Characterization of an Extremely Thermostable Restriction Enzyme, PspGI, from a Pyrococcus Strain and Cloning of the PspGI Restriction-Modification System in Escherichia coli

Abstract

An extremely thermostable restriction endonuclease, PspGI, was purified from Pyrococcus sp. strain GI-H. PspGI is an isoschizomer of EcoRII and cleaves DNA before the first C in the sequence 5′ ^CCWGG 3′ (W is A or T). PspGI digestion can be carried out at 65 to 85°C. To express PspGI at high levels, the PspGI restriction-modification genes (pspGIR and pspGIM) were cloned in Escherichia coli. M.PspGI contains the conserved sequence motifs of α-aminomethyltransferases; therefore, it must be an N4-cytosine methylase. M.PspGI shows 53% similarity to (44% identity with) its isoschizomer, M.MvaI from Micrococcus variabilis. In a segment of 87 amino acid residues, PspGI shows significant sequence similarity to EcoRII and to regions of SsoII and StyD4I which have a closely related recognition sequence (5′ ^CCNGG 3′). PspGI was expressed in E. coli via a T7 expression system. Recombinant PspGI was purified to near homogeneity and had a half-life of 2 h at 95°C. PspGI remained active following 30 cycles of thermocycling; thus, it can be used in DNA-based diagnostic applications.

Since the discovery of the first type II restriction endonuclease (24), these enzymes have played important roles in creating recombinant DNA molecules (7, 21). Over 100 type II restriction-modification (R-M) systems have been cloned so far (22). Among the cloned restriction endonucleases, some enzymes with the same or related DNA recognition sequences have similar amino acid sequences (30 to 100% identity) (30). Weak amino acid sequence similarities (16 to 20% identity) have been reported among some nonisoschizomers (18, 23). The methylases of different R-M systems are more conserved. Nine conserved sequence motifs were found among the aminomethylases (N4-cytosine and N6-adenine methylases) (14), and 10 were found among the cytosine-5 methylases (20). The aminomethylases are separated into three groups, α, β, and γ, based on the circular permutation of conserved motifs (14, 30).

Restriction enzymes BsoBI and AvaI have been used in isothermal strand displacement amplification to detect pathogens such as Mycobacterium tuberculosis (16, 29). Many thermostable restriction enzymes have been isolated from Thermus species and Bacillus stearothermophilus (1, 22). To isolate highly thermostable restriction enzymes, we screened extreme thermophiles living near deep-sea vents. One such isolate, Pyrococcus sp. strain GI-H, displays restriction enzyme activity in its cell extracts. Here we report the characterization of this extremely thermostable restriction enzyme, PspGI, and the cloning and expression of the pspGIR and pspGIM genes in Escherichia coli. Recombinant PspGI was purified, and its activity at high temperatures was determined.

MATERIALS AND METHODS

Bacterial strains.

ER1821(λDE3) is a T7 expression strain that is also deficient in methylation-dependent restriction (McrBC⁻ Mrr⁻ McrA⁻). Pyrococcus sp. strain GI-H was isolated from sediments near an oceanic hydrothermal vent and provided by H. W. Jannasch, Woods Hole Oceanographic Institute, Woods Hole, Mass. (26). Pyrococcus sp. strain GI-H cells were propagated in a medium comprised of 0.5× Difco marine broth mixed with an equal volume of Difco sea salts (40 g/liter), 0.01 M cysteine, 0.005 M BTP, and 10 g of sulfur per liter and incubated at 85°C in flasks without aeration or agitation. Cells in the late log phase of growth were collected by centrifugation and stored at −70°C until use.

Purification of PspGI, N-terminal amino acid sequencing, and thermocycling.

