Gene: ALDH4A1 (aka ALDH4, P5CDH). Human. UniProt P30038. 563 aa precursor;
N-terminal mitochondrial transit peptide (residues 1-24), mature chain 25-563.
ALDH4A1 is the mitochondrial NAD+-dependent delta-1-pyrroline-5-carboxylate
dehydrogenase (P5CDH), an aldehyde dehydrogenase (ALDH superfamily, ALDH4
family). It catalyzes the second (final) step of proline catabolism:
oxidation of L-glutamate-5-semialdehyde (glutamate gamma-semialdehyde, GSA,
in spontaneous equilibrium with its cyclic tautomer delta-1-pyrroline-5-carboxylate,
P5C) to L-glutamate, using NAD+.
The substrate P5C/GSA is generated from proline by proline dehydrogenase (PRODH),
and from ornithine by ornithine aminotransferase; thus P5CDH sits at the node
interconnecting the urea and TCA cycles (glutamate is fed into central metabolism)
[PMID:22516612 "GSA is the hydrolysis product of Δ1-pyrroline-5-carboxylate (P5C),
which is generated from proline by proline dehydrogenase (PRODH)"; "GSA is also
produced from ornithine by ornithine aminotransferase"].
Cloning/characterization: two full-length human P5CDh cDNAs encode a 563-residue
protein; expression in P5CDh-deficient yeast confers P5CDH activity and growth on
proline [PMID:8621661 "a mitochondrial matrix NAD(+)-dependent dehydrogenase,
catalyzes the second step of the proline degradation pathway"; "Both conferred
measurable P5CDh activity and the ability to grow on proline as a sole nitrogen
source"]. (Note: this 1996 paper used the older EC 1.5.1.12; current EC is 1.2.1.88.)
Human P5CDH is also the second enzyme of hydroxyproline (4-Hyp) catabolism.
By analogy to proline, trans-4-hydroxy-L-proline is oxidized to
1-pyrroline-3-hydroxy-5-carboxylate (3OH-P5C) by hydroxyproline oxidase; the
nonenzymatic hydrolysis product 4-hydroxyglutamate semialdehyde (OH-GSA) is then
oxidized by P5CDH to 4-erythro-hydroxy-L-glutamate (OH-Glu)
[PMID:22516612 "P5CDH is also the second enzyme of hydroxyproline catabolism in
humans"; "which is oxidized to 4-erythro-hydroxy-L-glutamate (OH-Glu) by P5CDH"].
The GO term GO:0003842 explicitly covers this dual activity ("The activity can also
oxidize other 1-pyrrolines, e.g. oxidation of 3-hydroxy-1-pyrroline-5-carboxylate to
4-hydroxyglutamate") per its OLS definition.
PMID:21998747 (Riedel et al. 2011) is primarily about HOGA1 (4-hydroxy-2-oxoglutarate
aldolase), the terminal enzyme of the 4-Hyp pathway, but its Figure 1 / pathway
description places 1P5CDH as the second of four mitochondrial 4-Hyp-degrading enzymes
PMID:21998747. This supports
GO:0019470 (trans-4-hydroxy-L-proline catabolic process) TAS for ALDH4A1.
First human P5CDH crystal structure solved by Srivastava et al. 2012 (PMID:22516612).
Classic ALDH fold: N-terminal NAD+-binding (Rossmann-like) domain, C-terminal
catalytic domain furnishing the essential cysteine nucleophile Cys348 and
substrate-binding Ser349, plus an oligomerization domain. Homodimer (domain-swapped),
~122 kDa in solution [PMID:22516612 "This domain furnishes the essential cysteine
nucleophile (Cys348) and several residues that bind GSA, including Ser349";
"The molar mass is estimated to be 122 kDa"]. Kinetics: Km(NAD+)=100 uM,
Km(L-P5C)=32 uM, kcat=10 s-1 for wild-type HsP5CDH.
Mechanism (general ALDH): nucleophilic attack by conserved Cys on the aldehyde C to
form a hemithioacetal, hydride transfer to NAD+ giving a thioacyl intermediate and
NADH, then hydrolysis to the carboxylic acid product.
Autosomal recessive; deficiency of P5CDH causes accumulation of P5C and proline
[file:human/ALDH4A1/ALDH4A1-uniprot.txt "Hyperprolinemia 2 (HYRPRO2)"; PMID:22516612
"Type II hyperprolinemia is an autosomal recessive disorder caused by a deficiency
in Δ(1)-pyrroline-5-carboxylate dehydrogenase (P5CDH; also known as ALDH4A1)"].
