TBCK is a case where the gene name asserts more than the evidence supports: "TBC1 domain-containing
kinase" carries "kinase" in the symbol, and GOA carries GO:0004672 protein kinase activity. The
supporting evidence code is NAS (non-traceable author statement) against the Manning human-kinome
census, not an experiment — and the same GOA record simultaneously carries a NOT annotation for
GO:0006468 protein phosphorylation from the very same reference. GOA therefore already holds both
sides of the question, incoherently.
Provenance of the NAS pair is the kinome census, which was an inventory of kinase-fold genes, not an
assay: PMID:12471243. UniProt's own comment on the entry (Q8TEA7) is
DOMAIN: The protein kinase domain is predicted to be catalytically inactive. {ECO:0000255} — i.e.
a sequence-model prediction, matching the negative NOT|involved_in GO:0006468 row.
The 2026 biochemistry closes this. Full-length human TBCK was expressed in Sf9 cells and purified for
the first time; the pseudokinase domain does not even bind nucleotide:
PMID:42546825. The DSF titrations with ATP, GTP, ADP, GDP and Mg2+ gave no ligand-induced
shift: PMID:42546825. Coupled-enzyme assays found no hydrolysis of either nucleotide:
PMID:42546825 and PMID:42546825.
Caveats weighed. This is a single laboratory publishing in a methods-oriented journal (Protein Expr
Purif), and PMID:41958985 is the bioRxiv preprint of the same work by the same authors — it is not
independent corroboration, and I have counted it as one body of evidence, not two. Against that: the
biochemistry agrees with a sequence argument that was already in UniProt and in the original disease
papers, the assays include the right positive control (asparagine synthetase), and no publication in
twenty years has reported a TBCK kinase activity or a TBCK substrate. On balance the kinase and
ATP-binding assignments are argued down rather than merely demoted.
The TBC domain, unlike the pseudokinase domain, retains its catalytic machinery:
PMID:42546825.
Disease genetics pointed the same way a decade earlier — the recurrent missense allele sits on the TBC
arginine finger: PMID:27040692 and PMID:27040692.
This was demonstrated in 2025-2026. A Fudan group identified the GAP and its target
(primary paper PMID:41207833, Sci Bull; summarised by the same authors in an Autophagy punctum):
PMID:41789809, PMID:41789809, and the loss-of-function phenotype is what a RAB5-GAP loss predicts:
PMID:41789809.
So the two halves of the protein point in opposite directions: the domain in the gene's name is dead, and
the domain that is not in the name is the catalytic one.
TBCK is a subunit of the five-protein FERRY complex, a Rab5 effector:
PMID:37267905.
The mRNA contact is made by other subunits, not by TBCK:
PMID:42546825. The regulation of intracellular mRNA localization annotation is
therefore a complex-level process TBCK participates in, not an activity of TBCK itself.
The original TBCK paper reported mTOR effects on knockdown:
PMID:23977024 and
PMID:23977024.
A human neural-progenitor model does not reproduce it:
PMID:39553985
and PMID:39553985.
The 2026 review of the field says as much:
PMID:41789809. Kept, but non-core, with the dispute recorded.
GO:0004672 protein kinase activity (NAS) and GO:0005524 ATP binding (IEA InterPro + NAS) →NOT|involved_in GO:0006468 protein phosphorylation (NAS) → ACCEPT. The negative assertion isGO:0005096 GTPase activator activity (IBA) → ACCEPT as the core molecular function.GO:0032006 regulation of TOR signaling (IMP) → KEEP_AS_NON_CORE, dispute recorded.GO:0008283 cell population proliferation and GO:0030036 actin cytoskeleton organization (IMP) →protein binding → over-annotated.Does the pseudokinase domain do anything? Class I pseudokinases commonly work as allosteric or
scaffolding modules, and TBCK's has a sizeable but low-confidence cavity in the AlphaFold model. The
rhodanese-like domain is likewise predicted dead:
PMID:42546825. Neither
domain has a described binding partner or function, yet disease alleles hit the protein throughout.