Androglobin is the fifth and most structurally unusual mammalian globin: a large (1667 aa) chimeric protein in which an N-terminal calpain-like (peptidase C2) region is followed by a circularly permuted globin domain whose two halves are separated by an IQ calmodulin-binding motif, and then by a long C-terminal region ending in a coiled coil. The globin domain is a genuine heme protein - the recombinant human domain purifies with heme and shows a hexacoordinate ferrous spectrum - but it is not a bis-histidyl globin: the proximal F8 histidine (His824) is retained while the distal E7 position is a glutamine (Gln792). It binds nitric oxide as an unusual five-coordinate complex, shows nitrite reductase activity in the isolated globin domain (at millimolar K_M, so its physiological significance is untested) that is enhanced by calmodulin binding at the IQ motif, and autoxidises three orders of magnitude faster than neuroglobin, so its measured chemistry points to nitric oxide handling rather than to oxygen storage or transport. The calpain-like region, by contrast, is a degenerate protease fold: it retains a cysteine at the aligned nucleophile position but has lost the catalytic histidine and asparagine, and MEROPS classifies androglobin as a non-peptidase homologue of family C2. Androglobin is expressed predominantly in post-meiotic male germ cells and, more broadly, in motile-ciliated cell types, where its transcription is driven by the ciliogenic factors FOXJ1 and RFX2. It localises to the sperm flagellum, midpiece and annulus, binds calmodulin, septin 10, CFAP69, SPEF2, TTC29 and CFAP47, and in the ciliate central apparatus is a component of the C1b/C1f supercomplex. Loss of androglobin in mice causes male infertility with defective spermatid maturation, abnormal head shaping and disrupted manchette, annulus and flagellum organisation, and bi-allelic human variants cause asthenozoospermia and oligoasthenoteratozoospermia with acrosome, mitochondrial-sheath and axonemal defects.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0007286 spermatid development | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted as a core biological process. The propagation was checked in all three of the ways this campaign has found defects, and came back clean in each. (i) The donor MGI:MGI:3605549 resolves to UniProtKB G3UZ78, ADGB_MOUSE, Swiss-Prot reviewed, 1657 aa - the true 1:1 ortholog of the 1667-aa human protein, not a paralog. (ii) A paralog transfer is structurally impossible on this node: PANTHER PTHR46298 "Androglobin-like" has exactly two reviewed (Swiss-Prot) members, human ADGB and mouse Adgb, both in subfamily PTHR46298:SF1, so there is no divergent clade member that could donate the wrong biology. The family's cached total is 1670 proteins over 3237 taxa; the count of two is the reviewed subset and is stated as such. (iii) Asking not merely whether the donor carries its own evidence but which term it carries, QuickGO shows mouse Adgb holds GO:0007286 by its own IMP from PMID:36995441 - the identical term, neither above nor below the propagated one - so no downward MODIFY is warranted. Human patient genetics corroborates the transfer independently: bi-allelic ADGB variants cause asthenozoospermia and oligoasthenoteratozoospermia with malformed spermatids. Reason: Sole-ortholog donor carrying its own experimental annotation at exactly this term, in a two-member reviewed family with no paralog risk, and corroborated by human genetics. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:3605549 · mouse Adgb (UniProtKB G3UZ78, ADGB_MOUSE, reviewed/Swiss-Prot, 1657 aa) SUPPORTS TRANSFER True 1:1 ortholog, resolved via UniProt xref:mgi-3605549 using the bare numeric id (the doubled MGI:MGI: form returns HTTP 400). It carries its own IMP to this exact term from PMID:36995441, so the propagation neither over-reaches nor lands above its donor. PANTHER:PTN002922608 · PANTHER internal tree node in PTHR46298 (Androglobin-like) SUPPORTS TRANSFER An internal tree node, not a protein. Recorded as such rather than as unresolvable. Note the two CC rows on this gene use a different node, PTN002922607, so PAINT placed the BP and CC terms at separate nodes within a family whose reviewed membership is only human and mouse ADGB. Supporting Evidence: PMID:35700329 it locates to the sperm flagellum, the annulus, and the midpiece, and that it is crucial for male fertility and sperm formation PMID:38385883 the spermatozoa from patients harboring ADGB variants showed multiple acrosome and flagellum malformations under light and electron microscopy |
