LMTK2 review notes

Setup and research provenance

• Review initiated 2026-08-09 for human LMTK2 (HGNC:17880; UniProtKB:Q8IWU2).
• just fetch-gene human LMTK2 --uniprot-id Q8IWU2 retrieved 43 live QuickGO rows and seeded 42 normalized annotation objects.
• Provider-generated deep research was attempted with Falcon (just deep-research-falcon human LMTK2) but the Edison service returned HTTP 402 Payment Required.
• A second attempt with Perplexity (just deep-research-perplexity human LMTK2) returned HTTP 401 insufficient quota.
• No provider-named deep-research file was created manually. This notes file records the primary-literature and database synthesis instead.

Evidence journal

Identity and naming

• The reviewed human record is Q8IWU2, a 1,503-aa product named serine/threonine-protein kinase LMTK2. [UniProtKB:Q8IWU2 "ID LMTK2_HUMAN Reviewed; 1503 AA."] [UniProtKB:Q8IWU2 "DE RecName: Full=Serine/threonine-protein kinase LMTK2;"]
• Use LMTK2 as the gene symbol and principal protein name. AATYK2, BREK, KIAA1079, KPI2, and LMR2 are search aliases, not separate genes or products. [UniProtKB:Q8IWU2 "GN Name=LMTK2; Synonyms=AATYK2, BREK, KIAA1079, KPI2, LMR2;"] The 2024 nomenclature review also identifies the legacy literature names KPI-2, Cprk, BREK, AATYK2, KIAA1079, and LMR2 as names formerly used for LMTK2. PMID:38191649
• “Lemur tyrosine kinase 2” is a historical misnomer and must not be read as evidence of tyrosine kinase activity: recombinant KPI-2/LMTK2 was reported to be strictly Ser/Thr-specific. PMID:16887929

GOA provenance and identifier audit

• The 43 QuickGO rows normalize to 42 review objects because two raw rows encode the same LMTK2–PPP1R2 IPI assertion (same product, qualifier, GO term, evidence, reference, and partner), differing only in submitting source and date. The IntAct row is [file:genes/human/LMTK2/LMTK2-goa.tsv "UniProtKB Q8IWU2 LMTK2 enables GO:0005515 protein binding molecular_function ECO:0000353 IPI PMID:12393858 UniProtKB:P41236 9606 Homo sapiens IntAct Serine/threonine-protein kinase LMTK2 20260725"] and the older UniProt row is [file:genes/human/LMTK2/LMTK2-goa.tsv "UniProtKB Q8IWU2 LMTK2 enables GO:0005515 protein binding molecular_function ECO:0000353 IPI PMID:12393858 UniProtKB:P41236 9606 Homo sapiens UniProt Serine/threonine-protein kinase LMTK2 20031216"]. Coalescing them preserves the biological assertion rather than dropping evidence.
• Both GO:0070853 “myosin VI binding” rows carry WITH/FROM=UniProtKB:P13533. [file:genes/human/LMTK2/LMTK2-goa.tsv "UniProtKB Q8IWU2 LMTK2 enables GO:0070853 myosin VI binding molecular_function ECO:0000353 IPI PMID:18029400 UniProtKB:P13533 9606 Homo sapiens UniProt Serine/threonine-protein kinase LMTK2 20110110"] [file:genes/human/LMTK2/LMTK2-goa.tsv "UniProtKB Q8IWU2 LMTK2 enables GO:0070853 myosin VI binding molecular_function ECO:0000353 IPI PMID:18429820 UniProtKB:P13533 9606 Homo sapiens UniProt Serine/threonine-protein kinase LMTK2 20110110"]
• P13533 is the reviewed human MYH6 entry, whereas the intended unconventional myosin-VI partner is reviewed human MYO6, Q9UM54. [UniProtKB:P13533] [UniProtKB:Q9UM54]
• This is a source-identifier miscitation, not a reason to reject the biological term: one cited paper directly identifies LMTK2 as a myosin-VI-binding transmembrane kinase. PMID:18029400 Preserve P13533 in the machine-sourced annotation, mark the source as bad/miscited in the propagation review, and use Q9UM54 only as corrected supporting context.

