Characterization and High-Level Periplasmic Expression of Thermostable α-Carbonic Anhydrase from Thermosulfurimonas Dismutans in Escherichia Coli for CO2 Capture and Utilization

Carbonic anhydrase (CA) is a diffusion-controlled enzyme that rapidly catalyzes carbon dioxide (CO2) hydration. CA has been considered as a powerful and green catalyst for bioinspired CO2 capture and utilization (CCU). For successful industrial applications, it is necessary to expand the pool of thermostable CAs to meet the stability requirement under various operational conditions. In addition, high-level expression of thermostable CA is desirable for the economical production of the enzyme. In this study, a thermostable CA (tdCA) of Thermosulfurimonas dismutans isolated from a deep-sea hydrothermal vent was expressed in Escherichia coli and characterized in terms of expression level, solubility, activity and stability. tdCA showed higher solubility, activity, and stability compared to those of CA from Thermovibrio ammonificans, one of the most thermostable CAs, under low-salt aqueous conditions. tdCA was engineered for high-level expression by the introduction of a point mutation and periplasmic expression via the Sec-dependent pathway. The combined strategy resulted in a variant showing at least an 8.3-fold higher expression level compared to that of wild-type tdCA. The E. coli cells with the periplasmic tdCA variant were also investigated as an ultra-efficient whole-cell biocatalyst. The engineered bacterium displayed an 11.9-fold higher activity compared to that of the recently reported system with a halophilic CA. Collectively these results demonstrate that the highly expressed periplasmic tdCA variant, either in an isolated form or within a whole-cell platform, is a promising biocatalyst with high activity and stability for CCU applications.


Introduction
Carbonic anhydrase (CA) is a zinc-metalloenzyme that catalyzes carbon dioxide (CO 2 ) hydration: CO 2 + H 2 O → HCO 3 − + H + [1]. CAs are widespread in the three domains of life and are classified into seven distinct families (α, β, γ, δ, ζ, η and θ), performing important physiological roles in various organisms from microbes to human [1]. CAs show diffusion-controlled kinetics with a k cat of up to 4.4 × 10 6 s −1 [2]. This feature makes them powerful and eco-friendly catalysts for bioinspired CO 2 capture and utilization (CCU) which is one of the promising routes to the mitigation of greenhouse gas emissions. The hydration of CO 2 , which is the rate-determining step of CO 2 capture into HCO 3 − , can be accelerated by the catalysis of CA [3]. The rapid formation of HCO 3 − can benefit CCU processes that

Comparative Analysis of Bacterial Thermostable α-CA Sequences
The amino acid sequences of tdCA and taCA were aligned along with those of other α-CAs from the selected Gram-negative thermophiles, Persephonella marina (pmCA), Caminibacter mediatlanticus (cmCA) and Sulfurihydrogenibium yellowstonense (sspCA) (Figure 1). The residues highly conserved across bacterial α-CAs are also present in tdCA including three zinc ligand (His-115, His-117, and His-134; tdCA numbering system), proton shuttle residue (His-90), two gate-keeper residues (Glu-121 and Thr-200) and two cysteine residues forming an intramolecular disulfide bond (Cys-49 and Cys-204) [29]. All of the selected CAs have an N-terminal signal peptide, suggesting the periplasmic localization in the original hosts. The average sequence identity across the five sequences is 49.9%, while the highest identity is found between the sequences of taCA and pmCA (62.0%). The sequence of tdCA shows the highest and the lowest identities with those of taCA (56.3%) and sspCA (44.4%), respectively, implying that tdCA is more closely related to taCA than the other thermophilic CAs. . Conserved or similar residues across three or more sequences are shaded in black or gray, respectively. The native signal sequences are enclosed in blue boxes. The two cysteine residues for the formation of the intramolecular disulfide bond are indicated by a closed circle (•). The three zinc ligand histidine residues (*) and proton shuttling histidine residue (#) are marked. The two gate-keeper residues are also indicated by an open circle (○).

