AIGR Gene Hypothesis Deep Research — S. pombe Trm4b (O13935): rRNA/mRNA and Mitochondrial Ribosome Functions

Target: Schizosaccharomyces pombe Trm4b / trm4b (gene directory trm402, SPAC23C4.17), UniProt O13935
Focus type: function_assignment
Hypothesis slug: trm4b-rrna-mrna-and-mitochondrial-ribosome-functions
Iterations completed: 3

Summary

The seed hypothesis proposes that S. pombe Trm4b methylates rRNA or mRNA at cytosine C5, and/or participates in rRNA processing or mitochondrial large-ribosomal-subunit (LSU) assembly — each to be judged independently of its established tRNA C49/C50 methylation, and each requiring positive evidence for a new substrate rather than merely the absence of a negative (tRNA-C34) result. Evaluating each sub-claim against primary literature and public bioinformatics resources, the answer is: Trm4b is strongly supported only as a nuclear tRNA C5 (m5C) methyltransferase; the mRNA-methylation claim is weakly supported by computational ortholog inference (ISS) alone; and the rRNA-methylation, rRNA-processing, and mitochondrial-LSU-assembly claims are refuted as paralog carry-over from NSUN4.

Three independent, positive lines of evidence — not just the absence of a positive assay — argue against the mitochondrial and rRNA claims. First, Trm4b is a 685-aa nuclear protein with no mitochondrial transit peptide, essentially identical in size and architecture to its true 1:1 ortholog, S. cerevisiae Trm4/NCL1 (684 aa, nuclear/nucleolar), whose curated GO set is tRNA-only. Second, the enzyme that actually performs mitochondrial 12S rRNA C911 methylation and coordinates mitoribosome monosome (SSU+LSU) assembly is NSUN4, a short (~381–384 aa) protein bearing a mitochondrial transit peptide (residues 1–25). Third, although Trm4b, NSUN2, and NSUN4 all share PANTHER family PTHR22808, the seed's open question — whether the IBA "with" identifier MGI:1919431 is NSUN4 — is confirmed: MGI:1919431 = Nsun4. The mitochondrial/rRNA signal thus enters Trm4b's record through shared-superfamily phylogenetic inference, not any S. pombe measurement.

The only non-tRNA function with a defensible rationale is mRNA (cytidine-5-)-methyltransferase (GO:0062152), which sits on Trm4b as an ISS projection from human NSUN2 (Q08J23). Because vertebrate NSUN2 does carry experimentally supported mRNA m5C activity, the projection is lineage-plausible — but it skips the yeast ortholog (ScTrm4/NCL1) that lacks mRNA activity entirely, making it a phylogenetically non-conservative, low-confidence inference. Recommended curation: retain the tRNA MF as core; keep GO:0062152 only as low-confidence ISS/non-core; add no rRNA or mitochondrial terms.

Key Findings

F001 — Trm4b's demonstrated function is restricted to tRNA C49/C50 m5C

The curated UniProt entry TRM4B_SCHPO (O13935) lists only tRNA substrates under FUNCTION and CATALYTIC ACTIVITY: 5-methylcytosine formation at C49 and C50 (PMID: 30646830, ECO:0000269, experimental) and at C60/61/62 in tRNA-Asp (PMID: 23074192). The primary tRNA-methylome paper states plainly that "Trm4b methylates C49 and C50, which both lie in the TΨC-stem," and — critically for this hypothesis — only the tRNA methylome was assayed; there was no rRNA, mRNA, or mitochondrial substrate panel. PomBase records the experimental molecular function as GO:0016428 tRNA (cytidine-N5)-methyltransferase (IDA/IMP, PMID:30646830), and the deletion is viable.

This finding anchors the entire evaluation. The single high-quality experimental study behind Trm4b defines its demonstrated substrate as tRNA and its demonstrated sites as C49/C50. Consistent with the seed's framing, it neither tested nor excluded rRNA/mRNA — but it provides zero positive support for the rRNA/mRNA/mitochondrial extensions.

