These notes rest on the UniProt record, the GOA rows and the cached publications;
the Falcon deep-research report arrived after the review was written and is
cross-checked at the end of this file.
F1QMN8_DANRE, 989 aa, submitter name "Toll-like receptor 21 precursor", gene name
tlr21 (synonym TLR21.1), cross-referenced to RefSeq NP_001186264, GeneID 402884 and
ZFIN ZDB-GENE-040220-4. NCBI gene 402884 is "tlr21 toll-like receptor 21" on
chromosome 16, so this is the right gene, and the length matches the
receptor characterised in the literature
PMID:24282308. PANTHER subfamily PTHR24365:SF545, shared with chicken TLR21.
The entry's DE line carries EC=3.2.2.6, inherited from the RefSeq record; this is
the source of a spurious enzymatic GO annotation, discussed below.
GOA carries GO:0061809 "NAD+ nucleosidase activity, cyclic ADP-ribose generating"
(IEA from EC:3.2.2.6) on this entry, because the TrEMBL DE line inherited that EC
number from RefSeq. This is removed. The claim presumably originates in the NADase
activity of some TIR domains (plant TIR proteins and human SARM1), and UniProt
states the general case explicitly for Toll-like receptors: "In some plant proteins
and in human SARM1, the TIR domain has NAD(+) hydrolase (NADase) activity (By
similarity). However, despite the presence of the catalytic Asp residue, the
isolated TIR domain of human TLR4 lacks NADase activity (By similarity). Based on
this, it is unlikely that Toll-like receptors have NADase activity"
(the same caution appears on every reviewed mouse entry of this family, e.g.
file:mouse/Tlr13/Tlr13-uniprot.txt lines 141-147). No NADase activity has been reported for any TLR21,
and the functional literature on this receptor concerns CpG-DNA sensing.
Compared tlr21-deep-research-falcon.md against the finished review. Right gene, right
accession, and it independently reaches the same conclusions on every substantive point:
CpG-DNA sensing with a GTCGTT preference distinct from zebrafish TLR9's GACGTT/AACGTT
preference; no response to other PAMP classes; intracellular rather than surface
localisation with ER-to-endosome trafficking via UNC93B1; functional analogy to mammalian
TLR9 without orthology; and cooperative CpG surveillance with TLR9 in the same animal.
Most usefully, it reaches the same verdict on the enzymatic row that this review removed:
[file:DANRE/tlr21/tlr21-deep-research-falcon.md "The EC number 3.2.2.6 assigned to F1QMN8 in UniProt appears to be an annotation error"].
That is now cited on the GO:0061809 REMOVE row as concurrence; the argument itself still
rests on the UniProt family statement about TIR-domain NADase activity and on the absence of
any enzymatic report for a TLR21.
One addition taken up, as a question rather than an annotation:
Not taken up: the CpG-ODN2007 protection result against Vibrio vulnificus is a ligand
treatment in whole animals, not a manipulation of this receptor, so it cannot support a
receptor annotation beyond what PMID:24282308 already supports; the interferon and IRF3/IRF7
outputs are extrapolated from chicken TLR21 and from pathway diagrams rather than measured for
the zebrafish receptor, so no interferon-production term is proposed here (unlike the chicken
review in this batch, where the measurement exists); the LRR-count and docking-model details
come from carp, catfish and chicken proteins; and the aquaculture-adjuvant material is
application rather than function.
Actions: every existing annotation unchanged, including the GO:0061809 REMOVE and the
GO:0005886 over-annotation call, and no new annotations. Changes are one new reference (the
Falcon file), one new supported_by quote and one new suggested question.