Analysis of a Multicomponent Thermostable DNA Polymerase III Replicase from an Extreme Thermophile*

This report takes a proteomic/genomic approach to characterize the DNA polymerase III replication apparatus of the extreme thermophile, Aquifex aeolicus. Genes (dnaX, holA, and holB) encoding the subunits required for clamp loading activity (τ, δ, and δ′) were identified. The dnaX gene produces only the full-length product, τ, and therefore differs from Escherichia coli dnaX that produces two proteins (γ and τ). Nonetheless, theA. aeolicus proteins form a τδδ′ complex. ThednaN gene encoding the β clamp was identified, and the τδδ′ complex is active in loading β onto DNA. A. aeolicus contains one dnaE homologue, encoding the α subunit of DNA polymerase III. Like E. coli, A. aeolicus α and τ interact, although the interaction is not as tight as the α−τ contact in E. coli. In addition, theA. aeolicus homologue to dnaQ, encoding the ε proofreading 3′–5′-exonuclease, interacts with α but does not form a stable α·ε complex, suggesting a need for a brace or bridging protein to tightly couple the polymerase and exonuclease in this system. Despite these differences to the E. coli system, the A. aeolicus proteins function to yield a robust replicase that retains significant activity at 90 °C. Similarities and differences between the A. aeolicus and E. coli pol III systems are discussed, as is application of thermostable pol III to biotechnology.

Chromosomal replicases of all cellular organisms studied thus far are composed of three components, the DNA polymerase, a ring-shaped DNA sliding clamp, and a clamp loader that uses ATP to assemble the sliding clamp onto DNA (1)(2)(3). In bacteria, the sliding clamp is a homodimer called ␤ (4). The ring-shaped ␤ dimer completely encircles DNA and slides along the duplex (5). The ␤ clamp also binds the DNA polymerase III, thereby tethering it to DNA for high processivity (5).
This report on the Aquifex aeolicus pol III 1 replicase is part of our continuing study of comparing and contrasting replicases from a variety of bacteria. Most knowledge of bacterial DNA polymerase III (pol III) structure and function has been obtained from studies of the Escherichia coli replicase, DNA polymerase III holoenzyme (reviewed in Ref. 6). Therefore, a brief overview of its structure and function is instructive for the comparisons to be made in this report. In E. coli, the catalytic subunit of DNA polymerase III is the ␣ subunit (129.9 kDa) encoded by dnaE; it lacks a proofreading exonuclease (7). The proofreading 3Ј-5Ј-exonuclease activity is contained in the ⑀ (27.5 kDa) subunit (dnaQ) that forms a 1:1 complex with ␣ (8,9). The pol III ␣Ϫ⑀ complex is found tightly associated to a third subunit, called , to form the heterotrimeric E. coli DNA polymerase III core (10). The subunit (holE, 8.6 kDa) is not essential for growth and is generally not conserved in bacteria (11).
The E. coli pol III ␣ subunit and pol III core subassembly act distributively on primed ssDNA and have only low activity; they are even further inhibited by the presence of SSB (7,12). However, after the ␤ clamp has been assembled onto a primed site, the efficiency of the pol III ␣ subunit is greatly stimulated, and ␣⅐␤ extends the primer at a rate of ϳ300 nucleotides/s with a processivity of 1-3 kb (9). The pol III ␣⑀ complex and pol III core subassembly are even further stimulated by ␤ and extend DNA at a rate of about 1 kb/s with a processivity that exceeds the entire 7.2-kb M13mp18 ssDNA template (9).
Mechanistic studies have outlined the overall mechanism of the clamp loader and are consistent with the structural analysis. The ␥ subunit is the only subunit that interacts with ATP and therefore is the motor of the clamp loader (20). The ␦ subunit alone can open one interface of the ␤ dimer (21,22). The ␦ clamp opener is sequestered within ␥ complex via association to ␦Ј (21). ATP binding to ␥ 3 results in a conformational change, releasing the ␤ interactive site on ␦ from ␦Ј (21,23). The ATP⅐␥ 3 ␦␦Ј species binds to ␤, opens the ring, and binds DNA (24,25). Then hydrolysis of ATP brings ␦ back onto ␦Ј, severing connections to ␤, allowing ␤ to close around DNA (21, 26 -28).
In E. coli, the dnaX gene encoding ␥ also encodes the subunit of DNA pol III holoenzyme (29 -31). (71.1 kDa) is the full-length product of dnaX, whereas ␥ is shorter (47.5 kDa), being truncated by a translational frameshift. can fully replace ␥ in the clamp loader, and the 3 ␦␦Ј complex is active in clamp loading (13). The C-terminal sequences unique to are required for interaction with the pol III ␣ subunit (32) and also with the replicative DnaB helicase (33,34). Therefore, within the holoenzyme, subunits must replace two (or all three) of the ␥ subunits in order to connect the two pol III core polymerases in the holoenzyme structure for simultaneous replication of both leading and lagging strands (60).
We have undertaken the study of other bacterial replication systems in an effort to delineate those features of prokaryotic replicases that are general to all bacteria. Study of the Gramnegative Thermus thermophilus dnaX gene showed that it produces both ␥ and , like E. coli dnaX (35,36). However, instead of a Ϫ1 ribosomal frameshift, T. thermophilus employs a transcriptional slippage mechanism that results in both Ϫ1 and Ϫ2 frameshifts (35,37). We have also examined the pol III replicase of a Gram-positive organism, Streptococcus pyogenes (38). This study showed that only one protein is produced from the S. pyogenes dnaX gene. This full-length protein (62.1 kDa) is intermediate in length between E. coli ␥ (47.5 kDa) and (71.1 kDa) but retains the capacity to bind the DNA polymerase, like E. coli . However, the strength of this interaction is weaker than in the E. coli system. Like other Gram-positive bacteria, the S. pyogenes DNA polymerase, pol C (ϳ165 kDa), contains an inherent 3Ј-5Ј-exonuclease activity instead of delegating this proofreading action to a separate ⑀ subunit as observed in the E. coli system (39). As in E. coli, the S. pyogenes, , ␦, and ␦Ј subunits are required to load ␤ onto DNA, and the clamp endows pol C with the same rapid speed and processivity as the entire E. coli DNA pol III holoenzyme (38). It is interesting to note that S. pyogenes also contains a second homologue to E. coli ␣, which we refer to as the DnaE polymerase (38). DnaE polymerase is similar in size to E. coli ␣ (120 kDa) and like E. coli ␣ it lacks 3Ј-5Ј-exonuclease activity (38). The processivity of S. pyogenes DnaE polymerase is stimulated by ␦␦Ј and ␤, but its intrinsic speed (60 nucleotides/s) is unaltered (38).
This report examines the pol III replication machinery of the extreme thermophile, A. aeolicus. By using the known genome sequence (41), we identify A. aeolicus replicase genes and produce and isolate recombinant ␣, ⑀, ␤, , ␦Ј, ␦, subunits and SSB. We then compare and contrast the function and assembly of these replicase subunits from a thermophile to the Gram-negative E. coli replicase and to our previous studies on the Grampositive S. pyogenes replicase.

EXPERIMENTAL PROCEDURES
Materials-Radioactive nucleotides were from PerkinElmer Life Sciences; unlabeled nucleotides were from Amersham Biosciences and The Upjohn Co. DNA oligonucleotides were synthesized by Invitrogen. M13mp18 ssDNA was purified from phage that was isolated by two successive bandings in cesium chloride gradients as described (42). M13mp18 ssDNA was primed with a DNA 30-mer (map position 6817-6846) as described (9). The pET protein expression vectors and BL21 (DE3) protein expression strain of E. coli were purchased from Novagen. DNA modification enzymes were from New England Biolabs. A. aeolicus genomic DNA and A. aeolicus cells were a gift of Dr. Robert Huber and Dr. Karl Stetter (Regensburg University, Germany). Protein concentrations were determined using Protein Stain (Bio-Rad) and BSA as a standard. Polyclonal antisera was produced by rabbits injected with purified E. coli ␥ or ␣ (5). Antibodies directed to A. aeolicus were purified from antisera by transfer of E. coli from an SDS gel to nitrocellulose followed by incubation with antisera and elution of purified antibody from the nitrocellulose membrane.
The pETAadnaE plasmid was transformed into the BL21 (DE3) strain of E. coli. Cells were grown in 50 liters of LB containing 100 g/ml kanamycin, 5 mM MgSO 4 at 37°C to A 600 ϭ 2.0, induced with 2 mM IPTG for 20 h at 20°C, and then collected by centrifugation. Cells were resuspended in 400 ml of 50 mM Tris-HCl (pH 7.5), 10% sucrose, 1 M NaCl, 30 mM spermidine, 5 mM DTT, and 2 mM EDTA. The following procedures were performed at 4°C. Cells were lysed by passing them twice through a French press (15,000 pounds/square inch) followed by centrifugation at 13,000 rpm for 90 min at 4°C. In this protein preparation, as well as each of those that follow, the induced A. aeolicus protein was easily discernible as a large band in an SDS-polyacrylamide gel stained with Coomassie Blue. Hence, column fractions were assayed for the presence of the A. aeolicus protein by SDS-PAGE analysis, which forms the basis for pooling column fractions.
