Focused Primary-Evidence Audit: Human NEK1 (Q96PY6) Uncertain Claims for GO Curation

Audit date: September 2026
Scope: Claim-by-claim re-scoping of six uncertain assertion clusters attached to human NEK1 (UniProt Q96PY6), with attention to the CFAP410 (C21ORF2) partnership, distinguishing direct human assays, ortholog studies, predictions, secondary discussion, and disease phenotypes.

Accessibility note. This audit was performed from verified PubMed abstracts (stored, verbatim) plus open-access full-text summaries. Where a specific figure/table/supplement was not directly opened, this is stated explicitly rather than inferred from absence. Verbatim passages below are taken from the papers' abstracts unless noted. Evidence classes are labeled throughout: [Direct human assay] / [Ortholog] / [Prediction] / [Secondary discussion] / [Disease phenotype].

Summary

This audit traced six clusters of uncertain NEK1 claims to their primary evidence, verified identifiers, and separated experiment from citation, human assay from ortholog, and demonstrated interaction from prediction. The headline conclusion is that several widely propagated NEK1 annotations are compound or mis-scoped, and correcting them materially changes what GO can defensibly assert.

The single most consequential correction concerns ATR/ATRIP Thr1989. NEK1 does not directly phosphorylate ATR or ATRIP at Thr1989. Liu et al. 2013 (PMID: 23345434) establish that T1989 is an ATR autophosphorylation site that NEK1 primes and stabilizes by maintaining ATRIP levels and ATR–ATRIP association. Summaries claiming an "ATRIP-T1989 NEK1 substrate" are doubly wrong: T1989 resides on ATR, not ATRIP, and it is autophosphorylation, not a NEK1-catalyzed event. The RAD54-Ser572 claim (Spies et al. 2016, PMID: 27264870) is genuinely disputed but not experimentally refuted in the peer-reviewed literature: the frequently-cited TLK1 paper (PMID: 37439356) tests entirely different residues (T41/T59/T700) and is therefore not a refutation; the only experimental non-replication is an unpublished bioRxiv preprint, and a December-2025 preprint reaffirms S572 while proposing a developmental NEK1/NEK3/NEK5 switch that could reconcile the conflict.

On localization, endogenous human NEK1 maps to the distal centriole tip throughout the cell cycle (June 2026 U-ExM study, PMID: 42265085), refining—not contradicting—older pericentriolar-material and basal-body descriptions, most of which relied on mouse/canine overexpression. The VDAC1-Ser193 phosphosite is real and supported by both cell-based and reconstituted in-vitro assays using human GST-VDAC1 and human-cell-derived NEK1, but its original mechanistic framing rests on the now-outdated assumption that VDAC1 is an obligatory component of the mitochondrial permeability transition pore. Finally, NEK1–CFAP410 is a genuine direct binary interaction via the NEK1 C-terminal domain, distinct from a non-obligatory NEK1–CFAP410–SPATA7 retinal association; and the neuronal importin-β1/α-tubulin phosphorylations are in-vitro-only, with physiological relevance inferred from phenotypes rather than demonstrated at the specific sites.

Claim-by-Claim Summary Table

Key Findings

Finding 1 — ATR/ATRIP Thr1989 is ATR autophosphorylation primed by NEK1, not a NEK1 substrate site

[Direct human assay] Liu et al. 2013 (PMID: 23345434) show that NEK1 associates with the ATR–ATRIP complex and is required to maintain ATRIP protein levels, ATR–ATRIP association, and basal ATR activity before DNA damage. In cells lacking NEK1, ATR fails to autophosphorylate at Thr1989. The verbatim abstract text states that "cells lacking Nek1 failed to efficiently phosphorylate multiple ATR substrates and support ATR autophosphorylation at threnine 1989, one of the earliest events during the ATR response" and that NEK1 "enhances the stability and activity of ATR-ATRIP before DNA damage, priming ATR-ATRIP for a robust DNA damage response."

