Structural and Kinetic Characterization of an Archaeal β-Class Carbonic Anhydrase

Abstract

The β-class carbonic anhydrase from the archaeon Methanobacterium thermoautotrophicum (Cab) was structurally and kinetically characterized. Analytical ultracentrifugation experiments show that Cab is a tetramer. Circular dichroism studies of Cab and the Spinacia oleracea (spinach) β-class carbonic anhydrase indicate that the secondary structure of the β-class enzymes is predominantly α-helical, unlike that of the α- or γ-class enzymes. Extended X-ray absorption fine structure results indicate the active zinc site of Cab is coordinated by two sulfur and two O/N ligands, with the possibility that one of the O/N ligands is derived from histidine and the other from water. Both the steady-state parameters k_cat and k_cat/K_m for CO₂ hydration are pH dependent. The steady-state parameter k_cat is buffer-dependent in a saturable manner at both pH 8.5 and 6.5, and the analysis suggested a ping-pong mechanism in which buffer is the second substrate. At saturating buffer conditions and pH 8.5, k_cat is 2.1-fold higher in H₂O than in D₂O, consistent with an intramolecular proton transfer step being rate contributing. The steady-state parameter k_cat/K_m is not dependent on buffer, and no solvent hydrogen isotope effect was observed. The results suggest a zinc hydroxide mechanism for Cab. The overall results indicate that prokaryotic β-class carbonic anhydrases have fundamental characteristics similar to the eukaryotic β-class enzymes and firmly establish that the α-, β-, and γ-classes are convergently evolved enzymes that, although structurally distinct, are functionally equivalent.

The thermophilic archaeon Methanobacterium thermoautotrophicum obtains energy for growth by the reduction of CO₂ to CH₄ and is also an obligate chemolithoautotroph; thus, this organism has a high demand for CO₂. Carbonic anhydrase, a zinc-containing enzyme catalyzing the reversible hydration of carbon dioxide (equation 1)

1

is expected to play an important role in the growth of M. thermoautotrophicum and may have several functions, including transporting HCO₃⁻ into the cell and providing CO₂ or HCO₃⁻ to enzymes that utilize these substrates.

Based on sequence comparisons, carbonic anhydrases belong to three genetically distinct classes (α, β, and γ) which appear to have independent origins (24). The most extensively studied enzymes are those from the α-class, which is composed primarily of mammalian carbonic anhydrases, but also includes enzymes from the green alga Chlamydomonas reinhardtii (19, 20) and the prokaryote Neisseria gonorrhoeae (13). The β-class enzymes are abundant in C₃ and C₄ monocotyledenous and dicotyledenous plants and green unicellular algae (24, 43), where they are essential for photosynthetic CO₂ fixation (6). The most recently identified class of carbonic anhydrase, the γ-class (24), is represented by the prototype Cam from the archaeon Methanosarcina thermophila (2). Even though sequences encoding putative γ-class carbonic anhydrases have been found in prokaryotes from both the Bacteria and Archaea domains (2, 52), Cam is the only γ-class enzyme that has been biochemically characterized (2, 3, 58).

Crystal structures for five α-class mammalian isozymes (CA I to V) (10, 16, 17, 23, 33, 38, 54) and the α-class enzyme from N. gonorrhoeae (26) reveal a monomer in which the dominating secondary feature is an antiparallel β-sheet. The γ-class Cam is remarkably distinct from the α-class carbonic anhydrases in that it is a homotrimer in which each monomer adopts a novel left-handed β-helix fold (28, 36). Even though the α- and γ-classes are notably different in both their tertiary and quaternary structures, both classes contain a catalytically essential zinc ion coordinated by three histidine residues. Recently, the structures of the β-class carbonic anhydrases from both the dicotyledenous plant Pisum sativum (pea) (35) and the red alga Porphyridium purpureum (41) have been solved. Both the P. sativum homo-octamer and the P. purpureum homodimer exhibit a predominantly α-helical secondary structure. Unlike enzymes from the α- and γ-classes, the active site zinc of these β-class enzymes is coordinated by two cysteines and one histidine residue. A conserved aspartate residue appears to serve as a fourth ligand in the P. purpureum enzyme, but not in the P. sativum enzyme.

