AIGR Gene Hypothesis Deep Research — Final Report

Gene: fogD (FogD) · Organism: Aspergillus ruber CBS 135680 (NCBITaxon:1388766) · UniProt: A0A017SE81 (286 aa)
Focus type: function_assignment · Hypothesis slug: sdr-lipid-functions-and-polyketide-release
Source: genes/ASPRC/fogD/fogD-ai-review.yaml (free-text)

Executive Judgment

Verdict: Over-annotated (for the six lipid GO terms) / Partially supported (for the underlying SDR chemistry).

The seed hypothesis asks a curator to weigh six lipid-metabolism GO terms on FogD —
acylglycerone-phosphate reductase (NADP+) activity (GO:0000140), triacylglycerol lipase
activity (GO:0004806), endoplasmic reticulum localization (GO:0005783), lipid-droplet
localization (GO:0005811), phosphatidic acid biosynthetic participation (GO:0006654), and
triglyceride catabolism (GO:0019433) — against what FogD was actually shown to do.

The clear answer from primary literature and public annotation provenance is that all six
lipid terms are electronic (IEA:TreeGrafter) carry-overs propagated from the experimentally
characterized Saccharomyces cerevisiae protein Ayr1, and none of them is supported by any
FogD-specific assay, localization experiment, or mutant phenotype. FogD's only
experimentally demonstrated role is as a secondary-metabolite short-chain dehydrogenase/
reductase (SDR) in the flavoglaucin biosynthetic cluster, where it temporarily reduces a
polyketide salicylaldehyde to a salicyl alcohol to permit downstream decoration
(PMID:32134669(https://pubmed.ncbi.nlm.nih.gov/32134669/)).

The important nuance — which the seed hypothesis correctly insists on — is that this is not a
case of "wrong graft" or "weak source." The PANTHER family assignment (PTHR44169, the Ayr1
family) is genuine, FogD retains an intact SDR catalytic tetrad, and SDR membership does not
by itself exclude hydrolase side-activity. But plausible-by-homology is not
demonstrated-for-FogD. At ~39% identity to Ayr1 (twilight zone), conserved fold and cofactor
chemistry transfer, whereas Ayr1's specific substrate identities and dual moonlighting
functions do not. The six lipid terms should therefore be treated as non-core electronic
predictions rather than as FogD functions, and the review should foreground the
secondary-metabolite reductase role.

Key Findings

Finding 1 — FogD's experimentally supported function is a secondary-metabolite SDR (polyketide salicylaldehyde reduction), not lipid metabolism

UniProt A0A017SE81 records the recommended name "Short-chain dehydrogenase fogD", EC
1.1.1.-, and the alternate designation "Flavoglaucin biosynthesis cluster protein D," with
the FUNCTION and PATHWAY comments framing the protein within secondary-metabolite biosynthesis
(ECO:0000269 / ECO:0000305 sourced to
PMID:32134669(https://pubmed.ncbi.nlm.nih.gov/32134669/)). The DISRUPTION PHENOTYPE is
equally telling: deletion of fogD impairs flavoglaucin and congener production and yields
very low accumulation of (8E,10E,12E)-3,5,7-trihydroxytetradeca-8,10,12-trienoic acid — a
polyketide intermediate, not a glyceride or phospholipid.

The mechanistic paper is unambiguous about the chemistry. In the flavoglaucin pathway, the
polyketide skeleton is "released as alkylated salicyl alcohols, which is a prerequisite for
consecutive hydroxylation and prenylation, before reoxidation to the final aldehyde products"
(PMID:32134669(https://pubmed.ncbi.nlm.nih.gov/32134669/)). This describes a reductive
tailoring step on a polyketide — a CH-OH-forming NADP(H)-dependent reduction — which is the
generic SDR chemistry FogD shares with its family, applied to a secondary-metabolite substrate.
This is distinct from the reductive thioester release mechanism sometimes seen in polyketide
synthase R-domains; the evidence here supports a discrete oxidoreductase forming an alcohol,
consistent with the general MF term GO:0016616 (oxidoreductase acting on CH-OH group of
donors, NAD/NADP acceptor).