| GO:0097225 sperm midpiece | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted. Same sole-ortholog donor as the spermatid-development row, and the donor holds GO:0097225 by its own IDA from PMID:35700329 - the same term, not a parent, so the ACRV1-style "propagation landed above its donor" defect is absent here. The mouse study localised Adgb along the whole flagellum including the midpiece by immunofluorescence in stage-12 and stage-15 spermatids and in epididymal sperm, and UniProt records the same placement. The donor's qualifier is located_in while PAINT uses is_active_in; that is the normal PAINT convention for a component where the gene product acts and is not worth a MODIFY on its own. Reason: Ortholog donor with its own IDA at precisely this term; localisation independently recorded in UniProt and reproduced in human spermatogenic cells. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:3605549 · mouse Adgb (UniProtKB G3UZ78, ADGB_MOUSE, reviewed/Swiss-Prot, 1657 aa) SUPPORTS TRANSFER Carries GO:0097225 by IDA from PMID:35700329, the identical term. Checked for the ACTR8-style complex projection: that reference annotates 8 annotations over only 2 distinct entities (Adgb and Sept10) and its phenotype terms stay on the perturbed gene, so it is parallel per-protein curation, not a projection. PANTHER:PTN002922607 · PANTHER internal tree node in PTHR46298 (Androglobin-like) SUPPORTS TRANSFER Internal tree node, not a protein. Shared with the sperm-annulus IBA row. Supporting Evidence: PMID:35700329 it locates to the sperm flagellum, the annulus, and the midpiece, and that it is crucial for male fertility and sperm formation file:human/ADGB/ADGB-uniprot.txt along the whole sperm flagellum. Detected in the annulus of the sperm |
| GO:0097227 sperm annulus | IBA GO_REF:0000033 | ACCEPT | Summary: Accepted. The donor holds GO:0097227 by its own IDA from PMID:35700329, again the exact term. This is the most specific and most informative of the three localisations: mouse Adgb and Sept10 co-localise in the annulus of stage-15 spermatids and mature sperm, and loss of Adgb mislocalises Sept10, which is the mechanistic link between the localisation and the manchette/annulus phenotype. UniProt records the same annulus localisation. Reason: Ortholog donor with its own IDA at precisely this term, and the annulus is where the gene's best-characterised interaction and phenotype are located. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:3605549 · mouse Adgb (UniProtKB G3UZ78, ADGB_MOUSE, reviewed/Swiss-Prot, 1657 aa) SUPPORTS TRANSFER Carries GO:0097227 by IDA from PMID:35700329, the identical term. PANTHER:PTN002922607 · PANTHER internal tree node in PTHR46298 (Androglobin-like) SUPPORTS TRANSFER Internal tree node, not a protein. Shared with the sperm-midpiece IBA row. Supporting Evidence: file:human/ADGB/ADGB-uniprot.txt along the whole sperm flagellum. Detected in the annulus of the sperm |
| GO:0004198 calcium-dependent cysteine-type endopeptidase activity | IEA GO_REF:0000002 | REMOVE | Summary: Removed. This is the textbook case of a domain's name being propagated into an activity, and the removal is argued positively rather than merely asserted. Three independent lines converge, and two of them do not depend on my alignment at all. First, the residues. Aligning the catalytic domains of six reviewed reference calpains (CAPN1, CAPN2, CAPN3, CAPN7, CAPN10, CAPN15) to ADGB 70-411 under four substitution and gap models, and reading the ADGB residue in each column anchored on the reference's own UniProt Active site feature, gives a sharply split result. The nucleophile cysteine is retained: 11 of 15 aligned columns have a Cys in ADGB and it is always the same residue, C132, with the ADGB position invariant even though which reference reaches it varies by model. The general base histidine and the asparagine are not: 1 of 22 and 0 of 22 respectively, scattered over seven and six different ADGB positions, and the single His hit appears under one model only at a position no other model selects. A papain-fold cysteine protease cannot turn over without the general base that deprotonates the nucleophile, so a retained Cys with no His and no Asn is a fold, not an active site. Second, UniProt says the same thing in a CAUTION, and annotates zero Active site features on ADGB while all six reference calpains carry three each. Third, and independently of sequence analysis, MEROPS classifies androglobin as a non-peptidase homologue of family C2 and makes ADGB the holotype of that entry; the identifier block corroborates it, with the reference calpains at C02.001/.002/.004/.018/.029/.031 against ADGB at C02.972. The term additionally requires calcium dependence, for which there is no evidence of any kind on this protein - the calcium link in the literature runs through calmodulin binding at the IQ motif, which is a different thing. Reason: An InterPro domain signature has been read as an activity. The calpain fold is present but degenerate: the catalytic His and Asn are absent across six references and four alignment models, UniProt flags it in a CAUTION and annotates no active-site residues, and MEROPS independently classifies ADGB as a non-peptidase homologue with ADGB as the holotype. No experimental evidence supports calcium-dependent cysteine protease activity for ADGB in any organism. This is an electronic inference removed on biological grounds, not a curator's experimental call being second-guessed. Propagation Review Root cause: PROPAGATION BAD Failure modes: PSEUDO OR SUBACTIVITY LOSS FUNCTIONAL DIVERGENCE Sources checked: InterPro:IPR001300 · Peptidase_C2_calpain_cat (calpain catalytic domain signature) SUPPORTS SOURCE BUT NOT TARGET The signature itself is sound and ADGB genuinely matches the calpain catalytic fold - UniProt, PROSITE PS50203 and SMART CysPc all place the domain at 70-411. What fails is the InterPro2GO mapping from that fold to a catalytic activity, because ADGB is a fold-retaining pseudoenzyme that has lost the general base and the Asn. The defect is in the propagation, not in the source: this is why the root cause is PROPAGATION_BAD and not SOURCE_WEAK_OR_INFERRED, which my own analysis would contradict. Supporting Evidence: file:human/ADGB/ADGB-uniprot.txt The calpain domain lacks the conserved active site residues. file:human/ADGB/ADGB-bioinformatics/RESULTS.md A papain-fold cysteine protease cannot catalyse without the general-base histidine that deprotonates the nucleophile. A retained Cys with no His and no Asn is a fold, not an active site. file:human/ADGB/ADGB-bioinformatics/RESULTS.md All 6 reference calpains: Cysteine. ADGB: **Non-peptidase homologue**, and ADGB is the *holotype* of that entry. PMID:22115833 The calpain-like region is homologous to the catalytic domain II of the large subunit of human calpain-7. |
| GO:0006508 proteolysis | IEA GO_REF:0000002 | MARK AS OVER ANNOTATED | Summary: Marked as over-annotated rather than removed, because two things are true at once and they pull in opposite directions. The route that produced this row is the same unsound one that produced the GO:0004198 row: InterPro2GO reading a degenerate calpain fold as a protease. On that basis alone the term would go. But there is a real, published observation that ADGB contributes to proteolysis of septin 10, and it must not be erased by a structural argument. Reading past the abstract shows exactly what was and was not shown. The assay is co-overexpression of ADGB and SEPT10 in HEK293 cells, with a dose-dependent 37 kDa SEPT10 fragment appearing in the presence of ADGB; the perturbations that reduced it were a whole-IQ-domain deletion and a whole-protease-domain deletion. There is no purified-protein cleavage assay, no protease inhibitor, and crucially no catalytic-residue point mutant - which is the experiment that would demonstrate catalysis, and for which there is no confident triad to target. The word "catalytic" does not appear in the paper, and the authors consistently write "contributes to" rather than "cleaves". A large deletion in a 1667-residue protein can abolish an activity by disrupting folding or complex assembly without the deleted region being the catalytic module. So the conclusion may well be right by a mechanism other than direct catalysis - ADGB delivering SEPT10 to another protease, for example - but it is not established for this gene by the route that produced the annotation. That is over-annotation, not error. Reason: Derived from a catalytic domain signature that MEROPS and UniProt both hold to be non-catalytic in ADGB, so the inference route is unsound. Not removed, because a real in vitro observation of ADGB-dependent SEPT10 cleavage exists; that observation rests on co-overexpression and whole-domain deletion rather than on any demonstration that ADGB itself catalyses the bond hydrolysis, so it cannot carry the term on its own either. Propagation Review Root cause: PROPAGATION BAD Failure modes: PSEUDO OR SUBACTIVITY LOSS Sources checked: InterPro:IPR001300 · Peptidase_C2_calpain_cat (calpain catalytic domain signature) SUPPORTS SOURCE BUT NOT TARGET Same signature and same failure as the GO:0004198 row: the fold match is real, the activity inference is not. Recorded separately here because the biological-process term, unlike the molecular-function term, is not contradicted by independent data - it is merely unsupported by the route that produced it. Supporting Evidence: PMID:35700329 Collectively, these in vitro data suggest a scenario in which overexpressed ADGB proteolytically contributes to cleavage of overexpressed SEPT10 in an O2-independent but CaM-dependent manner. file:human/ADGB/ADGB-uniprot.txt Interacts with septin SEPT10; contributes to in vitro file:human/ADGB/ADGB-uniprot.txt The calpain domain lacks the conserved active site residues. |
| GO:0019825 oxygen binding | IEA GO_REF:0000002 | KEEP AS NON CORE | Summary: Kept, but as a non-core function - and the reasoning here runs opposite to the reflex. The tempting move is to remove it: ADGB is a hexacoordinate globin whose distal position is occupied, and the primary biophysical paper states outright that the extreme autoxidation rate means O2 binding is not a physiological function of the protein. Reading the term's definition rather than its label changes the answer. GO:0019825 is defined, in full, as "Binding to oxygen (O2)". It asserts nothing about transport, delivery, storage or physiological role. And the same paper that argues against a physiological role did observe an oxy complex and measure its association rate constant of about 0.8 