Characterization of an African Swine Fever Virus 20-kDa DNA Polymerase Involved in DNA Repair*

African swine fever virus (ASFV) encodes a novel DNA polymerase, constituted of only 174 amino acids, belonging to the polymerase (pol) X family of DNA polymerases. Biochemical analyses of the purified enzyme indicate that ASFV pol X is a monomeric DNA-directed DNA polymerase, highly distributive, lacking a proofreading 3′-5′-exonuclease, and with a poor discrimination against dideoxynucleotides. A multiple alignment of family X DNA polymerases, together with the extrapolation to the crystal structure of mammalian DNA polymerase β (pol β), showed the conservation in ASFV pol X of the most critical residues involved in DNA binding, nucleotide binding, and catalysis of the polymerization reaction. Therefore, the 20-kDa ASFV pol X most likely represents the minimal functional version of an evolutionarily conserved pol β-type DNA polymerase core, constituted by only the “palm” and “thumb” subdomains. It is worth noting that such an “unfingered” DNA polymerase is able to handle templated DNA polymerization with a considerable high fidelity at the base discrimination level. Base excision repair is considered to be a cellular defense mechanism repairing modified bases in DNA. Interestingly, the fact that ASFV pol X is able to conduct filling of a single nucleotide gap points to a putative role in base excision repair during the ASFV life cycle.

African swine fever virus (ASFV) encodes a novel DNA polymerase, constituted of only 174 amino acids, belonging to the polymerase (pol) X family of DNA polymerases. Biochemical analyses of the purified enzyme indicate that ASFV pol X is a monomeric DNA-directed DNA polymerase, highly distributive, lacking a proofreading 3-5-exonuclease, and with a poor discrimination against dideoxynucleotides. A multiple alignment of family X DNA polymerases, together with the extrapolation to the crystal structure of mammalian DNA polymerase ␤ (pol ␤), showed the conservation in ASFV pol X of the most critical residues involved in DNA binding, nucleotide binding, and catalysis of the polymerization reaction. Therefore, the 20-kDa ASFV pol X most likely represents the minimal functional version of an evolutionarily conserved pol ␤-type DNA polymerase core, constituted by only the "palm" and "thumb" subdomains. It is worth noting that such an "unfingered" DNA polymerase is able to handle templated DNA polymerization with a considerable high fidelity at the base discrimination level. Base excision repair is considered to be a cellular defense mechanism repairing modified bases in DNA. Interestingly, the fact that ASFV pol X is able to conduct filling of a single nucleotide gap points to a putative role in base excision repair during the ASFV life cycle.
Despite the variety of existing DNA polymerases, there are a few basic principles that are common to all these enzymes, irrespective of their role either in DNA replication or in DNA repair. The basic chemistry of each individual reaction always involves a pair of divalent metal ions that are coordinated by carboxylate residues. Such a two-metal ion mechanism, originally proposed by Beese and Steitz (1) and probably extrapolative to all nucleotidyltransferases (2)(3)(4), appears to be either evolutionarily conserved or acquired by convergent evolution of nonhomologous proteins. In addition to the general deoxynucleotidyl transfer mechanism, it appears that some structural convergency could apply also for the interaction with DNA, a common substrate; an overall view of the crystal structures available for DNA-dependent polymerases always shows a hand-shaped structure, with "thumb," "palm," and "fingers" subdomains, defining at least one cleft for holding DNA (5)(6)(7)(8)(9)(10)(11)(12). Moreover, both DNA replicases and DNA repair enzymes are often multienzymatic proteins, having built-in nucleolytic activities that exist as individual structural modules, separated from the polymerization domain. Thus, the paradigmatic Escherichia coli DNA polymerase I (pol I) 1 has a proofreading 3Ј-5Јexonuclease, and a 5Ј-3Ј-exonuclease to remove the RNA from the Okazaki fragments (13). Most DNA replicases are also endowed with a proofreading 3Ј-5Ј-exonuclease domain, having an evolutionarily conserved pol I-type active site (14). Reverse transcriptases have an RNase H activity, required for second strand DNA synthesis (reviewed in Telesnitsky and Goff (15)). Even pol ␤, the smallest of the known DNA polymerases (39 kDa), has a 8-kDa N-terminal domain that can excise a 5Јterminal base-free deoxyribose phosphate residue from incised abasic sites by a ␤-elimination mechanism (16,17). However, considering that DNA synthesis has to be productively coordinated with these various degradative functions, specific differences in the geometry of the enzyme/DNA interaction should be expected. Additional differences in complexity and/or specificity at the structural level among DNA polymerases could be the consequence of the necessity to interact with accessory proteins to increase processivity, or with specialized molecules (as tRNA or terminal proteins) acting as primers.