The kinetic properties of the human α-class isozymes CA I, CA II, and CA III have been extensively investigated and follow a common zinc hydroxide mechanism for catalysis (39, 48). The catalytically active group in this mechanism model is the zinc-bound water, which ionizes to a metal-bound hydroxide ion that attacks CO₂. According to the proposed mechanism, the enzyme-catalyzed reaction occurs in two mechanistically distinct steps (where E = enzyme and B = buffer). The first step is the interconversion between carbon dioxide and bicarbonate (equations 2a and 2b), in which the rate is related to the steady-state parameter k_cat/K_m. The second step is the regeneration of the active form of the enzyme (equations 2c and 2d), involving the rate-determining intramolecular and intermolecular proton transfer events which are reflected in the steady-state parameter k_cat:

2a

2b

2c

2d

The catalytic mechanism of the M. thermophila γ-class Cam resembles that of human α-class CA II despite significant structural differences in the active sites of these two enzymes (1). The kinetic properties reported for the P. sativum and Spinacia oleracea (spinach) β-class carbonic anhydrases are also consistent with this mechanism (29, 30, 46). Thus, the kinetic analyses of enzymes from all three classes suggest convergent evolution of the catalytic mechanism (1, 36, 39).

The β-class was initially thought to be composed solely of carbonic anhydrases from monocotyledenous and dicotyledenous plants. A mitochondrial β-class carbonic anhydrase was discovered in C. reinhardtii (18), and other enzymes belonging to this class have since been identified in other algae (25, 62). Only two β-class carbonic anhydrases from the Bacteria domain have been purified (22, 53), and the subsequent purification of Cab, a β-class enzyme from the thermophilic archaeon M. thermoautotrophicum, establishes that this class of carbonic anhydrase extends to the Archaea domain (50). Recent work establishes that this class is widely distributed in metabolically diverse prokaryotes representing both the Bacteria and Archaea domains and has ancient origins (50, 52).

Even though the β-class of carbonic anhydrase is the only class with documented enzymes in all three domains, less is known about the biochemistry and overall structural aspects of this class than for either the α-class or the more recently identified γ-class. Herein we report on structural and kinetic studies of Cab, the first for any prokaryotic β-class carbonic anhydrase.

MATERIALS AND METHODS

Analytical ultracentrifugation.

Equilibrium centrifugation was performed with a Beckman model XLI ultracentrifuge. The radial distribution of protein was monitored by A₂₃₅, A₂₈₀, or A₂₉₅, depending upon the concentration loaded. Protein concentrations of Cab were estimated by using A₂₈₀ and an extinction coefficient (2,740 cm⁻¹ M⁻¹ based on 1 subunit) calculated from the deduced amino acid sequence of the cab gene. Cab was centrifuged at 8,000, 12,000, and 16,000 rpm for 22, 14, and 14 h, respectively. Establishment of equilibrium was verified by the coincidence of the final two scans at each speed. The ultracentrifugation data were fit by using NONLIN (Pharsight Corp., Mountain View, Calif.) and Sedenterp (37).

CD analysis.

Spectra were acquired at 37°C with an Aviv circular dichroism (CD) spectrophotometer, model 62DS. The concentration of the S. oleracea carbonic anhydrase was estimated by using A₂₈₀ and an extinction coefficient (23,260 cm⁻¹ M⁻¹ based on 1 subunit) calculated from the deduced amino acid sequence of the S. oleracea gene (excluding the 98 N-terminal amino acids encoding the chloroplast transit peptide sequence). Samples (10 μM) of the M. thermoautotrophicum Cab and S. oleracea carbonic anhydrases in 20 mM potassium phosphate (pH 6.8) containing 0.1 M KCl were placed in a cuvette with a 1-mm path length, and the data points obtained were from 320 to 202 nm in 1.0-nm increments. Five spectra were taken for each sample and averaged. The resulting spectra were normalized for direct comparison.

EXAFS.

Zn K-edge X-ray absorption spectroscopy (XAS) data of the as-isolated M. thermoautotrophicum ΔH carbonic anhydrase (Cab) were collected on beam line 7-3 at the Stanford Synchrotron Radiation Laboratory, with the SPEAR ring operating at 3.0 GeV and a 50- to 100-mA current (Table 1) (47). For XAS, 180 μl of Cab (50 mg/ml) in 50 mM potassium phosphate (pH 6.8) containing 35% glycerol was transferred to a Lucite cuvette covered with Mylar adhesive tape as an X-ray transparent window material, capped, and frozen in liquid nitrogen. Extended X-ray absorption fine structure (EXAFS) data analysis and curve fitting were performed by using EXAFSPAK (http://ssrl.slac.stanford.edu/exafspak.html) and Feff v7.0 (5, 44). Multiple-scattering contributions from outer-shell atoms of histidine ligands were quantified as described previously (15), with parameters derived from tetra(imidazole) zinc(II) perchlorate (7).

TABLE 1.

