The effect of the disease-related mutation P237H on HCAII folding has been investigated previously, and it was found that this mutation led to a ∼7.3 kcal/mol decrease of HCAII stability [5]. However, sequence alignment analysis indicated that the 237th residue is not fully conserved in CAII, and Ala and Thr also appear in the CAII sequence from the other species (Figure 1). An unresolved question is whether the position 237 of CAII has any amino acid residue preference? To elucidate this problem, the mutation-induced stability changes were evaluated by the prediction of FoldX [20], a well-established method that has been successfully applied to the analyses of protein folding [14,21], protein design [22], protein-protein interactions [23,24], protein-DNA binding [25] and evolution [26] in a variety of proteins. The prediction was carried out using the standard procedures, and the “RepairPDB” command was performed before calculation to minimize the FoldX free energy for the WT structure at 25 °C.

As shown in Figure 2, all the mutations tested had small ΔΔG values around 1 kcal/mol, and the largest change in stability was found to be caused by the P- > I mutation with a value of 1.22 kcal/mol. These results suggest that according to the FoldX prediction, the substitution of Pro at position 237 by any of the other residues did not significantly affect HCAII stability. In other words, Pro237 contributed little to the overall stability of the protein. The large discrepancy between the experimental data (ΔΔG = 7.3 kcal/mol) and the prediction (ΔΔG = 1.03 kcal/mol) of the P237H mutation suggested that the role of Pro237 in HCAII might not be well evaluated by FoldX. In this case, it is necessary to determine the changes in Gibbs free energy by experimental methods. Five typical mutations (P237A, P237T, P237N, P237I, P237F) were chosen for further analysis by folding studies. The choice of P237A and P237T was due to the appearance of these two residues in the sequence of the other species (Figure 1). The other three mutations were chosen because according to the FoldX results shown in Figure 2, P237F was the most stable one among the mutants, while P237I and P237 N were the most unstable ones among the possible 20 natural amino acids.

HCAII_pwt, which contains a C206S mutation to avoid the interference of unexpected disulfide formation, was used in this study. Previous studies have shown that HCAII_pwt have indistinguishable folding and functional properties from the wild type protein [27–31]. All recombinant proteins could be successfully obtained in the soluble fraction when overexpressed in E. coli. The activities of the mutants were similar to HCAII_pwt, ranging from 87% to 99% of the activity of HCAII_pwt (Table 1). The effect of the mutations on HCAII structure was investigated by circular dichrosim (CD) (Figure 3) and intrinsic fluorescence (data not shown, see also Figure 4) experiments. The spectra of the mutants were almost superimposed with those of HCAII_pwt, suggesting that the mutations did not affect either the secondary or the tertiary structures of HCAII_pwt.

The HCAII sequence contains seven Trp residues distributed throughout the folded protein structure, thus the conformational changes of the proteins can be sensitively monitored by intrinsic Trp fluorescence. The emission maximum wavelength of the intrinsic Trp fluorescence was measured at each GdnHCl concentration tested, and the results are presented in Figure 4. Consistent with previous observations [5,27,28], the transition curves of all proteins were a three-state process with a molten globular intermediate state (I) appearing between the native (N) and the unfolded (U) state. By fitting the transition curves into the three-state model N↔I↔U, the Gibbs free energy of the two transitions was obtained, and the changes in stability (ΔΔG) were calculated accordingly for each mutant (Table 1). The ΔG values of HCAII_pwt were similar to those in literature [4,5], and the minor deviations might be caused by different experimental procedures.

Most mutations slightly destabilize both the N↔I and I↔U transitions except P- > A mutation seems to stabilize the N↔I transition. The ΔΔG values of folding were between 0.9 and 2.9 kcal/mol. The most destabilized mutation was P- > F, while the least was P- > I. This observation was quite different from the FoldX prediction, which indicated that HCAII_P237I was the most unstable and HCAII_P237F was the most stable mutant. The large discrepancy between the theoretical and experimental ΔΔG values for the three mutants HCAII_P237T, HCAII_P237F and HCAII_P237H (Figure 5) suggested that the effects of these mutations could not be predicted correctly. One possible reason is that the FoldX has a correlation coefficient of 0.81 and a standard deviation of 0.46 kcal/mol [20], and another may be that the prediction can give reasonable data of a large data set but not for the details of a small set, as indicated by other authors [32]. Nonetheless, the large experimental ΔΔG values (>2 kcal/mol) caused by the P237T, P237F and P237H mutations implied that the position 237 of CAII should play a role in CAII stability, and the disease-related mutation P237H was the most deleterious.

