FLG (P20930) — Transition Metal Ion Binding (GO:0046914): Hypothesis Review

Focus type: function_assignment · existing_annotations[6].function_hypothesis
Seed hypothesis: FLG has transition metal ion binding (GO:0046914)
Term/evidence context: GO:0046914 "transition metal ion binding", IEA, GO_REF:0000002 (InterPro2GO)
Gene: human FLG / Filaggrin / UniProt P20930 (profilaggrin, 4,061 aa)

Summary

The IEA annotation of FLG to transition metal ion binding (GO:0046914) is an over-annotation — a family-level InterPro2GO carry-over rather than a member-specific, experimentally supported molecular function of the filaggrin gene product. It should be treated as non-core and is a strong candidate for removal or explicit flagging as unsupported in the gene review. The annotation traces to a single InterPro signature, IPR034325 "S-100", which is the only signature on P20930 mapped to GO:0046914 via the automated InterPro2GO pipeline (GO_REF:0000002). That entry's own description hedges — "Many S100 proteins have been found to bind transition metals, such as Zn²⁺ and Cu²⁺" — a statement about the family as a whole that is then applied uniformly to every S100-domain-containing member regardless of individual ligand chemistry.

Three independent lines of direct and comparative evidence converge to refute a real metal function. First, the crystal structure of the profilaggrin N-terminal fused-type S100 domain (PDB 4PCW, 2.2 Å; PMID: 25760235) was solved with bound calcium ions and functions through hydrophobic dimer-interface pockets that mediate protein–protein interactions (annexin II, stratifin/14-3-3σ, HSP27); a HETATM audit of the deposited coordinates finds 8 calcium atoms and zero transition-metal atoms. Second, a residue-level audit shows the domain lacks the ligand chemistry that all characterized metal-binding S100 proteins use: only 2 histidines (His60, His65) and zero cysteines in residues 1–91, versus the canonical His₃Asp/His₄–His₆ or Cys-cluster motifs required. Third, a geometric analysis confirms the two histidines lie 8–15 Å from any calcium site and cannot cluster to build a metal motif.

Finally, a comparative annotation audit demonstrates the artifact directly: all six human S100-fused-type structural proteins (FLG, TCHH, HRNR, RPTN, CRNN, FLG2) uniformly carry GO:0046914 as IEA:InterPro with no member-specific metal evidence, whereas the genuine transition-metal-binding S100 proteins (S100A7, S100A8, S100A9, S100B) instead carry the specific, experimentally-backed child term GO:0008270 (zinc ion binding) with NAS/TAS/IDA evidence. The most important caveat is that no study has directly assayed the profilaggrin S100 domain for Zn/Cu binding, so this is a strong structure-plus-sequence-plus-annotation inference rather than a direct negative experiment. The well-supported molecular function of this domain is calcium ion binding (GO:0005509) plus calcium-dependent protein binding.

Verdict: Over-annotated (refuted as a direct, core function).

Key Findings

Finding 1 — The GO:0046914 annotation is a single-signature, family-level InterPro over-annotation

The transition-metal-binding annotation on P20930 does not originate from any experiment on filaggrin or profilaggrin. It is produced automatically by the InterPro2GO mapping (GO_REF:0000002) from exactly one signature: IPR034325 "S-100", a domain-level entry that is the only signature on the protein linked to GO:0046914. The InterPro entry's own text makes a family-generalizing statement — "Many S100 proteins have been found to bind transition metals, such as Zn²⁺ and Cu²⁺" — which the pipeline applies to every protein containing an S-100 domain, without regard to whether that individual member retains the necessary ligand residues.

Against this generic inference stands the direct experimental characterization of the domain. The profilaggrin N-terminal fused-type S100 domain was crystallized at 2.2 Å (PDB 4PCW) with bound calcium ions (PMID: 25760235). Functionally, the domain operates as a dimer with hydrophobic pockets that mediate protein–protein interactions — its characterized partners are annexin II, stratifin (14-3-3σ), and HSP27. No zinc, copper, or other transition-metal binding was reported.

A computed sequence audit of the S100 domain (residues ~1–91) reinforces this: the domain contains only 2 histidines (His60, His65), both located in the EF-hand-2 calcium loop/helix IV region, and zero cysteines across the entire N-terminal 212 residues. Canonical S100 transition-metal sites require either a His₃Asp / His₄–His₆ cluster (the "His-Zn" group; PMID: 20306096, PMID: 23276281) or cysteine ligands (the "Cys-Zn" group; PMID: 20306096) at the dimer interface. Profilaggrin's S100 domain possesses neither. UniProt's curated binding-site residues for the domain (positions 62, 64, 66, 68, 73 = Asp, Asp, Asn, Lys, Glu) are the canonical EF-hand calcium ligands, not metal-cluster residues.