The disease mutation S352L abolishes catalytic activity and eliminates NAD+ binding
by inducing an ~8 Å rearrangement of the catalytic loop PMID:22516612.
S352L and variant P16L were originally reported by Geraghty et al. 1998
(PubMed:9700195; not in local publications cache).
Hsu et al. 2025 (Nat Cell Biol) propose that ALDH4A1 is a third, catalysis-independent
component of the mitochondrial pyruvate carrier
PMID:40355545.
Lines of evidence in that paper:
Three independent 2025 cryo-EM analyses of human MPC resolve only MPC1 and MPC2:
Important caveat that limits how much these structures can refute. All three were
determined from heterologously co-expressed, tagged MPC1 and MPC2 only
PMID:40044865
PMID:40691140.
ALDH4A1 was therefore never in the sample. Not seeing a subunit you did not put in is not
evidence that the subunit is absent in vivo. The honest statement of what these structures
do establish is narrower but still relevant: the pyruvate permeation pathway and its
conformational cycle are fully contained within the MPC1-MPC2 heterodimer, so ALDH4A1
cannot be part of the translocation path itself.
Two further points cut against the "third component" framing:
Recorded, not annotated. No GO annotation is proposed for the MPC role, for three
reasons: (i) it is a single-laboratory result with no independent replication; (ii) the
paper's own data show MPC1+MPC2 transporting without ALDH4A1, so a "component of the
pyruvate carrier" CC assertion would overstate what was shown; (iii) an involved_in
mitochondrial pyruvate transport BP annotation cannot presently be separated from the
indirect possibility that losing P5C dehydrogenase from the matrix perturbs pyruvate flux
metabolically. The claim is instead captured as a knowledge_gap, in suggested_questions
and suggested_experiments, and in the top-level description as a statement about the
state of the biology; PMID:40355545 is marked correctness: DISPUTED with the reasoning
recorded in review_notes.
This should be revisited if the native complex is isolated from mitochondria with a defined
stoichiometry, or if the proteoliposome stimulation is independently reproduced.
This entry supersedes the earlier action recommendations and the 2026-09-17 MPC
interpretation above; the historical text is retained as provenance. HGNC:406 is
Approved ALDH4A1, with previous symbol ALDH4 and alias P5CDh. The immutable record
is human UniProt P30038. Parent preflight and an independent byte comparison found
all five canonical files identical to main
ba3ff58d7d2de76dbe3c24b16e05e12369f463fc; canonical and both alias PR searches
were empty. All 17 seeded source objects and all three alternative products remain
unchanged. This is a full audit despite the previous COMPLETE label.
The genuine Falcon request used a 1200-second timeout and the configured
perplexity-lite fallback, concurrently with normal publication caching. Both
provider commands failed during retrieval of deep-research-client from PyPI
because DNS could not resolve the host, before either research provider was
contacted. No provider report was created. The normal gene-publication command
reused all ten requested canonical PMID records. The terminal logs are
/tmp/ALDH4A1-provider.log and /tmp/ALDH4A1-fetch.log; they are local operational
receipts, not independent biological evidence. The manual research is recorded
here rather than impersonating a provider artifact.
The notes-inclusive census also found the historical PubMed:9700195 citation
above. A normal fetch-pmid 9700195 attempt failed with the URL/DNS error
nodename nor servname provided, or not known and produced no cache
(/tmp/ALDH4A1-extra-fetch.log). That historical source remains an explicit
notes-only cache gate; it is not newly used to support a decision. All ten PMID
references in the YAML and all three cited Reactome records are cached. The
review remains DRAFT pending recovery of this notes citation.
The canonical full papers read were PMID:22516612, PMID:21998747,
PMID:34800366, PMID:40355545, PMID:40044865 and PMID:40691140. The local records for
PMID:8621661, PMID:4015840, PMID:32814053 and PMID:40101766 are abstract-only;
the metadata flags retain those distinctions. For the interaction-screen design,
the original full PMID:32814053 was additionally read at the
MDC accepted-paper route,
previously downloaded as /tmp/ACTA1-PMID32814053.pdf. This does not imply recovery
of its ALDH4A1-specific supplementary interaction entry. The Nature primary
PMID:40355545 article agrees
with the cached full paper. These external checks were performed on 2026-09-27.