The clarified cell lysate was heated to 65°C for 30 min, and the precipitate was removed by centrifugation at 13,000 rpm in a GSA rotor for 1 h. The supernatant (1.4 g, 280 ml) was dialyzed against buffer A (20 mM Tris-HCl (pH 7.5)), 10% glycerol, 0.5 mM EDTA, 5 mM DTT) overnight, and then diluted to 320 ml with buffer A to a conductivity equal to 100 mM NaCl. The dialysate was applied to a 150-ml fast flow Q-Sepharose column (Amersham Biosciences) equilibrated in buffer A, and eluted with a 1.5-liter linear gradient of 0 -500 mM NaCl in buffer A. Eighty fractions were collected. Fractions 38 -58 (1 g, 390 ml) were pooled, dialyzed versus buffer A overnight, and applied to a 250-ml heparin-agarose column (Bio-Rad) equilibrated with buffer A. Protein was eluted with a 1-liter linear 0 -500 mM NaCl gradient in buffer A. One hundred fractions were collected. Fractions 69 -79 (320 mg in 200 ml) were pooled and dialyzed against buffer A containing 100 mM NaCl. The ␣ preparation was aliquoted and stored frozen at Ϫ80°C. The Coomassie Blue-stained SDS-polyacrylamide gel of the final ␣ preparation is shown in Fig. 1.
Purification of ⑀ Encoded by dnaQ-The A. aeolicus dnaQ was identified in the genome sequence as a 202-residue protein of 17,132 Da having significant homology to E. coli dnaQ (41). A. aeolicus dnaQ was amplified by PCR using the following oligonucleotide primers. The upstream 35-mer (5Ј-GTGTGTCATATGCGAGACAATCTCCTTGATGGC-AG-3Ј) contains an NdeI site (underlined), and the downstream 44-mer (5Ј-GTGTGGATCCTCAAAACTTACCCTTTTCCAGCCTTTTTAAGGA-G-3Ј) contains a BamHI site (underlined). The PCR product was digested with NdeI and BamHI, purified, and ligated into the pET24a vector to produce pETAadnaQ.
The pETAadnaQ plasmid was transformed into E. coli strain BL21(DE3). A single colony was used to inoculate 12 liters of LB media supplemented with 200 g/ml ampicillin. Cells were grown at 37°C to A 600 ϭ 0.5 at which point 0.5 mM IPTG was added. After a 3-h induction, cells were collected by centrifugation and resuspended in 50 mM Tris-HCl (pH 7.5), 10% sucrose, 1 M NaCl, 30 mM spermidine, 5 mM DTT, 2 mM EDTA. Cells were lysed by two passages through a French press (15,000 pounds/square inch) followed by centrifugation at 13,000 rpm for 30 min at 4°C. The resulting supernatant (1038 mg) was incubated in a 65°C waterbath for 30 min. The solution was clarified by centrifugation at 13,000 rpm for 30 min. The supernatant (426 mg) was dialyzed against buffer A and loaded onto a 15-ml fast flow Q-Sepharose column equilibrated with buffer A. The column was eluted with a 150-ml linear gradient of 50 -500 mM NaCl in buffer A; 80 fractions were collected. Peak fractions (fractions 36 -44, 12.9 mg) were pooled, dialyzed against buffer A, and then loaded onto a 10-ml heparin-agarose column equilibrated with buffer A. The column was eluted with a linear gradient of 0 mM to 1 M NaCl; 80 fractions were collected. Peak fractions (fractions 60 -70) were pooled, aliquoted, and stored at Ϫ80°C.
Identification of A. aeolicus holA and Purification of ␦-The A. aeolicus holA gene was not identified previously by the genome sequencing project. We identified A. aeolicus holA by searching the A. aeolicus genome with the amino acid sequence of the E. coli ␦ subunit (encoded by holA). Although the resulting match had too low a score to be confident of its assignment as ␦, the studies of this report prove that the gene truly encodes ␦ subunit of the replicase. The A. aeolicus holA gene was amplified by PCR using the following primers. The upstream 36mer (5Ј-GTGTGTCATATGGAAACCACAATATTCCAGTTCCAG-3Ј) contains an NdeI site (underlined); the downstream 39-mer (5Ј-GTGT-GTGGATCCTTATCCACCATGAGAAGTATTTTTCAC-3Ј) contains a BamHI site (underlined). The PCR product was digested with NdeI and BamHI, purified, and ligated into the pET24 NdeI and BamHI sites to produce pETAaholA.
The pETAaholA plasmid was transformed into E. coli strain BL21 (DE3). Cells were grown in 50 liters of LB media containing 100 g/ml kanamycin. Cells were grown at 37°C to A 600 ϭ 2.0, then induced for 20 h upon addition of 2 mM IPTG, and collected by centrifugation. Cells from 25 liters of culture were lysed as described above for purification of ␣.
The cell lysate was heated to 65°C for 30 min, and the precipitate was removed by centrifugation. The supernatant (650 mg, 240 ml) was dialyzed against buffer A, adjusted to a conductivity equal to 160 mM NaCl by addition of 40 ml of buffer A, and applied to a 220-ml heparinagarose column equilibrated in buffer A containing 100 mM NaCl. The column was eluted with 1.0 liters linear gradient of 150 -700 mM NaCl in buffer A. One hundred and four fractions were collected. Fractions 45-56 were pooled (250 mg, 210 ml), diluted with 230 ml buffer A to a conductivity equal to 230 mM NaCl, and then loaded onto a 100-ml fast flow Q-Sepharose column equilibrated in buffer A containing 150 mM NaCl. The column was eluted with a 200-ml linear gradient of 150 -750 mM NaCl in buffer A; 73 fractions were collected. Fractions 16 -38 were pooled (95 mg, 40 ml), aliquoted, and stored at Ϫ80°C.
Identification of A. aeolicus holB and Purification of ␦Ј-The A. aeolicus holB gene was identified by Blast search using the E. coli ␦Ј sequence as a query. The A. aeolicus holB gene was amplified by PCR using the following primers. The upstream 41-mer (5Ј-GTGTGT-CATATGGAAAAAGTTTTTTTTGGAAAAAACTCCAG-3Ј) contains an NdeI site (underlined); the downstream 35-mer (5Ј-GTGTGTGGATC-CTTAATCCGCCTGAACGGCTAACG-3Ј) contains a BamHI site (un-derlined). The PCR product was digested with NdeI and BamHI, purified, and ligated into the pET24 NdeI and BamHI site to produce pETAaholB.
The pETAaholB plasmid was transformed into E. coli strain BL21 (DE3). Cells were grown at 37°C in 50 liters of media containing 100 g/ml kanamycin to A 600 ϭ 2.0 and then induced for 3 h upon addition of 2 mM IPTG. Cells were collected by centrifugation and were lysed using lysozyme by the heat lysis procedure. The cell lysate was heated to 65°C for 30 min, and precipitate was removed by centrifugation. The supernatant (2.4 g, 400 ml) was dialyzed versus buffer A and then applied to a 220-ml fast flow Q-Sepharose column equilibrated in buffer A. Protein was eluted with a 1-liter linear gradient of 0 -500 mM NaCl in buffer A; 80 fractions were collected. Fractions 23-30 were pooled and diluted 2-fold with buffer A to a conductivity equal to 100 mM NaCl and then loaded onto a 200-ml heparin-agarose column equilibrated in buffer A. Protein was eluted with a 1-liter linear gradient of 0 -1.0 M NaCl in buffer A; 84 fractions were collected. Fractions 46 -66 were pooled (1.3 g, 395 ml), dialyzed versus buffer A containing 100 mM NaCl, then aliquoted, and stored frozen at Ϫ80°C.
Purification of Encoded by dnaX-The A. aeolicus dnaX gene was amplified by PCR from genomic DNA using the following primers. The upstream 41-mer (5Ј-GTGTGTCATATGAACTACGTTCCCTTCGCGA-GAAAGTACAG-3Ј) contains an NdeI site (underlined); the downstream 36-mer (5Ј-GTGTGTGGATCCTTAAAACAGCCTCGTCCCGCTGGA-3Ј) contains a BamHI site (underlined). The PCR product was digested FIG. 1. Protein subunits of A. aeolicus pol III holoenzyme system. A, subunit preparations were analyzed in a 15% SDS-polyacrylamide gel stained with Coomassie Blue. Proteins were prepared as described under "Experimental Procedures." The positions of protein standards analyzed in the same gel are indicated to the left. B, the table gives the subunit gene name, mass predicted from the gene sequence for A. aeolicus pol III subunits, and their percent identify to corresponding subunits in the E. coli pol III systems (aligned using ClustalX). The function of subunits, or combinations of subunits, are given in the column at the right.
with NdeI and BamHI, purified, and ligated into the pET24 NdeI and BamHI sites to produce pETAadnaX.
