DnaK (P0A6Y8) Zinc-Ion-Binding Prediction — Focused Curation Report
Gene: DnaK (Hsp70 chaperone) · Escherichia coli K-12 · UniProt P0A6Y8 Prediction under review: BioReason-Pro SFT → zinc ion binding (GO:0008270) Focus type: computational_prediction · Slug: prediction-zinc-binding Reference context: doi:10.64898/2026.03.19.712954
Executive Judgment
Verdict: Over-annotated / weakly supported — treat as a non-core, likely artifactual assignment (effectively refuted as a molecular function).
DnaK does not possess a bona fide zinc-coordinating center. The prediction is not wholly baseless — there is a single existing IDA GO annotation (GO:0008270) on DnaK — but that annotation rests entirely on one low-specificity, abundance-biased proteomic Zn-blot screen (11985624) and is contradicted by DnaK's sequence and structure:
- DnaK has only one cysteine (Cys15) in 638 residues and no CXXC / C4 zinc-finger motif — it is structurally incapable of forming a classical cysteine-based zinc site.
- 0 of 63 DnaK PDB entries contain a bound zinc ion, despite extensive structural characterization of both the NBD and SBD.
- The genuine zinc-binding component of the DnaK/DnaJ/GrpE system is the co-chaperone DnaJ (P08622), which has a CR-type zinc finger coordinating 2 Zn²⁺ via 8 cysteines. The prediction is best explained as paralog/system confusion plus database carry-over.
Recommended lead: do not curate GO:0008270 as a DnaK function; if retained at all, downgrade to a low-confidence, in-vitro-only/adventitious note, not a mechanistic MF.
Most important caveat: an existing curated IDA annotation does exist (EcoliWiki, 11985624), so this is a case of challenging a weak legacy annotation rather than a term with zero support. The judgment rests on the low specificity of that assay and the absence of any structural/sequence basis.
Evidence Matrix
| Citation | Evidence type | Stance | Claim tested | Key finding | Context | Confidence / limitations |
|---|---|---|---|---|---|---|
| UniProt P0A6Y8 (computed here) | Sequence/computational | Refutes | DnaK has a Cys-based zinc motif | Only 1 Cys (Cys15); 0 CXXC motifs; no Metal-binding/Zinc-finger feature | E. coli DnaK, 638 aa | High; sequence is definitive that no canonical C4/C2H2 Zn finger exists |
| RCSB (63 entries, queried here) | Structural | Refutes | Zinc is present in DnaK structures | 0/63 DnaK PDB entries contain a ZN ligand | NBD + SBD crystal/NMR structures | High; large structural sample, all zinc-free |
| 11985624 (Katayama 2002) | Direct assay (proteomic ⁶⁵Zn-blot) | Qualifies (sole support) | DnaK binds Zn²⁺ in vitro | DnaK "newly identified" among 9 hits alongside AckA, GlyA, TktA/B, Tsf, ribosomal proteins | 2D-gel Zn-blot of total E. coli lysate | Low specificity; abundance-biased; many hits lack Zn-finger motifs |
| QuickGO annotation record (queried here) | Database | Qualifies | Provenance of DnaK Zn annotation | Only GO:0008270 on DnaK = IDA/EcoliWiki/11985624; identical ref also used for DnaJ | Annotation metadata | High; shows single-source, carry-over pattern |
| 15683252 (Shi 2005) | Structural/biochemical | Competing (DnaJ) | Which system member binds Zn | DnaJ has two C4-type zinc fingers (C144/147/161/164/183/186/197/200) binding 2 Zn²⁺ | E. coli DnaJ | High; localizes zinc function to DnaJ, not DnaK |
| 8662861 (Banecki 1996) | Direct assay (AAS/HgS titration) | Competing (DnaJ) | DnaJ zinc stoichiometry | Two Zn²⁺ per DnaJ monomer | E. coli DnaJ | High; establishes DnaJ as the zinc-binding partner |
| 23708608 (Qi 2013); 34453889; 33950017 | Structural | Refutes (by omission) | DnaK core MF is Zn-dependent | Full-length ATP-bound and domain structures: ATPase NBD + peptide SBD; no zinc site | E. coli DnaK | High; mechanism is ATP/Mg²⁺/K⁺- and peptide-driven, not zinc |
| InterPro (queried Iter 2) | Structural/evolutionary | Refutes | DnaK has a zinc-finger domain | Only Hsp70 domains (IPR012725, PF00012, NBD/SBD/PBD); no zinc-finger/metal domain | E. coli DnaK | High; comprehensive signature database |
| Ortholog Cys census (queried Iter 2) | Evolutionary | Refutes | Zinc binding conserved in Hsp70 | DnaK orthologs carry 0–5 Cys, 0 CXXC; M. tuberculosis DnaK has 0 cysteines | Hsp70 family (bacteria/yeast/human) | High; a zero-Cys ortholog cannot form a Cys zinc site |
GO Curation Implications
- GO:0008270 (zinc ion binding, MF): Lead — do not propagate / recommend removal or NOT-qualifier consideration for DnaK. The single IDA source is a low-specificity proteomic screen; there is no motif, no structural zinc, and no mechanistic role. If curator policy forbids removing an IDA, at minimum it should not be treated as a core molecular function and should carry an explicit caveat that it derives from a whole-proteome Zn-blot with probable non-specific binding.
- Do not substitute a vaguer term (e.g., "metal ion binding" or "protein binding") — no informative metal-binding term is supported.
