CFAP91 (MAATS1, AAT-1, C3orf15) review notes

Deep research status

• just deep-research human CFAP91 --provider falcon failed: Edison API authentication
  error ("Failed to authenticate", HTTP 403 Forbidden).
• just deep-research human CFAP91 --provider perplexity failed: provider not available
  in this environment ("Provider 'perplexity' not available. Available: falcon, asta,
  openscientist").
• No deep-research file was written. Literature research was done manually (PubMed search
  for CFAP91 / MAATS1 / FAP91 / AAT-1) using the cached publications in publications/.

Identity

• UniProt Q7Z4T9, Cilia- and flagella-associated protein 91; aliases MAATS1, AAT-1,
  C3orf15. CFAP91 family (PANTHER PTHR22455). Many isoforms from alternative promoters and
  splicing (AAT-1alpha/beta/gamma, AAT-1L/M/S).
• Disease: spermatogenic failure 51 (SPGF51, MIM:619177), autosomal recessive MMAF /
  astheno-teratozoospermia.

Literature findings (with provenance)

CSC and radial spoke base

• Chlamydomonas FAP91 was identified in the CaM-containing CSC; "FAP91 directly interacts
  with radial spoke protein 3 (an AKAP), which is located at the base of the spoke", and
  antibodies to FAP91 restored dynein activity in a sliding assay; "the CaM- and
  spoke-associated complex mediates regulatory signals between the radial spokes and
  dynein arms" PMID:17967944.
• CSC knockdown: "the CSC is required for spoke assembly and wild-type motility"; RS2
  assembly defective PMID:21613541.
• CSC cryo-ET: connects RS2, N-DRC and RS3S PMID:22740634.
• Tetrahymena CFAP91 knockout: "lack of Tetrahymena Cfap91 protein affects stable
  docking/positioning of the radial spoke RS3 and the base of RS2, and adjacent inner
  dynein arms"; Cfap61/Cfap251 diminished; "CFAP91-KO cells swam about a hundred times
  slower than wild-type cells"; "Tetrahymena Cfap91 localizes at the base of radial spokes
  RS2 and RS3" PMID:36552811 (abstract only).
• Human respiratory axoneme cryo-EM: CFAP91 "begins at the base of RS2, interacts with the
  N-DRC and then links the RS3 subunits CFAP251 and CFAP61" PMID:37258679.

Mouse and human sperm

• Martinez et al. 2020: six unrelated MMAF men with homozygous deleterious MAATS1
  variants; "severe central pair complex (CPC) and radial spokes defects"; WDR66 is "a
  physical and functional partner of CFAP91"; Trypanosoma TbCFAP91 is axonemal and its
  knockdown "impaired flagellar movement led to CPC defects" PMID:32161152 (abstract
  only).
• Wang et al. 2025 (mouse): "Cfap91 knockout (KO) mice exhibit impaired sperm flagellum
  formation and male infertility"; CFAP91 co-immunoprecipitates with CFAP251 and LRRC23,
  which fail to localize in KO sperm flagella; authors "establish CFAP91 as an essential
  scaffolder of RS3 assembly" PMID:40931011 (full text). Ependymal and tracheal cilia
  elongated normally in KO mice ("ciliary elongation was endurable in Cfap91-/- mice in
  contrast to disrupted sperm flagellum formation"); hydrocephalus was investigated in a
  susceptible background.
• CCDC113 interacts with CFAP91 and is needed for its distribution on doublet
  microtubules PMID:39671309.

Older AAT-1 work (somatic cells, overexpression)

• AAT-1 was cloned as an AMY-1 (MYCBP)-binding protein; "AAT-1alpha was found to bind to
  and be colocalized in mitochondria with AMY-1 in human HeLa and mouse GC-1 cells"; forms
  a complex with AKAP1 (S-AKAP84/AKAP149) and PKA RII; phosphorylated by PKA
  PMID:12223483 (abstract only).
• Further isoforms: "AAT-1L and AAT-1M were found to be localized diffusely in the
  cytoplasm, AAT-1S, like AAT-1alpha, was found to be localized in the mitochondria-like
  structure" PMID:15863901 (abstract only).
• These are the source of the cytoplasm and mitochondrion EXP rows. They concern specific
  isoforms in transfected somatic cell lines; the endogenous full-length protein in
  sperm and airway cilia is axonemal. AKAP1 is itself a mitochondrial AKAP, so a fraction
  of some isoforms could be mitochondrial, but this has not been shown for endogenous
  CFAP91 in germ cells.

Curation judgements

• Cytoplasm (EXP x2, IEA): kept as non-core (isoform behaviour in transfected cells).
• Mitochondrion (EXP x2, IEA): marked as over-annotated; the claim (a mitochondrial
  protein) exceeds the evidence (overexpressed isoforms in HeLa/GC-1 cells).
• Protein binding (IPI, HuRI, DPY30): removed per protein-binding policy (no functional
  context; DPY30 interaction not otherwise supported).
• Spermatogenesis (IMP): MODIFY to GO:0120316 sperm flagellum assembly, the specific
  process the MMAF phenotype reports.
• Axonemal central apparatus assembly (IMP): CPC loss in patient sperm and Trypanosoma
  knockdown is a recurrent secondary feature of MMAF; CFAP91 is located on the doublets
  at the RS2/RS3 base, not in the central apparatus. Marked as over-annotated.
• Radial spoke stalk (colocalizes_with, ISS): MODIFY to GO:0120341 radial spoke base 2 and
  GO:0120342 radial spoke base 3.
• NEW: GO:0062177 radial spoke assembly (comparator LRRC23, an RS3 subunit that CFAP91
  co-immunoprecipitates with, carries it by IMP/IBA); GO:0005200 structural constituent of cytoskeleton (consistent with the CFAP61 review).

