PP_1969

Summary: Catalytic activity is plausible for this radical-SAM family member, but the exact reaction is unresolved.

Reason: Under the project rule GO:0003824 is MODIFY where a reaction is known and KEEP_AS_NON_CORE where it is not. No reaction is established for PP_1969: GO:0061798 is itself UNDECIDED here, so there is no specific term to refine to, and the row is as precise as the evidence allows. Kept rather than left UNDECIDED, which the enum reserves for annotations whose supporting publications could not be read; the cluster and SAM sites are intact, so catalysis itself is not in doubt, only its reaction.

Summary: Molybdenum-cofactor pathway involvement depends on the unresolved GTP 3',8'-cyclase activity and has no gene-context support.

Reason: Restored to UNDECIDED (2026-10-16) together with GO:0061798. PP_1969 would do part of the work of this process only by catalysing the MoaA step, which is not established (see the GO:0061798 row). Nothing else ties it to the pathway: no molybdenum-cofactor or molybdoenzyme gene lies within six loci.

Summary: Metal-ion binding is consistent with the radical-SAM iron-sulfur motif but is non-core.

Reason: Following the RADICAL_SAM project rule for the generic domain-derived rows (projects/RADICAL_SAM.md): a lateral correction. The iron of a radical SAM protein is cluster-bound, and GO:0051539 is already carried by this entry.

Proposed replacements: 4 iron, 4 sulfur cluster binding

Summary: This broad iron-sulfur-cluster binding term is redundant with the more specific retained [4Fe-4S]-cluster binding annotation.

Reason: Following the RADICAL_SAM project rule for the generic domain-derived rows (projects/RADICAL_SAM.md): correct but general. MODIFY rather than TOO_GENERAL so that the replacement is stated explicitly, as in the project's other 30 reviews. UniProt's [4Fe-4S] cofactor line for Q88LG4 rests only on an ARBA rule, but the bioinformatics analysis shows the cluster-binding residues are intact, so the [4Fe-4S] assignment holds independently of that rule.

Proposed replacements: 4 iron, 4 sulfur cluster binding

Summary: Specific [4Fe-4S]-cluster binding is plausible but is not an exact pathway-function assignment.

Reason: UniProt and the conserved cysteine motif support a radical-SAM [4Fe-4S] cluster.

Summary: GTP 3',8'-cyclase activity is possible but not established for PP_1969: the MoaA ligand-binding residues are conserved, but the protein fails both function-level MoaA models and the score loss is not explained as benign.

Reason: Reverted from ACCEPT (2026-10-15) to UNDECIDED (2026-10-16) after scoring the function-level models. NCBIfam TIGR02666 (moaA): Q88LG4 scores 269.2 bits, 14.8 below the trusted cutoff (284.05) and 1.8 below the noise cutoff (270.95), against 370-403 bits for every reviewed MoaA tested, including both P. aeruginosa paralogs, and 81 bits for the KT2440 PqqE control. HAMAP MF_01225_B: normalized score 30.5, below the level-0 trigger cutoff 35.254 that every reviewed MoaA passes; the rule's size and 14 FT binding-residue conditions are all met, so the failure is in the profile score itself. The loss is not a missing non-catalytic segment. It is spread along the protein, mostly in the N-terminal radical-SAM region and the C-terminal twitch/tail, and falls at eight positions that every canonical MoaA conserves but both paralogs replace (e.g. Q88E69 R30 and R78), none of them a UniProt-annotated ligand position. All 17 annotated ligand positions align confidently and are identical. Ligand-site conservation therefore does not discriminate here: a MoaA-derived paralog keeps those residues whatever its present substrate. The shared PANTHER PTHR22960:SF0 (and the SFLD MoaA-like group and KEGG K03639) is the classification that produced this TreeGrafter row, and is treated as non-discriminative between paralogs, exactly as for PP_3571, gabD-II and PP_4594 in the same re-review batch. Unlike PP_2482, PP_1969 has no molybdenum-cofactor or molybdoenzyme gene in its neighbourhood. Three paralogs assigned one activity on sequence alone is the paralog over-annotation pattern. A blinded OpenScientist run judged the term supported from the same architecture, cysteine and subfamily evidence, without the model scores. The call needs an assay or complementation, not more sequence similarity.

Summary: cPMP synthase is the distinct MoaC reaction and should not be assigned to a stand-alone MoaA-family protein.

Reason: Wrong step of a two-enzyme sequence, verifiable from the term definition. GO:0061799 is defined as (8S)-3',8-cyclo-7,8-dihydroguanosine 5'-triphosphate = cyclic pyranopterin phosphate + diphosphate - that is, it consumes the product of the MoaA reaction, and is the MoaC step. This protein is a MoaA-type radical SAM enzyme: Pfam PF04055 Radical_SAM plus PF06463 Mob_synth_C with the MoaA_twitch domain (InterPro IPR010505, IPR040064 MoaA-like), the architecture that performs GTP 3',8-cyclase chemistry (GO:0061798, propagated by the same node and retained as UNDECIDED here pending target-specific evidence). It has no MoaC-type pyranopterin synthase fold. P. putida KT2440 encodes a dedicated moaC, so the cPMP step is accounted for by a separate gene. The joint annotation comes from InterPro IPR050105, which lumps MoaA and MoaC into one MoCo-biosynthesis grouping; the two activities are nonetheless distinct reactions on distinct substrates.