The 2026-09-27 audit below is the current assessment. The earlier notes are retained as history; their GMP/XMP/salvage rejections, blanket propagation claim, and whole-pathway interpretation of “step 2/2” are superseded.
UniProtKB:P30566 (PUR8_HUMAN). HGNC:291. EC 4.3.2.2. Chromosome 22q13.1.
Lyase 1 family, adenylosuccinate lyase subfamily.
Deep research status: falcon provider out of credits (HTTP 402); no
-deep-research-falcon.md generated. Review grounded in the cached UniProt
record, the seeded GOA, cached publications (PMID_10888601, 11428554, 16973378,
19405474, 27590927), and the two cached Reactome entries. PMID:19405474 is the
only one with full text available in the cache; the others are abstract-only.
ADSL is a homotetrameric cytosolic enzyme that catalyses two distinct
beta-elimination (fumarate-releasing) reactions in purine metabolism, both in
the same active site (three subunits contribute residues to each of the four
active sites):
Both activities reside at the same active site; a single competitive inhibitor
(APBADP) blocks both, showing the two substrates occupy the same site
[PMID:19405474 full text].
Key verbatim support:
- UniProt FUNCTION: "Catalyzes two non-sequential steps in de novo AMP synthesis"
and "converts succinyladenosine monophosphate (SAMP) to AMP and fumarate."
- PMID:19405474: "Adenylosuccinate lyase (EC 4.3.2.2) catalyzes two β-elimination
reactions in the de novo synthesis of purines: the cleavage of adenylosuccinate
(SAMP)1 to AMP and fumarate; and the conversion of 5-aminoimidazole-4-
(N-succinylocarboxamide ribonucleotide) (SAICAR) to 5-aminoimidazole-4-
carboxamide ribonucleotide (AICAR) and fumarate."
- PMID:19405474: "ASL, a catalyst of key reactions in purine biosynthesis, is
normally a homotetramer in which three subunits contribute to each of four
active sites."
- Reactome R-HSA-73800 / R-HSA-73828: "The active form of this enzyme is a
cytosolic tetramer" and it mediates both reactions in vivo.
Cytosol. IDA (HPA immunofluorescence, GO_REF:0000052), TAS (Reactome), IBA. Also
reported to associate with the purinosome (multi-enzyme DNPS complex) under some
metabolic conditions [Reactome R-HSA-73800; PMID:27590927].
Adenylosuccinate lyase deficiency (ADSLD; MIM:103050), autosomal recessive.
Accumulation of dephosphorylated substrates SAICA-riboside (SAICAr) and
succinyladenosine (S-Ado); psychomotor/mental retardation, epilepsy, autistic
features. S-Ado/SAICAr ratio in CSF correlates with severity PMID:10888601.
Many characterised missense variants reduce activity (PMID:19405474 etc).
MF (core):
- GO:0004018 SAMP lyase — IDA x3 (PMID:19405474, 16973378) + IBA + IEA → ACCEPT (core).
- GO:0070626 SAICAR lyase — IDA (PMID:27590927) + IBA + IEA → ACCEPT (core).
- GO:0003824 catalytic activity (IEA/InterPro) — too general; MARK_AS_OVER_ANNOTATED
(parent of the two specific MFs already present).
- GO:0042802 identical protein binding (IDA PMID:16973378; IEA) — homotetramer, so
self-association is real, but bare-binding term is uninformative vs the enzymatic
MFs and complex CC → KEEP_AS_NON_CORE (IDA), IEA duplicate KEEP_AS_NON_CORE.
CC (core = cytosol):
- GO:0005829 cytosol — IDA/TAS/IBA → ACCEPT.
- GO:0032991 protein-containing complex (IDA PMID:16973378) — homotetramer =
protein-containing complex; correct but very general → KEEP_AS_NON_CORE.
BP:
- GO:0006189 'de novo' IMP biosynthetic process (IEA) — ACCEPT core (SAICAR→AICAR
is step 2/2 of the IMP de novo pathway; UniProt PATHWAY).