The pETAadnaX plasmid was transformed into E. coli strain BL21 (DE3). Cells were grown in 50 liters of LB containing 100 g/ml kanamycin at 37°C to A 600 ϭ 0.6 and then induced for 20 h at 20°C upon addition of IPTG to 2 mM. Cells were collected by centrifugation and lysed as described for purification of ␣. The clarified cell lysate was heated to 65°C for 30 min, and the protein precipitate was removed by centrifugation. The supernatant (1.1 g in 340 ml) was treated with 0.228 g/ml ammonium sulfate followed by centrifugation. The subunit remained in the pellet which was dissolved in buffer B (20 mM Hepes (pH 7.5), 0.5 mM EDTA, 2 mM DTT, 10% glycerol) and dialyzed versus buffer B to a conductivity equal to 87 mM NaCl. The dialysate (1073 mg, 570 ml) was applied to a 200-ml fast flow Q-Sepharose column equilibrated in buffer A. The column was eluted with a 1.5-liter linear gradient of 0 -500 mM NaCl in buffer A; 80 fractions were collected. Fractions 28 -37 were pooled (289 mg, 138 ml), dialyzed against buffer A to a conductivity equal to 82 mM NaCl, and then loaded onto a 150-ml column of heparin-agarose equilibrated in buffer A. The column was eluted with a 900-ml linear gradient of 0 -500 mM NaCl in buffer A; 32 fractions were collected. Fractions 15-18 (187 mg, 110 ml) were dialyzed versus buffer A, then aliquoted, and stored at Ϫ80°C.
The pETAadnaN plasmid was transformed into E. coli strain BL21 (DE3). Cells were grown in 1 liter of LB containing 100 mg/ml kanamycin at 37°C to A 600 ϭ 1.0 and then induced for 6 h upon addition of 2 mM IPTG. Cells were collected (7 g) and lysed as described in the purification of ␣ above. The cell lysate was heated to 65°C for 30 min, and the protein precipitate was removed by centrifugation. The supernatant (39 mg, 45 ml) was applied to a 10-ml DEAE-Sephacel column (Amersham Biosciences) equilibrated in buffer A. The column was eluted with a 100-ml linear gradient of 0 -500 mM NaCl in buffer A; 75 fractions were collected. Fractions 45-57 were pooled (18.7 mg), dialyzed versus buffer A, and applied to a 30-ml heparin-agarose column equilibrated in buffer A. The column was eluted with a 300-ml linear gradient of 0 -500 mM NaCl in buffer A; 65 fractions were collected. Fractions 27-33 were pooled (11 mg, 28 ml) and stored at Ϫ80°C.
The pETAassb plasmid was transformed into E. coli strain BL21 (DE3). Cells were grown in 6 liters of LB media containing 200 g/ml ampicillin. Cells were grown at 37°C to A 600 ϭ 0.6, then induced at 15°C overnight in the presence of 2 mM IPTG, and collected by centrifugation. Cells were lysed as described above for purification of ␣ except cells were resuspended in buffer C (20 mM Tris-HCl (pH 7.9), 500 mM NaCl).
The cell lysate was heated to 65°C for 30 min, and the resulting precipitate was removed by centrifugation. The supernatant (1.4 g, 190 ml) was applied to a 25-ml chelating Sepharose column (Amersham Biosciences) charged with 50 mM nickel sulfate and then equilibrated in buffer C containing 5 mM imidazole. The column was eluted with a 300-ml linear gradient of 5-100 mM imidazole in buffer C. Fractions of 4 ml each were collected. Fractions 81-92 were pooled (ϳ240 mg in 48 ml) and dialyzed overnight against 2 liters of buffer B containing 200 mM NaCl. The dialysate was diluted to a conductivity equal to 92 mM NaCl using buffer A and then loaded onto an 8-ml MonoQ column equilibrated in buffer A containing 100 mM NaCl. The column was eluted with a 120-ml linear gradient of 100 -500 mM imidazole in buffer A. Seventy four fractions were collected. Fractions 57-70 were pooled (100 mg, 25 ml), aliquoted, and stored at Ϫ80°C.
Preparation of ␦␦Ј Complex-The ␦ (30 mg) and ␦Ј (30 mg) subunits were mixed in a volume of 23 ml of buffer A (final conductivity equivalent to 130 mM NaCl) and incubated at 24°C for 10 min. The mixture was applied to an 8-ml MonoQ column equilibrated in buffer A. The protein was eluted with an 80-ml gradient of 100 mM to 500 mM NaCl; 60 fractions were collected. The ␦␦Ј complex elutes later than either ␦ or ␦Ј. The ␦␦Ј complex was stored at Ϫ80°C.
Constitution of ␦␦Ј Complex-The reaction mixture contained 400 g of ␦␦Ј and 1.2 mg of in a volume of 200 l of buffer A containing 300 mM NaCl. The protein mixture was incubated at 24°C for 15 min and then injected onto an HR 10/30 Superose 12 column equilibrated with buffer A containing 300 mM NaCl. After the first 6.6 ml, fractions of 170 l were collected. Fractions were analyzed in 10% SDS-polyacrylamide gels stained with Coomassie Blue. The controls included ␦␦Ј complex alone (400 g) and alone (1.2 mg).
Quantitation of the Intracellular Concentration of A. aeolicus -A. aeolicus cells were a generous gift of Dr. Karl Stetter (Regensburg University). A. aeolicus cells were resuspended in buffer A to ϳ2 ϫ 10 11 cells/ml by direct counting as follows. Serial dilutions of the cell suspension (10 7 , 10 8 , 10 9 , and 10 10 ) were counted using a hemocytometer (Hausser Scientific) under a high power microscope (Olympus IX70), and the results were averaged. A 50-l aliquot was then cracked in 100 l of a solution containing 0.3% SDS, 30 mM Tris base, 30 mM DTT, 0.01% bromphenol blue, and 8% glycerol. Amounts equivalent to 2.33 ϫ 10 8 and 9.32 ϫ 10 8 cells were analyzed in two lanes of a 12% SDSpolyacrylamide gel, followed by transfer to nitrocellulose. Western blot analysis was then performed using rabbit polyclonal antibody directed against E. coli ␥ (which was purified against A. aeolicus as described earlier). Known amounts of pure recombinant A. aeolicus were analyzed in the same gel for use as a standard curve to quantitate the in A. aeolicus cells observed in the Western blot. The protein concentration of A. aeolicus was calculated from its absorbance at 280 nm in 6 M guanidine hydrochloride using the molar extinction coefficient determined from its 2 Trp, 14 Tyr, and 20 Phe content (⑀ 280 ϭ 5,690 ϫ 2 ϩ 1,280 ϫ 14 ϩ 2 ϫ 20 ϭ 29,340 M Ϫ1 cm Ϫ1 ).
Assays of ␣, ␦␦, Ϯ␤-Titration of ␣ into replication reactions were performed using circular M13mp18 ssDNA primed with a 700-fold molar excess of synthetic DNA 30-mer oligonucleotide. Reactions contained 138 fmol of primed M13mp18 ssDNA, 100 pmol of DNA 30-mer, 3.6 g of SSB (when present), 150 ng of ␦␦Ј, 0.4 g of ␤ (when present), and the indicated amount of ␣ in 50 l of 20 mM Tris-HCl (pH 8.8), 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 2 mM MgSO 4 , 0.1% Triton X-100, 4 mM CaCl 2 , 0.5 mM ATP, and 60 M each dATP and dGTP. Reactions were mixed on ice, and then the mixture was brought to 65°C for 2 min before initiating synthesis upon addition of 2 l of dCTP and [␣-32 P]dTTP (final concentrations, 60 and 40 M, respectively). Aliquots were removed and quenched at the times indicated upon adding EDTA and SDS to final concentrations of 20 mM and 0.5%, respectively. Quenched reactions were then analyzed in a 0.8% alkaline-agarose gel. Size standards were also included in the same gel. Gels were dried followed by analysis using a PhosphorImager.
Replication time course reactions contained 70 ng (25 fmol) of M13mp18 ssDNA, 100 pmol of 30-mer oligonucleotide, 2 g of SSB (when present), 100 ng of ␣, 200 ng of ␦␦Ј complex, and 40 ng of ␤ in 25 l of 20 mM Tris-HCl (pH 8.8), 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 10 mM MgSO 4 , 0.1% Triton X-100, 1 mM ATP, 4 mM CaCl 2 , 10% glycerol, and 60 M each dGTP and dATP. Replication time courses were performed at different temperatures as described above except that SSB was omitted where indicated, and the mixture was brought to the indicated temperature for 2 min before initiating synthesis by the addition of 2 l of dCTP and [␣-32 P]dTTP (final concentrations, 60 and 40 M, respectively). Aliquots were removed and quenched at the times indicated upon adding EDTA and SDS to 20 mM and 0.5%, respectively. Quenched reactions were then analyzed in 0.8% alkaline-agarose gels. Size standards were included in the same gel. Gels were dried and exposed to film.