Expression and Purification of Recombinant CAs
The genes for tdCA and taCA were expressed in E. coli BL21(DE3) using the strong T7lac promoter system. The hexahistidine (His6)-tagged recombinant CAs without the predicted signal peptides were expressed and accumulated in the cytoplasm of E. coli. Both recombinant CAs were successfully produced in soluble forms with the estimated molecular weights of 27.2 kDa and 27.0 kDa for tdCA and taCA, respectively, which correspond to the band positions in the protein gel ( Figure 2a). The expression level of tdCA was 20% lower than that of the taCA in E. coli BL21(DE3) strain, as revealed by the densitometric analysis of soluble CAs on the Western blot result (Figure 2a). When the recombinant enzymes were purified by His6-tag affinity chromatography and dialyzed against sodium phosphate buffer, nearly half of taCA enzymes formed insoluble precipitates (Figure 2b). The low solubility of taCA has been previously reported and was alleviated by increasing the ionic strength of buffer that may screen the attractive interactions between taCA molecules [9]. taCA is a basic protein with an isoelectric point (pI) value of 9.1 and shows highly T. dismutans (tdCA; OAQ21602), T. ammonificans (taCA; WP_013538320), P. marina (pmCA; ACO04804), C. mediatlanticus (cmCA; EDM23829), and S. yellowstonense (sspCA; ACD66216). Conserved or similar residues across three or more sequences are shaded in black or gray, respectively. The native signal sequences are enclosed in blue boxes. The two cysteine residues for the formation of the intramolecular disulfide bond are indicated by a closed circle ( ). The three zinc ligand histidine residues (*) and proton shuttling histidine residue (#) are marked. The two gate-keeper residues are also indicated by an open circle ( ).

Expression and Purification of Recombinant CAs
The genes for tdCA and taCA were expressed in E. coli BL21(DE3) using the strong T7lac promoter system. The hexahistidine (His 6 )-tagged recombinant CAs without the predicted signal peptides were expressed and accumulated in the cytoplasm of E. coli. Both recombinant CAs were successfully produced in soluble forms with the estimated molecular weights of 27.2 kDa and 27.0 kDa for tdCA and taCA, respectively, which correspond to the band positions in the protein gel ( Figure 2a). The expression level of tdCA was 20% lower than that of the taCA in E. coli BL21(DE3) strain, as revealed by the densitometric analysis of soluble CAs on the Western blot result (Figure 2a).

Expression and Purification of Recombinant CAs
The genes for tdCA and taCA were expressed in E. coli BL21(DE3) using the strong T7lac promoter system. The hexahistidine (His6)-tagged recombinant CAs without the predicted signal peptides were expressed and accumulated in the cytoplasm of E. coli. Both recombinant CAs were successfully produced in soluble forms with the estimated molecular weights of 27.  When the recombinant enzymes were purified by His6-tag affinity chromatography and dialyzed against sodium phosphate buffer, nearly half of taCA enzymes formed insoluble precipitates (Figure 2b). The low solubility of taCA has been previously reported and was alleviated by increasing the ionic strength of buffer that may screen the attractive interactions between taCA molecules [9]. taCA is a basic protein with an isoelectric point (pI) value of 9.1 and shows highly When the recombinant enzymes were purified by His 6 -tag affinity chromatography and dialyzed against sodium phosphate buffer, nearly half of taCA enzymes formed insoluble precipitates (Figure 2b). The low solubility of taCA has been previously reported and was alleviated by increasing the ionic strength of buffer that may screen the attractive interactions between taCA molecules [9]. taCA is a basic protein with an isoelectric point (pI) value of 9.1 and shows highly positive electrostatic surface potential, which is in sharp contrast to tdCA that has pI value of 6.5 and moderately charged surface potential ( Figure 3). According to a computational analysis on the relationship between electrostatic surface properties and protein solubility [30], it appears that the positively charged surface patches of taCA ( Figure 3) contribute to the low solubility, although a convincing biochemical explanation has not been provided. In contrast, tdCA showed no precipitation after purification and dialysis (Figure 2b), indicating its high solubility. Even after the boiling of soluble enzymes for 1 h, a significant fraction of tdCA still remained soluble while no soluble protein was observed for taCA ( Figure 2b). Because the boiled tdCA lost almost all of the activity (showing only 0.5% of the initial activity), it is suggested that tdCA can retain its solubility even after denaturation, followed by irreversible inactivation. positive electrostatic surface potential, which is in sharp contrast to tdCA that has pI value of 6.5 and moderately charged surface potential ( Figure 3). According to a computational analysis on the relationship between electrostatic surface properties and protein solubility [30], it appears that the positively charged surface patches of taCA ( Figure 3) contribute to the low solubility, although a convincing biochemical explanation has not been provided. In contrast, tdCA showed no precipitation after purification and dialysis (Figure 2b), indicating its high solubility. Even after the boiling of soluble enzymes for 1 h, a significant fraction of tdCA still remained soluble while no soluble protein was observed for taCA ( Figure 2b). Because the boiled tdCA lost almost all of the activity (showing only 0.5% of the initial activity), it is suggested that tdCA can retain its solubility even after denaturation, followed by irreversible inactivation. The predicted structure of tdCA was first visualized in an orientation showing the active-site cavity in the center of the structure, and the structure of taCA was then aligned according to the orientation of tdCA. Only one (chain A) of the two chains was taken for the visualization of taCA.

