Invariant Asp-1122 and Asp-1124 are essential residues for polymerization catalysis of family D DNA polymerase from Pyrococcus horikoshii.

Family D DNA polymerase has recently been found in the Euryarchaeota subdomain of Archaea. Its genes are adjacent to several other genes related to DNA replication, repair, and recombination in the genome, suggesting that this enzyme may be the major DNA replicase in Euryarchaeota. Although it possesses strong polymerization and proofreading activities, the motifs common to other DNA polymerase families are absent in its sequences. Here we report the mapping of the catalytic residues in a family D DNA polymerase from Pyrococcus horikoshii. Site-directed alanine mutants for 28 conserved aspartic acid or glutamic acid residues were screened for polymerization and 3'-5' exonuclease activities. We identified the invariant aspartates Asp-1122 and Asp-1124 within the most conserved motif as the catalytic residues involved in DNA polymerization. Alanine mutation at either site caused a loss of polymerization activity, whereas the conserved mutants, D1122E, D1124N, and D1124E, had slightly reduced polymerization activity. We also found that the 3'-5' exonuclease activity remains in D1122A and D1124A, indicating that the catalytic residues of DNA polymerization are different from those of the 3'-5' exonuclease activity. Furthermore we determined the molecular mass of the recombinant enzyme by gel filtration and proposed a heterotetrameric structure for this enzyme.

two subunits: a small subunit (DP1), which shows low but significant homology to eukaryotes DNA polymerase ␦ (7), and a large subunit (DP2), which is believed to be the catalytic subunit because of its weak polymerization activity (5). The interaction between the two subunits is essential for full enzyme activities. Strong polymerization and 3Ј-5Ј exonuclease activities are obtainable only when the two subunits were mixed or co-expressed together (5). Amino acid sequences of the large subunit have no homology to other DNA polymerases and do not contain common motifs of other DNA polymerase families despite the fact that members of PolD from the Euryarchaeota species are highly conserved. PolD has been mainly characterized in P. furiosus; nevertheless, many aspects of the enzyme properties, such as the catalytic residues for both polymerization and 3Ј-5Ј exonuclease activities, are still unknown.
It is of particular interest that the genes coding the two subunits of PolD in Pyrococcus (P. furiosus, P. horikoshii, P. abyssi, and P. woesei) exist in an operon containing three or four other ORFs that are important in DNA replication, DNA repair, and recombination (6,8). One upstream ORF (PH0124 in P. horikoshii) is a homologue to Orc1 (subunit 1 of the origin recognition complex) of yeast, and a downstream ORF (PH0119) is a homologue to yeast RadB, a protein involved in recombination. Moreover, this operon is located adjacent to a recently identified replication origin in the Pyrococcus genomes (8). The genes of replication factor C (ORF PH0112 and PH0113 in P. horikoshii) and putative DNA helicase (PH0109) are located only 6 -10 kilobase pairs away from the genes of PolD. This clustering of genetically essential genes conserved in Pyrococcus indicates that PolD is likely the main replicase in DNA replication in Euryarchaeota. Because the enzymes involved in DNA replication in Archaea are similar to the replicative enzymes in eukaryotes (9), further characterization of the proteins related to DNA replication, including PolD, will help reveal the DNA replication mechanism in Archaea, as well as that in eukaryotes, which has become complicated during evolution and is difficult to study. Fortunately, the thermostability of the replication proteins and clustering of the replication genes facilitate their analysis. In the present study, we successfully cloned, expressed, and purified the PolD from P. horikoshii (PolDPho). By site-directed mutagenesis, we identified amino acid residues Asp-1122 and Asp-1124 as the essential residues responsible for DNA polymerizing activity. Using gel filtration analysis, we determined the molecular mass of the recombinant PolD from P. horikoshii and proposed that PolD from P. horikoshii probably forms a heterotetrameric structure in solution.