Ten grams of cell paste was resuspended in 35 ml of sonication buffer (20 mM Tris-HCl [pH 7.5], 1 mM dithiothreitol [DTT], 0.1 mM EDTA) and lysed by sonication. PspGI was purified by chromatography through heparin-Sepharose, Mono Q, and Mono S (Pharmacia Biotech, Piscataway, N.J.), heparin-TSK (TosHass, Montegomeryville, Pa.), and Polycat A (Custom LC, Inc.) fast protein liquid chromatography columns. Column fractions were assayed for PspGI restriction activity on T7 phage DNA at 65°C for 1 h (the assay temperature of 65°C was used before the optimal temperature for PspGI digestion was known). The peak fractions of restriction enzyme activity were pooled. Proteins were resolved by sodium dodecyl sulfate (SDS)–10 to 20% polyacrylamide gel electrophoresis (PAGE) and detected by Coomassie blue staining. Purified PspGI was subjected to electrophoresis and electroblotted in accordance with published procedures (15). The membrane was stained with Coomassie brilliant blue R-250, and the protein band of approximately 31 kDa was excised and subjected to sequential degradation on an Applied Biosystems model 407A protein sequencer.

Cell extracts containing recombinant PspGI were heated at 70°C for 30 min. Denatured proteins and insoluble materials were removed by centrifugation. PspGI was purified by chromatography through heparin-Sepharose and Source Q columns.

The half-life of PspGI was measured by incubating the enzyme in 1× buffer 3 (New England Biolabs, Inc. [NEB]) plus 0.1% Triton X-100 and 100 μg of bovine serum albumin per ml at 95°C for 4 h. Samples were taken every 20 min and stored at 4°C until activity assays.

To measure the thermostability of PspGI, 100 U of the recombinant enzyme was subjected to 30 cycles of thermocycling at 95°C for 30 s, 60°C for 30 s, and 72°C for 30 s. Following thermocycling, 1 to 10 U of the original input enzyme was used to cleave 1 μg of T7 phage DNA at 75°C for 1 h.

Two primers were used to amplify the coding sequence from genomic DNA by PCR. The forward primer, based on N-terminal amino acid residues 4 to 10, had the following sequence: 5′ GTT GGA TCC AAC CTN GTN ATH GAY ATH 3′. The reverse primer, based on residues 21 to 27, had the following sequence: 5′ GTT CTG CAG GCY TCR TAD ATD ATY TCR TT 3′ (where R is A or G; Y is C or T; N is A, C, G, or T; H is A, C, or T; and D is A, G, or T). The PCR amplification conditions were as follows: 95°C for 3 min for one cycle; four cycles of 95°C for 20 s, 38°C for 30 s, and 72°C for 5 s; and then 20 cycles of 95°C for 20 s, 56°C for 30 s, and 72°C for 5 s.

Mapping of PspGI cleavage sites.

PspGI cleavage sites on T7 phage DNA were mapped by double digestion of T7 phage DNA with PspGI and endonucleases which cleave at known positions. PspGI restriction digestion was carried out at 75°C. One unit of PspGI is defined as the amount of enzyme required to digest 1 μg of T7 phage DNA to completion at 75°C for 1 h. The activity assay buffer (high-salt buffer) contained 100 mM NaCl, 10 mM MgCl₂, 1 mM DTT, 100 μg of bovine serum albumin per ml, and 0.1% Triton X-100. Medium-salt buffer was the same as high-salt buffer except that it contained 50 mM NaCl. Low-salt buffer contained 10 mM bis-Tris-propane-HCl, 10 mM MgCl₂, and 1 mM DTT.

DNA sequence analysis, inverse PCR, and plasmid construction.

Plasmid DNA was sequenced by the dideoxy termination method with an AmpliTaq DNA polymerase dideoxy terminator sequencing kit (PE Applied Biosystems, Foster City, Calif.). Inverse PCR was carried out as described previously (19). Inverse PCR products were gel purified and sequenced directly or cloned into pUC19 and then sequenced.

For the construction of pLG-PspGIM, two primers were used to amplify the pspGIM gene from genomic DNA. The forward and reverse primers had the following respective sequences: 5′ TAT GGA TCC GGA GGT GAA AAA AAT GAA GTC ATG GAG AGA GTC ATT TCA 3′ and 5′ GGA GGA TCC TTA ACT CTT GTG TAA TAC AAC AAT GTT 3′ (BamHI sites are underlined). PCR conditions were 95°C for 1 min, 60°C for 1 min, and 72°C for 1 min with 2 U of Vent DNA polymerase for 20 cycles. The PCR product was digested with BamHI and ligated into BamHI-cleaved and calf intestinal phosphatase-treated pLG339 (27). Vector pLG339 contains the pSC101 origin and is compatible with a ColE1 origin. The Dcm⁻ strain GM2163 was used as the host for transformation, since Dcm methylation blocks PspGI digestion. The level of resistance to PspGI digestion of plasmid DNA was used as an indicator of M.PspGI expression and modification in vivo.