Architecture, topology, and isoform boundary

• The reviewed product has two N-terminal transmembrane helices and a protein-kinase domain at residues 137–407. [UniProtKB:Q8IWU2 "FT TRANSMEM 11..31"] [UniProtKB:Q8IWU2 "FT TRANSMEM 43..63"] [UniProtKB:Q8IWU2 "FT DOMAIN 137..407"]
• Fluorescence-protease-protection experiments with terminal GFP fusions place both termini, and therefore the kinase active site, on the cytoplasmic face of intracellular membranes. PMID:23114966 This is direct topology evidence from a tagged-cell system, not merely a sequence prediction. PMID:23114966
• The current reviewed flat-file exposes one 1–1,503 chain and contains no ALTERNATIVE PRODUCTS or VAR_SEQ stanza in a literal audit. [UniProtKB:Q8IWU2 "FT CHAIN 1..1503"] Therefore this review should not invent an isoform field or claim isoform-specific function. This is a “no curated isoforms in this record” boundary, not proof that alternative human transcripts or proteoforms do not exist.

Direct biochemical core and PP1 regulatory assembly

• The founding study cloned a human 1,503-residue transmembrane KPI-2/LMTK2 product and directly demonstrated Ser/Thr kinase activity in autophosphorylation and added-substrate assays. PMID:12393858 PMID:12393858
• A second biochemical study used homogeneous recombinant LMTK2 kinase domain and a 1,154-peptide array, confirming strict Ser/Thr specificity and activity toward CFTR and phosphorylase substrates. PMID:16887929 PMID:16887929
• PP1C is the best established direct substrate: LMTK2 phosphorylated PP1C Thr320 and reduced phosphatase activity. PMID:12393858 This supports the specific Ser/Thr phosphatase-inhibitor interpretation, not only generic “protein phosphatase inhibitor activity.”
• LMTK2 is also a PP1-binding regulatory scaffold. Its C-terminal domain binds PP1C through a VTF motif, while inhibitor-2/PPP1R2 associates with the same broad tail region with or without PP1C. PMID:12393858 PMID:12393858 The functional unit is therefore a membrane-localized kinase/PP1/inhibitor assembly, although the cached abstract does not establish a single fixed stoichiometry.
• CDK5R1/p35 is a real regulator and interactor, but the experiments span a human-brain cDNA bait, primary rat cortical neurons, and transfected CHO/COS cells, so the interaction should not be generalized as a human-neuron IDA without qualification. PMID:12832520 PMID:12832520 PMID:12832520 The same study showed reciprocal kinase regulation: PMID:12832520

Myosin-VI-linked endocytic recycling

• In HeLa-cell depletion experiments, loss of LMTK2 or myosin VI trapped transferrin receptor in swollen early endosomes and strongly reduced tubule formation/delivery into the endocytic recycling compartment. The model boundary is explicit: PMID:18029400 The functional result is PMID:18029400
• The same full text places LMTK2 along the early/recycling route rather than in the late-endosome pathway. PMID:18029400
• The independent BREK/LMTK2 paper reports cytoplasmic vesicle and perinuclear recycling-endosome localization and a block specifically between early and recycling endosomes after depletion. Its cache is abstract-only PMID:18429820 and calls the model only “cells,” so do not invent a lineage or organism boundary beyond the curated human annotation. PMID:18429820 PMID:18429820

