Introduction

Although enzymes have been used as catalysts in a variety of commercial applications (e.g., drugs, medical diagnostics, pulp bleaching, starch processing, biosensors, and detergents), many enzymes lack practical utility despite their useful catalytic activities, primarily because their thermolability hinders their prolonged use as catalysts and long-term storage at ordinary temperatures. In particular, although enzymes identified in mesophiles have efficient catalytic activities at moderate temperatures, they are thermolabile and readily inactivated. In contrast, most enzymes isolated from extreme thermophiles are thermostable but are inefficient catalysts at moderate temperatures. Therefore, enzymes from both mesophiles and extreme thermophiles are often impractical for applications that require efficient enzymatic activity at room temperature (e.g., medical diagnostics and biosensors). Protein engineering is a promising approach to overcome this challenge. Useful catalysts can be obtained by enhancing the thermostability of thermolabile enzymes while maintaining efficient catalytic activities at moderate temperatures.

There are two main approaches for enhancing enzyme thermostability. In one approach, in silico rational design based upon structural information about target enzymes is used to predict variants with enhanced thermostability (Huang et al. 2012; Fang et al. 2014). The predicted variants are then enzymatically characterized in vitro to confirm their actual thermostability. This approach is effective for narrowing the list of candidates for enzymatic characterization but is inapplicable without structural information. The other approach uses random mutagenesis in vitro to construct an enzyme variant library followed by screening this library to identify thermostable variants. This approach often requires enzymatic characterization of numerous variants in vitro (Asako et al. 2008; Ben Mabrouk et al. 2013; Stephens et al. 2014). However, in several cases, thermophiles have been used as library hosts to facilitate the screening process because thermostable enzymes are distinguishable from thermolabile enzymes during the growth of thermophiles at high temperatures when target enzyme activities are detected in vivo (Matsumura and Aiba 1985; Liao et al. 1986; Hoseki et al. 1999; Tamakoshi et al. 2001; Fridjonsson et al. 2002; Brouns et al. 2005; Nakamura et al. 2005). Although not a main approach, some thermostable enzymes have also been generated via spontaneous mutations in thermophiles, a process termed as thermoadaptation-directed enzyme evolution. Examples include the generation of thermostable variants of hygromycin B phosphotransferase (Nakamura et al. 2005), α-galactosidase (Fridjonsson et al. 2002), 3-isopropylmalate dehydrogenase (Tamakoshi et al. 2001), and kanamycin nucleotidyltransferase (Liao et al. 1986).

Geobacillus kaustophilus HTA426 (JCM 12893) is an aerobic, Gram-positive, Bacillus-related thermophile for which genetic tools have been established (Suzuki et al. 2012; Suzuki and Yoshida 2012; Suzuki et al. 2013a, b) and a whole genome sequence is available (Takami et al. 2004). In a previous study (Suzuki et al. 2015), we constructed an error-prone thermophile, MK480, from G. kaustophilus HTA426 and then used this strain to examine the thermoadaptation-directed evolution of orotidine 5′-phosphate decarboxylase of Bacillus subtilis 168. This procedure successfully generated two mutant genes encoding thermostable variants, which suggested that this approach may efficiently generate thermostable variants from thermolabile enzymes. However, a single practical example is not sufficient to establish the versatility of this approach.

Therefore, this study examined the thermoadaptation-directed evolution of the chloramphenicol acetyltransferase (CAT) of Staphylococcus aureus. The gene encoding this enzyme is a proven but inefficient genetic marker for Geobacillus spp. presumably because of insufficient enzyme stability at high temperatures.

