AFF3 (also called LAF-4) is a 1226-residue, almost entirely intrinsically disordered nuclear protein of the four-member AF4/FMR2 (ALF) family, and the scaffolding subunit of super elongation complex-like 3 (SEC-L3), which it forms together with the P-TEFb kinase module (CDK9 and cyclin T1) and the ENL/AF9 proteins. Like its paralogues AFF1 and AFF4 it holds an elongation-competent assembly together on RNA polymerase II, and the purified complex carries high polymerase II C-terminal-domain kinase activity. Where AFF3 differs from those paralogues is in how it reaches chromatin. It has no DNA sequence preference of its own - the isolated protein binds double-stranded DNA indiscriminately in vitro, while a separate internal segment activates transcription strongly when tethered by a heterologous DNA-binding domain - and it is instead delivered to particular regulatory elements by sequence-specific zinc-finger proteins, ZFP281 to the Meg3 enhancer of the imprinted Dlk1-Dio3 locus and ZFP57 to the same locus's intergenic differentially methylated region. In human cells it binds the methylated CpG island downstream of the XIST promoter in a DNA-methylation-dependent manner and keeps XIST expression monoallelic. It concentrates in nuclear speckles, the compartment in which CDK9 and cyclin T1 also form foci. In B cells it binds immunoglobulin switch-region DNA through its C-terminal region and promotes recruitment of the deaminase AID, biasing class switch recombination toward the IgG isotypes. Expression is highest in lymphoid tissue and, during development, in the brain, somites and limb buds; loss of the gene in mouse impairs migration of cortical neurons, and mouse null animals additionally show skeletal, kidney and brain malformations. AFF3 protein abundance is set by a short degron that directs ubiquitin-dependent degradation, and the gene is unusually dosage-sensitive in both directions: heterozygous deletion, or promoter GCC-repeat expansion with hypermethylation and silencing, and stabilising degron variants all produce overlapping human syndromes featuring intellectual disability, seizures, mesomelic dysplasia of the lower limbs and horseshoe kidney. The gene is also a recurrent fusion partner of KMT2A/MLL and of RUNX1 in acute lymphoblastic leukaemia.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0006355 regulation of DNA-templated transcription | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted as core. This is the best-founded of AFF3's five IBA rows, and the only one whose WITH/FROM includes the gene's genuine mouse orthologue - mouse Aff3 holds this exact term by both IDA and IMP. Reason: All five tokens were resolved and each protein donor queried for its own evidence, so the usual objection that IBA donors carry nothing but the same family-level inference is testable here and is false - four of the four protein donors hold their own experimental annotation to this term or a descendant. MGI:MGI:106927 resolves to P51827, mouse Aff3, the true orthologue, which holds GO:0006355 by IDA and IMP from PMID:25162227. The remaining donors are paralogues, which is legitimate for an IBA - Drosophila lilli (Q9VQI9, the single fly AFF protein and therefore co-orthologue of all four human paralogues) by IMP and IGI, mouse Aff1 (O88573) by IDA to GO:0045893, human AFF1 (P51825) by IMP to GO:0032786 and GO:0032968. The row is kept at the unsigned parent rather than refined, and the reason is the recipient rather than the donors. Every signed donor term here is in fact POSITIVE - mouse Aff1's term IS GO:0045893, and human AFF1's GO:0032786 and GO:0032968 were each verified to sit under it - and GO:0045893 is itself a descendant of GO:0006355, so a positive child was available and unused. The earlier wording claimed all three were verified under GO:0045893, which was vacuous for GO:0045893 itself and untrue of GO:0032968, whose claim was missing from the guard until round 2 added it. An earlier draft of this reason argued instead that the donors disagreed in sign, misreading GO:0032786 as negative by proximity to GO:0032785; that was wrong and the ancestry guard refuted it. What actually forbids refining the row is that AFF3's own output runs in both directions - it represses XIST from the silent allele in two human cell lines, while with ZFP281 it establishes a permissive state at the Meg3 enhancer and its over-expression raises 84% of the transcripts it changes in mouse cortical cells. A positive-only term would be false for the repressive half, which is why the specific negative instance is proposed as a separate GO:0045892 row instead. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000829417 · PANTHER tree node covering the AF4/FMR2 family SUPPORTS TRANSFER A tree node, not a gene product, so it cannot be queried for its own evidence. It is the inference structure rather than an evidence source. FB:FBgn0041111 · Drosophila lilli (Q9VQI9, Swiss-Prot, 1673 aa) SUPPORTS TRANSFER Holds GO:0006355 by its own IMP and GO:0045944 by IMP and IGI. The sole AF4/FMR2 family member in Drosophila, hence co-orthologue of all four human paralogues. Three candidate accessions were returned for this FlyBase id and the reviewed one was used; the count is recorded rather than the first hit silently taken. MGI:MGI:106927 · mouse Aff3 (P51827, Swiss-Prot) - the true orthologue SUPPORTS TRANSFER Holds GO:0006355 by both IDA and IMP from PMID:25162227. Ten candidate accessions were returned for this MGI id and the single reviewed one was used. MGI:MGI:1100819 · mouse Aff1 (O88573, Swiss-Prot) SUPPORTS TRANSFER Holds the descendant GO:0045893 by IDA from PMID:9365243. A paralogue. Its term is POSITIVE regulation, as are human AFF1's GO:0032786 and GO:0032968, so the donor set agrees in sign - an earlier draft of this review asserted the opposite and was refuted by the ancestry guard. UniProtKB:P51825 · human AFF1 SUPPORTS TRANSFER Holds the descendants GO:0032786 and GO:0032968 by IMP. A paralogue of the recipient, not the recipient itself. Supporting Evidence: PMID:22547686 SEC-L2 and SEC-L3 control the expression of different subsets of genes PMID:30535390 Knockdown of AFF3 leads to de-repression of the inactive allele of XIST in terminally differentiated cells PMID:28180295 ZFP281 recruits AFF3 to the Meg3 enhancer within the imprinted Dlk1-Dio3 locus, thus regulating the allele-specific expression of the Meg3 polycistron file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0006355` regulation of DNA-templated transcription | 5 | 4 | 4 | |
