ACTL7B (actin-like protein 7B, T-ACTIN 1) is a 415-residue actin-related protein (ARP) expressed almost exclusively in the testis. The gene is intronless and sits head-to-head with its paralogue ACTL7A on chromosome 9q31, both having arisen by retroposition of a spliced actin-progenitor mRNA. The protein appears post-meiotically in round and elongating spermatids, where it is found in the cytoplasm, on the developing acrosome, at the region connecting the flagellum to the sperm head, and at lower abundance inside the nucleus of early spermatids and spermatocytes. Two independent mouse knockouts show that it is required for spermiogenesis: null males are infertile with severe oligoteratozoospermia, multiple morphological abnormalities of the flagellum, malformed sperm heads, detached acrosomes and disrupted membranes appearing from spermiogenesis step 9, after which abnormal spermatids are degraded. Its only well-supported molecular partners are the LC8-family dynein light chains DYNLL1 and DYNLL2, which ACTL7B binds in testis and whose intracellular distribution becomes abnormal in its absence, implicating ACTL7B in the positioning of dynein light chains during spermatid remodelling. Although it retains the actin fold and much of the actin nucleotide pocket, the surface actin uses for filament formation is not conserved, with the DNase-I-binding loop entirely replaced, so ACTL7B is not expected to behave as a conventional filament-forming actin, and no polymerisation, nucleotide-hydrolysis or actin-binding activity has been demonstrated for it. A nuclear role in chromatin regulation, possibly by substituting for conventional nuclear ARPs in nucleosome-remodelling complexes, has been proposed but so far rests on localisation, transcriptome changes in knockout testis, and computational docking.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005634 nucleus | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Phylogenetic propagation from PANTHER node PTN008986520. All four protein seeds were resolved and each carries its own experimental evidence for nucleus: mouse Actl7a (Q9QY84, Swiss-Prot; EXP and IDA), Saccharomyces ARP9 (Q05123, Swiss-Prot; IDA and IPI), Candida ARP9 (Q5A9X7, unreviewed TrEMBL; IDA) and a Trypanosoma brucei actin-like protein (Q57ZL0, unreviewed TrEMBL; IDA and HTP). The two TrEMBL entries are reported as such: their protein names are automatic by-similarity labels and are not used here as evidence of what the family does, although their curated GO annotations are real. What is missing from the seed list is the true orthologue, mouse Actl7b. The path is therefore paralogue plus deep nuclear ARPs at a node constrained only to Eukaryota, which is a weak basis for a compartment claim. The destination, however, is independently correct: two 2023-2024 mouse studies place ACTL7B itself inside the nucleus of early spermatids and spermatocytes, one of them reporting a conserved putative nuclear localisation sequence. Reason: The localisation is right and should be kept, but it is not where ACTL7B's demonstrated function lies, and the qualifier overstates the evidence. is_active_in asserts that the gene product's molecular function is executed in the nucleus, and no nuclear activity has been measured for ACTL7B: the proposed chromatin-remodeller subunit-swapping mechanism is computational docking, and the HDAC1 and HDAC3 mislocalisation phenotype shows a requirement, not an activity. located_in would be the defensible qualifier. Kept as non-core, and the better evidence line would be ISS or ISO from mouse Actl7b (PMID:36617158, PMID:38464253) rather than IBA from ARP9. Propagation Review Root cause: NO FAILURE NON CORE Failure modes: WRONG ORTHOLOG OR PARALOG Sources checked: MGI:MGI:1343051 · mouse Actl7a (Q9QY84), paralogue rather than orthologue SUPPORTS SOURCE BUT NOT TARGET Carries GO:0005634 by EXP (PMID:12672658, PMID:21278383) and IDA (PMID:36734600), so the source annotation is sound. But it is the ACTL7A paralogue, and the paper behind its nucleus call reported the opposite compartment for Actl7b: T-ACTIN 2 (Actl7a) nuclear, T-ACTIN 1 (Actl7b) cytoplasmic. Paralogue transfer is legitimate for IBA, but it means no orthologue-strength inference is available on this row. SGD:S000004636 · S. cerevisiae ARP9 (Q05123), RSC and SWI/SNF actin-related subunit SUPPORTS SOURCE BUT NOT TARGET Swiss-Prot, with GO:0005634 IDA and IPI of its own; also annotated to GO:0016586 (RSC complex) and GO:0016514 (SWI/SNF complex) at node PTN000940625. A genuine nuclear ARP, but one of the most divergent members of the family, 20.2% identical to ACTL7B and retaining only 7 of 20 actin nucleotide contacts, so it supports "some ARPs are nuclear", not "ACTL7B is nuclear". CGD:CAL0000196900 · C. albicans ARP9 (Q5A9X7) SOURCE WEAK OR INFERRED Unreviewed TrEMBL entry; gene symbol ARP9 comes from CGD. Carries its own IDA for nucleus. Reported as unreviewed because an unreviewed entry is weaker support and its protein name cannot be treated as functional evidence. UniProtKB:Q57ZL0 · T. brucei "Actin-like protein, putative" SOURCE WEAK OR INFERRED Unreviewed TrEMBL entry with no gene symbol; the name is an automatic by-similarity