Materials and methods

Bacterial strains, media, and primers

G. kaustophilus MK242 and MK480 (Table 1) were previously constructed (Suzuki et al. 2015) and used as genetically stable and error-prone strains, respectively. These strains were grown at 60 °C in Luria–Bertani (LB) medium. However, cells carrying plasmids derived from pGKE70 or pGKE75 (see below) were cultured in LB medium supplemented with 5 mg l−1 kanamycin (LK5). Escherichia coli DH5α was used for DNA manipulations. E. coli BL21-CodonPlus(DE3)-RIPL (Novagen, Darmstadt, Germany) was used to produce recombinant proteins. E. coli strains were grown at 37 °C in LB medium. Ampicillin (50 mg l−1) and chloramphenicol (Cm; 50 mg l−1) were added as necessary. The primer sequences are summarized in Table 2.

Table 1 G. kaustophilus strains used in this study
Full size table
Table 2 Primers used in this study
Full size table

Construction of plasmids

The plasmid pGKE70, which is capable of autonomous replication in E. coli and not in G. kaustophilus, was previously constructed (Suzuki et al. 2015). This plasmid was used to integrate genes into the trpE locus (GK2204) of G. kaustophilus and to induce gene expression in G. kaustophilus under the control of the P gk704 promoter (Suzuki et al. 2013b). The plasmid pGKE75 (Fig. 1) was constructed from pGKE73 (Suzuki et al. 2013a) by removing the SphI site in the pBST1 replication region using the QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies, CA, USA) and primers sphU and sphD. This plasmid, which is capable of autonomous replication in both E. coli and G. kaustophilus, was used as an E. coli–Geobacillus shuttle plasmid to induce gene expression in G. kaustophilus under the control of the P gk704 promoter. The cat gene, which encodes S. aureus CAT, was amplified from the plasmid pNW33N (Rhee et al. 2007) using primers catF1 and catR. This gene was then cloned between the SphI and BamHI sites of pGKE70 and pGKE75 to yield pGKE70-cat and pGKE75-cat, respectively. The cat gene was also amplified using primers catF2 and catR and cloned between the NdeI and BamHI sites of pET-16b (Novagen, Darmstadt, Germany) to yield pET-cat.

Fig. 1
figure 1

pGKE75 structure. Plasmid pGKE75, which replicates autonomously, can be shuttled between E. coli and G. kaustophilus. AmpR ampicillin resistance gene, pUC pUC replication region, oriT conjugative transfer origin, TK101, thermostable kanamycin nucleotidyltransferase gene, P gk704 the P gk704 promoter functional in G. kaustophilus (Suzuki et al. 2013b), pBST1 replicon for G. kaustophilus (Taylor et al. 2008). Unique restriction enzyme sites within multiple cloning sites are also indicated

Full size image

Plasmid introduction into G. kaustophilus

Derivatives of pGKE70 and pGKE75 were introduced into G. kaustophilus cells using conjugative DNA transfer from E. coli DH5α (Suzuki and Yoshida 2012; Suzuki et al. 2013a). Transformants carrying pGKE70 derivatives integrated at the trpE locus were selected using kanamycin resistance and tryptophan auxotrophy. Transformants carrying pGKE75 derivatives were selected using kanamycin resistance.

Generation of cat A138T mutant in G. kaustophilus MK480

G. kaustophilus MK480 harboring pGKE75-cat was incubated twice at 65 °C for 24 h on LK5 plates containing 10 mg l−1 Cm. The resulting cells (approximately 104 cells) were collected and then incubated at 65 °C on LK5 plates containing 20 mg l−1 Cm. Similar successive cultures were conducted on LK5 plates containing 30, 40, and 50 mg l−1 Cm (Table 3). Plasmid mixtures were isolated from cells resistant to 50 mg l−1 Cm, inserted into E. coli DH5α, and then reintroduced into G. kaustophilus MK242 using conjugative transfer. The transformants were incubated at 60 °C for 24 h on LK5 plates containing 10 mg l−1 Cm. At this point, pGKE75-cat derivatives were isolated from clones that exhibited obvious Cm resistance. When these plasmids were analyzed for P gk704 -cat sequences, the plasmid pGKE75-cat A138T, which carries a cat mutant (cat A138T), was identified.