On the other hand, the maximal simplicity at the structural/ functional level could be envisioned for a DNA polymerase involved in a minimal DNA synthesis reaction, exemplified by the repair of a single nucleotide gap. Such a "custom made" DNA nucleotidyltransferase would require three carboxylate residues involved in metal binding and catalysis, a DNA binding site providing a very precise interaction with the templating base and the primer terminus, necessary for an efficient and faithful catalysis, and several residues devoted to bind and select the appropriate dNTP substrates. Such a minimal DNAsynthesizing enzyme should be able to dissociate promptly after catalysis, to facilitate the subsequent action of a ligase. In this sense, additional DNA binding subdomains or motifs other than those forming the catalytic core would be inconvenient.
African swine fever virus (ASFV), an enveloped deoxyvirus (170 kb), which causes a fatal disease to domestic pigs (18 -20), not only encodes a eukaryotic-type (family B) DNA polymerase involved in viral DNA replication (21)(22)(23) but, as shown in this study, also codes for a new member of the pol X family of DNA polymerases, a peculiar group that includes pol ␤ and terminal transferase (TdT) from vertebrates (24), mitochondrial pol ␤ from protozoans (25), and DNA polymerase IV (pol IV) from yeast (26,27). Interestingly, the ASFV pol X is only 20 kDa but efficiently catalyzes a single nucleotide gap repair reaction. A three-dimensional structure prediction for ASFV pol X based upon extrapolation to the crystal structure of pol ␤ (9, 28, 29) is discussed, as well as a putative role in BER of the viral DNA.
Amino Acid Sequence Comparisons-Multiple alignment of the DNA polymerase sequences shown in Fig. 1 was done in two steps. The initial alignment of the putative ASFV pol X protein with sequences of TdT and pol ␤ from different sources was done using the computer programs PILEUP and PRETTY from the University of Wisconsin Genetics Computer Group (30). At the second step, the multiple alignment obtained was adjusted manually, refining it on the basis of the secondary structure elements of rat pol ␤, as deduced from its crystal structure (9,28).
Expression of ASFV Protein pO174L in E. coli-The open reading frame containing the putative DNA polymerase X gene O174L from ASFV (31) was cloned into the pRSET-A bacterial expression vector, which allows the expression of recombinant proteins as fusions with a multifunctional leader peptide containing a hexahistidyl sequence for purification on Ni ϩ2 -affinity resins (32). First, open reading frame O174L was polymerase chain reaction-amplified from a plasmid clone of ASFV strain BA71V (31) using oligonucleotides RO-17 (with a 5Ј extension that includes a BamHI restriction site) and RCP-2 (with a 5Ј extension that includes a PstI restriction site). The resulting 0.5-kb polymerase chain reaction product was cloned at the BamHI/PstI sites of vector pRSET-A. The E. coli strain JM109 was used as a host for transformation. The construction of the recombinant expression plasmid, named pRSET-pol X was confirmed by DNA sequencing. Expression of the His-tagged pO174L protein was carried out in the E. coli strain BL21(DE3) pLysS, which contains the T7 RNA