The 51 seeded source assertions, two alternative products and 20 original reference ID/title pairs are preserved. Final review decisions are evidence-specific; no NEW assertion is proposed. One Ser/Thr kinase core links substrate phosphorylation to neuronal microtubule dynamics and postsynaptic organization. Nuclear transcriptional responses, developmental outcomes, proliferating-cell functions and ciliary expression experiments remain contextual roles.
Early biochemical evidence establishes the enzyme class PMID:15917271, “CDKL5 is indeed a kinase”; the newer studies below resolve substrate and product context.
PMID:30266825(https://pubmed.ncbi.nlm.nih.gov/30266825/) directly tests human CDKL5-dependent MAP1S/CEP131 phosphorylation and a Thr-substituted peptide with substantial retained activity. Ser/Thr activity is appropriate; the serine-specific existing assertion is also correct. Tyr171 autophosphorylation is distinguished from the tested exogenous Tyr-peptide failures. Neither warrants an unrestricted trans-protein tyrosine kinase claim or a universal negative. The work uses the 1030-aa/115-kDa form for cellular re-expression. The independent structural study, PMID:29420175, uses a human 1–303 T169D/Y171E crystallization construct; its WT activity experiments are distinct. Structural divergence and lack of an established cyclin activator justify refining the EC-derived cyclin-dependent assertion to observed Ser/Thr activity, without declaring that a cyclin could never interact.
The seeded UniProt products are named 1/O76039-2 (displayed 960 aa) and 2/O76039-1 (alternative 1030 aa). The exact source file and product records remain unchanged. Primary transcript nomenclature differs: PMID:27315173, “hCDKL5_1 is the most abundant isoform expressed within the central nervous system (CNS)” also describes hCDKL5_5/115-kDa adult-testis expression. PMID:21748340(https://pubmed.ncbi.nlm.nih.gov/21748340/) independently identifies the 107-kDa alternative C terminus in its accessible abstract. The current UniProt tissue-specificity sentence attaches brain predominance to its named isoform 2, which does not align straightforwardly with this primary transcript/product distinction. Primary expression evidence is therefore cited explicitly rather than transferring product numbers between naming systems. Functional equivalence of both human products is not assumed.
PMID:30266824(https://pubmed.ncbi.nlm.nih.gov/30266824/) combines engineered human CDKL5 kinase-domain chemistry, rodent substrate proteins, mouse knockout neurons and human patient-derived neurons. The full Methods identify human 1–352 kinase, F89A/C152A analog-sensitive mutations, K42A negative control and a full-length 107-kDa construct. Phosphorylation-dependent MAP1S microtubule binding and MAP1S knockdown rescue connect the kinase to microtubule dynamics. Human EB2 phosphorylation corroborates conservation; rodent phosphosite numbers are not silently relabeled as human residues. These experiments do not make CDKL5 a microtubule motor or an intrinsic polymerase.
PMID:22922712(https://pubmed.ncbi.nlm.nih.gov/22922712/) supplies mouse hippocampal postsynaptic localization, silencing/kinase-dead and rescue experiments, plus patient-derived human neuronal spine/contact abnormalities. Its mEPSC recordings are rodent experiments. The NGL-1 fragment kinase assay uses human NGL-1 and Myc-CDKL5 immunocomplexes from HEK293T, rather than a demonstrated purified single-enzyme preparation. The reported NGL-1 mechanism is retained as a candidate route to synaptic organization, without overstating its direct substrate certainty. The later substrate paper questions aspects of earlier NGL-1/HDAC4 substrate assignments; this does not erase the primary neuronal organization phenotype or mean the 2012 paper lacked an in vitro phosphorylation assay.