Table 3 Cm resistance transition of G. kaustophilus MK480 harboring the indicated plasmid
Full size table

Generation of cat A138S and cat A138V mutants by in vitro site-directed mutagenesis

The upstream and downstream regions of the cat gene were amplified from the plasmid pET-cat using primers t7pro and 138sR (for upstream) and 138sF and t7term (for downstream), respectively. The two fragments were fused by overlap extension polymerase chain reaction (PCR) to yield the mutant gene cat A138S. The upstream and downstream regions were also amplified using primers T7pro and 138vR and 138vF and t7term, respectively, and fused to yield the cat A138V gene.

Preparation of recombinant proteins

E. coli BL21-CodonPlus(DE3)-RIPL harboring pET-cat was aerobically cultured at 20 °C for 16 h in liquid LB medium (100 ml) supplemented with lactose (1 %). Harvested cells were suspended in lysis buffer (10 mM Tris–HCl, pH 8.0; 10 ml) and sonicated. The cell lysate was clarified by centrifugation to eliminate cell debris and then applied to a TALON metal affinity resin column (Takara Bio, Shiga, Japan; 0.1-ml bed) equilibrated with the same buffer. The column was successively washed with buffers containing 20 mM (1 ml), 50 mM (0.5 ml), 100 mM (0.5 ml), and 150 mM (0.5 ml) imidazole. Recombinant proteins were subsequently eluted with a buffer containing 200 mM imidazole and dialyzed against storage buffer (10 mM Tris–HCl, pH 7.8, 0.2 M NaCl). The mutant cat genes cat A138T, cat A138S, and cat A138V were also cloned in pET-16b and used to prepare CATA138T, CATA138S, and CATA138V proteins, respectively, using the same procedure described above for the preparation of CAT. We assessed protein purity using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Protein concentration was determined using the Bradford method with bovine serum albumin as the standard.

CAT activity assay

CAT catalyzes the transfer of an acetyl group from acetyl-CoA to Cm, producing acetylated Cm and CoASH. Standard reaction mixtures (0.3 ml) contained 50 mM sodium phosphate (pH 7.0), 0.1 mM Cm, 0.1 mM acetyl-CoA, and 1 mM 5,5′-dithio-bis(2-nitrobenzoic acid). During the reaction, the product CoASH stoichiometrically reduces 5,5′-dithio-bis(2-nitrobenzoic acid) to produce 5-thio-2-nitrobenzoic acid. The mixture was preincubated for 1 min in a quartz cell inserted into a spectrophotometer (UV-2600, Shimadzu, Kyoto, Japan) equipped with a thermostat (TCC-240A, Shimadzu). Adding an enzyme and monitoring, the absorbance of the solution at 412 nm, which arises from 5-thio-2-nitrobenzoic acid, subsequently initiated the reaction. The concentration of 5-thio-2-nitrobenzoic acid was determined using its molar extinction coefficient at 412 nm (13,600 M−1 cm−1). The value was then used to determine the amount of CoASH produced during the reaction. One unit was defined as the amount of enzyme that produces 1 μmol of CoASH per minute. The specific activity was determined from three independent experiments and is presented as the mean ± standard deviation. Kinetic parameters were determined using Lineweaver–Burk analysis from the initial reaction rates when 50–200 μM Cm and 100 μM acetyl-CoA were used as substrates.

Thermostability assay

Protein solutions (50 μl; 0.2 g l−1 protein) were incubated at 65–80 °C for 1 h and then centrifuged (16,000×g for 10 min at 4 °C) to remove aggregated proteins. CAT activity retained in the supernatant was then analyzed. The temperature at which the residual activity equaled 50 % (T 1/2) was determined by linear regression of the thermal inactivation curves (CAT, 74–77 °C; CATA138T, 76–80 °C; CATA138S, 73–78 °C; CATA138V, 75–79 °C). For time-course analyses, protein solutions were incubated at 75 °C for 6 h, centrifuged, and then the supernatant was assayed for CAT activity.