polymerase gene under the control of the isopropyl ␤-D-thiogalactopyranoside (IPTG)inducible lacUV5 promoter, and a plasmid constitutively expressing T7 lysozyme (33,34). Cells were transformed with plasmid pRSET-O174L and grown overnight in LB medium at 37°C. Flasks containing LB broth were inoculated with 0.1 volume of an overnight culture of E. coli and incubated in a rotatory shaker at 35°C until the absorbance at 595 nm reached 0.7. Then, IPTG (Sigma) was added to a final concentration of 0.4 mM and incubation was continued for 3 h at 35°C. Cells were collected by centrifugation for 15 min at 1,900 ϫ g, and washed twice with buffer A (50 mM phosphate buffer, pH 7.5, 500 mM NaCl, 20 mM imidazole). After resuspension in the same buffer, cells were sonicated on ice. Then, the suspension was cleared by centrifugation for 10 min at 1,900 ϫ g, at 4°C, and an aliquot of the supernatant was withdrawn as a "total extract" fraction. The recombinant 20.3-kDa protein was soluble under these conditions, since it remained in the supernatant after a new centrifugation for 20 min at 14,500 ϫ g, at 4°C. Induction, overproduction, and solubility of the recombinant protein was analyzed by polyacrylamide gel electrophoresis (PAGE) in the presence of SDS, in a 7-20% polyacrylamide gradient, and visualized by Coomassie Blue staining.
Purification of ASFV pol X-Ni-NTA agarose beads (QIAgen), previously equilibrated in buffer A, were added to the soluble fraction containing the recombinant protein, obtained as described above. After stirring for 2 h at 4°C, the resin was loaded into a 1.6-cm diameter column, and extensively washed with buffer A. The recombinant ASFV protein pO174L was eluted from the column with buffer B (50 mM phosphate buffer, pH 7.5, 500 mM NaCl, 500 mM imidazole). The eluate was loaded onto a 5-ml glycerol gradient (15-30%) containing 50 mM Tris-HCl, pH 7.5, 20 mM ammonium sulfate, 180 mM NaCl, 1 mM EDTA, and 7 mM 2-␤-mercaptoethanol, and centrifuged at 62,000 rpm (Beckman SW.50 rotor) for 26 h, at 4°C. After centrifugation, 20 fractions were collected from the bottom of the tube, examined in Coomassie Blue-stained gels, and tested for DNA polymerase activity on activated DNA.
In Situ Gel Analysis of DNA Polymerase Activity-The assay was carried out essentially as described previously (35). The protein samples were electrophoresed in 7-20% gradient SDS-PAGE gels containing 1.5 mg/ml activated calf thymus DNA (Pharmacia Biotech Inc.) as template-primer, followed by in situ renaturation of proteins and incubation of the gel in a DNA polymerase assay mixture. Prior to renaturation, the gel was washed twice with 50 mM Tris-HCl, pH 7.5, for 15 min at 4°C. Renaturation was allowed to occur during 3 h at 4°C in buffer C (50 mM Tris-HCl, pH 7.5, 6 mM (AcO) 2 Mg, 40 mM KCl, 16% glycerol, 0.01 mM EDTA, 1 mM dithiothreitol (DTT), and 400 g/ml bovine serum albumin (BSA)). In situ polymerization was assayed with buffer D (buffer C plus 2 M each dNTP and 1.2 nM [␣-32 P]dATP) for 12 h at 30°C. After washing unincorporated [␣-32 P]dATP from the gel, and in situ precipitation of the DNA with buffer E (5% trichloroacetic acid, 1% sodium pyrophosphate), the gel was dried and the activity bands (radioactively labeled) were detected by autoradiography.