Tris, sulfuric acid, ultra pure guanidine hydrochloride (GdnHCL), p-nitrophenol acetate(p-NPA) and isopropyl-β-d-thiogalactopyranoside (IPTG) were purchased from Sigma. All the other chemicals were local products of analytical grade.

According to the previous report, a pseudo wild type HCAII (HCAII_pwt) was used in this research. This pseudo wild type protein was constructed with the mutation C206S to avoid possible interference from the folding of HCAII by incorrect disulfide formation since Cys206 is the only Cys in HCAII. Previous study has shown that HCAII_pwt has the same catalytic properties and folding as the wild type HCAII [27]. The mutated proteins were obtained by site directed mutagenesis using the following primers:

HCAII_P237I-For, 5′-CAATGGGGAGGGTGAAATCGAAGAACTGATG-3′;

Site-directed mutations were carried out using standard procedures. The genes were cloned into PET28b vector (Novagen) and a 6-His tag were added at the C-terminus of the protein to facilitate protein purification.

HCAII_pwt and the mutated proteins were overexpressed in E. coli Rosseta(DE3) in LB_kan at 37 °C, and the induction of overexpression was achieved by the addition of 0.5 mM IPTG and 0.5 mM ZnSO₄. The bacterial cells were harvested by centrifugation, sonicated and the target proteins were purified by Ni-NTA affinity chromatography (QIAGEN) as described previously [5]. The final products were collected on a Superdex 75 HR 10/30 (GE Healthcare Life Sciences), and only the peak containing the monomeric form were collected. The protein concentration were determined by measuring the absorbance at A₂₈₀ using ɛ_{280 nm} = 53,800 M⁻¹ cm⁻¹.

The enzymatic activity of HCAII_pwt and the mutants were determined by the esterase activity assay, which monitors the appearance of p-nitrophenolate anion spectrophototometrically during the hydrolysis of p-NPA [33]. The 1-mL assay mixtures contained 1 mM pNPA, 3% acetone and 10 mM Tris-H₂SO₄, pH 7.5. The reaction was started by the addition of 1.5 μM enzyme, and the hydrolysis of pNPA to pNP was monitored by following the increase in absorbance at 348 nm at 25 °C. The final value was obtained by subtracting the background values for the non-catalyzed ester hydrolysis.

The unfolding of HCAII_pwt and the mutated proteins were performed by incubating the proteins in 10 mM Tris-H₂SO₄ buffer, pH 7.5, in the presence of various concentrations of GdnHCl for 16 h at 25 °C. Then spectroscopic experiments were performed to monitor the structural changes of the samples. The final protein concentration was 0.8 μM. The unfolding data were fitted to a three-state folding model as described previously [5].

The intrinsic Trp fluorescence was measured using a 1 cm path-length quartz cuvette on a Hitachi F-4500 spectrophotometer at 25 °C. The excitation wavelength was 295 nm with both the entrance and exit slits of 5 nm, and the emission spectra were collected between 300 nm and 450 nm. The Far-UV circular dichroism (CD) spectra were recorded on a Jasco-715 spectrophotometer (Tokyo, Japan) over a wavelength range of 190–250 nm using the 0.1 cm path-length cells. The protein concentration for CD experiments was 1.5 μM. The presented spectra were the average of three repetitions.

The changes in the Gibbs free energy (ΔΔG) induced by mutations at position 237 were calculated by FoldX (version 3.0 beta3) [20]. The structure of the wild type protein (PDB ID: 2CBA) was minimized using the “RepairPDB” command to identify the residues that had bad torsion angles, van der Waal’s clashes or total energies belonging to the complex interface. Then individual mutations were built using “BuildModel” command and the ΔΔG values were extracted from the FoldX output files.