Collection of X-ray absorption spectroscopic data

Characteristic	Zn EXAFS result
Synchrotron radiation facility	Stanford Synchrotron Radiation Laboratory
Beamline	7-3
Current in storage ring (mA)	50–100
Monochromator crystal	Si[220]
Detection method	Fluorescence
Detector type	Solid-state arraya
Scan length (min)	25
No. of scans in average	12
Temp (K)	10
Energy standard	Zn foil, first inflection
Energy calibration (eV)	9,660.7
E0 (eV)	9,670
Pre-edge background (eV):
Energy range	8,657–9625
Gaussian center	8,638
Width	750
Spline background energy range; eV	9,333–9,902 (4)
(polynomial order)
	9,902–10,134 (4)
	10,134–10,366 (4)

^aThe 13-element Ge solid-state X-ray fluorescence detector at the Stanford Synchrotron Radiation Laboratory is provided by the National Institutes of Health Biotechnology Research Resource. 

Enzyme purification.

Cab was heterologously produced in Escherichia coli as previously described (50). Thawed cell paste (10 g) was suspended in 20 ml of buffer A (50 mM potassium phosphate [pH 6.8]) and passed twice through a chilled French pressure cell at 138 MPa. The cell lysate was centrifuged at 20,000 × g for 20 min to remove cell debris and then centrifuged at 100,000 × g for 2 h to remove membranes. The supernatant was loaded onto a 50-ml Q Sepharose (Fast Flow) anion-exchange column (Pharmacia) equilibrated with buffer A. After a 100-ml wash, the column was developed with a 400-ml linear gradient from 0 to 0.75 M KCl. The enzyme eluted between 450 and 550 mM KCl, and fractions containing active enzyme were pooled. The fractions containing active enzyme were raised to 1.5 M (NH₄)₂SO₄ and run on a 50-ml Phenyl-Sepharose column (Pharmacia) equilibrated with buffer A plus 1.5 M (NH₄)₂SO₄. After a 100-ml wash, the column was developed with a 400-ml linear gradient from 1.5 to 0 M (NH₄)₂SO₄ with the enzyme eluting at approximately 100 mM (NH₄)₂SO₄. The active fractions were pooled, desalted, and loaded onto a Mono Q 10/10 anion-exchange column (Pharmacia) equilibrated with buffer A. After a 30-ml wash, the column was developed with a 100-ml linear gradient from 0 to 1 M KCl. The enzyme eluted between 460 and 520 mM KCl, and fractions containing active enzyme were pooled, desalted, and stored at −20°C. Carbonic anhydrase activity was measured at room temperature by a modification of the electrometric method of Wilbur and Anderson (60). Protein concentrations were determined by the Bradford method with Bio-Rad dye reagent and bovine serum albumin (Sigma) as the standard (12).

Steady-state kinetics.

Initial rates of CO₂ hydration and HCO₃⁻ dehydration were determined by stopped-flow spectroscopy (KinTek, State College, Pa.) at 25°C by the changing pH indicator method (34). Saturated solutions of CO₂ were prepared by bubbling CO₂ into distilled, deionized water (32.9 mM) at 25°C. The CO₂ concentration ranged from 6 to 24 mM, and the HCO₃⁻ concentration ranged from 5 to 80 mM. The following buffer-pH indicator pairs (and wavelengths) were used: at pH 5.5 to 6.6, MES [2-(N-morpolino)ethanesulfonic acid] (pK_a = 6.1)-chlorophenol red (574 nm); at pH 6.8 to 7.2, MOPS [2-(N-morpholino)propanesulfonic acid] (pK_a = 7.2)–p-nitrophenol (400 nm); at pH 7.4 to 7.8, HEPES [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid] (pK_a = 7.5)-phenol red (557 nm); at pH 8.0 to 9.0, TAPS [N-tris(hydroxymethyl)methyl-3-aminopropanesulfonic acid] (pK_a = 8.4)–m-cresol purple (578 nm). The observed initial rates were corrected for the uncatalyzed rate of the reaction, which was at least five times lower than the catalyzed rate. The steady-state parameters k_cat and k_cat/K_m and their standard errors were then determined by fitting the observed initial rates to the Michaelis-Menten equation.

The buffer dependence was measured at pH 8.5 (TAPS) and 6.5 (MES) by varying the CO₂ concentration from 6 to 24 mM and the buffer concentration from 5 to 50 mM, maintaining the ionic strength at 0.1 M by addition of sodium sulfate. The observed initial rates were fit to the Michaelis-Menten equation. To determine the solvent deuterium isotope effect on the kinetic constants, identical buffer solutions were prepared in H₂O and D₂O. The solvent isotope effect was measured at pH 8.5 (50 mM TAPS). The pH was adjusted until the pH meter readings were identical. This method is appropriate, because the correction of a pH meter reading for pD (pD = meter reading + 0.4) (21) is approximately compensated for by the change in the acid dissociation constant in D₂O for weak acids whose pK_a values lie between 3 and 10 (pK_D − pK_H = 0.5 ± 0.1) (8). The CO₂ concentrations ranged from 7 to 27 mM based on the saturating solutions of CO₂ in D₂O at 25°C (38.1 mM). All fits described were performed with KaleidaGraph (Synergy Software, Reading, Pa.).