DNA Polymerization on Activated DNA-The incubation mixture contained, in 25 l, 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM DTT, 4% glycerol, 0.1 mg/ml BSA, 100 nM [␣-32 P]dATP and 100 nM of the other three dNTPs, 1.25 g of activated calf thymus DNA as a substrate, and different fractions (2 l) from a glycerol gradient containing ASFV pol X DNA polymerase. When indicated, the activated DNA was omitted, and the activating MgCl 2 was either omitted or substituted by MnCl 2 . Butylanilino-dATP, butylphenyl-dGTP, phosphonoacetic acid, and aphidicolin (specific inhibitors of eukaryotic-type (family B) DNA polymerases), or ddNTPs (specific inhibitors of family X DNA polymerases) were added at the indicated concentrations. After incubation for 5 min at 37°C, the reaction was stopped by adding 10 mM EDTA, 0.1% SDS, and the samples were filtered through Sephadex G-50 spin columns in the presence of 0.1% SDS. The excluded volume, corresponding to the labeled DNA, was counted (Cerenkov radiation). Polymerization activity was calculated as the amount of incorporated dNMP.
DNA Polymerization Assays on Defined DNA Molecules-Terminal transferase activity was evaluated by using 5Ј-labeled SP1 as substrate. The incubation mixture contained, in 10 l, 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM DTT, 4% glycerol, 0.1 mg/ml BSA, different concentrations (up to 400 M) of the four dNTPs, 200 ng of purified ASFV pol X, and 3.2 nM of 5Ј-labeled SP1. After incubation for different times at 37°C, the reactions were stopped by adding EDTA up to 10 mM. Samples were analyzed by 8 M urea, 20% PAGE and autoradiography. DNA-dependent polymerization was assayed on different primer-template structures, obtained by hybridization of 5Ј-labeled SP1 (or SP1p) to the template oligonucleotide indicated in each case. The incubation mixture contained, in 10 l, 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM DTT, 4% glycerol, 0.1 mg/ml BSA, 3.2 nM of the hybrid indicated in each case, and the indicated amount and concentration of ASFV pol X DNA polymerase and the four dNTPs, respectively. After incubation for different times at 37°C, the reactions were stopped by adding EDTA up to 10 mM. Samples were analyzed by 8 M urea, 20% PAGE and autoradiography. Quantitation was done by densitometric analysis of the band(s) corresponding to primer extension products. The catalytic efficiency and nucleotide substrate specificity of ASFV pol X was assayed on the template-primer SP1/SP1cϩ6 (3.2 nM), essentially as described above. In this case, insertion of the nucleotide complementary to the two first template bases (providing various concentrations of either dATP (up to 10 M) or ddATP (up to 10 M) or ATP (up to 400 M) as the sole nucleotide) was assayed in the presence of 200 ng of ASFV pol X and either MgCl 2 (10 mM) or MnCl 2 (4 mM) as metal activator. To analyze the base specificity (nucleotide insertion fidelity) of ASFV pol X, the oligonucleotide SP1 was hybridized to four variants of the SP1cϩ6 template oligonucleotide (SP1cϩ6 (T), SP1cϩ6 (G), SP1cϩ6 (C), SP1cϩ6 (A), differing in the first template base. Nucleotide insertion on each hybrid structure (3.2 nM) was comparatively studied as described above, by providing 200 ng of ASFV pol X and various concentrations of either the correct dNTP (up to 20 M) or each of the three wrong dNTPs (up to 400 M). To obtain "gapped" structures of defined length, a third oligonucleotide was hybridized to the corresponding template-primer oligonucleotide (3.2 nM). Thus, a 1-nucleotide gapped structure was obtained by hybridization of SP1 and D(g1)P oligonucleotides to the template oligonucleotide SP1cϩ18(g1); 6-nucleotide gapped structures were obtained by hybridization of SP1 and either D(g6) or D(g6)P to the template oligonucleotide SP1cϩ18(g6). DNA polymerization on these structures was essentially as described above, in the presence of the indicated amounts of ASFV pol X and different sets of dNTPs. When indicated, T4 DNA ligase (0.5 units) was added to the reaction mixture to seal the repaired DNA gap.
