TRMT6 review notes

Reviewed as a pair with TRMT61A (genes/human/TRMT61A/TRMT61A-notes.md), which carries the
full write-up of the mRNA-m1A substrate dispute. This file covers what is specific to TRMT6: it is
the non-catalytic subunit, and its GOA molecular-function coverage is essentially a single
uninformative RNA binding row.

The problem with TRMT6's MF annotations

TRMT6-goa.tsv contains exactly two distinct MF terms:

• GO:0005515 protein binding — IPI ×6 (three of them the TRMT61A interaction, the rest
  high-throughput interactome hits: KAT5, LMO3, PPIA, SETDB1, YWHAG)
• GO:0003723 RNA binding — HDA, from a proteome-wide mRNA interactome capture
  PMID:22658674

Neither says anything about what TRMT6 does. Its actual molecular role has been known at atomic
resolution since 2015 and is simply not represented.

What TRMT6 actually does

The human complex is a dimer of heterodimers, two TRMT6 + two TRMT61A. TRMT6 is a catalytically dead
paralogue of TRMT61A that has been repurposed as the substrate-binding half of the enzyme:

PMID:26470919

Crucially, it does not merely hold the tRNA — it delivers the target base into the partner subunit's
active site, in trans across the dimer interface:

PMID:26470919

and it does so by remodelling the tRNA so that a normally buried base becomes reachable:

PMID:26470919

The complex is obligate: activity requires both chains
PMID:16043508, and the two genes are separately essential in yeast
PMID:16043508.
Site specificity resides in the complex, not in TRMT61A alone
PMID:16043508.

How that is expressed in GO

Three pieces, all of which already exist in the ontology:

1. GO:0000049 tRNA binding (MF) — a direct, structure-supported specialisation of the HDA
   GO:0003723 row. Verified against QuickGO to be a descendant of GO:0003723.
2. GO:0140767 enzyme-substrate adaptor activity (MF) — definition: "An adaptor that brings
   together an enzyme and its substrate. Adaptors recruit the substrate to its enzyme, thus
   contributing to substrate selection and specificity." This is a near-verbatim description of
   the trans-delivery of A58 into the TRMT61A active site. It is the best available MF for what
   TRMT6 itself does, as distinct from what the complex does.
3. contributes_to GO:0160107 tRNA (adenine(58)-N1)-methyltransferase activity — the
   complex-level catalytic activity that TRMT6 enables but does not independently possess. The
   schema's contributes_to_molecular_function slot exists for exactly this case.

Plus GO:0031515 tRNA (m1A) methyltransferase complex in in_complex (already annotated, IBA/IEA/IPI).

So "subunit of the tRNA m1A58 methyltransferase complex" is expressible — the combination above
says it precisely. What is not expressible is the specific mechanism (bind tRNA, splay the loops,
present a buried nucleotide to a partner active site), which is why a narrower child of GO:0140767
is proposed in proposed_new_terms rather than a term for complex membership, which would be
redundant with GO:0031515.

GO:0008047 enzyme activator activity was considered and rejected: TRMT6 is not a regulator that
modulates an otherwise-functional enzyme, it is an obligate structural half of it.

The mRNA m1A dispute (shared with TRMT61A)

Summary only; see TRMT61A-notes.md for the full argument and quotes.

• The mRNA substrate assignment for the complex comes from two 2017 base-resolution studies
  PMID:29072297 and
  PMID:29107537
• It is challenged by PMID:42337368, whose conclusion is
  PMID:42337368
• That paper is a discriminating assay, not a blanket debunk — it confirmed the equally contested
  internal m7G: PMID:42337368.

TRMT6-specific note: GOA gives TRMT6 GO:0006397 mRNA processing (IDA ×2, the same two 2017 papers)
but no GO:0061953 MF row — the MF sits only on TRMT61A. That asymmetry is correct: the
catalytic activity belongs to the catalytic subunit. TRMT6's contribution to any mRNA methylation is
the same adaptor role, and is not separately annotated here.

The cancer literature asserting abundant TRMT6-deposited mRNA m1A
(PMID:42003777;
PMID:40897821) depends on the
mapping methods under challenge, and reports opposite directions of effect on target stability.
PMID:40897821's title is additionally self-contradictory — it calls TRMT6 an "m1A methylase" that
acts "by demethylating" its target
(PMID:40897821) — while its abstract describes deposition, not removal. No
annotation in this review is supported by it.

Curation position taken

• GO:0140767 enzyme-substrate adaptor activity → added as NEW (IDA, PMID:26470919). Not in GOA;
  proposed, not asserted as existing curation. It is the MF used by core_functions.
• GO:0003723 RNA binding (HDA) → MODIFY to GO:0000049 tRNA binding. True but uninformative;
  the replacement is supported by the crystal structure rather than by the poly(A)-capture experiment
  itself, and that is stated in the reason.
• GO:0005515 protein binding (IPI ×6) → MARK_AS_OVER_ANNOTATED, per project guidance. The one
  interaction that matters (TRMT61A) is already stated properly by GO:0031515.
• GO:0031515 tRNA (m1A) methyltransferase complex (IBA/IEA/IPI) → ACCEPT, core.
• GO:0030488 tRNA methylation (IEA) → ACCEPT, core BP.
• GO:0006397 mRNA processing (IDA ×2) → KEEP_AS_NON_CORE, dispute recorded, kept consistent
  with the parallel decision on TRMT61A. Not removed: experimental annotations, full text unseen.
• Nucleus / nucleoplasm → ACCEPT.

Terms verified

All GO ids checked against QuickGO before use: GO:0000049 MF, GO:0140767 MF (definition quoted
above), GO:0160107 MF, GO:0031515 CC, GO:0030488 BP, GO:0006397 BP, GO:0003723 MF,
GO:0005634 CC, GO:0005654 CC, GO:0060090 MF (parent chain of GO:0140767). All cited PMIDs
re-verified against PubMed metadata.