Activity and Stability Comparison of Recombinant CAs
The purified soluble enzymes were subjected to the activity test. Enzyme activity was measured by an assay that depends on the color change of a pH indicator upon the generation of a proton by CO2 hydration reaction. The specific activity of tdCA (3200 U/mg) was 2.7-fold higher than that of taCA (1200 U/mg) (Figure 4a). Considering the reported high catalytic efficiencies of taCA [9,26], it appears that tdCA is one of the most efficient catalysts for CO2 hydration. The predicted structure of tdCA was first visualized in an orientation showing the active-site cavity in the center of the structure, and the structure of taCA was then aligned according to the orientation of tdCA. Only one (chain A) of the two chains was taken for the visualization of taCA.

Activity and Stability Comparison of Recombinant CAs
The purified soluble enzymes were subjected to the activity test. Enzyme activity was measured by an assay that depends on the color change of a pH indicator upon the generation of a proton by CO 2 hydration reaction. The specific activity of tdCA (3200 U/mg) was 2.7-fold higher than that of taCA (1200 U/mg) (Figure 4a). Considering the reported high catalytic efficiencies of taCA [9,26], it appears that tdCA is one of the most efficient catalysts for CO 2 hydration. The enzymes were preheated for 15 min at 70 °C and cooled at 4 °C prior to the long-term incubation. Error bars represent standard deviations from two or three independent experiments. Asterisks indicate statistical significance determined by unpaired, two-tailed (for panel a and b) or one-tailed (for panel c) t-test ( # p < 0.1,* p < 0.05, ** p < 0.01). ND, not detectable.
Next, the stability of the enzyme was evaluated by measuring the remaining enzyme activity after heat treatment under high-temperature conditions. The decrease in activity represents the fraction of enzyme inactivated due to the heating [31]. After short-term (15 min) incubation at high temperature ranging from 70 to 90 °C, tdCA showed similar residual activities (43-53% of its initial activity) regardless of the incubation temperature, while the residual activity of taCA gradually decreased from 88% (at 70 °C) to 47% (at 90 °C) (Figure 4b). At first glance, the stability of tdCA seemed to be lower than that of taCA because the residual activities of tdCA were generally lower than those of taCA (Figure 4b). For instance, the residual activity of tdCA after 15 min incubation at 70 °C was 46%, which was much lower than that (88%) of taCA. At 100 °C, however, tdCA exhibited 18% residual activity, while taCA completely lost its activity (Figure 4b). These results prompted us to hypothesize that the purified tdCA enzymes were composed of heat-labile and heat-stable fractions and the sharp decrease of tdCA in residual activity after the short-term incubation was due to the inactivation of the heat-labile fraction. The remaining heat-stable fraction was assumed to be more stable than taCA. To test this hypothesis, we first incubated the intact enzymes at 70 °C for 15 min, and then, the heat-treated enzymes were taken as the initial samples (0 days) for the subsequent longterm stability test (up to five days) at 70 °C. The result demonstrated that the preheated tdCA was actually more stable than taCA (Figure 4c). The inactivation showed monophasic, first-order kinetics The enzymes were preheated for 15 min at 70 • C and cooled at 4 • C prior to the long-term incubation. Error bars represent standard deviations from two or three independent experiments. Asterisks indicate statistical significance determined by unpaired, two-tailed (for panel a and b) or one-tailed (for panel c) t-test ( # p < 0.1, * p < 0.05, ** p < 0.01). ND, not detectable.
Next, the stability of the enzyme was evaluated by measuring the remaining enzyme activity after heat treatment under high-temperature conditions. The decrease in activity represents the fraction of enzyme inactivated due to the heating [31]. After short-term (15 min) incubation at high temperature ranging from 70 to 90 • C, tdCA showed similar residual activities (43-53% of its initial activity) regardless of the incubation temperature, while the residual activity of taCA gradually decreased from 88% (at 70 • C) to 47% (at 90 • C) (Figure 4b). At first glance, the stability of tdCA seemed to be lower than that of taCA because the residual activities of tdCA were generally lower than those of taCA (Figure 4b).
For instance, the residual activity of tdCA after 15 min incubation at 70 • C was 46%, which was much lower than that (88%) of taCA. At 100 • C, however, tdCA exhibited 18% residual activity, while taCA completely lost its activity (Figure 4b). These results prompted us to hypothesize that the purified tdCA enzymes were composed of heat-labile and heat-stable fractions and the sharp decrease of tdCA in residual activity after the short-term incubation was due to the inactivation of the heat-labile fraction. The remaining heat-stable fraction was assumed to be more stable than taCA. To test this hypothesis, we first incubated the intact enzymes at 70 • C for 15 min, and then, the heat-treated enzymes were taken as the initial samples (0 days) for the subsequent long-term stability test (up to five days) at 6 of 13 70 • C. The result demonstrated that the preheated tdCA was actually more stable than taCA (Figure 4c). The inactivation showed monophasic, first-order kinetics in both tdCA and taCA (R 2 > 0.99), implying that the active enzymes from the heat-treated initial sample were composed of a single, heat-stable fraction [32]. The calculated half-life of tdCA (42.2 h) at 70 • C was 83% longer than that of taCA (23.1 h). It is worth noting that the stability, as well as the solubility of taCA, could be improved by increasing the ionic strength with the addition of 300 mM NaCl [9,20] while the stability of tdCA was not affected by the addition of salt (data not shown). Collectively, these data show that tdCA can be a better catalyst than taCA in terms of both activity and stability under general low-salt-containing aqueous conditions.

