AIGR Gene Hypothesis Deep Research — Final Report

Target: Drosophila melanogaster TyrRS (UniProt Q9VV60, FBgn0027080)

Hypothesis: "Q9VV60 binds resveratrol" (GO:1905594 — resveratrol binding)

Focus type: function_assignment

Summary

The proposed function assignment "F:resveratrol binding" (GO:1905594) for Drosophila TyrRS should not be applied to the gene review, and if present it should be removed. Three independent lines of evidence converge on this conclusion. First, the GO term itself is obsolete — GO:1905594 "resveratrol binding" was obsoleted on 2026-07-19 with the explicit rationale that "binding to resveratrol is not an evolved molecular function", has no replacement or "consider" term, and carries zero annotations across all organisms. Second, no organism carries this annotation, including the human ortholog YARS1, which is the only protein where a TyrRS–resveratrol interaction has actually been measured; GO curators translated that human experiment into the general term GO:0036094 "small molecule binding," not a resveratrol-specific term. Third, resveratrol is a plant stilbene phytoalexin that is not endogenous to Drosophila, so any binding is a xenobiotic/pharmacological event rather than a physiological molecular function.

The most important caveat is that binding is genuinely plausible on structural grounds. The fly tyrosine-binding pocket is fully conserved with the human pocket where resveratrol was co-crystallized: fly Q9VV60 shares 68.4% identity with human YARS1, and all nine reported human resveratrol-contact residues are identical in the fly protein. But plausibility-by-homology is not measured binding — it has never been tested in Drosophila — and homology does not rescue an obsolete, unannotatable GO term.

The gene's real, experimentally supported molecular function is tyrosine–tRNA ligation (GO:0004831), backed by a fly IDA experiment (PMID: 19561293). Verdict: REFUTED as a GO annotation; only weakly/partially supported as an in-vitro biochemical possibility that has no bearing on annotation.

Key Findings

Finding 1 — The fly tyrosine-binding pocket is fully conserved with the human pocket where resveratrol binds

The supplied sequence was verified independently: 525 residues, SHA-256 94682a9b…a01f — an exact match to the frozen cohort sequence. UniProt Q9VV60 is annotated as "Tyrosine–tRNA ligase, GN=TyrRS," and the protein carries the diagnostic class I aminoacyl-tRNA synthetase Rossmann-fold signatures: a HIGH-like motif region YWGTATTGKPH (around position 38) and a KMSKS-family motif KMSSS (around position 221).

A global Needleman–Wunsch alignment (BLOSUM62) of fly Q9VV60 against human YARS1 (P54577) gives 357/522 = 68.4% identity, with a near-identical catalytic core. Most decisively, all nine reported resveratrol-contact residues from the human co-crystal structure (PDB 4Q93; Y39, L72, A74, H77, V152, Y166, Q170, D173, I191) are identical in the fly protein at the corresponding positions (fly Y38, L71, A73, H76, V151, Y165, Q169, D172, I190), with conserved local sequence context around each.

This is the strongest evidence for the seed hypothesis. In human TyrRS the resveratrol pocket is the tyrosine active site, and that pocket is transferable to the fly enzyme by strict residue conservation. The direct human study (PMID: 25533949) states: "Here we present a 2.1 Å co-crystal structure of resveratrol bound to the active site of TyrRS." Because resveratrol is a tyrosine mimic occupying that pocket, and the pocket is conserved, fly TyrRS could plausibly bind resveratrol in vitro. However, this remains an inference by homology — not a measured binding constant, not an assay in fly protein, and not evidence of physiological relevance. Under the standards set by the decisive question, pocket transferability is a hypothesis, not experimental validation.

Finding 2 — GO:1905594 "resveratrol binding" is OBSOLETE and unannotatable

The QuickGO/EBI ontology record for GO:1905594 shows isObsolete=True, name="obsolete resveratrol binding", aspect=molecular_function, and the comment: "The reason for obsoletion is that binding to resveratrol is not an evolved molecular function." The term has replacedBy=None and consider=None. The ontology history records the obsoletion action on 2026-07-19 (the is_a binding parentage was deleted). A QuickGO annotation search returns 0 annotations to GO:1905594 across all organisms.

