AIGR Gene Hypothesis Deep Research — Final Report

Target gene: IRE1 (UniProt G0RBE3), Hypocrea jecorina / Trichoderma reesei strain QM6a (NCBITaxon:431241)
Focus type: function_assignment
Hypothesis slug: fungal-stress-death-and-signaling-complex
Source file: genes/HYPJE/IRE1/IRE1-ai-review.yaml

Summary

This report evaluates a two-part function-assignment hypothesis for Trichoderma reesei / Hypocrea jecorina IRE1 (UniProt G0RBE3, strain QM6a): (1) that the protein contributes to intrinsic apoptotic signaling after ER stress (GO:0070059), and (2) that it forms an IRE1-TRAF2-ASK1 complex (GO:1990604). Both propositions were tested independently against direct target evidence, fungal-family evidence, homolog identity, and mechanistic context, exactly as the seed requested. Both fail. The verdict is REFUTED / over-annotated by computational carry-over, with the important nuance that this is a PAINT/TreeGrafter release-and-source inconsistency (over-propagation from a mammalian subfamily), not a demonstrated experimental loss of an ancestral fungal function.

The IRE1-TRAF2-ASK1 complex (GO:1990604) is definitionally impossible to assemble in T. reesei because two of its three obligate subunits — TRAF2 and ASK1 (MAP3K5) — have no orthologs in the QM6a proteome, and ASK1 is absent kingdom-wide across Fungi (0 hits across taxid 4751). The single MATH/TRAF-domain protein in the proteome (G0RV09) is a ubiquitin carboxyl-terminal hydrolase, not a TRAF signaling adaptor. Every experimental demonstration of the complex in the literature is metazoan (human and rat cell systems). The intrinsic ER-stress apoptotic signaling role (GO:0070059) has never been demonstrated for the fungal protein: the sole primary characterization (PMID:15480788(https://pubmed.ncbi.nlm.nih.gov/15480788/)) establishes an adaptive UPR function only — kinase autophosphorylation, complementation of yeast ire1 mutants, HAC1 mRNA splicing, and chaperone induction — with no assay of IRE1-dependent death.

The two flagged annotations are attached experimentally (IDA) to the mammalian PANTHER subfamily PTHR13954:SF17 (human ERN1/IRE1α, O75460), whereas the fungal target G0RBE3 and the seed's own tree leaf (Sclerotinia sclerotiorum A7EHN1) belong to subfamily PTHR13954:SF6. The fungal proteins therefore carry the terms only as IEA:TreeGrafter family-level propagation across a subfamily boundary. The recommended curation action is de-propagation of both apoptosis terms while retaining the evidence-backed core: Ser/Thr kinase + endoribonuclease activity driving IRE1-mediated UPR (GO:0036498) at the ER membrane.

Key Findings

F001 — GO:1990604 (IRE1-TRAF2-ASK1 complex) cannot assemble: two of three obligate subunits are absent from T. reesei

The GO term GO:1990604 is defined (QuickGO) as a protein complex composed of IRE1 plus TRAF2 plus ASK1 — a strict three-subunit definition, so the absence of any single obligate partner is sufficient to refute complex formation. A UniProt REST query of the Trichoderma reesei QM6a proteome (taxid 431241) returned 0 proteins matching TRAF2 / "TNF receptor-associated factor" and 0 proteins matching ASK1 / "apoptosis signal-regulating kinase" (MAP3K5). The only MATH-domain (IPR002083) protein in the proteome, G0RV09, is annotated as a ubiquitin carboxyl-terminal hydrolase — it carries the MATH fold but is not a TRAF signaling adaptor and lacks the RING/zinc-finger architecture required for TRAF2 E3/adaptor function.

Every published assay that directly demonstrates the IRE1-TRAF2-ASK1 complex is metazoan: PMID:23000344(https://pubmed.ncbi.nlm.nih.gov/23000344/) (human bladder cancer T24 cells), PMID:31931206(https://pubmed.ncbi.nlm.nih.gov/31931206/) (human SH-SY5Y neuronal cells), PMID:34427546(https://pubmed.ncbi.nlm.nih.gov/34427546/) (rat), and PMID:33692842(https://pubmed.ncbi.nlm.nih.gov/33692842/) (human gastric cancer AGS/MKN-45). As the verbatim snippet from PMID:23000344 states, "Ursolic acid induces IRE1-TRAF2-ASK1 signaling complex formation to activate pro-apoptotic ASK1-JNK signaling," and PMID:31931206 shows that "knockdown of IRE1 with the LV-IRE1 shRNA suppressed the expression of IRE1, TRAF2, p-ASK1, and p-JNK" — both strictly in metazoan cell lines. There is no fungal counterpart.

