UniProt: P23526 (SAHH_HUMAN). Gene: AHCY / SAHH. HGNC:343. 432 aa. EC 3.13.2.1.
AHCY is adenosylhomocysteinase (S-adenosyl-L-homocysteine hydrolase, SAHH / AdoHcyase),
the NAD+-dependent enzyme that reversibly hydrolyses S-adenosyl-L-homocysteine (SAH/AdoHcy)
to L-homocysteine + adenosine [file:human/AHCY/AHCY-uniprot.txt "Reaction=S-adenosyl-L-homocysteine + H2O = L-homocysteine + adenosine"; RHEA:21708; EC=3.13.2.1].
Function is experimentally established: PMID:10933798 provides FUNCTION and CATALYTIC ACTIVITY
evidence (ECO:0000269) and is the IDA basis for GO:0004013.
Because SAH is a potent product inhibitor of essentially every SAM-dependent methyltransferase,
AHCY clears SAH and thereby sustains cellular transmethylation, coupling the SAM/methionine cycle
to homocysteine metabolism. Reactome frames this explicitly: "Adenosylhomocysteinase (AHCY) is a
tetrameric, NAD+-bound, cytosolic protein that regulates all adenosylmethionine-(AdoMet) dependent
transmethylations by hydrolysing the feedback inhibitor adenosylhomocysteine (AdoHcy) to homocysteine
(HCYS) and adenosine (Ade-Rib)" [Reactome:R-HSA-174401].
Deficiency causes Hypermethioninemia with S-adenosylhomocysteine hydrolase deficiency (HMAHCHD;
MIM:613752) — elevated SAH/SAM/methionine, failure to thrive, psychomotor/developmental delay,
facial dysmorphism, myopathy/myocardiopathy, hepatopathy [file:UniProt DISEASE; Reactome:R-HSA-5579084;
PMID:15024124]. Numerous loss-of-function missense variants (R49C, G71S, D86G, A89V, Y143C, Y328D,
W112* truncation) reduce catalytic activity and/or perturb tetramerization and nucleocytoplasmic
distribution [file:UniProt VARIANT features; PMID:19177456; PMID:28647132].
Molecular function
- GO:0004013 adenosylhomocysteinase activity — IBA, IEA, IDA (PMID:10933798), TAS (PMID:2596825): ACCEPT (core MF). IDA is the strongest.
- GO:0005515 protein binding — IPI x13 rows from large-scale interactome / OpenCell screens (PMID:25416956, 25910212, 28514442, 31515488, 32296183, 33961781, 35271311): uninformative bare "protein binding". Per policy, MARK_AS_OVER_ANNOTATED (do NOT REMOVE experimental IPIs). The biologically meaningful interaction (AHCYL1 homolog, homotetramer) is captured elsewhere; the IntAct partners (ANKRD40 Q6AI12, C1orf50 Q9BV19, APPBP2 Q92624) are not functionally interpreted.
Biological process
- GO:0033353 L-methionine cycle — IBA: ACCEPT (core; methionine/SAM cycle context). (UniProt DR still shows the older label "S-adenosylmethionine cycle" for GO:0033353; current ontology label is "L-methionine cycle".)
Cellular component
- GO:0005829 cytosol — IBA + IDA(HPA) + TAS(Reactome x2): ACCEPT (core location). Multiple independent lines.
- GO:0005737 cytoplasm — IEA SubCell: ACCEPT (broader parent of cytosol; consistent, less specific).
- GO:0005634 nucleus — IEA SubCell + EXP (PMID:28647132): KEEP_AS_NON_CORE. Real but minor/secondary pool shown by GFP tagging; not the principal site of catalysis.
- GO:0005783 endoplasmic reticulum — IEA SubCell + EXP (PMID:28647132): KEEP_AS_NON_CORE. Same GFP-tagging study; secondary distribution, not a known ER function.