RESULTS

Analytical ultracentrifugation analysis.

Gel filtration analysis had previously suggested that Cab is a tetramer (50). More precise analytical ultracentrifugation studies were performed to confirm the quaternary structure. A global fit of the data at several speeds and initial concentrations of enzyme could only be satisfactorily modeled as a single species. The molecular mass of 68,080 Da calculated from such a fit is approximately 91% of the tetrameric mass (74,810 Da) calculated from the deduced amino acid sequence. These results establish that Cab is a homotetramer.

CD spectroscopy.

CD spectroscopy was performed on both Cab and the β-class S. oleracea enzyme to characterize and compare the secondary structures. The spectra overlap throughout the far-UV region (Fig. 1). Cab has two intense negative bands centered at 223 nm ([θ] = −7,700.5) and 209 nm ([θ] = −9,970.0). This result is similar to the spectrum obtained for the S. oleracea enzyme (Fig. 1), which has two intense bands centered at 221 nm ([θ] = −11,078) and 209 nm ([θ] = −12,143). These results suggest that the dominant structure for both enzymes is α-helical (31, 32). The CD spectrum of the tetrameric Cab is less intense than that of the octameric S. oleracea enzyme, suggesting that the proportion of α-helical structure in the plant enzyme is greater than that of Cab.

FIG. 1.

Comparison of the CD spectra of the M. thermoautotrophicum and S. oleracea β carbonic anhydrases. Far-UV region spectra of the S. oleracea (solid line) and M. thermoautotrophicum (dotted line) enzymes are shown.

XAS.

The Zn K-edge X-ray absorption spectrum for Cab shows a shift in absorption edge position to lower energy compared to Cam, the γ-class carbonic anhydrase from M. thermophila (1) (Fig. 2A). This shift is indicative of increased electron density at the Zn atom, resulting from an increase in electron-donating ligands such as sulfur. Similarly, the shift of the main peak in the Fourier transform (FT) of β-class Cab (Fig. 2B) to a longer distance, compared with γ-class Cam, is indicative of a larger proportion of sulfur-containing ligands. The decrease in amplitude of the ca. 3- and 4-Å multiple-scattering peaks in the FT indicates that the Zn²⁺ site of β-class Cab contains fewer histidine ligands than γ-class Cam.

FIG. 2.

XAS analysis of Cab. (A and B) X-ray absorption edge spectra (A) and FTs (B) (k = 2 to 12 Å⁻¹) for Cab, the M. thermoautotrophicum β-class carbonic anhydrase (solid line), and Cam, the γ-class carbonic anhydrase from M. thermophila (dashed line) (1). (C and D) k³-weighted EXAFS (C) and FTs (D) (k = 2 to 12 Å⁻¹) of the β-class Cab (solid line) and calculated results for ZnS₂(N/O)₂ (dashed line; fit 4, Table 2).

The k³-weighted EXAFS of Cab (Fig. 2C) are best fit by assuming a Zn-S₂(N/O)₂ coordination environment (cf. fits 3 to 6, Table 2). Bond valence sum (BVS) analysis is an empirical method for correlating the sum of the strength of the metal-ligand bonds, as measured by the bond length, to the oxidation state of the metal (40, 57). BVS confirms that the Zn site has a coordination number of 4 (the BVS value should be approximately equal to the oxidation state of the metal, i.e., 2). The EXAFS can be fit assuming one of the O/N ligands results from histidine imidazole coordination (fit 6, Table 2). This fit results in a slight improvement in goodness-of-fit value (cf. fits 4 and 6, Table 2). However, the Debye-Waller factor (ς_as²) for the multiple scattering paths from outer shell atoms indicates that the contribution of this moiety to the overall EXAFS is small. As such, the population of histidine ligands must be disordered to account for the observed ς_as² values. This disorder could result from a heterogeneity among the Zn sites or from a Zn-histidine coordination geometry in which the imidazole ring is tilted such that the Zn-N-C bond angles differ significantly for the two sides of the imidazole ring. Thus, the EXAFS results indicate the zinc site of Cab is coordinated by two sulfur and two O/N ligands, with the possibility that one of the O/N ligands results from histidine (cf. fit 4, Table 2; Fig. 2D).

TABLE 2.

