AADACL2 is a 401-residue human member of the arylacetamide deacetylase (AADAC) family, one of five paralogs (AADAC, AADACL1/NCEH1, AADACL2, AADACL3, AADACL4) that belong to the GDXG or hormone-sensitive-lipase-like branch of the alpha/beta-hydrolase fold. It carries the complete catalytic apparatus of that clan - a Ser-Asp-His triad with Ser189 at the apex of a GDSSG nucleophile elbow, Asp341 and His371, an HGG oxyanion-hole loop at residues 111-113, and a Cys116-Cys338 disulfide - and those residues sit in register with the equivalent residues of the biochemically characterised relatives AADAC, NCEH1 and three mycobacterial carboxylesterases. It is therefore expected to be an active serine carboxylesterase, but no substrate has been identified and no catalytic activity has ever been measured for it. Transcript expression is essentially confined to skin, in contrast to the liver- and intestine-enriched AADAC and the ubiquitous NCEH1, so AADACL2 together with the also skin-expressed AADACL3 looks like a skin-associated expansion of the family. The protein enters the secretory pathway through an N-terminal hydrophobic segment, but whether that segment is a cleaved signal peptide that releases a soluble secreted enzyme or an uncleaved type-II signal anchor that retains it in a membrane - the arrangement demonstrated for AADAC and NCEH1 - has not been determined experimentally. Insertion/deletion variants in the gene have been associated with psoriasis in a Chinese Han cohort, and a chromosome 3 locus adjacent to the gene with ischemic stroke in indigenous African populations.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0016787 hydrolase activity | IBA GO_REF:0000033 | MODIFY | Summary: True of AADACL2, and true of the node it came from. PTN009058710 is the deep GDXG / alpha-beta-hydrolase-3 node, and its 16 protein donors span three Arabidopsis CXE carboxylesterases and the Arabidopsis isoprenylcysteine methylesterase ICME, two kynurenine formamidases that cleave an amide rather than an ester bond (mouse Afmid and yeast BNA7), Escherichia coli Aes, three Mycobacterium tuberculosis carboxylesterases, the bifunctional soybean 2-hydroxyisoflavanone dehydratase HIDH, the human, mouse and rat AADAC orthologs, mouse Nceh1, and one unreviewed Pseudomonas aeruginosa lipolytic enzyme. Tested donor by donor, bare hydrolase activity is not merely the cautious choice at that depth: it is the exact lowest common ancestor of the set, and no candidate term below it survives every donor. So this row is not an over-annotation, not a granularity defect and not a term-scoping error. What it is, is subsumed. AADACL2 separately carries GO:0052689 carboxylic ester hydrolase activity from IPR017157, the family-specific Arylacetamide deacetylase signature, and that is stronger and more direct evidence for this protein than a transfer from a clade this broad. The mechanism term GO:0017171 remains true of AADACL2 on its own residues - an annotated Ser189-Asp341-His371 triad in a GDSSG elbow with an HGG oxyanion loop, with all fifteen active-site-annotated donors placing their own catalytic nucleophile on position 189 and fourteen of the fifteen carrying a serine there - but it cannot be reached from this row, because the fifteenth, soybean HIDH, has a threonine at the nucleophile elbow. It is true of every donor PAINT cites at the family node PTN009058713 and carries IDA support from all three of them, so that is where it belongs, and it is recorded as a PAINT node-placement recommendation in knowledge_gaps rather than asserted here. Reason: All three human AADAC-family paralogs carry this row byte for byte: AADACL2, AADACL3 and AADACL4 each have GO:0016787 by IBA from GO_REF:0000033, transferred from PANTHER node PTN009058710, with the same 17 WITH/FROM tokens. That equality is measured for all three, not assumed, from two independent sources that have to agree: the committed GOA TSVs, and a per-accession QuickGO fetch for Q6P093, Q5VUY0 and Q5VUY2. It therefore has one correct treatment, and the donor set decides what that is. All 17 tokens were resolved and each donor's chemistry and nucleophile read off its own EC numbers, its own curated GO annotations classified by fetched ontology ancestry, and its own ACT_SITE features. GO:0016787 is true of 16 of the 16 donor proteins. GO:0052689 carboxylic ester hydrolase activity is true of 14 and false of 2: the kynurenine formamidases mouse Afmid (Q8K4H1) and yeast BNA7 (Q04066), EC 3.5.1.9, carry GO:0004061 arylformamidase activity, which sits under GO:0016810 on the carbon-nitrogen branch and not under the ester branch, by IMP and IDA respectively. GO:0017171 serine hydrolase activity is true of 15 and false of 1: soybean HIDH's nucleophile-elbow residue is Thr164. That last call rests on fold position rather than on a UniProt