AHSP is a small (102 residue, 11.8 kDa), all-alpha-helical, erythroid-restricted cytoplasmic protein that acts as a dedicated chaperone for free alpha-globin during haemoglobin assembly. Haemoglobin A is a tetramer of two alpha and two beta subunits, and alpha-globin that has not yet found a beta partner is unstable: it aggregates, loses haem, and its haem iron cycles through oxidation states that generate reactive oxygen species. AHSP binds monomeric alpha-globin with 1:1 stoichiometry through a hydrophobic surface that largely reproduces the alpha1-beta1 contact of the tetramer, engaging the G and H helices of the globin. Binding has three consequences. It keeps apo-alpha-globin soluble and shifts it from an aggregation-prone unfolded ensemble into a partially folded, haem-receptive state. It converts the haem iron of holo-alpha-globin into a ferric bis-histidyl hexacoordinate configuration in which both the proximal and distal histidines ligate the iron, a chemically inert arrangement that suppresses redox catalysis, blocks ferryl haem and protein-radical formation, and prevents haem loss and precipitation. And because the AHSP contacts are deliberately suboptimal relative to the alpha1-beta1 interface, beta-globin competitively displaces AHSP, so the protected alpha subunit is handed on to form HbA rather than sequestered. AHSP does not bind beta-globin or the assembled alpha2beta2 tetramer. Its expression is driven by GATA1 during erythroid maturation and rises through the erythroblast series. Loss of AHSP causes alpha-globin precipitation, raised reactive oxygen species and apoptosis in maturing erythroblasts, and the gene modifies the severity of beta-thalassaemia, where unpaired alpha-globin accumulates.
Definition: Binding to a free globin chain, holding it in a soluble, aggregation-resistant state in which the reactivity of its haem iron is suppressed, and releasing it to a partner globin subunit for assembly into a haemoglobin multimer.
Justification: GO's two chaperone molecular functions partition the space by the client's folding state, and neither covers AHSP's whole activity. GO:0044183 protein folding chaperone is for binding an unfolded protein "to fold it"; GO:0140309 unfolded protein holdase activity is for binding an unfolded protein and keeping it unfolded while escorting it elsewhere. GO:0044183 is annotated on this gene and is defensible, because AHSP does measurably assist alpha-globin folding - but it fits only the apo-globin regime, and only partially there, since what AHSP produces is a partially folded state it then holds rather than a folded protein it releases. GO:0140309 does not fit at all: AHSP holds haem-loaded alpha-globin in a crystallographically ordered, fully helical state, and it escorts nothing. GO already handles this shape of activity once - GO:0140713 histone chaperone activity sits under GO:0140597 and describes binding a folded client, keeping it soluble, and depositing it into its final assembly. The proposed term is its sibling with "nucleosome" replaced by "haemoglobin tetramer", and the clause about haem reactivity is what distinguishes it from a generic carrier: suppression of the client cofactor's redox chemistry is an integral part of the activity, not a downstream consequence, and no existing GO molecular function expresses it (GO:0016209 antioxidant activity requires the annotated entity to be itself oxidized or to trap radicals, which AHSP is not and does not). The obvious alternative - annotating the parent GO:0140597 directly and proposing nothing - was considered and rejected on two grounds. Its own parent GO:0140104 states in a comment that "a carrier moves with its substrate/cargo", and AHSP moves nowhere: the hand-off is competitive displacement of AHSP by beta-globin at the same interface, so a carrier annotation would assert the reverse of the measured mechanism. And every existing child of GO:0140597 concerns getting a client somewhere, none concerns changing the client's chemistry, so the carrier term alone would leave the redox half of the activity unexpressed in both branches.
Parent term: protein carrier activity
Supporting Evidence:
Definition: A protein complex formed by the stable binding of an alpha-haemoglobin-stabilizing protein to a free hemoglobin alpha chain, in 1:1 stoichiometry, in which the haem iron of the alpha chain adopts a hexacoordinate configuration. The complex is an intermediate in haemoglobin assembly and is distinct from the haemoglobin complex, which it does not contain.