Gel filtration column chromatography

Gel filtration was performed using an ÄKTA pure 25 system (GE Healthcare Japan) equipped with a Superdex 75 10/300 GL column (GE Healthcare Japan, Tokyo, Japan). Proteins were separated by isocratic elution in buffer (50 mM Tris-HCl, pH 7.0, 0.15 M NaCl) at a flow rate of 0.5 ml min−1.

Homology modeling of CAT structures

We built three-dimensional models of CAT and its variants, with the MODELLR program (http://salilab.org/modeller) using the crystal structure of E. coli CAT (PDB ID, 3CLA; sequence identity to S. aureus CAT, 41 %) as a template. Intramolecular interactions in the model structures were analyzed using the PIC program (http://pic.mbu.iisc.ernet.in).

Cm resistance assay

G. kaustophilus MK242, harboring either pGKE75-cat or pGKE75-cat A138T, was incubated at 50–70 °C for 24 h on LK5 plates with or without Cm (5 mg l−1). The resulting colonies were counted to determine the Cm resistance efficiency, which was defined as the ratio of the number of Cm-resistant colonies (grown on plates with Cm) to the total number of colonies (grown on plates without Cm).

Results

Thermoadaptation-directed evolution using previous procedures

Following the procedure described in the previous report (Suzuki et al. 2015), the cat gene was first integrated into the trpE locus of G. kaustophilus MK480 using plasmid pGKE70-cat. The resultant cells grew at 45 °C in liquid LK5 containing 10 mg l−1 Cm after 2 days of incubation. However, they did not grow at 60 °C, probably because the CAT protein produced in these cells was inactivated by thermal denaturation. However, these cells became Cm resistant after successive culture at 60 °C in the presence of 1, 5, 15, and 20 mg l−1 Cm. The resultant cells formed about 50 substantial colonies on LK5 plates containing 10 mg l−1 Cm even at 65 °C. The P gk704 -cat genes integrated at the trpE loci of 18 of these clones were amplified and sequenced. The results showed that none of the sequences contained mutations. This indicates that MK480 cells can become resistant to Cm without cat mutations through successive culture in the presence of the antibiotic.

Thermoadaptation-directed evolution using pGKE75-cat

To promptly distinguish between thermoadaptation-directed cat mutations and other factors responsible for Cm resistance at high temperatures, we exploited pGKE75-cat. This plasmid is capable of autonomous replication in both E. coli and G. kaustophilus and thereby shuttles between these bacteria. G. kaustophilus MK480 harboring pGKE75-cat was successively cultured at 65 °C in liquid LK5 containing 10–100 mg l−1 Cm. Successive culture was repeated over 13 times, generating cells stably resistant to 15 mg l−1 Cm at 70 °C. The pGKE75-cat derivatives were extracted from this cell mixture and reintroduced in G. kaustophilus MK242, which is genetically stable, unlike MK480. However, when the transformants were incubated at 60 °C on LK5 plates containing 10 mg l−1 Cm, no cells exhibiting obvious Cm resistance were identified. We also analyzed the P gk704 -cat sequences from several pGKE75-cat derivatives and confirmed that none of the plasmids contained mutations. These results suggest that pGKE75-based mutations can facilitate the elimination of false-positive clones (Cm-resistant cells because of factors other than cat mutations) and that successive culture in liquid media may be unsuitable for generating cat mutations.