Finding 2 — Comparative paralog and structural audit confirms systematic family over-annotation

Two complementary audits confirm the annotation is a family artifact. A direct HETATM audit of PDB 4PCW finds 8 Ca atoms and 0 atoms of any transition metal (no Zn/Cu/Mn/Fe/Ni); the only heteroatoms are calcium, a PEG fragment (2PE), and ordered waters. The structure that defines this domain's biophysics contains no transition metal.

A paralog audit shows that all six human S100-fused-type proteins carry GO:0046914 as IEA:InterPro, and none carries member-specific metal evidence — whereas the bona fide transition-metal-binding S100 proteins do not use the generic parent term at all, but the specific experimental child term GO:0008270 (zinc ion binding):

Because GO:0008270 is an is_a child of GO:0046914, a specific experimental term supersedes the generic IEA parent under standard GO annotation practice. FLG displays only the generic IEA parent because it has no experimental metal annotation to specialize it — the fingerprint of automated propagation rather than curated function.

Finding 3 — Structural geometry seals it: the two histidines cannot form a metal site

A coordinate-level geometric analysis of PDB 4PCW (four chains in the asymmetric unit) tested whether the domain's histidines could plausibly build a transition-metal site. The only resolved histidines are His59 and His64 (PDB numbering; = UniProt His60/His65). Their imidazole nitrogen atoms lie 8–15 Å from the nearest bound Ca²⁺, confirming they are not part of the EF-hand calcium sites.

Testing for a shared metal-coordination cluster found no pair of distinct histidines within the ~3.5–5 Å required to jointly coordinate one metal, except a single marginal His59–His59 cross-contact at ~5.2 Å that is a crystal-packing/tetramer contact rather than a functional site. All His64 pairs are >6 Å apart. There is no cluster of ≥3 histidines available to construct the canonical His₃Asp or His₄ S100 transition-metal motif. Geometry, not just sequence composition, excludes a metal site.

Mechanistic Model / Interpretation

The evidence separates the real function of the domain from the artifact of its annotation:

   InterPro2GO pipeline (GO_REF:0000002, IEA)
 │
 ▼
   IPR034325 "S-100" domain signature
   ("Many S100 proteins bind transition metals...")
 │  applied uniformly to ALL members
     ┌───────────┼───────────┬───────────┬───────────┬──────────┐
     ▼           ▼           ▼           ▼           ▼          ▼
   FLG         TCHH        HRNR        RPTN        CRNN       FLG2     ← all inherit GO:0046914 (IEA)
  (P20930)                                                             ← none has member-specific metal evidence

   ── vs ──

   Genuine metal-binding S100s (S100A7 / A8 / A9 / B)
     → carry SPECIFIC experimental term GO:0008270 (zinc ion binding)
       via His3Asp / His4–His6 interfacial clusters or Cys ligands

What FLG's S100 domain actually does (direct evidence, PDB 4PCW):

  Profilaggrin N-terminal S100 domain
│
├── binds CALCIUM via EF-hand ligands (D62,D64,N66,K68,E73)  → GO:0005509
├── forms a homodimer with hydrophobic binding pockets
└── engages protein partners: annexin II, 14-3-3σ, HSP27,     → Ca²⁺-dependent
    keratin filaments (cooperates with B domain; PMID 32893105)  protein binding

  Ligand inventory of residues 1–91:  His × 2 (His60, His65), Cys × 0
  Requirement for an S100 metal site:  His3Asp / His4–His6 cluster OR Cys ligands
  ⇒ MISSING → no transition-metal site

The domain's biology is calcium-sensing and calcium-dependent protein scaffolding in granular-layer keratinocytes, supporting epidermal barrier function. The "transition metal ion binding" label is a downstream consequence of automated family annotation, not a molecular activity of the gene product — precisely the frequency-biased, paralog-driven over-annotation that hypothesis-level review is designed to catch.

Evidence Base

How the literature fits together. The one structural paper on the FLG gene product itself (PMID: 25760235) shows calcium, not transition metal, and a protein-interaction function — the direct refutation. The biochemical follow-up (PMID: 32893105) reinforces a calcium-dependent protein-scaffolding role. The calprotectin and S100-survey papers (PMID: 20306096, PMID: 23276281, PMID: 27541598, PMID: 23431180) define the exact His/Cys ligand chemistry that real S100 metal-binders use and that FLG's domain provably lacks. The calbindin paper (PMID: 18359862) is the only entry hinting that isolated His residues can coordinate Zn — a weak analogy that does not apply given FLG's dispersed, calcium-loop histidines.