This is decisive for the curation question: an obsolete GO term with no replacement or "consider" alternative cannot be validly assigned to any gene product. The obsoletion rationale itself encodes the ontology curators' biological judgment that binding a plant-derived xenobiotic is not an evolved function of any protein. Resveratrol is a plant stilbene phytoalexin (PMID: 38903666: "These compounds function as phytoalexins, aiding plant defense against phytopathogens"; corroborated by PMID: 40644839), not a Drosophila metabolite.

Finding 3 — Neither the fly protein nor the directly assayed human ortholog carries a resveratrol-binding annotation

Current QuickGO molecular-function annotations for fly Q9VV60 comprise nine terms — GO:0000049 (tRNA binding), GO:0000166 (nucleotide binding), GO:0003723 (RNA binding), GO:0004812 (aminoacyl-tRNA ligase), GO:0004831 (tyrosine–tRNA ligase; including IDA PMID: 19561293 [Storkebaum PNAS 2009, fly CMT model] and TAS PMID: 26761199 [fly aaRS review]), and GO:0005524 (ATP binding) — and no resveratrol / GO:1905594 annotation.

Human YARS1 (P54577) carries 33 MF annotations and also has no GO:1905594. Importantly, the resveratrol study PMID: 25533949 supports GO:0004831 (IDA), GO:0005515 (IPI — the PARP1 interaction), and GO:0036094 "small molecule binding" (IDA) on the human protein. In other words, when GO curators encountered the single direct resveratrol experiment, they chose the general small-molecule-binding term, not a resveratrol-specific one. The study's own framing is pharmacological: "Resveratrol nullifies the catalytic activity and redirects TyrRS to a nuclear function."

Finding 4 — Q9VV60 is a single-isoform canonical cytoplasmic tyrosyl-tRNA synthetase with no resveratrol reference

The UniProt Q9VV60 record shows sequence length 525 (matching the frozen SHA-256), no alternative products / isoforms listed, proteinExistence="1: Evidence at protein level", catalytic activity tRNA(Tyr) + L-tyrosine + ATP = L-tyrosyl-tRNA(Tyr) + AMP + diphosphate + H(+) (Rhea), and subcellular location Cytoplasm (ARBA). There is no mention of resveratrol anywhere in the FUNCTION, CATALYTIC, or LOCATION comments. This rules out isoform confusion and confirms the analyzed sequence is the canonical, only protein product of the locus.

Mechanistic Model / Interpretation

The layered logic distinguishing "can bind" from "should be annotated" is summarized below:

    Q9VV60 (fly TyrRS, 525 aa, cytoplasmic, PE1)
                     |
     Class I aaRS Rossmann fold; tyrosine pocket
     (HIGH-like YWGTATTGKPH @38, KMSSS @221)
                     |
68.4% identity to human YARS1; ALL 9 resveratrol-contact
residues IDENTICAL (Y38/L71/A73/H76/V151/Y165/Q169/D172/I190)
                     |
+----------------------------+----------------------------+
|                                                         |
  BIOCHEMICAL PLAUSIBILITY                          CURATION / ONTOLOGY REALITY
  (resveratrol = tyrosine mimic,                    (GO:1905594 OBSOLETE 2026-07-19,
   occupies conserved Tyr pocket)                    "not an evolved MF", 0 annotations)
|                                                         |
  Plausible in vitro binding                        Term unannotatable; even human
  BUT never measured in fly;                        resveratrol data curated as
  xenobiotic, not endogenous ligand                 GO:0036094 small molecule binding
+----------------------------+----------------------------+
                     |
   GENUINE FUNCTION = tyrosine-tRNA ligation
   (GO:0004831, fly IDA PMID:19561293)

The key conceptual separation the decisive question demanded is between (a) whether the protein can physically recognize the compound (plausible), (b) whether that recognition is a physiological molecular function (no — resveratrol is a plant defense metabolite absent from flies), and (c) whether it merits the specific GO term (no — the term is obsolete and unannotatable). All three must hold for a valid function assignment, and only the weakest link (a) is satisfied.

