AIGR Gene Hypothesis Deep Research — Final Report

ComK (BACSU, UniProt P40396): ATP binding (GO:0005524) computational prediction

Gene: comK / Competence transcription factor · Organism: Bacillus subtilis 168 (BACSU, NCBITaxon:224308) · UniProt: P40396
Predicted term under review: ATP binding (GO:0005524), from BioReason-Pro SFT
Focus type: computational_prediction · Reference context: doi:10.64898/2026.03.19.712954

Summary

The BioReason-Pro SFT prediction of ATP binding (GO:0005524) for Bacillus subtilis ComK (P40396) is refuted. ComK is a compact, 192-amino-acid, single-domain sequence-specific DNA-binding transcription factor — the master regulator of genetic competence development — and it contains no ATP-binding motif or nucleotide-binding fold. Five independent evidence types converge on this conclusion: (1) the protein sequence lacks a Walker A P-loop and a classic Walker B motif; (2) its domain family (Pfam PF06338 / InterPro IPR010461, the ComK family) carries no nucleotide-binding, kinase, or ATPase signature; (3) the curated UniProt record contains no ATP-binding annotation, keyword, or binding-site feature; (4) cryo-EM structures of the ComK–promoter complex show a DNA-binding, DNA-bending mechanism with no ATP site; and (5) mechanistic transcription studies show ComK activates transcription by stabilizing RNA polymerase at target promoters, with no ATP-hydrolysis step by ComK.

The one genuine link between ComK and ATP is indirect and does not confer ATP binding on ComK: ComK is a substrate of the ATP-dependent MecA–ClpCP protease. In that complex, the ATPase activity resides entirely in the AAA+ chaperone ClpC, while ComK is the passive degradation target. Because ComK biology is routinely discussed alongside "the ATP-dependent ClpCP protease," a model that keys on co-occurrence can easily misattribute the ATPase's ATP-dependence to the substrate. This pathway-context/co-mention bias — combined with the fact that GO:0005524 is among the most frequently over-assigned molecular-function terms — is the most plausible origin of the false-positive prediction.

Recommended curation action: Do NOT add GO:0005524. Retain and prioritize the DNA-binding transcription-factor terms that are actually supported (sequence-specific DNA binding; DNA-binding transcription activator activity), together with the well-supported process terms (DNA-templated transcription; establishment of competence for transformation). The prediction should be recorded as an over-annotation to reject.

Key Findings

Finding 1 — ComK is a sequence-specific DNA-binding transcription factor with no ATP-binding motif (prediction REFUTED)

ComK's primary molecular function is sequence-specific DNA binding driving transcriptional activation, and there is no evidence of a nucleotide-binding fold anywhere in the 192-residue protein.

Sequence-level evidence. A direct scan of the P40396 sequence found no Walker A P-loop motif (consensus [AG]x4GK[ST], the glycine-rich loop that cradles the β/γ-phosphates of ATP/GTP) and no classic Walker B motif (the hhhhDE aspartate/glutamate that coordinates the catalytic Mg²⁺–water). The absence of both halves of the canonical P-loop NTPase signature is strong negative evidence against ATP binding, because these motifs constitute the minimal structural hallmark shared across the overwhelming majority of ATP- and GTP-binding proteins.

Domain-level evidence. ComK is the founding and essentially sole member of the ComK family (Pfam PF06338 / InterPro IPR010461), a family defined by the transcription-factor function of ComK and its orthologs across Bacilli. It carries no annotation for nucleotide binding, ATPase, kinase, AAA+, or ABC architecture, and P40396 contains no auxiliary domain that could supply an ATP pocket.

Database-level evidence. UniProt annotates P40396 with keywords DNA-binding, Activator, Repressor, Transcription regulation, and with GO terms GO:0003677 (DNA binding), GO:0006351 (DNA-templated transcription), and GO:0030420 (establishment of competence for transformation). There is no ATP-binding annotation (GO:0005524) in the curated record — the prediction conflicts with, rather than derives from, the curated evidence — and there is no EC number, cofactor, or nucleotide binding-site feature.

Structural evidence. Cryo-EM structures of the complex between ComK and its promoter DNA (PMID: 34016970) demonstrate that ComK functions through mechanical forces that alter DNA curvature — an allosteric, DNA-bending mechanism — not a catalytic nucleotide-dependent step. Verified quote: "Cryo-EM structures of the complex between ComK and its promoter demonstrate that this coupling is due to mechanical forces that alter DNA curvature." No ATP or nucleotide participates in the described functional mechanism.