Review follow-up (2026-10-16, PR #5064 review)

• GO:0120316 sperm flagellum assembly ancestry, checked by an OLS hierarchicalAncestors
  query and the term graph endpoint (2026-10-16):
  https://www.ebi.ac.uk/ols4/api/ontologies/go/terms/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FGO_0120316/hierarchicalAncestors.
  Direct parents: is_a GO:0044458 motile cilium assembly; is_a GO:0003006 developmental
  process involved in reproduction; part_of GO:0030317 flagellated sperm motility;
  part_of GO:0007286 spermatid development. GO:0030317 is_a GO:0060294 cilium movement
  involved in cell motility, which is_a GO:0003341 cilium movement. GO:0007286 is part_of
  GO:0048515 spermatid differentiation, which is part_of GO:0007283 spermatogenesis.
  So GO:0120316 is a descendant of cilium movement, flagellated sperm motility and
  spermatogenesis; the PR reviewer's doubt about the cilium movement and spermatogenesis
  paths was not borne out.
• GO:0062177 radial spoke assembly comparators, QuickGO annotation search
  (/QuickGO/services/annotation/search?goId=GO:0062177&goUsage=exact&taxonId=9606,10090,
  2026-10-16, 16 hits): LRRC23 human Q53EV4 (IMP PMID:37804054, IBA, IEA) and mouse
  O35125 (IMP PMID:34585727 and PMID:38091523, IBA, IEA, ISO); RSPH4A human Q5TD94 (ISS,
  IEA) and mouse Q8BYM7 (IMP PMID:32203505); RSPH9 human Q9H1X1 (ISS, IEA) and mouse
  Q9D9V4 (IMP PMID:30383886). The LRRC23 claim (IMP and IBA in human and mouse) holds;
  RSPH4A and RSPH9 carry the term but not by IBA, so the review text was corrected to say
  so.
• No GO term exists for the calmodulin- and spoke-associated complex (OLS search of GO,
  2026-10-16: "calmodulin and spoke associated complex", "calmodulin- and spoke-associated
  complex", "CSC complex", "spoke-associated", "calmodulin complex" returned no CSC term).
  Added a proposed_new_terms entry with the same name as the one in the CFAP61 review, so
  in_complex stays unset. GO:0120341 radial spoke base 2 and GO:0120342 radial spoke base 3
  labels confirmed verbatim in OLS the same day.
• core_functions now records GO:0120342 radial spoke base 3 as in_complex. The radial
  spoke base terms are is_a GO:0032991 protein-containing complex, so validation rejects
  them in locations, and in_complex holds a single term; GO:0120341 radial spoke base 2
  stays in the radial spoke stalk MODIFY row and is named in the core function
  description. The core function also drops GO:0003341 cilium movement because GO:0120316 entails it. The
  spermatogenesis -> sperm flagellum assembly MODIFY is a refinement (descendant via
  spermatid development), as now stated in its reason.
• Isoform fields on the cytoplasm/mitochondrion rows were not set. UniProt ALTERNATIVE
  PRODUCTS maps AAT-1L to Q7Z4T9-7, AAT-1M to Q7Z4T9-1 and AAT1-beta to Q7Z4T9-5, but has
  no synonym for AAT-1alpha or AAT-1S, which are the isoforms behind the mitochondrion
  rows. The PMID:15863901 cytoplasm row covers two isoforms (AAT-1L and AAT-1M), and all
  six GOA rows are on the canonical accession (no isoform column value), so setting a
  single isoform would misstate the GOA row. Left as prose.
• Also corrected an older line above: the cytoplasm rows are TOO_GENERAL in the review
  file (entailed by axoneme), not KEEP_AS_NON_CORE.

2026-10-25 ORGANELLES batch 3 (mitochondrion rows)

Rows: GO:0005739 EXP PMID:12223483, EXP PMID:15863901, IEA GO_REF:0000044 (SL-0173). No
sub-compartment rows. All three were already MARK_AS_OVER_ANNOTATED; re-checked and kept.

• The deciding point is the isoform. PMID:15863901 (abstract only, cached): "While AAT-1L and
  AAT-1M were found to be localized diffusely in the cytoplasm, AAT-1S, like AAT-1alpha, was
  found to be localized in the mitochondria-like structure". UniProt ALTERNATIVE PRODUCTS
  [file:human/CFAP91/CFAP91-uniprot.txt "Name=7; Synonyms=AAT-1L {ECO:0000303|PubMed:15863901};"]
  shows isoform 7 is the displayed canonical sequence, i.e. the isoform GOA annotates is the
  one the abstract calls cytoplasmic. The mitochondrial isoforms (AAT-1alpha, AAT-1S) have no
  UniProt isoform mapping.
• PMID:12223483 (abstract only): "AAT-1alpha was found to bind to and be colocalized in
  mitochondria with AMY-1 in human HeLa and mouse GC-1 cells" - overexpressed short isoform.
• Endogenous protein is axonemal (sperm, airway cilia). The IEA keyword mapping is correct
  for the entry, so the IEA row is over-annotated rather than a wrong mapping (no REMOVE).
• Rule 4: no NEW row; the only sub-mitochondrial hint is AKAP1 (OMM) as a binding partner.
• No is_active_in rows.