Engineering of tdCA for High-Level Expression
As previously shown, the expression level of tdCA was relatively low (Figure 2a), which necessitates further improvement of the expression level for the practical applications. In our efforts to improve the stability of tdCA by protein engineering, we incidentally found that the expression level of tdCA was significantly increased by the mutation S82Y (Figure 5a). The variant tdCA S82Y did not show any difference in both activity and stability compared to the wild-type tdCA (Figure 5b), implying that the mutation S82Y is neutral for the characteristics of tdCA. Next, we tried to express and translocate tdCA into the periplasm of E. coli via the Sec-dependent pathway, which is a better pathway than the Twin-arginine translocation pathway for CA secretion [33], by genetically fusing PelB signal peptide (SP PelB ) to the N terminus of tdCA. This design was originally intended to construct a whole-cell biocatalyst with periplasmic tdCA (see below). The expression level of SP PelB ::tdCA was remarkably improved compared to that of the wild-type, showing a thick band in the SDS-PAGE analysis (Figure 5a). Thus, the periplasmic expression may be used as an alternative strategy for improving the expression level of recombinant protein poorly expressed in the cytoplasm. When combined with the S82Y mutation, the total SP PelB ::tdCA S82Y , including the insoluble premature form, showed higher expression level compared to that of SP PelB ::tdCA, although the expression levels of mature forms were similar to each other (Figure 5a). It appeared that the excess amount of SP PelB ::tdCA S82Y was accumulated as an aggregate in the cytoplasm without the cleavage of signal peptide due to the saturation of the Sec-translocon capacity [34]. Consequently, the expression level of mature form from SP PelB ::tdCA S82Y increased 8.3-fold compared to that from wild-type tdCA according to the densitometric analysis on the Western blot result. This result can successfully lead to the reduction of the enzyme production cost.
A minor band for a protein with a smaller molecular weight compared to that of the mature form was observed in the gel electrophoretic analysis when SP PelB ::tdCA S82Y (or SP PelB ::tdCA) was expressed (Figure 5a). The truncated (or cleaved) form of tdCA S82Y , along with the premature tdCA S82Y in the soluble faction, was co-purified with the mature tdCA S82Y upon His 6 -tag affinity purification (Figure 5c). Because the heterogeneity and the low purity of purified enzyme sample might interfere with further biochemical characterization, the His 6 -tag was relocated to the N terminus of tdCA S82Y (SP PelB ::His 6 -tdCA S82Y ) to exclude the cleaved form of tdCA S82Y from the affinity purification. Interestingly, the expression analysis of SP PelB ::His 6 -tdCA S82Y showed no clear band for the cleaved form or the premature form of tdCA S82Y (Figure 5c). The N-terminal sequence following the signal peptide appeared to be critical for the efficient periplasmic translocation without the undesirable cleavage. As a result, SP PelB ::His 6 -tdCA S82Y was purified to apparent homogeneity (Figure 5c). Unfortunately, the specific activity of SP PelB ::His 6 -tdCA S82Y was only 65% of that of wild-type tdCA (Figure 5d). The change of amino acid sequences of tdCA S82Y at both N-and C-termini upon the relocation of His 6 -tag might slightly alter the conformation of the enzyme, leading to the activity change. Therefore, although SP PelB ::His 6 -tdCA S82Y showed high-level expression and allowed high-purity purification, it seems not to be adequate for the practical applications for CCU when compared to SP PelB ::tdCA S82Y . It might be useful for other specialized purposes, e.g., protein crystallization requiring a large amount of highly purified protein.