Thermoadaptation-directed evolution using pGKE75-cat and solid media

Cells producing functional CAT proteins inactivate Cm by acetylation and thereby reduce the concentration of intact Cm in the immediate environment. In liquid medium, however, Cm diffuses and contacts all cells equally. This may result in a weak growth bias for cells producing functional CAT. Based on this hypothesis, we cultured G. kaustophilus MK480 harboring pGKE75-cat on LK5 plates supplemented with Cm (Table 3). Until the fifth culture, cells grew poorly, with huddled colonies, only when spread on the plates at high cell density. However, at the sixth culture, a few of the cells (about 1 % of the total) formed substantial colonies when spread at low density. At that point, the pGKE75-cat derivatives were extracted from the cells and reintroduced into G. kaustophilus MK242 to eliminate false-positive clones. The cells were incubated at 60 °C on LK5 plates containing 10 mg l−1 Cm to check for Cm resistance. A total of 16 clones obviously resistant to Cm were identified. The pGKE75-cat derivatives were isolated from eight clones of these clones, and their P gk704 -cat sequences were analyzed. All eight carried the cat A138T mutant gene, which contains one mutation (G412A) responsible for an A138T replacement in the amino acid sequence (Fig. 2a). These plasmids had no mutations in the P gk704 promoter region.

Fig. 2
figure 2

Preparation of CAT protein and its variants. a Mutations in cat A138T, cat A138S, and cat A138V genes are shown in comparison with the cat gene. Deduced amino acid sequences are also indicated with underlined codons. b Purified recombinant CAT (lane 1), CATA138T (lane 2), CATA138S (lane 3), and CATA138V (lane 4) proteins were analyzed by SDS-PAGE with molecular markers (lane M; kDa)

Full size image

Thermostability of the CAT and CATA138T proteins

The cat and cat A138T genes were expressed in E. coli cells. The gene products, which contained N-terminal His10-tags, were purified to homogeneity by immobilized metal ion affinity chromatography (Fig. 2b). To assess their thermostability, protein solutions were incubated at elevated temperatures and analyzed for residual CAT activity (Fig. 3a). The CAT protein was completely inactivated when incubated at 77 °C, whereas CATA138T activity was substantially retained (47 %). Complete inactivation of CATA138T was observed at 80 °C. The T 1/2 values of CAT and CATA138T were 74.5 and 76.9 °C, respectively. In a time-course analysis, CAT was completely inactivated after 3 h of incubation at 75 °C, whereas CATA138T retained 45 % of its activity (Fig. 3b). These results show that CATA138T is more thermostable than CAT.

Fig. 3
figure 3

Thermal inactivation curves (a), time-course of inactivation (b), and temperature-dependent specific activity (c) of CAT (solid circles) and CATA138T (hollow circles). a After the protein solution was incubated at the indicated temperatures for 1 h, CAT activity retained in the supernatant was analyzed. The activity remaining after incubation at 65 °C was considered to be 100 %. b After the protein solution was the incubated at 75 °C for the indicated amount of time, CAT activity retained in the supernatant was analyzed. The activity without incubation was considered to be 100 %. c CAT-specific activity was determined at the indicated temperatures. Data are presented as the mean ± standard deviation (n = 3)

Full size image

Catalytic activity of the CAT and CATA138T proteins

The catalytic activities of CAT and CATA138T were analyzed at 30–70 °C. The enzymes exhibited almost identical temperature profiles, with their highest activity at 50–60 °C (Fig. 3c). The specific activities of CAT and CATA138T at 60 were 210 ± 5 and 200 ± 9 U mg−1, respectively. The kinetic parameters of CAT (k cat, 209 ± 17 s−1; K m for Cm, 162 ± 17 μM) and CATA138T (k cat, 193 ± 41 s−1; K m for Cm, 153 ± 50 μM) were also comparable at 60 °C. Thus, the A138T substitution had no effect on CAT activity.

Cm resistance conferred by the cat and cat A138T genes

G. kaustophilus MK242 carrying pGKE75-cat or pGKE75-cat A138T was analyzed for Cm resistance efficiency at 50–70 °C (Fig. 4). The efficiency conferred by cat A138T was twice that conferred by cat at 55–60 °C, although the cat gene conferred a slightly higher efficiency than cat A138T at 50 °C. Neither cat nor cat A138T conferred substantial Cm resistance above 65 °C. These data indicated that CATA138T is also more thermostable than CAT in vivo.