MATERIALS AND METHODS
Chemicals-The pET15b vector and ultracompetent E. coli XL2-Blue MRFЈ cells were purchased from Stratagene (La Jolla, CA). The pGE-MEX-1 vector was obtained from Promega (Madison, WI). The E. coli strain BL21-CondonPlus (DE3)-RIL competent cells were purchased from Stratagene. Vent DNA polymerase was purchased from New England Biolabs (Beverly, MA). Restriction enzymes were purchased from Takara Shuzo (Shiga, Japan) and Promega. The DNA ligation kit was purchased from Takara Shuzo and was used according to the manufacturer's recommendation. T4 polynucleotide kinase was obtained from Promega, and protease inhibitor mixture tablets (EDTA-free) were purchased from Roche Molecular Biochemicals.
Cloning of the Genes and Construction of Expression Vectors-The gene for the small subunit of PolD from P. horikoshii (DP1Pho) was amplified using PCR with primers S1 and S2 (Table I; Fig. 1). The amplified fragment was digested with NdeI and BamHI and inserted into an expression vector pET15b (Novagen, Madison, WI) to produce a vector named pET15b-S. The DP2Pho gene with the intein was split into two parts using PCR (Table I; Fig. 1). The upper part of the DP2Pho gene (DP2U, from 1 to 2347 base pairs) was amplified with primers L1 and L2. The lower part of the DP2Pho gene (DP2L, from 2348 to 4305 base pairs) was amplified using primers L3 and L4. After digestion with the restriction enzymes, DP2U and DP2L were inserted into the pGEMEX-1 vector sequentially to generate a vector pGEMEX-L(ϩ). The intein was removed using the overlap PCR method (10) using primers L3, L6, L5, and L4 ( Fig. 1; Table I), and the DP2L fragment without the intein generated by the overlap PCR replaced the SalI-NsiI fragment in pGEMEX-L(ϩ) to produce a vector pGEMEX-L(Ϫ) that did not have the intein.
To construct the co-expression vector for DP1Pho and DP2Pho, the BamHI site in pET15b-S was modified with PCR using primers S1 and S2M to introduce multiple sites of BamHI, NsiI, SalI, and SacII ( Fig. 1; Table I). The resulting plasmid was named pET15b-SM. To include the ribosomal binding site using PCR, a primer (pGEM-rbs) containing a SacII site was designed according to the sequences of pGEMEX-1 (Table  I; Fig. 1). The DP2Pho gene fragments together with a sequence from pGEMEX-1 were then subcloned sequentially into pET15b-SM. The final co-expression vector was named pET15b/SL (Fig. 1). The sequences of the genes were verified using the LIC-4200L(S)-2 DNA system (Li-Cor, Tokyo, Japan).
Site-directed Mutagenesis-Site-directed mutagenesis was carried out using the overlap PCR method (11) with the common primers L3 and L4 ( Fig. 1; Table I) and two site-specific complementary primers for each mutant. After site-directed mutation with PCR and digestion with SalI and NsiI, the mutant fragments were inserted back into pET15b/ SL. DNA sequencing was used to confirm that no spurious mutations had been introduced during PCR.
Expression and Purification of the Enzymes-The constructed coexpression vectors pET15b/SL and various mutant plasmids were transformed into host E. coli BL21-CodonPlus (DE3)-RIL. The transformed cells were grown in 2ϫ yeast tryptone medium containing ampicillin (100 mg/liter) at 37°C. When A 600 reached 0.6 -1.0, 2 mM isopropylthio-␤-D-galactoside was added to induce expression of the genes. After being cultured overnight at 30°C, the cells were harvested by centrifugation, resuspended in 50 mM Tris-HCl (pH 8.0) containing the protease inhibitor, and disrupted on ice by sonication. The disrupted cells were heated at 85°C for 30 min and centrifuged at 27,000 ϫ g for 20 min to remove cell debris and denatured proteins. Expression of both DP1Pho and DP2Pho in the supernatant was Sequences and restriction sites in the primer a S1 5Ј-TTTTGTCGACGTACATATGGATGAATTCGTAAAG-3Ј  Table I. checked by SDS-PAGE performed on a 10 -15% gradient gel using the Phast system (Amersham Pharmacia Biotech). Protein bands were visualized by staining with Coomassie Brilliant Blue R-250. The supernatant was dialyzed against buffer A (50 mM Tris-HCl, pH 8.0) and then against buffer B (50 mM Tris-HCl, pH 6.5). The dialysate was loaded onto a Hitrap Q column (Amersham Pharmacia Biotech), which was pre-equilibrated with buffer B using the fast protein liquid chromatography system (Amersham Pharmacia Biotech). The column was developed with a linear gradient of 0 -1000 mM. Fractions at around 450 mM NaCl, which contained the PolDPho complex, were dialyzed against buffer A to remove the salt. Proteins from pET15b/SL and mutant plasmids were further purified by loading them onto a nickel column (Novagen). After washing with a buffer containing 20 mM imidazole, the enzymes were eluted with the elution buffer containing 150 mM imidazole. The protein content was determined with the Bio-Rad protein assay dye reagent using bovine serum albumin as the standard protein.