PMID:20861382(https://pmc.ncbi.nlm.nih.gov/articles/PMC6633570/) was read through indexed primary Methods and selected Results despite challenge responses to local requests. It studies rat CDKL5a, a 934-aa construct, in rat neurons/electroporation and heterologous cells. Donor records match migration, dendrite/axon development, Rac signaling, binding and localization. The C-terminal rat fragment pulls Rac1 from lysate; this is not purified binary binding or a Rac GEF/GAP assay. PDGF-dependent ruffle colocalization keeps its original colocalizes_with qualifier. The rat and human products are not assumed identical.
PMID:34605059(https://pubmed.ncbi.nlm.nih.gov/34605059/) demonstrates CDKL5-dependent ELOA phosphorylation and damage-associated RNAPII repression, including endogenous depletion and WT/kinase-dead rescue. This supports nuclear activity beyond localization. Engineered NLS-tagged recruitment and phosphoproteomic experiments remain explicitly distinguished. MS2 reporter and endogenous-gene qRT-PCR readouts do not isolate the inhibited transcriptional stage, so the elongation-specific assertion is semantically generalized to GO:0000122. This is not asserted to be a formal direct-parent relation, and an elongation role is not disproven. Recruitment to DNA damage and effects on silencing do not by themselves justify a NEW DNA-repair process.
PMID:28740074(https://pubmed.ncbi.nlm.nih.gov/28740074/) provides endogenous human HeLa/MRC-5 centrosome/midbody localization and depletion experiments, murine CDKL5 rescue in human cells, timed prometaphase/cytokinesis and H2B-S14D rescue. The authors describe PMID:28740074, “CDKL5 contributes to cytokinesis by regulating HIPK2/H2B activity at the midbody”. Spindle defects are linked to centrosome accumulation and cytokinesis; the paper does not establish direct H2B phosphorylation by CDKL5 or microtubule nucleation by CDKL5. These data independently support contextual cell-cycle regulation even though the electronic cyclin-dependent MF premise needs refinement.
The human arm of PMID:29420175(https://pubmed.ncbi.nlm.nih.gov/29420175/) uses GFP-CDKL5 (NM_003159.2) in RPE-1 cells. It observes basal-body/tip/centrosome localization and inhibited ciliogenesis upon overexpression, reversed by depletion. Worm CDKL-1 length-control experiments are distinct. Native human physiological cilium necessity and the responsible substrate remain unresolved; the ciliary terms are retained outside the neuronal core.
The bounded machine-field extract preserves exact target pairs, interaction IDs, assay hosts, features, negative flags, query URLs, source hashes, selected ortholog records and HPA location fields. Eight positive IntAct records support the four 2017/2020/2021 assertions. HuRI uses human participants in yeast two-hybrid assays. BioPlex uses tagged CDKL5 bait in HEK293T complex assays; spoke expansion does not prove a binary interface. The GOLGA2 13–1002 feature belongs to that partner. No interaction partner's function is assigned to CDKL5.
The 2011 CDKL5–MLH1 assertion remains UNDECIDED: the main article is accessible, but the exact Table S2 target row/construct was not independently recovered. Later HuRI confirms the pair separately. Retention of the four supported generic-binding rows follows the fixed campaign instruction, which retains supported generic binding as non-core when no better molecular activity is established.
All publication files are authentic normal-fetch outputs; six files already on pinned main were reused byte-for-byte, including the existing screen caches and PANTHER files. No machine source was edited. Four supplementary normal PMID caches were fetched for neuronal mechanism and isoforms. PMID:15917271, PMID:16935860, PMID:27466189 and PMID:21748340 were read as authentic abstracts; source-specific full methods were not claimed. Their general kinase/isoform conclusions are supported independently by inspected full primary studies. PMID:20861382 has an abstract-only cache but separately accessed indexed full primary Methods/Results. PMID:21653829 has full main text but unresolved target supplementary evidence. The status of a local cache alone does not determine external primary access.