expression levels of mature forms were similar to each other (Figure 5a). It appeared that the excess amount of SPPelB::tdCAS82Y was accumulated as an aggregate in the cytoplasm without the cleavage of signal peptide due to the saturation of the Sec-translocon capacity [34]. Consequently, the expression level of mature form from SPPelB::tdCAS82Y increased 8.3-fold compared to that from wild-type tdCA according to the densitometric analysis on the Western blot result. This result can successfully lead to the reduction of the enzyme production cost.  6 -tag compared with that of wild-type tdCA. Error bars represent standard deviations from two or three independent experiments. Asterisks indicate statistical significance determined by unpaired, two-tailed t-test (** p < 0.01).

Ultra-Efficient Whole-Cell Biocatalysts Based on tdCA
Finally, the whole-cell CO 2 hydration activities of the recombinant E. coli strains with highly expressed periplasmic tdCA were measured to examine the potential of the strains as whole-cell biocatalysts. The whole-cell catalyst with periplasmic taCA was not considered for further testing because the expression level of periplasmic taCA was too low to be compared with that of periplasmic tdCA (data not shown). Instead, the highly active periplasmic whole-cell biocatalyst (RBS 2 SP PelB ::hmCA) previously constructed by using a halophilic CA (hmCA) from Hydrogenovibrio marinus and by engineering the ribosome binding site (RBS) was used for the comparison [33]. The strains with periplasmic tdCA showed much higher whole-cell activities than that of the RBS 2 SP PelB ::hmCA strain ( Figure 6). Notably, the activity (26.6 U/mL·OD 600 ) of SP PelB ::tdCA strain, which was similar to that (25.4 U/mL·OD 600 ) of the SP PelB ::tdCA S82Y strain, was 11.9-fold higher compared to that (2.2 U/mL·OD 600 ) of RBS 2 SP PelB ::hmCA. Although the activity (6.5 U/mL·OD 600 ) of the SP PelB ::His 6 -tdCA S82Y strain was much lower than that of the SP PelB ::tdCA or SP PelB ::tdCA S82Y strain, it was still three-fold higher than that of the RBS 2 SP PelB ::hmCA. Considering the reported k cat values of taCA (9.6 × 10 5 s −1 -1.6 × 10 6 s −1 ) [9,26] and hmCA (3.3 × 10 5 s −1 ) [12], the high activity of tdCA enzyme (Figure 4a) seems to be the primary factor that contributes to the remarkably high activity of the tdCA-based whole-cell biocatalysts. Thus, the engineered strains with highly expressed periplasmic tdCA could be used as ultra-efficient CO 2 -capturing whole-cell biocatalysts. Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 8 of 13 Figure 6. Whole-cell activities of recombinant strains with periplasmic CAs. Any enzymatic activity in the supernatant after centrifugation of the whole-cell solution was subtracted from the total activity to obtain pure whole-cell activity. Error bars represent standard deviations from three independent experiments. Asterisks indicate statistical significance determined by unpaired, two-tailed t-test (* p < 0.05). An asterisk right above a bar indicates statistical significance compared with RBS2SPPelB::hmCA sample.