label. Its GO record does carry IDA and HTP for nucleus, so the evidence provenance is real even though the name provenance is not. PANTHER:PTN008986520 · ancestral PANTHER node (taxon constraint 2759, Eukaryota) SUPPORTS TRANSFER Internal tree node, not a protein. Holds GO:0005634 by IBD from the four seeds above. A pan-eukaryotic node is a coarse level from which to assign a compartment to a mammal-specific, testis-restricted protein. Supporting Evidence: PMID:36617158 In mice, ACTL7B specifically localises to the developing acrosome, within the nucleus of early spermatids, and to the flagellum connecting region. PMID:38464253 This study confirms the intranuclear presence of ACTL7B in spermatocytes and round spermatids PMID:37800308 ACTL7B is detected in the cytoplasm and at lesser amounts in the nucleus of round and elongating spermatids |
| GO:0005737 cytoplasm | IBA GO_REF:0000033 | ACCEPT | Summary: Propagation from PANTHER node PTN001377938, which is restricted to the ACTL7 clade (taxon constraint 32524) and seeded by four proteins, all resolved: mouse Actl7b (Q9QY83, the true orthologue, with its own GO:0005737 IDA from PMID:12672658), mouse Actl7a, rat Actl7a and human ACTL7A. This is the best-founded row in the set, because the inference is orthologue-anchored and does not leave the family, and it agrees with the primary literature, where the bulk of ACTL7B is cytoplasmic in round and elongating spermatids and its reported structures (developing acrosome, flagellum connecting region) are cytoplasmic. Reason: Correct compartment, sound propagation, orthologue-anchored, and the compartment in which the demonstrated function, binding DYNLL1 and DYNLL2 and controlling their distribution during spermatid remodelling, is exercised, so is_active_in is appropriate here. The only criticism is that the term is uninformatively broad; the specific children supported by the mouse orthologue are proposed as new annotations below rather than by modifying this row. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:1343053 · mouse Actl7b (Q9QY83), true orthologue SUPPORTS TRANSFER Swiss-Prot, 86.7% identical to human ACTL7B, with GO:0005737 IDA from PMID:12672658 (T-ACTIN 1 found in the cytoplasm). Orthologue-strength support. UniProtKB:Q9Y615 · human ACTL7A (paralogue) SUPPORTS TRANSFER Paralogue with its own GO:0005737 IDA; adds no independent weight beyond the orthologue but is not misleading for a compartment this generic. MGI:MGI:1343051 · mouse Actl7a (paralogue) SUPPORTS TRANSFER Own EXP and IDA for cytoplasm (PMID:12672658, PMID:21278383, PMID:36734600). RGD:1304697 · rat Actl7a (Q641W9, paralogue) SUPPORTS TRANSFER Swiss-Prot, own EXP for cytoplasm. PANTHER:PTN001377938 · ACTL7-clade PANTHER node (taxon constraint 32524) SUPPORTS TRANSFER Clade-restricted node holding GO:0005737 by IBD from the four ACTL7 seeds, the right granularity for this family, in contrast to the pan-eukaryotic node used for the nucleus row. Supporting Evidence: PMID:12672658 the cellular locations of these two proteins are quite different (T-ACTIN-1 was found in the cytoplasm and T-ACTIN-2 was located in the nucleus) PMID:37800308 ACTL7B is detected in the cytoplasm and at lesser amounts in the nucleus of round and elongating spermatids |
| GO:0005198 structural molecule activity | IBA GO_REF:0000033 | ACCEPT | Summary: The WITH/FROM field contains no proteins, only two PANTHER tree nodes, so the row cannot be judged without opening the PAINT table for PTHR11937. Doing so is what makes this row informative. Node PTN000940351 carries GO:0005200 "structural constituent of cytoskeleton" by IBD, seeded from ten resolved proteins: conventional actins (human ACTB, mouse and rat cytoplasmic actin 2, yeast ACT1, two Dictyostelium actins), the Arp2/3 subunits ACTR2 and ACTR3, and the yeast dynactin-related ARPs ARP1/centractin and ARP10. All ten are Swiss-Prot. Node PTN008986528, which is ancestral to ACTL7B, carries two annotations from that parent node: GO:0005198 by IBA, and GO:0005200 as an IRD with negated=true. In other words the PAINT curators inspected this clade, explicitly refused to propagate the specific cytoskeletal constituent term into it, and let only the generic parent through. The residue-level analysis performed for this review reaches the same verdict from structure: the actin filament interface is not conserved in ACTL7B (26 of 74 contact positions identical, 34 non-conservative changes, the D-loop segment 38-45 entirely replaced), so ACTL7B is not a filament-forming actin and cannot be assigned the constituent term on fold grounds. Reason: This is the deliberately generalised term that PAINT arrived at after rejecting its specific child, and it is the only molecular function ACTL7B can currently carry. Every phenotype attributed to the protein is architectural, namely detached acrosomes, disrupted membranes, malformed flagella and sperm heads, and mislocalised dynein light chains, with no catalytic activity and no measured binding affinity for the human protein, so "structural molecule activity" is the honest level. Accepted with the explicit caveat that it must not be re-specialised to GO:0005200 without an experiment, and with a more