For the construction of pAII17-PspGIR, the pspGIR gene was amplified from genomic DNA by use of two primers with the following respective sequences: 5′ GGA GGA GTG CAT ATG GTT AGA AAT CTC CTT ATT GAT ATA ACA 3′ and 5′ GTG GGA TCC TTA CAC AAG AGT TAA TTG TTT TCC TCT TTT 3′ (NdeI and BamHI sites are underlined). PCR conditions were the same as those used for the amplification of the pspGIM gene. The PCR product was digested with NdeI and BamHI and gel purified from a low-melting-temperature agarose gel. Following β-agarase treatment and DNA precipitation, the DNA fragment was ligated into T7 expression vector pAII17 (11) with compatible ends. The ligated DNA was transformed into ER1821(λDE3)/pLG-PspGIM. ER1821(λDE3)/pLG-PspGIM/pAII17-PspGIR cells were cultured to the late log phase, and PspGI production was induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to a final concentration of 0.3 to 0.5 mM for 2 h. Cell extracts were prepared as described previously (31).

RESULTS

Purification of PspGI and mapping of PspGI sites on T7 phage DNA.

Approximately 2,000 U of PspGI endonuclease was purified to near homogeneity with six chromatographic columns. This protein (∼31 kDa) was electroblotted and sequenced. The N-terminal amino acid sequence of the first 28 residues was determined to be (Met)-Val-Arg-Asn-Leu-Val-Ile-Asp-Ile-Thr-Lys-Lys-Pro-Thr-Gln-Asn-Ile-Pro-Pro-Thr-Asn-Glu-Ile-Ile-Glu-Glu-Ala-Ile. The amino acid sequence was used to design degenerate primers for PCR amplification of the coding sequence (see below).

PspGI cleavage sites on T7 phage DNA were mapped to approximate positions 2400 and 8200 by double digestion of T7 phage DNA with PspGI and with endonucleases which cleave at known positions, such as ApaLI, BglII, BstBI, EcoNI, MluI, NruI, and StuI (data not shown). The sequence 5′ CCWGG 3′ (W is A or T) occurs in T7 phage DNA at positions 2366 and 8188. Dideoxy sequence analysis of the terminal base obtained from PspGI cleavage of the DNA substrate indicated that PspGI cleaves before the first C at 5′ ^CCWGG 3′ (data not shown). Thus, PspGI is an isoschizomer of EcoRII and a neoschizomer of BstNI (5′ CĈWGG 3′). Two other enzymes, SsoII and StyD4I, recognize a more degenerate sequence (5′ ^CCNGG 3′) (9, 17).

The recognition sequence of PspGI is the same as that of the Dcm methylase. The Dcm methylase modifies the internal C at the cytosine-5 position in 5′ CCWGG 3′ sites. Plasmids pBR322 and pUC19 prepared from a Dcm⁺ E. coli strain are mostly resistant to PspGI digestion. Lambda DNA is partially resistant to PspGI digestion, probably as the result of undermethylation of Dcm sites. T7 phage DNA is susceptible to PspGI digestion due to inhibition of Dcm methylase activity during T7 phage infection (13).

PspGI digestion can be carried out at 65 to 85°C. PspGI is most active in a restriction buffer with 100 mM NaCl. It has 10% relative activity in a low-salt buffer (no NaCl) and 80% relative activity in a medium-salt buffer (50 mM NaCl) (data not shown).

Cloning of pspGIR and pspGIM genes.

