Thermoadaptation-Directed Enzyme Evolution in an Error-Prone Thermophile Derived from Geobacillus kaustophilus HTA426

Abstract

Thermostability is an important property of enzymes utilized for practical applications because it allows long-term storage and use as catalysts. In this study, we constructed an error-prone strain of the thermophile Geobacillus kaustophilus HTA426 and investigated thermoadaptation-directed enzyme evolution using the strain. A mutation frequency assay using the antibiotics rifampin and streptomycin revealed that G. kaustophilus had substantially higher mutability than Escherichia coli and Bacillus subtilis. The predominant mutations in G. kaustophilus were A · T→G · C and C · G→T · A transitions, implying that the high mutability of G. kaustophilus was attributable in part to high-temperature-associated DNA damage during growth. Among the genes that may be involved in DNA repair in G. kaustophilus, deletions of the mutSL, mutY, ung, and mfd genes markedly enhanced mutability. These genes were subsequently deleted to construct an error-prone thermophile that showed much higher (700- to 9,000-fold) mutability than the parent strain. The error-prone strain was auxotrophic for uracil owing to the fact that the strain was deficient in the intrinsic pyrF gene. Although the strain harboring Bacillus subtilis pyrF was also essentially auxotrophic, cells became prototrophic after 2 days of culture under uracil starvation, generating B. subtilis PyrF variants with an enhanced half-denaturation temperature of >10°C. These data suggest that this error-prone strain is a promising host for thermoadaptation-directed evolution to generate thermostable variants from thermolabile enzymes.

INTRODUCTION

Enzymes catalyze numerous reactions that are difficult to perform using chemical catalysts and are commercially utilized in industry (1). The starch industry is a good example; it has a great demand for α-amylase, β-amylase, glucoamylase, pullulanase, and glucose isomerase for producing glucose and its related sugars. In addition, biosensors for monitoring of human blood glucose levels use glucose oxidase or glucose dehydrogenase. Recently developed detergents also contain protease, α-amylase, β-amylase, lipase, and/or cellulase. Thus, enzymes have potential for commercial use, but many are not practical despite possessing catalytic activities suitable for use. One reason is enzyme thermolability, which hinders the prolonged use of enzymes as catalysts and their long-term storage at around room temperature. Although enzymes identified in extreme thermophiles show excellent thermostability, they generally show low activity at moderate temperatures and accordingly possess low utility in applications that require efficient enzymatic activity at moderate temperatures (e.g., as biosensors and in chemical production and medical fields). Therefore, thermostability enhancement of thermolabile enzymes while maintaining catalytic activities at high levels at moderate temperatures is an important approach to expanding the commercial use of enzymes.

Thermostability enhancement has been achieved by two main approaches or their combination (2–4). One approach uses an in silico rational design to predict enzyme variants with enhanced thermostability (3). This approach requires three-dimensional structure information for target enzymes but allows the rapid prediction of candidates. The other approach uses in vitro random mutations to construct an enzyme variant library from which desired variants are screened. This approach is proven but may require in vitro enzymatic characterization of numerous enzyme variants (4). To facilitate the screening process, thermophiles have been used as library hosts because they allow in vivo enzymatic characterization at high temperatures without cell death. Examples are identification of thermoadaptive variants of Escherichia coli hygromycin B phosphotransferase (5), Streptoalloteichus hindustanus bleomycin-binding protein (6), Geobacillus stearothermophilus α-galactosidase (7), Saccharomyces cerevisiae 3-isopropylmalate dehydrogenase (8), and Staphylococcus aureus kanamycin nucleotidyltransferase (9–11) in Thermus thermophilus and/or G. stearothermophilus. Intriguingly, some of these variants have been generated, not via in vitro random mutations, but via spontaneous mutations in thermophiles, termed thermoadaptation-directed enzyme evolution (5, 7–9).

Spontaneous mutations primarily occur through formation of base-base mismatches caused by replication errors and DNA damage, such as deamination, 7,8-dihydro-8-oxoguanine (8-oxoG), and depurination lesions (12, 13). Deamination of adenine produces hypoxanthine and causes A · T→G · C transitions (14, 15), whereas deamination of cytosine and 5-methylcytosine produces U · G and T · G mismatches, respectively, leading to C · G→T · A transitions (12, 16–18). Oxidative damage to guanine produces the aberrant base 8-oxoG, which causes the C · G→A · T transversion because of the ability to form base pairs with adenine and cytosine (19, 20). The 8-oxoG lesion is also observed in the nucleotide pool as 8-oxo-dGTP, from which 8-oxoG can be misincorporated in opposing template adenines, causing the A · T→C · G transversion (20). Replication errors and depurination events can potentially cause any transversion or transition.

Because of the deleterious effects of mutations, base-base mismatches are corrected by several DNA repair systems. In E. coli, mismatches arising from replication errors are corrected by methyl-dependent mismatch repair that involves MutS, MutL, and MutH proteins (21). Although most organisms lack MutH homologs, MutL from species that lack MutH harbors a latent endonuclease activity similar to that of MutH (22), suggesting a MutH-independent mechanism (23). Mismatches arising from DNA damage are commonly corrected through base excision repair pathways, which are initiated by DNA glycosylases that excise damaged bases, such as Ung (uracil-DNA glycosylase) for uracil excision (17, 24), alkyladenine DNA glycosylase for hypoxanthine excision (16), MutM for 8-oxoG excision from C · 8-oxoG mismatches (20, 25, 26), and MutY for adenine excision from A · 8-oxoG mismatches (20, 25, 26). In addition, MutT homologs and transcription-coupled repair systems have been identified in several organisms. MutT degrades 8-oxo-dGTP to prevent 8-oxoG incorporation during DNA replication (20), and transcription-coupled repair corrects damage on the transcribed DNA strand, with preference over nontranscribed strands and nontranscribed double-stranded DNA (27). In E. coli, transcription-coupled repair is mediated by a transcription repair coupling factor encoded by mfd, which facilitates the removal of RNA polymerases stalled at DNA lesions that block transcriptional elongation and the recruitment of DNA repair proteins to lesions on transcribed strands (27, 28).