informative molecular function (dynein light chain binding) proposed below rather than by modifying this row. For contrast, the precomputed affinage record proposes GO:0008092 cytoskeletal protein binding as this gene's molecular activity; that is the same fold-to-activity leap, and no actin-binding or cytoskeletal-protein-binding measurement exists for ACTL7B. Propagation Review Root cause: NO FAILURE CORE Failure modes: GRANULARITY MISMATCH Sources checked: PANTHER:PTN008986528 · PANTHER node ancestral to ACTL7B SUPPORTS TRANSFER Holds GO:0005198 by IBA and, crucially, GO:0005200 as an IRD with negated=true, a documented curatorial rejection of the specific term for this clade. This node is the reason the row is a generic parent rather than the cytoskeletal constituent term. PANTHER:PTN000940351 · PANTHER node for conventional actins and Arp2/Arp3 SUPPORTS SOURCE BUT NOT TARGET Carries GO:0005200 by IBD seeded from ten Swiss-Prot proteins: human ACTB, mouse and rat cytoplasmic actin 2, yeast ACT1, two Dictyostelium actins, the Arp2/3 subunits ACTR2 and ACTR3, and the yeast dynactin ARPs ARP1/centractin and ARP10. All are genuine experimental annotations, so the node is sound at the source; the specific term was correctly blocked from descending into the ACTL7 clade. Supporting Evidence: file:human/ACTL7B/ACTL7B-bioinformatics/RESULTS.md | PTN008986528 | GO:0005200 | F | IRD (NEGATED) | PANTHER:PTN000940351 | PMID:36617158 KO mice were infertile, with severe and variable oligoteratozoospermia (OAT) and multiple morphological abnormalities of the flagellum (MMAF) and sperm head. file:human/ACTL7B/ACTL7B-deep-research-affinage.md - **molecular_activity:** GO:0008092 cytoskeletal protein binding |
| GO:0005856 cytoskeleton | IEA GO_REF:0000044 | MARK AS OVER ANNOTATED | Summary: Keyword-driven electronic annotation. The WITH/FROM token is not a gene but UniProtKB-SubCell:SL-0090, and that keyword comes from the single subcellular-location line in the UniProt entry, which is flagged ECO:0000250, by similarity. So the chain runs: ACTL7B has the actin fold, therefore UniProt infers it is cytoskeletal, therefore the keyword mapping asserts GO:0005856. No observation enters the chain at any point. The inference does not survive checking against what ACTL7B is actually reported to localise to. The compartments described in mouse, namely acrosomal vesicle (GO:0001669), the region connecting flagellum to head (GO:0120212 sperm head-tail coupling apparatus) and nucleus, are none of them descendants of GO:0005856 in the ontology. The paralogue ACTL7A is different in exactly this respect: it carries GO:0033011 perinuclear theca by IDA, and perinuclear theca is a descendant of cytoskeleton, so the cytoskeleton annotation is defensible for ACTL7A and not for ACTL7B. Independently, the filament interface analysis argues against ACTL7B being an F-actin subunit. Reason: Not demonstrably false, since a sperm cytoskeletal association may yet be found and the knockout phenotype is architectural, but as it stands the annotation asserts a compartment that no observation supports, derived from a by-similarity statement about an actin fold whose polymerisation surface ACTL7B has lost. Marked as over-annotated rather than removed because the term is broad enough that a future demonstration (for example in the perinuclear theca, as for ACTL7A) would rescue it. The informative and evidence-backed compartments are proposed as new annotations below. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK COMPARTMENT OR COMPLEX MISMATCH Sources checked: UniProtKB-SubCell:SL-0090 · UniProt subcellular location keyword "Cytoskeleton" SOURCE WEAK OR INFERRED A controlled-vocabulary term, not a gene product. It is attached to Q9Y614 only through the CC line "SUBCELLULAR LOCATION: Cytoplasm, cytoskeleton {ECO:0000250}", that is, by similarity to conventional actin, with no supporting experiment in the entry. Supporting Evidence: file:human/ACTL7B/ACTL7B-uniprot.txt SUBCELLULAR LOCATION: Cytoplasm, cytoskeleton {ECO:0000250}. file:human/ACTL7B/ACTL7B-bioinformatics/RESULTS.md | 38-45 | 8 | PRHQGVMV | RCPEAADA | 0/8 | 1/8 | 8/8 | |
| GO:0007010 cytoskeleton organization | IEA GO_REF:0000108 | REMOVE | Summary: An inter-ontology logical inference: the WITH/FROM field is literally the GO term GO:0005200, so this biological process annotation exists only because the molecular function "structural constituent of cytoskeleton" is on the gene. That molecular function is the legacy ProtInc TAS reviewed below, traced to a 1999 cloning paper that performed no protein-level experiment, and it is the very term that GO's own phylogenetic pipeline has negated for the clade containing ACTL7B (IRD at node PTN008986528). The inference step itself is mechanically correct; its input is not. No independent support exists for ACTL7B organising a cytoskeletal structure. The knockout phenotypes are spermatid morphogenesis failure, better captured by GO:0007286 spermatid development, and altered positioning of DYNLL1 and DYNLL2, which is protein localisation rather than cytoskeleton assembly. Reason: This annotation is entirely derivative of an annotation that should not exist. Remove the GO:0005200 TAS and this row has no basis at all; leave it in place and GO continues to publish a biological process for ACTL7B that rests on the 1999 naming of the gene as actin-like. Nothing is lost by removal, because the processes ACTL7B is genuinely required for are proposed explicitly below. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK FUNCTIONAL DIVERGENCE Sources checked: GO:0005200 · source term "structural constituent of cytoskeleton" on Q9Y614 (TAS, PMID:10373328) SOURCE WEAK OR INFERRED Not a gene product but the GO term the inference consumes. It is statement-level (TAS) evidence from a paper containing no protein-level experiment, and PAINT carries the same term as an IRD with negated=true for the ancestral node of this clade. An inference cannot be stronger than its input. Supporting Evidence: file:human/ACTL7B/ACTL7B-bioinformatics/RESULTS.md | PTN008986528 | GO:0005200 | F | IRD (NEGATED) | PANTHER:PTN000940351 | |
| GO:0005515 protein binding | IPI PMID:32814053 Interactome Mapping Provides a Network of Neurodegenerative ... | MARK AS OVER ANNOTATED | Summary: Partner resolved as HIP1 (huntingtin-interacting protein 1, O00291, Swiss-Prot), from a systematic yeast two-hybrid matrix of about 500 neurodegeneration-related baits that generated roughly 30,000 candidate interactions across about 5,000 proteins. HIP1 is the one of the four partners for which a mechanism is at least conceivable, because it carries a talin-like actin-binding module and could in principle engage an actin-fold surface. But there is no orthogonal assay for this pair, no reported testis context, and no follow-up: a PubMed search of the gene symbol returns 14 records, none of which revisits this pair. The bait was chosen for a brain-disease network in which ACTL7B, a spermatid-restricted protein, has no established presence. Reason: A single unreplicated high-throughput two-hybrid hit, annotated to the least informative molecular function term in the ontology. It is not evidence of a biological interaction at ACTL7B's own scale of expression, and "protein binding" records nothing about what ACTL7B does even if the contact is real. Flagged rather than removed because the underlying IntAct record is a genuine experimental observation and HIP1 is the one partner worth an actual co-immunoprecipitation in testis. Supporting Evidence: PMID:32814053 candidate interactions and is generated by systematic yeast two-hybrid interaction screening of ∼500 ND-related proteins |
| GO:0005515 protein binding | IPI PMID:32814053 Interactome Mapping Provides a Network of Neurodegenerative ... | MARK AS OVER ANNOTATED | Summary: Partner resolved as isoform 2 of LAMP2 (P13473-2, LAMP2B), a lysosomal membrane protein, from the same neurodegeneration two-hybrid matrix. LAMP2B is a single-pass membrane protein whose cytosolic tail is a short peptide, so a two-hybrid interaction with a 415-residue actin-fold protein has no interpretable structural basis, and none was proposed. No orthogonal assay, no testis context, no follow-up. Reason: Unreplicated two-hybrid hit on an uninformative term, with no mechanistic or expression rationale connecting a lysosomal membrane protein to a spermatid actin-related protein. Retained but flagged, since the IntAct record itself is a real observation and this review cannot see the primary data. Supporting Evidence: PMID:32814053 candidate interactions and is generated by systematic yeast two-hybrid interaction screening of ∼500 ND-related proteins |
| GO:0005515 protein binding | IPI PMID:32814053 Interactome Mapping Provides a Network of Neurodegenerative ... | MARK AS OVER ANNOTATED | Summary: Partner resolved as CASP6 (caspase-6, P55212, Swiss-Prot), one of the neurodegeneration baits of the two-hybrid screen and the partner recorded in UniProt as "Q9Y614; P55212: CASP6; NbExp=3; IntAct=EBI-25835070, EBI-718729;". A two-hybrid hit between a protease and a folded prey does not distinguish an enzyme-substrate encounter from an adventitious contact, and caspase-6 has no reported connection to ACTL7B or to spermiogenesis. Reason: Unreplicated two-hybrid hit from a disease-network screen, annotated to an uninformative term. Note that the NbExp=3 recorded by IntAct counts experiments within the same publication, not independent methods, so it should not be read as replication. Supporting Evidence: file:human/ACTL7B/ACTL7B-uniprot.txt Q9Y614; P55212: CASP6; NbExp=3; IntAct=EBI-25835070, EBI-718729; PMID:32814053 candidate interactions and is generated by systematic yeast two-hybrid interaction screening of ∼500 ND-related proteins |
| GO:0005515 protein binding | IPI PMID:32814053 Interactome Mapping Provides a Network of Neurodegenerative ... | MARK AS OVER ANNOTATED | Summary: Partner resolved as RAN (P62826, Swiss-Prot), the small nuclear GTPase, from the same two-hybrid matrix. RAN is small, abundant and ubiquitously expressed, and the annotation carries no information about specificity or context. The one reason not to dismiss it outright is that ACTL7B does have a nuclear pool and a reported conserved putative nuclear localisation sequence, so a RAN-dependent import route is a testable hypothesis, but that hypothesis has not been tested, and a two-hybrid hit is not evidence for it. Reason: Unreplicated two-hybrid hit annotated to an uninformative term, with a plausible but untested biological reading. If RAN-mediated import were demonstrated the appropriate annotation would be a specific import-related term, not "protein binding". Supporting Evidence: PMID:32814053 candidate interactions and is generated by systematic yeast two-hybrid interaction screening of ∼500 ND-related proteins PMID:38464253 identifies a putative nuclear localization sequence conserved across mammalian ACTL7B |