Curve-fitting results for Zn EXAFS^a

Fit	Shell	NS	Ras (Å)	ςas2 (Å2)	ΔE0 (eV)	fb	BVSc
1	Zn-O	4	2.10	0.0054	4.38	0.131	1.37
2	Zn-O	3	2.03	0.0031	−1.35	0.072	1.78
	Zn-S	1	2.32	−0.0011
3	Zn-O	3	2.01	0.0058	−3.65	0.080	2.45
	Zn-S	2	2.30	0.0035
4	Zn-O	2	2.00	0.0016	−4.86	0.073	2.03
	Zn-S	2	2.30	0.0022
5	Zn-O	2	1.99	0.0042	−5.91	0.081	2.67
	Zn-S	3	2.29	0.0057
6	Zn-S	2	2.30	0.0023	−4.42	0.070	2.03
	Zn-O,N	2	2.00	0.0016
	Zn-C	1	2.99	0.0092
	Zn-C	1	[3.06]	[0.0092]
	Zn-C	1	[4.15]	[0.0132]
	Zn-N	1	[4.19]	[0.0132]

^aThe sample used was M. thermoautotrophicum β-CA. The Stanford Synchrotron Radiation Laboratory file number is ZTC0A (k range = 2 to 12 Å⁻¹). Δk³_χ = 12.38. Each group is the chemical unit defined for the multiple scattering calculation. N_S is the number of scatterers (or groups) per metal. R_as is the metal-scatterer distance. ς_as² is a mean square deviation in R_as. ΔE₀ is the shift in E₀ for the theoretical scattering functions. Numbers in square brackets were constrained to be either a multiple of the above value (ς_as²) or to maintain a constant difference from the above value (R_as). 

^b

f is a normalized error (chi squared):

^c BVS = Σ_exp[(r₀-r)/B], B = 0.37, r₀(Zn²⁺-O) = 1.704, and r₀(Zn²⁺-S) = 2.09. 

Steady-state kinetic measurements.

The pH dependencies of both CO₂ hydration and HCO₃⁻ dehydration catalyzed by Cab were measured by stopped-flow spectroscopy with the changing pH indicator assay. The progress curves for the hydration of CO₂ and dehydration of HCO₃⁻ were consistent with Michaelis-Menten kinetics. The efficiency (k_cat/K_m) for CO₂ hydration (Fig. 3A) was several fold greater than that for HCO₃⁻ dehydration (Fig. 3B) over the pH range of 6.5 to 7.5, with a 20-fold difference in efficiency at pH 7.0.

FIG. 3.

pH dependence of CO₂ hydration and HCO₃⁻ dehydration. Activities were measured at 25°C in 50 mM buffer with an ionic strength of 0.1 M. (A) CO₂ hydration. Using Cab concentrations of 1.5 to 1.9 μM, the observed steady-state parameters k_cat (■) and k_cat/K_m (□) were determined over a pH range of 6.2 to 9.0. (B) HCO₃⁻ dehydration. Using Cab concentrations of 0.8 to 1.8 μM, the k_cat/K_m (■) was determined over a pH range of 5.5 to 7.5.

Both the steady-state parameters k_cat and k_cat/K_m for CO₂ hydration were pH dependent over a range from pH 6.2 to 9.0 and show an increase with increasing pH (Fig. 3A). The CO₂ + H₂O ⇌ HCO₃⁻ + H⁺ equilibrium becomes increasingly shifted towards HCO₃⁻ at increasing pH, requiring the HCO₃⁻ dehydration reactions to be performed over the pH range from 5.5 to 7.5. CO₂ hydration was not measured below pH 6.2 by this method for the same reasons. The k_cat/K_m for HCO₃⁻ dehydration decreased with increasing pH (Fig. 3B). Neither pH profile could be fitted to a theoretical titration curve with one, two, or three ionizations.

The rate of CO₂ hydration determined at both pH 6.5 and 8.5 was found to be strongly dependent on the concentration of buffer (Fig. 4). The steady-state parameter k_cat at both pH values was buffer dependent in a saturable manner. Replots of the k_cat values yielded an effective K_m of 4.8 mM for TAPS at pH 8.5 (Fig. 4A) and 12.7 mM for MES at pH 6.5 (Fig. 4B). These values are typical for the apparent K_m of zwitterionic buffers used in this pH range. These results indicate that the buffer behaves kinetically as a second substrate in a ping-pong mechanism, likely accepting a proton from the enzyme during CO₂ hydration. The rate constant k_cat/K_m was not dependent on the concentration of buffer (Fig. 4A and B insets).

FIG. 4.

Buffer dependence of CO₂ hydration. CO₂ hydration was measured at 25°C at pH 8.5 (A) and pH 6.5 (B) with an ionic strength of 0.1 M and a Cab concentration of 2.0 μM. The CO₂ concentration ranged from 6 to 24 mM, and the buffer concentration ranged from 5 to 50 mM. The observed steady-state parameters k_cat max and K_m, determined by using the Michaelis-Menten equation, were (1.3 ± 0.048) × 10⁴ s⁻¹ and 4.8 ± 0.7 mM, respectively, for TAPS at pH 8.5 and (4.5 ± 0.084) × 10³ s⁻¹ and 12.7 ± 0.6 mM, respectively, for MES at pH 6.5. (Insets) Double reciprocal plot of observed initial velocities versus CO₂ concentration at different concentrations of TAPS (A) or MES (B) (■, 50 mM; □, 20 mM; ●, 10 mM; ○, 5 mM).