Recently, we have studied genetic tools for the aerobic, Gram-positive, Bacillus-related thermophile Geobacillus kaustophilus HTA426 with the aim of using the strain for biotechnological applications and biological studies of the genus Geobacillus (29–32). In these studies, HTA426 cells efficiently produced descendants constitutively resistant to the antibiotics rifampin (Rif) and streptomycin (Str). Because Rif and Str resistance (Rif^r and Str^r) often arise from spontaneous mutations in certain genes (33–38), this observation suggests that the HTA426 strain has considerable mutability. Thus, in the present study, we evaluated G. kaustophilus mutability in detail and constructed an error-prone strain from the thermophile. We further investigated the generation of enzyme variants that are more thermostable than the parent enzyme using the error-prone strain.

MATERIALS AND METHODS

Bacterial strains and media.

Table 1 summarizes G. kaustophilus strains. G. kaustophilus MK242 was constructed from strain MK93 (29) using the same procedure as for construction of MK244 from MK72 (31). If not otherwise specified, G. kaustophilus strains were grown at 60°C in Luria-Bertani (LB) medium and minimal medium (MM). MM consisted of K₂SO₄ (0.3 g liter⁻¹), Na₂HPO₄ · 12H₂O (2.5 g liter⁻¹), NH₄Cl (1 g liter⁻¹), MgSO₄ (0.4 g liter⁻¹), MnCl₂ · 4H₂O (3 mg liter⁻¹), CaCl₂ · 2H₂O (5 mg liter⁻¹), FeCl₃ · ·6H₂O (7 mg liter⁻¹), 0.1% trace element solution (39), 10 mM Tris-HCl (pH 7.0), d-glucose (10 g liter⁻¹), and Casamino Acids (1 g liter⁻¹; Becton, Dickinson and Company, Franklin Lakes, NJ). Kanamycin (5 mg liter⁻¹) and uracil (10 mg liter⁻¹) were added when necessary. E. coli DH5α (TaKaRa Bio, Otsu, Japan) and pCR4Blunto-TOPO (Life Technologies, Rockville, MD) were used for DNA manipulations. E. coli BL21(DE3) and pET-16b (Merck KGaA, Darmstadt, Germany) were used for recombinant-protein production. E. coli cells were grown at 37°C in LB medium. Ampicillin (50 mg liter⁻¹) was added when necessary. Bacillus subtilis 168 was obtained from the Bacillus Genetic Stock Center (Columbus, OH) and grown in LB medium at 37°C.

TABLE 1.

G. kaustophilus strains

Strain	Relevant descriptiona	Doubling time (min)b
MK242	Derivative of the wild-type strain HTA426; ΔpyrF ΔpyrR ΔhsdM1S1R1 Δ(mcrB1-mcrB2-hsdM2S2R2-mrr) GK0707::Pgk704-bgaB	ND
MK480	Derivative of the strain MK242; ΔpyrF ΔpyrR ΔhsdM1S1R1 Δ(mcrB1-mcrB2-hsdM2S2R2-mrr) GK0707::Pgk704-bgaB Δ(mutS-mutL) ΔmutY Δung Δmfd	ND
MK242p70	MK242 derivative; trpE::pGKE70	20.2 ± 0.5
MK480p70	MK480 derivative; trpE::pGKE70	21.8 ± 0.5
ΔmutSLp70	MK242 derivative; Δ(mutS-mutL) trpE::pGKE70	20.4 ± 0.8
ΔmutMp70	MK242 derivative; ΔmutM trpE::pGKE70	21.6 ± 0.8
ΔmutYp70	MK242 derivative; ΔmutY trpE::pGKE70	20.0 ± 1.0
ΔmutTp70	MK242 derivative; ΔmutT trpE::pGKE70	21.9 ± 1.8
Δungp70	MK242 derivative; Δung trpE::pGKE70	20.4 ± 0.7
Δmfdp70	MK242 derivative; Δmfd trpE::pGKE70	22.7 ± 1.6
ΔmutSLcm	MK242 derivative; Δ(mutS-mutL) trpE::pGKE70-mutSL	21.9 ± 0.3
ΔmutMcm	MK242 derivative; ΔmutM trpE::pGKE70-mutM	22.6 ± 1.5
ΔmutYcm	MK242 derivative; ΔmutY trpE::pGKE70-mutY	22.9 ± 1.0
ΔmutTcm	MK242 derivative; ΔmutT trpE::pGKE70-mutT	22.9 ± 0.7
Δungcm	MK242 derivative; Δung trpE::pGKE70-ung	20.5 ± 0.9
Δmfdcm	MK242 derivative; Δmfd trpE::pGKE70-mfd	20.7 ± 0.7
MK242BSpyrF	MK242 derivative; trpE::pGKE70-BSpyrFwt	ND
MK480BSpyrF	MK480 derivative; trpE::pGKE70-BSpyrFwt	ND

^aΔ(mutS-mutL) encodes codons mutS_1–106 fused with mutL_534–630. ΔmutM, ΔmutY, ΔmutT, Δung, and Δmfd are deleted in codons mutM_11–255, mutY_14–347, mutT_13–150, ung_16–218, and mfd_33–1060, respectively. trpE is an internal gene for anthranilate synthase component I. The strains MK242 and MK480 lack the native pyrF gene encoding orotidine-5′-phosphate decarboxylase, but MK242^BSpyrF and MK480^BSpyrF harbor BSpyrF_wt, which encodes orotidine-5′-phosphate decarboxylase in B. subtilis.