| GO:0005200 structural constituent of cytoskeleton | TAS PMID:10373328 Cloning, mapping, and expression of two novel actin genes, a... | REMOVE | Summary: A ProtInc-era traceable author statement, dated 2003, tracing to the 1999 paper that cloned the gene. Reading that paper settles the matter: it reports cDNA selection and genomic sequencing, chromosomal mapping, Northern-blot tissue distribution and a mutation screen in familial dysautonomia patients. There is no protein-level experiment in it, no localisation study, and no cytoskeletal or polymerisation assay. The annotation is a restatement of the title's description of ACTL7A and ACTL7B as "two novel actin genes", a fold name entering GO as a molecular function. Three independent lines converge against it. First, GO's own phylogenetic curation has explicitly rejected this term for the clade containing ACTL7B: node PTN008986528 carries GO:0005200 as an IRD with negated=true and propagates only the generic parent GO:0005198, so GOA currently publishes for ACTL7B a term that PAINT has decided does not belong to its clade. Second, the structural requirement for the term is absent: of the 74 protomer-protomer contact positions computed from the F-actin filament, only 26 are identical in ACTL7B and 34 differ non-conservatively, and the largest contact segment, actin 38-45, the DNase-I-binding D-loop that forms actin's principal longitudinal contact, is entirely replaced (PRHQGVMV to RCPEAADA, 0 of 8 identical). At 35% interface identity ACTL7B sits with Arp2 and the nuclear ARPs, not with any conventional actin. Third, this row is the sole input to the spurious GO:0007010 IEA. This is not the mirror-image error of asserting "fold without function" without checking: the nucleotide pocket was checked too, and it is only partly degenerate (13 of 20 contacts retained, including the catalytic Gln137), which is why no nucleotide or ATP-binding claim is made or contested here. Reason: The evidence code is a statement, not an experiment, and the statement traced to is the gene's name. TAS is not an experimental code, so removing it does not overrule a curator who read data; the paper is cached and its abstract shows what it contains. GO has already reached the same conclusion for this clade through PAINT, and the structural analysis shows the fold-based route to the term is closed because the polymerisation surface is not conserved. Removing it also removes the basis of the GO:0007010 inference. The defensible level of this claim, GO:0005198 structural molecule activity, is already present on the gene by IBA and is accepted above, so no information is lost. Supporting Evidence: PMID:10373328 Two novel human actin-like genes, ACTL7A and ACTL7B, were identified by cDNA selection and direct genomic sequencing file:human/ACTL7B/ACTL7B-bioinformatics/RESULTS.md | 38-45 | 8 | PRHQGVMV | RCPEAADA | 0/8 | 1/8 | 8/8 | file:human/ACTL7B/ACTL7B-bioinformatics/RESULTS.md | ACTL7B (Homo sapiens) | query (human ACTL7B) | 42.9 | 26 | 13 | 34 | 1 | 35.1 | 52.7 | file:human/ACTL7B/ACTL7B-bioinformatics/RESULTS.md | PTN008986528 | GO:0005200 | F | IRD (NEGATED) | PANTHER:PTN000940351 | |
| GO:0015629 actin cytoskeleton | TAS PMID:10373328 Cloning, mapping, and expression of two novel actin genes, a... | REMOVE | Summary: The second ProtInc traceable author statement from the same 1999 cloning paper, and the same problem in the cellular component aspect. The paper performed no localisation experiment of any kind; its expression data are Northern blots of tissue RNA. GO:0015629 is defined as the part of the cytoskeleton "composed of actin and associated proteins", so the annotation asserts both that a structure made of actin exists at ACTL7B's location and that ACTL7B is in it. Neither has been shown. No published work reports ACTL7B binding F-actin or colocalising with an actin network, and the filament interface analysis argues it is not an actin filament subunit. The compartments actually reported for ACTL7B in mouse, namely developing acrosome, sperm head-tail coupling region and nucleus, are not actin cytoskeleton. The distinction from the paralogue matters and is not symmetric: ACTL7A is reported to support subacrosomal F-actin and carries perinuclear theca by IDA, so actin-related compartment claims have some footing there. ACTL7B has none of that evidence. Reason: Statement-level evidence from a paper with no localisation data, asserting membership of an actin-composed structure for a protein that has lost actin's polymerisation interface and has never been shown to associate with F-actin. This is the cellular component twin of the GO:0005200 defect and should go for the same reasons. The broader GO:0005856 cytoskeleton row is retained but flagged, so no compartment claim is silently strengthened by this removal. Supporting Evidence: PMID:10373328 Two novel human actin-like genes, ACTL7A and ACTL7B, were identified by cDNA selection and direct genomic sequencing PMID:36617158 In mice, ACTL7B specifically localises to the developing acrosome, within the nucleus of early spermatids, and to the flagellum connecting region. file:human/ACTL7B/ACTL7B-bioinformatics/RESULTS.md | 38-45 | 8 | PRHQGVMV | RCPEAADA | 0/8 | 1/8 | 8/8 | |