Because Dcm methylase blocks PspGI digestion and the transformation efficiency of the Dcm⁻ strain is rather low, the methylase selection method seemed less preferable for cloning the pspGIM gene. The endo-blue method (5) also failed to clone the functional pspGIR gene (17a), probably due to the partial protection conferred by the Dcm methylase. Therefore, a PCR cloning approach was used to clone directly the N-terminal coding region of the pspGIR gene. The first 28 amino acid residues were determined by protein sequencing of the purified PspGI protein. Degenerate primers were synthesized based on the amino acid sequence and used to amplify codons 4 to 27. The amplified product was cloned and sequenced to confirm that it matched the amino acid sequence derived from N-terminal protein sequencing. Two sets of primers were used to clone the adjacent DNA by inverse PCR. The pspGIR and pspGIM genes were amplified by PCR and resequenced directly from PCR products in the absence of cloning steps. The pspGIR and pspGIM genes are 819 and 1,299 bp long, respectively, and overlap by 15 bp. The predicted molecular mass of PspGI is 32 kDa, which matches closely with the apparent size of 31 kDa estimated from SDS-PAGE. The predicted mass of M.PspGI is 50.4 kDa. The R-M genes in the PspGI and EcoRII systems are convergent, whereas the R-M genes in the SsoII and StyD4I systems are divergent (Fig. 1).

FIG. 1.

Gene organization of PspGI, EcoRII, SsoII, and StyD4I R-M systems and of MvaI. aa, amino acids.

Comparisons of PspGI with EcoRII and SsoII and of M.PspGI with M.MvaI.

The EcoRII R-M system was cloned and sequenced previously (12, 25). SsoII and StyD4I are enzymes with a closely related recognition sequence, 5′ ^CCNGG 3′. The sequences of the SsoII and StyD4I R-M genes were published previously (9, 17). An amino acid sequence comparison of PspGI with EcoRII, SsoII, and StyD4I revealed 34% similarity (20% identity) over a stretch of 87 amino acid residues (Fig. 2; the respective GenBank accession numbers for PspGI, EcoRII, SsoII and StyD4I are AF067805, P14633, P34880, and D73442). The sequence of the mvaIR gene is not yet available for comparison.

FIG. 2.

Amino acid sequence alignment of PspGI (amino acid residues 88 to 184), EcoRII (residues 254 to 338), SsoII (residues 108 to 196), and StyD4I (residues 121 to 209). Identical residues are shown in bold. Similar residues are underlined. The sequence alignment was based on the Bestfit program (4).

Both M.PspGI and M.MvaI belong to the α group of aminomethyltransferases. M.PspGI shows extensive homology (53% similarity and 44% identity) to its isoschizomer M.MvaI (Fig. 3; the GenBank accession number for the mvaIM gene sequence is X16985). Conserved or identical residues are found throughout the proteins, even in the variable region (target-recognizing domain). The most conserved segment lies in N4 methylase motif IV, which contains the characteristic SPPY sequence. At the DNA level, the two methylase genes show 54% identity in their nucleotide sequences.

FIG. 3.

Amino acid sequence alignment of M.PspGI and M.MvaI. Identical or similar amino acid residues are shaded. Four plausible short segment deletions in M.PspGI are indicated by Del1 through Del4. The sequence alignment was made with the program MACAW. TRD, target-recognizing domain.

Expression of PspGI in E. coli and purification of recombinant PspGI.

We attempted to express the pspGIR gene in E. coli in the absence of M.PspGI modification. The pspGIR gene was amplified by PCR and ligated into pUC-based expression vector pRRS, and the ligated DNA was transformed into a Dcm⁺ RR1 variant. After the screening of over 240 transformants, one clone that produced a low level of PspGI in the crude cell extract was isolated. The specific activity of this enzyme was fivefold lower than that of the wild-type enzyme. We concluded that this isolate was probably a mutant. It is possible that the Dcm methylase did not fully modify all Dcm-PspGI sites in vivo, so that there was selection pressure to yield a less active mutant of PspGI.

To construct a premodified host, a BamHI fragment containing the pspGIM gene was cloned in low-copy-number plasmid pLG339 (27). The pspGIM gene was expressed constitutively under the control of the Tc promoter. Four clones with PCR inserts were isolated and shown to be partially resistant to PspGI and BstNI digestion. It was concluded that M.PspGI is partially active at the E. coli growth temperature (37°C). One of the plasmids was used to transform E. coli ER1821(λDE3) to premodify the host chromosome. A PCR fragment containing the pspGIR gene (flanked by NdeI and BamHI sites) was ligated into T7 expression vector pAII17 and transformed into the premodified host. When induced with IPTG, one clone produced approximately 50,000 U of PspGI per g of wet E. coli cells.