- GO:0042470 melanosome — IEA SubCell (from melanosome-fraction MS PubMed:17081065): MARK_AS_OVER_ANNOTATED. High-throughput organelle proteomics co-purification; not an established site of AHCY function.
- GO:0070062 extracellular exosome — HDA x3 (PMID:23533145, 19056867, 20458337): MARK_AS_OVER_ANNOTATED. Abundant cytosolic protein routinely detected in exosome/secretome MS; not a functional secreted location.
The canonical human symbol is AHCY, HGNC:343, Approved in the archived
HGNC primary-source subset projects/CLINGEN_MENDELIAN/hgnc-symbols.json.
SAHH and AdoHcyase are aliases; neither has a separate human gene directory.
The coordinator independently searched open PRs for all three names and found
no overlap. All five canonical files matched main
9541f70405e9e9bef718106c32ec1a54a8051bda byte-for-byte before authoring;
/tmp/AHCY-base/baseline.json records the source blobs and SHA-256 hashes.
The baseline has 25 seeded assertions plus one pre-existing NEW proposal,
not 26 independently seeded assertions. Original term, evidence, reference,
qualifier and isoform fields remain unchanged, including the retained prior NEW.
The earlier notes' raw GOA counts describe repeated partner records; the review
has seven aggregated generic-binding rows, and all seven were reassessed.
The required Falcon command with a 1200-second timeout and perplexity-lite
fallback was launched concurrently with normal publication caching. Initial
offline tool resolution failed because the required deep-research-client build
was unavailable in the local tool cache. One normal online attempt then failed
PyPI DNS resolution before either provider could run (both provider client exits
2; wrapper exit 1). This was not a fresh provider credit or scientific-response
failure. No provider artifact was generated or authored manually. The normal
publication command found all 13 originally declared PMID records already
cached and left them unchanged.
Normal additional fetches for PMID:12590576, PMID:9586999, PMID:19177456,
PMID:15024124 and PMID:33328229 returned Cached 0/5 with DNS errors. A separate
normal attempt for PMID:41549122 returned Cached 0/1 for the same reason.
The existing notes already cite the first four structural/disease sources;
the two later studies were identified during this audit. These six absent
records remain explicit publication gates. PMID:17081065 was already cached
and is now included in the reference list as the actual melanosome source.
No publication cache or machine source was edited. Reference availability
flags follow actual local metadata, with supplementary and extraction limits
stated separately; verified identity is not downgraded solely for a failed fetch.
The original abstract of PMID:10933798 reports substrate-driven closure of the
catalytic and NAD-binding domains; it is not a verbatim account of a complete
human steady-state hydrolysis assay. The established human IDA is retained with
curator deference and independent curated/structural corroboration. The human
structure records RCSB 1A7A and
RCSB 1LI4 were checked as primary deposited
structural evidence. They support the tetrameric cofactor-bound architecture;
the latter explicitly describes four reduced cofactors in the trapped
intermediate-analogue complex. NAD is a bound catalytic cofactor, not a net
substrate in the overall hydrolysis equation. The 1989 cDNA abstract,
PMID:2596825, supports enzyme identity and the 432-residue human sequence;
it is not presented as a kinetic experiment.
The local PTHR23420 PAINT table records all three IBD assertions at
PTN000602759, and its member file includes human P23526. The family also
contains an IKR at a different node, PTN001158976; no claim about that node's
exact descendants is made without the full tree. Human AHCY has direct catalytic
evidence, so the other-node negative assertion is not transferred to it. The
PAINT table's later methionine-cycle seed snapshot contains an additional
zebrafish seed compared with the original GOA; the original GOA is preserved.
Human self-evidence for activity/localization is legitimate descendant grounding,
not circularity. Source blocks use the ancestral node, not repeated extant seeds.
RHEA:21708 and EC:3.13.2.1 are supported by the exact human catalytic record;
ARBA:ARBA00087412 internals remain UNRESOLVED.