label - UniProt describes HIDH's ACT_SITE 164 as a proton acceptor (ECO:0000305) rather than naming a nucleophile, and 164 is the N-terminal-most member of a checked Thr-Asp-His triad - but the sequence corroborates it independently: 15 of the 16 donors read G-x-S-x-G at the nucleophile elbow while HIDH alone reads GETSG, so the elbow serine is substituted. The verdict does not depend on that rule: downgrading HIDH to undetermined still leaves GO:0017171 untrue of every donor, and downgrading every positionally-inferred nucleophile leaves it supportable for only the two donors UniProt labels explicitly. The two refutations lie on different axes, so neither candidate can be rescued by choosing the other, and because GO:0016788 and GO:0016810 are siblings whose only common ancestor below catalytic activity is GO:0016787, the term PAINT chose is exactly the lowest common ancestor of its donors. There is consequently no granularity and no term-scoping defect to correct on this row in any of the three genes. What remains true of it is redundancy: AADACL2 independently carries GO:0052689 from the subfamily-specific IPR017157 signature, which subsumes GO:0016787 and is better provenanced for this protein than a clade-wide transfer, so the row should collapse onto that term. The action is therefore MODIFY on redundancy grounds alone, and root_cause is EVIDENCE_CIRCULAR_OR_REDUNDANT - the target already has stronger direct evidence. No failure mode is recorded, because the propagation itself did not fail. Replacing the row with GO:0017171 would assert of the transfer something one of its own donors refutes, so the mechanism term is pursued as a node move to PTN009058713 instead, where it is true of every donor and held by IDA in all three of them. GRANULARITY_MISMATCH is deliberately not recorded. Read literally, its definition (a parent term that is true but uninformative) would fit any general term on any record; but the slot describes the biological shape of a propagation issue, and this propagation has none - the term is uninformative because the donor set is heterogeneous, not because the transfer could have been more specific. That follows the rule adopted for this campaign: ask whether the term is the LCA of its donors, and reserve GRANULARITY_MISMATCH for donors that agree while the term still sits above them. The schema's enum description does not yet distinguish the two readings and would be worth clarifying: all three paralogs now encode the row on the propagation-shape reading, which the enum text does not itself state, so the convention is carried by argument rather than by the schema. Propagation Review Root cause: EVIDENCE CIRCULAR OR REDUNDANT Sources checked: PANTHER:PTN009058710 · deep GDXG / alpha-beta-hydrolase-3 node SUPPORTS TRANSFER 16 protein donors across plants, fungi, bacteria and animals. GO:0016787 is true of all 16 and is their exact least common ancestor, so this node licenses the term it was used for and nothing more specific. UniProtKB:Q5NUF3 · soybean HIDH, 2-hydroxyisoflavanone dehydratase SUPPORTS TRANSFER Supports GO:0016787: bifunctional, with EC 3.1.1.1 and a GO:0106435 carboxylesterase activity IDA alongside EC 4.2.1.105, so it is a genuine if minor ester hydrolase and does not block the parent term. It is the single donor that refutes GO:0017171: UniProt annotates ACT_SITE 164 as a proton acceptor (ECO:0000305) rather than naming a nucleophile, so the call rests on fold position, but the elbow pentapeptide reads GETSG against G-x-S-x-G in the other 15 donors, and the verdict survives downgrading the call to undetermined. It matches neither IPR017157 nor the family node. MGI:MGI:2448704 · mouse Afmid, kynurenine formamidase SUPPORTS SOURCE BUT NOT TARGET EC 3.5.1.9 with GO:0004061 arylformamidase activity by IMP, on the GO:0016810 carbon-nitrogen branch rather than the ester branch, so it is one of the two donors that keep GO:0052689 off this node. It does not block a mechanism term: UniProt names its own nucleophile Ser162. SGD:S000002836 · yeast BNA7, kynurenine formamidase SUPPORTS SOURCE BUT NOT TARGET The second amide hydrolase, EC 3.5.1.9 with GO:0004061 by IDA. It resolves cleanly through xref:sgd-S000002836 to Q04066, whose ACT_SITE 110 UniProt labels the nucleophile and which reads as Ser, so it too is a serine hydrolase. UniProtKB:P22760 · human AADAC SUPPORTS TRANSFER 51.6 percent identical to AADACL2, triad mapping onto S189/D341/H371, and it holds GO:0017171 by IDA. It is a donor at the family node PTN009058713 as well, which is where that mechanism term can safely descend. UniProtKB:P9WK87 · NlhH carboxylesterase, Mycobacterium tuberculosis SUPPORTS TRANSFER Characterised EC 3.1.1.1 carboxylesterase at 35.7 percent identity; triad also maps onto S189/D341/H371. UniProtKB:Q9HTI0 · probable lipolytic enzyme, Pseudomonas