Assay of DNA Polymerizing Activity-The presence of dNTP in the reaction mixture is essential to synthesize DNA products by PolD. In the absence of dNTP, the substrate DNA is digested by the exonuclease activity. When the dNTP is omitted from the reaction mixture, only exonuclease activity is measurable. The DNA polymerizing activity was assayed by measuring the incorporation of [␣-32 P]dATP into the trichloroacetic acid-insoluble material. The standard reaction mixture (in 50 l) contained 20 mM Tris-HCl (pH 8.8, 25°C), 10 mM MgSO 4 , 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 0.1% Triton X-100, 200 g/ml heat-denatured salmon testes DNA, 0.25 mM dNTP mixture, 0.5 Ci (9.25 kBq) of [␣-32 P]dATP, and crude or purified enzymes. The reaction was performed at 70°C for 30 min, and then the radioactivity was measured by scintillation counting. At least two repeats were performed independently for each sample.
Assay of Primer Extension Activity-The primer extension ability was assayed using a 100-mer oligonucleotide whose sequences were taken from M13mp19 single-stranded DNA (5Ј-TCCTCTAGAGTCGA-CCTGCAGGCATGCAAGCTTGGCACTGGCCGTCCTTTTACAACGT-CGTGACTGGGAAAACCCTGGCGTTACCCAACTTAATCGCCTT-3Ј, from positions 6254 to 6353) annealed with a 5Ј end-labeled 34-mer oligomer (5Ј-GTAACGCCAGGGTTTTCCCAGTCACGACGTTGTA-3Ј, complementary to positions 6337-6304 of M13mp18) as a substrate. The 34-mer was radiolabeled at the 5Ј terminus with the bacteriophage T4 polynucleotide kinase and [␣-32 P]ATP. To make the substrate, 10 pmol of radiolabeled primer was mixed with 20 pmol of the 100-mer oligonucleotide in 20 l of buffer (7 mM Tris-HCl (pH 8.0), 50 mM NaCl, and 7 mM MgCl 2 ). The tube was boiled for 5 min and then cooled gradually to room temperature. The reaction mixture for polymerization (10 l) contained 20 mM Tris-HCl (pH 8.8, 25°C), 10 mM MgSO 4 , 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 0.1% Triton X-100, 0.25 mM dNTP, 0.5 pmol of labeled substrate, and 100 ng of enzymes. The reaction was performed at 70°C for 2 min. After the reaction, 10 l of stop buffer containing 95% formamide, 10 mM EDTA, and 1 mg/ml bromphenol blue was added to the mixture, and the tube was heated in boiling water for 5 min. The samples were loaded onto a 15% polyacrylamide gel (35 ϫ 42.5 cm) containing 7 M urea and 1ϫ TBE (89 mM Tris-HCl, 89 mM boric acid, and 2 mM EDTA (pH 8.0)) and electrophoresed for 1.5 h at 2500 V. Reaction products were visualized using a PhosphorImager (Bio-Rad).
Assay of 3Ј-5Ј Exonuclease Activity-The reaction mixture and procedures used to assay the 3Ј-5Ј exonuclease activity of the wild type and mutants were the same as for the primer extension assay, except that the dNTP was not added and 500 ng of heat-treated supernatant or 400 ng of purified enzymes was used. The reaction was performed at 70°C for 10 min or as indicated. To measure the reaction speed of exonuclease activity, 100 ng of enzyme was used, and the reaction was performed for 0.5, 1.0, 1.5, 2.0, and 2.5 min. The gel was electrophoresed for 3 h to get better separation of the products. After electrophoresis, the products were visualized and quantified using the PhosphorImager (Bio-Rad). The exonuclease activity was calculated according to the reduction of substrates, which was quantified by the ratio of signal strength of products to the total signal strength of each sample.
