ABCD4 (O14678) — Adjudication of Fatty-Acid & Peroxisome Function Claims

Focus: function_assignment · hypothesis fatty-acid-functions-and-peroxisome-organization
Source: genes/human/ABCD4/ABCD4-ai-review.yaml (free-text)
Date: 2026-09-21 · Iteration 1

Executive Judgment

Verdict: Over-annotated / Refuted for all four claims (as direct ABCD4 functions).

The seed hypothesis proposes that human ABCD4 retains (i) fatty acid transmembrane
transporter activity, (ii) fatty acid β-oxidation participation, (iii) very-long-chain
fatty acid (VLCFA) biosynthesis participation, or (iv) peroxisome organization.
Adjudicating each claim separately against its actual GO definition, the evidence
codes on ABCD4, direct substrate/localization experiments, and the ABCD-subfamily
phylogeny:

• None of the four claims has any experimental support on ABCD4. In EBI QuickGO
  (O14678, 55 annotations), every one of these terms is annotated only by IBA
  (Inferred from Biological Ancestor, ECO:0000318, GO_Central, GO_REF:0000033) — i.e.
  phylogenetically propagated from the ABCD1–3 ancestor, not from an ABCD4 experiment.
• Claim 3 is factually mis-stated: the biosynthetic term GO:0042761 is not
  annotated to ABCD4 at all; only the catabolic term GO:0042760 (also IBA-only) is
  present. ABC-D transporters import substrate for β-oxidation (catabolism), so a
  biosynthesis assignment is mechanistically backwards regardless.
• Claim 4 (peroxisome organization) is contradicted by direct localization data:
  ABCD4 localizes to the ER and lysosome, not peroxisomes, because it lacks the
  N-terminal peroxisomal-targeting region conserved in ABCD1–3 (PMID:19010322,
  27456980, 27766264).
• The experimentally grounded core function is lysosomal cobalamin (vitamin B12)
  export (PMID:22922874, 33845046, 31467407, 42303638), with MF term
  GO:0015420 supported by EXP/IDA/IMP.

The demonstrated cobalamin role does not, by itself, prove ABCD4 cannot carry a
fatty-acyl substrate — but the burden of a positive claim is unmet: there is no direct
ABCD4 fatty-acid transport/β-oxidation assay, no peroxisomal residence, and the
biosynthesis term is absent. Recommended action: treat all four claims as non-core;
do not assert them as ABCD4 functions.

Evidence Matrix

Provenance artifacts: /tmp/abcd4_claim_evidence_matrix.csv (QuickGO evidence codes) and the
computed pairwise Needleman-Wunsch %identity table below (ABCD1–4).

Computed ABCD subfamily divergence (NW global alignment, match+1/mismatch−1/gap−1):

ABCD4 is the divergent out-group of the subfamily and the shortest member (606 aa vs
659–745 aa), consistent with the missing N-terminal peroxisomal-targeting region — the
branch across which the IBA fatty-acid/peroxisome terms were propagated is exactly the
weak, long branch where phylogenetic inference is least reliable.

GO Curation Implications (leads — require curator verification)

Not recommending "protein binding" (GO:0005515) as a substantive call — a specific,
better-supported set of cobalamin terms is available.

Mechanistic Scope

• Immediate molecular function tested: ATP-driven transmembrane transport of a
  small-molecule substrate. ABCD4's demonstrated substrate is cobalamin, exported
  from the lysosomal lumen to the cytosol (handoff toward MMACHC), not a fatty
  acyl-CoA.
• Direct activity vs. downstream: The fatty-acid β-oxidation and VLCFA claims are
  pathway-level processes that in ABCD1–3 follow acyl-CoA import into peroxisomes.
  For ABCD4 there is neither the transport step (no assay) nor the compartment
  (not peroxisomal), so any such role would be inference, not direct activity.
• Peroxisome organization is an organelle-biogenesis process; ABCD4 is not a
  peroxisomal resident, removing the mechanistic basis for a direct role.