The solvent hydrogen isotope effects on the steady-state parameters for CO₂ hydration were measured at pH 8.5 and a concentration of TAPS buffer (50 mM) in which intermolecular proton transfer is not rate limiting. The solvent hydrogen isotope effect on k_cat was 2.1 ± 0.1, and no significant effect on k_cat/K_m (1.2 ± 0.1) was observed.

DISCUSSION

The overall folds of the monomeric mammalian and prokaryotic N. gonorrhoeae α-class carbonic anhydrases are similar, with the antiparallel β-sheet as the dominating secondary structure feature (26). The crystal structure of Cam, the prototypic γ-class carbonic anhydrase, reveals a homotrimer with the monomer adopting a novel left-handed β-helix fold (36). In stark contrast to the mainly β-sheet structures of α- and γ-class carbonic anhydrases, CD analysis of Cab and the S. oleracea enzyme suggests a predominantly α-helical structure. These prokaryotic and eukaryotic β-class carbonic anhydrases are from organisms at the phylogenetic extremes, suggesting this secondary structure feature is common to all β-class enzymes; indeed, the crystal structures of the P. sativum and P. purpureum β-class enzymes reveal a predominance of α-helical structure (35, 41). The content and arrangement of the predicted secondary structure elements for Cab are similar to those of the other β-class carbonic anhydrases (Fig. 5), suggesting a common ancestor, even though the amino acid sequence of Cab is only 23.3% identical to that of the S. oleracea enzyme.

FIG. 5.

Alignment of β-class carbonic anhydrase sequences. An alignment of the amino acid sequences of selected β-class carbonic anhydrases was generated with Clustal X (56). The three conserved zinc ligands are shaded in black, and the two other completely conserved amino acids are shaded in dark gray. The three active site residues not conserved in Cab are shaded in light gray. The open barrels indicate α helices as determined from the P. sativum structure (35), and the arrows represent β strands. The numbering refers to that of the M. thermoautotrophicum Cab amino acid sequence. GenBank accession numbers are as follows: P. sativum (pea), 115471; S. oleracea (spinach), 115472; Oryza sativa (rice), 606817; P. purpureum, 1395172; Coccomyxa sp., 1663720; E. coli CynT, 1657535; and M. thermoautotrophicum ΔH Cab, 1272331.

Nondenaturing polyacrylamide gel electrophoresis and gel filtration chromatography have previously suggested that the β-class carbonic anhydrases from monocotyledenous plants are homodimeric (43) and those from dicotyledenous plants are homooctameric (9); however, a more precise molecular mass has yet to be reported. In contrast to the enzymes isolated from higher plants, gel filtration studies of Cab and the E. coli CynT suggest these prokaryotic enzymes are homotetrameric (22, 50). The more precise analytical ultracentrifugation experiments reported here establish the homotetrameric composition of Cab. Thus, the β-class is distinct from the other two classes in that the β-class carbonic anhydrases are either dimeric, tetrameric, or octameric.

Based on electron microscopy of the chickpea leaf β-class carbonic anhydrase, Aliev et al. (4) presented a model for the quaternary structure of the β-class enzymes from dicotyledenous plants and proposed a 422 (dimer of tetramers) point group symmetry for the eight subunits. Two invariant cysteines (Cys-269 and Cys-272) in the deduced amino acid sequences of β-class carbonic anhydrases from dicotyledenous plants appear necessary for the oligomeric state of the P. sativum chloroplast carbonic anhydrase (9). Studies of these invariant cysteines fit well with the electron microscopy studies indicating a double-layered structure in which each layer is a tetramer. However, the recently solved structure of the P. satvium enzyme indicates the octamer does not have the predicted 422 point of symmetry, but rather has a 222 (dimer of dimers) symmetry (35). The dimeric P. purpureum enzyme, in which each monomer is composed of two internally repeated structures each having an active site (Fig. 5), appears as a tetramer with a pseudo 222 symmetry (41). Why alterations to the invariant Cys-269 and Cys-272 of the dicotyledenous plant enzymes convert the octamer into a tetramer is unclear.

In the P. sativum enzyme, pairs of monomers are joined together through extensive interactions mediated by the α1 and α2 helices and β2 strands (35). The dimer is therefore the basic building block, and nearly all of the protein-protein interactions responsible for forming the loosely packed octamer from dimers are mediated through interactions of the β5 strand. However, the second half of the β5 strand, which mediates most of the oligomerization interactions, is absent in several β-class enzymes, including Cab (Fig. 5). How the dimers are held together in the tetrameric Cab awaits solution of its crystal structure.