Fig. 4
figure 4

Cm resistance conferred by the cat and cat A138T genes. G. kaustophilus MK242 cells harboring pGKE75-cat (solid bars) or pGKE75-cat A138T (hollow bars) were incubated at 50–70 °C on LK5 plates with or without Cm (5 mg l−1). The Cm resistance efficiency was defined as the ratio of the number of Cm-resistant cells to the total number of cells. Data are presented as the mean ± standard deviation (n = 4). G. kaustophilus MK242 cells harboring pGKE75 had <0.1 % efficiency at all temperatures examined

Full size image

Enzymatic properties of CATA138S and CATA138V variants

To determine how the A138T replacement contributes to CATA138T thermostability, two CAT variants (CATA138S and CATA138V; Fig. 2) were prepared and their thermostability and activities were assessed. As shown in Fig. 5, CATA138S and CATA138V were more thermostable than CAT, with T 1/2 values of 75.3 and 76.5 °C, respectively. This result suggests that both of the hydroxyl group and the bulk of the threonine side chain contribute to CATA138T thermostability. With regard to specific activity, CATA138V had a somewhat higher specific activity at 60 °C (240 ± 7 U mg−1) than CAT and CATA138T, whereas CATA138S had a somewhat lower specific activity (140 ± 6 U mg−1).

Fig. 5
figure 5

Thermal inactivation curves (a) and time-course of inactivation (b) of CATA138S (solid circles), CATA138V (hollow circles), and CAT (dashed lines). a After the protein solution was incubated at the indicated temperatures for 1 h, CAT activity retained in the supernatant was analyzed. The activity remaining after incubation at 65 °C was considered to be 100 %. b After the protein solution was incubated at 75 °C for indicated time, CAT activity retained in the supernatant was analyzed. The activity without incubation was considered to be 100 %

Full size image

Subunit structures of CAT and its variants

Gel filtration analysis of recombinant CAT, CATA138T, CATA138S, and CATA138V proteins showed that all four exhibited single protein peaks with retention volumes almost identical to that of the standard conalbumin (75 kDa). This result is consistent with previous observations (Shaw and Brodsky 1968). Therefore, it is likely that these proteins form homotrimers, as does E. coli CAT (Leslie et al. 1988; Leslie 1990).

Discussion

Based on the known mutation frequency in G. kaustophilus MK480 (Suzuki et al. 2015), this strain presumably generates 10−5–10−4 mutations per gene, per culture. Thus, culture of this strain produces minor cell populations (<0.01 %) that may have a mutation in a gene and a major cell population (>99.99 %) that does not. This mutation frequency seems low, but it could allow the generation of a substantial number of mutants (e.g., 102–103 mutants from 107 cells). In this study, we examined the thermoadaptation-directed evolution of S. aureus CAT in the MK480 strain and successfully generated the mutant gene cat A138T by six successive cultures using improved procedures (i.e., plasmid-based mutations on solid media). The MK480 cells formed substantial colonies beginning with the sixth culture. This observation indicates that the cat A138T mutant was generated during the fifth or sixth culture by just a matter of odds and that no mutants encoding thermostable CAT variants were generated before then. As expected, enzymatic characterization showed that the CATA138T protein is more thermostable than CAT. Moreover, the cat A138T gene conferred Cm resistance to G. kaustophilus MK242 more efficiently than did the cat gene (Fig. 4). The cat gene has been used as a selectable marker for Geobacillus spp., but successful selection requires either an inconvenient two-step incubation (e.g., at 50 °C followed by 60 °C) or a long-term incubation (e.g., at 48 °C for 5 days) (Imanaka et al. 1982; Wu and Welker 1989). The cat A138T gene may, therefore, be more useful than cat as a selectable marker for genetic modifications in Geobacillus spp. Notably, CATA138T maintained efficient catalytic activity comparable to that of CAT (Fig. 3). This result shows that thermoadaptation-directed evolution using the MK480 strain can generate a mutation that enhances enzyme thermostability without affecting catalytic activity.