| GO:0003712 transcription coregulator activity | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted as AFF3's core molecular function, and kept rather than replaced. The single protein donor holds this exact term by its own IMP, and the term's own usage guidance describes AFF3's mode of DNA engagement precisely. Reason: Drosophila lilli (Q9VQI9) holds GO:0003712 by IMP from PMID:11171404, so the propagation lands exactly on its donor's experimental term rather than above or beside it. More importantly the term is right on AFF3's own evidence, and it is right for a specific reason. GO:0003712's usage comment states that most transcription coregulators do not bind DNA and that those which do usually bind it non-specifically or indirectly, whereas GO:0003700 requires selective binding of a specific genomic sequence motif within a cis-regulatory region. AFF3 matches the first description and fails the second - the isolated protein binds double-stranded DNA cellulose without sequence preference, its transactivating segment was only shown to activate transcription when tethered by a heterologous GAL4 DNA-binding domain, and at its two best-characterised targets it is delivered by sequence-specific zinc-finger proteins (ZFP281 at the Meg3 enhancer, ZFP57 at the IG-DMR) rather than recognising the element itself. The two terms were checked to be siblings - neither GO:0003700 nor GO:0003712 is an ancestor of the other - so this is a substantive choice between different claims, not a granularity preference. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000829417 · PANTHER tree node covering the AF4/FMR2 family SUPPORTS TRANSFER A tree node, not a gene product. FB:FBgn0041111 · Drosophila lilli (Q9VQI9, Swiss-Prot) SUPPORTS TRANSFER Holds this exact term by IMP from PMID:11171404, so the propagation neither overstates nor understates its donor. Supporting Evidence: PMID:28180295 Our study suggests that different zinc finger proteins can recruit AFF3 to different regulatory elements and differentially regulate the function of AFF3 in a context-dependent manner PMID:8555498 In vitro-translated LAF-4 was able to bind strongly to double-stranded DNA cellulose PMID:8555498 both LAF-4 and AF-4 had domains that activated transcription strongly when fused to the GAL4 DNA-binding domain file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0003712` is a descendant of `GO:0003700` | **false** — the two are SIBLINGS, so this is a wrong term, not a coarse one | |
| GO:0006354 DNA-templated transcription elongation | IBA GO_REF:0000033 | MODIFY | Summary: The biology is right but the term is polymerase-agnostic, and every measurement behind it is on RNA polymerase II. Modified to GO:0006368, a verified descendant, so the row asserts strictly more rather than something different. Reason: The WITH/FROM field here is worth dwelling on. It is byte-identical to the field on the same row of AFF1 and AFF4 - PANTHER:PTN000829417 plus UniProtKB:P51825 - but P51825 is AFF1, so the row is self-referential on AFF1 and paralogue-derived on AFF4 and on AFF3. Following that token one level deeper, AFF1's own evidence for GO:0006354 is a single EXP annotation from PMID:22547686, and that paper's measurement is explicitly on polymerase II - the SEC family complexes it purified, SEC-L3 among them, carry high polymerase II C-terminal domain kinase activity, and the gene sets it assigned to SEC-L3 were defined by mRNA-seq. There is no evidence anywhere in the chain for a role in elongation by any other polymerase. GO:0006368 was verified by fetching ancestors with is_a and part_of only to be a descendant of GO:0006354, so this is a downward move that asserts strictly more, and GO:0006368 is not otherwise annotated on AFF3, so the modification adds information rather than duplicating an existing row. Note the divergence within the family - the AFF1 review reached the same MODIFY independently, while the merged AFF4 review accepted the row unchanged. The rule applied here is that a broad term is modified only when the specific term its donor evidence supports is not already annotated on the gene, and that rule reproduces AFF1's answer on this row and AFF4's on the GO:0010468 row below. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: GRANULARITY MISMATCH Sources checked: PANTHER:PTN000829417 · PANTHER tree node covering the AF4/FMR2 family SUPPORTS TRANSFER A tree node, not a gene product. UniProtKB:P51825 · human AFF1 SUPPORTS TRANSFER Holds GO:0006354 by EXP from PMID:22547686 and is a paralogue of the recipient, not the recipient itself - the same token is self-referential on AFF1's own copy of this row. The source annotation is sound; only its polymerase-agnostic term is imprecise, and that imprecision is inherited rather than introduced by the transfer. Proposed replacements: transcription elongation by RNA polymerase II Supporting Evidence: PMID:22547686 The SEC family members demonstrate high levels of polymerase II (Pol II) C-terminal domain kinase activity; however, only SEC is required for the proper induction of the HSP70 gene upon stress PMID:22547686 biochemical isolation of SEC-like 2 (SEC-L2) and SEC-like 3 (SEC-L3) containing AFF2 and AFF3 in association with P-TEFb, ENL/MLLT1, and AF9/MLLT3 file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0006368` is a descendant of `GO:0006354` | true — the MODIFY is a downward move | |
| GO:0050877 nervous system process | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Kept as non-core, and deliberately not refined. Both protein donors hold the same, lower term - learning or memory - which is not supported for AFF3; and the specificity AFF3 does deserve lies in a developmental branch that is not reachable from this term at all. Reason: Querying the donors' own evidence gives an unusually clean picture. Mouse Aff2 (O55112) holds GO:0007611 by IMP from PMID:11923441 and Drosophila lilli holds GO:0007611 by IMP from PMID:18310460; GO:0007611 was verified to be a descendant of GO:0050877, so the propagation landed two levels above two agreeing donors. That is the shape of a granularity mismatch, but refining it would be wrong here, because AFF3 has no learning-or-memory data of its own. Two IMPs to a verified descendant of the propagated term are the row's proper basis, and they are what it is kept on. AFF3's own human genetics then divides in two, and the two halves have to be kept apart rather than offered together as corroboration. The cognitive half is on-branch and does corroborate the row - GO:0050890 cognition was verified to be a descendant of GO:0050877 and GO:0007611 to sit one step under cognition, so the intellectual disability and seizures of KINSSHIP syndrome and the population-scale association of a promoter GCC expansion with a 2.4-fold reduced probability of completing secondary education fall inside the term. The developmental half does not - GO:0021795 and GO:0001764 were both verified NOT to be descendants of GO:0050877, so the impaired migration of cortical neurons on loss of the mouse orthologue cannot be reached by refining this row and is not offered as its grounding; the appropriate specific term for it is proposed separately as