x 10^6 M^-1 s^-1 by stopped-flow. Oxygen binding is therefore demonstrated chemistry, and the annotation is true as written. What the data do refute are the physiological oxygen terms, and GOA does not carry them: GO:0005344 oxygen carrier activity requires binding and delivering oxygen to an acceptor or a location, and GO:0019826 oxygen sensor activity requires responding to changes in oxygen level. Neither is annotated and neither should be added - the authors' clause about "the possibility of oxygen sensing" is an explicit hedge, not evidence. Two further observations reinforce keeping this non-core: the measured autoxidation rate of at least 2.6 per second is three orders of magnitude faster than neuroglobin's, and the SEPT10 cleavage assay in cells was explicitly repeated under 0.2 percent O2 with no difference, so the one cellular process linked to ADGB is oxygen-independent. Curator precedent agrees with keeping it: the two closest comparators, human neuroglobin (Q9NPG2) and cytoglobin (Q8WWM9), are also hexacoordinate globins that are not oxygen transporters, and GOA gives each of them GO:0019825 by IDA. Removing the term from ADGB while its nearest structural analogues retain it on the same kind of evidence would be inconsistent. Reason: The term means only "binding to oxygen", and that binding has been measured on the human recombinant globin domain, so the annotation is correct as written. It is not a core function because the oxy complex is only transiently observable and autoxidises three orders of magnitude faster than neuroglobin's, and the protein's measured chemistry is oriented towards nitric oxide rather than oxygen. Supporting Evidence: PMID:38725499 This remarkable autoxidation rate strongly suggests that O2 binding is not a physiological function of the protein except for the possibility of oxygen sensing. PMID:38725499 With rapid autoxidation and high nitrite reductase activity, the globin appears to be more tailored toward nitric oxide homeostasis or buffering. |
| GO:0020037 heme binding | IEA GO_REF:0000002 | ACCEPT | Summary: Accepted as a core molecular function, and this is the half of the chimera where the domain-derived IEA turns out to be right. It is supported experimentally in human and structurally at residue level. Experimentally: the recombinant human globin domain has been expressed and purified with heme by three independent groups, showing a Soret band at 411 nm in the ferric state shifting to 423 nm on reduction with visible bands at 529 and 558 nm, an extinction coefficient determined by the hemochromogen method, and EPR characterisation. Structurally: aligning ADGB's globin domain to reviewed globins whose heme-iron ligands UniProt annotates, and requiring both residue identity and that the column land on one of ADGB's own annotated sites, the proximal F8 histidine is retained in 3 of 3 on-register references (myoglobin H94, neuroglobin H96, cytoglobin H113 all map to ADGB H824). The annotated-site condition is doing real work here rather than being decorative: sequence identity alone does not predict register, since cytoglobin at 18.0 percent is on-register for both ligands while haemoglobin beta at 20.3 percent is on-register for neither and produces a spurious distal "match" on a histidine UniProt does not annotate. One correction to UniProt belongs on this row rather than as a change to the GO term: the protein is a His/Gln hexacoordinate globin, not a bis-histidyl one. The distal E7 position is Gln792, confirmed by UniProt's own feature table, by the Swiss-Prot sequence read directly, and by three primary papers, while UniProt's FUNCTION prose still says bis-histidyl. That affects how the heme chemistry should be described, not whether heme is bound. Reason: Directly demonstrated on the human protein by three independent laboratories, and structurally well founded - the proximal heme-iron ligand is retained and is precisely the residue UniProt annotates. Core to what androglobin is. Supporting Evidence: PMID:22115833 The recombinantly expressed human globin domain exhibits an absorption spectrum characteristic of hexacoordination of the heme iron atom. PMID:39719941 These observations suggest that Gln30 in the heme distal site may coordinate to the heme iron in the ferrous state. file:human/ADGB/ADGB-bioinformatics/RESULTS.md **Proximal position (ADGB H824).** 3 reference columns land on it and 3 of 3 match: ADGB retains the proximal histidine. PMID:38725499 X = His for Cygb and Gln for Adgb |