- Well-supported core MF terms to prefer/retain instead: ATP binding (GO:0005524), ATP hydrolysis activity (GO:0016887), ADP binding (GO:0043531), unfolded protein binding (GO:0051082), ATP-dependent protein folding chaperone (GO:0140662). These reflect DnaK's actual activities.
- Paralog note for curators: the zinc-ion-binding annotation is biologically correct for DnaJ (P08622), and the shared 11985624 reference suggests the DnaK entry is a co-annotation carry-over.
Mechanistic Scope
The immediate molecular function being tested is direct coordination of a Zn²⁺ ion by DnaK. DnaK's actual direct activities are: (i) ATP binding and hydrolysis in the N-terminal nucleotide-binding domain (actin/hexokinase fold; uses Mg²⁺ and K⁺, not Zn²⁺), and (ii) binding of exposed hydrophobic segments of unfolded clients in the C-terminal substrate-binding domain, allosterically coupled to the nucleotide state. No step of this cycle requires or involves a structural zinc. Any zinc signal observed in a lysate blot is a downstream in vitro observation on an abundant protein, not a mechanistic feature of the chaperone cycle.
Conflicts and Alternatives
- Paralog confusion (most likely): DnaJ, the obligate co-chaperone in the DnaK/DnaJ/GrpE system, genuinely binds 2 Zn²⁺ via a CR-type zinc finger. Zinc binding is a real property of the system but resides in DnaJ.
- Database carry-over / frequency bias: the same reference (11985624) underlies the IDA on both DnaK and DnaJ; a model trained on such annotations would plausibly transfer "zinc ion binding" onto the more famous/abundant DnaK.
- Assay artifact: ⁶⁵Zn-blotting of 2D gels captures many abundant, acidic, or surface-exposed proteins non-specifically; the paper's own hit list (kinases, transketolase, ribosomal proteins, EF-Ts) is enriched for proteins with no canonical zinc-finger, signaling low specificity.
- No organism/isoform escape hatch: DnaK is a single-copy, well-conserved bacterial Hsp70; no isoform harbors extra cysteines that could form a zinc site.
- Negative literature check (Iter 3): targeted PubMed searches for a functional DnaK zinc/metal-binding site or a copper/zinc redox role returned no primary evidence (the only zinc-finger chaperone hit was thioredoxin-2, unrelated to DnaK). No study proposes zinc as mechanistically required for DnaK. The UniProt BP term "stress response to copper ion" (GO:1990169) reflects a physiological stress response, not direct metal coordination by DnaK.
Knowledge Gaps
- Is the Katayama Zn signal specific or adventitious? Checked: sequence (1 Cys), structure (0/63 PDB), and hit-list composition all argue non-specific. Matters because it is the sole primary support. Resolve with: purified DnaK ICP-MS/atomic-absorption zinc stoichiometry and competition/specificity controls (as were done for DnaJ).
- Does any DnaK conformer transiently bind Zn functionally? Checked: allosteric-cycle structures show no zinc site. Matters only if a regulatory zinc were proposed. Resolve with: metal-content analysis across ADP/ATP states.
- What exactly does EcoliWiki's IDA claim? Checked provenance (11985624 only). Matters for whether removal vs. caveat is appropriate. Resolve with: read full Katayama methods to confirm DnaK's measured affinity/specificity class.
Discriminating Tests
- ICP-MS / atomic absorption on purified recombinant DnaK (apo vs. Zn-loaded) — quantify Zn:protein stoichiometry; expect ~0 for a genuine non-binder, ~2 for DnaJ control.
- Cys15→Ser mutant + metal blot — a genuine Cys-based site would lose signal; a non-specific surface interaction would not.
- Side-by-side ⁶⁵Zn-blot of DnaK vs DnaJ vs a bona fide C4 zinc-finger control with EDTA/competitor titration to grade specificity.
- AlphaFold3/metal-aware docking or MIB/CHED metal-site prediction on P0A6Y8 — expect no high-confidence zinc site (bioinformatic corroboration).
Curation Leads (require curator verification)
- Action change: Flag GO:0008270 on DnaK as not a core function; propose removal or downgrade with an explicit provenance caveat. Do not replace with "metal ion binding" or "protein binding."
- Candidate references to verify (exact snippets):
- 11985624 — "nine zinc-binding proteins were newly identified including: acetate kinase (AckA), DnaK, serine hydroxymethyltransferase (GlyA)…" (verify this is the sole basis and note low specificity).
- 15683252 — "which coordinate with two Zn(II) ions to form an unusual topology of two C4-type zinc fingers…" (confirms zinc role belongs to DnaJ).
- 8662861 — "two Zn(II) ions interact with each monomer of DnaJ." (DnaJ stoichiometry).
- Suggested curator questions: (1) Is a single whole-proteome Zn-blot sufficient IDA support for a core MF given contradicting sequence/structure? (2) Should the DnaK/DnaJ shared reference trigger a carry-over review?
- Suggested experiment: purified-DnaK ICP-MS zinc stoichiometry with Cys15 mutant control.
- Positive terms to keep: ATP binding, ATP hydrolysis activity, unfolded protein binding, ATP-dependent protein folding chaperone.
Artifacts (computed provenance): artifacts/evidence_matrix.csv, artifacts/hsp70_cysteine_census.csv.
Provenance: All computed results (UniProt feature/GO parse, cysteine census, RCSB zinc-ligand query returning 0/63, QuickGO annotation provenance) were generated by executed code in Iteration 1; primary literature accessed via PubMed. No results were fabricated; the RCSB query returned HTTP 204 (empty) for DnaK-with-zinc.