The crucial curation point is that T1989 is explicitly described as ATR's own autophosphorylation, an event that NEK1 supports by stabilizing the complex. No assay in the paper demonstrates direct NEK1 phosphorylation of ATR or ATRIP at any residue. Consequently, any annotation stating that NEK1 phosphorylates "ATRIP-T1989" (or "ATR-T1989") as a direct substrate is a compound error and should be removed. The defensible GO annotation is a priming/stabilizing regulatory role on the ATR–ATRIP complex (protein stabilization / positive regulation of ATR signaling), not a kinase-substrate relationship at T1989.

Finding 2 — RAD54-Ser572 is disputed but not experimentally refuted in the peer-reviewed record

[Direct human assay] The positive claim originates from a single direct human study: Spies et al. 2016 (PMID: 27264870), which reports that "human Nek1 regulates homologous recombination (HR) by phosphorylating Rad54 at Ser572 in late G2 phase," supported by both cellular and in-vitro kinase assays and DR-GFP HR functional readouts.

The apparent refutation frequently attached to this claim—the TLK1/RAD54 paper, Ghosh et al. 2023 Nucleic Acids Research (PMID: 37439356)—does not test Ser572 at all. That paper shows "TLK1 phosphorylates RAD54 at three threonines (T41, T59 and T700)": a different kinase and different residues. Its rejection of S572 appears only in the Discussion as a written assertion, not as an experiment. Under the audit rule "do not treat PMID37439356 as refutation unless it actually tests this site," it must not be recorded as counterevidence to the phosphorylation event.

The chronology of the genuine dispute is summarized below.

The Löbrich preprint's proposed developmental switch (adult NEK1 vs. embryonic NEK3/NEK5 both converging on RAD54-S572) is a plausible reconciliation of the reproducibility conflict, because the non-replication used mouse NEK1-KO and immortalized lines that may reflect a different developmental/cell-context regime. Both the non-replication and the reaffirmation are non-peer-reviewed. It is essential to note that Gregorczyk 2023 (PMID: 37188479) mentions RAD54 only as [Secondary discussion] of a prior NEK1 target and performed no independent RAD54-Ser572 replication. Curation implication: retain the S572 phosphorylation as supported by one direct human study, flag it as disputed, and do not cite PMID37439356 as refutation.

Finding 3 — Endogenous human NEK1 localizes to the distal centriole tip; older PCM/basal-body claims are stage/species/overexpression artifacts

[Direct human assay, endogenous] The June 2026 study Streubel et al., Nat Commun 17:7330 (PMID: 42265085; doi 10.1038/s41467-026-73560-9; PMC13402741; identifier verified) mapped endogenous NEK1 by ultrastructure expansion microscopy (U-ExM) in human RPE1 cells and found it at the distal tip of centrioles throughout the cell cycle, colocalizing with CP110 and Cep97 and interacting with Cep97 and Cep78. NEK1 loss causes centriolar-microtubule hyperelongation from the distal tip without displacing CP110–Cep97, defining a parallel branch of centriolar-MT length control. NEK1 is removed from the basal body during ciliogenesis in a Cep78-dependent manner.

This reconciles the earlier literature once method, species, and cell-cycle stage are specified:

Curation implication: the precise, defensible localization for human NEK1 is the distal end of the centriole (with a basal-body presence during ciliogenesis that is actively removed). Pericentriolar-material/centrosome annotations derived from mouse/canine overexpression should be de-emphasized or qualified with the appropriate evidence and species tags.

Finding 4 — VDAC1-Ser193 is a genuine human phosphosite with in-vitro channel-gating consequences, but rests on an outdated permeability-transition-pore assumption

[Direct human substrate; human-cell kinase] Chen, Craigen & Riley 2009 (PMID: 19158487, Cell Cycle 8:257–267) report that "Nek1 regulates the pathway to mitochondrial cell death through phosphorylation of voltage dependent anion channel 1 (VDAC1) on serine 193." Interaction was shown by yeast two-hybrid, GST pull-down, and reciprocal IP; a portion of NEK1 localizes to mitochondria; the non-phosphorylatable S193A mutant causes cell death, while the phosphomimetic S193E transiently rescues mitochondrial membrane potential. Chen et al. 2010 (PMID: 20230784, BBRC) add reconstituted atomic-force-microscopy and liposome cytochrome-c conductance assays showing that NEK1 phosphorylation closes VDAC1 (S193A stays open; S193E stays closed, and VDAC1 "closes and prevents cytochrome c efflux when phosphorylated by Nek1").