GO Curation Implications

Lead (requires curator verification): remove or flag GO:0046914 on FLG as an unsupported InterPro2GO family over-annotation; do not promote it to a reviewed molecular-function annotation.

• The term is MF aspect but derives from a family-level InterPro signature whose GO mapping reflects that some S100 proteins bind Zn/Cu — not that filaggrin does.
• The same removal/flag rationale should be considered for the paralogous IEA annotations on TCHH, HRNR, RPTN, CRNN, and FLG2 — this is a systematic InterPro2GO family artifact, and correcting FLG alone leaves the family inconsistent. The mapping IPR034325 → GO:0046914 is itself a candidate for InterPro2GO refinement.

Mechanistic Scope

Immediate molecular function under test: direct coordination of a transition-metal ion (Zn²⁺/Cu²⁺/Mn²⁺/Fe) by the FLG gene product — necessarily via the N-terminal S100 domain, the only region with a candidate metal architecture.

• Direct gene-product activity that IS supported: Ca²⁺ binding by the N-terminal S100 (A) domain and calcium-modulated protein–protein interactions (annexin II, 14-3-3σ, HSP27, keratin) that regulate profilaggrin processing and cornified-envelope assembly.
• What GO:0046914 asserts but is NOT demonstrated: a folded, sequence-defined transition-metal coordination site on the profilaggrin gene product.
• Downstream / not the tested function: Filaggrin's canonical barrier role (proteolysis of profilaggrin into monomers, keratin aggregation, natural moisturizing factor generation) is a downstream processing pathway, not the S100 domain's metal chemistry. Serum zinc/selenium associations with atopic dermatitis severity (PMID: 36090737) are whole-organism micronutrient epidemiology, not evidence of FLG protein binding metal. Any weak, nonspecific chelation by histidine-rich degradation products would be a downstream metabolic property, not a molecular-function site — and is not the basis of the current IEA (which comes only from the S100 domain signature). The metal-relevant S100 domain exists only in profilaggrin; mature filaggrin repeat monomers do not contain it, so the annotation cannot even apply to the processed product.

Conflicts and Alternatives

• Database carry-over / frequency bias (primary alternative): GO:0046914 is an InterPro2GO generalization from the broad S-100 domain family. The transition-metal phenotype is concentrated in soluble, histidine-rich S100 paralogs (calprotectin S100A8/A9, S100A7, S100B), which drives the family mapping — classic over-annotation of a structural family member that lacks the site.
• Paralog/subfamily distinction: Profilaggrin is an S100 fused-type protein (SFTP) — a structural subgroup (with trichohyalin, hornerin, repetin, cornulin) whose S100 domain is specialized for hydrophobic-pocket protein binding (PMID: 32893105), not metal sequestration.
• Specific-term supersession pattern: True metal-binders carry the specific experimental term GO:0008270 (zinc ion binding), which suppresses the redundant generic IEA parent. FLG shows GO:0046914 only because it has no experimental metal term to supersede it — an artifact of the redundancy-filtering pipeline, not positive evidence.
• In-principle EF-hand His–Zn binding: PMID: 18359862 shows calbindin D28k can bind Zn²⁺ via His80. This raises a theoretical possibility, but FLG's two histidines sit in the EF-hand-2 calcium loop 8–15 Å apart from any partner, and calbindin is not S100-fused — a weak, non-physiological analogy.
• No conflicting positive evidence found: No primary study reporting Zn/Cu/Mn binding by the FLG gene product was located, and the deposited structure (4PCW) contains only Ca²⁺.

Limitations and Knowledge Gaps

1. No direct metal-binding assay of the profilaggrin S100 domain. Checked: PubMed and the 4PCW structure paper — only calcium was co-crystallized/tested. Matters because the verdict rests on inference from ligand absence rather than a direct negative experiment. Resolve with ITC/ICP-MS/XAS or a metal-competition fluorescence assay on the recombinant profilaggrin S100(AB) domain vs. Zn²⁺/Cu²⁺/Mn²⁺.
2. Solution vs. crystal behavior. Only the Ca-grown crystal is available; weak or transient transition-metal association could be missed. Resolve with metal-binding assays under varied conditions and metal-soaked XAS.
3. Behavior of the histidine-rich repeats toward metals. The full protein is globally histidine-rich, but the repeats have no S100 architecture and are not the basis of the IEA; any chelation would be nonspecific and downstream. Resolve with equilibrium dialysis/ICP-MS on a repeat construct (expected: weak, nonspecific).
4. Family-wide curation consistency and InterPro2GO provenance. Confirmed all six SFTP paralogs share the same IEA and that IPR034325 is the sole source. Matters because curators may prefer a mapping-level fix. Resolve by reviewing the IPR034325 → GO:0046914 mapping with InterPro/GOA.

