ygiQ (Q46861) notes

Status of the identification

A first pass of this review (RADICAL_SAM project Phase 2) could not verify that ygiQ is
rlmX: only the abstract of Ishiguro et al. was cached, and the abstract names rlmX
without giving any former gene name. The review was therefore left as a dark gene with
empty core_functions.

The full text of that paper is now cached
(publications/PMID_41380683.md, full_text_available: true) and settles the question
outright: the screen that found the enzyme was a screen of uncharacterised
methyltransferase candidates, the gene found was ygiQ, and the authors rename it in the
same paragraph [PMID:41380683 "screening 15 uncharacterized MTases and 6 known RNA";
"of a plasmid-encoded ygiQ gene, indicating that ygiQ is essential"; "Hereafter, we refer
to ygiQ as rlmX."]. Figure 2A is labelled ΔrlmX (ΔygiQ) and the deletion strain used is
the Keio-collection JW5501 (E. coli ΔygiQ (Knr) (Keio collection)), i.e. the b4469 /
Q46861 locus. The review below is rewritten on that basis; no reviewer-run deep research
was available, and the machine-generated OpenScientist report in this folder was used only
after checking it against the full text (see "OpenScientist report" below).

UniProt Q46861 and NCBI Gene 948963 still carry no rlmX gene name and no function, so the
gene symbol used in this repository stays ygiQ for now.

Identity and architecture

• 739 aa, UPF0313 family (HAMAP MF_01251; NCBIfam TIGR03904 SAM_YgiQ; PANTHER PTHR32331).
  Pfam PF08497 Radical_SAM_N, PF04055 Radical_SAM (core, ~372-650), PF11842 DUF3362;
  C-terminal 685-739 disordered. CxxxCxxC motif CFGGCSFC at 386-393.
  [file:ECOLI/ygiQ/ygiQ-uniprot.txt "Binds 1 [4Fe-4S] cluster. The cluster is coordinated
  with 3 cysteines and an exchangeable S-adenosyl-L-methionine."]
• The paper's own domain assignment matches but adds the cobalamin module that InterPro
  does not recognise: a Cbl-binding domain, the radical SAM (RS) domain and the DUF3362
  domain [PMID:41380683 "as a Cbl-dependent radical SAM enzyme, which is composed"; "of a
  Cbl-binding domain, an RS domain, and a DUF3362"]. Cbl-binding, SAM-binding and
  [4Fe-4S]-binding motifs were located by comparing an AlphaFold2 model of RlmX with
  crystal structures of Cbl-dependent RS MTases PMID:41380683.
• Sequence-similarity-network analysis places RlmX homologues (UPF0313) in a different
  subgroup (Mega 10) from the previously known Cbl-dependent RS MTases OxsB, TsrM, TokK and
  Mmp10 (Mega 1), while sharing their N-terminal-Cbl / C-terminal-cluster domain order.
  Homologues are widespread in bacteria and some archaea.

Reaction and cofactors

• RlmX installs a stereoselective methyl group at carbon 5'(S) of the ribose of two 23S
  rRNA residues in the peptidyl-transferase centre: dihydrouridine 2449 (giving D5Sm2449)
  and cytidine 2498 (giving Cm5Sm2498, after RlmM adds the 2'-O-methyl). This is a
  sugar-phosphate backbone methylation, not a 2'-O-methylation, and the authors
  describe it as unprecedented PMID:41380683.
• Mechanism (proposed, by analogy to other Cbl-dependent RS MTases): the [4Fe-4S] cluster
  reductively cleaves one SAM to the 5'-deoxyadenosyl radical, which abstracts the 5'(S)-H
  from D2449 or Cm2498; the methyl group then comes from methylcobalamin, and Cbl(I) is
  remethylated by a second SAM [PMID:41380683 "cleaves SAM to generate a 5′ -dAdo
  radical, which abstracts"; "a second SAM molecule to regenerate Me-Cbl (III). In our"].
  So two SAM are consumed per methyl transferred, one as radical source and one as methyl
  donor via Cbl.
• One [4Fe-4S] cluster. Iron quantification on reconstituted recombinant protein gave
  3.69 Fe per protein, i.e. a single cluster at ~92% occupancy PMID:41380683. Cluster ligands are C386, C390 and
  C393; each single Cys-to-Ala substitution abolishes activity PMID:41380683.
• Cobalamin binding was observed directly on the reconstituted protein by its 320 nm
  absorbance PMID:41380683, and G160N in the GG motif
  of the Cbl-binding domain abolishes methylation PMID:41380683.
• Both the cluster and Me-Cbl are needed for an active holoenzyme, and Me-Cbl is the methyl
  donor [PMID:41380683 "either the [4Fe-4S] cluster or Me-Cbl was indispensable for"; "of
  an active holoenzyme. Me-Cbl also functions as a methyl"]. Methyl viologen substitutes
  in vitro for the physiological electron donor (flavodoxin / flavodoxin reductase / NADPH
  are suggested).
• SAM dependence was shown directly, including transfer of a CD3 group from CD3-SAM
  PMID:41380683, and E525 contacts the SAM ribose; E525A impairs
  activity PMID:41380683.
• In vivo the modifications require ferric iron and cobalamin in the medium
  PMID:41380683.