General Culture Conditions, Bacterial Strains, and Plasmids Construction
The E. coli strains, plasmids, and primers used in this study are listed in Table 1. E. coli TOP10 was used for the DNA works, and E. coli BL21(DE3) strain was used for recombinant protein expression. Luria-Bertani (LB) medium supplemented with antibiotics was used for the culture at 37 °C and 220 rpm in a shaking incubator (Jeiotech, Daejeon, Korea). Fifty μg/mL ampicillin or 10 μg/mL streptomycin was supplemented to the culture media of recombinant strains or wild-type E. coli TOP10, respectively. The tdCA gene (GenBank accession number: OAQ21602) was chemically synthesized with codon optimization for E. coli (Genscript, Piscataway, NJ, USA) with a codon adaptation index (CAI) of 0.96 and was subcloned into pET-22b (+) (Novagen, Madison, WI, USA) using NdeI and XhoI restriction sites, resulting in pET-tdCA. The putative signal peptide of 20 amino acids was not included in the recombinant tdCA. The previously constructed taCA gene was also codon-optimized with a CAI of 0.88 [9]. For the mutation of S82Y, one-step polymerase chain reaction (PCR)-based mutagenesis was performed [35] using pET-tdCA as the template plasmid using the listed primers, resulting in pET-tdCA-S82Y. For the periplasmic expression, the tdCA gene was amplified by PCR and was subcloned into pET-22b(+) using NcoI and XhoI restriction sites, resulting in pET-PelB-ss::tdCA, pET-PelB-ss::tdCA-S82Y or pET-PelB-ss::Nhis_tdCA-S82Y. The recombinant tdCA derivatives except for SPPelB::His6-tdCAS82Y have a His6-tag sequence at the C terminus.

Expression of Recombinant CA Enzymes
E. coli BL21(DE3) strains transformed with the recombinant plasmids were incubated in LB medium at 37 °C and 180 rpm in the shaking incubator. At an OD600 of 0.6-0.8 measured using a UV-Vis spectrophotometer (Shimadzu, Kyoto, Japan), the medium was supplemented with 1 mM isopropyl-β-D-thiogalactopyranoside (Duchefa Biochemie, Haarlem, Netherlands) and 0.1 mM ZnSO4 (Junsei, Tokyo, Japan) for the induction of recombinant protein expression. The cells were further cultivated for 12 h at 37 °C and 180 rpm, harvested by centrifugation at 4 °C and 4000× g for 10 min, and resuspended in lysis buffer (50 mM sodium phosphate, 300 mM NaCl, and 10 mM imidazole; pH 8.0). The cells were lysed by an ultrasonic dismembrator (Sonics and Materials, Newtown, CT, USA) for 20 min on ice water. The lysate was centrifuged at 4 °C and 10,000× g for 10 Figure 6. Whole-cell activities of recombinant strains with periplasmic CAs. Any enzymatic activity in the supernatant after centrifugation of the whole-cell solution was subtracted from the total activity to obtain pure whole-cell activity. Error bars represent standard deviations from three independent experiments. Asterisks indicate statistical significance determined by unpaired, two-tailed t-test (* p < 0.05). An asterisk right above a bar indicates statistical significance compared with RBS 2 SP PelB ::hmCA sample.