aeruginosa SOURCE WEAK OR INFERRED The one donor with no reviewed entry, so its name is not usable as evidence and its single ACT_SITE is ECO:0000256. Its curated GO:0047619 acylcarnitine hydrolase activity is nevertheless IDA, so it still counts as an ester hydrolase donor. Proposed replacements: carboxylic ester hydrolase activity Supporting Evidence: file:human/AADACL2/AADACL2-bioinformatics/NODE_PTN009058710.md No donor refutes `GO:0016787`, and every more specific candidate is refuted by at least one donor. file:human/AADACL2/AADACL2-bioinformatics/NODE_PTN009058710.md GO:0016787 hydrolase activity: TRUE 16, FALSE 0, UNDETERMINED 0 file:human/AADACL2/AADACL2-bioinformatics/NODE_PTN009058710.md `GO:0016787` is not merely unrefuted, it is the exact lowest common ancestor of this donor set. file:human/AADACL2/AADACL2-bioinformatics/NODE_PTN009058710.md So HIDH is a genuine, if physiologically minor, carboxylic-ester hydrolase; it refutes the *serine* term only. file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md Every one of the fifteen sources places its own catalytic nucleophile on AADACL2 position 189 file:human/AADACL2/AADACL2-uniprot.txt FT ACT_SITE 189 file:human/AADACL2/AADACL2-bioinformatics/NODE_PTN009058710.md Fifteen of the sixteen donors read **G-x-S-x-G** at the elbow — `GDSAG` in twelve of them, `GQSAG` in ICME, `GHSAG` in Afmid and `GHSVG` in BNA7 — while HIDH alone reads **`GETSG`** file:human/AADACL2/AADACL2-bioinformatics/NODE_PTN009058710.md only_the_refuting_donor_undetermined: TRUE 15, FALSE 0, UNDETERMINED 1, true_of_every_donor=False file:human/AADACL2/AADACL2-bioinformatics/NODE_PTN009058710.md all three carry `GO:0017171` by IDA |
| GO:0016020 membrane | IBA GO_REF:0000033 | UNDECIDED | Summary: Cannot be resolved on present evidence, and the reason is worth recording rather than papering over. The node is PTN009058713 and all three of its protein sources are single-pass type-II membrane proteins whose N-terminal hydrophobic segment is an uncleaved signal anchor: human AADAC (TRANSMEM 6-23), mouse Aadac (6-26) and mouse Nceh1 (5-25). AADACL2's own UniProt record instead annotates that same positionally homologous segment as a cleaved signal peptide (SIGNAL 1..18, mature chain 19-401) and places the protein as Secreted, so the transfer moves a feature the target is said to lack. Of the 14 reviewed UniProtKB entries carrying the family signature IPR017157, 12 carry a transmembrane segment (11 of them annotated explicitly as a type-II signal anchor; the exception is mouse Aadacl3, with three plain helices), and AADACL2 is the single one with a cleaved signal peptide instead. That looks at first like a clean paralog-topology error, but measurement does not support removing the term. The annotated cleavage site puts histidine at position -1, which signal peptidase I essentially does not accept; the hydrophobic core is if anything more hydrophobic than AADAC's experimentally confirmed anchor (peak Kyte-Doolittle mean over a 19-residue window 1.94 versus 1.81); and the two N-termini are homologous across the entire segment with the same downstream TP-x-PDN-x-EE-x-W motif. This is the textbook signal-peptide versus type-II-signal-anchor ambiguity, which UniProt has resolved one way for AADACL2 and the other way for its four paralogs. Nothing experimental breaks the tie: the protein is detected by mass spectrometry, but all four observed peptides lie in the catalytic domain, the earliest beginning at residue 124, so none of them tests whether the N-terminal helix is removed. Reason: The annotation depends entirely on a topology that has never been determined for this protein, and the sequence evidence is genuinely two-sided rather than merely thin. Removing the term would assert secretion, which rests on a predicted cleavage site with a disallowed -1 residue; accepting it would assert membrane anchoring, which no evidence specific to AADACL2 supports. Note additionally that is_active_in is a stronger claim than the record can bear, and that even under the membrane hypothesis the bare parent term would be the wrong choice - AADAC's own PAINT annotation at its ortholog node is the specific GO:0005789 endoplasmic reticulum membrane. Propagation Review Root cause: UNRESOLVED Failure modes: COMPARTMENT OR COMPLEX MISMATCH Sources checked: PANTHER:PTN009058713 · AADAC-family node (AADAC, NCEH1, AADACL2, AADACL3, AADACL4) UNRESOLVED The node's shared feature is an N-terminal hydrophobic segment, which is a validated type-II anchor in the two characterised members and is annotated as a cleaved signal peptide in AADACL2; the node cannot distinguish the two readings. UniProtKB:P22760 · human AADAC SUPPORTS SOURCE BUT NOT TARGET Experimentally an ER type-II membrane glycoprotein with its active site in the lumen, but its anchor is