Measurement of Circular Dichroism (CD)-The CD of the wild type and mutant proteins was measured with a CD spectrometer (model 62A DS, Aviv Instrument, Lakewood, NJ) utilizing a 0.1-cm path length quartz cell in the far UV region. The measurement was carried out in 20 mM Tris-HCl (pH 8.0) buffer containing 50 -100 g/ml enzyme.
Measurement of Polymerization Fidelity-The fidelity of PolD was measured according to the method of Kunkel (12) with minor modifications. To make the gapped DNA, M13mp18 double-stranded DNA was digested with PvuI and PvuII, purified from a 2% agarose gel using the QIAquick mini-column (Qiagen, Hilden, Germany), and eluted in TE buffer (10 mM Tris buffer containing 1 mM EDTA). The digested M13mp18 DNA (1.0 g in 10 l) was added into 70 l of sterile distilled water and denatured for 10 min at 70°C. Then M13mp18 singlestranded DNA (2 g in 10 l) was added, and the tube was mixed and cooled down in an ice bath. After 10 min, 10 l of 20ϫ SSC buffer was added, and the mixture was heated at 60°C for 10 min. The gapped DNA was recovered by ethanol precipitation and confirmed by gel electrophoresis. The gap was filled by PolDPho with polymerization for 30 min at 70°C. The reaction was stopped by adding 1.5 mM EDTA, and the DNA was recovered by ethanol precipitation and dissolved in TE. The filling of the gap was confirmed by electrophoresis using agarose gel. E. coli strain XL2 Blue (MRFЈ) was transformed with the polymerized DNA, and strain MV1184 was used as lawn cells for transinfection. White plaques were counted, and the fidelity was calculated as a percentage of the number of white plaques against the total plaques formed.

Cloning and Expression of the Small and Large Subunit
Genes of PolDPho-According to the genomic sequences of P. horikoshii OT3 (13), the two ORFs PH0123 and PH0121 code for the small subunit DP1Pho (622 amino acids, 70 kDa) and large subunit DP2Pho (1434 amino acids, 163 kDa) of PolD, respectively. As in P. furiosus and P. abyssi, the two genes are arranged in tandem (with DP1Pho in the front) and are adjacent to ORFs that are important for DNA replication, recombination, and repair (6,8). Alignment and sequence analysis of DP2Pho with other PolD large subunits from the Archaea species revealed that DP2Pho contains a 166-amino acid miniintein (internal protein fragment) from Cys-955 to Gln-1120 ( Fig. 2; Ref. 14). The large subunit without the intein is 1268 amino acids in length and has a predicted molecular mass of 144 kDa.
The DP1Pho and DP2Pho genes were first cloned into pET15b and pGEMEX-1 separately (Fig. 1). The expression of DP1Pho with the pET15b-S vector in the E. coli strain BL21-CodonPlus (DE3)-RIL and its purification from soluble fractions were successful. The expression of DP2Pho using pGEMEX-L(Ϫ) and pGEMEX-L(ϩ) was also successful; however, the majority of the protein was in insoluble fractions. Purification of DP2Pho from soluble fractions proved difficult because of its insolubility and the degradation of the protein during purification.
To obtain a stable recombinant protein and active complex, we constructed the co-expression vector pET15b/SL. In this vector, the DP1Pho and DP2Pho genes were connected in tandem into two ORFs. The two genes are transcribed under the control of a single transcription promoter and terminator (T7 promoter and T7 terminator). After transformation and induction using the E. coli strain BL21-CodonPlus (DE3)-RIL, stable proteins were expressed as assayed by SDS-PAGE and were purified to near homogeneity with an anion exchange column and nickel column. Several milligrams of purified protein could be produced from a liter of culture. Fig. 3 shows the SDS-PAGE and Western blotting profile of His-tagged PolDPho. The band position of DP2Pho is as predicted at around 140 kDa, whereas DP1Pho migrated at around 90 kDa rather than the predicted 70 kDa. The band intensity of DP2Pho was about two times that of DP1Pho, as measured by density scanning of the protein bands on the SDS gel (data not shown), indicating a 1:1 molar ratio between the two subunits. Our success in purification of PolDPho with a histidine tag (His tag) at the N terminus of DP1Pho confirmed that DP1Pho and DP2Pho form a tight complex. Using a nickel column, the complex was bound to the resin through the His tag of DP1Pho, and DP2Pho was pulled down by a strong interaction between both subunits. This may also indicate that the N terminus of DP1Pho is located at the surface of the protein complex.