Sequence inspection confirms the enzyme is a bona fide catalytically competent SDR: the 286-aa
protein carries the N-terminal Rossmann cofactor motif TGCGQGGIG (residues ~8–16), the
catalytic serine (Ser145/146), the SDR signature YxxxK motif (YNATK, Tyr147/Lys151),
and the conserved Asn within an NCAG context (~res 82). The catalytic machinery is intact,
so the enzyme is a genuine active reductase — the question is only on which substrate, and
the direct evidence points to a polyketide, not acyl-DHAP or a triglyceride.

Finding 2 — The six lipid-metabolism GO terms are IEA:TreeGrafter carry-overs seeded by S. cerevisiae Ayr1

Every one of the six lipid terms on A0A017SE81 carries the evidence code IEA:TreeGrafter.
FogD maps to PANTHER family PTHR44169 "NADPH-DEPENDENT 1-ACYLDIHYDROXYACETONE PHOSPHATE
REDUCTASE" — the Ayr1 family — and to subfamily PTHR44169:SF3 "SHORT-CHAIN DEHYDROGENASE
SRDE." The set of six terms mirrors, almost exactly, the experimentally characterized dual
biology of yeast Ayr1:

• acyl-DHAP reductase (GO:0000140),
• novel TAG lipase (GO:0004806) — "identification of Ayr1p as a novel triacylglycerol
  lipase of yeast lipid droplets" (PMID:24187129(https://pubmed.ncbi.nlm.nih.gov/24187129/)),
• ER + lipid-droplet dual localization (GO:0005783, GO:0005811),
• and ether/phospholipid + TAG metabolism (GO:0006654, GO:0019433).

No FogD-specific assay, localization experiment, or mutant supports any of these six terms.
They are a coherent bundle precisely because they were transferred as a bundle from one
well-studied source protein. This is the signature of homology-based propagation, not of
independent characterization.

Finding 3 — Provenance chain verified: the six lipid terms are TreeGrafter-propagated (node PTN001211783) from experimentally annotated Ayr1

Querying the QuickGO annotation API for A0A017SE81 shows all six terms with evidence
IEA / ECO:0007826, reference GO_REF:0000118, assignedBy = TreeGrafter, and
withFrom = PANTHER:PTN001211783. The corresponding QuickGO record for S. cerevisiae Ayr1
(P40471) shows each of these terms is both experimentally supported and IBA-propagated
(GO_Central, GO_REF:0000033):

This confirms the seed hypothesis's core provenance concern: the FogD annotations trace back to
genuinely experimental Ayr1 annotations, propagated electronically. The historical
TreeGrafter provenance node PTN001211783 is the recorded withFrom source on FogD; it is
consistent with — but at the resolution of the public annotation index cannot be independently
disambiguated from — the seed's proposed ancestral IBD nodes PTN001213826 (ER) and
PTN001963740 (the other five claims). The ancestral IBD node annotations themselves are not
exposed in the public GOlr annotation index and could not be retrieved to complete a
node-by-node trace. This is a genuine, honestly reported provenance gap: the propagation is
real and Ayr1-seeded, but the exact internal graft-node lineage is not fully verifiable from
public endpoints.

Finding 4 — FogD shares only ~39% identity with Ayr1: conserved SDR chemistry, twilight-zone divergence

A global Needleman–Wunsch alignment of FogD (A0A017SE81, 286 aa) against S. cerevisiae Ayr1
(P40471, 297 aa) yields 111 identical residues over 281 aligned non-gap columns = 39.5%
identity (38.8% over FogD length). Both proteins retain the N-terminal Rossmann NAD(P)-binding
motif (FogD VTGCGQGGIG; Ayr1 VVTGASGG) and the SDR catalytic YxxxK motif (FogD
Tyr147/Lys151; Ayr1 Tyr157/Lys161). Crucially, FogD is placed in the distinct subfamily
PTHR44169:SF3 (SRDE), separate from the Ayr1 subfamily.

