ALG3 is the human Dol-P-Man:Man(5)GlcNAc(2)-PP-dolichol alpha-1,3-mannosyltransferase
(EC 2.4.1.258), the enzyme catalysing the first ER-lumenal step of dolichol-linked
oligosaccharide (LLO) assembly for protein N-glycosylation.
Deficiency causes ALG3-CDG (congenital disorder of glycosylation type Id / CDG1D;
MIM:601110), originally described as CDGS type IV. Characterised by microcephaly,
severe epilepsy, minimal psychomotor development, dysmorphism, and partial deficiency of
sialic acids in serum glycoproteins; the defect causes accumulation of the
Man5GlcNAc2-PP-Dol intermediate and transfer of truncated oligosaccharides
[PMID:10581255; PMID:15840742]. Pathogenic variants include G118D and R171Q.
Historical summary, superseded: the 2026-09-27 substantive review below
replaces this early action list. Current broad ER and glycoprotein-process rows
are ACCEPT, and generic binding rows are REMOVE for lack of functional specificity.
GOA MF term = GO:0052925 (dol-P-Man:Man(5)GlcNAc(2)-PP-Dol alpha-1,3-mannosyltransferase
activity) — this is the specific, correct current MF; used in IBA, IEA(EC/RHEA), and IDA.
Baseline and scope: HGNC:23056 / Q92685 / ALG3, approved aliases CDGS4,
D16Ertd36e, NOT56L and Not56. The root coordinator verified all five canonical
files against main ba3ff58d7d2de76dbe3c24b16e05e12369f463fc and separately checked
all five open-PR searches and alias directories. A subsequent comparison to main
9a41b2b3b426ab112d2b3256540aed3d43307bc0 showed no scope changes. This review
preserves every source field of all 22 seeded annotations, both alternative
products and all 15 original reference identities. There were zero prior NEW
rows, despite the earlier task estimate. The final review has 14 ACCEPT,
3 MODIFY and 5 REMOVE decisions, 20 assessed references and one integrated core.
The historical note's “IPI x11” counts source interactions, whereas the seeded
review has five collapsed source-paper binding rows.
Research provenance: the genuine default Falcon launch used the supported
writable UV tool/cache directories, --fallback perplexity-lite and a 1200-second
timeout. Both provider paths failed during dependency resolution because PyPI
DNS was unavailable, before a provider report was produced
(/tmp/ALG3-deep-research.log). No provider output has been authored. The parallel
normal GOA publication fetch found all six seeded PMIDs cached
(/tmp/ALG3-fetch-goa.log, exit 0). Manual primary-source research below supplies
the review; it does not replace missing machine caches.
PMID:10581255(https://pubmed.ncbi.nlm.nih.gov/10581255/), DOI
10.1093/emboj/18.23.6816, was verified
against PubMed/PMC1171744. The complete original seven-page article was read from
an author-uploaded PDF.
Figures 4-5 and Methods distinguish normal Dol-P-Man synthesis from impaired
mannose transfer onto supplied Man5GlcNAc2-PP-Dol in patient fibroblast microsomal
extracts. These are membrane enzyme assays with donor/acceptor controls, not a
purified-human-protein kinetic experiment. Figures 8-9 test normal and G118D
human cDNA complementation in yeast alg3 cells. Patient glycan profiles and the
mutant complementation retain some complete precursors: the Discussion on
page 6820 explicitly describes reduced rather than abolished activity. The old
review's complete-loss claim is corrected. The exact cached abstract also states
“due to its leaky nature, a residual formation of full-length LLOs.”
The original membrane assay and multipass sequence/ER retrieval discussion
support the ER context; no independent immunofluorescence experiment is claimed
from this paper. The lumenal face is the curator's biochemical inference and is
also positively recorded in the reaction pathway. The local publication remains
abstract-only; external access does not alter its protected metadata.
PMID:15840742(https://pubmed.ncbi.nlm.nih.gov/15840742/) was already cited in the
historical notes. The primary abstract identifies the intended human ALG3-CDG
study and reports correction of patient-fibroblast biochemistry after wild-type
ALG3 lentiviral expression. Its endocrine presentation supplies clinical context,
not an additional ALG3 endocrine function. The normal cache fetch failed DNS
(/tmp/ALG3-fetch-new.log).
The full original PMID:29547901 article
was recovered through indexed publisher text and an
author-uploaded full-text copy.