Detection of ␣ Interaction with and ⑀ by Elisa-The ␣Ϫ interaction was analyzed as follows: E. coli ␣ or A. aeolicus ␣ (2 g in 100 l) were placed in wells of a 96-well vinyl assay plate (Costar) and incubated for 12 h at 4°C in 20 mM Tris-HCl (pH 7.5), 1 mM EDTA, 2 mM DTT, 5 mM MgCl 2 , 100 mM NaCl, 20% glycerol. Solutions were removed, and wells were washed 4 times with 100 l of TBS. Wells were then blocked with 100 l of TBS containing 5% non-fat milk for 3 h at 23°C. Following the blocking step, wells were washed 4 times with TBS and then incubated with 4 g of either E. coli or A. aeolicus for 2 h at 23°C. Solutions were then removed, and wells were washed as above and then incubated overnight at 4°C with 100 l of 1:500 dilution of polyclonal rabbit antibody raised against E. coli ␥ and purified against A. aeolicus . Wells were washed three times with TBS (5 min each wash) and then incubated with anti-rabbit horseradish peroxidase-conjugated antibody (Sigma) for 30 min and developed with an ECL detection kit (Amersham Biosciences) as described by the manufacturer.
Detection of ␣ and ⑀ interaction was performed as described above with the following modifications. One hundred microliters of either E. coli ⑀ or A. aeolicus ⑀ (50 g/ml) were incubated in wells for 12 h at 4°C. Following the block step and the washes with TBS, 100 l of either E. coli ␣ or A. aeolicus ␣ (100 g/ml) were placed into the wells and incubated for 2 h at 23°C. Solutions were then removed, and wells were washed with TBS as above and then incubated overnight at 4°C with 100 l of 1:500 dilution of polyclonal rabbit antisera directed against E. coli ␣. Wells were then washed with TBS (3 times for 5 min) and further incubated with a 1:10,000 dilution of anti-rabbit horseradish peroxidase-conjugated antibody (Sigma) for 30 min and then developed with an ECL detection kit (Amersham Biosciences) and exposed to film.
Detection of ␣-Interaction by DNA Synthesis-DNA synthetic reactions contained 70 ng of M13mp18 ssDNA uniquely primed with a synthetic DNA 60-mer, 1 g of E. coli SSB, 100 ng of polymerase subunit (either S. pyogenes pol C or E. coli ␣), and when present 400 ng of (S. pyogenes or E. coli ), in 25 l of 20 mM Tris-HCl (pH 7.5), 4% glycerol, 0.1 mM EDTA, 5 mM DTT, 2 mM ATP, 8 mM MgCl 2 , 40 g/ml BSA, and 60 M each of dGTP and dCTP. Reactions were preincubated at 37°C for 2 min, and then synthesis was initiated upon addition of 1.5 l of dATP and [␣-32 P]TTP (final concentrations of 60 and 20 M, respectively). Reactions were allowed to proceed for 5 min prior to being quenched with an equal volume (25 l) of 1% SDS, 40 M EDTA. One-half of the reaction was analyzed for total DNA synthesis using DE81 filter paper as described (43).
Exonuclease Assay-Exonuclease assays contained 100 fmol of a 5Ј-labeled 50-mer DNA oligonucleotide and 50 ng of either E. coli ⑀ subunit or A. aeolicus ⑀ in 20 l of 20 mM Tris-HCl (pH 7.5), 4% glycerol, 0.1 mM EDTA, 5 mM DTT, 40 g/ml BSA, 8 mM MgCl 2 . Reactions were incubated at 37°C for the indicated times and were quenched upon the addition of 20 l of 95% formamide, 40 mM EDTA. Ten microliters of each reaction were analyzed in a 15% denaturing (6 M urea) polyacrylamide gel.
Calf Thymus DNA Synthesis Assay-Assays contained 2.5 g of activated calf thymus DNA and 0.2 of either E. coli ␣ subunit or A. aeolicus ␣ in a final volume of 25 l of 20 mM Tris-HCl (pH 7.5), 4% glycerol, 0.5 mM EDTA, 5 mM DTT, 40 g/ml BSA, 8 mM MgCl 2 , 1 mM ATP, 60 M each dCTP, dGTP, dATP, and 20 M [␣Ϫ 32 P]dTTP. Reactions were incubated at the indicated temperatures, and aliquots were removed at the times indicated. DNA synthesis was quantitated using DE81 paper as described (43).

RESULTS
The A. aeolicus Clamp Loader-Studies in the E. coli system demonstrated the need for five proteins to form a processive polymerase III: ␣, (or ␥), ␦, ␦Ј, and ␤ (9, 13, 46). A sixth protein, the ⑀ 3Ј-5Ј-exonuclease, further enhanced the speed and processivity of the holoenzyme constituted from these five proteins (9). The report on the A. aeolicus genome sequence (41) documented the presence of A. aeolicus dnaE (␣), dnaX (/␥), dnaN (␤), and dnaQ (⑀), but holA and holB genes encoding ␦ and ␦Ј were not identified. As we have reported previously, ␦ is a very poorly conserved subunit of the pol III replicase (38). We identified S. pyogenes holA by isolating a gene encoding a weak homologue to E. coli ␦ and then showing that the putative ␦ formed an active clamp loader complex with other S. pyogenes subunits (38). In the experiments below, we follow a similar strategy to identify genes encoding both A. aeolicus ␦ and ␦Ј.
A search of the A. aeolicus genome for the A. aeolicus holA gene encoding ␦, using E. coli ␦ subunit as the query, produced very weak matches, the best having about 18% identity to E. coli ␦ (see Fig. 2A). Likewise, a search for the A. aeolicus holB gene encoding ␦Ј, using E. coli ␦Ј as query, produced a similarly weak match (Fig. 2B). To determine whether the putative A. aeolicus holA and holB genes truly encode the ␦ and ␦Ј subunits of the replicase, we expressed and purified the encoded proteins and then tested them for clamp loading activity with the A. aeolicus dnaX product. The genes encoding the putative A. aeolicus ␦ and ␦Ј subunits were cloned into the T7-based pET11 expression plasmid. Purification of A. aeolicus ␦ and ␦Ј were performed as described under "Experimental Procedures," and a sample of each preparation is shown in the SDS-polyacrylamide gel of Fig. 1A.
To test ability of the putative A. aeolicus ␦ and ␦Ј subunits to form a clamp loader complex with A. aeolicus (␥), encoded by dnaX, we needed to express and purify the product of the A. aeolicus dnaX gene. A. aeolicus dnaX lacks a ribosomal frameshift sequence, and it also lacks the six or more contiguous A residues that are needed to produce ␥ in T. thermophilus by a transcriptional slippage mechanism (37). Consistent with the lack of sites for synthesis of the truncated ␥ subunit, expression of the A. aeolicus dnaX gene in E. coli showed production of only one protein that was the approximate size of the predicted full-length product (54.3 kDa). As will be shown later in this report, examination of A. aeolicus cell extracts revealed the presence of only full-length subunit, a truncated version (i.e. ␥) was not detected. The A. aeolicus subunit was purified as described under "Experimental Procedures," and a sample of the protein preparation is shown in Fig. 1A.
In the E. coli system, (or ␥) forms a heterotrimer with ␦ and ␦Ј (13). Even though the crystal structure of E. coli ␥ 3 ␦␦Ј shows that ␥ contacts both ␦ and ␦Ј directly, mixtures of ␥ (or ) with either ␦ or ␦Ј do not yield either heterodimer (i.e. ␦ or ␦Ј) complex (13). To examine the A. aeolicus , ␦, and ␦Ј subunits for protein interactions, combinations of these subunits were mixed and then analyzed by gel filtration on a Superose 12 column. The analysis in Fig. 3A demonstrates that a mixture of ϩ ␦Ј does not yield a heterodimer complex, instead the two proteins elute at distinct positions. This result is similar to observations in the E. coli system (13). Analysis of the ϩ ␦ mixture, in Fig. 3B, shows that the bulk of the and ␦ does not comigrate, although a slight amount of ␦ appears in the vicinity of indicating a weak interaction between them. The elution position of the free ␦ and ␦Ј subunits in Fig. 3, A and B, compared with molecular mass standards, indicates that ␦ and ␦Ј are each monomeric, similar to ␦ and ␦Ј of E. coli and S. pyogenes (13,38). The subunit elutes as an oligomer, possibly a trimer or tetramer, like E. coli and S. pyogenes (38,47).
Previous study of the E. coli and S. pyogenes pol III systems demonstrated that the ␦ and ␦Ј subunits form a 1:1 ␦␦Ј complex (13,38). Fig. 3D shows that A. aeolicus ␦ and ␦Ј also form a ␦␦Ј complex. Analysis of a mixture of ␦ and ␦Ј demonstrates that the two subunits coelute in earlier fractions than the elution position of either ␦ or ␦Ј alone (compare D with A and B) and thus form a ␦:␦Ј complex. Densitometric scans of the gel in Fig.  3D yield a stoichiometry of ␦ 1.1 ␦Ј 1.0 (a similar stoichiometry is obtained from an analysis using a 2-fold molar excess of ␦Јover ␦ and scanning the fractions containing ␦␦Ј complex).