To compare the effects of the intein and His tag on expression and activity, vectors containing the intein and vectors without the His tag were also constructed (named pET15b/SL(ϩ), pET11a/SL(Ϫ), and pET11a/SL(ϩ)) using a similar approach to that described above. Preliminary results showed that when using any of the three co-expression vectors, both DP1Pho and DP2Pho were also overexpressed in E. coli. The DNA polymer-ization activities of the heat-treated proteins were examined by measuring the incorporation of [␣-32 P]dATP into salmon testes DNA. We found that proteins using the intein-containing vectors, pET15b/SL(ϩ) and pET11a/SL(ϩ), had only 10 -20% of the DNA polymerization activity of proteins using vectors without the intein, pET15b/SL and pET11a/SL(Ϫ). This indicated that a small proportion of intein was self-spliced from the proteins. We also found that the addition of the His tag at the N terminus of DP1Pho had no effect on the DNA polymerization activities. Therefore pET15b/SL was used as the co-expression vector for further expression and mutagenesis.
Determination of the pH and Magnesium Ion Optima and Thermostability-The optimal pH and magnesium requirement for DNA synthesis were determined by measuring the incorporation of [␣-32 P]dATP into salmon testes DNA under different conditions or treatments. PolDPho showed the highest activity from pH 8.5 to 9.0, and the optimal magnesium concentration was 17.5 mM. PolDPho is highly thermostable; 50% activity remained after heating at 80°C for 60 min. The CD profiles of PolDPho at the near UV region measured at 25 and 90°C showed similar patterns, indicating no conformational alteration at high temperature (data not shown).
Determination of Native Molecular Mass by Gel Filtration-The molecular mass of PolDPho was assessed by gel filtration (Fig. 4). Purified protein (after anion exchange and nickel columns) was loaded onto a Superdex 200-pg (26/60) column (Amersham Pharmacia Biotech). PolDPho was eluted at around 151 ml. By calculation with elution volumes of calibration markers (128 ml of thyroglobulin, 145 ml of ferritin, 171 ml of catalase, 176.3 ml of aldolase, and 193.5 ml of albumin), the molecular mass of PolDPho was determined to be 421 kDa. After gel filtration, each fraction was concentrated 20 times, and the concentrated samples were subjected to DNA polymerization (by measuring the incorporation of [␣-32 P]dATP into salmon testes DNA), SDS-PAGE, and Western blotting analysis. The fraction activities varied coordinately with the UV absorption (Fig. 4) and protein band intensity (not shown). According to the ratio of band intensity between DP1Pho and DP2Pho (1:2) on SDS gel and the ratio of their molecular mass (1:2), we proposed that PolDPho has two large and two small subunits in one molecule, with a predicted molecular mass of 430 kDa.
Measurement of Primer Extension and Exonuclease Activity of the Wild Type Enzyme-The primer extension ability and 3Ј-5Ј exonuclease activities of PolDPho are shown in Fig. 5. For the primer extension assay, 100 ng of enzyme was used in one reaction. The products (35-84-mer) were observed after reacting for 2 min (Fig. 5A). In the 3Ј-5Ј exonuclease assay, more enzyme (400 ng) was used, and the reaction was performed for 30 min. The products were shorter than 34-mer (Fig. 5B).

Measurement of Fidelity by Kunkel Method-
The fidelity of DNA polymerization by PolDPho was assayed using the Kunkel method. Among a total of 122,202 plaques counted, 109 (0.89%) were white. The fidelity of Taq DNA polymerase, which lacks 3Ј-5Ј exonuclease activity, was assayed simultaneously as a control. The fidelity of Taq DNA polymerase was 1.65% (209/12,681) in our measurement, which was 1.9 times lower than PolDPho.
