CERKL

Summary: Do not infer CERKL catalytic activity from its kinase-family homology.

Reason: The PAINT assertion predicts lipid phosphorylation, but target-specific primary assays argue against transferring that catalytic function to CERKL. PMID:19158957 tested four human retinal CERKL variants in cell lysates and purified GST formats, with CERK controls, standard and photoreceptor-derived lipid substrates; long human variants were also tested by cellular phosphate labeling in murine 661W cells. No CERKL-dependent phosphorylation was detected. PMID:23501591 extended testing to very-long-chain ceramides and mixed-lipid phosphorus assays without positive CERKL activity. This is a target-function objection, not an objection to the number or identities of extant PAINT sources. PTN002285910 historical node placement was not reconstructed. The negative assays are scoped to the tested constructs, substrates and conditions; they do not establish that every CERKL product is catalytically inactive on every possible substrate. Preserve the original source fields and do not create a NOT assertion.

Summary: CERKL localizes to photoreceptor outer segments in rodent retina.

Reason: The exact mouse donor A2AQH1 carries an IDA for this term from PMID:23142158. Its full Results and Figure 3 describe CERKL immunolabeling in both inner and outer segments in mouse and rat, with primary-antibody omission controls. This supports the conserved localization transfer, without describing a direct human-retina assay or assigning a particular one of the eight human products. Segment localization does not establish a phototransduction activity.

Summary: CERKL localizes to photoreceptor inner segments in rodent retina.

Reason: The exact mouse donor A2AQH1 carries an IDA for this term from PMID:23142158. Its full Results and Figure 3 describe CERKL immunolabeling in both inner and outer segments in mouse and rat, with primary-antibody omission controls. This supports the conserved localization transfer, without describing a direct human-retina assay or assigning a particular one of the eight human products. Segment localization does not establish a phototransduction activity.

Summary: The supported interaction partner PPM1A is a protein phosphatase.

Reason: Three exact IntAct records join Q49MI3 with P35813: IM-29278-5301, IM-29278-24055 and IM-29278-28347. They report CERKL bait and PPM1A prey in human HEK293T or HCT116 anti-tag affinity-purification experiments; two records use spoke expansion. Refine the original generic MF to protein phosphatase binding at the resolution of the observed physical association. The current P35813 record identifies PPM1A and its protein Ser/Thr phosphatase activity. These assays do not establish a purified binary interface, phosphatase activation, a CERKL catalytic activity or a defined phosphatase substrate. Retain the original partner and evidence fields. Exact machine records and query hashes are preserved in CERKL-source-evidence.json. Spoke-expanded bait-prey edges represent co-complex association and can include indirect contacts. No separate low-throughput orthogonal validation was established in the inspected sources; three curated records are not three independent binary-binding validations.

Proposed replacements: protein phosphatase binding

Summary: The supported interaction partner PPM1A is a protein phosphatase.

Reason: The publication-linked author U2OS AP-MS network contains CERKL to PPM1A edge 609146. The exact gene identifiers agree with Q49MI3 and P35813; this is the AP-MS network, not a prediction from the integrated multimodal map. Refine the original generic MF to protein phosphatase binding at the resolution of the observed physical association. The current P35813 record identifies PPM1A and its protein Ser/Thr phosphatase activity. These assays do not establish a purified binary interface, phosphatase activation, a CERKL catalytic activity or a defined phosphatase substrate. Retain the original partner and evidence fields. Exact machine records and query hashes are preserved in CERKL-source-evidence.json. This second AP-MS context corroborates association, not a low-throughput orthogonal binary-binding assay.

Proposed replacements: protein phosphatase binding

Summary: The nuclear PAINT assertion is compatible with CERKL shuttling and nuclear complexes.

Reason: Retain the phylogenetic assertion with target support from endogenous human ARPE-19 nuclear staining, human-product nuclear shuttling and CERKL/SIRT1 nuclear colocalization and association. The RNA-binding study suggests a nuclear loading step but does not directly establish that step, so no nuclear RNA-transport mechanism is asserted. PTN008536821 history was not reconstructed; target Q49MI3 in the WITH list is legitimate experimental grounding, not circular evidence.