Is the A. aeolicus ␦␦Ј complex active as a clamp loader? To test this, a six-residue kinase recognition site was placed onto the C terminus of ␤, allowing it to be phosphorylated using [␥-32 P]ATP and cAMP-dependent protein kinase. The 32 P-␤ was then incubated with ␦␦Ј, ATP, and a circular plasmid having a single nick. After incubation at 70°C for 10 min, the reaction was applied to a 5-ml Bio-Gel A15m column. Bio-Gel A15m is a large pore resin that includes most proteins but excludes the large circular DNA molecule. Therefore, if ␦␦Ј is active in assembly of 32 P-␤ onto DNA, some of the 32 P-␤ should comigrate with the DNA in the excluded volume and resolve from the free 32 Fig. 4; about half of the 32 P-␤ is observed to comigrate with DNA in the excluded fractions (fractions 10 -15). As a control, a similar experiment was performed in the absence of ␦␦Ј complex. The result demonstrates that essentially no 32 P-␤ is assembled onto the DNA in the absence of ␦␦Ј. Hence, the A. aeolicus ␦␦Ј is capable of assembling A. aeolicus 32 P-␤ clamps onto DNA. The temperature resistance of the A. aeolicus ␦␦Ј complex and ␤ clamp will be demonstrated later in this report, along with the DNA polymerase.
DnaX Produces Only in A. aeolicus-The experiments of Fig. 5 address whether A. aeolicus cells produce a truncated product (i.e. ␥) or whether only full-length protein () is pro-duced. In Fig. 5A A. aeolicus whole cells were analyzed by Western using a rabbit polyclonal antibody raised against E. coli ␥ which cross-reacts with A. aeolicus . The A. aeolicus subunit is clearly visible in the analysis. The limit of detection in this assay (e.g. due to background) is such that if a lower molecular weight ␥-like product is present then it is present at less than 10% the intracellular level of . We have performed this Western several times and have also used even more cells, with results similar to those in Fig. 5. The inability to detect ␥ in A. aeolicus is consistent with the absence of recognizable signals for transcriptional slippage or a translational frameshift in the A. aeolicus dnaX gene sequence.
To determine the average number of molecules of present in an A. aeolicus cell, we directly counted the number of A. aeolicus cells in the cell suspension using a microscope. This provided us the number of cells that were analyzed in each lane of the Western in Fig. 5A. The amount of detected in cells by Western analysis was then quantitated by scanning the gel with a laser densitometer. The band intensity was converted to nanograms of by comparison with known quantities of recombinant A. aeolicus that had been analyzed in the same gel as the A. aeolicus cells. The result, in Fig. 5B, indicates that A. aeolicus has 110 -150 molecules of as trimer (assuming three subunits per clamp loader complex). Thus, this quantity could provide for 110 -150 clamp loading assemblies per cell. Similar quantitative work in the E. coli system indicated the presence of about twice as much ␥ (and ) than observed here for A. aeolicus (61). However, the same study estimated about 150 clamp loader complexes per E. coli cell due to limiting ␦Ј (61).
The A. aeolicus DNA Polymerase-The A. aeolicus ␣ subunit homologue has 40.7% identity to E. coli ␣. Cloning of the A. aeolicus dnaE gene into pET11 provided a high level of expression from which 320 mg of ␣ were obtained from 50 liters of induced cells. A sample of the protein preparation is shown in lane 1 of Fig. 1.
Like E. coli ␣, the A. aeolicus ␣ homologue lacks a region of homology to the E. coli ⑀ subunit exonuclease. Thus, it may be presumed that A. aeolicus ␣ binds to another protein that supplies the proofreading exonuclease function, similar to interaction of E. coli ␣ to the ⑀ 3Ј-5Ј-exonuclease. The A. aeolicus genome contains a dnaQ gene encoding a homologue to E. coli ⑀ (24.8% identity) (41). The A. aeolicus dnaQ gene was cloned into pET11 for expression and purification. A sample of the purified A. aeolicus ⑀ is shown in the SDS-polyacrylamide gel of Fig. 1.
In Fig. 6A, A. aeolicus ␣Ϫ⑀ interaction was analyzed in an Elisa type assay in which ⑀ is adhered to wells of a 96-well plate, followed by blocking additional sites with nonspecific protein. Following this, A. aeolicus ␣ is added, and then unbound ␣ is washed away. Rabbit antibody directed against E. coli ␣ cross-reacts with A. aeolicus ␣ and was used to detect whether A. aeolicus ␣ had bound to ⑀. The results demonstrate a high signal of ␣ bound to the well that was treated with ⑀. No ␣ is detected if the well is not pretreated with A. aeolicus ⑀. Similar results were obtained using E. coli ␣ and ⑀ (also shown in Fig. 6A). To determine whether A. aeolicus ␣ and ⑀ form a stable complex, we analyzed a mixture of ␣ and ⑀ (37 M ␣ and 44 M ⑀) by gel filtration on an FPLC Superose 12 column; however, the ␣ and ⑀ subunits do not comigrate (Fig. 6D). This result indicates that the ␣ and ⑀ subunits do not tightly interact under these conditions. This experiment has been repeated at a protein concentration of 150 M ␣ and 90 M ⑀, and in the presence of either , ␦␦Ј, or ␦␦Ј ϩ ␤, all with negative results (not shown).
A chromosomal replicase is generally expected to contain the 3Ј-5Ј-exonuclease activity in tight association with the DNA polymerase activity. The fact that we only detect weak interaction rather than a tight gel filterable complex between A. aeolicus ␣ and ⑀ may be due to any of several factors. For example, ␣ and ⑀ may interact tightly at high temperature. Another possibility is that a tight interaction between ␣ and ⑀ may occur in the presence of primed DNA. It is also possible that other proteins beyond those studied in this report are needed to act as a bridge between ␣ and ⑀. Yet another explanation may be that recombinant A. aeolicus ␣ or ⑀ is not properly folded (addressed below).
As one measure that the A. aeolicus ␣ and ⑀ subunits preparations are correctly folded, we examined them for catalytic activity. The A. aeolicus ␣ was active on gapped calf thymus DNA at 37°C but became more active at 65°C (Fig. 6B). By comparison, E. coli ␣ was more active than A. aeolicus ␣ at 37°C but was essentially inactive at 65°C. A. aeolicus ␣ retained significant activity even at 85°C (about 40% relative to 65°C). For exonuclease activity, a 5Ј 32 P-end-labeled DNA 50mer was used to determine whether the A. aeolicus ⑀ subunit was active. The results, in Fig. 6C, show an increase in mobility of the 5Ј 32 P-DNA oligonucleotide upon incubation with A. aeolicus ⑀, indicating the presence of 3Ј-5Ј-exonuclease activity. A similar experiment using E. coli ⑀ results in full degradation of the DNA within 10 min. The greater activity of E. coli ⑀ may be due to use of 37°C for the exonuclease assays, which is probably suboptimal for the A. aeolicus ⑀ subunit. The possibility that the A. aeolicus ␣ contains an intrinsic 3Ј-5-exonuclease activity was also tested in the exonuclease assay, but the result was negative (not shown). The presence of both polymerase and exonuclease activities suggests that recombinant A. aeolicus ␣ and ⑀ subunits are properly folded. Therefore, a lack of interaction between them is probably not due to one of them existing in a denatured state. Factors that may underlie the inability of A. aeolicus ␣ and ⑀ to form a complex are explored further under "Discussion." A. aeolicus ␣ and Interact-In the E. coli system, the subunit forms a tight contact with the ␣ subunit of the pol III core DNA polymerase, and this complex can be detected by gel filtration analysis at even low concentrations (100 nM) (46). Our studies in the S. pyogenes system demonstrate that the S. pyogenes subunit interacts with S. pyogenes pol C; however, the contact is weaker than in the E. coli system.
In Fig. 7, we examine the A. aeolicus and ␣ subunits for interaction between them. In Fig. 7A, A. aeolicus ␣interaction was analyzed in the Elisa type assay as described above for analysis of ␣-⑀. In this case, A. aeolicus ␣ is adhered to the wells, followed by blocking with nonspecific protein, and then adding A. aeolicus . Unbound was washed away, and antibody directed against E. coli (which cross-reacts with A. aeolicus ) was used to detect whether A. aeolicus had bound to ␣. The results demonstrate a high signal of bound to the well treated with ␣. No is detected if the well is not pretreated with A. aeolicus ␣. Hence, A. aeolicus ␣ and directly interact. Similar results were obtained using E. coli ␣ and (also shown in Fig. 7A).