F002 — The mitochondrial-ribosome / rRNA functions trace to NSUN4 (MGI:1919431) via paralog co-membership in PANTHER PTHR22808

PomBase's IBA (phylogenetic inference; GO_REF:0000033) provenance for trm4b's cytoplasm annotation (GO:0005737) uses a with= list of MGI:1919431, MGI:107252, UniProtKB:Q08J23, PANTHER:PTN000516077. Resolving these identifiers programmatically via mygene.info: MGI:1919431 = Nsun4 and MGI:107252 = Nsun2. The seed explicitly asked to verify — rather than assume — that MGI:1919431 is NSUN4; this is now confirmed: MGI:1919431 is Nsun4.

InterPro/PANTHER placement shows that Trm4b (O13935), NSUN2 (human Q08J23 / mouse Q1HFZ0), and NSUN4 (human Q96CB9 / mouse Q9CZ57) all map to PTHR22808, whereas NSUN5 (Q96P11/Q8K4F6) and NOP2/NSUN1 (P46087) map to a separate family, PTHR22807. Because NSUN4 co-resides in PTHR22808, tree-based propagation (IBA) can pull NSUN4-derived signals onto Trm4b even though NSUN4's function is compartmentally and mechanistically distinct. That NSUN4 function is well defined: PMID: 24516400 shows "NSUN4 methylates cytosine 911 in 12S rRNA" and that "the NSUN4/MTERF4 complex is required to assemble the SSU and LSU to form a monosome." The mito-rRNA and mitoribosome-assembly ideas in the seed are therefore NSUN4-specific, entering the record through shared-family inference, not a pombe measurement.

Reassuringly, an audit of Trm4b's actual IBA terms found that mito/rRNA terms were not, in fact, propagated onto Trm4b — only generic cytoplasm and tRNA binding were. NSUN5-type cytoplasmic rRNA m5C sits in a different family (PTHR22807) and cannot legitimately project onto Trm4b.

F003 — Trm4b's mRNA (cytidine-5-)-methyltransferase annotation (GO:0062152) is an ISS ortholog projection from human NSUN2, not a S. pombe measurement

PomBase annotation 87463 assigns GO:0062152 mRNA (cytidine-5-)-methyltransferase with reference GO_REF:0000024 (ISS) and with= UniProtKB:Q08J23 (NSUN2_HUMAN), assigned_by PomBase. Human NSUN2 (Q08J23) carries GO:0062152 by IDA:UniProtKB — i.e., the experimental anchor lives on the human protein and is projected onto Trm4b by sequence similarity (ISS). The review PMID: 42587746 notes that, of eight human m5C RNA methyltransferases, "only two, NSUN2 and NSUN6, were found to be capable of modifying the main portion of cellular mRNAs," which rationalizes why an mRNA projection would come from NSUN2 while also emphasizing that bulk mRNA m5C is comparatively rare and restricted. Trm4b's subcellular location is Nucleus (HDA:PomBase, PMID: 16823372) plus cytoplasm (IBA), with no mitochondrial localization — consistent with a nuclear tRNA/mRNA enzyme, not a mitochondrial one.

F004 — Architecture and localization cleanly separate Trm4b from mitochondrial NSUN4

A direct UniProt feature comparison makes the paralog-confusion mechanism explicit:

Trm4b (685 aa) is essentially the same length as ScTrm4/NCL1 (684 aa), the true 1:1 ortholog, and both are nuclear with no transit peptide. NSUN4, by contrast, is roughly half the length and carries a mitochondrial transit peptide (residues 1–25), localizing to the mitochondrion. Because Trm4b lacks mitochondrial targeting entirely, the mito-LSU-assembly claim cannot be a direct Trm4b function. This is positive architectural evidence against the mitochondrial claim, exactly the kind of evidence the seed asks for — not merely the absence of a positive tRNA-C34 result.