Recombinant PspGI was purified by heat denaturation of E. coli proteins at 70°C for 30 min, followed by chromatography. Recombinant PspGI was purified to >90% homogeneity (Fig. 4). Like native PspGI, recombinant PspGI had a specific activity of 3 × 10⁶ U/mg of protein on T7 DNA.

FIG. 4.

SDS-PAGE of purified recombinant PspGI. The apparent molecular mass of PspGI on SDS-PAGE was estimated to be 31 kDa (indicated by an arrow). Lane 1, protein size markers; lanes 2 and 3, E. coli cell extract containing PspGI before and after heat treatment at 70°C, respectively; lane 4, after heparin-Sepharose column chromatography; lane 5, after Source Q column chromatography. The protein gel was scanned with MicroTek ScanMaker and analyzed with the NIH Image analysis program.

PspGI temperature optimum for digestion and half-life at 95°C.

One unit of PspGI was used to cleave 1 μg of T7 phage DNA for 1 h at 25, 37, 50, 65, 75, 80, and 85°C. The DNA products were separated on an agarose gel (Fig. 5). PspGI did not cleave DNA at room temperature (Fig. 5, lane 1). It displayed partial activity at 37 to 65°C (Fig. 5, lanes 2 to 4). One unit of PspGI resulted in a complete digestion pattern at 75 to 85°C (Fig. 5, lanes 5 to 8). The optimum temperature for PspGI digestion is in the range of 75 to 85°C.

FIG. 5.

DNA cleavage activity at different temperatures. One unit of PspGI was used to cleave 1 μg of T7 phage DNA for 1 h at the specified temperatures. Lanes 1 through 7, 25, 37, 50, 65, 75, 80, and 85°C, respectively; lane 8, native PspGI digestion of T7 phage DNA at 85°C.

To measure the enzyme half-life, 1 U of recombinant PspGI was preheated at 95°C for 20 min to 4 h, and then DNA cleavage activity was assayed. The half-life of PspGI at 95°C was found to be approximately 2 h (Fig. 6). In DNA amplification reactions such as PCR, the DNA denaturation step is usually set at 95°C for 30 to 60 s. For a 30-cycle PCR, the enzyme would be heated at 95°C for 15 to 30 min. To test the thermostability of PspGI in DNA polymerase buffers, the enzyme was incubated in 1× Thermopol buffer (NEB) for 30 cycles of thermocycling. PspGI retained 100% activity after the thermocycling (data not shown).

FIG. 6.

Half-life of recombinant PspGI at 95°C. One unit of PspGI was heated at 95°C for 20 min to 4 h, and then the enzyme was used to cleave 1 μg of T7 phage DNA at 75°C. DNA cleavage products were separated on an agarose gel, transferred to a membrane, and detected with a Phototope-star chemiluminescence detection kit. X-ray films were exposed for 10 s, 30 s, 50 s, 1 min, and 2 min to obtain a linear response range and were scanned.

PspGI exhibits star activity when large numbers of units are used in DNA digestion. PspGI star activity was detected when >128 U of PspGI was used to digest 1 μg of T7 phage DNA. Minimal PspGI star activity was detected when <64 U of PspGI was used to cleave 1 μg of T7 phage DNA (data not shown).

To test PspGI activity in low-melting-temperature agarose, a T7 DNA band was excised from low-melting-temperature agarose and melted at 65°C. The DNA in the melted agarose was completely digested by PspGI at 75°C (data not shown).

DISCUSSION

Thermostable restriction enzymes such as TaqI isoschizomers have been characterized from Thermus sp. strain SM32 and Thermus filiformis Tok6A1 (1, 2). Native Tsp32II (from Tsp32 activity fraction II containing TaqI isoschizomers) and recombinant Tsp32I display partial activity at 85 to 90°C (2). Recombinant Tsp32I can be added during thermocycling (1a).