~39% identity sits squarely in the twilight zone of sequence homology. At this divergence,
the fold and the cofactor/catalytic chemistry are reliably conserved — so calling FogD an SDR
oxidoreductase is safe — but substrate identity and specialized moonlighting functions are
not reliably transferable. Ayr1's dual acyl-DHAP-reductase / TAG-lipase behavior and its
specific glyceride substrates are precisely the kind of fine specificity that diverges across
40%-identity paralogs, especially when the two proteins occupy different PANTHER subfamilies and
different biological contexts (primary lipid metabolism in yeast vs. secondary-metabolite
tailoring in a filamentous fungus).

Mechanistic Model / Interpretation

The two competing readings of FogD can be laid out directly:

          PTHR44169  (Ayr1 family, NADPH-dependent SDR fold)
                  |
+-------------------------+---------------------------+
|                                                     |
  S. cerevisiae Ayr1 (P40471)                        A. ruber FogD (A0A017SE81)
  subfamily = Ayr1                                   subfamily = PTHR44169:SF3 (SRDE)
  297 aa                                             286 aa
|                                                     |
  EXPERIMENTAL (multiple PMIDs):                     EXPERIMENTAL (PMID:32134669):
   - acyl-DHAP reductase  GO:0000140                  - flavoglaucin cluster SDR
   - TAG lipase           GO:0004806  <== moonlight    - reduces polyketide salicyl-
   - ER                   GO:0005783                     ALDEHYDE -> salicyl ALCOHOL
   - lipid droplet        GO:0005811                   - deletion impairs flavoglaucin
   - PA biosynthesis      GO:0006654                   - EC 1.1.1.-, intact SDR tetrad
   - TAG catabolism       GO:0019433                     (TGCGQGGIG / Ser / YNATK / NCAG)
|                                                     ^
|            ~39% identity (twilight zone)            |
+-----> IEA:TreeGrafter propagation (withFrom -------+
PANTHER:PTN001211783, GO_REF:0000118)
copies all six lipid terms to FogD

What transfers across 39% identity: the SDR three-dimensional fold, the Rossmann
NAD(P)-binding site, the catalytic Ser–Tyr–Lys triad, and therefore the generic capacity for
NADP(H)-dependent CH-OH ↔ C=O oxidoreduction (GO:0016616). This is safe.

What does NOT transfer: the specific substrates (acyl-DHAP, triacylglycerol,
phosphatidic-acid pathway intermediates), the hydrolase moonlighting activity, and the
organelle localization (ER + lipid droplet) that were measured for Ayr1. These are
context-specific properties of a yeast primary-metabolic enzyme, not portable predictions for a
fungal secondary-metabolite tailoring enzyme.

Reconciling the two views: the seed hypothesis is right that SDR membership does not exclude
hydrolysis, and right that a secondary-metabolite role or a missing signal peptide does not by
itself prove the absence of peripheral organelle association. But the burden of evidence runs
the other way: there is positive experimental evidence for the polyketide-reductase role
(mutant phenotype + pathway reconstitution logic in PMID:32134669) and zero FogD-specific
evidence for any of the six lipid activities/locations. The parsimonious model is that FogD is a
dedicated secondary-metabolite SDR whose lipid annotations are homology artifacts.

The distinction the seed asks us to draw — reductive thioester release vs. CH-OH donor
oxidoreduction — resolves toward the latter as the safe MF: the evidence supports FogD forming
an alcohol (salicyl alcohol) by NADP(H)-dependent reduction, i.e. GO:0016616-type chemistry,
rather than establishing a chain-releasing R-domain thioester reductase mechanism.