Justification: This complex is defined by four independent X-ray structures, counted by querying which PDB entries map both Q9NZD4 and P69905 rather than by reading UniProt's reference labels: 1Y01 (2.8 A, ferrous), 1Z8U (2.4 A, ferric), 3IA3 (3.2 A, the Pro30 cis-peptidyl structure) and 3OVU (2.83 A, a ternary complex with Staphylococcus aureus IsdH). Its 1:1 stoichiometry was measured independently by analytical ultracentrifugation, gel filtration and amino acid analysis. GO has no term for it, and this absence was re-checked after a text search had already missed GO:0031721 hemoglobin alpha binding on this same gene: seven independent query phrasings against QuickGO return no cellular-component term for an AHSP or alpha-globin complex, and GO:0005833 hemoglobin complex has zero children, so there is no hemoglobin sub-complex term anywhere to refine. ComplexPortal does not have it either - nine curated complexes contain AHSP (Q9NZD4) and not one of them contains alpha-globin (P69905) - so there is no upstream entry for GO to import. The absence is not merely a coverage gap, it produced a wrong annotation: the GO:0005833 hemoglobin complex NAS row removed in this review exists because a curator with a well-defined complex to record had only the tetramer term available, even though the reference being cited states in its abstract that AHSP does not bind the tetramer. GO already models the analogous case completely, which is the strongest argument for this term: GO:0031838 haptoglobin-hemoglobin complex is "A protein complex formed by the stable binding of a haptoglobin to hemoglobin", a direct child of GO:0032991, and its binding counterpart GO:0031720 haptoglobin binding sits immediately beside GO:0031721 in the same block of ids. So for haptoglobin GO has both the binding function and the complex; for AHSP it has only the binding function. The proposed name and definition deliberately follow GO:0031838's wording, and "hemoglobin alpha" rather than "alpha-globin" follows GO's own terminology in GO:0031721. Gene-symbol-style complex names are already used in this branch (e.g. GO:1990565 HSP90-CDC37 chaperone complex).
Parent term: protein-containing complex
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005515 protein binding | IPI PMID:15550245 Molecular mechanism of AHSP-mediated stabilization of alpha-... | MODIFY | Summary: The strongest binding datum AHSP has, and bare protein binding is the wrong term for it. This row records a 2.8 A crystal structure of AHSP bound to Fe(II) alpha-haemoglobin (PDB 1Y01), which maps the interface to the G and H helices of the globin and shows it recapitulating the alpha1-beta1 contact of HbA. Reason: CLAUDE.md forbids leaving an uninformative protein binding term where a more specific one exists, and a co-crystal structure is the strongest evidence a binding annotation can rest on. GO:0031721 hemoglobin alpha binding is defined as "Binding to a hemoglobin alpha chain", which is exactly what this structure resolves, and it is an is_a child of GO:0030492 hemoglobin binding, so the annotation is both more specific and definitionally exact - it does not inherit GO:0030492's tetramer wording. The partner is additionally recorded as a has_input extension. This term was missed on the first pass: an OLS text search for "alpha-globin binding" and "globin binding" cannot surface a term whose label is "hemoglobin alpha binding", because "globin" is not a token of "hemoglobin". Listing the is_a children of GO:0030492 finds it immediately. Proposed replacements: hemoglobin alpha binding Supporting Evidence: PMID:15550245 The crystal structure of AHSP bound to Fe(II)-alphaHb reveals that AHSP specifically recognizes the G and H helices of alphaHb through a hydrophobic interface that largely recapitulates the alpha1-beta1 interface of hemoglobin. PMID:15550245 The AHSP-alphaHb interactions are extensive but suboptimal, explaining why beta-hemoglobin can competitively displace AHSP to form HbA. |
| GO:0005515 protein binding | IPI PMID:16189514 Towards a proteome-scale map of the human protein-protein in... | MARK AS OVER ANNOTATED | Summary: Yeast two-hybrid screen hit with ZC3H12A (Regnase-1), an immune-cell endoribonuclease, from the CCSB human interactome pipeline. IntAct records this partner from two publications (PMID:16189514 and PMID:25416956) but both are CCSB Y2H against the same human ORFeome, so the repeat is assay reproducibility rather than orthogonal confirmation. MI-score 0.67. Reason: No orthogonal method, no co-purification, and no biological context linking the two proteins. AHSP is an erythroid-restricted 12 kDa cytosolic chaperone with one measured ligand; ZC3H12A is an inflammatory-response RNase of myeloid and lymphoid cells. Retained rather than removed because the interaction was seen in two separate screen publications, which is more than the singletons below, and because a Y2H positive is not evidence of absence. It should not be read as a functional partnership without a second method. Supporting Evidence: PMID:12066189 Stabilizing Protein (AHSP) is an abundant, erythroid-specific protein |