Next we designed another assay to examine A. aeolicus and ␣ for interaction, and this second method carries the advantage of demonstrating that the interaction is functional. The assay is based on the fact that E. coli ␣ has very low activity on a singly primed M13mp18 ssDNA template coated with SSB, but addition of E. coli provides over 10-fold simulation. This stimulation of E. coli ␣ is specific to E. coli , as use of S. pyogenes does not stimulate the reaction (Fig. 7B). Likewise, S. pyogenes ␣ is stimulated by S. pyogenes , but not by E. coli (Fig. 7B). In keeping with interaction between A. aeolicus ␣ and , A. aeolicus ␣ is stimulated by A. aeolicus . We were unable to examine whether E. coli or S. pyogenes stimulate A. aeolicus ␣ due to the 65°C reaction temperature required by the A. aeolicus system for activity.
In Fig. 7C, A. aeolicus ␣ and were analyzed for ␣⅐ complex formation by gel filtration on an FPLC Superose 12 column. The top two panels show the elution profiles of A. aeolicus ␣ and alone, and the bottom panel shows the analysis of a mixture of ␣ and . The result shows that the A. aeolicus ␣⅐ complex cannot be isolated by gel filtration and therefore suggests that they do not form as tight of a complex as E. coli ␣⅐, which is stable to gel filtration even at concentrations below 0.1 M. The concentrations of ␣ and used in the analysis of Fig. 7C were 37 and 46 M, respectively. We have also analyzed A. aeolicus ␣ and at yet higher concentrations (150 and 92 M, respectively) but still observe no interaction between them (data not shown).
A. aeolicus ␣ Functions with the ␤ Clamp-Does A. aeolicus pol III ␣ subunit function with A. aeolicus ␦␦Ј and ␤? This is tested in the experiments of Fig. 8 in which a singly primed M13mp18 ssDNA circle (7.2 kb) is utilized as substrate. In the E. coli system, this large ssDNA circular primed template must be coated with SSB in order for it to serve as an efficient substrate for pol III holoenzyme. Without SSB, E. coli pol III holoenzyme is encumbered by DNA secondary structure which SSB mostly eliminates by virtue of its tight interaction with the ssDNA template, providing about a 10-fold stimulation in synthesis relative to the absence of SSB. However, with SSB present, the E. coli holoenzyme extends the primer completely around the circular template within 11 s at 30°C to form the RFII species. In the absence of ␤, the E. coli core polymerase does not extend the primer more than a few hundred nucleotides (12). However, when ␤ is first assembled onto the primed site, core becomes rapid and processive in synthesis (5).
To develop this assay in the A. aeolicus system, we cloned the A. aeolicus ssb gene into pET11 and expressed and purified the SSB (a sample of the preparation is shown in lane 7 of Fig. 1). With the A. aeolicus SSB, ␤ clamp and ␦␦Ј clamp loader in hand, we examined their effect on the A. aeolicus pol III ␣ subunit in extension of a primed site on the M13mp18 ssDNA genome. As a basis for comparison we first examined the behavior of ␣ on singly primed SSB-coated M13mp18 ssDNA in the absence of ␤ subunit (Fig. 8A). In this experiment, product is labeled by polymerase-catalyzed incorporation of [␣-32 P]deoxyribonucleoside triphosphate. Reactions were incubated at a temperature of 65°C, as the experiment of Fig. 6 demonstrates that the A. aeolicus ␣ is more active at 65 than at 37°C. The results demonstrate that A. aeolicus pol III ␣ subunit by itself is capable of completely, or nearly completely, extending the single primer full circle provided it is supplied in a large amount and is given sufficient time (see the last time point at the highest ␣ concentration). The size of the product formed at any given time appears as a fairly tight band. Furthermore, the product length is dependent on the amount of ␣ added to the reaction. This pattern of product formation, and length dependence on ␣ concentration, indicates that synthesis is distributive. In other words, the ␣ polymerase samples the entire population of primed templates, synthesizing short stretches over and over on each template until all templates are finally completed. E. coli pol III core (and E. coli ␣ subunit) is also highly distributive in synthesis. However, E. coli core is unable to completely extend a single primer around M13mp18 ssDNA no matter the time or concentration of polymerase used due to presence of insurmountable barriers to extension (48). In fact, SSB inhibits chain extension by E. coli pol III core in the absence of ␤ (48).
Next, the time course of DNA synthesis by A. aeolicus pol III ␣ subunit was examined in the presence of both ␦␦Ј and the ␤ clamp (Fig. 8B). The results demonstrate formation of complete RFII product within 2 min at all concentrations of ␣. At the highest amount of ␣ used, formation of the RFII product occurs FIG. 6. Analysis of A. aeolicus ␣ and ⑀. A, Elisa dot blot assays of ␣Ϫ⑀ interaction. Wells were coated with either A. aeolicus ⑀ or no protein and then were blocked and treated with A. aeolicus ␣ or no additional protein prior to detection with antibody to E. coli ␣ and anti-rabbit antibody coupled to horseradish peroxidase. A similar experiment using E. coli ␣ and E. coli ⑀ is also shown. B, activity assays of A. aeolicus ␣ and E. coli ␣ (0.2 g each) were compared at different temperatures using activated calf thymus DNA as a substrate. C, the activity of A. aeolicus ⑀ and E. coli ⑀ was compared at 37°C using a 5Ј 32 P-end-labeled DNA 50-mer. Reactions were analyzed on a 12% denaturing (6 M urea) polyacrylamide gel, followed by autoradiography. The position of the 50-mer is indicated to the left of the gel. D, Superose 12 gel filtration analysis of a mixture of A. aeolicus ␣ and ⑀. Column fractions were analyzed in a 10% polyacrylamide gel stained with Coomassie Blue. Positions of ␣ and ⑀ are indicated to the right. 7. Analysis of A. aeolicus ␣ and . A, Elisa type assays of ␣Ϫ interaction. Wells were coated with either A. aeolicus ␣ or no protein and then were blocked and treated with A. aeolicus or no additional protein prior to detection with -specific antibody and anti-rabbit antibody coupled to horseradish peroxidase. A similar experiment using E. coli ␣ and E. coli is also shown. B, stimulates synthesis by ␣. The plot to the left demonstrates that E. coli stimulates ␣, but S. pyogenes has no effect (1st three lanes). Species-specific stimulation, an indication of specific ␣Ϫ interaction, is also observed by S. pyogenes pol C by S. pyogenes but not E. coli (lanes 5-8). Controls of E. coli alone and S. pyogenes alone are in lanes 4 and 7. The plot to the right shows that A. aeolicus ␣ is stimulated by A. aeolicus (lanes 9 -11). C, A. aeolicus ␣, , and a mixture of ␣ ϩ were analyzed by gel filtration. Column fractions were analyzed in a 10% SDS-polyacrylamide gel. The positions of ␣ and are indicated to the right of the gels. within 1 min indicating a speed of at least 120 nucleotides per s. Indeed, at the two highest ␣ concentrations, the synthetic time course is about the same indicating that the reaction is saturated with respect to ␣ at 2.64 g and above. At ␣ subunit concentrations lower than 0.88 g, less RFII product is formed at the 1-min time point compared with the 2-min time point. Furthermore, the immature products formed at 15 and 30 s become distinctly shorter as the ␣ subunit is titrated downward. The dependence of chain length on ␣ concentration indicates that ␣ does not remain attached to the ␤ clamp over the synthesis of the entire 7.2-kb circle. In other words, processivity of the ␣ subunit with the ␤ clamp is less than 7 kb. Most likely, ␤ remains tightly bound to DNA continuously as ␣ jumps from one ␤-containing template to the next during synthesis. This behavior is similar to that of the E. coli system when ␣ subunit is used with ␤ and ␥ complex instead of core or ␣⅐⑀ complex (9). In this case, the ␤ clamp confers a processivity of ϳ1-3 kb onto ␣, and the speed of the pol III ␣⅐␤ complex is about 300 nucleotides per s at 37°C (compared with about 1 kb/s when ␣⑀ complex is used at the same temperature).
The A. aeolicus pol III Replicase Is Thermostable-Next, we examined the thermostability of the A. aeolicus pol III replication system and its rate of synthesis at different temperatures.
These next experiments were performed using subsaturating pol III ␣ subunit in order to assess which temperature is most favorable, and therefore the rates of polymerization are less than the maximal rates observed in Fig. 8. The experiments were also performed in either the presence of SSB (a) or its absence (b). Proteins were preincubated with the DNA substrate for 2 min with two dNTPs to allow time for protein assembly onto DNA, and then synchronous synthesis was initiated upon addition of the remaining two dNTPs. Timed aliquots were quenched, and the products were analyzed in alkaline and native agarose gels. First we consider below the reactions that were performed in the presence of SSB.
