FH (fumarate hydratase, P07954) — curation notes

Context

Selected from projects/MITOTOL.md (section 3, "Same activity, unrelated families"), based on
Chen et al. 2026 PMID:42822426. That paper is computational and comparative. It is used here
only for evolutionary context, not as evidence for FH function.

Deep research

just deep-research-falcon human FH --fallback perplexity-lite FAILED (2026-10-09).
- Falcon did not return before its timeout.
- The perplexity-lite fallback failed with "Provider 'perplexity' not available. Available:
falcon, asta, openscientist".
- No deep-research file was created at review time.
- Update: the Falcon job outlived the recipe's 600 s timeout and later wrote
FH-deep-research-falcon.md. It agrees with the review: cytosolic FH removes fumarate released by
argininosuccinate lyase but does not catalyse a urea-cycle step, and nuclear FH (DNA-PK-phosphorylated
at Thr236) binds H2A.Z at double-strand breaks, where local fumarate inhibits KDM2B to promote NHEJ.
No annotation decisions changed.
- Literature research was done instead from the cached publications (GOA references) and PubMed
searches. Four papers were added: PMID:11865300, PMID:16098467, PMID:22014577 and
PMID:24280230.

Core biology (with provenance)

• Activity. FH reversibly hydrates fumarate to L-malate. PMID:30761759
  Human FH has been characterised by kinetics and by a 1.8 Å structure PMID:30761759.
• Homotetramer. Each active site is built from residues of three subunits, so disease
  variants that break oligomerisation lose activity PMID:29456767.
• Class II (FumC-type). The human cDNA is most similar to B. subtilis and E. coli FumC fumarases PMID:3828494. These are the iron-independent class II enzymes.
• Two products from one gene. Alternative transcription initiation gives products with and
  without the mitochondrial targeting sequence PMID:27037871.
• Mitochondrial matrix / TCA cycle PMID:27037871
• Cytosolic pool PMID:20231875
• Patient mutations affect both isoenzymes PMID:8200987
• Cytosolic FH and the urea cycle. FH consumes the fumarate co-product of argininosuccinate lyase (ASL).
  PMID:27554470
  When FH is lost, ASL runs in reverse PMID:24280230.
  Judgement: FH does a step on the fumarate branch (it removes ASL product), not a step of the
  ornithine-to-arginine cycle itself. Keep the urea cycle annotation as non-core. Mark arginine
  metabolic process as over-annotation, because FH acts on neither arginine nor its precursors.
• Nuclear DNA damage response / NHEJ.
• Human cells: FH enters the nucleus after IR or HU PMID:20231875. In yeast this function needs enzyme activity and can be complemented by fumarate.
• Mechanism PMID:26237645
• Direct binding to H2A.Z PMID:26237645
• FH at DSBs by ChIP PMID:26237645
• Outcome PMID:26237645
• Judgement: add NEW GO:2001034 (positive regulation of DSB repair via NHEJ) and GO:0035861
  (site of double-strand break). FH does part of the work, because its catalytic product is the
  signal. It does not join DNA ends, so a regulation term fits better than the NHEJ process term.
• Tumour suppressor (HLRCC). PMID:11865300
  Two oncometabolite mechanisms of fumarate:
• HIF prolyl hydroxylase inhibition PMID:16098467
• KEAP1 succination PMID:22014577
  These are consequences of FH loss and are not FH activities, so no GO term was proposed.
• Thermogenesis (mouse adipose knockout). PMID:27554470
  This is an indirect consequence of TCA failure. Positive regulation of cold-induced
  thermogenesis is marked as over-annotation.
• Evolution (MITOTOL). PMID:42822426
  The pair is class I ([4Fe-4S], O2-sensitive) and class II fumarases. Humans retain only class II
  (FH), so the IBA for fumarate hydratase activity has no paralog or family ambiguity in human.

Annotation decisions summary

• 59 GOA rows reviewed, plus 2 NEW.
• MODIFY:
• catalytic activity → fumarate hydratase activity
• protein binding with H2A.Z (PMID:26237645) → histone binding (GO:0042393)
• REMOVE: 13 generic protein-binding rows from HT Y2H screens and other large-scale screens.
• MARK_AS_OVER_ANNOTATED: arginine metabolic process (x2), cold-induced thermogenesis (x2), extracellular exosome.
• KEEP_AS_NON_CORE: urea cycle (x2).
• All other rows ACCEPT: MF, TCA, fumarate/malate metabolism, and mitochondrion/matrix/cytosol/nucleus/chromosome locations.

2026-10-28: ClinGen follow-up evidence audit

This entry supersedes the earlier decision summary where the decisions differ. The original notes and quotations above are preserved as session history. The current YAML and evidence audit record the revised decisions; the source guide identifies the evidence and its limits. No new publication quotations were added.