^bG. kaustophilus strains were cultured at 60°C in LB medium with monitoring of OD₆₀₀. The binary logarithm of OD₆₀₀ data during the logarithmic growth phase was plotted against incubation time to calculate doubling times. The data are presented as means ± standard deviations (SD) (n = 3 to 16). ND, not determined.

Construction of pGKE25 derivatives.

Primer sequences are listed in Table 2. The plasmid constructs pΔmutSL, pΔmutM, pΔmutY, pΔmutT, pΔung, and pΔmfd were used for in-frame deletions of the mutS-mutL, mutM, mutY, mutT, ung, and mfd genes, respectively, in G. kaustophilus MK242. For pΔmutSL construction, the upstream region (1.5 kb) of mutS (GK1306) was amplified using primers 1306upF and 1306upR, and the mutL (GK1307) downstream region (1.5 kb) was amplified using primers 1307dwF and 1307dwR. The two amplified fragments were combined using overlap extension PCR and were cloned into the BamHI site of pGKE25 (31) to yield pΔmutSL. In addition, ΔmutM (GK2728), ΔmutY (GK0463), ΔmutT (GK3067), Δung (GK3421), and Δmfd (GK0048) fragments were generated using appropriate primers and cloned into pGKE25 to yield pΔmutM, pΔmutY, pΔmutT, pΔung, and pΔmfd, respectively.

TABLE 2.

Primers used in this study

Primer	Sequence (5′-3′)	Target region	Underlined site
0048F	GCCGCATGCTTTCGTTGCATCGCTATTTAG	mfd full sequence	SphI
0048R	GCCGGATCCTTATGCCGTCACCGACTTCTC	mfd full sequence	BamHI
0048upF	GCCGGATCCCTTCGTGAAGGCGGGCGCAC	mfd upstream	BamHI
0048upR	GACCATTTCCTCACGCAGCCCGGCGACAAGCTG	mfd upstream
0048dwF	CAGCTTGTCGCCGGGCTGCGTGAGGAAATGGTC	mfd downstream
0048dwR	GGCGGATCCCGACGATGCCATAAATCGGG	mfd downstream	BamHI
0463F	GCCCCTGCAGGCACAAGAGAAACAGAGCG	mutY full sequence	Sse8387I
0463R	GCCGGATCCTTAATCGGGGCGGCGTAC	mutY full sequence	BamHI
0463upF	GCCGGATCCCTTGGATTTGGCGGATGAAC	mutY upstream	BamHI
0463upR	CCAAACACGTTGATGAAACTCGCGCGCTGG	mutY upstream
0463dwF	CCAGCGCGCGAGTTTCATCAACGTGTTTGG	mutY downstream
0463dwR	GCCGGATCCACATTCGTCCATGCGTGCTC	mutY downstream	BamHI
1306F	GGTGTGTACGATGGCATAATCCCCCTCCTGTC	mutSL full sequence
1306R	GACAGGAGGGGGATTATGCCATCGTACACACC	Pgk704 fragment
1306upF	GCCGGATCCTCCACCAAAACGAAAAAGTG	mutS upstream	BamHI
1306upR	CCACGTCGGATGGGAGAGCTGCACAACTTC	mutS upstream
1307R	GCCAGATCTCTACATCACCCGTTTAAATAG	mutSL full sequence	BglII
1307dwF	GAAGTTGTGCAGCTCTCCCATCCGACGTGG	mutS downstream
1307dwR	GCCGGATCCATACCGCGTTTTCCAATAGG	mutS downstream	BamHI
2204F	GCCGAATTCTGACGAATCGCCGTGGGCGC	trpE internal sequence	EcoRI
2204R	GGCGAATTCCTGTCCGGATGGCGATGCAC	trpE internal sequence	EcoRI
2728F	GCCGCATGCCGGAATTGCCGGAGGTG	mutM full sequence	SphI
2728R	GCCGGATCCCTAGCGCTGGCAGCGCGGGC	mutM full sequence	BamHI
2728upF	GCCGGATCCGCCGGAGTGAAAGTGGATAC	mutM upstream	BamHI
2728upR	GACCGTTTTTTCAATGATCGTTTCCACCTC	mutM upstream
2728dwF	GAGGTGGAAACGATCATTGAAAAAACGGTC	mutM downstream
2728dwR	GCCGGATCCTAATCGCAGAATGCGGATTC	mutM downstream	BamHI
3067F	GCCCCTGCAGGCGTGAACGAATTGCAACGGG	mutT full sequence	Sse8387I
3067R	GCCGGATCCTCAGCTCGGATCGAGCCG	mutT full sequence	BamHI
3067upF	GCCGGATCCAATACATGTGGCGCCGACAG	mutT upstream	BamHI
3067upR	GAGCCGATATGAAAGGTTTGTCACCCGTTG	mutT upstream
3067dwF	CAACGGGTGACAAACCTTTCATATCGGCTC	mutT downstream
3067dwR	GCCGGATCCTTCGCCTTGCATACTTCCTG	mutT downstream	BamHI
3421F	GCCGCATGCCGATTCTCAAAAACGAC	ung full sequence	SphI
3421R	GCCGGATCCTCATTCAGCGCGGGCGCCGATG	ung full sequence	BamHI
3421upF	GCCGGATCCCTTGGATGTACCATGCTTTG	ung upstream	BamHI
3421upR	GTTCTCGATTTGCCACTCCTCTTCAAGCAG	ung upstream
3421dwF	CTGCTTGAAGAGGAGTGGCAAATCGAGAAC	ung downstream
3421dwR	GCCGGATCCTTTCATCCGATGCGACGTCC	ung downstream	BamHI
P704F	GCCAAGCTTTTTCTTTTTCCTCCTTTGTTATC	Pgk704 fragment	HindIII
pyrFF	GTACATATGCGAAACAACCTGCCCATC	BSpyrFwt sequence	NdeI
pyrFR	GCCGGATCCTTAAGATTTGATTCCCTCCC	BSpyrFwt sequence	BamHI

Construction of pGKE70 derivatives.