The RlmX-specific Cys cluster is NOT a second [4Fe-4S] cluster

C453, C458 and C464 form an RlmX-specific Cys cluster in the RS domain. The authors place
it on the protein surface and attribute it to RNA recognition through zinc binding,
citing earlier work; all three residues are essential for methylation
[PMID:41380683 "resides on the surface of RlmX and is thought to contribute to"; "RNA
recognition through zinc binding.52 Interestingly, all three"]. No zinc was measured here,
and the iron quantification excludes a second [4Fe-4S] cluster. The OpenScientist report's
two-cluster model is therefore wrong, and this review annotates neither a second cluster
nor zinc ion binding.

Substrate and timing

• Holo-RlmX methylates deproteinized 23S rRNA but not the assembled 50S subunit
  PMID:41380683, so the
  modifications are installed early in 50S assembly on naked rRNA.
• In vitro transcribed 23S rRNA is a substrate; domain V (G2018-C2628) is the minimum
  efficient substrate, while shorter H89-93 and H89 transcripts are poor
  [PMID:41380683 "introduced into the domain V transcript (G2018-C2628), whereas"; "suggest
  that domain V is a minimum substrate for RlmX-mediated"]. RlmX thus has genuine
  structured-rRNA recognition, not just a local sequence requirement.
• Order of events: for D5Sm2449 the 5'(S)-methylation (RlmX) and dihydrouridylation (YhiN)
  occur in either order; for Cm5Sm2498 RlmX acts first, because prior 2'-O-methylation by
  RlmM blocks the 5'(S)-methylation.

Physiology

• D5Sm2449, Cm5Sm2498 and ho5C2501 (RlhA) are all elevated under anaerobic growth. rlmX
  mRNA is unchanged between conditions; RlmX protein is up, and the enzyme is additionally
  more active anaerobically, plausibly because the cluster is oxygen-sensitive
  PMID:41380683.
• Single ΔrlmX grows normally; ΔrlmX ΔrlhA is clearly retarded anaerobically and is
  rescued by wild-type rlmX but not by the catalytically dead G160N or C390A
  PMID:41380683.
• Ribosomes from the double knockout are ~half as productive in translation and transfer
  peptide at 64% of the wild-type rate PMID:41380683; 70S initiation and P-site tRNA binding are unaffected. Cryo-EM
  (PDB 8ZFG vs the double-knockout 70S) shows the modifications stabilise the P-site and
  PTC.
• ΔrlmX is less sensitive to erythromycin and clindamycin anaerobically, which the authors
  read as an indirect structural effect, not a resistance mechanism.

Curation decisions

• Catalysis is now established for Q46861, so the RADICAL_SAM generic-row rule applies in
  its normal branch: GO:0003824 MODIFY to the specific activity; GO:0051536 and GO:0005506
  MODIFY to GO:0051539; GO:0051539 ACCEPT and core.
• GO has no molecular-function term for ribose C5' methylation of RNA: every ribose
  methyltransferase term in GO (searched via QuickGO: GO:0008650, GO:0070039, GO:0070677,
  GO:0030743, GO:0090486, GO:0180021 and the tRNA series) is a 2'-O-ribose term, and the
  children of GO:0008649 rRNA methyltransferase activity are all base terms
  (GO:0016433/4/5/6, GO:0070037). A proposed_new_terms entry is used, with GO:0008649 as
  the proposed parent (its definition already admits ribose as the acceptor), and
  core_functions points at it via proposed_molecular_function.
• NEW rows added: GO:1904047 S-adenosyl-L-methionine binding, GO:0031419 cobalamin binding,
  GO:0019843 rRNA binding, GO:0031167 rRNA methylation. All four are activities RlmX itself
  performs, so the participation test is met; GO:0031167 is the single process term used,
  with GO:0000154 rRNA modification left out as its ancestor.
• Zinc ion binding (GO:0008270) is not added: the paper's zinc attribution for
  C453/C458/C464 is explicitly a literature-based expectation ("is thought to"), no metal
  was measured, and the mutant phenotypes are equally consistent with a direct RNA-binding
  role. Recorded as a knowledge gap and a suggested experiment instead.
• No anaerobic-growth or translation process terms are proposed: the growth and
  peptidyl-transfer phenotypes are downstream consequences of the modified ribosome, not
  steps RlmX carries out.