| GO:0005515 protein binding | IPI PMID:25156257 HSP70 sequestration by free Ξ±-globin promotes ineffective er... | MODIFY | Summary: The AHSP-alpha-globin interaction measured by bio-layer interferometry and curated in IntAct as a direct interaction (EBI-16247582, MI-score 0.71). The cached record for this reference is abstract-only and its abstract is about HSP70 sequestration by free alpha-globin, so the AHSP measurement is visible here only through the IntAct curation, not through the cached text. Reason: Same reasoning as the crystallographic row: a real, directly measured partner deserves a specific term rather than bare protein binding, and GO:0031721 hemoglobin alpha binding is the most specific one GO offers - the partner IntAct records is P69905, an alpha chain, not the tetramer. Deliberately not treated as a mis-attribution: per CLAUDE.md a curator who logged a bio-layer interferometry measurement read the full text, which is not in the cache, and a paper titled for one protein routinely assays another. Proposed replacements: hemoglobin alpha binding |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: FKBP1A from the Rolland 2014 binary interactome. In IntAct this partner is logged three times - two hybrid array, two hybrid prey pooling approach and validated two hybrid - which are three sub-methods of a single screen, not three experiments. MI-score 0.56. Reason: One Y2H experiment with both partners over-expressed in a heterologous nucleus, unreplicated by any orthogonal method. FKBP1A is a ubiquitous cytosolic prolyl isomerase and a frequent screen participant; nothing in the AHSP literature engages it. This is the third gene in this campaign where UniProt NbExp=3 resolves to one screen counted three ways. |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: UBE3A isoform 2 from the Rolland 2014 binary interactome; the same one-screen, three-sub-method logging pattern, MI-score 0.56. Reason: Single Y2H experiment, no orthogonal support. The accession was resolved rather than assumed: Q05086-2 is a declared UniProt isoform identifier for UBE3A isoform 2 (canonical Q05086, reviewed Swiss-Prot, 875 aa), not a truncated ORFeome clone standing in for the canonical protein, so this is not the partial-construct defect seen on ACRV1. It remains a single screen hit for a nuclear/cytoplasmic E3 ligase against an erythroid chaperone with one measured ligand. |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: The second of the two ZC3H12A screen rows, this one from the Rolland 2014 interactome (logged as array, prey pooling and validated two hybrid - one screen). Reason: Judged per partner and per reference, as the campaign requires. Treated identically to the PMID:16189514 ZC3H12A row: the two publications come from the same CCSB Y2H pipeline and ORFeome, so together they are one method seen twice rather than independent corroboration. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: ZNF20, a nuclear C2H2 zinc-finger protein of unknown function, from the HuRI reference interactome. One Y2H screen logged as three sub-methods, MI-score 0.56. Reason: Unreplicated screen hit with no shared compartment. ZNF20's only UniProt functional statement is "May be involved in transcriptional regulation"; AHSP is cytoplasmic. Y2H forces both proteins into the yeast nucleus at non-physiological levels, which is precisely the condition under which a compartment mismatch stops being informative. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: PSMB1, a non-catalytic beta subunit of the 20S proteasome core, from HuRI. One Y2H screen, three sub-method logs, MI-score 0.56. Reason: A proteasome core subunit is a common screen participant, and in the assembled 20S particle PSMB1 faces the interior chamber, so a stoichiometric binary contact with a cytosolic chaperone is not what the screen result would mean physiologically. There is a plausible biological story here - unpaired alpha-globin is degraded by the proteasome - but it is not what this datum tests, and no AHSP-proteasome experiment has been reported. Kept as over-annotated rather than removed so that a future targeted experiment has somewhere to attach. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Clavesin-1 from HuRI. One Y2H screen, three sub-method logs, MI-score 0.56. Reason: The clearest expression-domain mismatch in the set. CLVS1 is a neuron-restricted phosphatidylinositol 3,5-bisphosphate-binding protein required for normal late endosome/lysosome morphology in neurons; AHSP is restricted to the erythroid lineage. The two proteins have no cell type in common, so the interaction cannot occur in either protein's native setting even if the Y2H contact is real. Supporting Evidence: PMID:12066189 Stabilizing Protein (AHSP) is an abundant, erythroid-specific protein |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: VPS9D1, a VPS9-domain endosomal protein, from HuRI. One Y2H screen, three sub-method logs, MI-score 0.56. Reason: Unreplicated screen hit, no orthogonal method, no shared pathway with alpha-globin chaperoning. The four HuRI partners on this gene (ZNF20, PSMB1, CLVS1, VPS9D1) cohere as a screen's output rather than as a biology - nuclear, proteasomal, neuronal and endosomal - which is the signature the campaign has learned to read as assay design rather than gene function. |
| GO:0005515 protein binding | IPI PMID:40205054 Multimodal cell maps as a foundation for structural and func... | MODIFY | Summary: The AHSP-alpha-globin association recovered by affinity purification mass spectrometry (anti-tag co-IP) in the multimodal cell map, MI-score 0.71 - the same partner as the crystallographic and bio-layer interferometry rows, by a third, orthogonal method. Reason: Three independent methods in three independent publications converge on P69905 and on nothing else: x-ray diffraction, bio-layer interferometry and AP-MS, all scoring 0.71 while every other partner in IntAct scores 0.35-0.67 from a single method. Bare protein binding cannot be the conclusion for the one partner that a crystal structure resolves, so this moves to GO:0031721 hemoglobin alpha binding, with the partner preserved as a has_input extension. Erratum checked rather than assumed: PMID:40205054 carries an unflagged Publisher Correction (PMID:41039152) that fixes a duplicated equation in the Methods "Loss functions" section and does not touch the interaction data. Proposed replacements: hemoglobin alpha binding |