F002 — G0RBE3 is experimentally an adaptive UPR sensor, not a demonstrated apoptosis effector

The single primary characterization of the target protein, PMID:15480788(https://pubmed.ncbi.nlm.nih.gov/15480788/) ("The ire1 and ptc2 genes involved in the unfolded protein response pathway in the filamentous fungus Trichoderma reesei"), documents a canonical adaptive UPR sensor and nothing more. It shows: (i) functional complementation of Saccharomyces cerevisiae ire1 mutants; (ii) intrinsic kinase activity — "The T. reesei IREI protein has intrinsic kinase activity, as revealed by an in vitro autophosphorylation assay"; (iii) up-regulation of UPR targets bip1/pdi1 upon ire1 overexpression; and (iv) HAC1 mRNA splicing — "Splicing of the mRNA encoding the transcription factor HAC1 is also observed."

The UniProt curated record for G0RBE3 assigns FUNCTION = ER unfolded-protein sensing + endoribonuclease splicing of precursor mRNAs to induce UPR targets, and SUBCELLULAR LOCATION = ER membrane. The domain architecture (CDD Luminal_IRE1 cd09769, RNase_Ire1 cd10422, kinase PF00069, KEN/PUG) is the canonical UPR sensor. Critically, it contains no death domain and no dedicated TRAF-recruitment adaptor motif. No assay of IRE1-dependent apoptotic death has ever been reported in T. reesei. The demonstrated outputs (HAC1 splicing, chaperone induction) are the hallmark of the pro-survival, adaptive arm of the UPR, not the terminal/apoptotic arm.

F003 — Fungal programmed cell death runs through metacaspases/mitochondrial factors, not the metazoan IRE1-TRAF2-ASK1-JNK axis

Even granting that fungi can undergo regulated cell death, the executioner machinery differs from the mammalian model. Reviews and primary studies of yeast/fungal PCD attribute apoptosis to the metacaspases Yca1/Mca1, Nma111, the DNase Nuc1, and the AIF Ndi1, acting through mitochondrial permeability transition (PMID:37616576(https://pubmed.ncbi.nlm.nih.gov/37616576/); PMID:22770501(https://pubmed.ncbi.nlm.nih.gov/22770501/)). As the direct snippet from PMID:22770501 states, "Apoptosis is dependent on the Yca1 metacaspase, since loss of YCA1 abrogates cell death induced by oxidized Trx3." Because TRAF2 and ASK1 — the obligate partners of the mammalian IRE1 apoptotic complex — have no orthologs in T. reesei (0 hits each), the metazoan IRE1→TRAF2→ASK1→JNK death route has no structural basis in this organism. This is a mechanistic, not merely correlative, argument against transferring the mammalian apoptotic role.

F004 — Both flagged terms derive from mammalian primary experiments; ASK1 is absent kingdom-wide in Fungi

The defining primary sources for the IRE1 death module are mammalian. PMID:10650002(https://pubmed.ncbi.nlm.nih.gov/10650002/) (Urano et al., 2000, Science) established the IRE1-TRAF2 coupling: "The cytoplasmic part of IRE1 bound TRAF2, an adaptor protein that couples plasma membrane receptors to JNK activation," and showed IRE1α-/- fibroblasts are impaired in ER-stress JNK activation. PMID:12050113(https://pubmed.ncbi.nlm.nih.gov/12050113/) (Nishitoh et al., 2002, Genes Dev) established the complex and its death output: "ER stress activates ASK 1 through formation of an IRE1-TRAF2-ASK1 complex," driving neuronal death from expanded polyglutamine repeats.