GO:0021795. Marked non-core because the gene's core activity is transcriptional. Both the AFF1 and AFF4 reviews reached KEEP_AS_NON_CORE on this row independently. Propagation Review Root cause: NO FAILURE NON CORE Failure modes: GRANULARITY MISMATCH Sources checked: PANTHER:PTN000829417 · PANTHER tree node covering the AF4/FMR2 family SUPPORTS TRANSFER A tree node, not a gene product. FB:FBgn0041111 · Drosophila lilli (Q9VQI9, Swiss-Prot) SUPPORTS TRANSFER Its own experimental term for this branch is GO:0007611 learning or memory, two levels below the propagated term. The transfer of the general term is supported; the transfer of the donor's specific term would not be. MGI:MGI:1202294 · mouse Aff2 (O55112, Swiss-Prot) SUPPORTS TRANSFER Also holds GO:0007611 by IMP, from a Fmr2-knockout conditioned-fear study. A paralogue. Three candidate accessions were returned for this MGI id and the reviewed one was used. Supporting Evidence: PMID:39313615 In a cohort of 6,371 probands with neurodevelopmental problems of suspected genetic etiology, we observed a significant enrichment of AFF3 expansions compared with controls PMID:25162227 We showed here that Laf4 is expressed as early as E13.5 in the mouse neocortex and demonstrated a direct function for this gene in migration of cortical neurons file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0001764` is a descendant of `GO:0050877` | **false** — the developmental branch is unreachable from this term | |
| GO:0032783 super elongation complex | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted, and the term is exactly right - GO:0032783's definition explicitly admits an AFF family protein alongside ELL, EAF, P-TEFb and AF9 or ENL, which is precisely the composition of the AFF3-containing SEC-L3. The row is nevertheless under-evidenced as an IBA when direct human evidence exists. Reason: The definition was read rather than the label - GO:0032783 requires at minimum an ELL family factor, an EAF protein and an AFF family protein or distant relative, most likely also P-TEFb and AF9 or ENL - so SEC-L3 falls inside the term as written and no separate complex term is needed for the AFF3 branch of the family. The sole protein donor, Drosophila lilli, holds this exact term by IPI, so the propagation neither overstates nor understates its donor. What is unsatisfying about the row is its evidence code rather than its term - AFF3-containing SEC-L3 was purified from human cells and shown to contain P-TEFb, ENL/MLLT1 and AF9/MLLT3, and endogenous AFF3 co-immunoprecipitates CDK9 and cyclin T1 from human nuclear extracts with the reciprocal pull-down also performed on endogenous protein, so an IDA or IPI row is available and would be strictly better. IntAct independently records CDK9 in 6 records across 5 distinct publications and 3 distinct methods, with MI scores of 0.35 and 0.73, plus MLLT1 in two records from one publication. These counts are computed in intact_partners.py rather than counted by eye, because the hand-counted version of them was wrong. Recorded as a curation recommendation rather than acted on, since the term itself needs no change. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000829417 · PANTHER tree node covering the AF4/FMR2 family SUPPORTS TRANSFER A tree node, not a gene product. FB:FBgn0041111 · Drosophila lilli (Q9VQI9, Swiss-Prot) SUPPORTS TRANSFER Holds GO:0032783 by IPI from PMID:22195968. Recorded rather than leaned on, because AFF3's own human biochemistry for this term is stronger than the fly transfer. Supporting Evidence: PMID:22547686 biochemical isolation of SEC-like 2 (SEC-L2) and SEC-like 3 (SEC-L3) containing AFF2 and AFF3 in association with P-TEFb, ENL/MLLT1, and AF9/MLLT3 PMID:26214578 To determine whether AFF3 is present in SEC in adrenocortical cells, AFF3 was immunoprecipitated from nuclear extracts. CDK9 and cyclin T1, both components of P-TEFb, co-immunoprecipitated with AFF3 PMID:33961779 The ALF transcription factor paralogs, AFF1, AFF2, AFF3, and AFF4, are components of the transcriptional super elongation complex that regulates expression of genes involved in neurogenesis and development |
| GO:0005634 nucleus | IEA GO_REF:0000120 | ACCEPT | Summary: Accepted. The automatic call is correct and is independently confirmed by AFF3's own IDA, although its three WITH/FROM tokens turn out not to be three independent witnesses. Reason: The named rule was fetched from the ARBA endpoint rather than inferred from the GO_REF. ARBA00026330 has 1309 condition sets and exactly one reaches AFF3 - IPR007797 together with IPR043640 in a eukaryote gives GO:0005634 - which is a two-signature family conjunction for a localisation rather than a fold mapped to an activity, and it is therefore well founded. The independence of the combinatorial reference is weaker than it appears - the second token, InterPro:IPR043640, is one of the two signatures the rule itself conditions on, and the third, UniProtKB-SubCell:SL-0191, derives from UniProt's own SUBCELLULAR LOCATION line, which in this entry carries no evidence tag at all. So one signature is counted twice and the third witness is UniProt citing itself. This does not change the verdict, because the localisation is separately established experimentally on the human protein, but the apparent triple corroboration is illusory and is recorded as such. Propagation Review Root cause: NO FAILURE CORE Sources checked: ARBA:ARBA00026330 · ARBA rule, 1309 condition sets, one of which reaches AFF3 SUPPORTS TRANSFER The reaching condition set is IPR007797 AND IPR043640 AND Eukaryota. A family conjunction supporting a localisation, not a fold mapped to an activity. InterPro:IPR043640 · AF4/FMR2 C-terminal homology domain CIRCULAR OR REDUNDANT Not an independent witness - it is one of the two signatures the ARBA rule in the same WITH/FROM field already conditions on. UniProtKB-SubCell:SL-0191 · UniProt subcellular location keyword, Nucleus CIRCULAR OR REDUNDANT Derives from this entry's own SUBCELLULAR LOCATION line, which carries no evidence tag, so it is UniProt citing itself rather than a third method. Supporting Evidence: PMID:8555498 Antibodies to LAF-4 showed it to be a nuclear protein that showed an uneven, granular immunofluorescence pattern file:human/AFF3/AFF3-uniprot.txt CC -!- SUBCELLULAR LOCATION: Nucleus. file:human/AFF3/AFF3-bioinformatics/RESULTS.md IPR007797 AND IPR043640 AND taxon Eukaryota -> GO:0005634 |
