUT a differentiation block

  two BioPlex rows (28514442, 33961781) - one screening programme seen twice; PES1 is the BAIT and AFF4 the prey in both; HEK293T; IntAct MI-score 0.53, the joint-highest in the read queue and still below IntAct's 0.6 high-confidence cut

• ← related toIn all 24 read-queue rows inspected, the candidate is the BAIT and this project's protein is the prey - nobody has ever tagged KMT2A, AFF4, menin or PSIP1 and found any of them; plus seven bounded rejections

  AFF4 is the scaffold in 16 of the 30 read-queue rows and is the PREY in every one of them - no AFF4-directed experiment has recovered any read-queue member

• ← related toRe-bound at full text: AFF4 still has no haematopoietic cell-state result - the loose intersection is 265 records and the tight one 11, and every title read is a review, a complex-membership mention or a congress abstract book

  no full-text record places this protein in a haematopoietic cell state; the recall is its complex's literature

• ← testsThree hours after the fusion is degraded to 2.5-7.1% of its own sites, NOT ONE of the eight complex components profiled beside it falls below 68% retention genome-wide - menin 80.0%, DOT1L 68.7%, CDK9 83.6%, AFF4 92.7%, RNA Pol II 99.6%, pS2 98.3%, pS5 86.8% - so nothing in the complex is held on chromatin by the fusion at genome scale

  AFF4 retains 92.7% of its strongest peak calls genome-wide 3 h after fusion degradation - loss relative to the fusion's own 0.08. The SEC scaffold's genome-wide residence does not require the fusion, though the authors report AFF4 loss restricted to fusion target promoters.

• ← related toThe order in which the complex leaves fusion-occupied promoters, all fifteen channels on one scale

  AFF4 barely moves at fusion target promoters (0.821) while its SEC partner CDK9 moves to 0.687

• ← related toNo HA library exists in the AFF4 sequencing batch, so AFF4 is excluded from the peak partition by name

  AFF4 moves least of any complex channel at fusion promoters (0.821) and cannot be partitioned on peaks

• ← related toPartner-protein occupancy is not fusion occupancy: every KMT2A::AFF1 chromatin claim resting on an AFF1, AFF4 or SEC-subunit ChIP is an upper bound that includes free complex

  the same confound applies to AFF4 readouts

• ← bindsAFF4-MED10: AFF4 is prey in a 2004 Mediator MudPIT purification - a lead with an independent mechanistic reason to be true, and no experiment that tests it

  AFF4 prey in a MED10 Mediator MudPIT preparation, HeLa, miscore 0.35

• ← bindsAFF4-FMR1 - DECLINED, and the provenance is the finding: the 'publication' behind this IntAct row is HENA, an Alzheimer's data-integration dataset, not an experiment on AFF4

  single yeast two-hybrid fragment-pooling row citing a data-integration resource; DECLINED

• ← bindsAFF4-SDCBP (syntenin): the only prey-only row in this census typed 'direct interaction' - a proteomic phage-display hit on AFF4's C-terminal peptide, in vitro

  AFF4 C-terminal peptide binds syntenin PDZ1-2 on phage, in vitro, MI:0407

• ← bindsAFF4-CDK15: AFF4 is prey in a systematic CMGC-kinase TAP interactome - a kinase adjacent to the complex CDK9 works in, and nobody has followed it up

  AFF4 prey in a CDK15 tandem affinity purification, HEK293

• ← bindsAFF4-BTK - DECLINED. One osteosarcoma AP-MS row, and it is the only partner in the uncovered prey-only tail with an approved drug against it, which is exactly why the decline is written down

  single U2OS AP-MS row; DECLINED - BTK is not expressed in osteosarcoma

• ← bindsAFF1/AFF4-CAMKV - DECLINED, and it is a LABELLED REDISCOVERY: data/symbol-collisions.json already holds a CAMKV row saying BioPlex 2017 and 2021 are one programme and that CAMKV is deposited under split isoform accessions

  prey in AFF4 pull-downs across BioPlex 2.0, 3.0 and the Schaffer U2OS map; DECLINED

• part of →SEC — the AFF1/AFF4-scaffolded super elongation complex, the particle KMT2A::AFF1 is a permanent subunit of

  AFF4 is a scaffold subunit and is required for SEC stability and formation

• ← related toThe SEC-family target sets, reconstructed as a full three-way Venn: 27.5% of the family's transcriptional output is invisible to AFF4 depletion — and the two 'distinct' sibling sets overlap 8-20x more than chance

  AFF4 knockdown response set, 966 of 1,116 unique to it

• ← related toThree of the four AFF scaffolds have a named human Mendelian syndrome and the index fusion's partner has none — a 0/4 that is a fact about AFF1 and not about how hard anyone has looked

  AFF4 -> CHOPS syndrome, germline gain-of-function

• ← refutesAFF4 has never been placed in a haematopoietic cell state — every SEC cell-state result is in another tissue

  AFF4/SEC recruitment to MYC regulates its expression in myeloid AND lymphoid leukaemia and is proposed as a therapeutic target, while the same work isolates SEC-L2 and SEC-L3 carrying AFF2 and AFF3 with the same P-TEFb/ENL/AF9 partners (PMID 22547686) - so the +0.024 null is a redundancy measurement, not an absence of requirement