Discriminating Tests

1. Recombinant profilaggrin S100(AB) metal titration (ITC or competition with a fluorescent Zn chelator such as FluoZin/Mag-fura-2) ± saturating Ca²⁺ — the single most decisive experiment. Predicts no high-affinity (µM–nM) Zn/Cu site.
2. Structural superposition of 4PCW against a His-Zn S100 (e.g., S100A7, S100B-Zn, calprotectin) to confirm the interfacial His₃Asp geometry is absent (partly done here by residue and geometry audits).
3. XAS/anomalous diffraction of the domain soaked in Zn²⁺/Cu²⁺ to test for any adventitious site.
4. Comparative InterPro2GO audit across all SFTP paralogs — uniform inheritance of GO:0046914 with no member-specific metal data confirms systematic family over-annotation.

Proposed Follow-up Actions / Curation Leads

All items are leads requiring curator verification.

• Action change (lead): Do not retain GO:0046914 as a reviewed MF for FLG; classify as non-core / uninformative IEA (remove or flag as unsupported by member-specific evidence). Extend the same review to TCHH, HRNR, RPTN, CRNN, FLG2.
• Candidate reference to verify — PMID: 25760235: snippet "we determined a 2.2 Å-resolution crystal structure of the N-terminal fused-type S100 domain of human profilaggrin with bound calcium ions" → supports GO:0005509 calcium ion binding (upgrade rationale) and refutes a transition-metal role.
• Candidate replacement/kept MF term: GO:0005509 calcium ion binding (well supported); optionally calcium-dependent protein binding for the annexin II / 14-3-3σ / HSP27 / keratin interactions (PMID: 25760235, PMID: 32893105). Avoid bare "protein binding."
• Mechanistic reference — PMID: 20306096: snippet defining that metal-binding S100s use "three histidines and an aspartic acid" (His-Zn) or cysteines (Cys-Zn) — the ligand set FLG's S100 domain lacks (0 Cys, 2 His).
• Suggested curator question: "Is IPR034325 → GO:0046914 an appropriate mapping to propagate to structural S100-fused-type family members that lack the interfacial His/Cys metal ligands?" (candidate for InterPro2GO refinement / NOT qualifier).
• Suggested experiment: recombinant profilaggrin S100(AB) Zn²⁺/Cu²⁺ ITC/ICP-MS (Discriminating Tests #1).

Provenance / Artifacts

Computed in this run (executed code + output captured in the iteration log):

• S100 metal-ligand residue audit — S100 domain res 1–91: 2 His (60, 65), 0 Cys; N-terminal 1–212: 0 Cys; EF-hand2 Ca ligands 62/64/66/68/73 = D,D,N,K,E.
• PDB 4PCW HETATM metal audit — 8× Ca, 0× Zn/Cu/Mn/Fe/Ni (ligands: CA, 2PE [PEG], HOH). The profilaggrin S100 crystal structure contains only calcium.
• SFTP paralog + control annotation audit — all 6 S100-fused-type paralogs carry GO:0046914 IEA:InterPro; S100A7/A8/A9/B carry GO:0008270 (zinc ion binding) via NAS/TAS/IDA instead.
• PDB 4PCW His-geometry test — resolved S100 histidines (His59, His64 in PDB numbering = UniProt His60/His65) lie 8–15 Å from calcium and do not cluster (His64···His64 all >6 Å; one marginal His59···His59 crystal contact at ~5.2 Å) — insufficient to build a canonical His3Asp/His4 site.
• InterPro→GO mapping check — IPR034325 "S-100" is the sole P20930 signature mapped to GO:0046914; IPR002048 (EF-hand) → GO:0005509; IPR003303 (Filaggrin) → GO:0005198.
• UniProt GO set for P20930 confirms both GO:0046914 and GO:0005509 are IEA:InterPro (no experimental MF metal annotation).

Data sources: UniProt REST (P20930 + 8 comparators), InterPro REST (entry/protein), RCSB PDB 4PCW, PubMed.

Artifacts

• OpenScientist final report
• OpenScientist final report