| GO:0010468 regulation of gene expression | IEA GO_REF:0000002 | ACCEPT | Summary: Accepted. The term is true and the signature that supplies it is family-specific, but it is a redundant ancestor of the GO:0006355 IBA row on the same gene, so refining it would add nothing. Reason: Each of AFF3's three InterPro signatures was looked up separately rather than the mapping being attributed to interpro2go in general - IPR007797 (AF4/FMR2 family, 5800 proteins) supplies GO:0010468, IPR043640 supplies GO:0005634, and IPR043639 supplies nothing. This reproduces the merged AFF4 review's measurement exactly, and it also means that no molecular-function term is produced by any of AFF3's signatures, so the familiar fold-becomes-an-activity error does not occur on this gene. GO:0006355 was verified to be a descendant of GO:0010468 and is already annotated on AFF3 from a better-founded IBA whose donor set includes the true mouse orthologue with its own IDA and IMP. So the more specific term the evidence supports is already present, and modifying this row would merely duplicate it. That is the same rule applied to the GO:0006354 row above, where the specific term was absent and MODIFY therefore did add information. Note that the two sibling reviews split on this row - AFF1 modified it to GO:0006355 and AFF4 accepted it - and the rule adopted here reproduces AFF4's answer for a stated reason rather than by preference. Propagation Review Root cause: NO FAILURE CORE Sources checked: InterPro:IPR007797 · AF4/FMR2 family signature, 5800 proteins SUPPORTS TRANSFER A family-specific signature rather than a bare fold, and every characterised member of the family regulates transcription, so the mapping is sound. Its interpro2go target is a general parent, which is appropriate for a signature covering the whole family. Supporting Evidence: file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0006355` is a descendant of `GO:0010468` | true — the InterPro row is a redundant ancestor | PMID:22547686 SEC-L2 and SEC-L3 control the expression of different subsets of genes |
| GO:0005654 nucleoplasm | IDA GO_REF:0000052 | ACCEPT | Summary: Accepted. Human Protein Atlas immunofluorescence, consistent with the antibody localisation in the founding paper and refined by the nuclear-speckle localisation proposed as a separate row. Reason: Two independent human observations agree with this - the founding paper's antibody staining describes an uneven, granular nuclear pattern, and colocalisation with the speckle marker SC35 in adrenocortical cells identifies that granularity as nuclear speckles. GO:0016607 was verified to be a descendant of GO:0005654, so the proposed speckle row refines this one rather than contradicting it, and both are retained. Supporting Evidence: PMID:8555498 Antibodies to LAF-4 showed it to be a nuclear protein that showed an uneven, granular immunofluorescence pattern PMID:26214578 In adrenocortical cells, AFF3 colocalized with the SC35 protein (a nuclear speckle marker) file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0016607` is a descendant of `GO:0005654` | true — nuclear speck refines the nucleoplasm IDA | |
| GO:0005634 nucleus | IDA PMID:8555498 LAF-4 encodes a lymphoid nuclear protein with transactivatio... | ACCEPT | Summary: Accepted. A direct antibody localisation of the endogenous protein in lymphoid cells, and the observation the gene's UniProt SUBCELLULAR LOCATION line ultimately rests on. Reason: Abstract-only in the cache, but the abstract states the result in full. The reference annotates two entities across all of GOA - human AFF3 and mouse Aff3 - both to this same term, which is what a paper that worked in human and mouse lymphoid cell lines would be expected to produce, and not a projection pattern. Supporting Evidence: PMID:8555498 Antibodies to LAF-4 showed it to be a nuclear protein that showed an uneven, granular immunofluorescence pattern file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `PMID:8555498` | 2 | 2 | `GO:0005634` IDA on P51826 and P51827 | no | |
| GO:0034612 response to tumor necrosis factor | IMP PMID:20444755 Investigation of rheumatoid arthritis susceptibility genes i... | REMOVE | Summary: Removed. The sole supporting paper is a pharmacogenetic association study of patients' clinical response to TNF-blocking drugs. There is no TNF stimulus anywhere in it, no perturbation of the AFF3 gene product, and no cellular assay of any kind. Reason: This is the failure mode a disease-susceptibility gene invites, and the full text was read rather than the title. Three independent grounds. First, definitional - GO:0034612 requires a change in state or activity of a cell or organism as a result of a tumor necrosis factor stimulus, whereas the study administers TNF blockade to rheumatoid arthritis patients, the opposite perturbation, and never applies TNF to anything. Second, the evidence code - IMP requires a mutant phenotype, and the study genotypes eighteen common tag SNPs and regresses a six-month change in a clinical disease-activity score on them, which is neither a mutant nor a phenotype of the gene product. Third, there is no assay - the paper contains no cellular or biochemical experiment on AFF3 at all. Querying QuickGO by reference confirms this is the only annotation the paper carries anywhere in GOA, so nothing else depends on it, and it has already propagated - mouse Aff3 holds GO:0034612 solely by ISO with UniProtKB:P51826, that is, from this human row - so retracting the human row retracts the term from both species. The distinction being drawn is between the gene's normal molecular function and a statistical association between one of its variants and a drug response in patients. AFF3 does have a real, mechanistically characterised immunological function, in immunoglobulin class switch recombination, and that is proposed as a separate row on separate evidence; nothing here is a claim that AFF3 is uninvolved in immunity. Supporting Evidence: PMID:20444755 Eighteen single nucleotide polymorphisms (SNPs) mapping to 11 genetic loci were genotyped in 1012 patients with RA receiving treatment with etanercept, infliximab or adalimumab PMID:20444755 using the absolute change in 28 joint count disease activity score (DAS28) between baseline and 6-month follow-up as the outcome variable PMID:20444755 SNPs mapping to AFF3 and CD226 had a statistically significant association with the response to anti-TNF treatment file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `PMID:20444755` | 1 | 1 | `GO:0034612` IMP on P51826 | no | |