• Exact publication-and-protein IntAct queries support the 13 formerly removed generic-binding rows. Eleven now remain non-core. Binding to PIM1 and LMNB2 is refined to protein kinase binding and lamin binding, respectively, without transferring the partner's activity to FH. The FH–EXOSC5 BioID row is proximity evidence, with independent binary evidence for the same pair from PMID:32296183. The PMID:39468017 correction concerns affiliation only.
• Arginine metabolism and cold-induced thermogenesis remain non-core contextual annotations. The actual mouse donor and human FH rescue constructs in PMID:23643539 support fumarate coupling to ASL chemistry; ASL performs the arginine conversion. PMID:27554470 supports the mouse adipose cold-challenge phenotype. Neither result establishes a universal normal-human flux direction or a dedicated temperature-sensing activity of human FH.
• Cytosolic core wording is limited to reversible fumarate/malate catalysis and the supported metabolic contexts. The artificial FH(N) deletion of residues 1–21 in PMID:26237645 is distinct from the natural P07954-2 deletion of residues 1–43. Nuclear and cytosolic localization pools are not exclusive natural-product assignments.
• The positive NHEJ-regulation proposal remains because FH performs local fumarate-producing catalysis in the reported mechanism. Comparator and GO-CAM checks are qualified in the evidence audit. The experimentally supported chromosome annotation is refined to the site of a double-strand break; the broad electronic chromosome annotation is too general. The earlier authored NEW site-of-break row is retired in place to avoid duplicate active coverage, with its complete source and quotation content retained.
• PMID:29632342 corrects duplicated Figure 2d images and reports additional repetitions. Its operative correction text was read; corrected image pixels and deposited Figshare data were not inspected. This limitation remains explicit and does not invalidate every distinct assay in the original study.
• The exosome annotation is UNDECIDED because the exact FH supplementary entry was not recovered. No contamination claim or requirement for an exosome-specific function is inferred from that gap. Broad lyase activity is too general relative to the established fumarase activity. Mitochondrial IBA is retained with direct target support, without claiming a new reconstruction of its ancestral placement.

The candidate preserves all 59 imported source contexts, both pre-existing authored NEW objects, both natural products and all 78 inherited quotation associations. There are 61 stored annotation rows and 60 active rows: 34 ACCEPT, 17 KEEP_AS_NON_CORE, 5 MODIFY, 2 TOO_GENERAL, 1 UNDECIDED and 1 NEW. Three core functions and 34 references are retained. Two genuine publication caches were added (PMID:23643539 and PMID:29632342); existing cached sources and provider research files were not edited.

2026-10-28: Follow-up to PR #5658 review

This follow-up supersedes the preceding cold-thermogenesis decision and counts. Re-reading PMID:27554470 Discussion and Figure 8 supports the reviewer's concern: severe adipose mitochondrial dysfunction, ATP depletion and reduced BAT UCP1 establish a metabolic requirement for thermogenic capacity, but the experiments do not separate a positive regulatory FH step from that general requirement. Both GO:0120162 rows are now MARK_AS_OVER_ANNOTATED. The observed phenotype remains valid, and no claim is made that every TCA perturbation produces it.

The arginine annotations remain non-core for a different, positive reason. PMID:23643539 combines isotope tracing with compartment-specific rescue to link FH-catalysed fumarate turnover to ASL reaction flux. Arginine metabolic process includes participation in a pathway, not only immediate use of arginine as a substrate. ASL performs the arginine conversion; FH performs the coupled fumarate chemistry. This bounded interpretation does not assert normal-human flux or a unique natural-product assignment from an engineered construct.

Four newly added, reference-only correction notices were removed from positive supported_by lists. Their reference_review assessments and applicable additional_reference_ids remain, so the Figure 2d correction and affiliation notice retain their proper context. The evidence audit is now a cited file: reference in the relevant rows and core summaries. No publication quotation was added or altered.

The biological summary and cytosolic/nuclear core descriptions now describe the compartmental mechanisms directly. Product and model limits remain in annotation reasons, source notes and questions. The summary restores the supported enzyme classification and DNA-PK/Thr236/H2A.Z mechanism without asserting a newly checked alternative-initiation mechanism or ancestral family history.

The campaign's explicit instruction to retain supported generic binding as non-core remains in force. PIM1 and lamin-B2 refinement describes the measured partner class, not kinase activity or lamina assembly by FH; related screening and validation records from one paper are not treated as independent biological replications. The unresolved exosome target entry and source-specific chromosome dispositions are unchanged.

Current totals: 61 stored rows, 60 active; 34 ACCEPT, 15 KEEP_AS_NON_CORE, 5 MODIFY, 2 TOO_GENERAL, 2 MARK_AS_OVER_ANNOTATED, 1 UNDECIDED and 1 NEW. There are 35 references including the evidence audit, three core functions, two natural products and the same 78 inherited quotation associations.