PMID:22516612 directly measures recombinant human GSA/NAD+ turnover and human
apo/mutant structures. Human equilibrium analytical ultracentrifugation gives a
122 kDa dimer estimate, in agreement with its domain-swapped crystal dimer. The
ligand complexes used to analyze substrate recognition are mouse P5CDH; the
hydroxylated-substrate accommodation in Figure S8 is a structural model. The paper
states established hydroxyproline-pathway activity, but does not newly measure
human hydroxylated-GSA turnover. The current reference finding and hydroxyproline
reason now make that distinction. PMID:21998747 is a HOGA1 study whose introduction
and pathway diagram place P5CDH in the four-enzyme mitochondrial pathway, not a
new P5CDH enzyme assay. The preserved TAS process is supported as catalytic pathway
participation, with the corresponding cached Reactome reaction as corroboration.
Broad mitochondrial annotations retain ACCEPT at their original source resolution.
The generic oxidoreductase MF is refined by MODIFY to the directly measured
GO:0003842 subtype; the broad class remains biologically correct. Separate matrix
annotations supply compartment precision. Human homodimeric self-association is an established property of the
core enzyme and is also ACCEPT, without creating a separate binding core.
The live HPA subcellular page
reports supported mitochondrial staining in HaCaT and Hep-G2 cells and cytosolic
staining in A-431. The mitochondrial HTP paper has cached full text; its exact
ALDH4A1 supplementary row was not independently recovered. Its broad curated
assignment is retained with independent localization support, not a claim that
all organellar assignments were individually remeasured in this audit.
The original CACAO/GONUTS annotation record
traces the broad NAD+-aldehyde-dehydrogenase assignment to CACAO8806 and Table 1 of
PMID:4015840(https://gowiki.tamu.edu/wiki/index.php/PMID:4015840), with agar-overlay
staining using propionaldehyde or benzaldehyde. The full Table 1 was not recovered.
The abstract's emphasis on other ALDH forms therefore does not establish a wrong
protein assignment. The curated broad function is biologically sound; MODIFY refines it to the
independently measured human P5CDH reaction. This refinement does not claim that
the unread original table assayed GSA. The detailed original substrate assay
remains curator-deferred rather than independently verified.
The generic GO:0005515 record is REMOVE as uninformative, not because the
ALDH4A1-RBBP4 interaction is disproven. The immutable UniProt record lists three
experiments; the original screen performs repeated screens and pairwise retests.
The exact supplementary pair and a mechanistic RBBP4-dependent ALDH4A1 activity
remain unverified. No specific adaptor or scaffold function is substituted.
For propagation, PTN002684348 is the sole IBA source entity. Human self-evidence
is legitimate descendant grounding. The PAINT tree/node placement, ARBA internal
conditions and missing live InterPro entries were not reconstructed; their source
reviews explicitly retain UNRESOLVED internals where appropriate. Positive target
judgments rely on the independently established human reaction and location, not
fabricated verification of automated rules. The source mappings to EC:1.2.1.88,
RHEA:30235 and SL-0170 are consistent with the immutable human record.
The current decision adds one source-specific IDA proposal for GO:0141109
transporter activator activity. It does not add a transport-process annotation,
a pyruvate-carrier molecular function or an obligate third-subunit complex term.
ALDH4A1 performs the proposed regulatory work by binding the carrier and enhancing
its measured transport. This assignment does not rest on knockout necessity.
PMID:40355545 Figure 5 reports binding of recombinant ALDH4A1 to MPC1 and MPC2,
with human-cell association and interaction-defective deletion/rescue experiments.
The deletion of residues 182–199 loses MPC association and pyruvate-import rescue
while retaining the measured cellular proline-regulation phenotype; this is not a
complete purified-enzyme kinetic characterization of that deletion. Figure 6 and
Extended Data Figure 8 report enhanced MPC association/oligomerization with added
ALDH4A1. Figure 7 compares empty proteoliposomes, individual MPC subunits,
MPC1+MPC2, and MPC1+MPC2 with WT or catalytic S352L ALDH4A1. MPC1+MPC2 already
transports; added WT or S352L increases transport over time. The source explicitly
states [PMID:40355545, Results, "In vitro binding assay demonstrated the direct
interaction of ALDH4A1 with both MPC1 and MPC2"] and [PMID:40355545, Results,
"Either wild-type ALDH4A1 or mutant ALDH4A1(S352L) further enhanced pyruvate
transport by MPC1–MPC2 in a time-dependent manner compared with MPC1–MPC2"].
The reconstitution Methods use purified recombinant proteins, phosphatidylcholine
and cardiolipin, detergent removal and radiolabeled uptake/filter washing. The
recovered description does not provide an ALDH4A1-alone transport arm. It also
does not settle matched incorporated MPC amount/orientation for every comparison,
or distinguish stabilization during reconstitution from an increase in turnover
per incorporated carrier. Thus the evidence supports a transporter-activating
contribution but does not establish standalone carrier activity or a universal
native stoichiometry. The paper's gel filtration, crosslinking and native-gel data
are positive complex evidence; the previously stated requirement for any native
isolation is outdated. Those data still do not quantify a unique native 1:1:1
stoichiometry. A peer independently read these Results/Methods and reached the
same bounded interpretation.