Making Mutants to Map Catalytic Residues of Polymerization Activity-Amino acid residues in the region from Glu-800 to Glu-1195 of DP2Pho were targeted for site-directed mutagenesis. This region is clustered with negatively charged residues conserved in DP2 (Fig. 2). In addition, it contains the most conserved and unknown motif, which is disrupted by the mini-intein in P. horikoshii (from Cys-955 to Gln-1120 of DP2Pho), P. abyssi, and Halobacterium sp. NRC-1 (Fig. 2). Two common DNA polymerase motifs were proposed in this region through visual inspection (Ref.  (Table II). Mutant proteins were expressed and partially purified by heating, and the polymerization activities were examined by measuring the incorporation of [␣-32 P]dATP into salmon testes DNA. The 3Ј-5Ј exonuclease activity for each mutant was also assayed roughly by gel electrophoresis. Among all the 28 mutants, only D1122A and D1124A showed loss of polymerization activity, whereas the other mutants did not show a dramatic change in DNA synthesis in the preliminary screening (Table  II). Mutant proteins of D112A and D1124A were further purified by passing through anion exchange and nickel columns (see "Materials and Methods"), and the DNA polymerization activities were measured and compared with a wild type protein. When the purified proteins were used, D1122A and D1124A showed only 0.51 and 0.82% activities of the wild type enzyme, respectively.
To further confirm the catalytic importance of Asp-1122 and Asp-1124, various mutants were made at these two residues, in which aspartic acid (D) was changed to asparagine (N), glutamic acid (E), glutamine (Q), and arginine (R), respectively. Their polymerization activities are shown in Fig. 6, A and B. D1122E and D1124E showed 15 and 43% of the activity of the wild type PolDPho. D1124N had 17% of the activity, although we were unable to obtain mutant D1122N protein. However, D1122Q, D1122R, D1124Q, and D1124R showed activities of less than 1% of the wild type. We also made two mutants at Gly-1123 (glycine to alanine or arginine) and measured their polymerization activities. It is interesting that G1123A had only 13% of the activity of the wild type enzyme, whereas G1123R had 0.8% activity. It is likely that the reduction is due to the proximity of Gly-1123 to Asp-1122 and Asp-1124.
SDS-PAGE analysis using purified mutant protein showed the same profiles as the wild type, indicating that the mutant proteins in the region were properly expressed and stable (data not shown). To know whether the activity loss of the Asp-1122, Gly-1123, and Asp-1124 mutants was caused by conformational changes, we measured the CD spectra of the mutant proteins. Except for D1122R, which had a CD profile rather different from the wild type protein (data not shown), the other mutants were identical (D1122A, D1122E, G1123A, G1123R, D1124A, D1124E, D1124Q, and D1124R) or very similar (D1122E and D1122Q) to the wild type protein (Fig. 7, A-C).
The Mutants D1122A and D1124A Maintain the 3Ј-5Ј Exonuclease Activity-It is known that the polymerization and 3Ј-5Ј exonuclease domains are separate in family B DNA polymerase. If this is also the case in PolD, the mutants D1122A and D1124A should maintain the exonuclease activities. We checked the primer extension and 3Ј-5Ј exonuclease activities of both mutants by gel electrophoresis (Fig. 5). As expected, no DNA polymerization products from D1122A and D1124A could be detected (Fig. 5A); however, the 3Ј-5Ј exonuclease products were observed (Fig. 5B). We determined the reaction rates of exonuclease activities of the wild type enzyme, D1122A, and D1124A (Fig. 8). The reaction rates of the mutants D1122A and D1124A were 38 and 55% of the wild type, respectively. This shows that the exonuclease activities remain in the mutants, although they are slightly reduced. The data support the proposal that Asp-1122 and Asp-1124 are the catalytic residues of DNA polymerization and indicate that the catalytic residues of 3Ј-5Ј exonuclease and those of DNA polymerization are different in PolD.