At temperatures between 60 and 75°C, the A. aeolicus system yields significant levels of synthesis, and as the temperature is elevated, the rate of synthesis increases (Fig. 9A). Very little synthesis is observed at 55°C, which is presumably too low a temperature for the workings of this complex machinery from an extreme thermophile. The inactivity at low temperature may be ascribed, at least in part, to inactivity of the pol III ␣ polymerase, as simple gap filling assays using ␣ alone showed low activity at reduced temperature (see Fig. 6B). It remains possible that the clamp loading operation is also diminished at low temperature. The A. aeolicus SSB, however, would appear to remain functional at low temperature as it stimulates the E. coli pol III system at 37°C (data not shown).
Synthetic activity is greatly reduced at 80°C indicating that one or more components of the A. aeolicus pol III replicase denature at this temperature. DNA polymerase assays on gapped DNA at different temperatures also indicate that A. aeolicus pol III ␣ subunit loses activity at 80°C (Fig. 6B) suggesting that ␣ is the inactive component. However, it remains possible that yet other A. aeolicus components also lose activity at 80°C and higher. The thermostability of these proteins will be examined below. It is also possible that the primed template may lose efficiency in supporting synthesis at these high temperatures. In these experiments, we utilize a DNA 30-mer primer, the T m of which is below 80°C. However, we have raised the primer concentration in the assays to increase the occupancy of the primer on the ssDNA template at elevated temperature. To determine optimal primer concentration, we have titrated primer into reactions at different temperatures. The experiments shown here utilize 2 M primer; higher concentrations gave no greater synthesis at any temperature. We have also titrated a 90-mer DNA oligonucleotide primer into the assay at different temperatures but observed slightly lower activity than that obtained using the DNA 30-mer.
In the E. coli system, the primary role of SSB on synthesis is thought to be the removal of ssDNA secondary structure. Temperature can also remove secondary structure in ssDNA. It therefore seems reasonable to expect that SSB may not have as great of a stimulatory effect on synthesis at high temperature compared with a lower temperature. Ability of the A. aeolicus replication system to function above 60°C, where some secondary structure should melt without SSB, provides the opportunity to assess the effect of SSB on synthesis at high temperature. The replication reactions in Fig. 9B are identical to those in Fig. 9A, except that instead of adding SSB, only the SSB buffer is added to the reactions. The comparison demonstrates that synthesis in the absence of SSB is about the same as in its presence, consistent with the idea that SSB stimulates polymerase by removing secondary structure.
The temperature stability of each replicase component was determined in the experiment of Fig. 10. Each individual subunit, or complex (as indicated in each panel), was incubated for 2 min at elevated temperature in the absence of other components and then was shifted to ice before being assayed for activity. The thermostability of each component was tested under several different buffer conditions containing either 0.1% Triton X-100, 0.01% Nonidet P-40, 0.05% Tween 20, 4 mM CaCl 2 , 40% glycerol, or combinations of these reagents. Heattreated subunits were tested for activity by combining them with the other (unheated) subunits in the primed M13mp18 ssDNA synthesis assay. The assays contained ␣, ␦␦Ј, and ␤ (one of which was temperature-treated) and were conducted at 70°C for 1.5 min. Because this assay does not depend on the presence of SSB, we assayed the temperature stability of A. aeolicus SSB by substituting it for E. coli SSB in assays at 37°C using E. coli ␣, ␦␦Ј, and ␤ subunit.
The results in Fig. 10 show that A. aeolicus ␤ and SSB are by far the most thermostable components (B and D). Seventy five percent or more of their activity was retained under all conditions tested, even at 90°C. The ␦␦Ј clamp loader was also highly thermostable provided it was heat-treated in buffers lacking Tween (C). The ␣ DNA polymerase was the most heatsensitive of the A. aeolicus pol III replicase components (A). Under most conditions, the ␣ subunit lost 25-50% activity upon incubation at 80°C and lost 80 -100% activity upon 90°C treatment. However, polymerase activity was largely stabilized in the presence of 4 mM CaCl 2 and 40% glycerol, retaining over 50% activity after incubation at 90°C. As expected from the thermostability results of ␣ and ␦␦Ј, an equimolar mixture of ␣ and ␦␦Ј was generally thermolabile, but was most stabilized to heat treatment by 4 mM CaCl 2 and 40% glycerol (E).

Similarities to the E. coli pol III Replicase
The results of this study confirm that bacteria that grow at very high temperatures utilize the same overall strategy for processive replication as is observed in the mesophile, E. coli. Hence, A. aeolicus utilizes a pol III polymerase that is tethered to DNA by a ␤ sliding clamp which, in turn, requires a clamp loader complex for assembly onto DNA. Previous studies of thermophilic DNA polymerases have focused mainly on homologues to DNA polymerase I (reviewed in Ref. 62). However, recent studies (35,36) demonstrated the presence of ␥ and proteins in a thermophile which suggested they may use a multicomponent replicase like E. coli pol III holoenzyme for chromosomal replication. Since those studies, the genomes of several thermophilic organisms have been sequenced. These genomes contain at least some of the genes encoding subunits homologous to subunits of the E. coli pol III holoenzyme. This report confirms the presence of a working pol III machinery in the hyperthermophile, A. aeolicus, by reconstituting the machinery from recombinant subunits.
Studies in the E. coli pol III system showed that the ␥ (or ), ␦, and ␦Ј subunits are required to load the ␤ sliding clamp onto DNA (13). The A. aeolicus dnaX gene encoding was identified in the A. aeolicus genome report, but the holA and holB genes encoding ␦ and ␦Ј were not (41). We noted in our earlier study (38) of the S. pyogenes replicase that the ␦ subunit sequence was highly divergent from E. coli ␦. Sequence searches using the E. coli ␦ and ␦Ј sequences yield weak matches to open reading frames in the A. aeolicus genome and the current study positively identifies these genes as holA and holB by expressing the proteins they encode and demonstrating their function in the clamp loading reaction. Of the A. aeolicus proteins studied in this report, these two subunits display the least percent identity to the corresponding E. coli subunit sequences (see Fig.  1B).
Despite the rather wide sequence variations in A. aeolicus ␦ and ␦Ј compared with their E. coli counterparts, they are remarkably similar to the behavior of E. coli proteins in ability to form ␦␦Ј and ␦␦Ј complexes which remain associated during gel filtration analysis (13). Also like the E. coli system, the A. aeolicus ␤, once assembled onto DNA by ␦␦Ј, remains stably attached to DNA during gel filtration. Another similarity between these systems is that the A. aeolicus DNA polymerase III ␣ subunit, like E. coli ␣, lacks sequences that provide proofreading 3Ј-5Ј-exonuclease activity. The E. coli ␣ subunit is found tightly associated with the ⑀ 3Ј-5Ј-exonuclease which provides the proofreading function. Indeed, homologues to ⑀ are quite prevalent among bacteria. The dnaQ gene encoding A. aeolicus ⑀ was identified in the original A. aeolicus genome sequence report, and in this report we have cloned the gene and purified the corresponding protein. The A. aeolicus ⑀ homologue has 3Ј-5Ј-exonuclease activity, and our studies demonstrate that A. aeolicus ␣ and ⑀ interact. However, the association between the A. aeolicus ␣ and ⑀ subunits is not sufficiently strong to isolate an ␣⅐⑀ complex on a gel filtration column (as in the E. coli system). Inability to identify a tight A. aeolicus ␣⅐⑀ interaction will be elaborated upon below.
The present study demonstrates that A. aeolicus pol III ␣ subunit, when combined with ␤, is highly stimulated in synthesis of a primed M13mp18 ssDNA genome and extends the primer at approximately the same speed as E. coli ␣⅐␤. The current study also demonstrates that A. aeolicus ␣⅐␤ (plus ␦␦Ј) is not fully processive during primer extension around M13mp18, again like E. coli ␣⅐␤. Studies in the E. coli system demonstrate that the ⑀ proofreader is needed for optimal performance of the ␣ subunit (9). Whereas E. coli pol III core and ␣⑀ complex function with ␤ to synthesize DNA at a speed of about 1 kb/s (at 37°C), with a processivity greater than the 7.2-kb M13mp18 genome, the ␣ subunit travels with ␤ at a speed of about 300 nucleotides/s with a processivity of 1-3 kb (9). Hence, ⑀ binding to ␣ would appear to result in a more active form of the polymerase. These comparisons suggest that A. aeolicus ␣ may also bind another protein(s) to form a pol III "core" that has similar speed and processivity with ␤ as E. coli core.

Differences to the E. coli pol III Replicase
Subunit Interactions-It is widely anticipated that the polymerase and proofreading exonuclease activities will be tightly associated in chromosomal replicases because of the need for high fidelity to accurately duplicate an entire genome. Although A. aeolicus ␣ and ⑀ interact, the expectation that A. aeolicus ␣ and ⑀ would form a tight ␣⅐⑀ complex, as in the E. coli system, was not met. Because replicases across the evolutionary spectrum contain or have tightly associated proofreading exonuclease activity, it seems likely that A. aeolicus ␣⅐⑀ forms a tight complex in vivo. For example, the higher temperature at which A. aeolicus grows may strengthen the interaction between these two proteins. Alternatively, another protein may be required in A. aeolicus which serves as a brace, binding to both ␣ and ⑀, thereby stabilizing the ␣⅐⑀ contact (neither nor ␦␦Ј served this function; data not shown). Finally, ␣ and ⑀ may associate tightly when present on a primed template.