F005 — The true ortholog ScTrm4/NCL1 carries no rRNA/mRNA/mito functions, weakening the mRNA-ISS projection

A programmatic scan of the full GO set for P38205 (S. cerevisiae Trm4/NCL1) — the 1:1 experimentally characterized ortholog of Trm4b — returns only tRNA-centric terms: MF GO:0016428 tRNA (cytidine-N5)-MTase (IDA:SGD) and GO:0000049 tRNA binding (IDA:SGD); BP GO:0030488 tRNA methylation, GO:0002127 tRNA wobble cytosine methylation, GO:0006400 tRNA modification, GO:0070301 response to hydrogen peroxide; CC nucleus/nucleolus/nucleoplasm/cytoplasm. An explicit search for rRNA / mRNA / mitochondrion / ribosome terms returned NONE. The mRNA (cytidine-5-)-methyltransferase activity appears only on vertebrate NSUN2 (Q08J23, IDA) — not on the yeast ortholog. This is the strongest single argument that the mRNA-ISS annotation on Trm4b is a phylogenetically non-conservative, lineage-boundary projection: the closest experimentally characterized fungal relative shows tRNA specificity and no mRNA activity, implying mRNA m5C is a metazoan NSUN2-lineage innovation.

Mechanistic Model / Interpretation

The evidence resolves into a clean model in which one PANTHER superfamily (PTHR22808) contains functionally divergent m5C RNA methyltransferases, and annotation-propagation machinery has partially blurred those boundaries onto Trm4b.

      PANTHER PTHR22808  (m5C RNA methyltransferases)
┌───────────────────────────────┬────────────────────────────────┐
│                               │                                │
   Trm4/NSUN2 clade                 NSUN4 clade                     (vertebrate
   (nuclear, tRNA/mRNA)            (mito, 12S rRNA)                  expansion)
│                               │
  ┌─────┴─────┐                    ┌────┴─────┐
  │           │                    │          │
ScTrm4      Trm4b (O13935)      Human/Mouse NSUN4
(P38205)    685 aa, NUCLEAR     ~381–384 aa, MITO transit pep
684 aa      no transit pep      → 12S rRNA C911 m5C
tRNA-ONLY   tRNA C49/C50 m5C    → MTERF4 complex → mitoribosome
    (PMID:30646830)       SSU+LSU monosome assembly
│                   (PMID:24516400)
│
   Human NSUN2 (Q08J23): tRNA + mRNA m5C (IDA)
│  ISS projection (GO_REF:0000024, with=Q08J23)
▼
   Trm4b GO:0062152 mRNA (cytidine-5-)-MTase  ← weak, non-core

   Separate family PTHR22807: NSUN5 (large-subunit rRNA), NOP2/NSUN1 (rRNA)
   → NOT the source of the trm4b annotations

Direct, demonstrated Trm4b activity: SAM-dependent 5-methylcytosine deposition on tRNA at C49/C50 (and C60–62 in tRNA-Asp), acting as a nuclear tRNA (cytidine-N5)-methyltransferase (class I Rossmann-fold MTase, NOL1/NOP2/SUN domain). This is the primary molecular function.

Inferred, non-core: mRNA C5 methylation (GO:0062152) — plausible at the lineage level because human NSUN2 has this activity, but supported only by ISS from a human paralog and absent from the true fungal ortholog. Best treated as low-confidence inference.

Not supported (paralog carry-over): rRNA methylation, rRNA processing, and mitochondrial LSU assembly. These are NSUN4 (mito 12S rRNA) and, in a different family, NSUN5/NOP2 functions. They reach Trm4b only through shared-superfamily phylogenetic inference (IBA with=MGI:1919431=Nsun4) and are contradicted by Trm4b's nuclear, non-mito architecture.

The key discriminator is subcellular targeting plus ortholog identity. A mitochondrial LSU-assembly role requires a mitochondrial protein; Trm4b has no transit peptide and is nuclear. A cytoplasmic rRNA-processing role requires a nucleolar rRNA MTase (NSUN5/NOP2, family PTHR22807 — not Trm4b's family). Trm4b fits neither.