The downstream biology described in the human/rodent literature — resveratrol acting as a tyrosine-free-mimic (cis-RSV) or tyrosine-like-mimic (trans-RSV) to modulate TyrRS levels, nuclear translocation, and PARP1-mediated histone ADP-ribosylation (PMID: 35688816, PMID: 25533949) — is stress signaling downstream of binding and must be separated from the binding event itself. Even where documented, it is human/rat biology, not fly.

Evidence Base / Evidence Matrix

GO Curation Implications

Recommended action (lead requiring curator verification): REMOVE / DO NOT ADD "F:resveratrol binding" (GO:1905594).

• The term is obsolete (2026-07-19) with no replacedBy and no consider term. Assigning an obsolete term is invalid regardless of the underlying biology.
• The obsoletion comment — "binding to resveratrol is not an evolved molecular function" — is itself the ontology's curator-level judgment that this should never be an MF annotation for any gene.
• No downgraded resveratrol-specific replacement should be created. If a curator wished to capture the (human, in vitro) small-molecule interaction, the precedent set on the human ortholog is GO:0036094 "small molecule binding" (IDA) — but that is supported only by the human experiment, and there is no direct fly evidence to transfer it with an experimental code. For the fly protein, no small-molecule-binding annotation is warranted from resveratrol data.
• Retain the core molecular function: GO:0004831 "tyrosine–tRNA ligase activity" (fly IDA PMID: 19561293), along with the standard supporting terms already present (GO:0004812, GO:0005524, GO:0000166, GO:0000049, GO:0003723).

GO decision table:

Mechanistic Scope

The immediate molecular event under test is direct physical recognition of the small molecule resveratrol by the TyrRS protein. The evidence establishes that in human TyrRS this occurs in the tyrosine active-site pocket, with resveratrol acting as a tyrosine mimic that competitively occupies the substrate site and blocks aminoacylation. This binding is the proximal event.

Everything else described in the literature is downstream: loss of catalytic (aminoacylation) activity, nuclear translocation of TyrRS, activation of PARP1, histone serine-ADP-ribosylation, DNA-repair modulation, and neuroprotection versus neurodegeneration (isomer-dependent). These are pathway consequences and phenotypic outcomes measured in human cells and rat neurons, not molecular functions of the fly protein and not the binding event itself. For curation of the fly gene, only the proximal binding claim is in scope — and it has never been measured in Drosophila.

Conflicts and Alternatives

1. Organism transfer gap (primary conflict). The only direct experiment is on human YARS1, not fly Q9VV60. Assigning a fly annotation from a human in vitro experiment would require an ISS/ISO-style justified transfer — but the target term is obsolete, so no transfer is possible regardless.
2. Xenobiotic vs. physiological function. Resveratrol is a plant phytoalexin (PMID: 38903666, PMID: 40644839); flies do not synthesize it and it is not a natural ligand. A druggable pocket that happens to accept a xenobiotic is not an evolved molecular function — exactly the GO obsoletion rationale.
3. Isomer dependence. PMID: 35688816 shows cis- and trans-resveratrol bind the same pocket but drive opposite conformational/functional outcomes. A bare "resveratrol binding" annotation would obscure this and misrepresent the biology.
4. Database carry-over risk. The original "F:resveratrol binding" prediction most likely arose from automated similarity transfer of the human structural result to the fly ortholog — precisely the frequency-bias / paralog-overannotation failure mode the objective warns against. Neither UniProt nor GO curators applied this term to any organism.

No evidence was found that refutes physical binding capability; the conflicts are all about relevance and annotatability, not about whether the pocket could accept the ligand.