The Mutants D1122E and D1124N Have Lower Magnesium Affinity than Does the Wild Type-We measured the magnesium requirements for polymerization activity of the mutants D1122E and D1124N. As shown in Fig. 9, the magnesium concentration required for D1122E to exhibit half of its own full activity was about 4.5 mM, and for D1124N it was more than 7.5 mM. Both are significantly higher than that of the wild type (about 2.5 mM). These facts may indicate that the affinities of D1122E and D1124N to magnesium became lower than the wild type protein and that the invariant residues Asp-1122 and Asp-1124 are probably involved in binding the magnesium that is important for polymerization activity. DISCUSSION In the present study, several advances were presented for the characterization of PolD. We identified two aspartate residues, Asp-1122 and Asp-1124, as the catalytic residues for DNA polymerization in PolD. They are located in the most ϩϩϩ ϩϩϩ a The underline indicates that purified proteins were used for the assay; the other samples used a heat-treated supernatant.
b Reactions were performed using 0.5 g of supernatant or 0.2 g of purified proteins by the acid precipitate method.
c Reactions were performed using 0.5 g of supernatant or 0.4 g of purified proteins. d ϩ, ϩϩ, and ϩϩϩ, indicate 1-20%, 21-60%, and 61-100% of wild type enzyme activity, respectively. e -, indicates undetectable or very low activity (Ͻ1% of the wild type enzyme). f ND, not determined. conserved region of DP2, YAHPY(F)FHA(S)AKRRNCDG-(S)DED (Fig. 2). The change of Asp-1122 and Asp-1124 to alanine caused a dramatic decrease in polymerization activity. The CD analysis confirmed that the change in activity was not because of structural alteration. We presented evidence that Asp-1122 and Asp-1124 are probably involved in magnesium binding by analyzing the Mg 2ϩ dependence for the mutants D1122E and D1124N. We further demonstrated that the mutants D1122A and D1124A maintained exonuclease activity, indicating that neither Asp-1122 nor Asp-1124 is the catalytic residue of the exonuclease activity.
Interestingly, the most conserved region is inserted by a mini-intein in three of the eight Archaea species with PolD, P. horikoshii, P. abyssi, and Halobacterium sp. NRC-1 but not others. Inteins tend to be inserted into important motifs for enzymatic activities in archaeal DNA replication proteins, such as family B DNA polymerase and replication factor C, as described by Cann and Ishino (3). This is indirect evidence that the most conserved motif in PolD has catalytic importance.
The catalytic sites of family B DNA polymerase have been located by mutagenesis (15). The metal binding residues are invariant aspartic acid residues in the most conserved motif -YGDTDS-in region I (or motif C). The catalytic residues of family A DNA polymerase have also been determined by mutagenesis and by structural analysis (16). An invariant aspartic FIG. 7. CD profiles of the wild type enzyme and the mutants. The mean residue ellipticity was expressed as deg cm 2 dmol Ϫ1 . A, the wild type enzyme (E) and the mutants D1122A (q), D1122E (Ⅺ), and D1122Q (ϩ). B, the wild type enzyme (E) and the mutants G1123A (q) and G1123R (f). C, the wild type enzyme (E) and the mutants D1124A (f), D1124N (q), D1124E (‚), D1124Q (OE), and D1124R (ϩ).