- GO:0044208 'de novo' AMP biosynthetic process (IBA + IEA) — ACCEPT core (SAMP→AMP
is AMP-from-IMP step 2/2; UniProt PATHWAY).
- GO:0006164 purine nucleotide biosynthetic process (IC PMID:10888601) — ACCEPT
(correct, general parent).
- GO:0006167 AMP biosynthetic process (IDA PMID:11428554; IEA) — ACCEPT/KEEP; the
IDA in blood cells measures AMP-producing activity.
- GO:0009152 purine ribonucleotide biosynthetic process (IEA InterPro) — ACCEPT
(correct general parent).
- GO:0044209 AMP salvage (IEA Ensembl, GO_REF:0000107) — MISLEADING. ADSL's AMP
production (SAMP→AMP) is part of DE NOVO synthesis (and the purine nucleotide
cycle), not the salvage (hypoxanthine/adenine phosphoribosyltransferase) pathway.
→ MARK_AS_OVER_ANNOTATED (Ensembl ortholog electronic transfer, biologically off).
- GO:0006177 GMP biosynthetic process (IEA Ensembl) — REMOVE. GMP is made from IMP
by IMPDH + GMPS; ADSL is not on the GMP branch. Wrong IEA ortholog transfer.
- GO:0097294 'de novo' XMP biosynthetic process (IEA Ensembl) — REMOVE. XMP is made
from IMP by IMPDH; ADSL not involved. Wrong IEA ortholog transfer.
- GO:0009060 aerobic respiration (IEA Ensembl) — REMOVE. ADSL is a cytosolic purine
biosynthesis enzyme; fumarate release is not participation in aerobic respiration.
Over-propagated ortholog IEA.
- GO:0001666 response to hypoxia (IEA Ensembl) — REMOVE. No evidence ADSL itself is
a hypoxia-response effector; electronic ortholog transfer, not supportable.
- GO:0007584 response to nutrient (IEA Ensembl) — REMOVE. Electronic ortholog
transfer; not supportable for human ADSL.
- GO:0042594 response to starvation (IEA Ensembl) — REMOVE. Same.
- GO:0014850 response to muscle activity (IEA Ensembl) — REMOVE. ADSL is highly
expressed in muscle and participates in the purine nucleotide cycle there, but
"response to muscle activity" (a stimulus-response BP) is not a supportable
molecular role for ADSL from an electronic ortholog transfer. Over-propagation.
The GO_REF:0000107 (Ensembl Compara) block of stimulus/response and off-pathway
biosynthesis terms is the classic over-propagation cluster and is treated as such.
All 30 seeded annotations, 13 original references, two core functions, and the
cached UniProt record were audited. The starting document was INITIALIZED with
30 prior decisions and no NEW rows. Its five gene files matched main
62134e998e0e4fbda34cc79564fb081e8dbbd951 byte for byte before editing. The
review YAML base blob was ee68b5cccadfbdfee8322b3240660e0092fbc8ec.
An open-PR ADSL-symbol query had no matches. Three name/alias-search matches
(#2498, #2516, #3147) were checked by their actual file lists; none touched ADSL.
The NCBI human record confirms ADSL,
HGNC:291, and aliases ASL, AMPS and ASASE. Historical “ASL” in the enzyme papers
means adenylosuccinate lyase, while ASL is also the current symbol of the distinct
argininosuccinate lyase gene. The current review retains human UniProt P30566
and the supplied alternative-product records unchanged.
Final actions: 23 ACCEPT, 3 KEEP_AS_NON_CORE, 4 UNDECIDED; 0 NEW. All 30
source-field objects are unchanged; only their review judgments were revised.
The two cores describe the AMP-forming and SAICAR-cleaving reactions. No new
process annotations were manufactured from disease phenotypes or pathway gaps.
Human ADSL cleaves SAMP to AMP and fumarate, and SAICAR to AICAR and fumarate.
The SAICAR reaction lies in the shared pathway constructing IMP, upstream of
both adenine and guanine nucleotides. UniProt's “step 2/2” is the CAIR-to-AICAR
segment, not the whole de novo IMP pathway. Its separate AMP-from-IMP segment
also has two steps. These local pathway counts do not define the scope of the
corresponding GO processes.