uncleaved and AADACL2's equivalent segment is annotated as cleaved. MGI:MGI:1915008 · mouse Aadac SUPPORTS SOURCE BUT NOT TARGET TRANSMEM 6-26, type-II signal anchor. MGI:MGI:2443191 · mouse Nceh1 SUPPORTS SOURCE BUT NOT TARGET TRANSMEM 5-25, type-II signal anchor; ER-luminal orientation shown experimentally. Supporting Evidence: file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md So the membrane claim is a transfer of a feature that AADACL2's own record says it does not have. file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md This is the classic signal-peptide / type-II-signal-anchor ambiguity, and UniProt has resolved it one way for AADACL2 and the other way for its four paralogs. file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md No peptide spans the annotated cleavage site and none starts at the putative mature N-terminus (residue 19). file:human/AADACL2/AADACL2-uniprot.txt FT SIGNAL 1..18 PMID:35736449 AADAC is a type II membrane glycoprotein, facing with its active side to the lumen of the endoplasmic reticulum (ER) |
| GO:0003824 catalytic activity | IEA GO_REF:0000117 | MODIFY | Summary: The most general molecular-function term in the ontology, one step below the root, produced by ARBA rule ARBA00027533. It is not wrong - the protein almost certainly is an enzyme - but it conveys nothing that the other six rows do not, and it is redundant with the two GO:0016787 rows and with GO:0052689 already present on the same record. The protein's own feature table supports a specific mechanistic statement instead: an annotated Ser189-Asp341-His371 catalytic triad, an HGG oxyanion-hole motif at 111-113, a GDSSG nucleophile elbow and a PROSITE PS01174 GDXG-lipase serine match. Reason: A root-adjacent term, fully redundant with GO:0052689 carboxylic ester hydrolase activity already present on this record from the family-specific IPR017157 signature, which is the term the three general molecular-function rows on this protein should all collapse onto. GO:0017171 serine hydrolase activity is also satisfied residue for residue by this protein - an annotated Ser189-Asp341-His371 triad with an HGG oxyanion hole and a GDSSG nucleophile elbow - but it is not carried by any existing row and is pursued instead as a PAINT node-placement recommendation, so GO:0052689 is the replacement proposed here, matching the treatment given to AADACL3 and AADACL4. Proposed replacements: carboxylic ester hydrolase activity Supporting Evidence: file:human/AADACL2/AADACL2-uniprot.txt DR PIRSF; PIRSF037251; Arylacetamide_deacetylase; 1. file:human/AADACL2/AADACL2-uniprot.txt FT MOTIF 111..113 file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md So this is not a case of a family name propagating onto a fold-only relic: the nucleophile, the acid, the base and the oxyanion loop are all present |
| GO:0005576 extracellular region | IEA GO_REF:0000044 | UNDECIDED | Summary: Mapped mechanically from UniProt's Secreted subcellular-location statement (SubCell SL-0243), which in turn rests on a single predicted feature, SIGNAL 1..18 with mature chain 19-401, carrying ECO:0000255 sequence-model evidence, that is, a signal-peptide predictor rather than an experiment. Two measurements weaken that prediction. The proposed cleavage site places histidine at position -1, where signal peptidase I overwhelmingly requires a small neutral residue (Ala, Ser, Cys, Gly or Thr). And the hydrophobic core is at least as hydrophobic as the experimentally validated uncleaved type-II signal anchor of the 51.6 percent identical paralog AADAC: peak mean Kyte-Doolittle over a 19-residue window is 1.94 for AADACL2 residues 1-19 against 1.81 for AADAC residues 5-23, and the two N-termini are homologous over the whole segment. The Human Protein Atlas, whose protein-class assignment is a majority vote over several secretome and membrane predictors, does place AADACL2 - and only AADACL2 among the five human paralogs - in the secreted class, but that is not independent corroboration: those are predictors of the same family as the one UniProt used, and the same HPA pipeline fails to classify AADAC, an experimentally validated single-pass membrane protein, as a membrane protein at all. No experiment has localised AADACL2, and the four peptides that establish its expression all lie in the catalytic domain, the earliest at residue 124, so they do not test whether the N-terminal helix is removed. Reason: The sole basis is a predicted signal-peptide cleavage site whose -1 residue is one signal peptidase I does not use, and the same segment is a validated uncleaved membrane anchor in the closest characterised paralog. GOA simultaneously carries GO:0016020 membrane for this protein. The two terms are not disjoint in GO - a shed ectodomain can legitimately be in both - but here they are two incompatible readings of one hydrophobic helix, and only one of them can describe the mature protein. Accepting extracellular region would settle by fiat a topology question that no experiment has addressed; removing it would equally assert the alternative. Supporting Evidence: file:human/AADACL2/AADACL2-uniprot.txt CC -!- SUBCELLULAR LOCATION: Secreted. file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md histidine at -1, which is not a residue signal peptidase I accepts file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md A classifier that misses one of the family's two experimentally validated membrane proteins cannot be used to argue that AADACL2 is a negative. |