The plasmid pGKE70 (Fig. 1) was constructed to integrate genes into the trpE locus (GK2204) and to force gene expression under the control of the P_gk704 promoter (29). To construct pGKE70, the P_gk704 fragment was excised from pGAM48 (29) and the trpE internal sequence (positions 157 to 1360) was amplified using primers 2204F and 2204R. The P_gk704 fragment was cloned between the HindIII and SphI sites of the plasmid pTK19 (30), and the trpE fragment was then cloned in the MunI site to obtain pGKE70. To construct pGKE70-mutSL, the mutS-mutL and P_gk704 fragments were amplified using primers 1306F and 1307R and P704F and 1306R, respectively. These two fragments were combined to create the P_gk704-mutSL fragment, which was trimmed using HindIII and BglII and cloned between the HindIII and BamHI sites of pGKE70 to generate pGKE70-mutSL. In addition, mutM, ung, and mfd were amplified and cloned between SphI and BamHI sites of pGKE70 to generate pGKE70-mutM, pGKE70-ung, and pGKE70-mfd, respectively. mutY and mutT were amplified and trimmed using Sse8387I and BamHI and subsequently cloned between the PstI and BamHI sites of pGKE70 to generate pGKE70-mutY and pGKE70-mutT, respectively. The gene encoding orotidine-5′-phosphate decarboxylase of B. subtilis (BSpyrF_wt) was amplified from the strain 168 chromosome and subcloned between the SphI and BamHI sites of pGKE70 to generate pGKE70-BSpyrF_wt.

FIG 1.

pGKE70 structure. pGKE70 was constructed to integrate genes into the trpE locus and to force gene expression under the control of the P_gk704 promoter. Amp^R, ampicillin resistance gene; pUC, pUC replicon; oriT, conjugative-transfer origin; TK101, thermostable kanamycin nucleotidyltransferase gene; P_gk704, the P_gk704 promoter functional in G. kaustophilus (29); and trpE, an internal and defective gene for anthranilate synthase component I of strain HTA426. Restriction enzyme sites unique to multiple cloning sites are also indicated.

Plasmid introduction into G. kaustophilus.

Derivatives of pGKE25 and pGKE70 were introduced into G. kaustophilus cells using ternary conjugative transfer from E. coli DH5α (31). Transconjugants integrating pGKE25 derivatives in the chromosome were selected using uracil prototrophy. Transconjugants integrating pGKE70 derivatives at the trpE locus were selected using kanamycin resistance and tryptophan auxotrophy.

Construction of G. kaustophilus mutants deficient in DNA repair genes.

DNA repair genes in G. kaustophilus MK242 were deleted by two-step reciprocal crossovers using pyrF-based counterselection (30) with plasmids pΔmutSL, pΔmutM, pΔmutY, pΔmutT, pΔung, and pΔmfd. The mutants were transformed with pGKE70, yielding the ΔmutSL^p70, ΔmutM^p70, ΔmutY^p70, ΔmutT^p70, Δung^p70, and Δmfd^p70 mutants. Strain MK480 was constructed by simultaneous deletions of mutSL, mutY, ung, and mfd genes in MK242. Strains MK242 and MK480 were also transformed with pGKE70 to generate strains MK242^p70 and MK480^p70, respectively. For gene complementation, ΔmutSL, ΔmutM, ΔmutY, ΔmutT, Δung, and Δmfd mutants were transformed with plasmids pGKE70-mutSL, pGKE70-mutM, pGKE70-mutY, pGKE70-mutT, pGKE70-ung, and pGKE70-mfd to generate the ΔmutSL^cm, ΔmutM^cm, ΔmutY^cm, ΔmutT^cm, Δung^cm, and Δmfd^cm mutants, respectively.

Southern blotting.

Total DNA was digested using restriction endonucleases and separated on a 0.8% (wt/vol) agarose gel by electrophoresis. DNA was transferred to a nylon membrane and hybridized with digoxigenin (DIG)-labeled DNA probes that were synthesized using a PCR DIG Probe synthesis kit (Roche, Basel, Switzerland). Hybridized DNA was detected by the chromogenic method using a DIG Nucleic Acid Detection Kit (Roche).

Mutation frequency assay.

G. kaustophilus, E. coli, and B. subtilis strains were cultured overnight in liquid LB medium and mixed with glycerol to a final concentration of 20%. The mixture was divided into aliquots and stored at −80°C until use. Glycerol stocks were thawed, and an aliquot (10⁶ cells containing <1 Rif^r and Str^r cell stochastically) was inoculated into liquid LB medium (20 ml) in an Erlenmeyer flask (100 ml) with a silicon plug. The flask was incubated with shaking at 180 rpm, and the optical density at 600 nm (OD₆₀₀) was monitored using an OD-MonitorA instrument (Taitec, Saitama, Japan). On reaching early stationary phase (2 h after the OD₆₀₀ reached 1.0), cells were spread on LB plates containing efficacious Rif (10 and 50 mg liter⁻¹ for G. kaustophilus and B. subtilis; 50 and 250 mg liter⁻¹ for E. coli) and Str (10 and 50 mg liter⁻¹ for G. kaustophilus and E. coli; 500 and 3,000 mg liter⁻¹ for B. subtilis). Cells were also spread on LB plates without Rif or Str to determine viable-cell concentrations. After subsequent incubation for 24 h, grown colonies were counted, and the numbers of Rif^r or Str^r cells per 10⁹ viable cells were determined.