Evidence Base

GO Curation Implications

Leads requiring curator verification:

1. GO:0016428 tRNA (cytidine-N5)-methyltransferase activity (MF) — RETAIN (core). Experimentally supported (IDA/IMP, PMID:30646830). Ensure the annotation captures C49/C50 site specificity.
2. tRNA methylation BP terms — RETAIN, but re-examine GO:0002127 (tRNA wobble base C34 cytosine methylation) on Trm4b: PMID:30646830 assigns C34 methylation to the paralog Trm4a, so this BP term may belong to Trm4a, not Trm4b.
3. GO:0062152 mRNA (cytidine-5-)-methyltransferase activity (MF) — RETAIN WITH CAVEAT / mark non-core. Keep the ISS evidence code and with= (Q08J23); do not upgrade to an experimental code. Add a curator note that S. pombe splits Trm4 into Trm4a/Trm4b and that the true fungal ortholog ScTrm4/NCL1 lacks mRNA activity, making this a phylogenetically non-conservative projection.
4. rRNA methylation / rRNA processing (e.g., GO:0031167, GO:0006364) — DO NOT ADD. No positive evidence; canonical rRNA m5C (NSUN5/NOP2) is in a different PANTHER family (PTHR22807).
5. Mitochondrial LSU assembly / mito rRNA methylation / mitochondrial translation (e.g., GO:0000466, GO:0032543) — DO NOT ADD. These are NSUN4-specific; Trm4b is nuclear/cytoplasmic with no transit peptide. Any such term appearing via IBA is paralog-conflation and should be removed or NOT-qualified.
6. CC: nucleus (HDA, PMID:16823372) and cytoplasm (IBA) are appropriate; no mitochondrion term is supported.

"Protein binding" is not needed as a fallback — the informative, well-supported MF is the tRNA C5-methyltransferase activity.

Mechanistic Scope

• Direct molecular activity: SAM-dependent cytosine-C5 methyltransferase acting on tRNA at C49/C50 (and C60–62 in tRNA-Asp); nuclear.
• Inferred (ortholog) activity: mRNA C5 methylation, by analogy to human NSUN2 — a molecular activity (not a downstream phenotype), but unverified in S. pombe and absent from the yeast ortholog.
• Not the gene product's function (downstream/paralog): mitochondrial 12S rRNA methylation and mitoribosome (SSU/LSU) assembly are NSUN4 activities; large-subunit cytoplasmic rRNA m5C is an NSUN5 activity. Neither belongs to Trm4b. No S. pombe loss-of-function phenotype implicates Trm4b in rRNA maturation or mitoribosome assembly (the deletion is viable), so the rRNA/mito claims lack even loss-of-function support.

Conflicts and Alternatives

• Paralog confusion (primary alternative): NSUN4 (MGI:1919431) co-resides with Trm4b in PTHR22808, so propagation of NSUN4's mito-rRNA/LSU-assembly terms would be a family-level (IBA) artifact, not a real Trm4b function. Confirmed: MGI:1919431 is Nsun4.
• Family boundary protects against rRNA over-annotation: the canonical rRNA-m5C enzymes NSUN5 and NOP2/NSUN1 are in PTHR22807, distinct from Trm4b's PTHR22808 — rRNA-processing terms cannot legitimately reach Trm4b by orthology.
• Sub-functionalization: S. pombe uniquely splits Trm4/NSUN2 into Trm4a and Trm4b (PMID:30646830). A human-NSUN2-based ISS (mRNA activity) may map to Trm4a, Trm4b, both, or neither — a real organism-specific caveat for the mRNA term.
• Node discrepancy: the seed cites PANTHER node PTN000516076; the actual PomBase with-refs use PTN000516077. Both are ancestral nodes within PTHR22808; not decision-changing, but worth verifying against the current PANTHER tree.
• In-vitro / cross-species extrapolation: even NSUN2's mRNA activity is a mammalian in-vivo finding; extrapolating substrate breadth across ~1 billion years of divergence to fission yeast is speculative.