Evidence Base / Evidence Matrix

How the literature bears on the findings. The single deep-read primary paper,
Biosynthesis of the Prenylated Salicylaldehyde Flavoglaucin Requires Temporary Reduction to
Salicyl Alcohol for Decoration before Reoxidation to the Final Product
(PMID:32134669(https://pubmed.ncbi.nlm.nih.gov/32134669/)), directly establishes FogD's real
role and is the anchor for Findings 1 and the refutation of the lipid terms as core.
PMID:24187129(https://pubmed.ncbi.nlm.nih.gov/24187129/) and
PMID:10617610(https://pubmed.ncbi.nlm.nih.gov/10617610/) are Ayr1 (yeast) studies that
explain the origin of the propagated terms rather than supporting them for FogD — they are the
"competing/source-of-graft" evidence.

GO Curation Implications

Lead action (requires curator verification): treat the six lipid terms as non-core
electronic predictions, not FogD functions. Concretely:

Positively supported terms (leads to ADD as core):

• MF: GO:0016616 — oxidoreductase activity, acting on the CH-OH group of donors, NAD or
  NADP as acceptor (safe, evidence-backed generalization of the demonstrated reduction). EC
  1.1.1.- from UniProt is consistent.
• BP: GO:0044550 — secondary metabolite biosynthetic process (and, if a curator accepts
  cluster-level annotation, a flavoglaucin/polyketide-specific child). Evidence: mutant
  phenotype + pathway role in PMID:32134669(https://pubmed.ncbi.nlm.nih.gov/32134669/).

The recommendation deliberately avoids "protein binding" and instead anchors on the
evidence-backed oxidoreductase MF and secondary-metabolite BP.

Mechanistic Scope

The immediate molecular function under test is a specific NADP(H)-dependent oxidoreduction.
The direct evidence (PMID:32134669) supports FogD catalyzing the reduction of a flavoglaucin
polyketide salicylaldehyde to a salicyl alcohol — an on-pathway, reversible tailoring step
that primes the intermediate for hydroxylation and prenylation before final reoxidation. This
is CH-OH/NADP oxidoreductase chemistry (GO:0016616), performed on a secondary-metabolite
substrate.

Separated from this direct activity are the downstream / inferred properties that the seed
hypothesis correctly flags:
- Lipid substrates and hydrolase activity (acyl-DHAP, TAG) are inferred only from Ayr1 homology.
- ER and lipid-droplet localization are inferred from Ayr1, with no FogD imaging.
- Phosphatidic-acid biosynthesis and triglyceride catabolism are pathway memberships copied
from Ayr1's biology, not FogD phenotypes.

None of these downstream/inferred properties has been observed for FogD itself. The
loss-of-function evidence that does exist (deletion impairs flavoglaucin) points exclusively
to the secondary-metabolite role.

Conflicts and Alternatives

1. Paralog carry-over (primary conflict). The six lipid terms are database carry-over from
   Ayr1 via TreeGrafter. This is the single most important alternative explanation and it is
   well supported: identical bundle of terms, identical withFrom source, IEA evidence only.
2. Organism/context difference. Ayr1 is a yeast primary-metabolic enzyme (glycerolipid
   metabolism); FogD is a filamentous-fungus secondary-metabolite tailoring enzyme. Different
   biological compartments and substrate pools make wholesale function transfer unlikely.
3. Subfamily divergence. FogD sits in PTHR44169:SF3 (SRDE), a distinct subfamily from
   Ayr1, reinforcing that specialized function has diverged even though the family is shared.
4. The seed's own caution is valid and retained. SDR fold does not exclude hydrolase
   side-activity, and absence of a signal peptide does not exclude peripheral membrane/organelle
   association. So the six terms cannot be declared impossible — only unsupported for FogD.
   The honest position is "not demonstrated," not "disproven."
5. Mechanism ambiguity (thioester release vs alcohol formation). The data favor discrete
   CH-OH-forming reduction (GO:0016616) over a chain-releasing thioester-reductase mechanism,
   but the exact substrate-tethering state at the moment of reduction is not fully resolved.