Mechanistic evidence. ComK activates transcription by stabilizing RNA polymerase binding at the target promoter (PMID: 14762007; verified quote: "ComK stabilizes the binding of RNA polymerase to the comG promoter"). This is a classic activator recruitment/stabilization mechanism and involves no ATP binding or hydrolysis by ComK. Foundational work (PMID: 7783616; verified quote: "we demonstrate that ComK specifically binds to DNA fragments containing promoter and upstream sequences of the genes it affects") established the sequence-specific DNA-binding activity that defines ComK, and mutational studies of the K-box AT-boxes (PMID: 17468244) confirmed base-specific DNA recognition.

Together, five independent evidence types — sequence, domain, curated database, structure, and mechanism — all point away from ATP binding and toward DNA binding. This is the central finding that refutes the seed hypothesis.

Finding 2 — ComK's only genuine ATP connection is as a substrate of the ATP-dependent MecA–ClpCP protease (explains the false positive)

The prediction most plausibly arises from pathway-context/co-mention bias, not from any intrinsic ComK property. In B. subtilis, ComK protein levels are controlled by regulated proteolysis: a ternary complex of the adaptor protein MecA and the ATP-dependent protease ClpCP targets ComK for degradation (PMID: 12598648, PMID: 19767395).

The decisive point for curation is the division of labor within this complex:

• ClpC is an HSP100/Clp AAA+ ATPase; it binds and hydrolyzes ATP to power substrate unfolding and translocation. The ATP-binding activity belongs to ClpC.
• MecA is the adaptor that recognizes ComK, delivers it to ClpC, and stimulates ClpC's ATPase activity.
• ComK is the passive substrate. It contributes no ATP-binding or ATPase activity; it is the protein being degraded.

Verified quotes: "A complex of ClpC with the protease ClpP and the adaptor protein MecA also controls competence development by regulated proteolysis of the transcription factor ComK" (PMID: 12598648); "the degradation of the competence transcription factor ComK is mediated by a ternary complex involving the adaptor protein MecA and the ATP-dependent protease ClpCP" (PMID: 19767395).

Because ComK is routinely discussed in the same sentences as the "ATP-dependent protease ClpCP," a text- or co-occurrence-influenced model can mis-attribute the ATP-dependence of the degradation machinery to ComK itself. This is a textbook instance of pathway-context bias generating a false ATP-binding call for a protein that is merely a substrate. Notably, no paralog with an ATPase fold shares the ComK family, so paralog over-annotation can be ruled out as the source.

Mechanistic Model / Interpretation

The following schematic separates what ComK actually does from the ATP-dependent process in which ComK is a substrate, clarifying why GO:0005524 is a misassignment.

   ComK's ACTUAL molecular function (no ATP involved)
   ───────────────────────────────────────────────────
 ComK (192 aa, PF06338)
        │
     sequence-specific DNA binding
(K-box: paired AT-boxes, AAAA-N5-TTTT)
        │
        ▼
┌───────────────────────────────────┐
│  bends / curves promoter DNA       │  (cryo-EM, PMID 34016970)
│  stabilizes RNA polymerase binding │  (PMID 14762007)
└───────────────────────────────────┘
        │
        ▼
  activation of the K-regulon → competence
 (GO:0030420, GO:0006351)

   NO Walker A / Walker B / P-loop  →  NO ATP binding by ComK


   The ATP link is EXTERNAL to ComK (ComK = substrate)
   ───────────────────────────────────────────────────
      MecA (adaptor) ── recognizes ──► ComK (SUBSTRATE)
                          │
ClpC (AAA+ ATPase) ── ATP hydrolysis ──► unfolds / translocates
                          │
ClpP (protease) ─────────────────► degrades ComK

      ATP is bound & hydrolyzed by ClpC, NOT by ComK
      (PMID 12598648, PMID 19767395)

Interpretation. The two findings form a single coherent narrative. ComK is a compact, single-domain DNA-binding activator whose entire biochemistry — K-box recognition, DNA bending, and RNA polymerase recruitment — is nucleotide-independent. Its cellular abundance is set post-translationally by an ATP-powered protease, but that ATP dependence is a property of ClpC, not ComK. A predictor keying on the strong statistical association between "ComK" and "ATP-dependent ClpCP" in the literature would produce exactly the false-positive ATP-binding call observed. Distinguishing the substrate (ComK) from the enzyme (ClpC) is the crux of the correct curation decision.