Generation of BSpyrF_e1 and BSpyrF_e2.

G. kaustophilus MK242 and MK480 were transformed with pGKE70-BSpyrF_wt to generate MK242^BSpyrF and MK480^BSpyrF, respectively. The strains were cultivated in liquid MM (without uracil) at 60°C for 24 h, and then the grown cells were cultivated at 65°C for 24 h. Cells were spread on MM plates and incubated at 65°C to isolate single clones. From the clones, BSpyrF_wt genes were amplified and sequenced to identify the mutant genes BSpyrF_e1 and BSpyrF_e2.

Preparation of BSpyrF_wt, BSpyrF_e1, and BSpyrF_e2 proteins.

The BSpyrF_wt, BSpyrF_e1, and BSpyrF_e2 genes were amplified using primers pyrFF and pyrFR and cloned in pCR4Blunt-TOPO. After sequence checking, the genes were subcloned between the NdeI and BamHI sites of pET-16b. Using the resulting plasmids, E. coli BL21(DE3) was transformed and cultured at 37°C in liquid LB medium containing 1% lactose. After 6 h of incubation, the cells were harvested and stored at −80°C until use. All of the subsequent operations were performed at 4°C. Cells were suspended in buffer (20 mM sodium phosphate, 0.5 M NaCl, pH 7.4), followed by sonication. The cell lysates were clarified by centrifugation (18,000 × g for 10 min at 4°C) and applied to a Talon column (0.2 ml; TaKaRa Bio) equipped with the same buffer. The column was washed with buffer containing 20 mM imidazole (2 ml) followed by 50 mM imidazole (0.5 ml). Recombinant proteins were subsequently eluted with buffer containing 200 mM imidazole and dialyzed against 10 mM sodium phosphate (pH 7.0) with 0.5 M NaCl. The protein purity was evaluated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Thermostability assay of BSpyrF_wt, BSpyrF_e1, and BSpyrF_e2 proteins.

For analysis of thermal denaturation curves, protein solutions (20 μl; 0.4 g liter⁻¹ protein) were incubated at 30.0 to 55.6°C for 1 h using a gradient temperature heater (Thermal Cycler Dice Gradient; TaKaRa Bio) and then centrifuged (18,000 × g for 10 min at 4°C) to remove aggregated proteins. Proteins retained in the supernatant were analyzed by the Bradford method using bovine serum albumin as a standard. T_1/2, the temperature at 50% residual protein, was calculated by linear regression using thermal denaturation data (BSpyrF_wt, 35.6 to 45.6°C; BSpyrF_e1, 48.0 to 53.2°C; BSpyrF_e2, 48.0 to 54.4°C). For time course analysis, protein solutions were incubated at 37°C for 7 days. After centrifugation, proteins retained in the supernatant were analyzed by the Bradford method.

RESULTS

G. kaustophilus mutability.

G. kaustophilus MK242 was constructed as a laboratory strain that lacks genes related to pyrimidine biosynthesis (pyrF and pyrR) and DNA restriction modification in the HTA426 strain. It was transformed with pGKE70 for coordinating analytical conditions with other G. kaustophilus mutants (see below). The resulting mutant, MK242^p70, was used to assess G. kaustophilus mutability (Fig. 2). The mutability assay was based on generation frequencies of Rif^r and Str^r cells, because Rif^r and Str^r often arise from mutations in rpoB (encoding RNA polymerase β subunit) and rpsL (encoding ribosomal subunit protein S12) genes, respectively (33–38). The generation frequencies of Rif^r cells (per 10⁹ viable cells) from MK242^p70 cells were 930 ± 350 (for 10 mg liter⁻¹ Rif) and 520 ± 270 (for 50 mg liter⁻¹ Rif), and those of Str^r cells were 120 ± 60 (for 10 mg liter⁻¹ Str) and 70 ± 40 (for 50 mg liter⁻¹ Str). These frequencies were higher than those of E. coli and B. subtilis, confirming substantial mutability of G. kaustophilus.

FIG 2.

Mutability of G. kaustophilus, E. coli, and B. subtilis. Generation frequencies of Rif^r or Str^r cells (per 10⁹ viable cells) from G. kaustophilus MK242^p70, E. coli DH5α, and B. subtilis 168 were analyzed using Rif and Str at low (solid bars) and high (open bars) concentrations. Analysis was performed using four independent culture experiments (n = 4), and the data are presented as means and standard errors (SE).

Mutability of G. kaustophilus mutants.

The G. kaustophilus HTA426 genome (40) contains genes that may be involved in DNA repair (mutS, mutL, mutM, mutY, mutT, ung, and mfd). To evaluate gene functions, six deletion mutants (ΔmutSL, ΔmutM, ΔmutY, ΔmutT, Δung, and Δmfd) were constructed. After correct deletions were verified by Southern blotting (Fig. 3), these mutants were further transformed with pGKE70 or its derivatives to construct six mutants (ΔmutSL^p70, ΔmutM^p70, ΔmutY^p70, ΔmutT^p70, Δung^p70, and Δmfd^p70) and their complementary mutants (ΔmutSL^cm, ΔmutM^cm, ΔmutY^cm, ΔmutT^cm, Δung^cm, and Δmfd^cm) and analyzed for generation frequencies of Rif^r and Str^r cells (Fig. 4). All mutants showed growth rates comparable with that of strain MK242^P70 (Table 1), but mutation frequencies were markedly enhanced in the ΔmutSL^p70 (30- to 400-fold), ΔmutY^p70 (400- to 2,000-fold), Δung^p70 (100- to 400-fold), and Δmfd^p70 (100- to 400-fold) mutants. Their complementary mutants showed substantially suppressed frequencies, confirming the absence of polar effects accompanying the deletions. For the mutM and mutT genes, both deletion and complementation had relatively small effects on the frequencies. These data suggest that the mutSL, mutY, ung, and mfd genes have important roles in DNA repair but mutM and mutT have small contributions.