PMID:19405474(https://pubmed.ncbi.nlm.nih.gov/19405474/) was read in the full
cached article: Methods measures SAMP loss at 282 nm; Figure 3/Table 1 report
AMP-direction kinetics; Figure 6 and the oligomeric-state results connect
assembly defects with activity. K246E has severe oligomerization/activity
impairment. This article experimentally assays SAMP, while its introduction
summarizes SAICAR chemistry. The review does not call that introductory
statement a second substrate-kinetics experiment. The active tetramer has four
active sites, each assembled from three subunits.
PMID:10888601(https://pubmed.ncbi.nlm.nih.gov/10888601/) provides human
full-length/alternative-transcript and disease-mutant evidence, including
proportional decreases against both substrates; its cache is abstract-only.
PMID:16973378(https://pubmed.ncbi.nlm.nih.gov/16973378/) reports recombinant
human enzyme kinetics and analytical-ultracentrifugation tetramer mass; E. coli
is the expression host. The functional complex term is accepted at its original
broad resolution, and the two identical-protein-binding rows remain non-core
assembly properties.
PMID:11428554(https://pubmed.ncbi.nlm.nih.gov/11428554/) supports blood-cell
ADSL activity and AMP-maintenance context. PMID:27590927(https://pubmed.ncbi.nlm.nih.gov/27590927/)
combines recombinant-enzyme substrate preparation with CRISPR-edited HeLa
cells and metabolite/purinosome measurements. Both caches remain abstract-only.
Their experimentally curated annotations describe established ADSL chemistry
and are accepted with curator deference; knockout substrate accumulation is
not mislabeled as a purified-enzyme IDA assay.
Both cached Reactome events, R-HSA-73800 and R-HSA-73828, explicitly identify
ADSL as the cytosolic catalyst. The current HPA subcellular page
reports enhanced cytosolic localization with HPA000525 in A-431, U-251MG and
U2OS, including siRNA validation. No more specific compartment is substituted.
The live GO:0006177 definition
covers “The chemical reactions and pathways resulting in the formation of GMP,
guanosine monophosphate.” It does not restrict membership to IMPDH and GMPS.
GO:0097294 covers XMP
formation “from simpler precursors.” ADSL performs the SAICAR cleavage in that
larger biosynthetic sequence; it is not assigned either terminal enzyme activity.
GO:0044209 covers AMP
formation from its derivatives without de novo synthesis. Its parents include
AMP biosynthesis and purine ribonucleotide salvage. The definition's examples
of precursors do not restrict the process to APRT or adenosine kinase. Salvaged
purines can reach IMP through HPRT and then AMP through ADSS and ADSL. The
ADSL reaction performs chemical work in that route.
All four IBA rows were checked against the cached PTHR43172 PAINT table.
Their IBD node is PTN000154581. Structured source entries use that node
only; human ADSL among the descendant experimental sources supports inherited
function and is not circular evidence. The broad catalytic-activity and
purine-ribonucleotide-biosynthesis InterPro mappings are valid at their original
family-level resolution.
Compara donors were traced to mouse UniProt P54822 / ENSMUSP00000023043 for
the three pathway rows, and rat UniProt A0A8I5ZTN5 / ENSRNOP00000081561 for
the response/respiration and homomeric-binding rows. The six indexed mouse
Adsl GO-CAM activities (MGI:103202) were read, including their substrate and
output connections:
60ad85f700003015, activity 60ad85f700003072: SAICAR lyase in GMP60d5209a00001406, activity 60d5209a00001482: SAICAR lyase in de novo60897d8500000531, activity 60897d8500001245: SAMP lyase in AMP salvage,60ff660000001341, activity 60ff660000001352: AMP salvage from adenosine,60ad85f700002694 and 60ad85f700002947: additional de novo AMP and IMPThe cached models are under gocams/<model>/<model>-src.yaml; they were read
without modification. No direct human P30566 index hit was found. The human
judgments rely on conserved, directly established chemistry as well as the
mouse curator's explicit process modeling.