| GO:0045503 dynein light chain binding | ISS PMID:37800308 Actl7b deficiency leads to mislocalization of LC8 type dynei... | NEW | Summary: Proposed new annotation. This is the only informative molecular function ACTL7B has, and GO records none of it. In mouse, anti-ACTL7B beads pulled DYNLL1 and DYNLL2 out of whole wild-type testis extract, a reciprocal pull-down with anti-DYNLL2 beads recovered ACTL7B, and a GFP pull-down from cells expressing ACTL7B-eGFP enriched the same partners. That is endogenous, native-tissue, reciprocal and orthogonally corroborated evidence, qualitatively different from the four two-hybrid rows above. Human ACTL7B is 86.7% identical to mouse Actl7b over the whole chain, so ISS from the orthologue is the appropriate evidence code for the human gene. Reason: GOA currently gives ACTL7B four "protein binding" rows for unreplicated two-hybrid hits from a neurodegeneration screen, and nothing at all for the LC8 dynein light chains, which are the only partners supported by native-tissue co-immunoprecipitation in both directions plus a functional consequence in the knockout. Adding GO:0045503 replaces an uninformative picture with a specific one. The mouse record should carry the corresponding IPI annotation from this paper; it currently does not. Supporting Evidence: PMID:37800308 In the co-immunoprecipitation using the anti-ACTL7B-coupled beads, we identified LC8 light chains, dynein light chain 1 (DYNLL1) and its paralog dynein light chain 2 (DYNLL2). PMID:37800308 In the eluate of the co-immunoprecipitation using anti-DYNLL2-coupled beads, DYNLL2 and ACTL7B were detected, further supporting the ACTL7B-DYNLL2 interaction. PMID:37800308 To identify ACTL7B-protein interactions, anti-ACTL7B antibody was coupled to Dynabeads and used for co-immunoprecipitation on protein extracts from whole wild-type testes. |
| GO:0007286 spermatid development | ISS PMID:36617158 Actin-related protein ACTL7B ablation leads to OAT with mult... | NEW | Summary: Proposed new annotation. ACTL7B has no biological process annotation at all except the GO:0007010 row recommended for removal. Two independent mouse knockout lines, published in 2023 by different groups, agree that Actl7b-null males are infertile with a spermiogenesis defect: severe oligoteratozoospermia with multiple morphological abnormalities of the flagellum and sperm head in one, and detached acrosomes, disrupted membranes and flagellar malformations appearing from spermiogenesis step 9 followed by spermatogenic arrest in the other. Expression is restricted to round and elongating spermatids in both mouse and human. Reason: This is the process ACTL7B is required for, established by two independent knockouts, and it is missing from GOA while a fold-derived "cytoskeleton organization" inference occupies the biological process aspect. Proposed as ISS from mouse Actl7b because the knockouts are murine; the mouse record itself should carry IMP from both papers. Supporting Evidence: PMID:36617158 KO mice were infertile, with severe and variable oligoteratozoospermia (OAT) and multiple morphological abnormalities of the flagellum (MMAF) and sperm head. PMID:36617158 In conclusion, this work identifies ACTL7B as a key regulator of spermiogenesis that is required for male fertility. PMID:37800308 Severe spermatid defects, such as detached acrosomes, disrupted membranes and flagella malformations start to appear after spermiogenesis step 9 |
| GO:0001669 acrosomal vesicle | ISS PMID:36617158 Actin-related protein ACTL7B ablation leads to OAT with mult... | NEW | Summary: Proposed new annotation. The mouse orthologue localises specifically to the developing acrosome during spermiogenesis, and the independent knockout line shows detached acrosomes among the earliest defects, which is consistent with the protein being present at that structure. This is the informative cellular component that the generic cytoplasm and cytoskeleton rows fail to capture. Reason: GOA gives ACTL7B only cytoplasm, nucleus, cytoskeleton and actin cytoskeleton, of which two are recommended for removal or flagging. Acrosomal vesicle is the specific, evidence-backed compartment, transferred by ISS from the mouse orthologue. Note that GO:0001669 is not a descendant of GO:0005856, which is part of why the cytoskeleton row is marked as over-annotated. Supporting Evidence: PMID:36617158 In mice, ACTL7B specifically localises to the developing acrosome, within the nucleus of early spermatids, and to the flagellum connecting region. PMID:37800308 Severe spermatid defects, such as detached acrosomes, disrupted membranes and flagella malformations start to appear after spermiogenesis step 9 |
