ACTRT3

Q: Should the GO:0005200 rejection sweep in PANTHER family PTHR11937 be extended to the ARP-T branch, and if so should the general parent GO:0005198 be substituted as it was for ACTL7A and ACTL7B? GO:0005200 is asserted once in this family, at PTN000940351, and has since been explicitly negated at eight descendant nodes covering thirteen human genes - ACTR5, ACTR6, ACTR8, ACTL6A, ACTL6B, ACTL7A and ACTL7B on 2025-08-05, then ACTR2, ACTR3, ACTR3B, ACTR3C, ACTR1A and ACTR1B on 2026-04-16. The genes still receiving it from that node are the conventional actins ACTA1, ACTA2, ACTC1 and ACTG2, ACTR10 (which has its own orthologue in the seed set), and five un-adjudicated divergent genes: ACTL9, ACTL10, ACTRT1, ACTRT2 and ACTRT3. One node-level edit at the ancestor of that divergent set would settle all five human genes at once, following the template PAINT already used at PTN008986528, where the specific term was negated and GO:0005198 asserted in its place on the same day. Stated here once for all five genes rather than repeated per gene.

Q: Can GO:0005200 be re-earned by the perinuclear-theca proteins on evidence of their own, and if so should ACTL7A's removal be revisited at the same time? Read by its definition - the action of a molecule that contributes to the structural integrity of a cytoskeletal structure - GO:0005200 arguably fits the theca proteins directly, because GO places GO:0033011 perinuclear theca under GO:0005856 cytoskeleton and knockouts of ACTRT3 and ACTL7A both disrupt theca or acroplaxome structure. That is a different and better basis than the phylogenetic transfer from filament actins that currently carries the term. It cannot be settled gene by gene: ACTL7A is a theca protein with GO:0033011 by IDA whose merged review removed GO:0005200, so acting on the theca argument for ACTRT3 alone would make the two reviews inconsistent. The set to decide together is ACTRT1, ACTRT2, ACTRT3, ACTL7A and ACTL9. Note the complication that GO:0033011's own definition cautions that most theca constituents are not traditional cytoskeletal proteins.

Suggested experts: Hubert Schorle, Su-Ren Chen

Q: Should UniProt's tissue-specificity line for ACTRT3, "Ubiquitously expressed", be revised? The line is attributed to PMID:11750065, whose actual statement is that the mRNAs are expressed in all tested human tissues but in smaller amounts than actin - a 2001 transcript survey with an explicit abundance caveat and no protein data. Three sources cross-referenced from the same UniProt entry point the other way: HPA records "Tissue enriched (testis)" with testis the only tissue above its specificity threshold at 54.1 nTPM; Bgee leads with sperm; and in mouse the orthologue protein is expressed exclusively in the testis, particularly in haploid germ cells. A phrase such as "broadly expressed at low level, strongly enriched in testis; protein detected in male germ cells" would carry the same evidence without the misleading summary. This matters beyond UniProt: the merged ACTL8 review quotes the line to argue that the divergent actin clade is not uniformly testis-restricted, an inference the underlying data do not support.

Suggested experts: Masahiko Harata, Tadashi Baba

Q: Should the mouse Actrt3 GO:0005515 protein binding rows be upgraded to GO:0005522 profilin binding? Both rows carry WITH/FROM UniProtKB:Q9DAD6, which is PROF3_MOUSE (Pfn3), so the partner is already recorded as a profilin and the specific molecular-function term is available. Making the change on the mouse entry would let the human annotation arrive by the normal ortholog route rather than needing to be proposed here, and would remove one more uninformative protein-binding row from this family.

Suggested experts: Tadashi Baba

Q: Does ACTRT3 actually bind a nucleotide, and does it hydrolyse one? ACTRT3 has the joint best-preserved nucleotide site of the eight divergent human actin-like and actin-related-T proteins - 17 of 19 computed beta-actin ATP and cation contacts chemically compatible, with the entire phosphate-binding P1 loop, the magnesium-coordinating glutamine and the adenine contacts conserved - which was predicted from sequence in 2001 and is confirmed by measurement here. No nucleotide-binding term is proposed, because following the merged ACTR1A and ACTR10 reviews the term should name the ligand actually observed, and no structure of ACTRT3 exists and no binding has been assayed. The question is therefore genuinely open rather than settled either way, and answering it would decide whether the family-wide absence of nucleotide-binding annotation across the human actin set is a curation gap or a biological fact.

Suggested experts: Masahiko Harata

Q: Is the qualifier on the GO:0015629 IBA row correct, given that no molecular activity of ACTRT3 has been measured anywhere? The row uses is_active_in, which asserts that the gene product's molecular function is executed in the actin cytoskeleton. Nothing has been measured about ACTRT3's activity in any compartment, and its protomer interface argues against the filament role its donors have. located_in would state what the inference actually supports. The same observation applies to ACTL9, ACTL10, ACTRT1 and ACTRT2, which receive the identical row from the identical node.