Batch 1 of the INNATE_IMMUNITY project (Toll/TLR axis). Reviewed alongside TLR8
(Q9NR97), with which TLR7 shares most of its annotation set and its PAINT node.
Deep research: TLR7-deep-research-falcon.md (Falcon) arrived after the
review was written; see the cross-check section at the end. The notes below are
built from the UniProt record, cached publications, QuickGO and, where noted, open
PMC full text.
Endolysosomal single-pass type I membrane pattern-recognition receptor: a
leucine-rich-repeat ectodomain, one transmembrane helix, and a cytoplasmic TIR
domain. UniProt Q9NYK1: signal 1–26, chain 27–1049, extracellular 27–839,
transmembrane 840–860.
Dual ligand-binding architecture. TLR7 is not a simple "ssRNA receptor". The
crystal structures resolve two spatially distinct sites:
The endogenous site-1 ligand is 2',3'-cGMP, not 3',5'-cGMP. This matters for
the GO:0070305 response to cGMP row (see below).
Guanosine analogues and deoxyguanosine are RNA-independent agonists acting at the
same site:
ssRNA sensing and antiviral function.
- PMID:14976261
- PMID:16188996
Localisation. ER at rest, relocating to endosomes/lysosomes; UNC93B1 is the
trafficking chaperone.
- PMID:22164301
- PMID:12738885
- PMID:33432245
Outputs and disease. MyD88 myddosome → NF-κB (proinflammatory cytokines) and
IRF7 (type I IFN). Loss of function causes IMD74 (severe COVID-19); gain of
function causes SLEB17 (lupus).
- PMID:32706371
- PMID:32706371
GO:0070305 response to cGMP (IMP, PMID:35477763) is chemically wrong.
GO:0070305 is defined as a response to "cGMP (cyclic GMP, guanosine
3',5'-cyclophosphate)" (QuickGO). The TLR7 site-1 ligand released by
RNase T2/PLD3/PLD4 is guanosine 2',3'-cyclic monophosphate
PMID:38697119, and the cited lupus paper says exactly that
("guanosine and 2',3'-cGMP"). MODIFY to GO:1905108 guanosine binding, which
is what the structures and the deoxyguanosine work establish directly. GO has
no term for 2',3'-cGMP binding or for a response to it — raised as a question.
GO:0071260 cellular response to mechanical stimulus (IEP, PMID:19593445)
cites an unrelated paper. PMID:19593445 is "Expression of the Bcl-2 protein
BAD promotes prostate cancer growth" (PLoS ONE 2009); the cached record has
full text and contains zero occurrences of "TLR" or "toll". The same row
exists on TLR8 and on TLR3 (where it was also left UNDECIDED). Marked
UNDECIDED, with reference_review.correctness: WRONG_IDENTIFIER.
GO:0034158 toll-like receptor 8 signaling pathway on TLR7. An Ensembl
Compara transfer from mouse Tlr7 (UniProtKB:P58681). GO:0034158 is defined as
"The series of molecular signals initiated by a ligand binding to the
endolysosomal toll-like receptor 8" (QuickGO) — by definition it cannot be a
process TLR7 carries out. TLR7 already has GO:0034154 four times over.
REMOVE.
GO:0043235 signaling receptor complex (IDA, PMID:23382219) cites the SNX17
PX-FERM structure paper. I retrieved the open PMC copy (PMC3581954, 244 kB of
text) and it contains no mention of TLR7 or of Toll-like receptors. The claim
itself is true — TLR7 forms an activated m-shaped homodimer PMID:27742543 and
associates with UNC93B1 PMID:33432245 — so the row is kept (ACCEPT) with the
real support attached and the reference flagged as a wrong identifier.
GO:0003725 double-stranded RNA binding (IDA, PMID:16111635). The cached
record is abstract-only and says "RNA signals through human TLR3, TLR7, and
TLR8"; dsRNA is TLR3's ligand, while TLR7's second site binds single-stranded
uridine-containing RNA PMID:27742543. MODIFY to GO:0003727.
GO:0035197 siRNA binding (IEA from mouse Tlr7 + ISS). siRNA is a duplex;
the structurally defined TLR7 sites bind guanosine and single-stranded
uridine-containing RNA. Immunostimulatory siRNA plausibly acts through
single-stranded regions or degradation products rather than duplex binding.
MARK_AS_OVER_ANNOTATED.
Type II interferon. Two rows. The IDA PMID:16286015 row rests on a paper
that measured IFN-α/β and IFN-λ only — "Human TLR-7-, -8-, and -9-mediated
induction of IFN-alpha/beta and -lambda Is IRAK-4 dependent" — so it is
MODIFY→GO:0034346 positive regulation of type III interferon production
(matching the decision already taken in the TLR9 review). The ARBA IEA row is
independently supported for TLR7 by PMID:32706371 (IFN-γ abolished in TLR7
loss-of-function patients' PBMCs), but that is a 7-day PBMC readout and the
IFN-γ producers are not the TLR7-expressing cells, so it is kept as non-core.
Plasma membrane. The is_active_in GO:0005886 IBA sits at the TLR7/8/9
PAINT node and there is a melanocyte surface IDA PMID:19740627. TLR7's
intracellular confinement is functionally essential, and the endolysosome is
where it signals, so both rows are kept as non-core rather than accepted —
consistent with the TLR9 review. Whether is_active_in plasma membrane should
stand at that node is raised as a question.
GO:0005515 protein binding (IPI, PMID:33432245, UNC93B1). Uninformative;
TLR7 is the client of UNC93B1 and the biology is trafficking, not an activity
of TLR7. REMOVE (mirrors the TLR3 review's handling of the same paper).
Removal does not dispute the interaction.
Reactome bulk rows. 19 GO:0010008 endosome membrane and 19
GO:0036020 endolysosome membrane TAS rows, plus Golgi and ER membrane rows,
all from the Reactome TLR7/8/9 cascade models. Endosome/endolysosome membrane
are accepted as core; ER membrane is accepted (resting location); the two Golgi
membrane rows are kept as non-core because Golgi transit is a modelling
assumption of the UNC93B1 route with no direct evidence for human TLR7 (and
Golgi markers explicitly failed to colocalise with the paralogue TLR8,
PMID:22164301).
Compared the Falcon report (review-level sources: Zheng 2023, Hamerman & Barton
2024, von Hofsten 2024, Lind 2022; citation keys are not PMIDs) against every
review decision, the description and core functions.
referencesGO:0038187 IMP row.