DNA Gel Retardation Assay-5Ј-Labeled oligo SP1 or SP1p either alone or hybridized to different oligonucleotides were used to analyze the interaction of ASFV pol X with DNA. The incubation mixture, in a final volume of 20 l, contained 12 mM Tris-HCl, pH 7.5, 1 mM EDTA, 20 mM ammonium sulfate, 0.1 mg/ml BSA, 20 ng of ASFV pol X, and 3.2 nM of different single or hybrid oligonucleotides. After incubation for 5 min at 4°C, the samples were subjected to electrophoresis in 4% (w/v) polyacrylamide gels (80:1 monomer:bis), containing 12 mM Tris acetate, pH 7.5, and 1 mM EDTA, and run at 4°C in the same buffer at 8 V/cm, essentially as described previously (36). After autoradiography, DNA polymerase-DNA complexes were detected as a mobility shift (retardation) in the migrating position of the labeled DNA.

RESULTS
ASFV Protein pO174L Belongs to the Family X of DNA Polymerases-By complete DNA sequencing of the avirulent ASFV BA71V strain and subsequent data base searches for amino acid sequence similarities, it had been reported that the ASFV open reading frame designated as O174L potentially encoded a new member of the pol X family of DNA polymerases (31). This prediction was tested by the multiple alignment of this sequence with those of pol ␤ and TdT from several sources (24,25) and yeast pol IV (37), as representative members of DNA polymerase family X. A critical aspect to obtain the multiple alignment shown in Fig. 1 was the final adjustment carried out, as described under "Materials and Methods," on the basis of the reported crystal structure of rat pol ␤ (9, 28, 29). Thus, in general, the regions having a higher sequence similarity correspond to three-dimensional portions having a defined secondary structure, whereas the most variable regions among the DNA polymerases aligned correspond to randomcoiled portions connecting well ordered regions.
It has been reported that the N-terminal ϳ170 residues of yeast pol IV bear no homology to mammalian pol ␤ (26). According to the alignment shown here, the N-terminal 170 amino acid residues of yeast pol IV can be significantly aligned with the N-terminal portion of TdTs (similarity is indicated with gray boxes in Fig. 1). This finding together with the fact that this extra N-terminal domain had already been reported to be absent in pol ␤ in previous alignments of the pol X family (24,38) (see also Fig. 1), points to the existence of some specific structural/functional relationships among yeast pol IV and TdTs. Like pol ␤, ASFV pol X lacks this N-terminal extension, but also its two neighbor subdomains, i.e. the so-called "8 kDa" and "fingers" (green and yellow areas, respectively, in Fig. 1), present in all the other sequences aligned. The alignment of ASFV pol X starts at the C-terminal half of ␣-helix I, at the very end of the "fingers" subdomain, and extends through the "palm" and "thumb" subdomains (red and magenta areas, respectively, in Fig. 1). The regions corresponding to some stretches of secondary structure elements of both palm and thumb subdomains (as J-1-K-2, 5-M-N-6 -7, and O-3 10 ; see Fig.  1 for nomenclature) could be aligned without significant gaps among the different sequences, including ASFV pol X. This conservation at the primary structure level is probably the reflect of superior order constraints to form a conserved catalytic core. Another stretch of secondary structure elements (L-3-4) involves the same number of amino acid residues among ASFV pol X and pol ␤s, whereas in the case of TdTs and yeast pol IV, several amino acids (from 5 to 28 residues) are inserted among these elements. Mitochondrial pol ␤ differs from cellular pol ␤s in larger intervening sequences connecting secondary structure elements, particularly that (26 amino acid residues) between ␤-strand 7 and ␣-helix O. Contrarily, and contributing to its reduced size, ASFV pol X lacks some intervening amino acid sequences, such as that among ␤-strand 2 and ␣-helix L, which is particularly extensive in the case of yeast pol IV. In support of its functional significance, the inclusion of ASFV pol X in this multiple alignment outlines a very limited number (15) of invariant residues (indicated in white letters in Fig. 1), that include the catalytic triad of aspartates involved in metal binding (reviewed in Joyce and Steitz (39)). Interestingly, the thumb is the subdomain showing the highest similarity among the sequences compared, containing eight invariant residues. On the contrary, the palm subdomain, that contains seven invariant residues, including the three critical aspartates, allows larger differences in the number of residues connecting several secondary structure elements.