The prior homocysteine-production NEW was reassessed rather than retained merely
to align the core table. AHCY itself performs the SAH hydrolysis that releases
L-homocysteine. The live L-methionine-cycle definition and parent relations explicitly include this
interconversion. The homocysteine-metabolism parent page places GO:0071269 and
GO:0033353 as separate children; the sulfur-amino-acid and L-amino-acid
biosynthesis parent pages independently show GO:0071269 beneath them. Direct
child-page requests failed, so no claim of an exhaustive live relationship
export is made. The curated HAMAP AdoHcyase rule explicitly maps SAH hydrolysis to
GO:0071269. The local Pseudomonas ahcY source has that HAMAP assignment;
yeast MET17 in the primary GO-CAM browser and HSU1 in the indexed UniProt
record provide other direct homocysteine-forming enzyme comparators. This is
positive annotation practice for the performer role, not a proposed gap based
on absent annotations. The local GO-CAM index contains no human P23526 activity.
The proposal describes L-homocysteine formation, not human de novo methionine
biosynthesis. The two earlier cores are merged into one catalytic unit with
two process contexts, rather than presenting the same MF twice.
PMID:28647132 explicitly states endogenous nuclear and cytoplasmic localization.
Its abstract does not quantify a universal minor nuclear fraction. Both nuclear
rows are now ACCEPT as an established compartment of the enzyme, and the single
core includes nucleus and cytosol. The ER claim has explicit UniProt provenance
to the same paper, but its relevant experiment is outside the accessible abstract;
both ER rows remain UNDECIDED pending that result. This does not assert absence.
The broad cytoplasm and cytosol source assertions remain at their stated resolution.
The melanosome source is PMID:17081065; its stage-resolved fraction-MS abstract
names selected validation targets but does not expose AHCY's peptide record.
Likewise, PMID:19056867 (human urinary exosomes), PMID:20458337 (B-cell exosomes)
and PMID:23533145 (expressed-prostatic-secretion urinary exosomes) establish
sample-level proteomics without exposing the AHCY-specific identification in the
available extraction. These four annotations are UNDECIDED. The previous
categorical contamination/co-isolation explanations are withdrawn: cytosolic
abundance does not disprove an extracellular or organellar pool, and a separate
compartment-specific catalytic function is not required for localization.
Seven generic GO:0005515 rows now use REMOVE under the explicit binding policy,
without rejecting any interaction. Their source-specific partners are traced:
ANKRD40/Q6AI12, APPBP2/Q92624 and C1orf50/Q9BV19 in the 2014 binary map; the
first and third in the 2015 variant screen, 2017 BioPlex map, 2019 variant map
and 2021 BioPlex maps; C1orf50 alone in HuRI; ANKRD40 in OpenCell. Source
methods and accessible Results were checked, but individual AHCY pair/allele
supplementary rows remain uninspected. No adapter, inhibitor or partner-specific
function is invented from these records. The separately studied AHCYL1 and FTO
interactions are not substituted as if those were the partners in these sources.
PMID:33328229(https://pubmed.ncbi.nlm.nih.gov/33328229/) was verified through
PubMed abstract/figure captions and indexed original PMC7744083 Results. Nuclear
fractionation, endogenous BMAL1 association and chromatin recruitment primarily
use mouse liver/MEFs. Human U2OS depletion affects reporter amplitude; 293T
transfection supports co-IP. This strengthens the functional nuclear context
without assigning methyltransferase activity or proposing circadian-process NEW.
PMID:41549122(https://pubmed.ncbi.nlm.nih.gov/41549122/) was checked against
PubMed, indexed original PMC12848013 Results and the
publisher PDF.