Proteome-scale queries confirm the fungal machinery gap. T. reesei QM6a encodes only 1 MAP3K-named kinase (G0RDL7, not ASK1) and 1 MATH-domain protein (G0RV09, a ubiquitin hydrolase, not TRAF2). A query for "apoptosis signal-regulating kinase" across the entire Fungi kingdom (taxid 4751) returned 0 hits — ASK1/MAP3K5 is a metazoan innovation. Notably, the seed's tree leaf Sclerotinia sclerotiorum A7EHN1 is itself an IRE1 ortholog (InterPro IPR045133; PANTHER PTHR13954:SF6, the same subfamily as G0RBE3), not a TRAF2 or ASK1 protein — confirming the tree neighborhood is composed of fungal UPR sensors, not apoptotic adaptors. This directly answers the seed's question "do target fungal proteins genuinely correspond to TRAF2 and ASK1?" — No.

F005 — The experimental complex annotation sits on the mammalian PTHR13954:SF17 subfamily, not the fungal SF6 subfamily of G0RBE3

Human ERN1/IRE1α (UniProt O75460) carries GO:1990604 (IRE1-TRAF2-ASK1 complex) with IDA (direct experimental) evidence and GO:0070059 by ISS, and belongs to PANTHER PTHR13954:SF17. The fungal IRE1 target G0RBE3 and the seed's tree leaf A7EHN1 both belong to PTHR13954:SF6. Thus the experimentally-supported complex annotation resides on a distinct mammalian subfamily branch, and the fungal proteins carry the terms only as IEA:TreeGrafter family-level (not subfamily-consistent) propagation. This is exactly the release/source inconsistency the seed flags: the current PAINT IBDs for GO:0070059 and GO:1990604 are placed at a mammalian internal node (PTN000359344 per the seed) that is not an ancestor of the fungal SF6 leaf.

Programmatic caveat: the PANTHER family.do UI returned HTTP 403; the OAI familymsa service returned HTTP 200 (2 MB MSA) but full PAINT/IBD internal-node PTN annotations were not machine-parsed in this run. The subfamily assignments above (SF6 vs SF17) are from UniProt/InterPro/PANTHER records rather than a re-derived tree. The exact PTN node IDs (PTN001017826 vs PTN000359344) remain the seed's assertion to verify against the served tree.

Mechanistic Model / Interpretation

The core issue is a mismatch between what the fungal protein does and what the mammalian-derived GO terms require.

   MAMMALIAN IRE1α (human ERN1, O75460; PTHR13954:SF17)
   ----------------------------------------------------
   ER stress ──> IRE1α oligomerization
    │
┌───────────┴───────────┐
   (adaptive arm)          (terminal/apoptotic arm)
   XBP1 mRNA splicing      cytoplasmic tail binds TRAF2
   -> UPR gene induction        │
            recruits ASK1 (MAP3K5)
            = IRE1-TRAF2-ASK1 complex  <-- GO:1990604
                 │
            ASK1 -> JNK -> apoptosis    <-- GO:0070059


   FUNGAL IRE1 (T. reesei G0RBE3; PTHR13954:SF6)
   ---------------------------------------------
   ER stress ──> IRE1 oligomerization + autophosphorylation (kinase)
    │
     HAC1 mRNA splicing (RNase_Ire1 cd10422)
    │
     Hac1 TF -> bip1 / pdi1 chaperone induction   <-- GO:0036498 (UPR)
    │
    X   NO TRAF2 in proteome (0 hits)
    X   NO ASK1 in proteome / in ALL Fungi (0 hits, taxid 4751)
    X   NO death domain / TRAF adaptor motif

   Fungal PCD (if any) executes via a SEPARATE route:
oxidative/ER stress -> mitochondrial permeability transition
            -> metacaspase Yca1/Mca1, Nma111, Nuc1, Ndi1

The adaptive arm is conserved between the mammalian and fungal proteins — both perform kinase autophosphorylation and site-specific mRNA splicing (XBP1 in mammals, HAC1 in fungi) to drive UPR gene induction. The terminal apoptotic arm is a metazoan elaboration that depends on partners (TRAF2, ASK1) that fungi never acquired.

The correct reading is conditional/adaptive UPR conservation with loss (or, more accurately, non-acquisition) of the metazoan death-signaling output — best framed as over-propagation across a family node rather than a demonstrated experimental refutation of a fungal death phenotype.