The multiple alignment shown in Fig. 1 (further described under "Discussion") strongly indicates an evolutionary relationship between ASFV pol X and DNA polymerases from family X. However, based on quantitation of the amino acid sequence similarity in the portion aligned (approximately 25, 23, and 21% of ASFV pol X residues align with identical residues in TdTs, yeast pol IV and pol ␤s, respectively), it is not possible to infer if ASFV pol X has a closer structural/functional relationship with a particular subgroup of the DNA polymerase X family. Moreover, poly(A) polymerases, template-independent enzymes like TdTs, have been recently shown to share significant homology to the palm subdomain of family X DNA polymerases (4). Therefore, it was critical to know whether the small ASFV pol X, lacking the first 141, 294, and 318 residues of human pol ␤, human TdT and yeast pol IV, respectively, was still endowed with DNA polymerase activity.
ASFV Protein pO174L Has Intrinsic DNA Polymerase Activity-The expression plasmid pRSET-pol X has the ASFV gene O174L, potentially coding for a pol X-like protein, under the control of a promoter for phage T7 RNA polymerase (see "Materials and Methods"). After induction of E. coli cells transformed with plasmid pRSET-pol X, a new polypeptide migrating at the expected position for protein pO174L (ϳ20 kDa) was observed in the crude extracts ( Fig. 2A, left panel, lane T). This protein, making up to 4% of total E. coli protein, remained soluble under the extraction conditions used ( Fig. 2A,  The predicted DNA polymerase activity of the recombinant pO174L was first examined using an in situ activity gel assay, carried out as described under "Materials and Methods." This technique takes advantage of the high resolution of the SDS-PAGE to separate the different polymerase species. Moreover, in this particular case, a neat conclusion was expected based on the low molecular mass (20 kDa) of the putative polymerase in comparison with the endogenous E. coli DNA polymerases. As shown in Fig. 2A (right panel), in addition to the radioactive bands corresponding to E. coli pol I and to its proteolytic derivatives, an activity band with the electrophoretic mobility of pO174L was observed in the crude extracts (T) and soluble fractions (S) from IPTG-treated cultures, but not in the uninduced cultures (Ϫ). Moreover, the purified protein (Ni-NTA fraction) showed DNA polymerase activity in the in situ assay, confirming that the overproduced 20-kDa pO174L has intrinsic DNA polymerase activity.
As an additional purification step, the Ni-NTA fraction was sedimented through a glycerol gradient, and the collected fractions were individually assayed for DNA polymerase activity on IV (37), mitochondrial pol ␤ from Chrithidia fasciculata (Mit. pol ␤) (25), and ASFV pol X (31). Relevant similarity among yeast pol IV and TdTs in the N-terminal portion is indicated by gray boxes. According to rat pol ␤ structural data (9,28), the alignment can be divided in four subdomains: 8-kDa (rat pol ␤ residues 1-86; green area), fingers (rat pol ␤ residues 87-148; yellow area), palm (rat pol ␤ residues 149 -260; red area), and thumb (rat pol ␤ residues 261-335; magenta area). ␣-Helices (lettered) and ␤-strands (numbered) are indicated at the top of the aligned sequences. The portions of the alignment that correspond to these secondary structure elements are included in blue areas. The protease-sensitive region defining the separation between the 8-kDa and the 31-kDa catalytic domain of rat pol ␤ is indicated with an arrow. Two helix-hairpin-helix motifs are between ␣-helices C and D (8-kDa subdomain), and between ␣-helices F and G (fingers subdomain). Invariant residues at the N-terminal extension, 8-kDa, and fingers subdomains are in boldface type and indicated with an asterisk at the top of the aligned sequences. Residues of the palm and thumb subdomains that are invariant among all the sequences compared (including ASFV pol X) are indicated with white letters (over a black background) and with an asterisk at the top of the alignment. Any other identities with ASFV pol X in these two subdomains are in boldface type.  2. A, expression of ASFV pO174L and in situ analysis of DNA polymerase activity. Left panel, Coomassie Blue staining after SDS-PAGE separation of control noninduced (Ϫ) and IPTG-induced extracts