FIG. 8. Measurement of the exonuclease reaction rate of the wild type and the mutants D1122A and D1124A. One hundred nanograms of enzymes was added to the mixture in each tube, and the reactions were stopped after 0.5, 1.0, 1.5, 2.0, and 2.5 min (lanes 1-5). The reduction of substrate was counted using the PhosphorImager. A, a gel profile of the time course reaction using the wild type enzyme. B, assessment of reaction rates of endonuclease activity of the wild type (E), D1122A (Ⅺ), and D1124A (q). The substrate reduction was measured in at least two independent experiments.  and D1124N (؉). Two hundred nanograms of enzyme was used in each reaction. Each value was calculated in two independent measurements. acid residue in motif A (Asp-705 in E. coli DNA polymerase I) and one in motif C (Asp-882 in E. coli DNA polymerase I) are involved in metal binding. An additional carboxylate triad (Glu-710 and Glu-883) was recently found to be obligatory for catalysis in E. coli DNA polymerase I (17). The sequences of family B and family A DNA polymerase in the catalytic regions are not conserved, but protein structural conservation has been proposed (15). In the superposition of family A on family B DNA polymerase, the pair of negatively charged residues, DE, of a ␤-␤ turn in E. coli DNA polymerase I matched the two negatively charged residues in human polymerase ␣, which is a family B DNA polymerase (15). For PolD, sequences similar to motif A and motif C were proposed (3) according to visual inspection. However, mutagenesis at the proposed invariant residues in our study did not cause drastic changes in DNA polymerization. Instead, the invariant aspartic acid residues at Asp-1122 and Asp-1124 in the most conserved motif were found to be the catalytic sites. Therefore the most conserved motif is responsible for the polymerization catalysis. In this motif, glycine (or serine) is inserted between two aspartic acids. This is similar to the catalytic residues in motif C of family B DNA polymerase in which a threonine is inserted between two aspartic acid residues. Structurally, the threonine (T) was proposed to be at the bottom of two ␤ sheets. The glycine (or serine) with small side chains in the DP2 motif may perform the same functions as threonine in motif C of family B DNA polymerase. Structural analysis of PolDPho may eventually verify this hypothesis.
We also determined the molecular mass of PolDPho as 421 kDa. There are two possible combinations of subunit composition, in theory, based on this molecular mass; one is two large and two small subunits (predicted molecular mass of 430 kDa), and the other is four small subunits and one large subunit (predicted molecular mass of 428 kDa). The second combination does not seem likely considering the band intensity in the SDS-PAGE analysis (Fig. 3). Thus, we propose that PolDPho forms a heterotetrameric structure containing two large subunits and two small subunits in one molecule. Using gel filtration, we determined the molecular mass of the His-tagged small subunit to be around 70 kDa, corresponding to a monomer (data not shown). The small subunit could not form a tertiary structure; therefore the small subunit alone is not involved in the dimerization of the PolDPho complex. Further investigation is required to determine whether the large subunit alone or both subunits are involved in this process and whether the heterotetrameric structure exists in vivo. The structure of recombinant PolDPho seemed to be different from that of recombinant PolD from P. furious, which was reported to form a heterodimer (5). However, from the gel filtration results of the native PolD from P. furiosus, an early elution of active protein occurred before the heterodimeric peak (4), and the specific activity of the early peak seemed to be higher than that of the major peak. We suppose that the early elution peak might be the heterotetramer of the PolD from P. furiosus.
The structures of the main DNA polymerases seem to be conserved in the three domains of life. It is known that the ␣ ⑀ subassembly of holoenzyme III of E. coli is an asymmetric dimer including two ␣ subunits that are responsible for the leading strand and lagging strand synthesis, respectively, of the DNA replication folk. In eukaryotes, DNA polymerase ␦ and ⑀ are responsible for longer DNA synthesis in both strands after synthesis of short DNA fragments by DNA polymerase ␣ (18). It has been demonstrated that yeast (Schizosaccharomyces pombe) recombinant and native DNA polymerase ␦ complexes composed of four subunits form dimeric structures (19). There has also been evidence that budding yeast Saccharomyces cerevisiae DNA polymerase ⑀ complex forms a dimeric structure in vivo and that the dimerization is essential for DNA replication (20). Therefore it is likely that the major DNA replication polymerases in the three domains of life are in a dimeric structure in general.
Finally, a co-expression system has been developed that has allowed us to produce sufficient amounts of proteins and to generate various mutant plasmids and enzymes. Several factors may contribute to the success of this system. One is the use of the His tag at the N terminus of the small subunit, which allows purification of the complex by pulling down the large subunit using the nickel column. The second is the use of an induction temperature of 30°C rather 37°C and an isopropylthio-␤-D-galactoside concentration of 2 mM. Perhaps the most important factor is the use of the E. coli strain BL21-Condon-Plus (DE3)-RIL as the expression host. Previously, the strain BL21 (DE3) was used, but the expression level was extremely low. The improvement of protein expression by using strain BL21-CondonPlus (DE3)-RIL is reasonable because DP1Pho and DP2Pho genes contain many rare codons for E. coli. This co-expression system is also applicable for making other mutants in both subunits for further biochemical analysis. The system is now being evaluated for the commercial production of PolDPho.