HDAC5 (human, Q9UQL6) review notes

Deep research status: falcon (Edison) deep research failed with an authentication
error on 2026-10-24, and no other provider key was available. No -deep-research-*.md
file exists for this gene. The notes below come from the cached publications in
publications/, the UniProt record (HDAC5-uniprot.txt), and targeted web searches
(used only to resolve PMIDs, which were then cached with just fetch-pmid).

Why this gene was reviewed

The rule-based screen projects/FUNCTION_REQUIREMENT_RULES.md (FRR:0009) flagged
HDAC5's experimental GO:0001216 DNA-binding transcription activator activity (IMP,
PMID:19071119) because HDAC5 has no sequence-specific DNA-binding domain. Related rows
that assert DNA binding: GO:0000978 (IDA, PMID:24413532) and GO:0000976
contributes_to (IDA, PMID:19351956).

Identity and domain architecture

• Class IIa histone deacetylase, HD type 2 subfamily; 1122 aa; N-terminal extension
  with the MEF2-binding motif and the signal-responsive 14-3-3 sites (Ser259, Ser498),
  C-terminal HDAC catalytic domain from about residue 832
  PMID:10220385.
• UniProt: "Belongs to the histone deacetylase family. HD type 2 subfamily."
• No DNA-binding domain anywhere in the protein (UniProt features list only the
  deacetylase domain, NLS/NES, disordered and coiled-coil regions).

Catalytic activity: weak, probably not the main mechanism

• Original cloning paper: recombinant HDAC5 immunoprecipitates deacetylate all four
  core histones but less well than HDAC1/HDAC6, and HDAC5 co-purifies with HDAC3
  PMID:10220385
  PMID:10220385
• Activity of class II HDAC immunoprecipitates depends on the associated
  SMRT/N-CoR-HDAC3 complex
  PMID:11804585
• Vertebrate class IIa HDACs carry His in place of the catalytic Tyr; the reverse
  mutation boosts activity about 1000-fold in HDAC4, and HDAC5 behaves the same way
  PMID:17956988
  [PMID:17956988 "this \"gain of function phenotype\" could be extended to HDAC5 and -7"].
• Non-histone deacetylation reported: GCM1 (placenta) PMID:23867755
  and alpha-tubulin in injured sensory axons PMID:26157139.
  Neither study separates intrinsic HDAC5 catalysis from activity recruited through HDAC3.
• Judgement: the histone deacetylase activity rows are kept as over-annotations
  (MARK_AS_OVER_ANNOTATED): measurable in vitro activity of uncertain physiological
  relevance. Protein lysine deacetylase activity is kept as non-core.

Core mechanism: signal-regulated corepressor of MEF2 and other TFs

• Binds MEF2 and represses MEF2-dependent myogenesis
  PMID:10983972.
• Signal-dependent nucleocytoplasmic shuttling: CaMK phosphorylation and 14-3-3 binding
  export HDAC5 from the nucleus and derepress MEF2 targets
  PMID:11081517
  PMID:10869435
  PMID:15923258.
• Kinases: CaMK, AMPK (Ser259/Ser498; GLUT4 regulation) PMID:18184930,
  PKN1/2 at Thr292 in the NLS PMID:20188095, GRK5 in cardiomyocytes PMID:18711143.
• Other corepressor partners: HDAC3 / N-CoR / SMRT, HP1 PMID:12242305, BCoR
  PMID:10898795, RFXANK/ANKRA2 (MHC II repression) PMID:16236793, GATA3 (FUT7
  repression) PMID:17075044,
  GCM1 PMID:23867755, XBP1s PMID:20408817.
• Endothelium: HDAC5 represses FGF2 and SLIT2 and restrains sprouting angiogenesis,
  independent of MEF2 binding and of deacetylase activity
  PMID:19351956.
• Brain: loss of HDAC5 in mice sensitises to chronic cocaine and stress
  PMID:17988634.

The flagged DNA-binding/TF rows

• GO:0001216 (IMP, PMID:19071119, ARUK-UCL). The paper (abstract-only in cache) shows
  HDAC4/5 enhance HIF-1alpha transactivation by binding HIF-1alpha's inhibitory domain,
  displacing FIH-1 and favouring p300 binding
  PMID:19071119.
  That is a coactivator mechanism; HDAC5 does not bind DNA. Decision: MODIFY to
  GO:0003713 transcription coactivator activity (same as the HDAC4 review), with a
  WRONG_TERM source defect. This is not overruling the experiment: the experiment is
  accepted; only the activity class is corrected.
• GO:0000978 (IDA, PMID:24413532, UniProt). Abstract-only. HDAC5 is recruited to the
  ESR1 locus within an MTA1 complex; there is no evidence of direct, sequence-specific
  DNA binding. Decision: MODIFY to GO:0003712 transcription coregulator activity
  (direction of the effect differs between cell lines in that paper), with a source
  defect.
• GO:0000976 contributes_to (IDA, PMID:19351956). ChIP shows HDAC5 at the FGF2 and
  SLIT2 promoters, where it represses. Note GO:0000976 is_a GO:0043565
  sequence-specific DNA binding (checked with oaklib), so even with contributes_to the
  term implies DNA binding by the complex. Decision: MODIFY to GO:0003714 transcription corepressor activity. This deliberately differs from the HDAC4 review, which proposed
  GO:0000976 as the replacement for HDAC4's GO:0000978 row; that choice may be worth
  revisiting for HDAC4 (not edited here).
• GO:0140297 IPI (PMID:12242305, BHF-UCL) has WITH = mouse CBX5 (HP1alpha, Q61686,
  checked against UniProt REST). HP1 is a chromodomain adaptor, not a DNA-binding
  transcription factor. The claim "HDAC5 binds DNA-binding TFs" is true (MEF2, HIF1A,
  GCM1, GATA3), so the row is accepted as a claim, but a WRONG_TERM source defect is
  recorded for the evidence.

Localization

Nucleus (repressive state) and cytoplasm (after phosphorylation/14-3-3 binding); UniProt
"Shuttles between the nucleus and the cytoplasm." In cultured DRG neurons
PMID:26157139.

Open questions

• Does HDAC5 have a physiological substrate efficiently processed by its His-containing
  active site, as Lahm et al. proposed for class IIa enzymes?
• Is the HIF-1alpha coactivation seen in overexpression reproduced at endogenous levels?