General Culture Conditions, Bacterial Strains, and Plasmids Construction
The E. coli strains, plasmids, and primers used in this study are listed in Table 1. E. coli TOP10 was used for the DNA works, and E. coli BL21(DE3) strain was used for recombinant protein expression. Luria-Bertani (LB) medium supplemented with antibiotics was used for the culture at 37 • C and 220 rpm in a shaking incubator (Jeiotech, Daejeon, Korea). Fifty µg/mL ampicillin or 10 µg/mL streptomycin was supplemented to the culture media of recombinant strains or wild-type E. coli TOP10, respectively. The tdCA gene (GenBank accession number: OAQ21602) was chemically synthesized with codon optimization for E. coli (Genscript, Piscataway, NJ, USA) with a codon adaptation index (CAI) of 0.96 and was subcloned into pET-22b (+) (Novagen, Madison, WI, USA) using NdeI and XhoI restriction sites, resulting in pET-tdCA. The putative signal peptide of 20 amino acids was not included in the recombinant tdCA. The previously constructed taCA gene was also codon-optimized with a CAI of 0.88 [9]. For the mutation of S82Y, one-step polymerase chain reaction (PCR)-based mutagenesis was performed [35] using pET-tdCA as the template plasmid using the listed primers, resulting in pET-tdCA-S82Y. For the periplasmic expression, the tdCA gene was amplified by PCR and was subcloned into pET-22b(+) using NcoI and XhoI restriction sites, resulting in pET-PelB-ss::tdCA, pET-PelB-ss::tdCA-S82Y or pET-PelB-ss::Nhis_tdCA-S82Y. The recombinant tdCA derivatives except for SP PelB ::His 6 -tdCA S82Y have a His 6 -tag sequence at the C terminus.

Expression of Recombinant CA Enzymes
E. coli BL21(DE3) strains transformed with the recombinant plasmids were incubated in LB medium at 37 • C and 180 rpm in the shaking incubator. At an OD 600 of 0.6-0.8 measured using a UV-Vis spectrophotometer (Shimadzu, Kyoto, Japan), the medium was supplemented with 1 mM isopropyl-β-D-thiogalactopyranoside (Duchefa Biochemie, Haarlem, Netherlands) and 0.1 mM ZnSO 4 (Junsei, Tokyo, Japan) for the induction of recombinant protein expression. The cells were further cultivated for 12 h at 37 • C and 180 rpm, harvested by centrifugation at 4 • C and 4000× g for 10 min, and resuspended in lysis buffer (50 mM sodium phosphate, 300 mM NaCl, and 10 mM imidazole; pH 8.0). The cells were lysed by an ultrasonic dismembrator (Sonics and Materials, Newtown, CT, USA) for 20 min on ice water. The lysate was centrifuged at 4 • C and 10,000× g for 10 min. The pellets were designated the insoluble fraction (IS) and the supernatants were designated the soluble fraction (S).

Purification of Recombinant CA Enzymes.
Prior to affinity purification of target enzymes, heat-labile endogenous host proteins were heat-precipitated by incubating the lysates at 60 • C for 20 min. After centrifugation of the heat-treated lysates at 4 • C and 10,000× g for 10 min, the supernatants were mixed with Ni 2+ -nitrilotriacetic acid agarose beads (Qiagen, Germantown, MD, USA), and the His 6 -tagged target proteins were purified according to the manufacturer's instructions. The enzymes were eluted using elution buffer (50 mM sodium phosphate, 300 mM NaCl, and 250 mM imidazole; pH 8.0). The purified recombinant CAs were thoroughly dialyzed against 20 mM sodium phosphate buffer (pH 7.5) at 4 • C. In the case of taCA, precipitated enzyme after dialysis was removed by centrifugation at 4 • C and 10,000× g for 10 min, and the soluble enzymes in the supernatant were used for further biochemical analyses. The enzyme concentrations were adjusted to 5 µM before the experiments.