Comparison table: expected features of an ATP-binding protein vs. observed for ComK

Evidence Base / Evidence Matrix

GO Curation Implications

Lead requiring curator verification:

1. Do NOT add GO:0005524 (ATP binding). The molecular-function prediction is a false positive. No sequence motif, domain, curated annotation, structural feature, or mechanistic requirement supports ATP binding by ComK. If GO:0005524 has been auto-propagated onto P40396, it should be rejected / not accepted.

2. Retain and prioritize the informative Molecular Function terms that are actually supported:

3. GO:0003677 — DNA binding (already curated; supported by PMID: 7783616).

4. GO:0043565 — sequence-specific DNA binding and GO:0001216 — DNA-binding transcription activator activity as the precise MF terms (K-box specificity, PMID: 17468244; RNAP stabilization, PMID: 14762007). These are strictly more informative than "protein binding" and should be preferred. ComK also acts as a repressor in some contexts.

5. Retain the Biological Process terms: GO:0006351 (DNA-templated transcription) / GO:0006355 (regulation of transcription) and GO:0030420 (establishment of competence for transformation).

6. Frame the ATP link correctly, if annotated at all: ComK's relationship to ATP is as a substrate of an ATP-dependent protease (ClpCP). Any ATP-related annotation belongs to ClpC, not ComK. This relationship should not become an ATP-binding MF term on ComK.

Summary GO decision table:

Mechanistic Scope

Immediate molecular function under test: whether ComK directly binds ATP (GO:0005524). The answer is no. ComK's direct, primary molecular activity is sequence-specific binding to K-box DNA elements (paired AT-boxes) and, through that binding, DNA bending and stabilization of RNA polymerase at target promoters.

What must be kept separate (not ComK's direct molecular function):
- Downstream developmental outcome: establishment of genetic competence, DNA uptake, and homologous recombination — consequences of ComK's transcriptional activity, not evidence for ATP binding.
- Pathway context: ATP-dependent proteolysis of ComK by MecA–ClpCP — an external regulatory process in which ComK is the substrate and ClpC is the ATPase.
- Upstream regulation: DegU binding to the comK promoter (PMID: 15598897) governs comK expression, unrelated to any ComK ATP activity.

The ATP-binding hypothesis fails precisely because it conflates a pathway-level ATP dependence (proteolysis) with a gene-product-level molecular function (nucleotide binding).

Conflicts and Alternatives

• Substrate-vs-enzyme confusion (most likely artifact): The strongest ComK–ATP link in the literature is that ComK is a substrate of the ATP-dependent MecA–ClpCP protease (PMID: 12598648, PMID: 19767395). Attributing ClpC's ATP dependence to ComK is the error.
• Frequency / prior bias: GO:0005524 is one of the most common MF terms; a model with a strong prior can over-assign it in the absence of any real signal.
• Paralog / family confusion — ruled out: ComK is the sole defining member of PF06338; there is no ATP-binding paralog within the family that could seed annotation transfer.
• Organism-specific differences — none: ComK's role as a competence transcription factor is conserved across Bacilli; no ortholog is reported to have acquired a nucleotide-binding domain.
• In vitro-only artifact — none: No in vitro study reports ATP binding by ComK; the relevant assays (DNA binding, transcription activation) are ATP-independent for ComK.
• Database carry-over — absent: UniProt P40396 does not carry GO:0005524, so the prediction contradicts, rather than inherits from, the curated record.

All conflicts point the same direction: away from ATP binding by ComK.

Limitations and Knowledge Gaps

1. Motif scan scope. The Walker A/B scan used canonical consensus patterns; degenerate or atypical nucleotide pockets can occasionally escape simple motif matching. Why it matters: a hidden atypical site would be missed by motif search alone. Resolution: the gap is largely closed by the DNA-bound cryo-EM structure and by the ComK-family domain assignment, neither of which reveals a nucleotide pocket; a full HMMER/InterProScan and structure-based pocket analysis (e.g., on the AlphaFold model) would formally confirm.
2. Direct negative biochemical assay. No published experiment explicitly reports the absence of ATP binding by ComK (negative results are rarely published). Why it matters: absence of ATP binding is inferred from motif/domain/structure/mechanism rather than a dedicated binding assay. Resolution: an ITC, DSF/thermal-shift, DRaCALA, or ATP-agarose pulldown assay would provide a direct negative control.
3. Structure interpretation is literature-based. The cryo-EM conclusion is drawn from the published report (PMID: 34016970); the raw coordinates were not re-analyzed here for ligand density (none expected). Resolution: inspect the deposited PDB/EMDB entry for any nucleotide ligand.
4. Provenance of the prediction. The exact features BioReason-Pro SFT used are unknown. Why it matters: confirming co-mention/frequency bias would strengthen the "reject" recommendation. Resolution: inspect model attributions if available.