| GO:0005737 cytoplasm | IBA GO_REF:0000033 | ACCEPT | Summary: PAINT propagation from mouse Ahsp. The donor was resolved and its own evidence queried: MGI:MGI:2158492 is Q9CY02 AHSP_MOUSE, reviewed Swiss-Prot, 102 aa - the same length as the human protein and a one-to-one orthologue, not a paralogue - and it carries its own IDA to GO:0005737 from PMID:12066189, i.e. exactly the propagated term, not a generalisation of it. Reason: Correct term, correct level, sound donor. PTHR15914 has a single subfamily and contains no human paralogue, so none of the family-heterogeneity or mis-clustered-member failure modes can apply. The cytoplasm is where every measured AHSP activity occurs, so is_active_in is the right qualifier. |
| GO:0005737 cytoplasm | IEA GO_REF:0000120 | ACCEPT | Summary: The combinatorial IEA pipeline (GO_REF:0000120 - combined automated methods, not ARBA), combining an Ensembl Compara projection from mouse Ahsp with the UniProt subcellular location keyword SL-0086. The mouse source is the same orthologue as the IBA row and holds its cytoplasm annotation by its own IDA. Reason: Redundant with the IBA row and with UniProt's own subcellular location statement, which is itself experimentally supported (ECO:0000269|PubMed:12066189), but correct. Kept because the provenance differs and a correct annotation from a second pipeline is not an over-annotation. |
| GO:0005833 hemoglobin complex | NAS PMID:12066189 An abundant erythroid protein that stabilizes free alpha-hae... | REMOVE | Summary: The one clearly wrong annotation on this gene. AHSP is asserted to be part_of the haemoglobin complex, citing the discovery paper - whose own abstract states that AHSP forms a complex with free alpha-haemoglobin but NOT with haemoglobin A. Reason: Three independent grounds. (1) The cited reference says the opposite: AHSP forms a stable complex with free alpha-haemoglobin but not with beta-haemoglobin or haemoglobin A (alpha2beta2). (2) UniProt says the opposite, from two curated crystallographic papers. (3) The term definition excludes it - GO:0005833 is "an iron-containing, oxygen carrying complex... made up of two pairs of associated globin polypeptide chains", whereas the AHSP complex is a 1:1 heterodimer of one non-globin chain with one globin chain, and it is specifically not oxygen-carrying because binding drives the haem iron into the inert hexacoordinate ferric state. NAS is a non-experimental author-statement code, so the rule against overruling an unverifiable experimental call does not apply. The useful part of this finding is the cause: GO has no term for the complex AHSP actually forms, so the nearest available CC term was the tetramer - see proposed_new_terms. Supporting Evidence: PMID:12066189 Stabilizing Protein (AHSP) is an abundant, erythroid-specific protein that forms a stable complex with free alpha-haemoglobin but not with beta-haemoglobin or haemoglobin A (alpha(2)beta(2)). PMID:12192002 AHSP and alpha-globin are both monomeric in solution as determined by analytical ultracentrifugation and bind each other to form a complex with 1:1 subunit stoichiometry |
| GO:0006457 protein folding | IBA GO_REF:0000033 | ACCEPT | Summary: PAINT propagation from mouse Ahsp, which carries its own IDA to GO:0006457 from PMID:12066189 - the same term, not a parent of it. Reason: Sound donor, sound level, and independently supported on the human protein: purified human AHSP shifts human apo-alpha-globin from an unfolded ensemble to a partially folded state and dissociates its aggregates. The molecular-function counterpart of this process term is missing from GOA and is proposed below as a NEW GO:0044183 row. Supporting Evidence: PMID:20860551 AHSP forms a heterodimeric complex with Ξ±o that inhibits Ξ±o aggregation and promotes Ξ±o folding in the absence of haem. |
| GO:0006457 protein folding | IEA GO_REF:0000002 | ACCEPT | Summary: InterPro2GO mapping. IPR015317 is "Alpha-haemoglobin stabilising protein" - a family-specific signature covering the AHSP orthogroup, not a bare fold - and the live interpro2go file confirms it maps to exactly the four terms that appear as GO_REF:0000002 rows on this gene. Reason: A family-specific signature whose single experimentally characterised member holds this term by IDA is a well-founded source, unlike a fold-level signature. Redundant with the IBA row but not wrong. |
| GO:0020027 hemoglobin metabolic process | NAS PMID:12066189 An abundant erythroid protein that stabilizes free alpha-hae... | MODIFY | Summary: A correct but unnecessarily vague parent term. What AHSP participates in is specifically the formation of haemoglobin, not its turnover or catabolism: it chaperones the alpha subunit into the tetramer and is required for normal haemoglobin production in human erythroid cells. Reason: GO:0042541 hemoglobin biosynthetic process was verified to be a descendant of GO:0020027 via QuickGO ancestors, so this is a specificity upgrade within the same branch rather than a change of meaning, and it removes the implication that AHSP is involved in haemoglobin catabolism, for which there is no evidence. Human support: shRNA knockdown of AHSP in hemin-induced K562 cells and EPO-induced CD34+ erythroblasts caused alpha-globin precipitation and a significant fall in haemoglobin formation. Proposed replacements: hemoglobin biosynthetic process Supporting Evidence: PMID:18179859 RNA interference-mediated knockdown of AHSP expression resulted in considerable alphaHb precipitation, as well as in a significant decrease in HbF formation. |