Recombinant binding and hydrolase-retaining separation mutants support a distinct
adenosine-dependent AHCY/FTO mechanism in tumor models. The authors exclude
intrinsic methyltransferase activity. The publisher records replacement of
incorrect supplementary Figures S4 and S7 on 2026-09-10; the corrected panels
were not independently compared. This is a correction, not a retraction. The
mechanism is acknowledged in prose and an expert question; it does not replace
the catalytic core or create a new universal role. Broader physiological and
assembly-state scope remains a question.
The standalone biological summary no longer labels nuclear AHCY universally
minor or conflates all methylation effects with SAH turnover. Disease context
remains enzyme-deficiency biology, not treatment advice. Final validation,
independent review, notes-inclusive source gates and exact publication manifest
are recorded after the draft is stable.
The later primary-source recovery below supersedes the earlier same-session ER
and urinary-exosome uncertainty. Independent reviewer annotation_a4galt recovered
the author manuscript of PMID:28647132(https://www.researchgate.net/publication/316917659),
which I also opened and checked. Results/Figure 6F show mCherry-AHCY colocalization
with GFP-calreticulin 1 in HEK293T cells. Both ER assertions are now NON_CORE,
scoped to ectopic expression without membrane-topology or lumen claims. Figure 1A
provides endogenous U2OS nuclear/cytoplasmic localization.
Both reviewers independently recovered indexed original
PMC2637050 Table 1, PMID:19056867:
AHCY has 10 unique peptides and 28 spectra (the Pep and ID columns). This supports
the corresponding urinary-exosome HDA as NON_CORE at the preparation's resolution,
without claiming extracellular catalytic activity or exosome biogenesis.
The author retains the cached abstract quote as sample context; numerical target
evidence is explicitly attributed to the external primary table. Local cache
full-text flags for both papers remain unavailable.
The independent reviewer also read the melanosome
coauthor-hosted main PDF
and B-cell exosome author manuscript.
The latter describes centrifugation, sucrose flotation and anti-MHCII purification,
with at least two unique peptides in two of three experiments for the protein list.
AHCY's supplementary rows were not recovered for either source or PMID:23533145.
These three source-specific uncertainties remain; no contamination claim is made.
The prior authored NEW GO:0071269 is retained with IC, not IDA. The explicit
supporting entity is GO:0004013, the established human IDA catalytic annotation
whose original PMID:10933798 is retained. The human curated reaction establishes
that AHCY performs the product-forming step. This is an inference from an existing
experimentally supported MF, not a claim that I read a new direct human hydrolysis
assay in the conformational-study abstract. The official GO IC guidance
permits reuse of the supporting annotation's primary reference for a single-source
inference; its special multiple-source GO_REF is not assigned here.
The schema's supporting_entities records GO:0004013. The comparator, performer,
parent and local GO-CAM checks above remain the basis for the process scope.
All 25 machine-seeded assertions are unchanged; only the prior authored NEW's
evidence code and supporting entity are revised outside review objects.
Root independently read all 26 decisions, all 23 reference assessments and the
integrated core; it accepted the IC correction and recovered source-scoped
ER/exosome refinements. Final action counts are 12 ACCEPT, 3 KEEP_AS_NON_CORE,
7 REMOVE, 3 UNDECIDED and 1 NEW. The independent consultation involved no edits.
Final targeted validation passed with two warning groups: uncached YAML citations
PMID:33328229 and PMID:41549122, plus differing exosome actions across sources.
The latter is intentional: one target identification was recovered while two
remain unresolved. History validation and HTML rendering passed. The explicit notes-inclusive audit additionally retains
PMID:9586999, PMID:12590576, PMID:15024124 and PMID:19177456 as missing-source
gates: six total. No cache was fabricated or rewritten. Status remains DRAFT.
All 25 seeded source objects, 20 original reference id/title pairs, UniProt and
GOA bytes passed preservation checks; this notes file retains its original
content as an unchanged prefix. All five baseline blobs also match current main
d2d8c9043b082a62378eff620ec0122d4118173b. The final four-file publication manifest
is /tmp/AHCY-local-manifest.json; no Git or remote mutation was performed.