PMID:6480832(https://www.jci.org/articles/view/111553), present in the donor
models, is a mouse pharmacological muscle study of disrupted purine nucleotide
cycling. Its accessible abstract is not a GMP/XMP flux experiment. The
pathway judgments therefore use the model's reaction sequence together with
human chemistry, rather than overstating that paper's experiment.
PMID:25681585(https://pmc.ncbi.nlm.nih.gov/articles/PMC4405794/) was recovered
through indexed original Methods/Results/Discussion. Its mouse cardiac HIF-1alpha
and hadacidin experiments support a cycling/salvage context; the induced
regulatory enzymes include AMPD and HPRT, not an established induction of ADSL.
The full cached PMID:29079593(https://haematologica.org/article/view/8359),
including Figures 5 and 7, provides human/mouse erythrocyte salvage-enzyme and
isotope evidence. PubMed and the publisher identifier were independently
verified. The major hypoxic effect was decreased purine deamination. Aspartate
label in fumarate has alternative metabolic explanations, and these results
are not described as an ADSL-specific knockout or proof of enhanced ADSL flux.
The prior blanket removal of response annotations is superseded by an audit of
the actual donor studies. A metabolic enzyme can participate in physiological
responses; cytosolic location alone does not settle the annotation.
Human hypoxia studies are relevant follow-up evidence, not replacements for
unread rat assays. Externally indexed Results/Figure 4 of
PMID:32439803(https://pmc.ncbi.nlm.nih.gov/articles/PMC7363121/) show endogenous
ADSL proximity during hypoxic purinosome assembly, without a corresponding
increase in de novo purine synthesis. Proximity is not direct binding or flux.
PMID:40033100(https://www.nature.com/articles/s41556-025-01627-8) reports
hypoxia-linked ADSL phosphorylation and local fumarate inhibition of STING in
breast cancer. Its identifier and abstract were verified, but the main article
is subscription-restricted. The specialized mechanism remains a follow-up
lead; no new immune-response annotation or third core was added.
The required genuine Falcon launch (--timeout 1200 --fallback perplexity-lite)
ran with supported temporary UV tool/cache directories in parallel with normal
publication caching. Both providers failed before research because PyPI
deep-research-client dependency resolution encountered DNS errors. Neither
produced a report. This is a new failure distinct from the earlier 402 recorded
in the historical notes; no provider file was authored manually.
The five original publications were already cached and left unchanged. The
additional existing cache for PMID:29079593 was read unchanged. Normal
fetch-pmid batches for 6480832, 25681585, 3777158, 690130, 3689310,
32439803, 40033100, 8887278, and 7128902 completed with DNS failures and
cached 0/9. These are required publication-cache gaps: the review must
remain DRAFT until fetched normally. Primary identifiers and external
reading are recorded above independently of this machine-cache requirement.
No citation was removed to conceal a missing cache, and no cache was fabricated.
Local full_text_unavailable flags remain true for absent/abstract-only caches,
including papers whose full original text was recovered externally.
All original source fields, reference identifiers/titles, alternative products,
and machine-generated gene source files are preserved. The handoff manifest
records validation results, generated history, file hashes, and the explicit
cache gate. No Git, remote PR, shared-project, or source-cache edits were made.
Set the YAML status to DRAFT under the literal GeneReviewStatusEnum, which
reserves COMPLETE for reviews without validation warnings. The already documented
source-cache warnings remain unresolved. All biological judgments, source
assertions, reference assessments and core functions are unchanged. This status
label correction does not imply that source retrieval or automated review has
subsequently succeeded.
The root catalytic MF is refined to both measured substrate-specific lyase
activities (GO:0004018 and GO:0070626), preserving both reactions. This is based
on the human assays, not a blanket rule that all broad annotations are non-core.
Human tetramer formation is now ACCEPT on both self-association rows and integrated
into the two enzyme cores: each catalytic site receives residues from three subunits
PMID:19405474. The broad complex CC remains compatible with that functional assembly.
The rat aerobic-respiration transfer is now MARK_AS_OVER_ANNOTATED. The
primary rat abstract proposes that the
purine nucleotide cycle supplies citric-acid-cycle intermediates. The
live GO definition and parents
describe oxygen-coupled respiratory energy release. The issue is whether the
purine-cycle reaction is itself in that pathway, not whether ADSL must be a
respiratory-chain protein or every participant must perform redox chemistry.