Limitations and Knowledge Gaps

Discriminating Tests

1. In vitro substrate panel with purified recombinant FogD — assay NADPH-dependent
   reduction of (a) the flavoglaucin polyketide salicylaldehyde intermediate, (b) acyl-DHAP,
   and (c) a triacylglycerol hydrolysis readout. This directly separates the demonstrated
   secondary-metabolite reductase activity from the Ayr1-derived lipid activities.
2. Subcellular localization of FogD in A. ruber (fluorescent fusion + organelle markers)
   to test ER/lipid-droplet claims directly rather than by homology.
3. Structural comparison — an AlphaFold model or crystal structure of FogD superposed on
   Ayr1, examining the substrate-binding pocket and any residues implicated in Ayr1's lipase
   moonlighting, to predict whether hydrolase capability plausibly survived divergence.
4. Complementation swap — test whether FogD rescues a yeast ayr1Δ lipid phenotype, and
   whether Ayr1 rescues a fogD flavoglaucin defect, to measure functional interchangeability.
5. PANTHER node-lineage retrieval — obtain the internal IBD graft nodes (PTN001213826,
   PTN001963740) to complete the provenance chain the seed requested.

Proposed Follow-up Experiments / Actions (Curation Leads — require curator verification)

• Reframe the review so the core function is the secondary-metabolite SDR: propose core
  MF GO:0016616 (CH-OH/NADP oxidoreductase) and core BP GO:0044550 (secondary metabolite
  biosynthetic process), each citing PMID:32134669(https://pubmed.ncbi.nlm.nih.gov/32134669/).
• Downgrade the six lipid terms (GO:0000140, GO:0004806, GO:0005783, GO:0005811, GO:0006654,
  GO:0019433) from any core status to non-core IEA:TreeGrafter predictions, with a note that they
  are Ayr1-seeded homology transfers unsupported by FogD-specific evidence.
• Candidate reference + snippet to verify:
  PMID:32134669(https://pubmed.ncbi.nlm.nih.gov/32134669/) — "The polyketide skeleton was
  released as alkylated salicyl alcohols, which is a prerequisite for consecutive hydroxylation
  and prenylation, before reoxidation to the final aldehyde products." Use as the primary
  support for the reductase MF and secondary-metabolite BP.
• Provenance note to add: QuickGO A0A017SE81 — all six lipid terms IEA:TreeGrafter,
  GO_REF:0000118, withFrom=PANTHER:PTN001211783; source Ayr1 (P40471) carries these terms
  experimentally (e.g. TAG lipase PMID:24187129(https://pubmed.ncbi.nlm.nih.gov/24187129/)).
• Suggested curator questions: (i) Should the six lipid terms be retained as IEA or removed
  given the strong competing experimental role? (ii) Is a flavoglaucin/polyketide-specific BP
  child preferred over generic GO:0044550? (iii) Can the ancestral IBD nodes be retrieved to
  finalize provenance?
• Suggested experiments: the in vitro substrate panel and localization assay above.

Conclusion

FogD is genuinely a member of the Ayr1 SDR family and a catalytically intact NADP(H)-dependent
oxidoreductase, but its only experimentally supported role is temporary reduction of a
flavoglaucin polyketide salicylaldehyde to salicyl alcohol — a secondary-metabolite tailoring
step. The six lipid-metabolism GO terms in the seed hypothesis are IEA:TreeGrafter
over-annotations propagated from experimentally characterized yeast Ayr1 and are not
supported by any FogD-specific evidence. At ~39% identity (twilight zone) and in a distinct
PANTHER subfamily, conserved SDR chemistry transfers but Ayr1's specific lipid substrates,
hydrolase moonlighting, and organelle localization do not. The recommended curation lead is to
foreground GO:0016616 (MF) and GO:0044550 (BP) as core, and to treat all six lipid terms
as non-core electronic predictions rather than FogD functions — while acknowledging that they
are unsupported rather than formally disproven.

Artifacts

• OpenScientist final report
• OpenScientist final report