The four published curation files, two protected machine records and 14 existing
publication caches match the frozen receipt for head
eed05cf5717858dbd26c89caa783d36125f12241. The six missing records are now
present as exact normal-fetch bytes from Actions run 36292249952, source head
3f234ebd4b540057fb287b27efc11341cfb102b1, artifact 10924402765, ZIP SHA-256
d8908403ad407495b26d23af7007ee4da1fc643ebd61728d99ce9755bcfacdc2.
tmp/source3-canonical-import-receipt.json records the import. All six paths are
absent at the exact PR base; no existing source or history was overwritten.
Three records are abstract-only: PMID:9586999, PMID:12590576 and PMID:15024124.
Their recovered abstracts were read. The first describes an inhibitor-bound
hydrolase structure and intersubunit cofactor contacts; the second explicitly
reports four subunits with bound NADH in the trapped catalytic conformation.
The clinical report documents low human enzyme activity and methionine-cycle
metabolite abnormalities, including hypermethylated leukocyte DNA. It does not
justify a universal prediction that every methylation endpoint decreases when
AHCY is deficient. Their full-paper Methods were not newly recovered.
PMID:19177456 now has full XML-derived text. Its Methods and Results directly
compare recombinant human WT, R49C and D86G proteins expressed in E. coli,
including tagged/shorter-tag preparations, hydrolytic activity, bound cofactor,
native PAGE and gel filtration. R49C can form tetramers under the specified
reducing/temperature conditions while displaying much lower catalytic activity;
D86G has solubility/aggregation limitations. Those observations support the
existing catalytic, assembly and variant context without turning conditional
mutant behavior into a new normal-human function.
PMID:33328229 now has full XML-derived text, including Results, figure legends
and Methods. Mouse liver/MEF experiments establish endogenous nuclear/chromatin
association and BMAL1-linked methylation/transcription effects. Methods explicitly
identify mouse GFP-AHCY and mouse WT/K186N rescue constructs. A human 293T host
therefore does not by itself establish a human AHCY construct. The human U2OS
experiment uses three siRNAs against endogenous AHCY and a circadian reporter.
These distinctions agree with the earlier functional nuclear interpretation;
AHCY hydrolyzes SAH rather than catalyzing histone methylation itself.
PMID:41549122 now has full XML-derived text. The read Results and Methods identify
recombinant human AHCY and FTO, purified binding and FTO demethylation assays,
GST removal before oligomer separation, and D245A/Y193H controls that retain
hydrolase activity while impairing the reported FTO-dependent effect. These
experiments support the tumor-model mechanism already described in the review.
They do not establish an intrinsic methyltransferase reaction or a universal
replacement of the canonical tetrameric hydrolase role. The
publisher change history
was independently rechecked: incorrect supplementary S4/S7 files were replaced
on 2026-09-10. The corrected panels were not independently compared; the text
cache does not resolve that quantitative image-level limitation.
Only the two top-level references with newly available full text have their
availability flags changed to false and their access/scope notes updated.
All 26 annotation objects (25 seeded and the existing IC NEW), annotation actions,
the integrated core, 23 reference identities, correctness/relevance judgments,
original quotations, alternative products and biological description remain
unchanged. No new source reference, annotation or core process is introduced.
The notes-inclusive census contains 20 distinct PMIDs and two Reactome records,
all cached. Historical DNS-failure entries remain as provenance and are superseded
by this recovery. YAML DRAFT remains appropriate for the intentional source-specific
exosome-action advisory; the source-access publication gates are closed.
Final targeted validation passed with exactly one curation warning: GO:0070062
has NON_CORE for the recovered urinary-exosome identification and UNDECIDED for
the two unresolved preparations. This preserves source-specific evidence rather
than forcing uniform decisions. There are no missing-reference warnings. History
validation and rendering passed; all reference-level full-text flags match local
metadata. The pkg_resources notice is a runtime dependency deprecation.