Conflicts and Alternatives

• Paralog over-annotation (primary explanation): ABCD4 shares the ABC-D
  half-transporter architecture with ABCD1 (X-ALD, VLCFA-CoA), ABCD2, and ABCD3
  (branched acyl-CoA). GO's IBA pipeline propagates the ancestral fatty-acid/peroxisome
  functions to ABCD4. This is exactly the "frequency bias / paralog carry-over" the
  seed warns about.
• Historical naming: ABCD4 = "P70R / PMP69," named as a peroxisomal-membrane-protein
  relative, seeding legacy peroxisome CC annotations (PMID:9302272, 14533738) later
  overturned (PMID:19010322, 22922874).
• Seed's own term error: the VLCFA claim names the biosynthetic process
  (GO:0042761), which is not annotated; only catabolic (GO:0042760) exists — reinforcing
  that the claim is loose paralog carry-over rather than a curated ABCD4 assignment.
• Human loss-of-function phenotype is cobalamin-specific: cblJ patients (ABCD4
  biallelic mutations) present with combined methylmalonic acidemia + homocystinuria
  (PMID:22922874, 23141461, 33729671) — the cobalamin signature — with no reported
  VLCFA accumulation or peroxisomal-disease phenotype (contrast X-linked ALD from
  ABCD1). This is indirect negative evidence against a physiologically important
  fatty-acid role, though not a systematic VLCFA measurement.
• Sequence divergence (computed): ABCD4 shares only ~34% identity with ABCD1–3
  (vs 49% among the peroxisomal trio) and lacks their N-terminal targeting extension,
  so IBA transfer of ABCD1–3 attributes across this long branch is unreliable.
• Not fully excluded: no experiment has directly tested whether ABCD4 can bind/move
  a fatty-acyl species in vitro. Absence of evidence is not proof of substrate
  exclusivity — but it is insufficient to assert the function.

Knowledge Gaps

1. Direct substrate specificity of ABCD4. Checked: no reconstitution/transport assay
   with fatty-acyl-CoA substrates found. Matters because it would positively confirm or
   exclude a residual lipid-transport activity. Resolve with proteoliposome transport
   assays comparing cobalamin vs. (V)LCFA-CoA.
2. Whether IBA annotations should be experimentally overridden. Checked: GO_Central
   IBA present for all four. Matters for curation policy: IBA can be retained but should
   not outrank direct experimental CC/MF. A NOT-qualifier or removal request is the lead.
3. Fate of legacy peroxisome CC annotations. Checked: old IDA still present in
   QuickGO. Matters because CC drives downstream inference. Resolve by curator review
   citing PMID:19010322/22922874/27456980.

Discriminating Tests

• In vitro transport / ATPase stimulation of purified ABCD4 (±LMBD1) with cobalamin
  vs. C22:0/C24:0-CoA and branched acyl-CoA — the decisive substrate-specificity test.
• Comparative localization (endogenous, tagged CRISPR knock-in) vs. peroxisomal
  (PMP70/ABCD3) and lysosomal (LAMP1) markers — confirms non-peroxisomal residence.
• VLCFA / plasma C26:0 profiling in cblJ (ABCD4-deficient) patient fibroblasts vs.
  X-ALD (ABCD1) cells — if ABCD4 contributed to VLCFA handling, C26:0 should rise; the
  prediction under this report is no VLCFA accumulation in cblJ.
• Phylogenetic branch-length / ancestral-state check confirming the fatty-acid GO
  terms trace to the ABCD1–3 clade rather than a demonstrated ABCD4 attribute.

Curation Leads (require curator verification)

• Action changes: Mark GO:0005324, GO:0006635, GO:0042760, GO:0007031 (and any
  GO:0015245 if present) as non-core / candidate removal or NOT on ABCD4, noting they
  are IBA-only paralog propagation. Do not add GO:0042761 (VLCFA biosynthesis).
• Retain as core: GO:0015420 (MF), GO:0015889 + GO:0009235 (BP), GO:0005765 (CC).
• Reconcile CC: deprecate/annotate-superseded the legacy peroxisome CC terms in
  favor of lysosome/ER.
• Candidate references + snippets to verify:
• PMID:19010322 — "P70R-HA was localized to the endoplasmic reticulum (ER), not to peroxisomes."
• PMID:27456980 — "ABCD4 does not localize to peroxisomes but rather, the endoplasmic reticulum (ER), because it lacks the NH2-terminal hydrophilic region required for peroxisomal targeting."
• PMID:27766264 — "ABCD1-3 and ABCD4 are localized to peroxisomes and lysosomes, respectively."
• PMID:22922874 — "ABCD4, a gene that codes for an ABC transporter, which was previously thought to have peroxisomal localization and function."
• Suggested questions for curator: Are the four IBA fatty-acid/peroxisome terms
  worth retaining given definitive lysosomal/cobalamin evidence, or should they be
  removed/NOT-qualified to prevent downstream propagation?
• Suggested experiments: proteoliposome substrate panel; cblJ-fibroblast VLCFA
  profiling (see Discriminating Tests).

Limitations

IBA/evidence-code snapshot reflects QuickGO on 2026-09-21 and may differ from the
review's frozen release. No wet-lab exclusion of fatty-acyl transport by ABCD4 exists;
this report distinguishes absence of positive evidence from proven exclusivity.

Artifacts

• OpenScientist final report
• OpenScientist final report