ALAS2 (human) — review notes

UniProtKB: P22557 (HEM0_HUMAN). Gene: ALAS2 (HGNC:397; synonyms ALASE, ASB). X chromosome (Xp11.21).
EC 2.3.1.37. Belongs to the class-II PLP-dependent aminotransferase family.

Core biology (grounded in ALAS2-uniprot.txt + cached publications)

• ALAS2 is the erythroid-specific 5-aminolevulinate synthase, the first and rate-limiting
  enzyme of heme biosynthesis in erythroid cells.
  PMID:32499479
  PMID:34492704
• Catalyses the PLP-dependent condensation of succinyl-CoA and glycine to form aminolevulinic acid
  (ALA), with CoA and CO2 as by-products.
  [file:human/ALAS2/ALAS2-uniprot.txt "Catalyzes the pyridoxal 5'-phosphate (PLP)-dependent condensation of succinyl-CoA and glycine to form aminolevulinic acid (ALA), with CoA and CO2 as by-products"]
  PMID:32499479
  Rhea:RHEA:12921; EC=2.3.1.37.
• ALAS2 is PLP-dependent; PLP is covalently attached to active-site Lys391 as an internal aldimine
  PMID:32499479.
  UniProt: MOD_RES 391 N6-(pyridoxal phosphate)lysine; COFACTOR pyridoxal 5'-phosphate.
• Two isozymes: ALAS1 (housekeeping/ubiquitous, 3p21.2) and ALAS2 (erythroid, Xp11.21).
  PMID:32499479
• Homodimer (obligate; each active site formed from both subunits).
  PMID:32499479
• Localization: mitochondrion. UniProt SUBCELLULAR LOCATION = Mitochondrion inner membrane,
  peripheral membrane protein; localizes to the matrix side of the inner membrane.
  [file:human/ALAS2/ALAS2-uniprot.txt "Localizes to the matrix side of the mitochondrion inner membrane"]
  Reactome (R-HSA-189442) and InterPro place it in the mitochondrial matrix. The N-terminal 49-aa
  transit peptide targets the protein to mitochondria.
• Regulation: IRE in the 5'UTR couples translation to iron availability
  PMID:2050125;
  protein stabilized by hypoxia / proteasome inhibition (LXXLAP prolyl-hydroxylation/vHL pathway)
  PMID:16234850;
  competitively inhibited by itaconyl-CoA PMID:34492704; C-terminal autoinhibitory loop
  PMID:32499479.
• Interactions: SUCLA2 (β subunit of ATP-specific succinyl-CoA synthetase; UniProt Q9P2R7) —
  physiologically meaningful (channels/supplies succinyl-CoA).
  PMID:10727444
  BANP (Q8N9N5) from a large-scale Y2H interactome map PMID:25416956 — not functionally characterized.

Disease

• Loss-of-function missense mutations → X-linked sideroblastic anemia (XLSA / SIDBA1, MIM 300751);
  many characterized in vitro to reduce enzyme activity (K156E PMID:21252495; 10 missense in
  PMID:21309041).
• C-terminal frameshift/deletion gain-of-function mutations → X-linked (dominant) protoporphyria
  (XLDPT/XLDPP, MIM 300752); can also aggravate other erythropoietic disorders (e.g. CEP modifier,
  Y586F PMID:21653323).

GOA MF term

• GOA carries GO:0003870 5-aminolevulinate synthase activity (verified in ALAS2-goa.tsv).
  This is the exact current term; used for core_functions molecular_function.