The live GO:0141109 definition and parents
were checked on 2026-09-27. The definition requires binding and increased
transporter activity; its is_a parents are transporter regulator activity and
molecular function activator activity. It is not a carrier-activity assertion.
The proposed term is not an ancestor or descendant of any retained seeded term,
and there is no second NEW term. The target accession/name search of the cached
gocams/index.tsv found no ALDH4A1/P30038 activity; it therefore supplied no
existing target model to resolve or contradict this role.
The comparator check considered transporter-associated proteins whose regulatory
role is distinct from substrate translocation. Primary UniProt annotation displays
show the same term on rat Pdzk1/Q9JJ40
(source RGD), mouse Cltrn/Q9ESG4
(source GO_Central), and rat Atp1b2/P13638
(source RGD). Their roles span a binding scaffold, a carrier-binding trafficking/
activity regulator, and a noncatalytic pump partner. They establish that GO uses
this MF for transporter-associated regulatory proteins; they do not transfer
those proteins' particular mechanisms to ALDH4A1. These were annotation-display
checks, not independent re-reviews of every donor experiment or exact evidence
chain. Local authored NEW proposals in other reviews were not treated as curated
comparator assertions. QuickGO API requests did not yield an independently
usable annotation export, so no comprehensive species-wide absence claim is made.
The MPC1-MPC2 structural papers constrain the translocation pathway to the
heterodimer. They do not refute accessory regulation. Co-expression of two tagged
subunits does not prove that endogenous host ALDH4A1 could never be present;
the previous categorical sample-exclusion sentence is withdrawn. PMID:40044865
and PMID:40691140 full Methods were read; PMID:40101766 remains limited to its
primary abstract/publisher record. PMID:40355545 is VERIFIED for the bounded
binding/activation claim, rather than DISPUTED merely for lacking an independent
replication. Native occupancy, stoichiometry, physiological distribution and the
precise activation mechanism remain explicit questions. The newer activity is
kept outside the integrated defining P5CDH catalytic core at this stage.
The current 18-entry review has 14 ACCEPT, two enzyme-specificity MODIFY actions,
one generic-binding REMOVE and one NEW. The catalytic core integrates proline and hydroxyproline degradation under
the same enzyme MF in the matrix. No second broad core duplicates that chemistry.
Validation note: live AmiGO labels GO:0003842 with an explicit (NAD+) suffix,
whereas the repository validator expects the shorter name already present in
the source snapshot. Authored labels use the validator-compatible name for the
same term ID; NAD+ chemistry remains explicit in the prose. No source term ID or
label was changed.
Final checks: targeted gene validation and rendering passed. All 35 attached
quotations matched the canonical sources under whitespace normalization. The
17 seeded source objects, 18 original reference identity/title pairs, three
alternative products, UniProt and GOA bytes were preserved. Parent independent
review accepted the biological decisions after the two enzyme-specificity
refinements and added core matrix evidence. The YAML validator has no curation
warnings; DRAFT is retained for the notes-only PMID:9700195 cache gate.
This entry supersedes the earlier PMID:9700195 cache gate. The source6 normal-fetch record was imported unchanged after archive/run and exact-byte verification; its SHA256 aab24fdf9bae81f8c5361c73c65dc03a8a0b1f36d3b819ebc5a9853312aa9895 and Git blob c3e6aee5591fcbf1f046e1469b8ebe4db25a6773 match tmp/source6-canonical-import-receipt.json. Before editing, all five local canonical gene files were matched to published PR #3271 head 8b86954c33fecbad3810eea861a2db0749e01c9d through GitHub blob IDs.
The recovered abstract, not a full paper, was read. PMID:9700195 reports variants identified in four people and tests human P5CDH constructs in a deficient Saccharomyces cerevisiae strain. Wild-type human enzyme restores activity and growth on proline; S352L and G521fs(+1) do not. P16L produces functional enzyme and is interpreted as a population polymorphism. This corroborates the existing S352L catalytic-loss interpretation without converting the heterologous yeast host into a direct human-cell assay, generalizing inactivity to P16L, or changing the independently assessed MPC-regulatory evidence. No functional action, core, source assertion or reference identity is changed.