Results/Figures 3-5 and Methods identify stable human HEK293 transfectants,
reciprocal co-IP of ALG3 with the CREB3 precursor, and partner-specific processed
ALG3 species in other human-cell assays. The Discussion presents a prerequisite
for CREB3 processing as an interpretation of precursor-selective binding. It
does not resolve ALG3 as a protease or demonstrate an adaptor bridge. Homodimer
and partner associations remain positive findings; the five generic binding rows
are REMOVE under the uninformative-term policy, not because the interactions
are disproven. No NEW signaling process or separate homodimer assertion is added.
A related original PMID:30192950 study
was read through its author-uploaded full text,
including Figure 2, Results, Discussion and Methods. HT-29/SKBR3 fractionation
and antibodies against different regions identify processed species in ER,
cytosolic and nuclear fractions. Individual cleavage sites and some topology
interpretations additionally rely on predictions. The work asks which species
are active and questions the catalytic assignment; it does not directly refute
mannose transfer. The controlled 1999 microsomal experiment and human-cDNA
complementation remain positive evidence. The separate processing observations
are retained as context and an expert question, without mapping the paper's
transcript nomenclature to UniProt isoforms by assumption or assigning an
unmeasured nuclear molecular activity. Normal fetch failed DNS
(/tmp/ALG3-fetch-isoform.log).
The four interaction-survey sources were also assessed individually:
All five binding-row sources retain their IPI/source identity. Short assay or
finding snippets replace paper-title-only support. The annotation-reviewer peer
confirmed the bounded policy approach but did not independently recover the
2018 full source; the full-source recovery above was performed in this lane.
PMID:38597022(https://pmc.ncbi.nlm.nih.gov/articles/PMC11251843/) was read in the
original PMC manuscript, Methods and Results. Plasma glycan profiles and one
homozygous R266C patient fibroblast line reveal abnormal glycan extension and
increased UPR/ERAD readouts. The source explicitly acknowledges the single
ALG3 cell line and uses ALG9-CDG as an additional comparator. These downstream
responses support the consequence of glycosylation failure; they do not establish
ALG3 as the IRE1 signaling or ERAD execution component. No new UPR or ERAD term
is asserted. Normal cache fetch failed DNS (/tmp/ALG3-fetch-new.log).
PMID:40789468(https://pmc.ncbi.nlm.nih.gov/articles/PMC12451169/) is the final
peer-reviewed 2025 JBC paper, DOI 10.1016/j.jbc.2025.110582. Indexed original
Results/Discussion/limitations show AKT phosphorylation of ALG3 Ser11/Ser13,
including human MCF10A and breast cancer cell work, isolated ALG3/recombinant
AKT1 assays and site substitutions. Glycoprotein phenotypes are rescued
differentially by wild-type and phosphorylation-site-mutant ALG3. The authors
explicitly did not directly measure phosphorylation-dependent transferase
activation or protein folding. ALG3 is the kinase's substrate; no kinase or
signaling MF is assigned to ALG3. The question of purified enzyme kinetics
remains open. Normal fetch failed DNS (/tmp/ALG3-fetch-2025.log).
PMID:41807832(https://www.nature.com/articles/s41589-026-02164-7), DOI
10.1038/s41589-026-02164-7, was inspected through indexed original Methods/Results
using queries combining the DOI with ScALG3, D71N, GgALG12 and Results. The study
selected yeast ScALG3, human HsALG9 and chicken GgALG12. Purified ScALG3
processes synthetic lipid-linked substrates, with the D71N mutant used for a
substrate-bound structural complex. This corroborates ortholog catalytic
mechanism but is not a human ALG3 structural assay or a direct measurement of
human transmembrane-helix count. The shared normal fetch already failed during
ALG12 review (/tmp/ALG12-fetch-new.log); it was not duplicated.
The protected PTHR12646 PAINT table places the three relevant IBD assertions at
PTN000291297: ER, glycoprotein biosynthesis and the precise catalytic activity.
All three remain ACCEPT. Structured IBA source entities are PTN-only; target
experimental evidence is legitimate ancestral grounding, not circularity.
The InterPro IPR007873 generic MF is true but refined to the target's specific
chemistry. The ARBA00085866 output was verified in the local rule catalog; its
exact historical target-matching condition was not recovered, so that source
link remains explicitly unresolved without manufacturing a rule failure.
RHEA:29527, EC:2.4.1.258 and UniProtKB-SubCell:SL-0097 agree with the protected
human record.