| GO:0030097 hemopoiesis | NAS PMID:12066189 An abundant erythroid protein that stabilizes free alpha-hae... | MODIFY | Summary: Too general, and redundant. AHSP is erythroid-specific and acts only in the erythroid lineage; the gene already carries GO:0030218 erythrocyte differentiation, a descendant of this term, from two independent pipelines. Reason: Hemopoiesis covers myeloid and lymphoid development that AHSP has no part in. GO:0030218 is the correct level, is a verified descendant of GO:0030097, and is already annotated by IBA with a mouse IMP behind it and by InterPro2GO, so the parent term adds nothing. Also supported by how the gene was found in the first place - as a target of the erythroid transcription factor GATA-1. Proposed replacements: erythrocyte differentiation Supporting Evidence: PMID:12066189 a screen for genes induced by the essential erythroid transcription factor GATA-1 |
| GO:0030218 erythrocyte differentiation | IBA GO_REF:0000033 | ACCEPT | Summary: PAINT propagation from mouse Ahsp, which holds this term by its own IMP from PMID:12066189 - the knockout phenotype of reticulocytosis, abnormal erythrocyte morphology and inclusion bodies of denatured haemoglobin. Reason: The donor's evidence sits at exactly this term and the phenotype is erythroid-specific. Retained as core rather than non-core because erythropoiesis is not one arena among many for this gene - it is the only setting in which AHSP is expressed, and the null phenotype is an erythrocyte defect. Human corroboration comes from AHSP knockdown in CD34+ erythroblasts. Supporting Evidence: PMID:12066189 AHSP-gene-ablated mice exhibit reticulocytosis and abnormal erythrocyte morphology with intracellular inclusion bodies that stain positively for denatured haemoglobins |
| GO:0030218 erythrocyte differentiation | IEA GO_REF:0000002 | ACCEPT | Summary: InterPro2GO from the AHSP family-specific signature IPR015317. Reason: Correct and well-founded for the same reason as the other IPR015317 rows: the signature is specific to the AHSP orthogroup and the one experimentally characterised member holds this term by IMP. Redundant with the IBA row rather than wrong. |
| GO:0030492 hemoglobin binding | IEA GO_REF:0000002 | MODIFY | Summary: InterPro2GO from IPR015317. Other than bare protein binding this was, until now, the only molecular function GOA carried for AHSP - but the term is one level too general, and its definition names a molecule AHSP provably does not bind. Reason: GO:0030492 is defined as binding to "hemoglobin, an oxygen carrying, conjugated protein containing four heme groups and globin", i.e. the tetramer - and AHSP forms a complex with free alpha-haemoglobin but not with haemoglobin A. Its is_a child GO:0031721 hemoglobin alpha binding is defined as "Binding to a hemoglobin alpha chain", which is exactly right, and it is supported by four X-ray co-crystal structures, 1:1 solution stoichiometry and bio-layer interferometry. This also removes an asymmetry the first version of this review carried: GO:0005833 is removed partly because its definition requires a tetramer, so a tetramer-defined binding term could not be accepted on the same page with only a broad-synonym mitigation. The general term is not wrong, merely superseded, which is why this is MODIFY rather than REMOVE. Proposed replacements: hemoglobin alpha binding Supporting Evidence: PMID:12066189 Stabilizing Protein (AHSP) is an abundant, erythroid-specific protein that forms a stable complex with free alpha-haemoglobin but not with beta-haemoglobin or haemoglobin A (alpha(2)beta(2)). |
| GO:0030492 hemoglobin binding | NAS PMID:12066189 An abundant erythroid protein that stabilizes free alpha-hae... | MODIFY | Summary: The same term from the discovery paper as a non-traceable author statement. Unlike the GO:0005833 row from the same reference, this one is not wrong - but like the InterPro row above it names the tetramer rather than the chain AHSP actually binds. Reason: Same reasoning as the GO_REF:0000002 row: GO:0031721 hemoglobin alpha binding is the exact term and an is_a child of this one. Worth noting that the same paper supports both the replacement and the removal of GO:0005833 - it states in one sentence that AHSP binds free alpha-haemoglobin and not the tetramer, which simultaneously licenses the alpha-chain term and refutes the complex term. The evidence has also moved a long way past NAS since 2002: four X-ray co-crystal structures, solution stoichiometry and affinity, and bio-layer interferometry. An evidence-code upgrade to IDA cannot be expressed for a term already in GOA, so it is recorded in suggested_questions instead. Proposed replacements: hemoglobin alpha binding |
| GO:0050821 protein stabilization | IBA GO_REF:0000033 | ACCEPT | Summary: PAINT propagation from mouse Ahsp, which holds GO:0050821 by its own IDA from PMID:12066189. This is the process term that best captures what AHSP does. Reason: Donor, term and level all check out, and GO:0050821's definition - "any process involved in maintaining the structure and integrity of a protein and preventing it from degradation or aggregation" - is an exact description of the AHSP mechanism. The only limitation is that the process term has no matching molecular function in GO, which is the subject of the ONTOLOGY knowledge gap below. Supporting Evidence: PMID:12066189 Moreover, AHSP specifically protects free alpha-haemoglobin from precipitation in solution and in live cells. |
| GO:0050821 protein stabilization | IEA GO_REF:0000002 | ACCEPT | Summary: InterPro2GO from the family-specific signature IPR015317. Reason: Correct and well-founded; redundant with the IBA row. Four terms, one family-specific signature, and all four are terms the one characterised orthologue holds experimentally - an InterPro2GO mapping working exactly as intended. |