of E. coli BL21 (DE3) cells transformed with the recombinant plasmid pR-SET-pol X, the latter extracts analyzed as crude (T) and soluble (S) fractions, obtained as described under "Materials and Methods." The electrophoretic analysis of a highly purified fraction obtained after Ni-NTA chromatography is also shown. The electrophoretic migration of a collection of molecular mass markers is shown on the left. An arrow shows the expected position for protein pO174L (20 kDa). Right panel, autoradiography of the in situ gel analysis carried out as described under "Materials and Methods," after the SDS-PAGE separation (using an activated DNA-containing gel) shown in the left panel. The DNA polymerase activity band corresponding to the electrophoretic migration of protein pO174L is indicated as ASFV pol X. The activity bands corresponding to the endogenous E. coli DNA polymerase I (pol I) and its Klenow fragment are also indicated. B, Cosedimentation of ASFV pol X activity with a 20-kDa polypeptide. The recombinant ASFV protein pO174L, eluted from the Ni-NTA column, was sedimented on a glycerol gradient (15-30%) as described under "Materials and Methods." The inset shows a SDS-PAGE analysis followed by Coomassie Blue staining of gradient fractions 8 -19. Quantitation of the ASFV pol Xstained band corresponding to each fraction is expressed as optical density (a.u., arbitrary units). Arrows indicate the sedimentation position of several molecular mass markers centrifuged in control gradients. DNA polymerase activity present in each fraction, assayed on activated DNA as described under "Materials and Methods," is indicated as dNMP incorporation (pmol). activated DNA, as described under "Materials and Methods." As shown in Fig. 2B, the single activity peak observed perfectly overlapped with the protein peak, sedimenting as a 20-kDa polypeptide, corresponding to monomeric ASFV pol X. Fractions 13-15 were collected and pooled for further in vitro analysis of ASFV pol X activity. Table I summarizes some of the requirements for the DNA polymerase activity of ASFV pol X, evaluated on activated DNA as template. As expected, exogenous addition of DNA was an absolute requirement. Divalent metal ions, such as Mg 2ϩ (10 mM optimal concentration) or Mn 2ϩ (4 mM optimal concentration), were absolutely required, and both equally activated the polymerization reaction. Also, as expected, different specific inhibitors of the eukaryotic-type class (family B) of DNA polymerases, as phosphonoacetic acid, butylphenyl-dGTP, butylanilino-dATP, and aphidicolin (reviewed in Brown and Wright (40)), either only moderately affected or did not inhibit polymerization by ASFV pol X. On the other hand, ddNTPs strongly inhibited DNA synthesis at dideoxy/deoxy ratios even lower than those reported to inhibit pol ␤ and yeast pol IV (27).
Polymerization by ASFV pol X Is Template-Primer-dependent-Although ASFV pol X can be included in the ␤-type DNA polymerases (pol X family) based on structural criteria, this group includes also TdTs, a kind of DNA synthesizing enzymes that are not template-dependent. As shown in Fig. 3, ASFV pol X has no terminal transferase activity since it is not able to add dNMPs to the 3Ј-end of a single-stranded DNA molecule (SP1). Besides, both in the presence (Fig. 3) or in the absence (not shown) of dNTPs, ASFV pol X was unable to release dNMPs from the 3Ј-end of SP1 (as it would be detected by a reduction in size of the 5Ј-labeled molecule), indicating the lack of a 3Ј-5Ј-exonuclease activity. On the other hand, when the SP1 oligonucleotide was hybridized to a second oligonucleotide providing a 6-nucleotide template sequence (SP1cϩ6), ASFV pol X catalyzed the 5Ј-3Ј extension of the primer strand (SP1) up to six additional nucleotides. This result clearly indicates that DNA polymerization by ASFV pol X is template-primer-dependent. On the SP1/SP1cϩ6 template-primer structure, the DNA polymerization pathway was completely distributive, as the size of the elongated products was reduced by lowering the enzyme/DNA ratio (see Fig. 3). This behavior of ASFV pol X is very similar to those of mammalian pol ␤ and yeast pol IV (27,41,42), the only known DNA polymerases displaying distributive synthesis under similar reaction conditions. Apparently, these three enzymes dissociate from the DNA after each round of nucleotide addition.