Context-dependent cargo and transport roles

• CFTR is a strong direct human-airway substrate and trafficking phenotype: endogenous association and Ser737 phosphorylation were tested in human airway epithelial models, and LMTK2 loss reduced CFTR endocytosis while increasing surface density and chloride secretion. PMID:24727471 PMID:24727471 PMID:24727471 This is compelling cargo-specific biology, but it is not yet evidence that CFTR regulation is the universal core function of LMTK2 across tissues.
• In HeLa cells, LMTK2–PP1C signaling altered the GSK3β–KLC2 phosphorylation pathway and Smad2 cargo binding/signaling. PMID:21996745 PMID:21996745 Treat this as a mechanistically coherent non-neuronal transport/signaling context, not a universal cargo definition.
• A later study found endogenous LMTK2–KLC1 association in HEK293 cells and rat cortical neurons, then used rat neurons to show KLC-dependent axonal transport of LMTK2 and LMTK2-dependent transport of p35/cdk5. PMID:31068217 PMID:31068217 This supports the biological plausibility of the rat-ortholog growth-cone and neuronal-cell-body transfers, but those locations remain non-core for the human review.
• The knockout phenotype is species-restricted: Brek/Lmtk2-null male mice were infertile with azoospermia and defective late spermatogenesis. PMID:17158803 PMID:17158803 This is a genuine organismal role but should not displace the direct biochemical and trafficking core in the human review.
• A 2024 human bronchial epithelial study extends the PP1 mechanism into TGF-β1/Smad3 signaling. PMID:39316683 PMID:39316683 This is another well-bounded epithelial signaling context rather than a reason to make generic TGF-β signaling the core function.

Annotation action rationale

• Accept as core: specific serine/threonine kinase activity, serine/threonine phosphorylation, membrane localization, early/recycling-endosome localization, endocytic/receptor recycling, and myosin VI binding. The enzyme specificity is direct PMID:16887929 and the trafficking mechanism is replicated PMID:18429820; the two myosin-VI annotations remain biologically valid despite their P13533 provenance defect.
• Modify to informative terms: broad protein kinase activity to serine/threonine kinase activity; broad protein phosphorylation to serine- and threonine-phosphorylation processes; generic phosphatase inhibitor activity to protein serine/threonine phosphatase inhibitor activity; and generic PPP1CC “protein binding” rows to protein phosphatase 1 binding. These changes expose the experimentally established specificity PMID:12393858 PMID:12393858
• Repair source-contorted trafficking terms: modify “early endosome to late endosome transport” to endocytic recycling, because the cited experiment reports failure to move from early to recycling endosomes PMID:18429820. Keep both generic perinuclear rows as contextual rather than minting duplicate Golgi/recycling-endosome recommendations; those specific locations already have their own source rows.
• Keep as non-core: ATP binding is intrinsic but generic for the kinase reaction; autophosphorylation is demonstrated but subordinate to substrate phosphorylation; perinuclear localization is true but nonspecific; transferrin transport is the assay cargo/readout of recycling; and rat-transferred growth-cone/neuronal-cell-body locations are ortholog/model context. The source explicitly presents transferrin as an example cargo PMID:18029400 and the rat provenance is present in the rows [file:genes/human/LMTK2/LMTK2-goa.tsv "UniProtKB Q8IWU2 LMTK2 located_in GO:0030426 growth cone cellular_component ECO:0000265 IEA GO_REF:0000107 UniProtKB:D3ZBH5|ensembl:ENSRNOP00000050872 9606 Homo sapiens Ensembl Serine/threonine-protein kinase LMTK2 20260724"] [file:genes/human/LMTK2/LMTK2-goa.tsv "UniProtKB Q8IWU2 LMTK2 located_in GO:0043025 neuronal cell body cellular_component ECO:0000265 IEA GO_REF:0000107 UniProtKB:D3ZBH5|ensembl:ENSRNOP00000050872 9606 Homo sapiens Ensembl Serine/threonine-protein kinase LMTK2 20260724"].
• Mark Golgi localization as over-annotated: the earlier p35/cprk study reports a Golgi-associated signal PMID:12832520, but later imaging found very little LMTK2 overlap with the Golgi marker GM130 PMID:18029400.
• Mark generic binding as over-annotated: PPP1R2/inhibitor-2 and CDK5R1/p35 interactions are real, but root-level “protein binding” is not an informative molecular function. Preserve the evidence in findings/notes rather than elevating GO:0005515 to the core. PMID:12393858 PMID:12832520
• Add only evidence-bounded new annotations: endosome membrane follows from the direct topology/localization synthesis PMID:23114966; positive regulation of endocytosis is restricted to the human-airway CFTR perturbation PMID:24727471; kinesin binding is an orthology transfer because the binding constructs were mouse LMTK2/KLC1/KLC2 PMID:31068217 and remains neuronal/non-core; positive regulation of TGF-β receptor signaling is restricted to the human bronchial epithelial Smad3 result PMID:39316683.