| GO:0016020 membrane | IEA GO_REF:0000120 | UNDECIDED | Summary: The same claim as the PAINT membrane row, reached independently by ARBA rule ARBA00028763 and by the InterPro family signature IPR017157 (Arylacetamide deacetylase). Of the 14 reviewed UniProtKB entries carrying IPR017157, 12 have a TRANSMEM feature (11 of them annotated as a type-II signal anchor), one (human AADACL3) has neither a signal peptide nor a transmembrane segment, and exactly one - AADACL2 - has a cleaved signal peptide and no transmembrane segment. So the signature-to-GO mapping is a reasonable summary of the family and a poor summary of this member. It is also directly contradicted within the same GOA record by GO:0005576 extracellular region, which UniProt's own SubCell pipeline generated from the same protein's SIGNAL feature: one database asserts both readings of one hydrophobic helix. Reason: Kept open for the same reason as the PAINT row. The family-majority topology is a real argument for membrane anchoring, and the sequence measurements do not exclude it - AADACL2's N-terminal hydrophobic core scores higher than AADAC's validated anchor, and the predicted cleavage site is atypical. But nothing specific to AADACL2 supports it either, and the same record carries the alternative reading of the same helix. This should be resolved experimentally, not by choosing which pipeline to trust. Supporting Evidence: file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md Of the 14 reviewed members, AADACL2 is the only one whose N-terminal hydrophobic segment UniProt calls a cleaved signal peptide. file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md Hydropathy cannot distinguish them, and the N-termini are homologous file:human/AADACL2/AADACL2-uniprot.txt DR InterPro; IPR017157; Arylacetamide_deacetylase. |
| GO:0016787 hydrolase activity | IEA GO_REF:0000002 | MODIFY | Summary: InterPro2GO mapping from IPR013094, Alpha/beta hydrolase fold-3 (Pfam PF07859, which matches this protein twice). That is a bare fold-level domain signature shared by esterases, lipases, amidases, a threonine-nucleophile dehydratase and catalytically dead relatives alike, so hydrolase activity is the most the signature itself can license and the mapping is correct for the signature. It is nonetheless redundant on this record: the same protein matches the subfamily-specific IPR017157, whose mapping supplies GO:0052689 carboxylic ester hydrolase activity, and residue-level evidence the fold signature cannot see - the complete Ser189-Asp341-His371 triad, the HGG oxyanion loop, the GDSSG elbow, and the eight closest WITH/FROM donors placing their own catalytic residues on exactly those three positions - rules out the fold-retaining pseudoenzyme reading that would otherwise be the reason to stop at the parent. Reason: Correct for the signature and redundant for the protein - a redundancy objection, not an under-specificity one. The IPR013094-to-GO:0016787 mapping should not change: most proteins matching a bare alpha/beta-hydrolase-3 domain warrant nothing more specific, and the same fold signature also matches soybean HIDH, whose physiologically important reaction is a dehydration. What makes the row replaceable on AADACL2 is that AADACL2 also matches the family-specific IPR017157 and already carries GO:0052689 from it, so this row adds only an ancestor of a term the record has. The replacement is a statement about what AADACL2's annotation set should contain, not a claim that a fold signature could license the child term. GO:0017171 serine hydrolase activity is separately true of AADACL2 on its own residues, but a fold that includes a threonine-nucleophile member cannot deliver it either, so it is raised as a PAINT node-placement recommendation rather than proposed here - the same treatment as the IBA row above, and the same treatment AADACL3 and AADACL4 receive. GRANULARITY_MISMATCH is deliberately not recorded here either, for the reason set out on the IBA row above: a signature whose own member set is heterogeneous has no granularity defect to record. Propagation Review Root cause: EVIDENCE CIRCULAR OR REDUNDANT Sources checked: InterPro:IPR013094 · Alpha/beta hydrolase fold-3 (fold-level domain) CIRCULAR OR REDUNDANT A fold signature whose GO mapping is necessarily the general hydrolase parent, and which reaches outside the subfamily whose chemistry is being asserted: it also matches soybean HIDH, a threonine-nucleophile dehydratase outside IPR017157. It adds nothing beyond the IPR017157-derived GO:0052689 row on the same protein. Proposed replacements: carboxylic ester hydrolase activity Supporting Evidence: file:human/AADACL2/AADACL2-uniprot.txt DR Pfam; PF07859; Abhydrolase_3; 2. file:human/AADACL2/AADACL2-uniprot.txt DR PIRSF; PIRSF037251; Arylacetamide_deacetylase; 1. file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md Every source that fails lies at 26.5% identity or below, which is where this global alignment loses register in the C-terminal half |
| GO:0052689 carboxylic ester hydrolase activity | IEA GO_REF:0000002 | ACCEPT | Summary: The single most informative row on this record, and it holds up. IPR017157 is the family-specific Arylacetamide deacetylase signature (PIRSF037251), not a bare fold, and every biochemically characterised protein it covers is a carboxylic-ester hydrolase: AADAC (diglyceride and acetyl esterase activity) and NCEH1/KIAA1363 (cholesterol ester, acetyl monoalkylglycerol ether and retinyl ester hydrolase), plus their rodent, bovine, rabbit and orangutan orthologs. The same holds for the closer non-family relatives in this gene's WITH/FROM column - Escherichia coli Aes, three Mycobacterium tuberculosis carboxylesterases and three Arabidopsis CXE carboxylesterases - though not for the two formylkynurenine formamidases there, mouse Afmid and yeast BNA7, which act on amide bonds and fall outside IPR017157. AADACL2 retains the machinery to do the same chemistry. Two caveats belong on the record rather than in the action: no ester-hydrolase activity has ever been measured for AADACL2 itself, so this remains a prediction; and the family is not confined to ester bonds, since AADAC's name-giving activity is hydrolysis of the amide bond of arylacetamide xenobiotics, so the bond type AADACL2 prefers is an open question rather than a settled one. Reason: A family-specific signature, not a fold signature, supporting the activity that every characterised member of the family possesses, in a protein whose catalytic triad and oxyanion hole are intact and in register with those members. It is the most informative activity statement currently on the record, it is the term the other three molecular-function rows should collapse onto, and core_functions records it as AADACL2's core molecular function - the same choice made for AADACL3 and AADACL4, whose catalytic registers are identical. Two caveats stay on the record rather than in the action: it is prediction-grade, since no ester-hydrolase activity has ever been measured for AADACL2; and the family is not confined to ester bonds, since AADAC's name-giving activity is hydrolysis of the amide bond of arylacetamide xenobiotics, so the bond type AADACL2 prefers is an open question. The mechanism term GO:0017171 is the statement that would survive if AADACL2 turned out to prefer amide bonds, and it is separately evidenced on this protein's own residues, but it is not on any existing row; it is raised as a PAINT node-placement recommendation for PTN009058713 instead of being asserted as a second activity. Supporting Evidence: file:human/AADACL2/AADACL2-uniprot.txt DR PIRSF; PIRSF037251; Arylacetamide_deacetylase; 1. file:human/AADACL2/AADACL2-bioinformatics/RESULTS.md independently characterised bacterial carboxylesterases and with the 51.6%-identical human paralog AADAC PMID:35736449 AADAC, is known to hydrolyze amide and ester bonds of a number of xenobiotic substances, as well as clinical drugs and of endogenous lipid substrates such as diglycerides file:human/AADACL2/AADACL2-deep-research-affinage.md No mechanistic discoveries found in literature. |
Loading supporting content…
Download this section (compressed HTML)Q: Should GO:0017171 serine hydrolase activity move from the ortholog-specific PANTHER nodes PTN002745055 (AADAC) and PTN002745068 (NCEH1) down to the shared family node PTN009058713? Measured against the node's own donors, the term is true of all three of them - human AADAC, mouse Aadac and mouse Nceh1 - and all three of them hold it by IDA, whereas at the far deeper PTN009058710 that AADACL2, AADACL3 and AADACL4 actually inherit their only activity term from, it is refuted by soybean HIDH's threonine nucleophile. One node move would give all three paralogs a mechanism term they currently inherit from nothing, and the reciprocal half is worth reporting too: GO:0016020 membrane sits at that same family node although the type-II signal anchor it rests on is the feature AADACL2's own record says it lacks.