Two caveats govern curation. First, both papers frame VDAC1 as "a key component of the mitochondrial permeability transition pore"—an assumption refuted since 2007 (Baines et al. showed VDAC1/2/3 are dispensable for MPT). The phosphosite and gating data survive this correction, but the mechanistic narrative ("controls the PTP") does not. Second, on species: the substrate was cloned human GST-VDAC1, and Ser193 sits inside the barrel wall accessible from the cytoplasmic side. The kinase source was corrected during this audit: it was not purified bacterial NEK1 but rather (i) immune complexes of endogenous NEK1 immunoprecipitated from human cells and (ii) myc-/GFP-tagged NEK1 expressed in human cells (HK2, HEK293), phosphorylating bacterially-purified human GST-VDAC1, with a kinase-dead K33A specificity control that could not phosphorylate GST-VDAC1, plus heat-inactivated and irrelevant-substrate (GST-Hzwint-1) negatives. The endogenous-IP kinase is unambiguously human; the exact species of the cloned tagged cDNA is not stated. Independent corroboration of the NEK1→VDAC1 axis comes from a separate De Benedetti-lab study (Cell Cycle ~2020, PMC7028156).

Curation implication: annotate NEK1 phosphorylation of human VDAC1-Ser193 (supported; human substrate; human-cell kinase; K33A control) while dropping obligatory-PTP language and the "regulation of permeability transition pore" mechanistic gloss.

Finding 5 — NEK1–CFAP410 is a direct binary interaction via the NEK1 C-terminal domain, distinct from a non-obligatory NEK1–CFAP410–SPATA7 association

[Direct human interaction + Prediction] Gregorczyk et al. 2023 (Rouse lab), Life Science Alliance 6(7):e202201740 (PMID: 37188479; doi 10.26508/lsa.202201740) functionally characterize the association of C21ORF2 (= CFAP410) with NEK1 in human cells. They report that endogenous NEK1 and C21ORF2 form a tight complex; a C21ORF2-interaction domain (CID) at the NEK1 C-terminus (~aa 1208–1286) is necessary for the association; pathogenic mutations there disrupt the complex; and AlphaFold predicts an extended interface between the C21ORF2 leucine-rich-repeat domain and the NEK1-CID. Functionally, NEK1 mutations that inhibit kinase activity or weaken C21ORF2 association compromise ciliogenesis, and C21ORF2—like NEK1—is required for homologous recombination (traffic-light-reporter HR assay in U2OS). A 2025 Frontiers structural study modelled HsCFAP410-NTD(1–150) + HsNEK1-CTD(1208–1286) by AlphaFold-multimer. This is therefore a direct, mapped binary interaction, with the atomic interface being a prediction, not an experimental structure.

[Ortholog/association] Separately, Wheway et al. 2015 (PMID: 26167768, Nat Cell Biol) place NEK1, CFAP410, and SPATA7 together in a retinal ciliopathy module: endogenous NEK1 pulls down C21ORF2 from bovine retinal lysates; GST-SPATA7 recovers endogenous C21ORF2 and NEK1 from bovine retina; the three colocalize at the basal body in human TERT-RPE1 cells; and nek1-null zebrafish ciliopathy features are partially rescued by human wild-type C21ORF2. The abstract text confirms "Biochemical approaches place C21orf2 within key ciliopathy-associated protein modules." These are association/colocalization/cross-species functional data; SPATA7's recovery of NEK1 could be bridged through C21ORF2, and no assay demonstrates a simultaneous obligatory heterotrimer. Eblimit 2015 (PMID 25398945) is a separate SPATA7–RPGRIP1 study and is not evidence for a NEK1 trimer.

Curation implication: annotate a direct NEK1–CFAP410 binary interaction (NEK1-CTD/CID ↔ CFAP410-NTD/LRR; interface predicted) and, separately, a non-obligatory NEK1–CFAP410–SPATA7 basal-body association (bovine + human, association-level evidence). Do not assert an obligatory trimer.