Another difference to the E. coli system is the relatively low affinity interaction between A. aeolicus ␣ and . Although we show herein that A. aeolicus ␣ and directly interact, they do not form a complex that can be isolated by gel filtration as is readily demonstrated for E. coli ␣⅐ complex (32). In the E. coli system the C-terminal sequences of are required for binding to ␣ and the DnaB helicase (32)(33)(34)40). In this regard, the A. aeolicus subunit is only 54.3 kDa, significantly less than E. coli (71.1 kDa) and only 7 kDa larger than E. coli ␥ (47.5 kDa), and this discrepancy between the length of E. coli and A. aeolicus is predominantly in the C-terminal region. For example, the N-terminal 40 kDa of A. aeolicus contains the greatest extent of the homology to E. coli and corresponds to the sequences needed to bind ␦, ␦Ј, and ATP for motor protein function. However, it remains possible that A. aeolicus ␣interaction increases at the high temperatures where this thermophile lives.
SSB-The SSB generally stimulates its cognate DNA polymerase. For example, T4 gp32 protein, E. coli SSB, and eukaryotic RPA all significantly enhance DNA synthesis by their respective DNA polymerase holoenzyme. The basis for this enhancement by SSB is generally believed to be the elimination of DNA secondary structure "road blocks" to polymerase translocation. Theoretically, SSB should slow polymerase chain extension, because some of the energy of nucleotide polymerization must be expended to displace tightly bound SSB from template ssDNA during its conversion to duplex. If SSB displacement is rate-limiting, or partially rate-limiting, its displacement will slow the intrinsic rate of DNA synthesis. If this is the case, the observed SSB stimulation of synthesis is probably due to elimination of kinetic barriers (i.e. DNA hairpins) that exert much more effect on polymerase speed than the barrier of SSB displacement. Thus, the net effect of substituting the SSB displacement barrier for the DNA secondary structure barrier is a stimulation of synthetic rate.
At high temperatures many DNA secondary structures may spontaneously melt. Therefore, for organisms that grow at high temperature, the DNA melting role of SSB may be less significant to polymerase progression. Indeed, the results of this report demonstrate that at elevated temperature, where most DNA secondary structure is probably eliminated, SSB has no significant effect on the efficiency of synthesis. Of course, it is possible that SSB may stimulate the hypothetical A. aeolicus core polymerase (in the presence of ␤, ␦␦Ј). It seems likely that SSB is needed for other important roles and thus is retained in thermophiles even if there is no significant need to melt DNA hairpins. For example, SSB coating of ssDNA on the lagging strand may be important to protect DNA against digestion by nucleases. Also, E. coli SSB helps in ordering the handoff of an RNA primer from primase to the DNA polymerase (51). This primase-to-polymerase switch has been conserved in evolution and involves specific protein-protein contacts with SSB. SSB is also required in some types of DNA repair and recombination (52).

Similarities to Other Replicases
Absence of , , and Subunits-Besides replicase subunits that play catalytic roles in clamp loading and polymerization, the E. coli DNA pol III holoenzyme contains three small subunits that are not absolutely required for these processes. The smallest of these, (8.6 kDa), associates directly with ⑀ subunit in the heterotrimeric pol III core polymerase (53). Absence of shows no defect in polymerase action or function of polymerase with the clamp loader and clamp in vitro (9,53). Consistent with this, the holE gene encoding can be deleted without noticeable consequence to E. coli growth and viability (11). Moreover, homologues to are not widespread among bacteria.
A search of the A. aeolicus genome does not yield a significant match to E. coli , and in fact a search of the entire GenBank TM yields only one significant match to . 2 In common with the genomic sequences of most other organisms, A. aeolicus shows no significant sequence matches to the E. coli and subunits of the clamp loader within DNA pol III holoenzyme. Biochemical studies in the E. coli system show that these subunits are not required for clamp loading action (46,54). What do these subunits do? The subunit binds to ␥ and increases the stability of ␥␦Ј complex and the ␥␦␦Ј complex (54,55). The subunit binds to in the ␥ complex clamp loader (54). The subunit also binds to SSB (56,57). The -to-SSB contact is involved in displacement of primase from the RNAprimed site, thereby helping the polymerase and primase to trade places on the RNA primer (51). This -SSB interaction also aids clamp loading under conditions of elevated ionic strength, and stimulates polymerase elongation as well (56,57). A PsiBlast search of GenBank TM using E. coli and sequences as queries shows a somewhat broader distribution of homologues to these two proteins among different bacterial species, although the list is still confined to only a few organisms. 3 It remains quite possible that yet other organisms may contain homologues to these two subunits, but sequence changes, combined with their small size, may preclude their identification by sequence comparison. This scenario is underscored by the difficulty of recognizing the essential ␦ subunit that is present apparently in all bacteria but difficult to identify by sequence searches.
Lack of ␥ Subunit-Examination of dnaX genes in Gen-Bank TM shows that the dnaX gene of some bacteria contains a frameshift sequence but several others do not. A. aeolicus dnaX contains neither the ribosomal translational frameshift sequence nor the transcriptional slippage sequence. Consistent with this, we demonstrate in this report that only the fulllength protein, designated , is detected by Western analysis of A. aeolicus whole cell extracts. Production of only the fulllength product of dnaX is by no means unique as we have shown previously (38) that only is produced from dnaX in the Gram-positive organism, S. pyogenes (as recombinant protein in E. coli).
The genes encoding clamp loader subunits in the T4 phage, yeast, human, and archaeal systems are not known to produce a truncated product. Hence, it may be more appropriate to ask why E. coli, and some other bacteria, go through the trouble of producing a truncated product from dnaX. The evolution of both translational and transcriptional mechanisms to produce ␥ (E. coli and T. thermophilus, respectively) would suggest that the truncated product plays an important role in the organisms that produce it. However, genetic studies show that the dnaX gene in E. coli can be mutated such that ␥ is not produced, with no growth defects (58). This dnaX mutant strain should make only the complex clamp loader, in which the three ␥ subunits are replaced by three subunits. This should also have the effect of dedicating the clamp loader to the replication fork because the C-terminal domains of will connect the clamp loader to core polymerases and the replicative DnaB helicase.
The fact that some bacteria have evolved frameshifting strategies to produce ␥ suggests that there is some role(s) for a ␥ complex clamp loader. For example, ␤ clamps must not only be loaded onto DNA but also need to be unloaded from DNA so they can be reused (45). The ␥ complex has been shown to be capable of removing ␤ clamps from DNA (21). Thus, a clamp loader that is physically separate from the replication fork may be important to clamp recycling. Furthermore, recent studies (59) reveal that ␤ interacts with all the E. coli DNA polymerases and also with other proteins such as MutS and DNA ligase. Therefore, there may be circumstances in which there is an advantage to having a clamp loader that is not dedicated to the replication fork via strong attachment to ␣ and DnaB helicase. In this regard, we would like to make the correlation that , produced by organisms that do not make ␥ (i.e. A. aeolicus and S. pyogenes), does not form as tight a contact with ␣ as observed (between ␣ and ) in the E. coli system. Hence, it is conceivable that the resulting complex may not be as committed to action at a replication fork and thus would be available to act at other sites, either to recycle ␤ from DNA or to load ␤ onto DNA for use in different processes, such as repair and recombination.
Thermophilic pol III and Its Use in Technology-Current DNA amplification techniques make use of relatively distributive single subunit DNA polymerases whose physiological role is likely to be more similar to the repair enzyme, DNA polymerase I, than a chromosomal replicase. It seems quite likely that a rapid pol III-type enzyme endowed with a sliding clamp for much higher speed, processivity, and fidelity would provide advantages over polymerases of a more distributive type. For example, a rapid and processive thermophilic DNA replicase may yield larger products in "long chain PCR" applications and could perhaps deliver more reliable performance in these procedures. The PCR format cycles the reaction through a high temperature DNA denaturing step. Except for ␣, the components of the A. aeolicus pol III replicase are quite thermostable and would likely withstand these high temperature treatments. The ␣ subunit is the most thermolabile component and is less capable of withstanding temperatures used in polymerase chain cycling reactions. However, A. aeolicus ␣ can withstand temperatures in the 50 -80°C range and thus may function well in high temperature isothermal amplification methods.
We have recently made the discovery in the E. coli system that ␤ functions with numerous proteins, including DNA polymerase I (59). Hence, it may very well be possible to derive benefit upon addition of the highly thermostable A. aeolicus ␤ clamp and ␦␦Ј clamp loader to current PCR protocols utilizing thermophilic DNA polymerase I homologues. These components should be capable of withstanding the repeated cycles through 95°C in the PCR procedure. Technological applications of these thermophilic replication components to new and existing procedures will be an exciting prospect for future study.