| GO:0035116 embryonic hindlimb morphogenesis | IMP PMID:18616733 Triangular tibia with fibular aplasia associated with a micr... | KEEP AS NON CORE | Summary: Kept as non-core. My first reading was that a single case report of a microdeletion could not carry a gene-specific IMP; that was wrong, because an independent paper establishes the deletion contains only AFF3, and three further lines of evidence converge on the lower limb specifically. Reason: The concern that this is a contiguous-gene deletion is answered directly - PMID:25162227 states that the 500 kb 2q11.1 microdeletion encompasses only the LAF4 gene. The phenotype is lower-limb and mesomelic, matching the term rather than its parent - the patient has fibular agenesis and triangular tibiae, the same paper's mouse in-situ shows expression in the limb buds and the zeugopod, an independent study reports mouse expression in the developing brain, somites and limb buds, and knockin mice modelling the KINSSHIP microdeletion reproduce lower mesomelic limb deformities while homozygous Aff3 knockouts show skeletal anomalies. GO:0035116's definition is simply the generation and organisation of the hindlimbs, the posterior limbs of an animal, so there is no taxon or anatomy problem for a human lower limb. On the direction of effect - AFF3's limb phenotypes arise both from loss of the gene and from stabilisation of the protein by degron variants, and a gene whose increase and decrease both disrupt limb development is participating in limb development either way, so involved_in is right. What the gain-of-function data do not license is any claim about a new molecular activity of the variant protein, and none is made here - the KINSSHIP substitutions confer resistance to degradation, not a new function. Marked non-core because AFF3's core activity is transcriptional and this is a downstream developmental role. Supporting Evidence: PMID:25162227 A human microdeletion of 500 kb on chromosome 2q11.1 encompassing only the LAF4 gene has been detected by array comparative genomic hybridization on peripheral lymphocytes PMID:18616733 We report on a girl with fibular agenesis, severely abnormal, triangular tibiae PMID:18616733 In situ hybridization analysis of Laf4 in mouse embryos revealed expression in the developing brain, in the limb buds and in the zeugopod corresponding to the limb phenotype PMID:33961779 Whereas homozygous Aff3 knockout mice display skeletal anomalies, kidney defects, brain malformations, and neurological anomalies PMID:33961779 knockin animals modeling one of the microdeletions and the most common of the missense variants identified in affected individuals presented with lower mesomelic limb deformities like KINSSHIP-affected individuals and early lethality PMID:24763282 By whole-mount in situ hybridization the mouse AFF3 ortholog shows strong regional expression in the developing brain, somites and limb buds in 9.5-12.5dpc mouse embryos |
| GO:0003711 transcription elongation factor activity | IDA PMID:22547686 The super elongation complex family of RNA polymerase II elo... | NEW | Summary: Proposed. The purified AFF3-containing SEC-L3 carries high polymerase II C-terminal domain kinase activity, which is an activity of the assembled complex that AFF3 contributes to rather than one AFF3 has alone - hence the contributes_to qualifier. Reason: The catalytic subunit of SEC-L3 is CDK9, not AFF3, so the elongation-stimulating activity belongs to the complex; contributes_to is the qualifier GO provides for exactly this case. The activity was measured on the purified complexes rather than inferred, and the same paper defines SEC-L3's composition. GO:0003711 was checked not to be redundant with the GO:0003712 row - the two are siblings under GO:0140110, and neither ancestor closure contains the other - so this is an additional statement rather than a refinement. Note the paper's own negative in the same sentence, which is respected here - only AFF4-containing SEC, not SEC-L3, is required for stress induction of HSP70, so no claim is made that SEC-L3 substitutes for SEC. One step in this is an interpretation rather than a measurement and is flagged as such - what was measured is polymerase II C-terminal domain kinase activity of the purified complexes, and reading that as elongation-factor activity relies on the established role of CTD Ser2 phosphorylation in releasing paused polymerase rather than on an elongation assay performed on SEC-L3. The contributes_to qualifier already keeps the claim off AFF3 itself; the experiment that would convert the inference into a measurement is set out in suggested_experiments. Both the AFF1 and AFF4 reviews reached GO:0003711 as the complex-level contribution for their genes, from independent evidence, and the sibling relation is now checked on both legs - GO:0003711 and GO:0003712 were each verified to sit under GO:0140110 with neither containing the other. Supporting Evidence: PMID:22547686 The SEC family members demonstrate high levels of polymerase II (Pol II) C-terminal domain kinase activity; however, only SEC is required for the proper induction of the HSP70 gene upon stress PMID:22547686 biochemical isolation of SEC-like 2 (SEC-L2) and SEC-like 3 (SEC-L3) containing AFF2 and AFF3 in association with P-TEFb, ENL/MLLT1, and AF9/MLLT3 |
| GO:0030674 protein-macromolecule adaptor activity | IPI PMID:26214578 Combined transcriptome studies identify AFF3 as a mediator o... | NEW | Summary: Proposed. AFF3 is simultaneously in complex with both modules SEC-L3 joins - the P-TEFb kinase and the ENL/AF9 proteins - in human cells, and holding those modules together is what the complex needs from the AFF subunit. Reason: The evidence is complex-level rather than interface-level, and that is stated plainly rather than hedged. Endogenous AFF3 immunoprecipitated from human nuclear extracts brings down CDK9 and cyclin T1, with the reciprocal co-immunoprecipitation also performed on endogenous protein; the purified AFF3-containing SEC-L3 contains P-TEFb together with ENL/MLLT1 and AF9/MLLT3; and IntAct independently records CDK9 in 6 records across 5 distinct publications and 3 distinct methods (anti tag coip, pull down, tap), with MI scores of 0.35 and 0.73, plus MLLT1 in two records from one publication. What is not established for AFF3, in contrast to AFF1 and AFF4, is which residues form each interface - there is no AFF3 structure with a partner bound - so the claim is made at the level of the activity and not of the surface. GO:0030674 was verified NOT to be a descendant of GO:0005515 - its only ancestors are GO:0003674 and GO:0060090 - so this is a move into the molecular-adaptor branch rather than a refinement of protein binding. AFF3 has no GO:0005515 rows in GOA at all, which is why this is proposed as a new row rather than as a replacement, and it is a coverage gap rather than a correction - two of AFF3's replicated IntAct partners are the very SEC modules it bridges and GOA has curated neither. The scale of that gap was measured within the family in a single QuickGO call, so a zero could not be a rejected query - AFF4 carries 15 GO:0005515 rows and AFF1 carries 3, while AFF3 carries none, which is a coverage asymmetry rather than a biological one given that AFF3's CDK9 contact is the most replicated of the three in IntAct. For IPI the WITH/FROM field takes the interacting partners, so it names CDK9 (UniProtKB:P50750) and cyclin T1 (UniProtKB:O60563), the two proteins the co-immunoprecipitation actually brought down; both were confirmed to be reviewed human entries by asserting primaryAccession on