| GO:0032880 regulation of protein localization | ISS PMID:37800308 Actl7b deficiency leads to mislocalization of LC8 type dynei... | NEW | Summary: Proposed new annotation, and the weakest of the four proposals, offered because it is the one process term that reflects a measured molecular consequence rather than a morphological endpoint. In Actl7b-null testis, DYNLL1 and DYNLL2 are mislocalised while their protein levels are unchanged, and conversely expressing ACTL7B in HEK cells redistributes both light chains. Loss of function and gain of function therefore agree that ACTL7B influences where the LC8 light chains are. Reason: The knockout phenotype most directly attributable to ACTL7B's one solid interaction is a protein-localisation phenotype, and GO has no term on this gene expressing it. Proposed at the regulation-of-protein-localization level rather than something more specific because the mechanism, whether ACTL7B tethers, transports or excludes the light chains, is not resolved. Supporting Evidence: PMID:37800308 Taken together, these results suggest that the localization of DYNLL1 and DYNLL2 is altered in the absence of ACTL7B, whereas the amount of protein is unchanged. PMID:37800308 This supports the notion that ACTL7B interacts with and controls DYNLL1 and DYNLL2 localization in the cell. |
Loading supporting content…
Download this section (compressed HTML)Q: Does ACTL7B bind ATP or ADP? The nucleotide pocket retains the P-loop serine and lysine and the catalytic Gln137 but has lost G15, D157, R210, T303 and Y306, placing it between the nucleotide-binding cytoplasmic ARPs and the nucleotide-poor nuclear ARPs. No binding or hydrolysis assay has ever been reported for any ACTL7-family protein.
Suggested experts: Clement TM, Witke W
Q: Is the interaction with DYNLL1/DYNLL2 direct, and is ACTL7B the cargo or the adaptor? The co-immunoprecipitations establish association in native testis in both directions, but not directness, stoichiometry, or which partner positions the other.
Suggested experts: Merges GE, Schorle H, Witke W
Q: Does ACTL7B associate with F-actin at all? It has lost the polymerisation interface, but the paralogue ACTL7A is reported to support subacrosomal F-actin, so the two paralogues may differ on exactly this point. A direct actin-binding measurement would settle whether any actin-related GO term belongs on ACTL7B.
Suggested experts: Clement TM, Eddy EM
Q: Does the nuclear pool of ACTL7B occupy an actin-related-protein position in an INO80 or SWI/SNF complex, as computational docking predicts? A pull-down of ACTL7B from spermatocyte nuclei followed by mass spectrometry would answer this directly, and would decide whether the is_active_in nucleus qualifier is justified.
Suggested experts: Ferrer P, Clement TM
Q: Should the GO:0005634 nucleus annotation carry located_in rather than is_active_in? No molecular function has been demonstrated for the nuclear pool, so the current qualifier asserts more than the evidence supports.
Q: Curation debt in the mouse record: MGI:1343053 / Q9QY83 has only five GO annotations and nothing from either 2023 knockout paper. Should the mouse gene carry IMP for spermatid development and IPI for DYNLL1/DYNLL2 from PMID:36617158 and PMID:37800308? The human ISS annotations proposed here depend on those.
Q: Curation debt in UniProt: Q9Y614 still has no FUNCTION comment despite two knockout papers, and its only subcellular-location line is ECO:0000250 by similarity. It also records four IntAct interactions without any indication that they all come from a single neurodegeneration-focused two-hybrid screen of a protein not expressed in brain.
Q: Are the two legacy ProtInc TAS annotations from PMID:10373328 worth a systematic sweep? The same 1999 paper supplies GO:0005200 and a cytoskeletal component term to both ACTL7B and ACTL7A, so this is a reproducible artefact of one cloning paper rather than a single bad row.
Experiment: Measure nucleotide binding and hydrolysis on recombinant human ACTL7B, using isothermal titration calorimetry or fluorescent nucleotide exchange for binding and a phosphate-release assay for hydrolysis, with an Arp2/3 subunit as a positive comparator and a nuclear ARP as a negative one. This is the cheapest experiment that would resolve the largest ontological ambiguity about the family, since it determines whether nucleotide-binding molecular function terms could ever apply.
Hypothesis: ACTL7B has retained enough of the actin nucleotide pocket to bind ATP or ADP, or alternatively has joined the nuclear ARPs in losing nucleotide binding altogether.
Type: biophysical nucleotide binding and ATPase assay
Experiment: Test polymerisation and actin co-assembly directly with pyrene-actin polymerisation in the presence of ACTL7B, sedimentation of ACTL7B with F-actin, and a chimera in which the ACTL7B segment corresponding to actin 38-45 is replaced by the actin D-loop, to ask whether the interface is the reason for any negative result.
Hypothesis: ACTL7B cannot polymerise or co-assemble into actin filaments, because the D-loop and the wider protomer-protomer interface are not conserved.