Limitations and Knowledge Gaps

• No direct fly binding measurement. No ITC, SPR, thermal shift, or co-crystal exists for fly TyrRS + resveratrol. What was checked: PubMed literature, UniProt, and GO annotations — none report a fly assay. This matters because "missing target experiment ≠ evidence of absence"; binding remains plausible but unmeasured. Resolution: a direct biophysical binding assay on recombinant Q9VV60.
• Docking not performed as a substitute. Per the decisive question, docking would only generate hypotheses, not validation; it was intentionally not used to claim binding. Structural inference here is limited to residue-level conservation mapping.
• PDB 4Q93 contact residues taken from the human structural literature, not re-derived from an independent structure determination in this work; the residue-numbering transfer to fly relies on the pairwise alignment.
• GO obsoletion date (2026-07-19) reflects the ontology state at analysis time; curators should confirm the current QuickGO/OLS status at review time, though obsoletion is not expected to reverse.
• Functional consequence in flies unknown. Whether resveratrol modulates TyrRS catalysis or any nuclear/stress role in Drosophila has not been tested; the human PARP1 axis may not be conserved.

Discriminating Tests

1. Direct biophysical binding assay on recombinant fly Q9VV60 (isothermal titration calorimetry or surface plasmon resonance) with cis- and trans-resveratrol separately, using L-tyrosine as a positive competitor control and a non-binding stilbene analog as negative control. This directly tests the seed hypothesis in the target organism.
2. Thermal shift / DSF of Q9VV60 ± resveratrol vs. ± tyrosine to confirm active-site engagement.
3. Aminoacylation inhibition assay: does resveratrol inhibit fly TyrRS tyrosylation activity (as it does human)? A competitive Ki would confirm active-site occupancy.
4. Co-crystallography or cryo-EM of fly TyrRS + resveratrol to verify the contact residues predicted identical to human.
5. Ontology/annotation audit: confirm GO:1905594 obsolete status and the absence of any organism annotation in the current release before finalizing the review.

Proposed Follow-up Experiments / Actions (Curation Leads)

All leads require curator verification.

• Action change: Remove/reject "F:resveratrol binding" (GO:1905594) from the TyrRS review; do not substitute a resveratrol-specific term. Reason: term obsolete 2026-07-19, no replacement, zero annotations, xenobiotic interaction, no fly data.
• Retain: GO:0004831 "tyrosine–tRNA ligase activity" as the experimentally supported core molecular function — fly IDA, PMID: 19561293.
• Candidate reference to verify (human precedent — do not transfer to fly): PMID: 25533949, snippet: "Here we present a 2.1 Å co-crystal structure of resveratrol bound to the active site of TyrRS." Note this was curated on human YARS1 as GO:0036094 small molecule binding, not resveratrol binding.
• Supporting rationale reference: PMID: 38903666, snippet: "These compounds function as phytoalexins, aiding plant defense against phytopathogens" — supports the "not an evolved animal MF" judgment.
• Isomer caveat reference: PMID: 35688816 — cis- vs. trans-resveratrol produce opposite TyrRS conformational effects; a generic binding term would be misleading.
• Suggested curator question: Was the original "F:resveratrol binding" prediction an automated similarity transfer from the human structure? If so, flag as ortholog over-annotation.
• Suggested experiment (for the record): direct ITC/SPR binding assay on recombinant fly Q9VV60 with cis/trans-resveratrol and tyrosine competition controls.

Conclusion

The seed hypothesis is biochemically plausible but curationally refuted. Fly TyrRS almost certainly can accommodate resveratrol in its conserved tyrosine pocket (68.4% identity to human, all nine contact residues identical), but this has never been measured in Drosophila, represents a xenobiotic pharmacological interaction rather than an evolved function, and the specific GO term (GO:1905594) is obsolete, replacement-less, and unannotated in every organism — including the human protein where resveratrol binding was actually demonstrated. The genuine, experimentally supported molecular function of Q9VV60 is tyrosine–tRNA ligation (GO:0004831). The "F:resveratrol binding" assignment should be removed and not replaced.

Artifacts

• OpenScientist final report
• OpenScientist final report