FIG 3.

In-frame deletions in DNA repair genes in G. kaustophilus MK242. Correct deletions to generate the ΔmutSL (A), ΔmutM (B), ΔmutY (C), ΔmutT (D), Δung (E), and Δmfd (F) mutants were verified by Southern blotting. The open boxes represent DNA probes. The dotted lines indicate the deletion regions. The numbers indicate DNA fragment lengths (kb).

FIG 4.

Effects of deletion and complementation of mutSL, mutM, mutY, mutT, ung, and mfd genes on G. kaustophilus mutability. The generation frequencies of Rif^r or Str^r cells were evaluated using Rif and Str at 10 mg liter⁻¹ (solid bars) and 50 mg liter⁻¹ (open bars). Analyses were performed in three to five independent experiments for each condition (n = 3 to 5). The data are presented as means and SE of fold changes relative to the mean generation frequency of the MK242^p70 strain.

Mutations generated in G. kaustophilus.

The rpoB and rpsL genes in Rif^r and Str^r cells that were derived from strain MK242^p70 were sequenced to analyze mutations generated in cells (Table 3). Rif^r cells contained any of the 12 single mutations in the rpoB gene, and Str^r cells contained any of the six single mutations in the rpsL gene. All mutations occurred independently and are probably responsible for the antibiotic resistance. Overall, A · T→G · C and C · G→T · A transitions were predominant, whereas A · T→C · G, A · T→T · A, C · G→A · T, and C · G→G · C transversions were less frequent. We also analyzed Rif^r and Str^r clones derived from ΔmutSL^p70, Δung^p70, ΔmutY^p70, and Δmfd^p70 mutants (Table 3). In addition to single mutations observed in MK242^p70 cells, three single mutations in rpoB and one single mutation in rpsL were identified. Some Str^r cells showed no mutation in rpsL. Cells may contain mutations in rrn for 16S rRNA because it is known that certain rrn mutations result in Str^r in E. coli (37, 38). The ΔmutSL^p70 and Δmfd^p70 mutants showed markedly increased frequencies of A · T→G · C and A · T→C · G mutations, respectively. The ΔmutY^p70 mutant showed high frequencies of the C · G→A · T transversion, suggesting that MutY suppresses the transversion by repairing A · 8-oxoG mispairs, similar to E. coli (20, 25) and B. subtilis homologs (26).

TABLE 3.

Mutations generated in G. kaustophilus mutants

Mutation	Mutation frequencya
	MK242p70	ΔmutSLp70	ΔmutYp70	Δungp70	Δmfdp70	MK480p70
rpoB mutations in Rifr
T1393C	3/48	2/19
C1405A	2/48		1/22	3/21
A1406G					5/20
A1406T	3/48
G1414T	2/48
A1415T	2/48
A1416T	1/48
C1433A			3/22
C1433T	2/48	1/19			1/20
C1444G				1/21
C1444T	14/48	2/19	3/22	3/21	1/20
A1445G	4/48	11/19		1/21
A1445T	1/48
C1460A	1/48	1/19	15/22			24/32
C1460T	13/48	2/19		13/21	13/20	8/32
rpsL mutations in Strr
A167C	2/48				9/21
A167G	18/48	1/22		2/22	10/21
A168T	1/48
A301G	13/48	4/22	6/22	7/22	1/21	20/32
A302G	11/48	10/22	4/22	10/22	1/21	3/32
C311A	3/48	1/22	9/22	1/22
C311T			2/22	1/22		7/32
Not identified		6/22	1/22	1/22		2/32
Mutation spectra
A · T→C · G	2/96 (2.1)				9/41 (22)
A · T→G · C	49/96 (51)	28/35 (80)	10/43 (23)	20/42 (48)	17/41 (41)	23/62 (37)
A · T→T · A	8/96 (8.3)
C · G→A · T	8/96 (8.3)	2/35 (5.7)	28/43 (65)	4/42 (9.5)		24/62 (39)
C · G→G · C				1/42 (2.4)
C · G→T · A	29/96 (30)	5/35 (14)	5/43 (12)	17/42 (40)	15/41 (37)	15/62 (24)

^aG. kaustophilus mutants that were grown on LB plates containing 50 mg liter⁻¹ Rif or Str were analyzed for rpoB and rpsL sequences, respectively. The fraction represents sequences carrying the mutation per total number of sequences determined. Mutation spectra summarize sequences that arise from the mutation per total sequences carrying mutations (excluding those not identified). Values in parentheses indicate percentages (%). Empty cells indicate that the mutation was not observed.

The error-prone thermophile G. kaustophilus MK480.

G. kaustophilus MK480 was constructed from strain MK242 by simultaneous deletions of the mutSL, mutY, ung, and mfd genes, which were confirmed to be genes involved in functional DNA repair systems in G. kaustophilus, as described above. The properties were evaluated using the mutant MK480^p70, which showed 700- to 9,000-fold-increased generation frequencies of Rif^r and Str^r cells (Fig. 5A). The growth curves of MK242^p70 and MK480^p70 were almost identical (Fig. 5B), with comparable doubling times (Table 1). The predominant mutations were A · T→G · C, C · G→T · A, and C · G→A · T (Table 3).

FIG 5.