None of these gaps materially weakens the refutation; they are avenues for formal confirmation.

Discriminating Tests / Proposed Follow-up Experiments

Concrete, prioritized actions to definitively distinguish ATP binding from the established DNA-binding function:

1. ITC or thermal-shift (DSF) with ATP on purified ComK (± Mg²⁺). Expected if hypothesis true: measurable K_d / thermal stabilization. Predicted outcome: no binding — a clean direct negative control. Highest-value, low-cost test.
2. Inspect the ComK–DNA cryo-EM structure / PDB (PMID: 34016970) for any bound nucleotide density or P-loop-like pocket. Expected: none.
3. Structure-based pocket/ligand-site prediction on an AlphaFold model of P40396 and superposition against known P-loop NTPases. Expected: a DNA-binding surface, not an ATP pocket.
4. Full domain/HMM re-scan (InterProScan, HMMER vs. Pfam-A). Expected: PF06338-only architecture; no AAA+/P-loop/kinase domain.
5. ATP-agarose or DRaCALA binding assay on purified ComK. Predicted: negative — formally closes the biochemical gap.
6. Attribution audit of the BioReason-Pro prediction: test whether the ATP call tracks co-mention with ClpCP/MecA text, confirming frequency/pathway-context bias as the mechanism of the false positive.

Curation Leads (require curator verification)

• Action: Reject the ATP binding (GO:0005524) prediction for P40396; do not add. Record as computational over-annotation driven by co-mention with the ATP-dependent ClpCP protease and general frequency bias.
• Candidate MF terms to retain/add (more informative than "protein binding"):
• GO:0003677 DNA binding (retain) — snippet to verify: "we demonstrate that ComK specifically binds to DNA fragments containing promoter and upstream sequences of the genes it affects" (PMID: 7783616).
• GO:0043565 sequence-specific DNA binding / GO:0001216 DNA-binding transcription activator activity — snippets to verify: "Cryo-EM structures of the complex between ComK and its promoter demonstrate that this coupling is due to mechanical forces that alter DNA curvature" (PMID: 34016970); "ComK stabilizes the binding of RNA polymerase to the comG promoter" (PMID: 14762007).
• BP terms to retain: GO:0006351 / GO:0006355 (transcription / regulation of transcription), GO:0030420 (establishment of competence).
• Note for the record (optional): ComK is a substrate of the ATP-dependent MecA–ClpCP protease — snippets to verify: "A complex of ClpC with the protease ClpP and the adaptor protein MecA also controls competence development by regulated proteolysis of the transcription factor ComK" (PMID: 12598648); "the degradation of the competence transcription factor ComK is mediated by a ternary complex involving the adaptor protein MecA and the ATP-dependent protease ClpCP" (PMID: 19767395). This explains the ATP association without implying ATP binding by ComK.
• Suggested curator question: Confirm whether the ComK cryo-EM/PDB entry shows any bound nucleotide (expected: none); and whether GO:0005524 was propagated from an electronic pipeline.
• Suggested experiment: ATP-agarose / DRaCALA / ITC assay on purified ComK to formally close the gap (expected negative).

Conclusion

The ATP-binding (GO:0005524) prediction for Bacillus subtilis ComK (P40396) is refuted. ComK is a 192-amino-acid, single-domain (ComK family, PF06338/IPR010461) sequence-specific DNA-binding transcription factor with no Walker A/B motifs or nucleotide-binding fold. Sequence, domain, curated-database, structural (cryo-EM, PMID: 34016970), and mechanistic (PMID: 14762007) evidence unanimously support DNA binding and RNA polymerase stabilization rather than ATP binding. The only genuine ATP link is that ComK is a substrate of the ATP-dependent MecA–ClpCP protease, in which the ATPase is ClpC — the most plausible source of the frequency/pathway-context bias behind this misassignment. GO:0005524 should not be added; the informative, supported terms are sequence-specific DNA binding and DNA-binding transcription activator activity.

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist comK ATP evidence matrix
• OpenScientist comK GO decision table