Finding 6 — Neuronal importin-β1 and α-tubulin are in-vitro-only NEK1 substrates; physiological relevance is inferred, not demonstrated at the sites

[In-vitro substrate; in-vivo phenotype] Mann et al. 2023, Science Advances (PMID: 37585529; doi 10.1126/sciadv.adi5548) report that the NEK1 interactome/expression proteomics is enriched for microtubule cytoskeleton and nucleocytoplasmic transport, and that α-tubulin and importin-β1 are phosphorylated by NEK1 in vitro. The in-vivo/cellular data (Drosophila motor phenotypes; iPSC-motor-neuron knockdown, kinase inhibition, and patient mutation) show disrupted microtubule homeostasis and nuclear import, rescued by microtubule-stabilizing drugs. Critically, the rescue operates via general MT stabilization and does not demonstrate that the specific in-vitro phosphosites are required for the phenotype.

Curation implication: annotate α-tubulin and importin-β1 as in-vitro NEK1 substrates with an in-vitro qualifier; the physiological "regulates nuclear import / MT homeostasis" role should be annotated from the in-vivo phenotype evidence, kept logically separate from the specific phosphorylation events.

Mechanistic Model / Interpretation

NEK1 is a large, multi-compartment NIMA-family kinase whose annotations have accreted across two decades, multiple species, and shifting mechanistic assumptions. The audit resolves the confusion by binning each claim on two axes: directness of evidence (direct human assay → ortholog → prediction → secondary discussion → disease phenotype) and compartment/pathway.

          HUMAN NEK1 (Q96PY6)
                  |
   ┌──────────────┬──────────────┼───────────────┬────────────────┐
   │              │              │               │                │
 DNA REPAIR    ATR PRIMING   CENTRIOLE/CILIA   MITOCHONDRIA    NEURONAL/ALS
   │              │              │               │                │
 RAD54-S572   ATR-ATRIP      distal centriole  VDAC1-S193     importin-β1,
 (disputed;   (T1989 = ATR   tip (endogenous,  (human sub-    α-tubulin
  1 direct    autophos.,     human, U-ExM);    strate; human   (in-vitro only;
  human study; NEK1 primes/  CFAP410 direct    kinase; K33A    phenotype rescue
  not refuted) stabilizes,   binary partner    control; PTP    via MT stabiliz.,
      NOT a substrate (CID/NTD, interface framing outdated) not site-specific)
      at T1989)      predicted); non-
            obligatory SPATA7
            association

Three interpretive threads unify the findings:

1. Citation ≠ experiment. Two of the most damaging errors in the current annotation set arise from treating a written claim as an experiment: the "ATRIP-T1989 substrate" (which mis-reads an autophosphorylation-priming result) and the "PMID37439356 refutes S572" attribution (which mis-reads a Discussion sentence in a paper that tested different residues). Both should be corrected on evidentiary, not biological, grounds.

2. Species and expression regime drive apparent contradictions. The localization "conflict" dissolves once endogenous human U-ExM data are separated from mouse/canine overexpression; the S572 reproducibility conflict may dissolve along a developmental (adult vs. embryonic) axis with paralog redundancy (NEK1 vs. NEK3/NEK5).

3. Outdated framing outlives valid data. The VDAC1-S193 phosphosite and gating data remain valid even though the permeability-transition-pore rationale that motivated them is obsolete. Curators should preserve the molecular event and retire the pathway gloss.

Evidence Base

Non-peer-reviewed primary sources referenced (must be flagged as preprints):
- De Benedetti lab, Evidence that Nek1 does not phosphorylate Rad54-S572, bioRxiv doi 10.1101/2022.04.01.486731 — the only experimental non-replication of S572; remains unpublished.
- Löbrich lab, Nek family members regulate Rad54 during homologous recombination in developing mice, bioRxiv doi 10.64898/2025.12.19.695536 (posted 2025-12-19; the 10.64898 prefix is bioRxiv's new DOI prefix, verified real, not a typo of 10.1101) — reaffirms S572 with an adult-NEK1 / embryonic-NEK3–NEK5 developmental switch.