| GO:0031514 motile cilium | IEA GO_REF:0000044 | ACCEPT | Summary: Accepted. This is the Swiss-Prot subcellular-location keyword pipeline working correctly: SL-0117 is "Flagellum", whose own curated GO mapping is GO:0031514, and UniProt's SUBCELLULAR LOCATION for ADGB is "Cell projection, cilium, flagellum". Unusually for a keyword-derived row, the term is supported well beyond the sperm flagellum that motivated it, which is why the broader motile-cilium term is the right level rather than an over-reach. ADGB expression across human tissues is specifically associated with motile-ciliated cell types, its promoter is a direct target of the ciliogenic transcription factors FOXJ1 and RFX2, and overexpressing full-length ADGB increases the number of ciliated cells. In Tetrahymena, Adgb is a subunit of the C1b/C1f supercomplex of the ciliary central apparatus. A useful counterweight kept in view: deleting Adgb in Tetrahymena caused only minor changes in ciliary motility, unlike deletion of its partners Spef2A or Cfap69, so ADGB is a resident of motile cilia rather than a core motor or structural component of the projection - which is exactly what a located_in term claims and no more. Reason: Correct keyword-to-GO mapping, and independently supported in three systems - human motile-ciliated cell types with FOXJ1/RFX2 regulation, mouse sperm flagellum, and the ciliate central apparatus. Supporting Evidence: PMID:33453283 specifically being associated with cell types forming motile cilia PMID:33453283 cell culture studies indicated an ADGB-dependent increase in the number of ciliated cells upon overexpression of the full-length protein, confirming a ciliogenesis-associated role of ADGB in mammals PMID:34083607 C1b/C1f contains several evolutionarily conserved proteins: Spef2A, Cfap69, Cfap246/LRGUK, Adgb/androglobin |
| GO:0007283 spermatogenesis | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Correct, but kept as non-core. Ensembl Compara transfer from the 1:1 mouse ortholog, whose own annotation is an IMP from the knockout study - Adgb-null males are infertile with reduced sperm concentration and motility, abnormal spermatid elongation and increased apoptosis in the cauda epididymis. Checked for the ACTR8-style complex projection and it is absent: the supporting reference annotates only 2 distinct entities in total and its phenotype terms, GO:0007283 IMP and GO:0007286 IMP, sit on Adgb alone, the gene actually knocked out. The transfer is also corroborated directly in human rather than resting on the mouse alone: bi-allelic ADGB variants have been found in infertile men in three independent cohorts. Kept as a non-core parent of the more informative spermatid development term GO:0007286, which is what carries the process in core_functions. Reason: Ortholog transfer from a knockout-based IMP, corroborated by human patient genetics in three independent cohorts; no projection artefact. Supporting Evidence: PMID:35700329 it locates to the sperm flagellum, the annulus, and the midpiece, and that it is crucial for male fertility and sperm formation file:human/ADGB/ADGB-uniprot.txt formation and maturation of elongating spermatids, thus playing an |
| GO:0007286 spermatid development | IEA GO_REF:0000107 | ACCEPT | Summary: Accepted. This is the same conclusion as the IBA row above, reached by a second and independent pipeline: Ensembl Compara transferring mouse Adgb's IMP from PMID:36995441. Worth recording explicitly rather than passing over, because it means GO:0007286 is annotated twice on this gene and the duplication is not additional evidence - both rows trace to the same single ortholog and the same paper. Across the whole record, 8 of the 13 GOA rows rest on two publications and one donor organism. That is not an error and I have not marked it as one, but anyone counting independent support for ADGB should know the effective N. Reason: Same well-founded ortholog transfer as the IBA row, by a different pipeline. Correct, but it is a duplicate rather than corroboration. Supporting Evidence: PMID:36995441 Spermatogenesis was also abnormal, with malformation of both elongating and elongated spermatids |
| GO:0036126 sperm flagellum | IEA GO_REF:0000107 | ACCEPT | Summary: Accepted. The donor holds GO:0036126 by its own IDA from PMID:35700329, so the transfer lands on exactly the term the mouse data support. Human immunofluorescence independently places ADGB in the flagella of human spermatogenic cells, and the human loss-of-function phenotype - short, angled, coiled and absent tails with disrupted 9+2 axonemal structure and a disorganised mitochondrial sheath - is what a flagellar protein's loss looks like. The same human study also reports acrosomal localisation, which GOA does not capture; see suggested_questions rather than a proposed annotation here, since it rests on one study's immunofluorescence. Reason: Ortholog IDA at the identical term, independently reproduced by immunofluorescence in human spermatogenic cells. Supporting Evidence: PMID:41834962 ADGB was found to be highly expressed after puberty in both human and mouse testes and was localized to the acrosome and flagella of spermatogenic cells. PMID:35700329 it locates to the sperm flagellum, the annulus, and the midpiece, and that it is crucial for male fertility and sperm formation |