Annotation review disposition summary

• MF 5-aminolevulinate synthase activity (GO:0003870): core; multiple EXP/IDA/IMP + IBA/IEA → ACCEPT.
• BP heme biosynthetic process (GO:0006783): core; ACCEPT (IBA/IEA/ISS/NAS).
• BP heme B biosynthetic process (GO:0006785, IDA): more specific; ACCEPT (heme b is the product of the pathway ALAS2 initiates).
• CC mitochondrion / mitochondrial matrix / mitochondrial inner membrane: ACCEPT (matrix is most informative; IMM is where it sits peripherally on the matrix side).
• BP hemoglobin biosynthetic process (GO:0042541): keep as non-core (downstream physiological role; heme feeds hemoglobin).
• BP erythrocyte differentiation (GO:0030218): keep as non-core (heme flux governs erythroid maturation).
• BP protein binding (GO:0005515, IPI): uninformative MF term → MARK_AS_OVER_ANNOTATED (SUCLA2, BANP, vHL interactions are real but the bare term adds nothing).
• MF transferase activity (GO:0016740, IEA InterPro): too general vs the specific acyltransferase → MODIFY to GO:0003870.
• BP porphyrin-containing compound metabolic process (GO:0006778) / tetrapyrrole biosynthetic process (GO:0033014): broad IEA parents of heme biosynthesis → ACCEPT (correct but general).
• MF pyridoxal phosphate binding (GO:0030170): secondary MF, correct cofactor → ACCEPT.
• BP intracellular iron ion homeostasis (GO:0006879, ISS): mark as non-core (ALAS2 consumes iron pathway indirectly via heme; ISS from ALAS1 ortholog); KEEP_AS_NON_CORE.
• BP response to hypoxia (GO:0001666, IDA) / intracellular oxygen homeostasis (GO:0032364, NAS): O2-dependent stability regulation from PMID:16234850 → KEEP_AS_NON_CORE (regulatory response, not core catalytic function).

2026-09-27 — ClinGen Mendelian campaign re-review

This dated audit supersedes the disposition and mechanistic wording in the earlier
notes above; the earlier journal entry is retained as history. All 40 seeded
assertions, qualifiers and reference identities were preserved. The approved
symbol is ALAS2 (HGNC:397), UniProt P22557; ALAS-E / ALASE and the previous symbol
ASB were checked separately for directory and open-PR overlap. The coordinator
verified all five canonical files against main
c7078166039c9abd5c62704489283403eb520007; no overlapping PR was found.

The review and annotation-reviewer workflows were applied to every annotation,
all references and the core synthesis. A peer independently checked the two
oxygen/localization papers; the coordinator independently reviewed the final
biology. Exact local preflight hashes are in /tmp/ALAS2-root-baseline.json and
the final publication manifest records the changed-file hashes.

Primary evidence and source boundaries

• The full cached human structure/kinetics study PMID:32499479 supports the
  PLP-dependent reaction, homodimeric active sites and C-terminal regulation.
  Its structural constructs include human residues 79–587 and 143–587 expressed
  in insect cells, with additional bacterial expression for kinetics/mutagenesis.
  PLP binding belongs within the catalytic core, not in a second redundant core.
• PMID:34492704 was read in its cached main full text. It reports recombinant
  ALAS2 inhibition and mouse erythroleukemia/macrophage experiments, including
  human ALAS2 overexpression in mouse cells. Detailed recombinant constructs
  remain in unrecovered supplementary Methods. Itaconyl-CoA inhibits ALAS2;
  it is not established as an alternative substrate. No claim of direct human
  erythroblast or patient testing is inferred from these experiments.
• Full publisher PMID:10727444(https://www.jci.org/articles/view/6816) was read
  on 2026-09-27: Methods use mature human ALAS2 bait (residues 54–587) and human
  SUCLA2 in yeast, then tagged human proteins in CHO mitochondrial preparations.
  Physical association is established. The paper proposes efficient substrate
  use or import assistance; it does not demonstrate succinyl-CoA channeling.
  This corrects the categorical channeling statement in the earlier notes.
• PMID:14643893(https://pubmed.ncbi.nlm.nih.gov/14643893/) remains locally
  abstract-only; a publisher preview was also read at
  DOI 10.1016/s1357-2725(03)00246-2.
  It directly reports functional exon-4-skipped ALAS2, retained SUCLA2 binding
  and import dependent on the first 49 residues. Import across the inner
  membrane is not itself a membrane-residency assay. The seeded inner-membrane
  IDA is nevertheless retained by explicit curator deference to the established
  UniProt peripheral/matrix-face localization. Cached GO-CAM
  67369e7600001710 repeats that PMID for the location; it is not an independent
  experimental confirmation. The question of the exact attachment assay remains.
• PMID:16234850(https://pubmed.ncbi.nlm.nih.gov/16234850/) was independently
  verified from its primary abstract, but full Results/Methods were not
  recovered. K562 hypoxia, proteasome inhibition, proline-mutant stabilization,
  normoxic vHL co-immunoprecipitation and ubiquitination concern regulated ALAS2.
  The proposed hydroxylase pathway does not directly establish ALAS2
  hydroxylation or an oxygen-sensor function. The IDA response-to-hypoxia remains
  non-core. Intracellular oxygen homeostasis, GO:0032364, requires maintaining
  intracellular oxygen levels, so that NAS assertion is UNDECIDED pending the
  full source rather than being retained on the basis of a different observation.
• The human mutation studies PMID:21252495, PMID:21309041 and
  PMID:21653323 were read at cached primary-abstract scope. Their actual
  enzyme assays support the activity. The ten-variant study does not show the
  same defect for every variant, and Y586F was assessed as a modifier of
  UROS-associated congenital erythropoietic porphyria. The IRE/promoter paper
  PMID:2050125 is not treated as a direct assay of lineage specification.
• Cached full main texts PMID:25416956 and PMID:34800366 establish their
  interaction-map and MitoCoP designs, respectively. The exact ALAS2-BANP
  supplementary pair and ALAS2-specific MitoCoP entry were not independently
  recovered. No negative interaction result or contamination conclusion follows.
  Broad mitochondrial HTP remains ACCEPT with independent human localization.