Comparison of mutability (A) and growth curves (B) between the G. kaustophilus strains MK242^p70 and MK480^p70. (A) Generation frequencies of Rif^r or Str^r cells from MK480^p70 were evaluated using Rif and Str at 10 mg liter⁻¹ (solid bars) and 50 mg liter⁻¹ (open bars). The data are presented as means and SE (n = 5) of fold changes relative to the mean generation frequency of the MK242^p70 strain. (B) The strains MK242^p70 and MK480^p70 were cultured at 60°C in LB medium with monitoring of the OD₆₀₀.

Generation of BSpyrF_e1 and BSpyrF_e2 genes in G. kaustophilus MK480.

Because G. kaustophilus MK242 and MK480 are auxotrophic for uracil owing to a deficiency in the native pyrF gene, we investigated thermostability-directed evolution of the BSpyrF_wt protein (encoded by B. subtilis pyrF) using the strains MK242^BSpyrF and MK480^BSpyrF. The MK242^BSpyrF strain grew at 60°C in liquid MM (without uracil) but not at 65°C, probably because of BSpyrF_wt thermal denaturation (Fig. 6A). Similar observations were made for MK480^BSpyrF. However, after cultivation at 60°C, MK480^BSpyrF cells grew even at 65°C, whereas MK242^BSpyrF cells did not. MK480^BSpyrF cells efficiently formed colonies on MM plates. Among the colonies grown, eight were analyzed for BSpyrF_wt sequences to identify the mutant genes BSpyrF_e1 (6/8 clones) and BSpyrF_e2 (2/8 clones). BSpyrF_e1 carried a C497T mutation responsible for A166V substitutions in the BSpyrF_wt amino acid sequence (Fig. 6B). In addition to the C497T mutation, BSpyrF_e2 carried a C169A mutation responsible for the L57I substitution.

FIG 6.

Generation of thermostable variants BSpyrF_e1 and BSpyrF_e2 from BSpyrF_wt. (A) Phenotype transition from prototrophy to auxotrophy for uracil in G. kaustophilus MK480^BSpyrF. G. kaustophilus MK480^p70, MK480^BSpyrF, and MK242^BSpyrF were incubated in MM at 60°C (left) and 65°C (middle). MK480^BSpyrF and MK242^BSpyrF cells that had been grown at 60°C were incubated in MM at 65°C (right). (B) Mutations in BSpyrF_e1 and BSpyrF_e2 genes. BSpyrF_e1 carried C497T, and BSpyrF_e2 carried C169A and C497T mutations. The C169A and C497T mutations are responsible for L57I and A166V amino acid substitutions, respectively.

Thermostability of BSpyrF_wt, BSpyrF_e1, and BSpyrF_e2 proteins.

Recombinant BSpyrF_wt, BSpyrF_e1, and BSpyrF_e2 proteins were produced as proteins fused with an octahistidine tag at the N terminus and purified to homogeneity using immobilized metal affinity chromatography (Fig. 7A). To assess their thermostability, the protein solution was incubated at high temperatures and analyzed for the amounts of protein free of thermal aggregation (Fig. 7B). The BSpyrF_wt protein was completely aggregated when incubated at 45°C, whereas BSpyrF_e1 and BSpyrF_e2 were almost completely retained in the supernatant. Complete aggregation of BSpyrF_e1 was observed at >53°C and that of BSpyrF_e2 at >54°C. The T_1/2 values for BSpyrF_wt, BSpyrF_e1, and BSpyrF_e2 were 40.3, 50.0, and 51.0°C, respectively, showing that these variants were much more thermostable than the BSpyrF_wt protein. This finding was further supported by time course analysis of protein aggregation (Fig. 7C). After incubation at 30°C for 1 day, BSpyrF_wt was completely aggregated, whereas BSpyrF_e1 retained 77% of its total amount. BSpyrF_e2, which retained 84% at 30°C and 38% at 37°C even after 7 days, was notably more stable. These results suggest that this approach has potential for markedly improving the storage stability of thermolabile enzymes at ordinary temperatures.

FIG 7.

Thermostability assay of BSpyrF_wt, BSpyrF_e1, and BSpyrF_e2 proteins. (A) SDS-PAGE analysis of purified recombinant BSpyrF_wt (lane 1), BSpyrF_e1(lane 2), and BSpyrF_e2 (lane 3) proteins with molecular markers (lane M; kDa). (B) Thermal denaturation curves of BSpyrF_wt (solid circles), BSpyrF_e1 (open circles), and BSpyrF_e2 (solid squares) proteins. The protein solution was incubated for 1 h at the indicated temperatures and centrifuged. Proteins retained in the supernatant were analyzed by the Bradford method. The data are presented as means and SE (n = 4 to 8). The amount of protein remaining after 30°C incubation was taken to be 100%. (C) Time course analysis of BSpyrF_wt (solid circles), BSpyrF_e1 (open circles), and BSpyrF_e2 (solid squares) aggregations. The protein solution was incubated at 30°C (left) and 37°C (right) for the indicated periods. After centrifugation, proteins retained in the supernatant were analyzed by the Bradford method. The data are presented as means (n = 4; SE are less than 3% [not shown]).