Live GO:0052925 resolves
the Dol-P-Man donor and Man5-to-Man6 acceptor reaction. Its alpha-1,3 parent
GO:0000033 is below GO:0000030 mannosyltransferase activity; the two Reactome MF
refinements and InterPro refinement preserve chemistry while improving
specificity. Broad valid ER and glycoprotein-biosynthesis assertions remain
ACCEPT at their original resolution. Cached Reactome R-HSA-446188 is the normal
sixth-mannose step; R-HSA-4720473 describes impaired variants in that normal
reaction context and does not show normal activity in all mutants.
R-HSA-446193 has an isolated wording error calling terminal glucoses GlcNAcs;
its pathway assignment is otherwise supported, and the cache is unchanged.
GO-CAM model 65c57c3400000687, activity 65d7e4ac00000326, already assigns
human Q92685 the exact activity, lumenal-side location and LLO biosynthesis using
the original human study. This is a second representation of the same evidence,
not an independent experiment. The two previous duplicate cores are consolidated
into one reaction core, retaining the two existing process assertions and using
the already seeded lumenal location. Zero NEW rows; zero new core BP terms.
ALG3 itself performs mannose transfer. Necessity-only stress, proliferation and
CREB3-processing observations are not used to add processes, and no pathway gap
is inferred from comparator absence.
DRAFT is retained pending five required normal caches: PMID:15840742,
PMID:30192950, PMID:38597022, PMID:40789468 and PMID:41807832. Identity/content
verification by primary web access is distinct from local cache availability;
VERIFIED reference assessments do not claim a missing cache exists. All other
PMID flags follow protected metadata. Missing records are neither fabricated nor
removed from the citation census to silence checks. Sources proposed for a later
recovery batch do not modify any already dispatched job.
Final local verification: just validate human ALG3 passed with one warning
category listing exactly the five missing references above. History validation
and HTML rendering passed. The immutable-source comparison passed for all
22 annotations, all 15 original reference identities and both alternative
products; all 50 cached supporting-text occurrences matched case-sensitively
after whitespace normalization. No casefold-only quote passes, YAML anchors or
aliases, or trailing YAML whitespace remain. The final four-file manifest records
byte-based Git blob IDs and SHA256 hashes; no protected source, Git or remote
state was changed.
Preflight confirmed all five canonical files against published head
92a9f6c44a0b03e01ae2bd1c7ddc9497f5fb8dbf. The five recovered publication
paths were absent from main 21364698b8b0c1099c7f26384bfd35b13596d282.
The normal fetcher ran in source7 run 36299519155, and the coordinator imported
the exact verified artifact bytes. No publication, UniProt, GOA or historical
record was edited. The earlier failed local-fetch statements above remain a
chronological record; the five missing-cache gates are now closed.
Actual recovered evidence was read before updating reference assessments:
All five reference identifiers and titles match the recovered normal records.
VERIFIED is retained on primary-source identity and supporting content; a failed
terminal fetch had not invalidated the separately documented primary browser
reading. Three local full-text flags are now corrected to available; the two
abstract-only flags remain unavailable. Seven short exact cached findings are
attached, including the explicit limitations.
For review comment 5854233340, the two annotation rationales now state the
cached abstract's positive catalytic/orthology or interaction findings. The
earlier figure-level external readings remain documented above with their actual
source routes. Generic binding removal does not deny the interactions or
identify ALG3 as a protease. Broad valid ER and glycoprotein-process assertions
remain ACCEPT; unresolved historical ARBA matching remains distinguished from
independently supported target biology. No changes to the 22 source assertions,
their 14 ACCEPT / 3 MODIFY / 5 REMOVE actions, the two alternative products, the
integrated core or any reference identity are required. No NEW was added.
The recursive authored-source census includes notes, DOI links, publisher and
author-uploaded PDF routes: 11 required PMIDs and three Reactome records are
present. The PDF/publication routes identify the already listed 10581255,
29547901 and 30192950; no additional DOI-only work was found. There is no
provider report in the gene directory. Unrelated bibliography entries in the
protected UniProt source are not promoted to review claims.
Final checks: just validate human ALG3 passes with zero warnings, as do the
new history validation and HTML rendering. Status is COMPLETE. All 57 cached
supporting-text occurrences match case-sensitively after whitespace
normalization; the 22 assertions and actions, 20 reference identities/titles,
two alternative products, one core, immutable inputs and published history are
unchanged. The manifest lists four authored files plus five exact normal caches.