Q: Is AADACL2 secreted or membrane-anchored? Its N-terminal hydrophobic segment is annotated as a cleaved signal peptide, while the positionally homologous segment in AADAC and NCEH1 is an experimentally validated uncleaved type-II signal anchor. The predicted cleavage site places histidine at -1, which signal peptidase I does not accept, and the segment is more hydrophobic than AADAC's anchor. GOA carries both extracellular region and membrane for this protein as a direct result.
Suggested experts: Achim Lass, Ulrike Taschler
Q: What does AADACL2 hydrolyse? The Ser-Asp-His triad, oxyanion hole and GDXG elbow are intact and in register with characterised carboxylesterases, but the family straddles bond types - AADAC's name-giving activity is amide hydrolysis of arylacetamides while its lipid activity is on diglycerides, and NCEH1 acts on cholesterol, retinyl and acetyl monoalkylglycerol esters. Given that AADACL2 is skin-restricted, epidermal barrier lipids and cutaneous xenobiotics are the obvious first substrate panels.
Suggested experts: Achim Lass, Carina Wagner
Q: Does the psoriasis association at AADACL2 have a mechanism? An exome-wide indel screen in 32,043 Chinese Han individuals named AADACL2 as a previously unreported psoriasis gene, and the gene's expression is restricted to skin. Whether the associated variants alter the catalytic triad region, the N-terminal targeting segment or expression level is unknown, and there is no functional assay in which to test them.
Suggested experts: Liangdan Sun
Q: Are AADACL2, AADACL3 and AADACL4 a skin-associated expansion of the AADAC family with distinct substrates, or largely redundant? AADACL2 and AADACL3 are both skin-expressed, AADACL4 is choroid-plexus-restricted, and none has been assayed. All three currently share exactly the same two propagated annotations, which means the record cannot distinguish them at all.
Experiment: Express and purify recombinant AADACL2 (mature construct and full length) and label with a fluorophosphonate activity-based probe, comparing wild type with an S189A mutant. Competitive activity-based protein profiling against a panel of lipid and xenobiotic esters and arylacetamides, read out as probe displacement, then identifies substrate classes without needing to guess the physiological substrate first. Include AADAC and NCEH1 as positive controls, since both are established fluorophosphonate-reactive serine hydrolases.
Hypothesis: AADACL2 is a catalytically active serine hydrolase whose Ser189 nucleophile is accessible to active-site-directed probes.
Type: activity-based protein profiling
Experiment: Express epitope-tagged AADACL2 (tag at the C-terminus so it survives either topology) in primary keratinocytes or an immortalised keratinocyte line, and ask three questions in parallel - whether the protein accumulates in conditioned medium, whether it partitions with membranes after carbonate extraction, and whether it is protease-protected in digitonin- versus saponin-permeabilised cells. Determine the mature N-terminus of the secreted or membrane form by Edman sequencing or N-terminal peptide mass spectrometry, which will also test UniProt's cleavage position after residue 18. Run AADAC in parallel as the validated type-II anchor control.
Hypothesis: The N-terminal hydrophobic segment of AADACL2 is a cleaved signal peptide, so the mature protein is released into the medium rather than retained in a membrane.
Type: subcellular fractionation and protease protection
Experiment: Profile AADACL2 protein and transcript across a calcium-induced keratinocyte differentiation time course and across epidermal layers, then knock the gene out in organotypic skin equivalents and measure the neutral lipid and wax/cholesterol ester composition of the stratum corneum by mass spectrometry, together with barrier function by transepidermal water loss. Because AADACL3 is also skin-expressed, test single and double knockouts to expose redundancy.
Hypothesis: AADACL2 contributes to lipid ester turnover during epidermal terminal differentiation, where its expression is concentrated.
Type: differentiation time course plus knockout lipidomics in organotypic skin
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: No substrate and no measured activity. AADACL2 has never been assayed - not against an ester, an amide, a lipid or a xenobiotic - so although the catalytic machinery is demonstrably intact, which bond it hydrolyses, on which substrate, and to what end are all unknown. No biological process can be assigned, and consequently the entire GO record consists of two IBA and five IEA rows with no experimental evidence of any kind.
OPEN BIOLOGY WHOLLY_DARK
What is known: What is established is the machinery and the expression. The Ser189-Asp341-His371 triad, the HGG oxyanion loop and the GDSSG nucleophile elbow are annotated on the protein and are in register with the catalytic residues of the 51.6 percent identical human AADAC, mouse and rat Aadac, mouse Nceh1 and three characterised Mycobacterium tuberculosis carboxylesterases. The protein is detected by mass spectrometry (four peptides) and its transcript is essentially skin-restricted. What is missing is any biochemistry at all.