| GO:0044183 protein folding chaperone | IDA PMID:20860551 AHSP (Ξ±-haemoglobin-stabilizing protein) stabilizes apo-Ξ±-ha... | NEW | Summary: Proposed molecular function. UniProt annotates AHSP with the Chaperone keyword and GOA carries GO:0006457 protein folding as a process, but the gene has no chaperone molecular function at all - the only MF terms GOA carries are protein binding and hemoglobin binding. This row closes that gap with the term GO itself designates as the replacement for the obsoleted GO:0051082 unfolded protein binding. Reason: Direct evidence with purified human components: recombinant full-length human AHSP (102 residues) and alpha-globin prepared from human blood. AHSP dissociates soluble apo-alpha aggregates, forms a 1:1 heterodimer, and shifts the globin from unfolded to partially folded forms. GO:0051082 was checked at QuickGO and is obsolete with no replaced_by, listing GO:0044183 and GO:0140309 under "consider"; GO:0140309 was rejected because its definition is a protein carrier activity that "escorts" the client and its term comment requires the client to be kept unfolded, neither of which AHSP does. GO:0044183 is the honest partial fit - the folding it produces is partial and haem-dependent rather than to the native state, so the residual mismatch is filed as an ONTOLOGY knowledge gap with a proposed term rather than being papered over. Species discipline: IDA rather than IMP because the assay is biochemical, and human rather than mouse material throughout. Supporting Evidence: PMID:20860551 AHSP forms a heterodimeric complex with Ξ±o that inhibits Ξ±o aggregation and promotes Ξ±o folding in the absence of haem. PMID:20860551 Addition of AHSP leads to dissociation of soluble Ξ±o aggregates, formation of a stable Ξ±o:AHSP dimer, and a dramatic shift from unfolded to partially folded forms of Ξ±o. PMID:20860551 Hb A, Ξ±h and Ξ²h were obtained from human blood. PMID:12192002 AHSP and alpha-globin are both monomeric in solution as determined by analytical ultracentrifugation and bind each other to form a complex with 1:1 subunit stoichiometry |
| GO:1903427 negative regulation of reactive oxygen species biosynthetic process | IMP PMID:18179859 Reduction of AHSP synthesis in hemin-induced K562 cells and ... | NEW | Summary: Proposed process term for the redox arm of AHSP's function, which GOA does not represent at all. Binding to AHSP converts the alpha-subunit haem iron to a bis-histidyl hexacoordinate ferric state that is chemically inert: it lowers the subunit redox potential from +40 to -78 mV, slows the reaction of the iron with hydrogen peroxide, and abolishes detectable ferryl haem and protein-based radicals. Reason: The generic ROS term is the correct choice here rather than a species-specific child, because the literature implicates several species in sequence - autoxidation by molecular oxygen, hydrogen peroxide, ferryl haem and protein-based radicals - and no single one is the target. GO:1903427 was verified at QuickGO to be current and a descendant of GO:2000378, and the more specific biosynthetic branch is used because what AHSP suppresses is the generation of the species, not their downstream metabolism. The molecular-function framing was tested and rejected: GO:0016209 antioxidant activity requires the annotated entity to be itself more easily oxidized than the substance it protects or to trap free radicals, and all seven of its children are redox enzymes, whereas AHSP has no redox-active group and acts by remodelling its client's haem coordination sphere. That mismatch is filed as an ONTOLOGY knowledge gap. Evidence code chosen deliberately: this is an involved_in claim about a cellular process, so the row is anchored on the cellular loss-of-function datum - shRNA knockdown of AHSP in human K562 cells and human CD34+ erythroblasts raises reactive oxygen species - which is IMP, and in human cells, so the code is not borrowed from the abundant mouse knockout literature. The in vitro work on purified human AHSP and human HbA alpha-subunits is cited as additional support rather than as the row, because a two-protein reconstitution demonstrates the chemistry that underlies the process rather than the process itself; taking it as IDA for a BP term would have overstated what was assayed. Supporting Evidence: PMID:19706593 AHSP forms a specific complex with alphaHb and suppresses the heme-catalyzed evolution of reactive oxygen species by converting alphaHb to a conformation in which the heme is coordinated at both axial positions by histidine side chains (bis-histidyl coordination). PMID:23264625 AHSP binding also dramatically reduces the redox potential of Ξ±-subunits, from +40 to -78 mV in 1 m glycine buffer, pH 6.0, at 8 Β°C, demonstrating independently that AHSP has a much higher affinity for Fe(III) versus Fe(II) Ξ±-subunits. PMID:15931225 The alpha-haemoglobin-stabilizing protein (AHSP) binds alpha-haemoglobin (alphaHb), inhibits the ability of alphaHb to generate reactive oxygen species and prevents its precipitation on exposure to oxidant stress. PMID:26995402 In the presence of Alpha-hemoglobin stabilizing protein (AHSP), which stabilizes the Ξ±-subunit in a redox inactive hexacoordinate conformation (thus unable to undergo the redox ferric/ferryl transition), Cys93 oxidation was substantially reduced in both proteins. PMID:18179859 AHSP-knockdown cells demonstrated an increased ROS production and increased rate of apoptosis. |
Loading supporting contentβ¦
Download this section (compressed HTML)Q: Should the GO:0005833 hemoglobin complex annotation on AHSP be removed consortium-wide, and should the AHSP-alpha-globin heterodimer be given its own complex term at the same time? The annotation contradicts the abstract of the reference it cites and the definition of the term, but the underlying cause is that GO has no term for the complex that actually exists.
