Elucidating the role of metal ions in carbonic anhydrase catalysis

The study addresses why substitution of the native metal ion in metalloenzymes with chemically similar, non-native metals often causes large changes in catalytic activity. Carbonic anhydrase II (CA II), a highly efficient zinc metalloenzyme that catalyzes reversible hydration of CO2, serves as the model system. CA II has a single metal-binding site and a well-characterized active site with hydrophobic and hydrophilic regions separated by an entrance conduit. The native Zn2+ acts as a Lewis acid, lowering the pKa of bound water to generate a zinc-bound hydroxide at physiological pH. Prior work has examined metal binding and specificity through coordination stereochemistry and principles such as HSAB and the Irving–Williams series, yet the atomic-level role of metal ions in catalysis remains unclear. Non-native metal substitutions (e.g., Co2+, Ni2+, Cu2+) lead to drastic activity differences despite similar general chemical properties. This study aims to connect metal coordination geometry and electrostatic effects to specific catalytic steps and water network organization in CA II.

Background literature establishes the ubiquity and functional importance of metalloproteins and the roles of metal ions in catalysis (acidity/electrophilicity/nucleophilicity). Metal-binding affinity and specificity have been interpreted via coordination stereochemistry and semi-empirical frameworks (hard and soft acids and bases; Irving–Williams series). CA II is a canonical zinc enzyme with well-resolved structures and extensively studied kinetics and mechanism, including roles of hydrophobic/hydrophilic regions and proton transfer via water networks and His64. Prior structural studies of metal-substituted CA II suggested that altered coordination geometries may influence catalysis, but direct evidence linking these geometries to catalytic intermediates and activity changes was lacking. Reports also noted that metal substitution can induce alternative activities (e.g., copper enabling nitrite reduction).

• System: Human carbonic anhydrase II (CA II) with native Zn2+ and substitutions by Co2+, Ni2+, and Cu2+; apo-CA II as control.
• Activity context: These metals yield ~100% (Zn2+), ~50% (Co2+), ~2% (Ni2+), and 0% (Cu2+) of native catalytic activity.
• Sample preparation: Native Zn-CA II expressed in E. coli BL21 (DE3) pLysS; purified via sulfonamide affinity chromatography. Apo-CA II prepared by zinc chelation (pyridine-2,6-dicarboxylic acid; MOPS pH 7.0) and further purification; zinc loss verified by esterase assay and crystallography.
• Metal substitution: Apo-CA II crystals soaked in metal salt solutions (100 mM CoCl2, 100 mM NiCl2, 10 mM CuCl2; 1.3 M sodium citrate, 50 mM Tris-HCl pH 7.8). pH 11.0 conditions obtained with CAPS buffer.
• Crystallization: Hanging drop vapor diffusion with 1.3 M sodium citrate, 50 mM Tris-HCl pH 7.8 at ~20 °C; crystals ~30×100×200 µm3.
• Cryo-trapping intermediates: High-pressure cryocooling under controlled CO2 pressures (0 and 20 atm) using HPC-201 system. Crystals soaked in cryo-solution with 35% glycerol, coated with mineral oil, pressurized with CO2 for ~5 min, then plunged into liquid nitrogen (77 K). Samples stored in LN2.
• X-ray data collection: PLS-II beamline, λ=0.9793 Å, 100 K, ADSC Q270 CCD, 1° oscillation, 360 images, crystal-to-detector 120 mm. Dose <5×10^5 Gy to avoid radiation damage.
• Data processing and refinement: HKL2000 for indexing/integration/scaling. Refinement with CCP4/REFMAC5; starting models PDB 5DSR and 5YUK. 5% Rfree set. Iterative manual rebuilding in COOT; anisotropic B-factors. Systematic occupancy refinement of His64 in/out conformations by generating 99 models with 1% increments; selecting occupancy minimizing R-factor. Water molecules refined based on density (≥1σ in 2Fo−Fc) with careful validation; transient waters retained when appropriate.
• Assays: Esterase activity (4-nitrophenyl acetate, monitored at 348 nm) to confirm apo status and reactivation by Zn2+ addition.
• Structural deposition: Multiple PDB entries provided for apo-, Zn-, Co-, Ni-, and Cu-CA II at 0 and 20 atm CO2 and at pH 7.8 and 11.0.