Supplying intermediates supports a metabolic connection without demonstrating
this pathway assignment. The rat phenotype is not denied; full specificity controls
remain inaccessible, and no replacement regulation term is invented. Independent
peer review supported this scope correction and prompted the explicit non-redox caveat.
The reviewer questioned the species of PMID:6480832, but the
primary PubMed abstract explicitly reports
28 AICAriboside-treated mice and 22 saline controls. That source is correctly mouse;
the later PMID:3777158 study is rat. Reference notes now name exact primary verification
routes for all nine missing caches, distinguishing abstract versus full-section access.
VERIFIED means inspected identifier/content, not a successful normal cache fetch.
The missing-source requirements remain unchanged. Four IBA source comments are now
term-specific, and the blood-cell AMP row uses measured activity plus independent
reaction support instead of an aim-only quotation. All 30 source assertions and 23
reference identities, downloaded sources and published history remain unchanged.
Peer review also distinguished the dual-substrate assays reported in PMID:10888601 from the reaction descriptions in PMID:19405474, whose measured kinetics concern adenylosuccinate. The two relevant review rows now cite the measured dual-substrate result.
The earlier nine-record cache gate is now closed. Standard fetch output from
GitHub Actions run 36286975328 (head
5946477c8ac79ade0709264c775ea1262b108438), artifact 10920674630, was
transported without refetching or modifying records and verified against ZIP
SHA-256 c0ffe4a66b80278af34b44aab6a3ae354ffd5699236b3a486ca95527be5e9713.
The local import receipt is
tmp/verified-reference-records/local-import-receipt.json. Only this gene's
nine required records belong in its publication manifest.
full_text_unavailable flags are now false.All 30 source assertions, annotation decisions, two cores, 23 reference
identifiers/titles and machine-generated gene files are unchanged. Cache
availability is distinct from citation verification and does not resolve the
remaining contextual-transfer uncertainties. Previous history and dated notes
are retained; this entry supersedes their missing-record status. Targeted
validation, rendering and the new history check are recorded in the closure
manifest. COMPLETE is used only if the targeted validation has no warnings.
The current-head review at 87cbc5210d17dde649c9bf68c12b8065c57b3e8e
requested direct cached support for three physiological judgments. The starvation
row now quotes PMID:690130 for loss of liver/spleen activity, unchanged kidney/brain/
muscle activity, and restoration after refeeding. Its rationale and reference
assessment explicitly rest on the recovered abstract. The earlier dated record
of external full-article access remains historical provenance, rather than a
requirement for reproducing this decision. The cache remains abstract-only.
The aerobic-respiration row now quotes the authors' anaplerotic interpretation
from PMID:3777158. The muscle-response row quotes the metabolite-based activity
estimate and the reported stimulation-dependent synthetase/lyase change. These
excerpts support the existing distinction between metabolic support for respiration
and an unresolved source-specific muscle-response transfer; full controls remain
unavailable. The source spelling “anapleurotic” is preserved in the exact quotation.
The hypoxia rationale now states the actual PMID:8887278 results: increased AMP
deamination, decreased cortical IMP reamination, greater ADSL than synthetase
activity, and cycle limitation mainly by synthetase and AMP supply. A non-limiting
step need not be unregulated, so these findings alone do not settle the precise
ADSL-specific response or its human transfer. UNDECIDED remains appropriate. The
AMP-biosynthesis row now quotes PMID:11428554's explicit AMP-level discussion and
the measured human adenylosuccinate-to-AMP reaction from PMID:19405474, replacing
the less informative B-lymphocyte sentence and reaction-description fragment.
All 30 original source objects, actions, two cores, 23 reference identifiers/titles,
availability flags and machine-generated sources are preserved. No new publication
is introduced. The notes-inclusive source census, case-sensitive normalized quote
checks, targeted validation, history validation and render are recorded in the
follow-up manifest. COMPLETE is retained only with zero validation warnings.