the fetch. Making the row ISS on AFF1 and AFF4 instead, whose bridging interfaces are structurally mapped, was considered and declined - the assay here is a co-immunoprecipitation of AFF3 itself from human cells, so IPI is the correct code and what was missing was the entity rather than the code. The unmapped-interface caveat belongs in this reason and in the knowledge gap, which is where it is, not in a weaker evidence code. Supporting Evidence: PMID:26214578 The interactions were confirmed by reciprocal co-IP of endogenous CDK9 and overproduced AFF3-flag (Figure 7b, middle) or endogenous AFF3 (Figure 7b, bottom) PMID:26214578 To determine whether AFF3 is present in SEC in adrenocortical cells, AFF3 was immunoprecipitated from nuclear extracts. CDK9 and cyclin T1, both components of P-TEFb, co-immunoprecipitated with AFF3 PMID:22547686 biochemical isolation of SEC-like 2 (SEC-L2) and SEC-like 3 (SEC-L3) containing AFF2 and AFF3 in association with P-TEFb, ENL/MLLT1, and AF9/MLLT3 file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0030674` is a descendant of `GO:0060090` | true — adaptor activity is a separate MF branch, not a refinement of protein binding | file:human/AFF3/AFF3-bioinformatics/RESULTS.md | CDK9 (P50750) | 6 | 5 | 3 | 0.35 and 0.73 | |
| GO:0003690 double-stranded DNA binding | IDA PMID:8555498 LAF-4 encodes a lymphoid nuclear protein with transactivatio... | NEW | Summary: Proposed. A direct in vitro assay on the human protein, curated in UniProt's own FUNCTION comment and present as an Ensembl cross-reference on the UniProt entry, but absent from GOA. Reason: In-vitro-translated human LAF-4 binds strongly to double-stranded DNA cellulose, which is a direct assay of the term as defined - GO:0003690's definition is simply binding to double-stranded DNA, with no specificity clause. Human ChIP corroborates chromatin engagement at a defined element, the methylated CpG island downstream of the XIST promoter, and mouse Aff3 holds the same term by IDA. This is one of two GO terms carried on AFF3's UniProt entry as IEA:Ensembl but missing from GOA, and the two resolve oppositely - this one is correct and the accompanying GO:0003700 is not, which is raised separately as a UniProt correction rather than as a GO action. The distinction matters and is deliberate - binding double-stranded DNA is not the same claim as recognising a cis-regulatory sequence motif, and only the first is supported. Supporting Evidence: PMID:8555498 In vitro-translated LAF-4 was able to bind strongly to double-stranded DNA cellulose PMID:30535390 Quantitative PCR after chromatin immunoprecipitation (ChIP-qPCR) demonstrates that AFF3 is associated with the XIST DMR in both IMR-90 and HEK293T file:human/AFF3/AFF3-uniprot.txt DR GO; GO:0003690; F:double-stranded DNA binding; IEA:Ensembl. |
| GO:0016607 nuclear speck | IDA PMID:26214578 Combined transcriptome studies identify AFF3 as a mediator o... | NEW | Summary: Proposed. AFF3 colocalises with the speckle marker SC35 in human adrenocortical cells, a refinement of the existing nucleoplasm IDA and the specific identity of the granular pattern described in 1996. Reason: GO:0016607 was verified to be a descendant of GO:0005654, so this refines the existing HPA row rather than conflicting with it, and the two are compatible. The colocalisation is reported as a baseline observation and is distinguished in the paper's own text from the separate over-production experiment in which strong AFF3 excess disrupts SC35 distribution and redraws CDK9 and cyclin T1 - only the colocalisation is used here. The paper's discussion notes that the localisation reproduces an earlier report in HeLa cells, so it is not a single-cell-line observation. One limitation is recorded rather than glossed - the text does not state the expression level in the specific panel showing colocalisation, so whether the signal is endogenous or from a moderate transfection cannot be read off the cached article. Supporting Evidence: PMID:26214578 In adrenocortical cells, AFF3 colocalized with the SC35 protein (a nuclear speckle marker) PMID:26214578 In H295R adrenocortical cells, AFF3 is located in nuclear speckles; its overproduction interferes with the organization and/or biogenesis of these nuclear speckles |
| GO:0045892 negative regulation of DNA-templated transcription | IMP PMID:30535390 AFF3-DNA methylation interplay in maintaining the mono-allel... | NEW | Summary: Proposed. Knockdown of AFF3 in two human cell lines de-represses the silent XIST allele, so AFF3 acts as a repressor at a defined element - a signed statement the existing unsigned GO:0006355 row cannot make. Reason: This is human throughout, with two independent shRNAs in HEK293T and two more in IMR-90, and it is mechanistically anchored - AFF3 ChIP-qPCR places the protein on the methylated CpG island downstream of the XIST promoter in both lines, the binding is lost on demethylating treatment, and knockdown raises XIST transcript levels and the number of XIST RNA clouds. The signed term is proposed rather than a modification of the GO:0006355 row because AFF3's regulatory output is not uniformly negative - at the Meg3 enhancer and genome-wide it acts with ZFP281 to permit expression - so the unsigned parent remains the right general statement and this row adds the specific repressive instance. No claim about X-chromosome inactivation itself is made - the observation is about maintenance of monoallelic XIST expression in already-differentiated cells, not about initiating inactivation. Supporting Evidence: PMID:30535390 Knockdown of AFF3 leads to de-repression of the inactive allele of XIST in terminally differentiated cells PMID:30535390 AFF3 can directly bind to the XIST DMR, located at the CpG island downstream of the XIST promoter in terminally differentiated cells PMID:30535390 Quantitative PCR after chromatin immunoprecipitation (ChIP-qPCR) demonstrates that AFF3 is associated with the XIST DMR in both IMR-90 and HEK293T |
| GO:0021795 cerebral cortex cell migration | ISS PMID:25162227 Laf4/Aff3, a gene involved in intellectual disability, is re... | NEW | Summary: Proposed. Knockdown of the mouse orthologue impairs radial migration of cortical neurons in organotypic slices, with a shRNA-resistant rescue construct restoring it. This is the specific nervous-system role that the existing GO:0050877 row cannot express. Reason: ISS with mouse Aff3 (P51827) as the supporting entity, because the perturbation was done in mouse; the human evidence is genetic and clinical rather than cellular. The mouse result is well controlled - a rescue construct not recognised by the shRNA abolishes the phenotype, and the same manipulation does not affect cell survival or axon growth - and the term matches the assay's own language, orderly movement of cells in the cerebral cortex, rather than over-specifying a migration mode. This term was verified NOT to be reachable by refining GO:0050877, whose branch is organ-system process rather than development, which is why it is a separate row and not a modification of that one. Supporting Evidence: PMID:25162227 we discovered that Laf4 is required for cortical cell migration PMID:25162227 We showed here that Laf4 is expressed as early as E13.5 in the mouse neocortex and demonstrated a direct function for this gene in migration of cortical neurons file:human/AFF3/AFF3-bioinformatics/RESULTS.md | `GO:0001764` is a descendant of `GO:0050877` | **false** — the developmental branch is unreachable from this term | |