Type: in vitro polymerisation and co-sedimentation
Experiment: Map the interaction with recombinant pull-downs to establish directness, mutate the candidate LC8-binding motif in ACTL7B, and generate a knock-in mouse carrying that separation-of-function mutation to ask whether loss of light-chain binding alone reproduces the null phenotype.
Hypothesis: The ACTL7B-DYNLL1/DYNLL2 interaction is direct and is the proximate cause of the spermatid phenotype, rather than a downstream consequence of spermatid collapse.
Type: interaction mapping plus separation-of-function knock-in mouse
Experiment: Immunoprecipitate ACTL7B separately from purified spermatid nuclei and from cytoplasmic fractions and identify partners by mass spectrometry, to test the ARP-substitution hypothesis and to find the structural partners in the acrosome and head-tail coupling region. Compartment-resolved interactomes are needed because the two pools are predicted to have different partners.
Hypothesis: Nuclear ACTL7B occupies an actin-related-protein subunit position in INO80 or SWI/SNF-family nucleosome remodellers, while cytoplasmic ACTL7B works with dynein light chains and acrosomal partners.
Type: compartment-resolved immunoprecipitation mass spectrometry
Experiment: Screen ACTL7B in cohorts of men with idiopathic oligoteratozoospermia and multiple morphological abnormalities of the flagellum, and stain human spermatids for ACTL7B, to replace the current inference-only annotation set with human evidence.
Hypothesis: Loss-of-function variants in human ACTL7B cause oligoteratozoospermia and multiple morphological abnormalities of the flagellum, as they do in mouse.
Type: human genetics cohort screen with immunohistochemistry
Experiment: Test whether spermatid-expressed ACTL7A rescues the Actl7b-null phenotype, and the converse. The paralogues differ in expression breadth, compartment and phenotype, and four of the eight protein-level WITH/FROM tokens behind ACTL7B's cellular component annotations are ACTL7A orthologues, so a rescue experiment would tell curators whether such transfers are ever safe.
Hypothesis: ACTL7A and ACTL7B are not functionally interchangeable, despite sharing around 59% sequence identity and a common ancestor.
Type: transgenic cross-rescue in mouse
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: Every functional statement about ACTL7B comes from mouse. There is no human loss-of-function evidence, no human localisation study, and no biochemical assay on the human protein; the only human-specific data are transcript and protein detection in testis and a polymorphism screen that could not establish causality.
OPEN BIOLOGYCURATION MF_DARK
What is known: The human protein exists and is testis-enriched (UniProt PE 1, HPA tissue-enriched testis), and the mouse orthologue is 86.7% identical over the whole chain, so ISS transfer is reasonable and is what this review proposes. The gap is that nothing about the human protein has been measured directly, so every proposed annotation is an inference across species.
Significance: ACTL7B knockout phenocopies two leading causes of idiopathic human male infertility, so the absence of human data is the obstacle to using the gene diagnostically.
Gap: Whether ACTL7B binds a nucleotide is unknown. The pocket is intermediate: 13 of 20 nucleotide contacts computed from the F-actin structure are conserved, including the P-loop serine and lysine and the catalytic Gln137, but G15, D157, R210, T303 and Y306 are not. No nucleotide-binding or hydrolysis assay has ever been performed on ACTL7B or on any ACTL7-family protein.
OPEN BIOLOGY MF_DARK
What is known: This is a genuinely open question, not a hidden negative. The conservation level (65%) sits between the ATP-binding cytoplasmic ARPs Arp2 and Arp3 (85% and 70%) and the nucleotide-poor nuclear ARPs (35-50%), so the sequence alone cannot decide it. GOA does not annotate ATP binding for ACTL7B, so nothing is currently wrong; the point is that nothing is known either.
Significance: Whether the ACTL7 clade retains nucleotide binding determines whether these proteins can be conformationally regulated like actin or are static structural elements, which in turn determines what kind of molecular function terms could ever apply to them.
Gap: The nuclear pool of ACTL7B has a localisation but no activity. Intranuclear ACTL7B is documented in spermatocytes and round spermatids, a conserved putative nuclear localisation sequence has been described, HDAC1 and HDAC3 lose intranuclear localisation in its absence, and knockout testis shows transcriptional changes, but the only proposed mechanism, substitution for conventional nuclear ARPs on the HSA domains of INO80 and SWI/SNF remodellers, is a computational docking result in a preprint.
NARROWING BIOLOGY MF_DARK
What is known: The localisation is observed in vivo and the HDAC phenotype is a real knockout observation, which is why the nucleus annotation is kept. What does not exist is any measured interaction with a chromatin-remodelling complex, any demonstration that ACTL7B occupies an ARP position in one, or peer-reviewed publication of the modelling work.
Significance: If confirmed, germline-specific ARP substitution in nucleosome-remodelling complexes would be a new mechanism for the chromatin reorganisation of spermiogenesis. If not, the nuclear pool may be incidental, and GO would be carrying an is_active_in nucleus annotation for a protein with no nuclear activity.
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)