Transfer provenance and process scope

The actual ISS/Compara donor is mouse Alas2 P08680, not an ALAS1 ortholog.
The primary UniProt donor record
and MGI comparative graph
were read on 2026-09-27. The graph explicitly records its generation date as
2023-03-10; it is historical evidence provenance, not a freshly generated graph.
It links mouse heme synthesis, hemoglobin synthesis and erythrocyte
differentiation to IMP PMID:9446639, and intracellular iron homeostasis to
IMP PMID:10562540 with allele MGI:2180151.

PMID:9446639(https://pubmed.ncbi.nlm.nih.gov/9446639/) was verified from the
primary abstract: mouse Alas2 disruption leaves early erythroid markers while
reducing heme, globin and hemoglobinized cells. Its full source was not recovered.
For PMID:10562540(https://pubmed.ncbi.nlm.nih.gov/10562540/), the primary abstract
and then the author-uploaded full original
were read. Figure 4 distinguishes embryonic cytoplasmic iron deposits from
adult-chimera ring sideroblasts. Table II directly quantifies approximately
threefold higher iron per cell in null E10.5 embryos, not higher total iron per
embryo. Discussion leaves the intervening iron-import mechanism unresolved.
Thus iron homeostasis and late differentiation remain contextual NON_CORE
annotations; ALAS2 is not assigned iron transport or lineage transcription.

Live GO:0042541
includes pathways producing hemoglobin with its heme groups. ALAS2 catalyzes an
actual step in producing that component, a defining erythroid specialization.
The three seeded hemoglobin-biosynthesis assertions are ACCEPT. They do not
require that ALAS2 translate globin or assemble the final protein. The integrated
core covers heme B and hemoglobin biosynthesis using the same catalytic unit.
Live GO:0032364 distinguishes
oxygen steady-state maintenance from oxygen-dependent stability.

IBA reviews preserve ancestral nodes PTN000343737 / PTN000343740 alone in
source_entities. The PAINT tree/MSA was not re-executed. The target's own
experimental evidence is legitimate descendant support, not circularity.
InterPro, EC, Rhea, UniProt subcellular mapping and the mouse ortholog sources
are separately described; ARBA00034153 internals remain UNRESOLVED while direct
human evidence establishes the target reaction. Cached human GO-CAM
67369e7600001710 includes ALAS2 synthase activity and its heme B pathway role.
No NEW annotation is proposed; the existing catalytic/process coverage suffices.

The four generic GO:0005515 annotations now use REMOVE under the explicit
annotation-reviewer policy, replacing the earlier prohibited OVER decisions.
This removes an uninformative term, not the experimentally reported SUCLA2/vHL
associations or the unresolved BANP pair. No specific adaptor, substrate
channeling or regulatory MF is manufactured from binding alone.

Research attempts and cache gates

Publication caching and default Falcon research were launched concurrently.
just fetch-gene-pmids human ALAS2 succeeded, finding all 11 original records
already cached (/tmp/ALAS2-fetch.log); those bytes were preserved. The first
provider invocation inherited offline mode and could not resolve its uncached
client. A normal network-enabled attempt with a 1200-second timeout and
--fallback perplexity-lite was then made. Both client launches failed before
provider contact because PyPI DNS lookup failed (/tmp/ALAS2-provider-online.log).
No provider report was created, and these notes are manual research.