No evidence was found that supports the seed's rRNA/mito extensions over the paralog-confusion alternative.

Limitations and Knowledge Gaps

Discriminating Tests

1. RNA bisulfite sequencing of trm4bΔ vs WT S. pombe on rRNA and (tRNA-depleted) mRNA — directly tests rRNA/mRNA m5C dependence on Trm4b. Most decisive single experiment.
2. Mitochondrial fractionation + GFP imaging to confirm Trm4b is excluded from mitochondria (predicted: excluded), formally closing the mito-LSU claim.
3. In vitro methylation of rRNA/mRNA substrates by recombinant Trm4b (predicted: little/no activity vs strong tRNA activity).
4. Ortholog complementation: test whether human NSUN2 (mRNA-competent) but not NSUN4 rescues trm4bΔ, and whether Trm4b rescues ScTrm4Δ.
5. Polysome/mitoribosome profiling in trm4bΔ (predicted: no mitoribosome-assembly defect).
6. Comparative PANTHER-tree audit of PTHR22808 to confirm mito/rRNA terms are anchored below the Trm4b/NSUN2 ancestor (predicted: yes — hence no legitimate IBA propagation).

Proposed Follow-up Actions / Curation Leads

Leads requiring curator verification:

• Reference to verify (core tRNA function): PMID: 30646830 — snippet: "Trm4b methylates C49 and C50, which both lie in the TΨC-stem." Supports the retained tRNA MF and shows only tRNA was assayed.
• Reference to verify (mRNA ISS basis): UniProtKB:Q08J23 (human NSUN2), which carries GO:0062152 by IDA — supports keeping GO:0062152 as ISS, not experimental.
• Reference to verify (paralog distinction): PMID: 24516400 — snippets: "NSUN4 methylates cytosine 911 in 12S rRNA" and "the NSUN4/MTERF4 complex is required to assemble the SSU and LSU to form a monosome." Grounds the mito-LSU-assembly function in NSUN4, not Trm4b.
• Reference to verify (mRNA competence is lineage-restricted): PMID: 42587746 — snippet: "only two, NSUN2 and NSUN6, were found to be capable of modifying the main portion of cellular mRNAs."

Candidate action changes:
- Retain tRNA MF/BP; retain mRNA-ISS only with a low-confidence/non-core caveat; do not add any rRNA-processing, mito-rRNA-methylation, or mitochondrial-LSU-assembly term.
- If such terms already appear via IBA, treat as paralog carry-over and remove or NOT-qualify.
- Re-examine GO:0002127 (wobble C34) on Trm4b, since PMID:30646830 assigns C34 methylation to Trm4a.

Suggested curator question: "Is the mRNA-MTase ISS meant to represent Trm4b specifically, given S. pombe's Trm4a/Trm4b split, or is it a genus-level NSUN2 projection that skips the yeast ortholog lacking the activity?"

Suggested experiment: trm4bΔ RNA-BS-seq on rRNA + mRNA (definitive test of the entire hypothesis).

Bottom Line

Trm4b (O13935) is well supported only as a nuclear tRNA C5 (m5C) methyltransferase acting at tRNA C49/C50. The hypothesis's rRNA-methylation, rRNA-processing, and mitochondrial-LSU-assembly extensions have no positive experimental support in any organism and are best explained as paralog carry-over from NSUN4 (confirmed MGI:1919431 = Nsun4) via shared PANTHER family PTHR22808 — contradicted by Trm4b's 685-aa nuclear, transit-peptide-free architecture that matches the tRNA-only ortholog ScTrm4/NCL1. The only non-tRNA annotation with a defensible rationale — mRNA (cytidine-5-)-methyltransferase (GO:0062152) — is a weak ISS projection from human NSUN2 that even the yeast ortholog lacks, and should be retained only as low-confidence, non-core inference; no rRNA or mitochondrial terms should be added.

Artifacts

• OpenScientist final report
• OpenScientist final report