The recursive citation census covered all non-derived gene files and links, with the HTML treated as a derived duplicate. There is no provider or hypothesis report and no nested PDF artifact. The author-PDF link https://edoc.mdc-berlin.de/id/eprint/19322/1/19322oa.pdf is explicitly linked to the already cached PMID:32814053; it creates no separate unidentified source. All 11 PMIDs used by the authored review and notes as evidence/context are now cached, as are the three authored Reactome reaction references. Fifteen normalized DOI strings were checked against publication metadata or the paired PMID/DOI in the immutable UniProt record. Six absent raw-UniProt bibliography records remain unused by the review: PMID:1395511, PMID:1286669, PMID:8493898, PMID:23186163, PMID:24275569 and PMID:25944712. They concern historical protein characterization or broad proteomics, but no reviewed assertion depends on their uninspected results. They remain inventoried raw-source provenance, rather than new evidence claims or completion gates. All raw Reactome cross-references are cached. Current-main presence checks were made at 30a9290881824baaacab2b681d7808f572c4cef6; the newly recovered PMID:9700195 is explicitly included in this follow-up manifest.
The sole missing required cache is therefore resolved. Status is changed to COMPLETE subject to the final warning-free schema, ontology, reference and best-practices validation. All 17 seeded source objects, the one prior NEW assertion, all 18 decisions, the integrated catalytic core, three alternative products and protected UniProt/GOA bytes remain unchanged. No additional provider attempt, scientific claim or new annotation is introduced by this bounded follow-up. The earlier genuine provider/fetch failures remain historical provenance.
Final source6 follow-up checks: full gene validation passed with no curation warnings; all 35 case-sensitive source quotes and all preserved source/isoform/reference/core objects passed. COMPLETE records the closed cache gate and successful validation, without claiming that the open mechanistic questions have been experimentally resolved.
The five canonical files matched published head 7bcd63b89f37004ce686e24f83bdc2625eede75f before this follow-up. Formal review 5329918746 and full comment 5854960148 were read. This entry supersedes the earlier argument for replacing the experimental GO:0004029 assertion with GO:0003842: that replacement conflated independently established GSA chemistry with the substrate scope of a different historical assay.
The original CACAO8806 record and AL4A1/P30038 page were rechecked through their indexed original pages. They explicitly identify human ALDH4A1 and Table 1 agar-overlay assays with propionaldehyde or benzaldehyde. The primary table itself remains unavailable. GO:0004029 is therefore retained as KEEP_AS_NON_CORE, deferring to the experimental curator while making no physiological-substrate claim. Its GSA replacement and unrelated GSA quote are removed. PMID:4015840 has a verified identifier and curator trail, but its exact primary isozyme support remains UNVERIFIED. The abstract's emphasis on other ALDH forms is not evidence of a wrong-gene annotation. The separate direct-human P5CDH assertion and integrated catalytic core remain intact.
The earlier statement that the PTN was the sole IBA source entity is also superseded. The immutable GOA row records MGI:MGI:2443883|PANTHER:PTN002684348|UniProtKB:P30038. All three are now inventoried in the propagation review. The mouse descendant's individual experiment and exact PAINT node placement remain un-reconstructed; target self-evidence is legitimate grounding for the inherited assertion. No donor-count argument or inferred tree topology is introduced.
Cached PMID:9700195 is now represented in references, with its exact abstract quote and human-allele/yeast-host distinction. S352L and G521fs(+1) lack detectable activity in deficient yeast, whereas P16L remains functional. The already proposed GO:0141109 receives the explicit enables qualifier, and the PMID:40355545 assessment distinguishes its measured binding/transport-enhancement result from the uncorroborated universal third-subunit framing. No new annotation is introduced.
The optional broad-location and self-association suggestions do not warrant action changes. Mitochondrion describes the core enzyme's compartment at the original assay resolution; the narrower matrix summary does not make that same location non-core. Human homodimerization is an integral property of the catalytic unit and need not have a separate binding-only core entry. Both locations and self-association remain ACCEPT for these biological reasons.
All 17 original source assertions, three alternative products and the integrated core are preserved. The sole action change is the historical assay MODIFY to KEEP_AS_NON_CORE. All 11 required PMIDs and three Reactome records remain cached; the prior DOI-inclusive census and six explicitly unused raw-bibliography exclusions still apply. There are no new provider artifacts or source dependencies.
Full targeted validation passed without curation warnings. All 35 case-sensitive, whitespace-normalized cached quotations and source-preservation checks passed. Independent review of every changed biological block and this notes addition concurred, including a separate check of the original indexed CACAO source. Status remains COMPLETE.