| GO:0097225 sperm midpiece | IEA GO_REF:0000107 | ACCEPT | Summary: Accepted, and it duplicates the IBA row for the same term by a second pipeline from the same donor and paper. The donor's own evidence is an IDA at this exact term. The human ultrastructural phenotype is consistent - a disorganised mitochondrial sheath is a midpiece defect - though that is phenotype rather than localisation and does not by itself establish the component. Reason: Ortholog IDA at the identical term; duplicate of the IBA row by a second pipeline. Supporting Evidence: file:human/ADGB/ADGB-uniprot.txt along the whole sperm flagellum. Detected in the annulus of the sperm |
| GO:0097227 sperm annulus | IEA GO_REF:0000107 | ACCEPT | Summary: Accepted, duplicating the IBA row for the same term. The donor carries an IDA at this term, and the annulus is the location with the strongest mechanistic support: ADGB and SEPT10 co-localise there in stage-15 spermatids and mature sperm, and SEPT10 mislocalises when Adgb is lost. Septins are the structural constituents of the sperm annulus, so an ADGB-septin interaction at that site coheres with everything else known about the gene. Reason: Ortholog IDA at the identical term, with an interaction partner and a loss-of-function mislocalisation phenotype both anchored at this same structure. Supporting Evidence: file:human/ADGB/ADGB-uniprot.txt along the whole sperm flagellum. Detected in the annulus of the sperm PMID:35700329 Finally, in vitro data suggest that Adgb contributes to Sept10 proteolysis in a CaM-dependent manner. |
| GO:0005516 calmodulin binding | IPI PMID:36995441 ADGB variants cause asthenozoospermia and male infertility. | NEW | Summary: Proposed new annotation, and the clearest gap in this gene's record: GOA carries no calmodulin-binding annotation for ADGB in either human or mouse, despite calmodulin being the one partner that binds a defined, named motif in the protein and the one whose disruption is directly tied to human disease. UniProt annotates an IQ calmodulin-binding motif at residues 906-935, sitting between the two halves of the circularly permuted globin domain - an arrangement that is the structural signature of the whole protein family. Four independent lines support the interaction. In human patients, ADGB variants causing asthenozoospermia disrupt binding of ADGB to calmodulin, and UniProt records this with ECO:0000269 against the human entry. In cells, a truncated construct covering the globin and IQ domains co-immunoprecipitates calmodulin robustly, and mass spectrometry of the isolated globin-domain pulldown shows prominent enrichment of endogenous calmodulin. Biophysically, fluorescein-labelled calmodulin mutants show a 35 percent quenching with the N-lobe label against 10 percent with the C-lobe label on adding the purified ADGB globin domain, consistent with energy transfer to the heme. Functionally, calmodulin binding raises the domain's nitrite reductase catalytic efficiency roughly two-fold, and deleting the IQ motif abolishes that activation. Recorded as IPI against the human patient-variant study, with calmodulin-1 as the supporting entity. Note the one negative in the record, kept in view rather than suppressed: co-immunoprecipitation of full-length ADGB did not detect calmodulin under basal conditions in HEK293 cells, which the authors attribute to interference from the long C-terminal region. That argues the interaction is regulated or conformationally gated, not that it is absent. Reason: Fills the most conspicuous molecular-function gap in the record. ADGB has a UniProt-annotated IQ motif, human disease variants that disrupt calmodulin binding, direct co-immunoprecipitation and mass-spectrometric evidence, biophysical binding data on the purified human domain, and a measured functional consequence - yet no GO annotation in either species. Supporting Evidence: PMID:36995441 The variants disrupted the binding of ADGB to calmodulin. PMID:35700329 a truncated construct covering the globin and IQ domains displayed robust ADGB-CaM interaction PMID:39719941 The circularly permutated globin domain of Adgb was well folded with a heme group, which can interact with CaM via the IQ motif. |
| GO:0070026 nitric oxide binding | IDA PMID:38725499 The circularly permuted globin domain of androglobin exhibit... | NEW | Summary: Proposed new annotation. The purified human androglobin globin domain binds nitric oxide, and does so in a way that is unusual enough to be diagnostic rather than generic globin chemistry: the complex is five-coordinate, which in the globin superfamily is highly atypical and is instead the signature of hemoproteins with nitric-oxide storage or sensing roles such as soluble guanylate cyclase and cytochrome c prime. The characterisation used optical spectroscopy, EPR, stopped-flow and femtosecond laser flash photolysis; the NO-bound Soret peak is blue-shifted to 395 nm against the 420-430 nm typical of hexacoordinate ferrous-NO globins, and the measured affinity is in the 10^-10 to 10^-11 M range. Adding NO to the ferric protein produced no optical change, so the binding is specific to the ferrous state rather than a general heme artefact. The term is well precedented for this chemistry. GOA annotates GUCY1B1 to GO:0070026 by IEA - the code is stated because the neighbouring holders are experimental and it must not be read across - and soluble guanylate cyclase is precisely the five-coordinate NO hemoprotein the authors name as ADGB's comparator, alongside cytochrome c prime. The experimental precedent comes from CBS, THAP4 (nitrobindin) and the haemoglobin complex CPX-2158, which hold the term by IDA. This is a molecular function of ADGB that GOA does not record at all, and unlike the oxygen term it is the activity the biophysics actually points to. Stated caveat: the measurements are on the isolated recombinant globin domain, not full-length androglobin, and no in vivo NO-related phenotype has been demonstrated for ADGB in any organism. Reason: Direct, multi-method biophysical demonstration on the purified human globin domain of an atypical five-coordinate NO complex with sub-nanomolar affinity. Not represented anywhere in GOA for this gene. Supporting Evidence: PMID:38725499 Adgb binds NO as a five-coordinate heme iron PMID:38725499 With rapid autoxidation and high nitrite reductase activity, the globin appears to be more tailored toward nitric oxide homeostasis or buffering. |
| GO:0098809 nitrite reductase activity | IDA PMID:38725499 The circularly permuted globin domain of androglobin exhibit... | NEW | Summary: Proposed new annotation. Two independent laboratories have measured nitrite reductase activity on the purified human androglobin globin domain, and a third study has probed it by mutagenesis. The ferrous heme reacts with nitrite to generate nitric oxide, followed spectroscopically as loss of the 423 nm Soret with a rise near 386 nm; the kinetics fit Michaelis-Menten, giving kcat/KM of 0.046 mM^-1 s^-1 for the domain alone rising to 0.102 on calmodulin binding, and mutating the distal glutamine to tyrosine further enhances the activity. The activity is also modulated by the redox state of the CD-loop disulfide. Term choice was checked against the definitions rather than the labels. GO:0050421 "nitrite reductase (NO-forming) activity" was considered and rejected: its definition is cytochrome-c-specific, describing the cd1-type enzyme, and does not describe a deoxyferrous globin reducing nitrite. GO:0098809 is the correct level, and that is confirmed by curator practice rather than only by reading the definitions: GOA already annotates human neuroglobin, cytoglobin and myoglobin to GO:0098809 by IDA for exactly this chemistry, and none of the three uses the GO:0050421 child. Stated caveats, which are the authors' own. All measurements are on the isolated recombinant globin domain rather than full-length androglobin, they use dithionite as the reductant, and the KM values are in the millimolar range - well above physiological nitrite. The original authors write that it remains unclear whether this is a natural function of Adgb. It is proposed here as a demonstrated in vitro activity, not as an established physiological role, and it is not listed among the core functions for that reason. Reason: Replicated in vitro catalytic activity on the human protein's globin domain, measured by two independent groups with full Michaelis-Menten kinetics and probed by mutagenesis by a third. Absent from GOA. Proposed with the isolated-domain and millimolar-KM caveats explicit. Supporting Evidence: PMID:38725499 With rapid autoxidation and high nitrite reductase activity, the globin appears to be more tailored toward nitric oxide homeostasis or buffering. PMID:39719941 the binding of CaM enhanced the nitrite reductase activity of Adgb PMID:39719941 it remains unclear whether this is a natural function of Adgb |
Loading supporting content…
Download this section (compressed HTML)Q: UniProt's FUNCTION line for Q8N7X0 describes androglobin as a "chimeric globin with a bis-histidyl six-coordinate heme-iron atom", but bis-histidyl requires two histidines and UniProt's own feature table annotates the distal heme-iron ligand as Gln792, with only the proximal His824 being a histidine. Three primary papers agree the distal residue is a glutamine - Reeder et al. give the ferric coordination as His-Fe(III)-X with "X = His for Cygb and Gln for Adgb", Nie et al. label it the distal Gln30 in their construct numbering, and Ren et al. name it in their title and mutate it as Q12H/Q12Y. Should the FUNCTION text be corrected to His/Gln hexacoordinate? The prose appears to predate the 2024-2026 biophysics.
Suggested experts: UniProt curation team, David Hoogewijs, Brandon Reeder
Q: Human ADGB carries no experimental GO annotation of any kind, despite Evidence at protein level and more than a decade of published human biochemistry: heme binding by the human recombinant globin domain, nitric-oxide and nitrite chemistry on that domain by two independent laboratories, calmodulin binding demonstrated biophysically and disrupted by human disease variants, and co-immunoprecipitation of CFAP69, SPEF2, TTC29 and CFAP47 with human ADGB. All 13 current rows are electronic or phylogenetic, and 8 of them trace to a single mouse ortholog and two papers. Could GOA curate the human primary literature directly rather than relying on Compara and PAINT transfer?