The frozen incremental manifest contains four curation files and six exact new
source caches, with no Git or remote mutations by this author.
The complete formal review 5329581554 and detailed comment 5854310687 were read. The live PR remained open at 4e54f8f589a9f3523a15b483f5addef521155444, and all five local gene files matched that exact imported tree before editing. This follow-up preserves all 25 machine-seeded source objects and their decisions, plus the existing IC process assertion.
The NAD-binding proposal is supported by direct human structural evidence, not merely by a request to convert every prose detail into GO. The cached abstract of PMID:12590576 reports bound NADH in the neplanocin-trapped enzyme. Its original RCSB 1LI4 deposition, read on this date, identifies human P23526, a 432-residue unmutated protein expressed in E. coli, a 2.01-Angstrom structure and nicotinamide ligand. The expression host is not the protein species. The deposited ligand name is not used to infer an oxidation state contrary to the paper. The machine UniProt record independently assigns one NAD cofactor per subunit to this study. The full journal article remains unavailable locally; the direct primary structure deposition and explicit abstract suffice for the bounded binding assertion.
The live GO:0051287 definition and parents include either NAD+ or NADH and place the term under adenyl nucleotide binding (GO:0030554). This matches a redox-cycling cofactor and avoids redundant separate oxidized/reduced binding proposals. No existing AHCY MF is an ancestor or descendant of this nucleotide-binding term. The cached GO-CAM index contains no P23526/AHCY entry; this is an index search, not a claim of exhaustive absence from current GO-CAM production. The current PTHR23420 PAINT table has no NAD-binding IBD.
Comparator checks targeted rat Ahcy, mouse Ahcy and yeast SAH1. The indexed original MGI comparative Ahcy graph, whose page states a 2023-03-10 snapshot, explicitly has a rat Ahcy NAD-binding IDA. It does not display a mouse experimental NAD-binding row. The combined live QuickGO query and direct SGD route were inaccessible, so no comprehensive current mouse/yeast absence is inferred. The rat curated precedent rules out a universal orthologue convention against this term. The PSEPK ahcY review has an authored NEW, which is not treated as a pre-existing external curator assertion. These comparisons inform convention; the proposed human IDA is grounded in the human structure, not transferred from rat.
PMID:12590576, PMID:15024124 and PMID:19177456 are promoted from notes-only sources to explicit references with their exact cache titles and source-scoped assessments. The first human deficiency report (PMID:15024124) supports the disease description, without supplying an extra developmental-process annotation. The full cached PMID:19177456 Methods/Results were reread: human NM_000687.1 constructs, bacterial expression, native PAGE and gel filtration support WT tetramers. Its 0.7–0.9 cofactor molecules per monomer and 95.7% NADH concern R49C; four cofactors per WT tetramer and predominantly oxidized WT cofactor are cited prior findings. Mutant measurements are not relabeled as WT measurements. The single catalytic core receives structural and oligomer evidence without creating a redundant cofactor-binding core.
The existing GO:0071269 IC row now also quotes the machine-record pathway step directly. Its evidence type and provenance remain unchanged. All source-specific exosome/melanosome judgments are retained. DRAFT is intentional because the schema defines COMPLETE as having no validation warnings, while the legitimate differing exosome actions produce an advisory. The earlier genuine Falcon/fallback attempts failed before provider contact because dependency DNS resolution failed; the lack of a provider artifact is documented provenance, and no report is fabricated to satisfy a filename convention. No new source cache is needed for these changes.
The independent read-only peer checked the human structure, formal term, no ancestor/descendant overlap and the 2009 WT-versus-mutant attribution; no blocker was found. Full targeted validation passed with only the previously justified exosome-action advisory. All 36 current supporting excerpts match their local source text; original reference identities, source fields, two alternative products, raw records and prior histories are preserved. The recursive authored PMID/DOI/Reactome census remains source-closed.