PspGI is the most thermostable restriction enzyme discovered so far. To our knowledge, this is the first report of R-M genes cloned and expressed from the extreme thermophile Pyrococcus. Recombinant PspGI has a half-life of approximately 2 h at 95°C. This high degree of thermostability makes this enzyme useful for DNA cleavage during DNA amplification reactions.

In the amino acid sequence analysis of four restriction enzymes that recognize 5′ CCWGG 3′ and 5′ CCNGG 3′, we found a conserved segment of 87 amino acid residues. We propose that this segment is part of a common DNA recognition domain for 5′ CC_GG 3′ If this prediction is true, this motif should also be found in the amino acid sequences of the BstNI, MvaI (5′ CĈWGG 3′), and ScrFI (5′ CĈNGG 3′) restriction endonucleases.

Multiple amino acid substitutions were introduced to change a thermolysin-like protease from B. stearothermophilus to a boiling-resistant extremozyme (28). The amino acid substitutions were mainly located in the surface loop at the N terminus. The substitutions made to the protease (T56 to A, G58 to A, S65 to P, and A69 to P) were considered to stabilize the protein by a reduction of the entropy of the unfolded state. The introduction of a salt bridge or disulfide bond can also increase the thermostability of the protease (28).

M.PspGI shows 53% similarity (44% identity) to M.MvaI. M.PspGI is isolated from Pyrococcus, whereas M.MvaI is derived from a plasmid of the mesophile Micrococcus variabilis. The homologous amino acid sequences between M.PspGI and M.MvaI may reflect functional or structural constraints on the proteins. Conserved aminomethylase motifs have also been found among the type II methylases in the genome of Methanococcus jannaschii (20a). Four small segment deletions may have occurred in M.PspGI to make it more compact or rigid and thermostable (Fig. 3, Del1 through Del4). A comparison of the M.PspGI and M.MvaI amino acid sequences showed that there are 24 Ala residues in M.PspGI and 12 Ala residues in M.MvaI. Both M.PspGI and M.MvaI have 20 Pro residues. The increased number of Ala residues in M.PspGI may contribute to its thermostability. The local tertiary structures resulting from the nonsimilar residues are likely to contribute to the thermostability of M.PspGI.

In the published sequence of Pyrococcus horikoshii OT3 (10), there are four putative type II methylases (PH0338, PH0584, PH0905, and PH1032) that contain conserved motifs of aminomethyltransferases. The putative methylase PH1032 is very similar to M.DpnII (58% similarity and 47% identity); therefore, it may be an M.DpnII isoschizomer. Open reading frames adjacent to the putative methylases may encode the cognate endonucleases. Another putative methylase (PH0039) contains 10 conserved motifs of cytosine-5 methylase (when a TTG codon is used as the start codon for this open reading frame, the translated protein will include cytosine-5 methylase motifs I to X). Two open reading frames upstream of the PH0039 genes may encode the cognate endonuclease. The P. horikoshii OT3 genome does not appear to encode homologs of PspGI and M.PspGI.

The extreme thermostability of PspGI should allow the simultaneous amplification of mutant DNA and cleavage of amplified wild-type DNA. The loss of the PspGI site can be used as an indicator of mutation in the coding or noncoding sequences. A two-step amplification method has been developed to detect ras oncogene mutations in a small fraction of cells (3). This method introduces a restriction site at the codon of interest by use of mismatched primers (e.g., introducing a BstNI site for the detection of a codon 12 mutation in the ras oncogene). The mutation is then detected by the loss of the restriction site (6, 8). With the two-step procedure, the amplified PCR product containing the wild-type sequence was cleaved by BstNI. The amplified mutant DNA was resistant to BstNI digestion and was gel purified and then reamplified with a mutation-specific primer. PspGI and BstNI are neoschizomers (enzymes with the same recognition sequence but cleaving at different bases). Presumably, one should allow adequate time for the completion of restriction digestion before proceeding to the next cycle. Otherwise, the wild-type sequence can also serve as a template for the next round of amplification. For DNA amplification in gene chips, PspGI activity can be minimized at room temperature and can be activated by heating the reaction mixture to 65 to 85°C.