• Coordination geometries without CO2 pressurization: Zn- and Co-CA II (pH 11.0) show tetrahedral coordination (His94, His96, His119 plus water); Ni-CA II shows octahedral (three waters bound); Cu-CA II shows trigonal bipyramidal (two waters bound). Apo-CA II has water at the vacant site.
• Substrate/product binding:
  • Apo- and Zn-CA II at 20 atm CO2 show similar CO2 binding, indicating the protein scaffold alone positions CO2. In Zn-CA II, CO2 lies 2.9 Å from the Zn-bound water (WZn) in an optimal nucleophilic attack arrangement; the hypothetical position (2) is only 0.36 Å from the CO2 carbon, favoring attack. Product HCO3− in Zn-CA II likely binds monodentate, enabling weak interaction and fast dissociation.
  • Co-CA II at pH 11.0 exhibits dual occupancy at 20 atm CO2: ~50% CO2 (tetrahedral coordination) and ~50% HCO3− (octahedral coordination), with HCO3− bidentate and an extra water bound. At pH 7.8, HCO3− binds with full occupancy and octahedral geometry even without added CO2 (likely from ambient CO2), showing stronger product binding than Zn-CA II. HCO3− binding weakens as pH increases, consistent with deprotonation of Co2+-bound water (formation of hydroxide) facilitating product dissociation.
  • Ni-CA II at 20 atm CO2 maintains octahedral coordination with bidentate HCO3− and an additional water, similar to Co-CA II, but shows steric clash between one Ni-bound water and the CO2 binding position. The nucleophilic attack geometry is distorted: distance between position (2) and CO2 carbon is 1.55 Å (vs 0.36 Å for Zn), indicating less favorable alignment. HCO3− binding affinity is largely pH-insensitive; product displacement likely requires two incoming waters rather than deprotonation-triggered release.
  • Cu-CA II shows no clear CO2 or HCO3− density at 20 atm; the active site indicates severe steric hindrance from a Cu2+-bound water preventing proper CO2 orientation. Even with proper CO2 orientation, the spare Cu-bound water is ~3.9 Å from CO2 and the attack geometry is highly distorted; distance between position (2) and CO2 carbon is 2.93 Å. These features explain complete inactivity.
• Water network and electrostatic effects (long-range ~10 Å):
  • Apo- and Zn-CA II share the primary proton transfer network (WZn→W1→W2→His64) organized by the protein scaffold. Upon CO2 binding, W1 disappears and an intermediate water W1′ appears, with Zn2+ fine-tuning EC water dynamics to stabilize W1′ and facilitate coupling between proton transfer and solvent exchange. Zn-CA II shows modified dynamics of W2 and His64 relative to apo, optimizing proton transfer and water/substrate/product exchange.
  • Co-CA II resembles Zn-CA II at 0 atm CO2, but upon CO2/HCO3− binding and octahedral expansion, the EC water dynamics differ and W1′ is less stabilized. Proton transfer likely proceeds via an altered pathway (WCo,octa→W2→His64) while HCO3− remains bound; subsequent deprotonation to hydroxide promotes product dissociation and restoration of tetrahedral geometry.
  • Ni-CA II’s persistent octahedral geometry eliminates W1 due to steric hindrance, alters W2 dynamics, and destabilizes W1′, reflecting a different electrostatic environment. Proton transfer likely follows WNi′→WNi→W2→His64.
  • Cu-CA II retains a well-defined potential proton transfer path (WCu→W1→W2→His64) and similar W2/His64 dynamics to Zn-CA II, indicating that inactivity stems from poor substrate binding and attack geometry rather than disrupted proton transfer network.
• Activity correlation: Native Zn-CA II (tetrahedral) is most efficient; Co-CA II (~50% activity) undergoes tetrahedral→octahedral conversion affecting product release and proton transfer; Ni-CA II (~2% activity) maintains octahedral geometry, causing steric hindrance and inefficient attack, strong product binding, and altered water dynamics; Cu-CA II (0% activity) exhibits trigonal bipyramidal coordination with severe steric conflicts preventing effective substrate binding and nucleophilic attack.

The findings directly link metal ion coordination geometry and long-range electrostatic effects to each catalytic stage in CA II: substrate binding, nucleophilic attack, product binding and displacement, and proton transfer. The protein scaffold alone positions CO2 and establishes a baseline water network for proton transfer, but the metal ion is essential for generating and retaining the active hydroxide and for fine-tuning solvent networks over ~10 Å distances. Tetrahedral Zn2+ optimizes both attack geometry and dynamic water connectivity, accounting for high catalytic efficiency. Co2+ supports substrate conversion in tetrahedral geometry but expands to octahedral upon product formation, strengthening product binding and necessitating a modified proton transfer pathway to trigger product release. Ni2+ enforces octahedral coordination throughout, introducing steric hindrance to substrate binding, distorting the attack geometry, stabilizing strong bidentate product binding, and perturbing water dynamics, all reducing efficiency. Cu2+ provides a clear example where the coordination geometry creates insurmountable steric barriers and poor attack geometry, rendering the enzyme inactive despite an intact proton transfer network. These results resolve why chemically similar metal substitutions yield dramatically different activities and highlight how coordination geometry and electrostatics govern metalloenzyme catalysis.

This work provides structural evidence that metal ions in CA II influence catalysis beyond simple Lewis acid effects, through their characteristic coordination geometries and long-range electrostatic modulation of active-site water networks. Tetrahedral Zn2+ enables optimal nucleophilic attack and efficient solvent-coupled proton transfer and product exchange; Co2+ partially preserves these features but undergoes octahedral expansion that affects product release; Ni2+ maintains octahedral coordination that hinders multiple catalytic steps; Cu2+ imposes steric and geometric constraints that prevent catalysis. These insights offer direct constraints for theoretical and computational modeling of metalloenzymes and inform drug discovery targeting metalloenzymes, engineering of natural enzymes, de novo metalloenzyme design, and supramolecular catalyst development. Future studies could explore broader metal series, time-resolved measurements across catalytic cycles, and targeted mutations to disentangle protein versus metal contributions to water network dynamics.