DISCUSSION

Among the types of DNA damage, deamination, depurination, and destruction of deoxyribose residues increase at high temperatures (13, 15, 41, 42). Because of such high-temperature-associated DNA damage, thermophiles may undergo more frequent DNA damage during growth, but the hyperthermophile Sulfolobus acidocaldarius exhibits mutability comparable with that of mesophiles, reflecting efficient functioning of DNA repair genes in the hyperthermophile (43). However, our preliminary observation suggests that G. kaustophilus HTA426 may have relatively high mutability, leading us to evaluate its mutability in detail. As expected, a mutation assay using Rif and Str revealed the substantial mutability of G. kaustophilus. This high mutability may partially be due to high-temperature-associated DNA damage and insufficient DNA repair genes, because mutation analysis of G. kaustophilus showed frequent A · T→G · C and C · G→T · A transitions, which can increase at high temperature through deamination of adenine and cytosine, respectively (12, 14–18). Although spontaneous adenine deamination is much less frequent than cytosine deamination in general (15, 16), A · T→G · C transition was more frequent than C · G→T · A transition in G. kaustophilus. In the ΔmutSL^p70 mutant, the frequency of A · T→G · C was as high as 80%. These observations imply a defect in deaminated-adenine (i.e., hypoxanthine) repair in G. kaustophilus. In addition, gene deletion analysis indicated small contributions of mutM and mutT to DNA repair. In E. coli (20) and B. subtilis (26), MutM suppresses the C · G→A · T transversion cooperatively with MutY, and MutT suppresses A · T→C · G transversion. Although MutY and Mfd contribute to suppressing these transversions in G. kaustophilus (Table 3), it is likely that G. kaustophilus is exposed to relatively high risks of the transversions.

We constructed G. kaustophilus MK480 as an error-prone thermophile by deleting functional DNA repair genes that we had identified and investigated thermoadaptive gene mutations in the error-prone strain using BSpyrF_wt as a model. G. kaustophilus MK480^BSpyrF was originally auxotrophic for uracil at 65°C but became prototrophic after culture in MM at 60°C, followed by 65°C, along with BSpyrF_e1 and BSpyrF_e2 generation. This phenotype transition can be explained by generation and selection, during incubation at 60°C, of mutant genes that encode enzyme variants that are more thermostable than the BSpyrF_wt protein. The thermostability assay confirmed that the variants BSpyrF_e1 and BSpyrF_e2 are actually more thermostable than BSpyrF_wt and that BSpyrF_e2 is more thermostable than BSpyrF_e1. These observations indicate that both A166T and L57I substitutions contribute to BSpyrF_wt thermostabilization. In the (β/α)₈ barrel fold of the BSpyrF_wt protein (44), the residues L57 (replaced in BSpyrF_e2) and A166 (replaced in BSpyrF_e1 and BSpyrF_e2) constitute β-strand β3 and α-helix α2, respectively (Fig. 8). The side chain of L57 hydrophobically interacts with the side chains of the F39, V47, and L59 residues to occupy the void between β-strand β3 and α-helix α2 (Fig. 8A). Similar hydrophobic interactions were observed for the A166 residue, the side chain of which interacts with Y170 and T179 to occupy the void between α-helix α6 and β-strand β7 (Fig. 8B). Therefore, it is possible that thermostability enhancement by L57I and A166V substitutions involves optimizing hydrophobic interactions between α-helices and β-strands in the barrel fold, thereby making the structure robust.

FIG 8.

Three-dimensional structures surrounding the L57 (A) and A166 (B) residues of the BSpyrF^wt protein.

Thermoadaptation-directed evolution using strain MK480 requires no construction of random mutant libraries because mutations are generated in cells during successive culture. Although the mutations may be less frequent than mutations obtained using in vitro mutagenesis approaches, one amino acid alteration potentially confers sufficient enhancement of enzyme thermostability, as shown by BSpyrF_e1 and other examples (45). Moreover, strain MK480 can employ in vivo enzymatic characterization at high temperatures to select thermoadaptive enzymes, as described below: (i) when cell growth is accelerated by catalytic activity of target enzymes, candidate genes encoding thermoadaptive enzymes can be concentrated, because thermolabile enzymes undergo thermal denaturation during cell growth at high temperatures, as in BSpyrF_e1 and BSpyrF_e2 generation, and (ii) when catalytic activities of target enzymes can be detected by chromogenic or fluorogenic assay in vivo, candidate genes can be selected by a plate assay without cell death. Even in other cases, candidate genes may be efficiently selected by the activity-independent method described by Chautard et al. (46). The availability of high-temperature in vivo assays is a notable advantage of strain MK480 over error-prone mesophiles for thermoadaptation-directed evolution. In addition, strain MK480 showed 700- to 9,000-fold-increased mutability. This enhancement is comparable to that achieved by abolishing (5,000-fold) or silencing (2,000-fold) DNA repair genes in E. coli (47, 48), while G. kaustophilus had higher intrinsic mutability than E. coli (Fig. 2). Therefore, G. kaustophilus MK480 has higher mutability than E. coli strains engineered for increased mutability, an additional advantage over error-prone mesophiles.

There are some reports of thermoadaptation-directed enzyme evolution in thermophiles (5, 7–9), although not using an error-prone thermophile. A 3-isopropylmalate dehydrogenase variant having five substitutions with a 12°C-higher T_1/2 than the parent enzyme was generated by iterative incubation for approximately 10 days in T. thermophilus (8). T. thermophilus cells were also used to generate an α-galactosidase variant possessing one substitution with an enhanced T_1/2 of 3°C by 11 days of incubation (7). Compared with these examples, strain MK480 generated thermoadaptive variants with excellent thermostability enhancement (>10°C) in a short time (2 days), probably due to high mutability. These data suggest that strain MK480 has an advantage over not only error-prone mesophiles, but also thermophiles used for thermoadaptation-directed evolution to date. It is also noteworthy that in strain MK480, every mutation can theoretically occur, given that although the A · T→C · G, A · T→T · A, and C · G→G · C mutations were not observed in the strain, they were found in the parent strain (MK242^p70) or the Δung mutant (Δung^p70). Moreover, G. kaustophilus allows a wide range of selection temperatures (42 to 74°C), which is especially favorable for thermoadaptation-directed evolution of very thermolabile enzymes. Thus, strain MK480 is a practical resource for generating thermostable enzyme variants by thermoadaptation-directed evolution and provides new opportunities for the facile generation of thermostable enzymes.