CAT members are currently classified into three types on the basis of their substrate specificities (Biswas et al. 2012). S. aureus CAT is classified as type III. The three-dimensional structure of another type III CAT, from E. coli, has been determined by X-ray crystallographic analysis (Leslie et al. 1988; Leslie 1990). We used homology modeling based on a crystal structure of E. coli CAT to create structural models of S. aureus CAT and CATA138T and then used these models to analyze the molecular mechanisms responsible for the enhancement of thermostability by the A138T replacement. In the model of S. aureus CAT (Fig. 6a), residue A138 resides on β-sheet 7 and is located on the protein surfaces far from the active center. Although the side chains of I134, P135, and L164 surround this residue, no hydrophobic interactions were observed within 4 Å. However, in the model of CATA138T (Fig. 6b), the hydroxyl group of T138 forms hydrogen-bonding interactions with the backbone amides of residues E84 (oxygen atom) and L86 (nitrogen atom). In addition, the methyl group in the T138 side chain appears to gather I134, P135, and L164 side chains using hydrophobic interactions. Because CATA138S and CATA138V showed higher thermostability than CAT, we also constructed models of CATA138S and CATA138V. In the initial CATA138S model (Fig. 6c), the hydroxyl group of S138 formed no hydrogen bonds. However, hydrogen bonds were possible with the backbone amides of residue E84 when the hydroxyl group was rotated. In the CATA138V model (Fig. 6d), hydrophobic interactions were observed within 4 Å among the side chains of V138, I134, P135, and L164. Taken together, these models suggest that the mutations in CATA138S and CATA138V enhanced thermostability by improving hydrogen-bonding and hydrophobic interactions, respectively. Likewise, the mutation in CATA138T enhanced thermostability by improving both interactions, thereby making the structure more robust. Meanwhile, it is likely that the A138T replacement had no effect on the flexibility of the active center, because CATA138T maintained efficient activity.

Fig. 6
figure 6

Structural models of the area surrounding residue 138 of CAT (a), CATA138T (b), CATA138S (c), and CATA138V (d). The structures were constructed using homology modeling and rendered using PyMOL software. The side chains of I134, P135, L164, and residue 138 are shown using sticks with spheres. Residue 138 is shown in yellow. Sticks show the backbone amide bonds of L86 and E84. Oxygen and nitrogen atoms are rendered in red and blue, respectively. Hydrogen bonds involving the hydroxyl group of T138 are shown as dashed lines (color figure online)

Full size image

We note that the in vitro random mutagenesis approach could potentially identify the A138T replacement as a thermostability enhancement for the CAT protein. However, spontaneous mutation in the MK480 strain has the following advantages for generating thermostable enzyme variants, compared with in vitro random mutagenesis. During directed evolution, mutant libraries are constructed with low mutation frequency (a few mutations per gene), which is preferred because enzymes retaining function appear to decline exponentially with increasing numbers of amino acid substitutions (Bloom et al. 2005; Drummond et al. 2005). Whereas only a small number of random mutations can be made at a time using in vitro random mutagenesis (Bloom et al. 2005), MK480 cells generate mutations only through cultivation, without the construction of random mutant libraries. This allows simple operations and the generation of a substantial number of mutants while maintaining a low mutation frequency. Moreover, MK480 cells can simultaneously select active thermostable clones using appropriate selection pressures and high-temperature culture. Thus, thermoadaptation-directed evolution using the MK480 strain is convenient for readily identifying a crucial mutation for moderate thermostability enhancement.