The genuine Falcon provider attempt timed out at its configured 90-second bound; the explicit perplexity-lite fallback returned HTTP 401 insufficient quota. No provider report was produced or fabricated. The orphaned task-specific Falcon client was stopped after the wrapper exited and retained the output pipe. This manual evidence review and notes are the research output. Independent bounded peers checked kinase specificity/cyclin scope and transcription/nuclear assay scope. All action decisions were then reviewed individually by the whole-gene reviewer.
This appended assessment supersedes the opening statement that only the neuronal setting is represented in the core. The original journal is preserved. A second Ser/Thr-kinase core now describes the human cultured-cell response to induced DNA breaks, without generalizing it to constitutive neuronal transcription or DNA repair.
The full CDKL5 primary study, PMID:34605059 supports an elongation model more strongly than the earlier short rationale conveyed. ELOA is both an elongation factor and a component of an RNAPII-directed ubiquitin ligase; CDKL5 phosphorylates its Ser311 site. DRB prevents both proteins from accumulating at laser-damage tracks (Figures 3 and 6; associated Results). These observations make an elongation mechanism plausible. They do not show that CDKL5-dependent phosphorylation changes ELOA's elongation or ligase activity, or that Ser311 is necessary or sufficient for repression. ELOA recruitment itself persists after CDKL5 disruption and for the S311A mutant. The Discussion explicitly leaves the relevant substrate mechanism unresolved.
The assay order matters. In Methods, MS2 foci, FokI is induced for 3 hours before doxycycline is added for another 3 hours to induce reporter transcription. The separate FokI recruitment experiment activates transcription before damage. Thus recruitment's requirement for ongoing transcription does not identify the stage inhibited in the silencing assay. Figure 8 establishes CDKL5-dependent repression of endogenous SLCO5A1/RYR2, with WT but not kinase-dead K42R rescue and cleavage/repair controls, but RNA abundance does not isolate initiation, promoter escape or elongation.
The reviewer also correctly identifies prior elongating-polymerase biology. In Shanbhag et al., PMID:20550933, selected full Results/Figure 3 show an ATM-dependent decrease in Ser2-phosphorylated RNAPII while the total-polymerase proxy persists at the reporter. The authentic abstract of Pankotai et al., PMID:22343725 reports DNA-PK-dependent inhibition of both elongation and reinitiation. These primary studies establish the surrounding mechanism; neither identifies which transcriptional step CDKL5 controls. Full Pankotai Methods/figures were not read.
Row 31 therefore retains MODIFY to GO:0000122, with the original GO:0034244/IDA/PMID:34605059/involved_in fields untouched. Fresh QuickGO ancestry excludes GO:0000122. The original term has GO:0034243 and GO:0032785 as its immediate is_a parents; the replacement is a lateral change to the demonstrated RNAPII repression, not a claim about formal parentage. The elongation model remains an explicit open mechanistic question, not a disproved function.
The nuclear core distinguishes two established observations: ELOA-site phosphorylation, including endogenous phosphosite loss after CDKL5 disruption/depletion, and kinase-dependent transcriptional repression. It does not join them into an untested Ser311-to-repression causal chain. The existing nucleus is_active_in IBA row is promoted from non-core to ACCEPT on this independently established catalytic role, preserving its source fields and the historical-node reconstruction limitation. NLS-engineered recruitment/screening experiments remain distinct from the endogenous functional readouts; this is not direct imaging of endogenous CDKL5.
Two smaller clarifications preserve their actions. The GO_REF:0000108 cell-cycle inference becomes stale after refining its GO:0004693 premise, whereas independent PMID:28740074 experiments support the retained contextual process. Re-curation from that experimental source is now a question. Also, the live GO:0106310 is_a closure does not include GO:0004674. Serine and Ser/Thr activities remain compatible because both substrate chemistries were observed, without asserting the suggested ontology child relation.
The source evidence JSON now includes the exact finite ontology results, query URLs and response hashes. Both additional background caches were produced by the unchanged normal fetch tool; the existing source caches and all source assertions remain unchanged. No new literature quotation was introduced by this follow-up.