Significance: AADACL2 sits in a family whose two characterised members are pharmacologically important - AADAC hydrolyses clinically used arylacetamide prodrugs and procarcinogens, NCEH1/KIAA1363 is a drug-target serine hydrolase in cancer and platelet biology - and AADACL2 is the only family member whose expression is restricted to skin (AADACL3 is skin and placenta, AADACL4 choroid plexus). A skin-expressed serine esterase is a plausible participant in epidermal barrier lipid processing or in cutaneous xenobiotic handling, and the gene surfaced independently in a psoriasis indel screen, so the missing biochemistry is the barrier to interpreting a genetic signal that already exists.
What would resolve it: Serine-hydrolase activity-based protein profiling of the recombinant protein against a lipid and xenobiotic ester panel, which would establish both that it is catalytically active and what class of bond it prefers.
Provenance (the field's own admissions):
Gap: Whether AADACL2 is secreted or membrane-anchored is undetermined, and GOA currently asserts both. The N-terminal hydrophobic segment is annotated as a cleaved signal peptide (SIGNAL 1..18) by UniProt and is the positional homolog of an experimentally validated uncleaved type-II signal anchor in AADAC and NCEH1. The predicted cleavage site places histidine at -1, which signal peptidase I does not use, and the hydrophobic core is more hydrophobic than AADAC's validated anchor, so the sequence does not discriminate.
OPEN BIOLOGYCURATION CC_DARK
What is known: What is established: the protein enters the secretory pathway through an N-terminal hydrophobic helix, and it is expressed and detectable. What is missing is one experiment: no protease-protection, fractionation, surface-labelling or conditioned-medium assay has been performed on AADACL2, and the four observed peptides all lie in the catalytic domain (earliest residue 124), so mass spectrometry has not tested the N-terminal fate either.
Significance: The two possibilities imply different biology - a soluble enzyme acting in the extracellular space or stratum corneum, against a membrane-tethered enzyme acting on lipids of the secretory pathway. GO does not forbid a protein from carrying both terms, but these two are alternative readings of a single hydrophobic helix, so only one can describe the mature protein. The curation-side half of the gap is that GOA currently carries both, from three separate pipelines, with nothing recording that they are alternatives rather than complements.
What would resolve it: A protease-protection or medium-versus-membrane fractionation experiment on tagged AADACL2 in a keratinocyte line, plus N-terminal sequencing of the mature protein, would settle it and license exactly one of the two existing annotations.
Provenance (the field's own admissions):
Gap: The mechanistic molecular function of the AADAC family is annotated only at ortholog-specific PANTHER nodes and not at the node the family shares, so three human genes carry no mechanism term they are entitled to. GO:0017171 serine hydrolase activity sits at PTN002745055 (AADAC) and PTN002745068 (NCEH1); AADACL2, AADACL3 and AADACL4 inherit only bare hydrolase activity from the far deeper PTN009058710, while what they do inherit from the shared family node PTN009058713 is the much less transferable GO:0016020 membrane.
OPEN CURATION MF_DARK
What is known: The catalytic triad is conserved in all five human family members - each of their UniProt records annotates the equivalent Ser, Asp and His - and for AADACL2 that conservation was confirmed by alignment to eight independently characterised relatives. Which node can carry the term was then measured donor by donor. At PTN009058713 the donors are human AADAC, mouse Aadac and mouse Nceh1, all IPR017157 members, all serine nucleophiles, all carboxylic ester hydrolases, with GO:0017171 held by IDA in all three of them - so the term is true of every donor there and experimentally supported by each. At PTN009058710 it is false of one donor, soybean HIDH, whose nucleophile-elbow residue is a threonine where the other 15 donors carry a serine, and HIDH matches neither IPR017157 nor the family node. So the term cannot descend to the deep node but is safe at the family node. The N-terminal anchor runs the other way: present in 12 of the 14 reviewed IPR017157 entries, 11 of them an explicit type-II signal anchor, but absent from AADACL2's record. Which of the two terms is safe at the family node is therefore not in doubt; only the annotation has not been made.
Significance: This is a single tree-level edit that would correct three human genes at once, and the pattern is general: PAINT curators place well-evidenced mechanism terms conservatively at ortholog nodes while localisation terms, which transfer far less reliably, are placed at family nodes. The net effect is records where the location is asserted and the activity is not. It also explains a curation trap: because the mechanism term is genuinely true of the gene, it is tempting to attach it to whatever generic row is at hand, which misattributes it to a transfer that cannot support it. AADACL2, AADACL3 and AADACL4 now all state the recommendation as a node move instead.
What would resolve it: Move GO:0017171 serine hydrolase activity down to PTN009058713 and drop GO:0016020 membrane from that node pending experimental topology data.
Provenance (the field's own admissions):
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)