| GO:0045190 isotype switching | ISS PMID:36001653 AFF3, a susceptibility factor for autoimmune diseases, is a ... | NEW | Summary: Proposed. AFF3 binds immunoglobulin switch-region DNA through its C-terminal region and promotes recruitment of the deaminase AID; the knockout mouse loses IgG2c, IgG1 and IgG3 but not IgM. Reason: The process term is chosen over the regulation branch on the strength of GO:0045190's own usage comment, which directs gene products involved in the mechanism of the switch recombination itself to this term and reserves the regulation term for cytokines and co-stimulatory molecules acting from outside. AFF3 binds the switch regions and delivers the deaminase that initiates the reaction, which is mechanism-level participation. ISS with mouse Aff3 as the supporting entity, since the knockout and the switch-region binding were done in mouse; the human arm of the same study is an expression association in B cells and is not sufficient on its own. On the molecular function this process pairs with, an adaptor or coregulator term was considered and declined as over-claiming. GO:0030674 would assert a bridge between the switch-region DNA and the deaminase, and the paper explicitly leaves that mechanism open, offering cohesin and P-TEFb as alternative routes; GO:0003712 belongs to transcription regulation whereas class switch recombination is a DNA recombination reaction. What was measured is occupancy of the switch regions, so double-stranded DNA binding is the activity this function rests on. This is also the substantive immunological function of a gene whose only previous immune annotation, response to tumor necrosis factor, is removed in this review as unsupported - the removal is not a claim that AFF3 has no immune role. Supporting Evidence: PMID:36001653 the AFF3 protein binds to the IgM and IgG1 switch regions via a C-terminal domain, and Aff3 deficiency reduces the binding of AID to the switch regions less efficiently PMID:36001653 Aff3-deficient mice exhibit low serum levels of immunoglobulins, predominantly immunoglobulin G2c (IgG2c) followed by IgG1 and IgG3 but not IgM PMID:36001653 AFF3 directly regulates CSR by facilitating the recruitment of AID to the switch regions PMID:36001653 We detected significant increases in the signals near the switch regions of IgM and IgG1, indicating that AFF3 can bind to these regions PMID:36001653 While the mechanism by which AID is recruited to switch regions is still unclear, the following mechanisms have been proposed PMID:36001653 AFF3 may regulate CSR by facilitating the interaction of AID with cohesin factors |
| GO:0001822 kidney development | ISS PMID:33961779 Variants in the degron of AFF3 are associated with intellect... | NEW | Summary: Proposed, as a non-core developmental role. Homozygous Aff3 knockout mice have kidney defects, and human patients show horseshoe kidney from stabilising degron variants - hypoplastic kidney too, per UniProt's DISEASE line - and urogenital malformations from deletion. Reason: Three independent lines converge and they come from opposite directions of gene dosage, which is what makes this a participation claim rather than a phenotype claim. The mouse null is the primary evidence and is why the row is ISS with mouse Aff3 as supporting entity; independently, KINSSHIP is named partly for the kidney phenotype - PMID:33961779 spells the acronym out as KI for horseshoe kidney, while UniProt's DISEASE line for the syndrome records horseshoe or hypoplastic kidney, and the broader wording is UniProt's rather than the paper's. The second supporting quote is weaker than it looks and is flagged rather than leaned on - the deletion patient's finding is recorded as urogenital tract malformation, which is broader than the kidney, so the mouse null's explicit kidney defect is doing the work and the human quote is corroboration of organ system rather than of organ. As with the limb row, nothing is inferred about a new molecular activity of the variant protein - the degron substitutions confer resistance to degradation. Proposed with restraint, as the weakest of the developmental rows here, and explicitly non-core. Supporting Evidence: PMID:33961779 Whereas homozygous Aff3 knockout mice display skeletal anomalies, kidney defects, brain malformations, and neurological anomalies PMID:18616733 We report on a girl with fibular agenesis, severely abnormal, triangular tibiae, urogenital tract malformations, failure to thrive, convulsions and recurrent apnoeas leading to respiratory arrest at the age of 4 months file:human/AFF3/AFF3-uniprot.txt facial features, horseshoe or hypoplastic kidney, and failure to |
Loading supporting content…
Download this section (compressed HTML)Q: PMID:22547686 is the paper that isolated the AFF3-containing SEC-L3, and its only annotation anywhere in GOA is GO:0006354 EXP on AFF1, the paralogue it contrasts SEC-L3 against. AFF3 receives the term back only as an IBA whose WITH/FROM points at AFF1. This is the second instance of the same shape in this family - the merged AFF4 review found that PMID:20159561, titled for AFF4, likewise produced one annotation and it was on AFF1. Should the family be re-swept for papers whose annotations landed on AFF1 rather than on the paralogue they characterise, and should AFF3 receive direct IDA or IPI rows for super elongation complex membership and for elongation from this paper?
Suggested experts: GO Central, PAINT curators
Q: AFF3's UniProt entry carries GO:0003700 DNA-binding transcription factor activity by IEA:Ensembl, projected from mouse Aff3, and GOA does not. The mouse source rows rest on ChIP-qPCR of over-expressed HA-tagged Laf4 at a single promoter, which the authors themselves frame as testing a potential direct transcriptional regulator, and GO:0003700's usage comment warns against this case. AFF3 has no sequence motif, binds double-stranded DNA indiscriminately in vitro, and is recruited by ZFP281 and ZFP57. Should the Ensembl-projected GO:0003700 cross-reference be retracted in favour of GO:0003712, which AFF3 already holds? Note that the companion projection on the same entry, GO:0003690, is correct and is missing from GOA - the two should move in opposite directions.