Core/non-core synthesis

• Core function 1: LMTK2 is a two-pass integral-membrane, cytoplasm-facing Ser/Thr kinase that phosphorylates PP1C Thr320 and defined cargo-associated substrates such as CFTR Ser737.
• Core function 2: LMTK2 directly binds PP1 catalytic subunits through a C-terminal VTF motif and recruits inhibitor-2/PPP1R2 into a membrane-associated regulatory assembly PMID:12393858. This separates PP1 docking/scaffolding from the catalytic inhibitory event.
• Core function 3: LMTK2 binds myosin VI and is required for cargo movement/tubulation from early endosomes into the Rab11-positive recycling compartment. This is supported by direct binding and depletion phenotypes in human HeLa cells PMID:18029400 PMID:18029400.
• CFTR endocytosis, KLC/Smad transport, axonal p35/cdk5 delivery, bronchial TGF-β/Smad3 signaling, and mouse spermatogenesis are credible extensions with explicit tissue, cell-line, or species boundaries. They should remain non-core/contextual until their breadth and physiological hierarchy are established.

Gaps and discriminating experiments

1. Separate catalytic from scaffold function in recycling. Rescue LMTK2-depleted cells with wild type, kinase-dead, PP1-binding-motif mutant, and myosin-VI-binding-region mutant proteins; quantify early-to-Rab11 recycling-compartment flux, tubule formation, and transferrin-receptor return. Existing depletion establishes necessity but does not partition the mechanism. PMID:18029400
2. Define endogenous PP1-complex composition and isoform preference. Use endogenous co-immunoprecipitation/cross-linking mass spectrometry and matched phosphatase assays for PPP1CA, PPP1CB, PPP1CC, and PPP1R2, coupled to Thr320 phosphosite measurements. The founding result proves direct PP1C binding and inhibition but not a fixed endogenous stoichiometry. PMID:12393858
3. Map physiological substrate breadth. Compare acute degron loss with kinase-dead rescue in human airway, neuronal, and non-neuronal cells by quantitative phosphoproteomics, then validate direct sites with purified LMTK2. Current direct examples include PP1C Thr320 and CFTR Ser737 PMID:12393858 PMID:24727471, but tissue-general substrate hierarchy is unresolved.
4. Resolve human proteoforms before any isoform-specific curation. Combine long-read RNA sequencing, N/C-terminal proteomics, and topology assays in LMTK2-expressing tissues; map any recovered proteoforms to the current 1,503-aa chain. The present reviewed record supplies only [UniProtKB:Q8IWU2 "FT CHAIN 1..1503"], so isoform-specific claims would currently outrun the curated record.
5. Test neuronal localization directly in human cells. Validate growth-cone/cell-body localization and KLC-dependent p35/cdk5 transport in human iPSC-derived neurons, because the existing transfer is from rat Lmtk2 [file:genes/human/LMTK2/LMTK2-goa.tsv "UniProtKB:D3ZBH5|ensembl:ENSRNOP00000050872"] and the transport experiments used rat cortical neurons PMID:31068217
6. Correct upstream interaction provenance. Submit or flag the two GOA WITH/FROM=P13533 records for repair to Q9UM54 while retaining the cited LMTK2–myosin-VI evidence; the current identifiers resolve to MYH6 and MYO6 respectively. [UniProtKB:P13533] [UniProtKB:Q9UM54]