• related to →Nuclear speckles — the compartment everyone assumes KMT2A::AFF1 uses, on evidence that skips AFF1 itself

  AFF4 is one of the three AFF proteins actually shown in nuclear speckles

• related to →A two-compartment switch for a SEC cargo: nucleolar retention versus AFF4-induced nucleoplasmic droplets, set by one methylated arginine

  AFF4 builds the nucleoplasmic SEC compartment that pulls Tat out of the nucleolus

• depends on →The case that the KMT2A-fusion condensate is an artefact — what 1,6-hexanediol, overexpressed tags and in-vitro droplets actually prove

  the AFF4 phase-separation claim inherits the assay limits recorded there

• related to →AFF4 has never been placed in a haematopoietic cell state — every SEC cell-state result is in another tissue

  the localisation gap matches the cell-state gap: no AFF4 experiment in any haematopoietic system

• ← bindsPBX2 - an AFF4 co-purifying protein in two rounds of one AP-MS atlas, and the PBX paralogue the leukaemia literature explicitly excluded

  co-purifies with tagged AFF4 in BioPlex 2.0 and 3.0 (HEK293T, anti-tag co-IP, spoke-expanded); not a measured binary contact

• ← related toThe four AFF4 two-hybrid partners are three rows of one screen, not three experiments - and in that same screen GOLGA2 has 249 partners while AFF4 has five

  AFF4's five two-hybrid partners in Rolland 2014, and the prey degrees that make four of them unusable

• ← bindsAP2B1 (beta-2-adaptin) - the only one of AFF4's four two-hybrid partners that is not a hub, and the only one with an echo in this graph's endocytosis biology

  yeast two-hybrid hit in Rolland 2014; one screen, never replicated, but the only one of the four preys that is not a hub

• ← bindsGOLGA2 (GM130) - a Golgi matrix protein with 249 partners in the one yeast screen that reported it binding AFF4, which had five

  yeast two-hybrid hit in Rolland 2014; prey degree 249 in the same screen, cis-Golgi protein, not usable

• ← bindsMTUS2 (TIP150) - a microtubule plus-end tracking protein with 153 partners in the one yeast screen that reported it binding AFF4

  yeast two-hybrid hit in Rolland 2014; prey degree 153 in the same screen, cytoskeletal protein, not usable

• ← bindsTRAF2 - a cytoplasmic NF-kB brake reported as an AFF4 two-hybrid hit, 97 partners in the screen that found it, and the live instance of the brief's own accession trap

  yeast two-hybrid hit in Rolland 2014; prey degree 97 in the same screen, cytoplasmic, not usable

• ← bindsJMY - one AP-MS row with AFF4, in HCT116 only, from a paper whose whole argument is that interactomes differ between cell lines

  single BioPlex 3.0 AP-MS row, HCT116 arm only, spoke-expanded; absent from the 293T arm of the same study

• ← bindsLRRK2 - the AFF4 row is one of 78 hits on an in-vitro protein array, but the protein has a real AML literature that arrived through USP7 and interferon, not through the interactome

  one in-vitro protein-array row (host taxid:-1), one of 78 array hits for LRRK2 in a ROCO-family study

• ← related toThe index fusion's 3' partner has ONE observed binary interaction in IntAct and forty-two residues of solved structure - every SEC contact attributed to AFF1 is an AFF4 result wearing AFF1's name

  the paralogue that carries the SEC's structural evidence: nine PDB entries, five of them against CDK9 and cyclin T1

• ← related toAFF1 reaches P-TEFb only through AFF4, and the index fusion's elongation phenotype is therefore a claim about AFF4 availability

  if true, AFF4 availability rather than AFF1 abundance sets elongation output at fusion targets

• ← bindsAR / androgen receptor - the best-supported partner in the absence set by independent-publication count, and the reason a menin inhibitor is in prostate cancer trials

  one spoke-expanded BioID proximity row in LAPC4; no corroboration

• ← bindsMED19 - the Mediator subunit curated against an engineered MLL-AFF4 fusion protein, and the one REST uses to deliver G9a

  MED19-bait anti-tag co-IP, MI:0914 association, HeLa (IntAct EBI-394580, PMID 15175163)

• ← bindsPIP4K2A - the lipid kinase at a genome-wide childhood-ALL susceptibility locus, whose product is the ligand a PHD finger reads

  PIP4K2A-bait AP-MS in HEK293T (BioPlex 2.0/3.0) and U2OS (PMID 40205054) - two platforms

• ← bindsEPB41L3 / DAL-1 - a membrane-skeleton tumour suppressor that binds and modulates two arginine methyltransferases, and whose own promoter methylation is a clinical test

  EPB41L3-bait AP-MS, BioPlex 3.0 HEK293T; weakest arm of the entry - no nuclear pool reported

• ← related toA two-compartment switch for a SEC cargo: nucleolar retention versus AFF4-induced nucleoplasmic droplets, set by one methylated arginine

  AFF4 is the SEC component that builds the active-state droplets in this system