Suggested experts: UniProt curators, Ensembl Compara
Q: On the GO:0050877 nervous system process node, both protein donors' own experimental term is GO:0007611 learning or memory, two levels below the propagated term, while the recipient AFF3's documented nervous-system role is developmental and sits in a branch not reachable from GO:0050877 at all. Is a node-level split warranted, keeping the behavioural term where the behavioural evidence is and giving the AFF3 branch a developmental term instead?
Suggested experts: PAINT curators
Q: Mouse Aff3 holds GO:0016604 nuclear body and GO:0005829 cytosol by ISO GO_REF:0000119 with UniProtKB:P51826, while human AFF3 currently carries neither term. These look like reflections of human annotations that have since been withdrawn. The same projection route has a live consequence for the GO:0034612 removal proposed here - mouse Aff3 holds response to tumor necrosis factor solely by ISO from the human row, so retracting the human row should retract the mouse one, and there is no other support for it in either species. Should ISO projections be re-checked against their current source, does the pipeline retract a projection when its source annotation disappears, and can the GO:0034612 pair be withdrawn together?
Suggested experts: MGI, GO Central, BHF-UCL
Q: AFF4's merged review asks whether GO:0032783 should reach AFF3 at all, on the premise that AFF3 has not been shown to be a subunit of the super elongation complex. That premise is refuted by AFF3's own literature - PMID:22547686 reports the biochemical isolation of AFF3-containing SEC-L3 together with P-TEFb, ENL/MLLT1 and AF9/MLLT3, and PMID:26214578 immunoprecipitates endogenous AFF3 from human nuclear extracts and recovers CDK9 and cyclin T1, with the reciprocal pull-down also on endogenous protein. GO:0032783's definition explicitly admits an AFF family protein, so SEC-L3 falls inside the term as written. Should that question be withdrawn for AFF3, and should the complex-membership row be upgraded from IBA to a direct experimental annotation rather than restricted to an AFF1/AFF4 node? The same correction does not obviously extend to AFF2, which was not examined here.
Q: The nine-residue AFF3 degron whose variants cause KINSSHIP syndrome is not a feature in UniProt - the entry records the individual VARIANT rows at positions 231, 233 and 235 but no corresponding region feature. Should a degron or short-sequence-motif feature be added, given that the region is the mechanistic basis of a named syndrome and of the gene's dosage sensitivity?
Suggested experts: UniProt curators
Experiment: Map the CDK9/cyclin T1 and ENL/AF9 binding regions of AFF3 by co-immunoprecipitation from a deletion series expressed at near-endogenous level in an AFF3-knockout human line, then test whether a variant that loses only the P-TEFb contact still assembles SEC-L3 and whether one that loses only the ENL/AF9 contact does. This would convert the adaptor activity proposed here from a complex-membership inference into a mapped bridging function.
Hypothesis: AFF3 bridges the P-TEFb kinase module and the ENL/AF9 proteins through non-overlapping segments, as AFF1 and AFF4 do, so single segment deletions should separate the two contacts.
Type: interaction mapping
Experiment: Perform PRO-seq or NET-seq before and after acute degradation of endogenous AFF3, using a degron tag rather than shRNA so that direct and adaptive effects are separable, in a human line, and ask whether pausing index rises and Ser2-phosphorylated polymerase II falls specifically at AFF3-occupied genes. This is the measurement that would justify a positive regulation of transcription elongation term for AFF3, which is deliberately not proposed in this review.
Hypothesis: AFF3 promotes elongation at its own target genes, which the current evidence shows only for the complex rather than for the gene.
Type: genomics
Experiment: In primary human B cells stimulated to switch, perform AFF3 ChIP-qPCR at the IgM and IgG switch regions and measure AID occupancy and switching efficiency after AFF3 knockdown. This would replace the ISS row proposed here with human experimental evidence, and would test whether the risk-allele expression association has a mechanistic counterpart.
Hypothesis: Human AFF3 binds immunoglobulin switch regions and promotes AID recruitment, as the mouse orthologue does.
Type: immunology
Experiment: Compare AFF3 ChIP-seq peaks in a human line with and without depletion of candidate recruiters, taking ZNF281 and ZFP57 first and then factors whose motifs are enriched under AFF3 peaks, and in parallel test purified AFF3 for sequence preference by SELEX or protein binding microarray. A null result in vitro combined with recruiter-dependent loss in vivo would settle the coregulator-versus-transcription-factor question that the UniProt GO:0003700 cross-reference turns on.
Hypothesis: The AFF3 zinc-finger recruitment model generalises - AFF3 has no intrinsic sequence specificity and its genomic distribution is set entirely by partner factors.
Type: genomics
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: No structure of AFF3 in complex with any partner exists, so which segments of AFF3 contact P-TEFb, ENL and AF9 is unknown, in contrast to AFF1 and AFF4 where the cyclin T1 and ELL interfaces are mapped.
OPEN BIOLOGY RESIDUAL_SUBGAP
Significance: The adaptor activity is asserted here from simultaneous complex membership rather than from mapped surfaces. Interface mapping would also determine whether the four AFF paralogues bind the shared modules through the same or different segments, which is the structural question behind their non-redundancy.
Provenance (the field's own admissions):
Gap: Whether human AFF3 binds immunoglobulin switch regions and promotes AID recruitment in human B cells has not been tested; the mechanism is established in mouse, and the human arm of the same study is an expression association only.
OPEN BIOLOGY BP_DARK
Significance: The isotype switching row proposed here is ISS for that reason. AFF3 is a shared rheumatoid arthritis and type 1 diabetes susceptibility locus, so whether the switching mechanism operates in human cells bears directly on how the association is interpreted.
Provenance (the field's own admissions):
Gap: GO has no term for maintenance of allele-specific silencing by a protein that reads DNA methylation at a differentially methylated region, which is what AFF3 does at the XIST DMR and at imprinted gametic DMRs.
OPEN ONTOLOGY
Significance: The behaviour is captured here only as negative regulation of transcription, which loses both the allele specificity and the methylation-reading mechanism. It is recorded as a gap rather than resolved with a term request, because the boundary against genomic imprinting, whose own definition already covers propagation of imprints through life, would need curator input.
Provenance (the field's own admissions):
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)