Structure of a ¹²⁹Xe-Cryptophane Biosensor Complexed with Human Carbonic Anhydrase II

The α-carbonic anhydrases (CAs) are zinc metalloenzymes that catalyze the reversible hydration of CO₂ in forming HCO₃^-. The active site of an α-CA contains a catalytically essential Zn²⁺ coordinated by three histidine residues at the bottom of a 15 Å deep cleft, and the tightest binding CA inhibitors developed to date contain a sulfonamide moiety that coordinates to Zn²⁺ as a sulfonamidate anion.¹ Notably, human isozyme II (CAII) is an ideal model system for exploring new inhibitor designs, some of which can be exploited in biosensing applications.^2-4 Here, CAII is utilized for the structure-based design of a xenon (¹²⁹Xe) biosensor for potential use as a magnetic resonance imaging (MRI) contrast agent.

The ¹²⁹Xe isotope has a spin-1/2 nucleus, a >200-ppm chemical shift window in water, and a natural isotopic abundance of 26% (commercially available up to 86%), which makes it an appealing biomolecular probe for MRI. Moreover, ¹²⁹Xe can be laser polarized to enhance MRI signals ∼10,000-fold.⁵ Although current in vivo MRI applications are limited to functional lung imaging through the diffusion of Xe gas,⁶ the encapsulation of ¹²⁹ Xe within a cryptophane cage (K_D ≈ 30 μM at 37 °C in phosphate-buffered solution)⁷ facilitates its use as a biosensor that can be targeted to specific proteins using an appropriate affinity tag.^8,9 For example, racemic biosensor 1 (Figure 1a) has been designed to bind to the CA isozymes (K_D = 60 ± 20 nM against CAII in solution), and yields a distinctive ¹²⁹Xe-MRI spectrum when bound to CAII.¹⁰ Here, we report the X-ray crystal structure of the CAII-1-Xe complex at 1.70 Å resolution.

Figure 1.

(a) The M_oM_o and P_oP_o enantiomers of the cryptophane-A-derived CA biosensor. The benzenesulfonamide moiety serves as an affinity tag that targets the Zn²⁺ ion, and the R₁ substituents contain triazole propionate moieties that enhance aqueous solubility. (b) Stereoview of a simulated annealing omit map showing 1-M_oM_o (blue) and 1-P_oP_o (red) bound in the active site (1.9 σ contour, teal). A Bijvoet difference Fourier map (2.0 σ, black) confirms the encapsulation of Xe (yellow). Coordination interactions with Zn²⁺ (grey sphere) are indicated by dotted lines.

For structure determination, CAII was overexpressed in E. coli and purified as described,¹¹ then incubated with a two-fold excess of 1, concentrated to 10 mg/mL, and crystallized by the hanging drop vapor diffusion method. Crystals were cryoprotected in 15% glycerol and subsequently pressurized under 20 atm Xe for 30 min prior to flash cooling and X-ray data collection. The structure was refined to final R_work and R_free values of 0.23 and 0.25, respectively.

Biosensor 1 coordinates to the active site Zn²⁺ ion as the sulfonamidate anion, displacing the zinc-bound hydroxide ion of the native enzyme as previously observed in other complexes of CAII with benzenesulfonamide derivatives.^1,2,12 The crystallographic occupancies of 1 and Zn²⁺ are refined at 0.5. It is unusual to observe diminished Zn²⁺ occupancy in a CAII-inhibitor complex, but the molecular origins of this effect are not clear.

The encapsulation of Xe within the cryptophane cage of 1 is confirmed by inspection of the Bijvoet difference Fourier map calculated from anomalous scattering data (Figures 1b and S1.) X-ray diffraction data was collected at a wavelength λ = 0.9795 Å, which is far from the Xe L_I edge of 2.27 Å.¹³ Nevertheless, the anomalous scattering component f” is 3.4 e^- for Xe, so the anomalous signal is still prominent at the wavelength of data collection. A second Xe binding site is observed in a hydrophobic pocket defined by A116, L148, V218, L157, V223 and F226 (Figure S2). The crystallographic occupancies of these Xe sites refine to 0.50 and 0.37, respectively. Anomalous scattering peaks are absent from crystals not subject to Xe pressurization.

Notably, 1 contains a chiral axis and the electron density map reveals the binding of equal populations of both enantiomers, (each refined with an occupancy of 0.25; Figure 1)^14-16 Overall, the binding of 1 does not cause any significant structural changes in the active site, and the root-mean-square deviation is 0.34 Å for 256 Cα atoms between the current structure and the unliganded enzyme (PDB 2CBA).¹⁷

The total surface area of 1 is ∼1500 Ų, of which ∼500 Ų becomes solvent inaccessible due to contacts of 1 within the active site cleft of CAII designated molecule I in Figure 2. The surrounding CAII molecules in the unit cell (molecules II-IV), sequester an additional ∼540 Ų of the surface area of 1 from solvent. Some structural changes are observed near the outer rim of the active site cleft where the cryptophane binds. The most notable change is observed for Q136, which rotates ∼180° to make van der Waals contacts with the cryptophane and the symmetry-related cryptophane bound to molecule III in the crystal lattice. Other residues at the active site rim of molecule I that make close contacts with the cryptophane are G132 and P202. Additional structural changes in the crystal lattice result from the binding of 1 to molecule I: in molecule II, H36 rotates ∼90° to make a van der Waals contact with the cage, and Q137 of molecule III rotates ∼90° to donate a hydrogen bond to an ether oxygen atom of 1.

Figure 2.

The unit cell of CAII crystals in space group C2 contains four molecules: I (x,y,z), II (x+1/2,y+1/2,z), III (-x,y,-z) and IV (-x+1/2,y+1/2,-z). The binding of 1 in the active site cleft of molecule I buries ∼500 Ų. Crystal contacts bury an additional 540 Ų of the surface of 1 as follows: 270 Ų with molecule III, and 240 Ų and 30 Ų with the front and back faces of molecule II, respectively. Molecule IV does not contact 1 bound to molecule I.

Although the pendant propionates appear to be more disordered than the cryptophane and are characterized by correspondingly weaker electron density, a hydrogen bond between a propionate moiety and Q53 of molecule II is observed. The relative dearth of strong cryptophane-protein interactions may explain why the affinity of 1 measured by ITC is only slightly better than that measured for the parent triazole-benzenesulfonamide lacking the cryptophane (K_D = 100 ± 10 nM).¹⁰

Limited hydrogen bond interactions between CAII and the cryptophane moiety of 1 may be advantageous for the use of cryptophanes as ¹²⁹Xe biosensors. Translational and rotational freedom, the consequence of a flexible linker between the cryptophane and the benzensulfonamide, could allow the cage to reorient rapidly in situ, independently of the protein, to result in decreased correlation times and narrower line widths that increase the sensitivity of ¹²⁹Xe NMR measurements in solution.⁸

In conclusion, this work reveals the first experimentally determined structure showing how an encapsulated ¹²⁹Xe atom can be specifically directed to a biomedically relevant protein target. The possible implications for cancer diagnosis are profound, given that CA isozymes IX and XII are overexpressed on the surface of certain cancer cells.¹⁸ Moreover, a search of the Protein Data Bank reveals that with its molecular mass of 1554, the 1-Xe complex is one of the largest synthetic organic ligands ever cocrystallized with a protein. Thus, this work demonstrates the feasibility of preparing crystalline complexes between proteins and nonbiological, nanometer-scale ligands.¹⁹