XAS analysis clearly indicates that the zinc coordination of β-class Cab is distinct from that of the α- and γ-class carbonic anhydrases (cf. fit 4, Table 2; Fig. 2D). The active site zinc of the α- and γ-class enzymes is coordinated by three histidines and at least one water molecule (1, 28, 61). In the catalytic mechanism of carbonic anhydrase, the zinc-bound water ionizes to a metal-bound hydroxide ion that attacks CO₂ (equation 2a). EXAFS results suggest that the active site zinc of β-class Cab is coordinated by two cysteine residues and two oxygen/nitrogen ligands, with the possibility that one of the oxygen/nitrogen ligands derives from histidine (Table 2, fit 4). Our results are nearly identical to EXAFS results previously reported for the S. oleracea enzyme (11, 46). Cys-32, His-87, and Cys-90 of Cab are completely conserved in all known β-class carbonic anhydrase sequences (Fig. 5). The recently solved crystal structures of the P. sativum and P. purpureum β-class carbonic anhydrase confirm that the two conserved cysteines and the conserved histidine are ligands to the active site zinc (35, 41). The second oxygen/nitrogen ligand would be expected to be a water molecule; however, the fourth ligand of the recently solved P. purpureum crystal structure is a conserved aspartate corresponding to Asp-34 of Cab. Nonetheless, several pieces of evidence suggest that the conserved aspartate is unlikely to act as an essential fourth ligand to the active site zinc in β-class carbonic anhydrases. First, this aspartate has previously been shown not to be essential for zinc coordination or catalytic activity. Site-directed mutagenesis studies with the S. oleracea enzyme (11) indicate that alterations to the two conserved cysteine residues and the conserved histidine residue result in inactive variants lacking zinc; however, the variant in which the conserved aspartate was replaced with asparagine retained the active site zinc. Site-directed mutagenesis of the S. oleracea enzyme and Cab indicates that the conserved aspartate is not absolutely required for catalytic activity (41; K. S. Smith, C. J. Ingram-Smith, and J. G. Ferry, unpublished data). Second, the P. purpureum carbonic anhydrase was crystallized at pH 6.75, and previous reports indicate that essentially no activity was detected for this enzyme below pH 7.0 (62), suggesting that the published structure of the P. purpureum enzyme may be of an inactive enzyme. Therefore, Asp-34 is unlikely to be an essential fourth ligand to the active site zinc in Cab. The second oxygen/nitrogen ligand in Cab is most likely a deprotonated water molecule that serves as the zinc hydroxide attacking CO₂.

Even though the three zinc ligands (Cys-32, His-87, and Cys-90 of Cab) and two other active site residues (Asp-34 and Arg-36 of Cab) are conserved among all β-class carbonic anhydrases (Fig. 5), key active site residues (Gln-151, Phe-179, and Tyr-205 of the P. sativum enzyme) conserved among the enzymes from dicotyledenous and monocotyledenous plants, algae, and E. coli CynT, are absent in Cab (35). Although the roles of these residues in the P. sativum enzyme have not yet been investigated experimentally, Kimber and Pai (35) propose that Gln-151 may electrophilically activate the CO₂ molecule by forming a hydrogen bond with CO₂ through its side chain amide. Phe-179 and Tyr-205 form part of an extensive hydrophobic patch whose function may be to ensure that the binding energy of inhibitor molecules is as unfavorable as possible (35). Other members of the same phylogenetic clade as Cab, which consists primarily of sequences from both archaea and gram-positive bacteria species, are also missing these active site residues present in P. sativum (51, 52). In Cab, Gln-151, Phe-179, and Tyr-205 of the P. sativum enzyme are substituted for by histidine (His-23), lysine (Lys-53), and valine (Val-72), respectively (Fig. 5). These substitutions imply that the active site of Cab differs substantially from those of other β-class enzymes. Whether His-23, Lys-53, and Val-72 in Cab play a role similar to that proposed for Gln-151, Phe-179, and Tyr-205 in the plant enzymes remains to be investigated.