Suggested experts: GO editorial office, UniProt
Q: Can the holdase/foldase split introduced when GO:0051082 was obsoleted be revised so that chaperones acting on folded or partially folded clients are expressible? AHSP fails both GO:0044183 (it does not fold its client to the native state) and GO:0140309 (it does not keep its client unfolded, and it does not escort it anywhere). GO:0140713 histone chaperone activity shows the pattern already exists for one client class; a globin sibling would follow the same precedent.
Suggested experts: GO editorial office
Q: GO:0031721 hemoglobin alpha binding is the exact term for AHSP, yet neither GOA nor InterPro2GO uses it - both stop at the tetramer-defined parent GO:0030492, and IPR015317 is a signature specific to the AHSP orthogroup, so the mapping could safely be made more specific at source. Should the InterPro2GO mapping for IPR015317 be retargeted from GO:0030492 to GO:0031721, which would fix the annotation for every AHSP orthologue at once rather than gene by gene?
Suggested experts: InterPro, GO editorial office
Q: ComplexPortal holds nine complexes containing AHSP and none containing alpha-globin, even though the AHSP-alpha-globin heterodimer is defined by four X-ray structures and a measured 1:1 stoichiometry while the nine have no structure between them. Is the heterodimer missing because it falls outside ComplexPortal's scope as a transient assembly intermediate, or is it simply uncurated? The answer determines whether the proposed GO complex term should be authored directly or imported.
Suggested experts: ComplexPortal, GO editorial office
Q: Several AHSP annotations carry NAS or IBA evidence for terms that now have direct human experimental support - the hemoglobin-binding rows have four X-ray co-crystal structures behind them, GO:0050821 has purified human protein assays. There is currently no way in this schema to propose an evidence-code upgrade for a term already present in GOA, since NEW is rejected for such terms. Is a distinct action for "right term, weak code, here is the evidence" worth adding?
Suggested experts: ai-gene-review maintainers
Q: AHSP's PANTHER family PTHR15914 contains no human paralogue and a single subfamily, and the four IBA rows all trace to one mouse orthologue with its own experimental evidence from a single 2002 paper. Is this the right shape for a PAINT node, or would the annotations be better placed as ISO from mouse, making the single-study dependence visible?
Suggested experts: GO Central PAINT curators
Experiment: Ribosome profiling with selective AHSP pulldown in differentiating human CD34+ erythroblasts, to test whether AHSP associates with nascent alpha-globin chains still on the ribosome. The apo-alpha-globin chaperone activity is established in vitro with purified components but has never been placed in time relative to translation and haem acquisition in cells.
Hypothesis: AHSP acts on alpha-globin co-translationally or immediately post-translationally, before haem insertion, rather than only on the mature holo-subunit.
Type: OTHER
Experiment: Compare the subcellular distribution of endogenous AHSP with that of a size-matched inert reporter of similar molecular weight and charge in staged primary erythroblasts, using fractionation plus two independent antibodies, and test whether an AHSP fusion large enough to prevent passive diffusion still reaches the nucleus. This distinguishes a real nuclear function from diffusion and from antibody cross-reactivity.
Hypothesis: The nuclear AHSP pool reported in late erythroblasts is passive entry of a small protein rather than a regulated, functional localisation.
Type: OTHER
Experiment: Test whether AHSP co-immunoprecipitates with 20S proteasome subunits from human erythroblast lysates under conditions of alpha-globin excess (beta-globin knockdown or thalassaemic cells), and whether AHSP loss changes the rate of alpha-globin degradation. This is the one screen partner on this gene with a plausible mechanism behind it, and it is currently supported only by a single unreplicated binary assay.
Hypothesis: The proteasome interaction recovered by yeast two-hybrid reflects a genuine route by which AHSP-bound alpha-globin that fails to find a beta partner is delivered for degradation.