Evidence Base (Evidence Matrix)

How the literature groups. (1) The single fungal-target paper (PMID:15480788) supports only the adaptive UPR core and is silent on death. (2) The mammalian mechanistic papers (PMID:10650002, 12050113, 23000344, 31931206, 33692842, 34427546) establish the IRE1-TRAF2-ASK1 apoptotic module exclusively in metazoan cells and are the true origin of the flagged GO terms. (3) The fungal/comparative PCD papers (PMID:37616576, 22770501) show that fungal regulated cell death executes via metacaspase/mitochondrial factors, providing a positive alternative mechanism that does not require IRE1 nucleation.

GO Curation Implications

Leads requiring curator verification (confirm against the live GO/PANTHER release):

Net: the informative retained core is MF kinase + endoribonuclease and BP IRE1-mediated UPR at the ER membrane. This satisfies the objective's requirement to avoid "protein binding" as a terminal recommendation — the retained terms are specific catalytic and process terms. The two apoptosis terms should be de-propagated, ideally with an annotation-exception flag to block re-propagation, framed as a TreeGrafter over-propagation rather than a demonstrated biological loss-of-function.

Mechanistic Scope

The immediate molecular function under test is whether G0RBE3 directly (a) participates in a stable IRE1-TRAF2-ASK1 protein complex and (b) initiates an intrinsic apoptotic signaling cascade in response to ER stress.

• Direct gene-product activity (established): ER luminal misfolded-protein sensing → oligomerization/auto-activation; trans-autophosphorylation (Ser/Thr kinase); site-specific endoribonucleolytic cleavage of HAC1 mRNA. These are intrinsic, adaptive, homeostatic outputs at the ER membrane.
• Complex membership (GO:1990604) — not direct, and physically impossible: requires partners (TRAF2, ASK1) that are not encoded in the genome. This is not a downstream-phenotype question; it is a hard structural impossibility.
• Apoptotic signaling (GO:0070059) — not established: any conditional fungal death under severe ER stress would, on current evidence, be executed by metacaspase/mitochondrial machinery, not by an IRE1-nucleated TRAF2/ASK1 module. The seed correctly notes that adaptive UPR function and ER localization do not exclude conditional death signaling — but the burden of evidence for the specific GO terms is not met, and the required partners are absent.

Conflicts and Alternatives

1. Database carry-over (primary alternative explanation). Both flagged terms are IEA:TreeGrafter. The experimental IRE1-TRAF2-ASK1 evidence belongs to human ERN1 (O75460, PTHR13954:SF17); the fungal target is PTHR13954:SF6. Family-level propagation crossed a subfamily boundary onto a branch whose ancestor lacks the IBD placement — the best explanation is database carry-over, not biology.
2. Paralog/component confusion. The only MATH-domain protein in T. reesei is a ubiquitin hydrolase (G0RV09), not a TRAF adaptor; no MAP3K5/ASK1 ortholog exists. Fungal MAP3Ks (Ssk2/Ste11/Bck1) are not ASK1 orthologs and are not ER-stress apoptosis kinases.
3. Organism-specific machinery differences. TRAF2 and ASK1 are metazoan innovations; ASK1 returns 0 hits kingdom-wide in Fungi. The mammalian death module has no fungal structural equivalent.
4. Distinct fungal death route. If T. reesei undergoes regulated cell death, the conserved eukaryotic route is metacaspase + mitochondrial (Yca1/Nma111/Nuc1/Ndi1; PMID:37616576, PMID:22770501), which does not require IRE1 as a nucleator.
5. Not a demonstrated loss-of-function. There is no evidence the fungal lineage ever possessed an IRE1-TRAF2-ASK1 death module, so this is best framed as absence/never-present in fungi + mispropagation, not "loss of an ancestral function."