Spontaneous mutagenesis rarely causes adjacent mutations within one amino acid codon and hence seldom generates several amino acid replacements (e.g., only a GCT → TGG mutation responsible for Ala → Trp replacement). However, once a crucial mutation is identified, further enhancement of thermostability and improvement in catalytic activity may be possible via the saturation mutagenesis approach targeting the original mutation site (Asako et al. 2008; Labrou 2010). This idea is supported by the fact that the CATA138V variant exhibited somewhat higher activity than CAT or CATA138T. In vitro random mutagenesis, which can generate mutations with high frequency and thereby explore multiple interacting mutations inaccessible to low frequency mutagenesis (Drummond et al. 2005), is also a powerful approach for further enhancement of thermostability. Moreover, when the S. aureus CAT structure is elucidated, it may be possible to enhance CATA138T thermostability via a rational design approach, optimizing the amino acid configuration surrounding the T138 residue. Therefore, cooperation among protein engineering, using the MK480 strain and these other approaches, would further expand the opportunities for generating useful thermostable enzymes.

In a previous study (Suzuki et al. 2015), mutations occurred at a gene in the MK480 chromosome. However, cat mutations were generated only when cultured on solid media. Although it is unclear why growth on solid media was efficient for mutations, definite growth biases based on selective pressures could be an important advantage of solid media. In addition to the strategic switch from liquid to solid media, the identification of cat A138T required a switch from chromosome-based mutations to plasmid-based mutations. Because plasmids can be readily isolated and reintroduced into fresh host cells to examine the activity and thermostability of encoded proteins, in contrast to genes integrated into chromosomes, this strategy efficiently eliminated false-positive clones that adapted to Cm by factors other than cat mutation. Moreover, plasmids enabled direct sequencing of target genes without preliminary PCR amplification. This avoids PCR errors, which are indistinguishable from gene mutations during successive culture. The fact that the generation of cat A138T was only achieved by using the improved procedure suggests its crucial role in certain cases.

This study has also revealed two new challenges for thermoadaptation-directed evolution using the MK480 strain. One challenge is the formation of biased mutations. Together with the previous study, which generated orotidine-5′-phosphate decarboxylase variants (Suzuki et al. 2015), two C/G → T/A transitions and one C/G → A/T transversion occurred in the target genes. Such biased mutations reflect intrinsically frequent C/G → T/A transition in G. kaustophilus (Suzuki et al. 2015). For a more equal distribution of mutations, it may be effective to expose cells to mutagenic agents such as hydrogen peroxide or N-ethyl-N-nitrosourea during successive culture. Mutagenic agents could also enhance mutation frequencies.

The other challenge is involuntary adaptation. Although MK480 cells harboring pGKE75-cat A138T were able to grow in media containing 50 mg l−1 Cm at 65 °C, MK242 cells that were freshly transformed with pGKE75-cat A138T could not grow under the conditions. In also thermoadaptation-directed evolution using previous procedures or pGKE75-cat with liquid media, MK480 cells harboring cat gene acquired apparent Cm resistance without cat mutations. These observations suggest that G. kaustophilus can adapt to Cm due to factors other than cat mutations. The plasmid-based method effectively eliminated false-positive clones that adapted to Cm, but such adaptation is still troublesome because it reduces selective pressure. This adaptation may involve rRNA gene mutations that circumvent Cm actions targeting the bacterial 23S ribosome (Mankin and Garrett 1991) or the activation of drug efflux pumps responsible for extracellular elimination of Cm (McMurry et al. 1994). However, it more likely arises from the induction of protein chaperones (or chaperone-like factors) that suppress CAT thermal denaturation because MK480 cells without the cat gene remained sensitive to Cm even after similar successive culture (Table 3). The identification of protein chaperones and/or chaperone-like factors induced in MK480 cells may be helpful in clarifying the mechanism of adaptation. This may also help to overcome the adaptation challenge and extend the utility of thermoadaptation-directed enzyme evolution in strain MK480.