Summary: Nuclear localization is observed in human and mouse cells.

Reason: The exact A2AQH1 donor has IDA nuclear annotations from PMID:21151604 and PMID:23142158. The first study also directly stains endogenous CERKL in human ARPE-19 nuclei, with recombinant-antigen pre-absorption controls. Localization varies between cells and conditions; this does not mean every product is constitutively nuclear.

Summary: Nucleolar accumulation is a documented expression-context observation.

Reason: PMID:19158957 reports nucleolar localization of tagged human retinal products in transfected cells, and the authentic abstract of PMID:19501188 independently reports wild-type CERKL nucleolar accumulation. The latter corroborates the location, but its full Methods were not recovered, so its construct, host and product mapping are not independently assigned. In contrast, PMID:21151604 Figures 4J-L and 5D-F find endogenous CERKL excluded from fibrillarin-positive nucleoli in human ARPE-19 and mouse 661W cells. Retain the expressed-product location as non-core, while leaving its physiological significance and expression dependence unresolved. No ribosome-biogenesis function is inferred.

Summary: CERKL acts in cytoplasmic ribonucleoprotein and stress-response contexts.

Reason: Retain the PAINT active-in assertion: human CERKLa directly binds mRNA in vitro and occurs in cytoplasmic compact mRNPs and stress granules in cells. Endogenous human ARPE-19 staining independently supports cytoplasmic localization. PTN008536821 history was not reconstructed, and the target in its evidence list is not circular. The in vitro RNA interaction and cellular associations are distinguished from a demonstrated RNA-transport motor mechanism.

Summary: The cytoplasmic electronic annotation agrees with target experiments.

Reason: The mouse A2AQH1 IDA source from PMID:21151604 and the original SL-0086 mapping agree with endogenous human ARPE-19 staining and cellular human CERKLa ribonucleoprotein studies. Retain cytoplasm as a broad location encompassing these demonstrated activities, without generalizing the localization to all products in every cell.

Summary: ER-associated CERKL is observed in an expression-dependent experimental context.

Reason: PMID:19158957 Figure 4E explicitly reports calnexin overlap for human retinal CERKL products expressed in COS-7 cells; Figure 5 and sucrose-gradient Methods independently place transfected human HeLa CERKLa-HA in PDI-containing fractions. The signal is not restricted to the ER, and these assays do not establish endogenous retinal localization, membrane topology or a resident ER function. In the original Results, the absence of precise overlap refers to endosomes and mitochondria, not to the positive ER/Golgi comparison. PMID:24498393 Results and Figure S1 instead describe CERKLa aggregates partly around the ER without matching marker colocalization. Retain the experimentally supported conditional location as non-core; neither exclusivity nor a universal steady-state distribution is asserted by the source located_in relation. The later negative context limits generalization but does not erase the earlier positive imaging and fractionation. ORGANELLES check (2026-11-05, ER batch 2). The row maps SL-0095 from a UniProt location line that is untagged and written for isoform 2 (CERKLa) only. That restriction is UniProt's, not the source's - PMID:19158957 tested several isoforms and found that none had a restricted localization, so the isoform-free GOA row is not an isoform defect. CERKL is a soluble protein with no signal peptide, TM segment or ER-retention motif, so any ER pool is on the cytosolic face, and all of the evidence is on transiently overexpressed tagged constructs (calnexin overlap in COS-7; HA-tagged protein in PDI-containing sucrose fractions that also hold Golgi markers); no endogenous ER staining exists. The organelle claim is kept as a minor, non-core association, as for AHCY in ER batch 1; this is a close call, and MARK_AS_OVER_ANNOTATED would be the action if the overexpression pattern is judged an artefact. No sub-compartment is added (a cytosolic-face pool stays at the root).

Summary: Golgi-associated CERKL is observed in an expression-dependent experimental context.