Type: OTHER
Experiment: Submit the heterodimer to ComplexPortal with the four co-crystal structures (1Y01, 1Z8U, 3IA3, 3OVU), the analytical ultracentrifugation stoichiometry and the isothermal titration calorimetry affinity, which together specify composition, stoichiometry and binding constant. A curated complex entry is the normal upstream route to a GO cellular component term. The gap is measurable rather than assumed: ComplexPortal currently holds nine complexes containing AHSP (Q9NZD4) and none containing alpha-globin (P69905), so the one AHSP assembly with four crystal structures behind it is the one that is missing.
Hypothesis: The AHSP-hemoglobin alpha complex can be given a defined stoichiometry and composition in ComplexPortal, which would let GO import a complex term rather than requiring one to be authored de novo.
Type: OTHER
What is not known β curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: GO has no molecular function term for a chaperone that binds a folded or partially folded client, holds it soluble in situ without unfolding it or translocating it, and releases it to an assembly partner. GO:0044183 requires the chaperone to fold an unfolded client; GO:0140309 requires it to keep the client unfolded and to escort it to an acceptor or a location.
OPEN ONTOLOGY MF_DARK
What is known: What is firmly established is the client's state in both regimes. Crystallography shows AHSP-bound alpha-globin as an ordered, fully helical globin in both the ferrous and ferric complexes, and NMR shows apo-alpha-globin shifted from unfolded to partially folded on AHSP binding and held there with conformational flexibility retained. So the client is neither unfolded, which excludes the holdase term, nor folded to the native state by AHSP, which makes the folding term only a partial fit.
Significance: AHSP is among the best structurally characterised chaperones in the human proteome - eight PDB entries, four of them X-ray co-crystal structures of the complex - and GO can currently say only that it binds haemoglobin. The gap is the same one that stranded the small heat-shock proteins when GO:0051082 was obsoleted without a suitable holdase term being created (repo issue #2222), but AHSP is blocked twice over: those genes hold genuinely unfolded clients and fail only the carrier clause, whereas AHSP fails both the unfolded clause and the carrier clause. Any resolution that only relaxes "escorts it to an acceptor molecule" would still not reach this gene.
What would resolve it: Create the proposed globin chaperone activity term as a sibling of GO:0140713 histone chaperone activity under GO:0140597 protein carrier activity, or more generally revise the holdase/foldase split so that the client's folding state is not the sole differentia. See proposed_new_terms.
Provenance (the field's own admissions):
Gap: GO has no molecular function term for suppressing the redox reactivity of a bound client's cofactor. AHSP does not itself react with oxygen or peroxide; it reconfigures the haem coordination sphere of the protein it binds.
OPEN ONTOLOGY MF_DARK
What is known: The chemistry is quantified: the alpha-subunit redox potential falls from +40 to -78 mV on AHSP binding, the rate of the initial hydrogen peroxide reaction with the iron drops, and ferryl haem and protein-based radicals become undetectable. What is missing is only the term, not the knowledge.
Significance: GO:0016209 antioxidant activity is the term an automated pipeline or an LLM will reach for - the affinage record for this gene proposes exactly that - and it is wrong, because its definition requires the antioxidant to be itself more easily oxidized than the substance protected or to trap free radicals. All seven is_a children of GO:0016209 are redox enzymes. Annotating AHSP there would assert a chemistry the protein does not have, and would place a haem-free, cofactor-free polypeptide among peroxidases and superoxide dismutases.
What would resolve it: Either a molecular function under GO:0003674 for cofactor-reactivity suppression in a bound client, or acceptance that this is only expressible as a biological process - the route taken here, with GO:1903427 negative regulation of reactive oxygen species biosynthetic process. The clause is also folded into the proposed globin chaperone activity definition.
Provenance (the field's own admissions):
Gap: Whether AHSP has any function in the nucleus of maturing erythroblasts is unknown. A single imaging study reports the protein becoming predominantly nuclear in late erythroblasts and being expelled with the nucleus at enucleation, and no molecular activity, partner or consequence has been attached to that pool.
OPEN BIOLOGYCURATION CC_DARK
What is known: Everything measured about AHSP is cytoplasmic: the complex with alpha-globin, the haem redox chemistry, the hand-off to beta-globin, and UniProt's experimentally supported subcellular location. The nuclear observation is an antibody-based localisation only.
Significance: If real and functional, a nuclear pool would be a second activity for a protein currently understood as doing one thing. If it is passive entry of an 11.8 kDa protein with no NLS, or an antibody artefact, then annotating GO:0005634 from this study would place AHSP in a compartment where none of its known chemistry can occur. The distinction cannot be made from the published data, which is why no nucleus annotation is proposed here.
What would resolve it: Fractionation of staged primary erythroblasts with a tagged or endogenous-epitope AHSP and a second, independent antibody; and a size-matched inert control to test whether nuclear entry is passive.
Provenance (the field's own admissions):
Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)