Protein Quantification
The purified enzyme was denatured in denaturing buffer (6 M guanidine hydrochloride GuHCl/20 mM sodium phosphate; pH 7.5), and the absorbance of the denatured protein was measured at 280 nm in a quartz crystal cuvette. The concentration of the purified protein was determined using the measured absorbance, and the calculated extinction coefficient at 280 nm by ProtParam (http://web.expasy.org/protparam/) [36].

CO 2 Hydration Assay
The CO 2 hydration activity was assayed by a colorimetric method [12,37]. 10 or 20 µL of the sample was added to the disposable cuvette containing 600 µL of 20 mM Tris buffer (pH 8.3) supplemented with 100 µM phenol red. The reaction was performed at 4 • C inside the spectrometer by adding 400 µL of CO 2 -saturated deionized water prepared in ice-cold water. The absorbance change was monitored at 570 nm. The time (t) required for the absorbance to decrease from 1.2 (corresponding to pH 7.5) to 0.18 (corresponding to pH 6.5) was determined. The time (t 0 ) for the uncatalyzed reaction was also measured by adding a corresponding blank buffer instead of an enzyme sample. The enzyme unit (U) was calculated, as (t 0 − t) / (t × 5) as previously described [33].

Thermostability Test
The samples were incubated at the indicated temperatures for the appropriate time and then immediately cooled on ice. The activities of the incubated samples were measured and compared with the activities of the non-incubated samples. The residual activities were calculated and presented as relative residual activity (%).

Whole-cell Activity
The whole-cell activity was measured as previously described [33]. After cultivation, cells were harvested and resuspended in phosphate-buffered saline (PBS; 8 g/l NaCl, 0.2 g/l KCl, 1.44 g/l Na 2 HPO 4 , and 0.24 g/l KH 2 PO 4 ) at a cell concentration of 8-14 OD 600 . For whole-cell activity measurement, the obtained kinetic data after CO 2 hydration assay were corrected by subtracting the absorbance of the whole-cell from the measured absorbance at 570 nm. Any activity of leaked enzyme from the prepared cells was estimated by measuring the activity of supernatant after the centrifugal removal of the cells from the cell suspension and was subtracted from the total activity to obtain the pure whole-cell activity.

In Silico Calculations
The multiple sequence alignment was performed using ClustalX 2.0, and the aligned sequences were shaded with Boxshade 3.21 (https://embnet.vital-it.ch/software/BOX_form.html). The three-dimensional structure of tdCA was constructed by protein threading modeling using I-TASSER [38]. The best threading templates used for the construction included CAs from Neisseria gonorrhoeae (PDB ID: 1KOP), Sulfurihydrogenibium azorense (PDB ID: 4X5S), and Thermovibrio ammonificans (PDB ID: 4C3T, 4COQ). The calculation and visualization of the surface electrostatic potential were performed on UCSF Chimera using Coulombic surface coloring [39]. Signal peptide cleavage sites were predicted by the SignalP 5.0 server (http://www.cbs.dtu.dk/services/SignalP/) [40]. Densitometric analysis of the protein band on Western blot was performed using ImageJ [41].

Conclusions
The recombinant tdCA was successfully expressed and purified in E. coli BL21(DE3). The solubility of tdCA was higher than taCA. The CO 2 hydration activity of purified tdCA was 2.7-fold higher than that of taCA. The purified tdCA enzymes appeared to be composed of heat-labile and heat-stable fractions. Although tdCA easily lost a fraction of activity by short-term heat treatment due to the inactivation of heat-labile fraction, the higher stability of the remaining heat-stable fraction and the higher activity make tdCA a better catalyst compared to taCA under low-salt aqueous solutions. The relatively low expression level of tdCA was overcome by the S82Y mutation and the periplasmic secretion via the Sec-dependent pathway. The combined strategy resulted in at least an 8.3-fold higher expression level of tdCA compared to that of wild-type tdCA. In addition, the E. coli cells with the periplasmic tdCA variant displayed 11.9-fold higher whole-cell activity compared to that of the recently reported system with a halophilic CA. These results demonstrate that the highly expressed periplasmic tdCA can be used as a promising biocatalyst for CCU applications under low-salt conditions.