Although the active sites of the plant β-class carbonic anhydrases and Cab may be significantly different, the kinetic data presented here suggest that the fundamental catalytic mechanism for Cab is similar to that reported for other β-class enzymes. The steady-state parameter k_cat/K_m is not dependent on the concentration of buffer, which was shown to act as a second substrate. This result is consistent with the α-class human CA II zinc hydroxide mechanism in which the interconversion of CO₂ and HCO₃⁻ (equations 2a and 2b), reflected in k_cat/K_m, is separate from the intramolecular and intermolecular proton transfer steps (equations 2c and 2d) (14, 34, 48). The pH profile of k_cat for the hydration of CO₂ (Fig. 3A) increases with pH, indicating that an unprotonated form of the enzyme is required for catalytic competence, consistent with nucleophilic attack of a zinc-bound hydroxyl group on CO₂. For human CA II, the pH profile of k_cat/K_m reveals the pK_a of the zinc-bound water and the pH profile of k_cat reveals the pK_a of the proton shuttle residue. The pH profiles of both k_cat and k_cat/K_m in the direction of CO₂ hydration for Cab (Fig. 3A) show more complicated behavior and could not be fitted to theoretical curves with one, two, or three ionizations. Nearby ionizable groups may influence these pH profiles to produce multiple pK_a values, or the pK_a values are lower than 6.0. The steady-state parameter k_cat of the β-class S. oleracea carbonic anhydrase was found to be pH dependent, with an apparent pK_a of approximately 8.5 in the absence of sulfate (46). Similar to Cab, the pH profiles for both k_cat and k_cat/K_m in the direction of CO₂ hydration for the β-class P. sativum carbonic anhydrase are pH dependent, although the profiles were not fitted to theoretical titration curves (29).

No significant hydrogen isotope effect was observed on the steady-state parameter k_cat/K_m for Cab. This result suggests that D₂O imposes no major structural changes in the enzyme and that the catalytic steps up to and including the first committed step of the reaction (equations 2a and 2b) do not contain a rate-contributing proton transfer step, consistent with the zinc-hydroxide mechanism. The solvent hydrogen isotope effect on k_cat observed for Cab suggests that an intramolecular proton transfer step is at least partially rate determining (equation 2c). Similar isotope effects on k_cat were reported for the bovine CA III α-class carbonic anhydrase (45), the γ-class enzyme Cam (1), and the P. sativum β-class carbonic anhydrase (30). The observed solvent hydrogen isotope effect on k_cat of 2.1 for Cab is smaller than the value of 3.8 reported for human α-class CA II (55), but similar to the value reported for human α-class CA IV (27), which follows a mechanism similar to that of CA II.

M. thermoautotrophicum grows optimally at temperatures between 65 and 75°C, and it is expected that the optimal temperature for enzyme activity would fall in this range; however, the decreased solubility of CO₂ at these temperatures under atmospheric pressure precludes the determination of accurate kinetic parameters above 25°C. In fact, Cab is the most thermostable carbonic anhydrase yet characterized, retaining greater than 90% activity after incubation at 85°C for 15 min (50). Optimal activity aside, the catalytic efficiency (k_cat/K_m) for CO₂ hydration (Fig. 3A) was several fold greater than that for HCO₃⁻ dehydration (Fig. 3B) over the pH range of 6.5 to 7.5, suggesting that the physiological role of Cab is to convert CO₂ to HCO₃⁻.

The chemolithoautotrophic M. thermoautotrophicum fixes CO₂, and synthesis of oxaloacetate is an important reaction in the CO₂-fixation pathways for the methanoarchaea. Oxaloacetate is the starting point of an incomplete reductive citric acid cycle that terminates at α-ketoglutarate and provides precursors for cell material and coenzyme biosynthesis (49). M. thermoautotrophicum possesses two enzymes, pyruvate carboxylase and phosphoenolpyruvate (PEP) carboxylase, for the synthesis of oxaloacetate (49). Bicarbonate has been shown to be the substrate for both of these enzymes; thus, the role of Cab may be to concentrate HCO₃⁻ in the vicinity of these enzymes. Similarly, eukaryotic carbonic anhydrase has been shown to provide bicarbonate to both pyruvate carboxylase and PEP carboxylase. The α-class human CA V is a mitochondrial enzyme that provides HCO₃⁻ for pyruvate carboxylase in the liver, kidney, and pancreatic islets (42, 59). In the photosynthesis of C₄ plants, carbonic anhydrase provides bicarbonate to PEP carboxylase for the initial carboxylation reaction in the fixation of CO₂ into C₄ acids (6) by rapidly converting CO₂ entering the mesophyll cells from the atmosphere to HCO₃⁻.

Conclusions.

Previously, only the plant β-class carbonic anhydrases had been characterized structurally or kinetically. Here we present the first structural and detailed study of a β-class carbonic anhydrase (Cab) from a prokaryote and the first from a chemolithotrophic thermophile. Cab and the enzymes from dicotyledenous plants represent the greatest extremes on the phylogenetic tree of the β-class of carbonic anhydrases (51, 52). The results presented here reveal remarkable similarities between the eukaryotic and prokaryotic enzymes that unite the β-class. Both Cab and the plant enzymes follow a zinc hydroxide mechanism for catalysis. The dominant structure for β-class enzymes is α-helical, and the active site is coordinated by two sulfur and two O/N ligands. These results firmly establish that the α-, β-, and γ-classes are convergently evolved enzymes that, although structurally distinct, are functionally equivalent.