Limitations and Knowledge Gaps

1. Proteome searches were name/domain-based. The 0-hit results for TRAF2/ASK1 rely on UniProt annotation text and InterPro/MATH-domain matching, not exhaustive HMM/structural homology profiling. A remote possibility of a highly diverged TRAF-like adaptor missed by keyword search remains. Resolution: run HMMER/Foldseek profile searches with TRAF2 (MATH+RING) and ASK1 (MAP3K catalytic + thioredoxin-binding region) profiles against the QM6a proteome and predicted structures.
2. PANTHER tree not fully re-derived programmatically. family.do returned HTTP 403; the OAI familymsa service returned HTTP 200 but PAINT/IBD PTN internal-node annotations were not machine-parsed. Subfamily assignments (SF6 vs SF17) are from UniProt/InterPro/PANTHER records. The exact PTN node IDs (PTN001017826 vs PTN000359344) remain the seed's assertion. Resolution: obtain the served PTHR13954 v19 tree and PAINT IBD placements directly to confirm the node is not ancestral to the SF6 fungal leaf.
3. Negative for apoptosis = untested, not disproven. No T. reesei study tested and failed to find IRE1-dependent death. The refutation is definitional (missing partners) plus absence of positive evidence, not a target-specific negative assay. This distinction matters for how the curation note is worded.
4. HYPJE-specific PCD literature is thin. Fungal PCD evidence is largely from S. cerevisiae and plants; direct Trichoderma PCD-mechanism data are sparse.

Discriminating Tests

Proposed Follow-up Experiments / Actions (Curation Leads)

Candidate action changes (require curator verification)
- Remove / de-propagate GO:1990604 (IRE1-TRAF2-ASK1 complex, CC) from G0RBE3; flag as annotation exception to block re-propagation.
- Remove or down-grade to flagged non-core GO:0070059 (intrinsic apoptotic signaling, ER stress, BP) on G0RBE3.
- Retain and emphasize GO:0036498 (IRE1-mediated UPR), Ser/Thr kinase + endoribonuclease MF terms, and ER membrane CC.

Candidate references with exact snippets to verify
- PMID:15480788 — "The T. reesei IREI protein has intrinsic kinase activity, as revealed by an in vitro autophosphorylation assay." and "Splicing of the mRNA encoding the transcription factor HAC1 is also observed." → basis for retained UPR core.
- PMID:10650002 — "The cytoplasmic part of IRE1 bound TRAF2..." → establishes mammalian (not fungal) origin of the TRAF2 link.
- PMID:12050113 — "ER stress activates ASK 1 through formation of an IRE1-TRAF2-ASK1 complex" → defines GO:1990604 in a mammalian system.
- PMID:22770501 — "Apoptosis is dependent on the Yca1 metacaspase..." → fungal death route is metacaspase-based.

Suggested curator questions
- Is the served PTHR13954 v19 IBD for GO:0070059/GO:1990604 placed at a node ancestral to the fungal SF6 leaf, or only to mammalian SF17? (Expected: only mammalian.)
- Should the review record this as a TreeGrafter over-propagation (source inconsistency) rather than a biological loss-of-function claim? (Recommended framing: yes.)

Suggested experiments — see Discriminating Tests table: Δire1 ER-stress survival/apoptosis assay in T. reesei; proteome HMMER/Foldseek sweep for TRAF2/ASK1; served-tree/PAINT re-derivation.

Provenance

• UniProt REST fetch of G0RBE3 (domains, GO codes, curated FUNCTION/CC) — executed this run.
• UniProt proteome queries for TRAF2 / ASK1 / MATH-domain proteins in taxid 431241 — executed this run (0 / 0 / 1 hits; sole MATH hit G0RV09 = Ub hydrolase).
• UniProt kingdom-wide ASK1 query (taxid 4751 → 0 hits) and T. reesei MAP3K enumeration (1 hit, G0RDL7) — executed this run.
• A7EHN1 identity check (fungal IRE1, PTHR13954:SF6, IPR045133) and human ERN1 O75460 GO/subfamily check (GO:1990604 IDA; PTHR13954:SF17) — executed this run.
• PANTHER programmatic access test: family.do HTTP 403; OAI familymsa HTTP 200 (PAINT node-level not parsed) — executed this run.
• QuickGO definitions for GO:0070059, GO:1990604, GO:0036498 — executed this run.
• PubMed abstracts reviewed: PMID 15480788, 10650002, 12050113, 23000344, 31931206, 33692842, 34427546, 22770501, 37616576.

Prepared as a hypothesis-level curation lead. All directional GO recommendations require curator verification against the live GO/PANTHER release. Computational proteome/subfamily results are reported conservatively; the PANTHER tree was not fully re-parsed programmatically in this run (family.do HTTP 403).

Artifacts

• OpenScientist final report
• OpenScientist final report