Suggested experts: GOA curation team, UniProt curation team
Q: Is androglobin itself the protease in the SEPT10 cleavage reaction? MEROPS classifies ADGB as a non-peptidase homologue of family C2 with ADGB as the holotype, UniProt flags the loss of active-site residues in a CAUTION, and alignment against six reviewed calpains under four substitution models finds the catalytic histidine and asparagine absent while only the nucleophile cysteine (C132) is retained. The cleavage evidence comes from co-overexpression of both proteins in HEK293 cells with whole-domain deletions, not from a purified-protein assay or a catalytic-residue point mutant. An equally consistent model is that androglobin targets SEPT10 to a different protease in a calmodulin-dependent manner.
Suggested experts: Anna Keppner, David Hoogewijs, MEROPS curation team
Q: Should GO:0001669 acrosomal vesicle be annotated for ADGB? Immunofluorescence in human and mouse spermatogenic cells reports androglobin at the acrosome as well as the flagella, and human patients with bi-allelic ADGB variants show acrosome loss by electron microscopy. It is not proposed as an annotation here because the localisation rests on a single study, but it is the most likely missing cellular component if it can be independently confirmed.
Suggested experts: GOA curation team
Q: Is the nitrite reductase activity physiologically relevant, or an in vitro property of the isolated domain? All measurements use the recombinant globin domain with dithionite as reductant and millimolar KM values; the original authors write that it remains unclear whether this is a natural function. Against that, testicular oxygen tension is low (~10-15 mmHg), which is the regime where globin nitrite reductase activity is thought to matter, and androglobin is expressed in exactly those post-meiotic cells.
Suggested experts: Brandon Reeder, Ying-Wu Lin
Q: A PAINT-specific observation: within PANTHER PTHR46298, the biological-process term GO:0007286 sits at node PTN002922608 while the two cellular-component terms sit at PTN002922607. The family's reviewed membership is only human and mouse androglobin, so no harm results here, but it is worth confirming the two-node split is intentional rather than an artefact, since androglobin orthologs exist across 3237 taxa and any future expansion of the reviewed set would inherit whichever placement is wrong.
Suggested experts: GO Reference Genome/PAINT curators
Experiment: Express and purify the isolated ADGB calpain-like region (residues 70-411) and full-length androglobin, and assay them directly against recombinant SEPT10 in the presence and absence of calcium and calmodulin. Include a C132S substitution as the catalytic-residue control - the point mutant that the existing cell-based work never made - and a panel of class-specific protease inhibitors. In parallel, repeat the HEK293 co-overexpression experiment with the C132S full-length protein: if the 37 kDa SEPT10 band still appears, androglobin is not the catalyst, and the whole-domain deletion result reflects loss of a binding or scaffolding function rather than of catalysis.
Hypothesis: Androglobin does not itself hydrolyse peptide bonds; the SEPT10 fragment observed on co-overexpression is produced by a cellular protease that androglobin recruits or licenses in a calmodulin-dependent manner.
Type: in vitro reconstitution and structure-function
Experiment: Measure oxygen association and dissociation kinetics and the autoxidation rate on full-length androglobin, and on the globin domain in complex with calcium-loaded calmodulin, by stopped-flow and flash photolysis, comparing directly against the published isolated-domain values. Repeat with the Q792H variant, which restores the canonical distal histidine and has already been shown to improve stability, to test whether the distal glutamine is what destabilises the oxy complex.
Hypothesis: The extreme autoxidation rate measured for the isolated globin domain is a property of the truncated construct, and the full-length protein or the calmodulin-bound complex stabilises the oxy form enough for oxygen chemistry to be physiologically accessible.
Type: biophysical kinetics
Experiment: Generate Adgb knock-in mice carrying the Q792H equivalent (restoring the ancestral distal histidine) and, separately, a proximal H824A heme-binding-null allele, and compare sperm counts, motility, annulus and manchette organisation and axonemal ultrastructure against wild-type and full knockout. A heme-null allele that phenocopies the knockout would establish that the globin chemistry, not merely the protein's presence, is what spermatid development requires; a heme-null allele with a normal phenotype would show the reverse and would reframe androglobin as a cytoskeletal adaptor that happens to carry a heme.
Hypothesis: The distal glutamine is functionally required in vivo, and androglobin's role in spermatid maturation depends on its heme chemistry rather than only on its scaffolding interactions.
Type: mouse genetics
Experiment: Quantify nitric oxide output in wild-type versus Adgb-null spermatids using a cell-permeant NO probe under controlled oxygen tensions spanning the testicular range, with and without nitrite supplementation, and confirm by measuring downstream nitrosation. Complement with quantitative proteomics of S-nitrosylated targets in wild-type versus knockout testis to identify whether any of the known androglobin partners - septin 10, CFAP69, SPEF2, TTC29, CFAP47 - are among them, which would connect the heme chemistry to the cytoskeletal phenotype.
Hypothesis: Androglobin generates nitric oxide from nitrite in post-meiotic germ cells under the low oxygen tension of the testis.
Type: cell biology and redox proteomics
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)