The two genuinely traced donor papers were requested normally using
just fetch-pmid 10562540 9446639. Both failed DNS; no records were fabricated
(/tmp/ALAS2-donor-fetch.log). Their verified primary identities/access routes
are recorded separately from cache availability. The notes-inclusive census is
13 PMIDs with exactly these two missing caches, plus one already-cached
Reactome record, R-HSA-189442. The review remains DRAFT pending those cache gates
after completed validation. No immutable source or prior history was edited.

Final checks: just validate human ALAS2 passed with one aggregate warning
covering the two missing PMID caches; just validate-history and just render
passed. All 40 source objects, four alternative products and 19 original
reference identity pairs were asserted unchanged. Final actions are 29 ACCEPT,
five KEEP_AS_NON_CORE, four REMOVE, one MODIFY and one UNDECIDED. The multiline
UniProt matrix-face/peripheral quote was checked against exact source bytes.

2026-09-27 — source5 recovery and PR #3267 evidence follow-up

This entry supersedes the two missing-cache gates reported in the earlier dated
entry. The published review was checked byte-for-byte at head
f0ec347e93276d4a69d832bdb567440720a08be6; the five canonical gene files and
published history matched before editing. All 40 source assertions, four
alternative products, reference identities, annotation actions and the single
catalytic core are retained.

The normal recovery run
ran at commit 60c5e96f8317dd1e7d8325242d8037b81e272c59. Its original artifact
10926007634 has ZIP SHA256
cd022c3e045b00798f8e29ff86904be885a4ed9e8b539ba11d44f6b954149400.
The independently verified import copied both records without alteration:

• publications/PMID_10562540.md — SHA256 e2bf6f57b3192e5a06a9fe0b7a259fbe8464570ff49d4bcd87abf332b527ebe2.
• publications/PMID_9446639.md — SHA256 9bada88eacf51376ce40f1fd55715f2240b78b17bb77b2ac24b33ef3d3b784f1.

Both caches have full_text_available: false; the YAML therefore retains
full_text_unavailable: true. PMID:10562540's cached abstract directly contains
the previously quoted sentence about diffuse cytoplasmic iron accumulation,
resolving the in-repository evidence-verification item in
review comment 5853651866.
The earlier external full-paper Figure 4/Table II assessment remains separately
identified; the recovered cache does not contain those figure/table passages.
PMID:9446639's cached abstract preserves the distinction between unchanged early
erythroid markers and reduced heme, globin and hemoglobinized cells. No new
full-text experiment is inferred, and no biological decision changes.

Four ISS source comments now refer only to the actual mouse UniProt P08680
donor; the nonexistent Ensembl identifier claim is removed from those four
comments. Compara rows retain their real seeded Ensembl source identifiers.
The generic-binding row sourced to PMID:14643893 now also quotes that paper's
own splice-variant/SUCLA2-affinity abstract, with the later human structure paper
as independent support. Its REMOVE action concerns the uninformative term,
not the reported association.

The core/non-core distinction remains mechanistic: ALAS2 catalyzes a step in
biosynthesis of heme, a chemical component of hemoglobin. The late erythroid
maturation phenotype is a contextual consequence of heme supply and does not
identify ALAS2 as a cell-fate regulator. The oxygen-homeostasis row remains
UNDECIDED: the available source establishes oxygen-dependent regulation of
ALAS2, while the full paper needed to adjudicate any additional oxygen
steady-state evidence remains inaccessible. The reviewer explicitly accepted
this as a defensible source-access judgment.

The recursive authored YAML/notes/provider-artifact census contains 13 PMIDs
and Reactome R-HSA-189442, all now cached. No provider report or nested research
artifact exists. DOI scanning decodes URLs and Markdown escapes and removes
unbalanced URL punctuation: the sole DOI, 10.1016/s1357-2725(03)00246-2, matches
the actual cached metadata for PMID:14643893 on the published head. It adds no
new citation obligation. Source identities and hashes are checked against
exact published objects; the two recovered records are the only new source
files required for this follow-up. Earlier failed-fetch reports remain as
historical provenance, with their results stated in the dated text above.

Final follow-up checks: just validate human ALAS2 completed with exit 0 and no
curation warnings; schema/history validation and HTML rendering passed. All
quoted passages now have local source records and pass exact normalized-text
checks. With no outstanding citation-cache gate, status is COMPLETE. No
biological action, core term, machine-fetched source or published history was
rewritten.