Reason: PMID:19158957 Figure 4E reports overlap of expressed human CERKL with GM130 in COS-7 cells. Its Figure 5 and Methods show human HeLa CERKLa-HA co-fractionating most strongly with TGN38 and also with GM130; overlap across multiple fractions does not establish organelle exclusivity or membrane topology. The subsequent sentence describing imprecise overlap concerns endosome and mitochondrial markers. In contrast, PMID:24498393 Results and Figure S1 report the CERKLa-containing aggregates absent from cis- and trans-Golgi compartments in its COS-7 assay. PMID:19501188 independently supports CERKL nucleolar accumulation, but a CERK-versus-CERKL localization contrast in its abstract is not proof that CERKL can never associate with Golgi compartments. Retain the positive source-backed location as non-core with the contradictory expression context explicit; no constitutive or endogenous retinal Golgi residence is claimed.

Summary: Do not infer CERKL catalytic activity from its kinase-family homology.

Reason: The official current IPR001206 record genuinely maps its DAG-kinase catalytic-domain signature to GO:0016301, so the source chain is faithful. A domain match alone nevertheless does not establish catalysis in CERKL. Full target-specific studies PMID:19158957 and PMID:23501591 repeatedly failed to detect CERKL-dependent phosphorylation with working CERK controls across standard/retinal lipids and broader substrate panels. The source-supported lipid/RNA-related roles are separate from a kinase mechanism. Reject this catalytic transfer to CERKL rather than recasting the domain mapping as a typo. The negative assays are scoped to the tested constructs, substrates and conditions; they do not establish that every CERKL product is catalytically inactive on every possible substrate. Preserve the original source fields and do not create a NOT assertion.

Summary: CERKL reduces oxidative-stress-induced apoptosis in a defined human cell assay.

Reason: PMID:19158957 Figure 7 measures caspase-dependent PARP cleavage in human HeLa cells expressing human 532-aa CERKLa. Protection occurs at 300 micromolar hydrogen peroxide over 12-24 hours, but not significantly at 400 micromolar, at 48 hours, or with the other tested death stimuli. The separate COS-7 propidium-iodide readout includes necrotic as well as apoptotic death. These target experiments support the conserved negative-apoptosis assertion without assigning direct antioxidant chemistry or caspase inhibition. The PAINT node history was not reconstructed; self-target evidence is valid.

Summary: CERKL associates with selected sphingolipids in protein-lipid overlays.

Reason: The exact mouse donor A2AQH1 annotation traces to PMID:23501591. Its Figure 8 directly tests lysates of human HEK293T cells expressing human CERKLa or a mixture of four retinal products against immobilized lipids: sphingomyelin, glucosylceramide and galactosylceramide are positive, whereas the empty-vector background excludes the retinoic-acid signal. Retain binding at this assay resolution; the experiment does not isolate a purified binary interaction or establish lipid transport/catalysis. Earlier PMID:19158957 reported negative overlays with both purified GST-CERKLa and transfected HEK293T lysate, so binding is not uniformly observed across protocols. The retina lipidomics model retains roughly 35% Cerkl transcription and is a knockdown rather than a complete null. The source transfer remains an unchanged mouse-donor IEA; the independently inspected positive lipid-overlay experiment itself used human CERKLa-containing lysate. Its negative lipid panel included ceramide, lactosylceramide, GM1 and free fatty acids, so this is not unrestricted lipid recognition.

Summary: Endogenous human CERKL can concentrate around the nucleus.

Reason: The original mouse donor A2AQH1 carries an IDA from PMID:21151604. Full Results and Figure 4 also directly show endogenous human ARPE-19 CERKL concentrated in the perinuclear cytoplasm, supported by antigen-pre-absorption controls. Retain the location without equating it with a specific organelle or a unique product.

Summary: Human CERKLa binds mRNA directly through its N-terminal region.

Reason: PMID:24498393 Figure 4C-D uses purified bacterial His-MBP-tagged human CERKLa and mRNA from human retina or COS-7 cells, with BSA/MBP controls and unlabeled competition. The 1-256 fragment binds while the 252-532 fragment does not. The paper identifies NM_201548.4; its NP_963842.1 translation exactly matches current Q49MI3-2, independently checked against the unchanged UniProt sequence. This informative MF is absent from the 16 original assertions. The cap-affinity assay instead uses a HEK293T immunoprecipitate, so direct cap recognition remains unresolved. The specific RNA targets and the roles of the other seven products are not established.