Limitations and Knowledge Gaps

• Full-text/supplement access. Several conclusions rest on abstracts, open-access summaries, and verified snippets rather than direct inspection of every figure and supplementary methods table. Specifically, the Gregorczyk 2023 figures/supplement were not directly inspected in this environment, and the exact recombinant-cDNA species used in the VDAC1 tagged-kinase assays is not stated in the accessible text. These are explicitly identified inaccessible data, not inferred absences.
• S572 remains formally unsettled. The two most decision-relevant primary sources (the non-replication and the reaffirmation) are both preprints. No peer-reviewed experiment has directly re-tested S572 with modern phospho-specific tools since Spies 2016. The developmental-switch hypothesis is attractive but unvalidated.
• VDAC1 kinase-cDNA species. The endogenous-IP kinase is unambiguously human, but the tagged-NEK1 construct's species of origin is not documented in accessible text; this is a minor residual uncertainty.
• NEK1–CFAP410 interface is predicted, not solved. The atomic interface is AlphaFold-multimer only; no experimental structure exists. The mapped domains (NEK1-CID ↔ CFAP410-NTD/LRR) and mutational-disruption data are solid, but the interface geometry is a prediction.
• In-vitro substrates (importin-β1, α-tubulin). No evidence connects the specific in-vitro phosphosites to the in-vivo phenotype; the physiological requirement of these phosphorylations is unestablished.

Proposed Follow-up Experiments / Actions

Immediate curation actions (defensible now):
1. Remove any annotation stating NEK1 directly phosphorylates ATR or ATRIP at Thr1989. Replace with a priming/stabilization of ATR–ATRIP regulatory annotation (Liu 2013, direct human).
2. Retain but flag NEK1→RAD54-Ser572 as a single-source, disputed direct human finding (Spies 2016). Do not cite PMID37439356 as counterevidence. Note both preprints in the evidence trail.
3. Update NEK1 cellular-component to distal centriole / centriole distal end (Streubel 2026, endogenous human, U-ExM), with basal-body presence during ciliogenesis. Add species/method qualifiers to legacy PCM/centrosome annotations derived from overexpression.
4. Annotate NEK1 phosphorylation of human VDAC1-Ser193 (Chen 2009/2010; human substrate; human-cell kinase; K33A control) and strip obligatory-permeability-transition-pore language from the mechanistic description.
5. Annotate a direct NEK1–CFAP410 binary interaction (NEK1-CTD/CID ↔ CFAP410-NTD/LRR; interface predicted) and, separately, a non-obligatory NEK1–CFAP410–SPATA7 basal-body association (Wheway 2015). Do not assert an obligatory trimer.
6. Annotate α-tubulin and importin-β1 as in-vitro NEK1 substrates with an in-vitro qualifier; keep the neuronal MT-homeostasis/nuclear-import role as a separate phenotype-based annotation (Mann 2023).

Experimental follow-ups (to resolve residual uncertainty):
- Obtain and inspect the full text/supplement of Gregorczyk 2023 and the VDAC1 papers to confirm construct species and interface-mapping details.
- Peer-reviewed re-test of RAD54-S572 with phospho-specific antibodies/MS across adult vs. embryonic and human vs. mouse systems to test the developmental-switch hypothesis directly.
- Experimental structure (cryo-EM/X-ray) or crosslinking-MS of the NEK1-CID · CFAP410-NTD complex to replace the AlphaFold prediction.
- Site-directed (phospho-null/phosphomimetic) rescue of importin-β1/α-tubulin in NEK1-deficient motor neurons to test whether the specific in-vitro sites are physiologically required.
- Monitor the publication status of both preprints (doi 10.1101/2022.04.01.486731 and doi 10.64898/2025.12.19.695536) for peer-reviewed resolution.

Prepared for GO curation, September 2026. All PMIDs and DOIs in this report were verified during the audit; preprint sources are explicitly marked as non-peer-reviewed.

Artifacts

• OpenScientist final report
• OpenScientist final report

Citations

1. PMID:23345434
2. PMID:27264870
3. PMID:37439356
4. PMID:42265085
5. PMID:37188479
6. PMID:21211617
7. PMID:26167768
8. PMID:19158487
9. PMID:20230784
10. PMID:37585529