Interestingly, ASFV pol X was completely unable to use a template-primer structure with a mismatch at the primerterminus (SP1p/SP1cϩ6 in Fig. 3). A similar result was obtained using Mn 2ϩ as metal activator (data not shown). It has been observed in other systems that inefficient mismatch extension is mainly due to a slow nucleotide insertion rate rather than to a low affinity for mismatched template-primer structures (43). In keeping with this, ASFV pol X showed a similar capacity to bind either SP1/SPcϩ6 or SP1p/SP1cϩ6, as determined by gel-shifting analysis (data not shown). Therefore, the inefficiency in extending a mismatch displayed by ASFV pol X might be interpreted as the requirement for a strict orientation of the primer-terminus at the polymerization active site for an efficient catalysis. Nucleotide Insertion by ASFV pol X Is Template-directed-The ability of ASFV pol X to select among the four deoxynucleotides (base discrimination) to catalyze a faithful templatedirected DNA synthesis, was preliminarily evaluated on the four template-primer structures depicted in Fig. 4, obtained as described under "Materials and Methods." The four dNTPs, at different concentrations, were individually assayed as a substrate for each of the four template-primer structures, thus representing the 16 possible template-substrate nucleotide pairs (4 matched ϩ 12 mismatched). Fig. 4 shows that, in all the cases, the labeled primer strand could be extended only by adding the correct (complementary) deoxynucleotide, but not by adding even an excess (400 M) of each of the three wrong (noncomplementary) deoxynucleotides. Similar results were obtained when Mn 2ϩ was used instead of Mg 2ϩ as metal activator.
Insertion Efficiency of Nucleotide Substrates-The efficiency of ASFV pol X in using different substrates (deoxynucleotides versus ribonucleotides and dideoxynucleotides) was evaluated using the oligonucleotide hybrid SP1/SP1cϩ6 as a templateprimer (Fig. 5A), which offers a T as the first and second nucleotide on the template. The insertion efficiency of the different complementary nucleotide (adenine) substrates was measured by quantifying the elongation products (16-mer/17mer) of the 5Ј-labeled primer (see "Materials and Methods") at different nucleotide concentrations. These experiments were carried out using Mg 2ϩ or Mn 2ϩ as the metal activator, since, in general, Mn 2ϩ -activated polymerases are error-prone and capable of incorporating "nonphysiological" substrates (44 -46). In this sense, a structural basis for metal ion mutagenicity has been recently proposed using pol ␤ as a model (47). As shown in Fig. 5B, dATP and ddATP were inserted with a similar efficiency, whereas incorporation of ATP was about 100-fold less efficient. When the concentration of dATP or ddATP was progressively increased up to 10 M, the catalytic efficiency (K cat / K m ) was also linearly increased. However, our approach does not allow a direct estimation of the true nucleotide incorporation rate, since the distributivity of the ASFV pol X is probably the rate-limiting step. Interestingly, the use of Mn 2ϩ as metal activator did not affect the insertion of ddATP and ATP, but produced a 2-3-fold reduction in the insertion efficiency of dATP.
Therefore, as expected for a DNA polymerase, ASFV pol X is able to "read" the 2Ј-position of the nucleotide, preferring deoxynucleotides versus ribonucleotides; moreover, this enzyme, like other members of the pol X family, does not discriminate nucleotides on the basis of a strong selection for the 3Ј-OH group.
Gap Filling by ASFV pol X-Pol ␤, a distributive enzyme on template-primer molecules, has been shown to be processive when filling short DNA gaps. Such a processive gap filling by pol ␤ was structurally and functionally related to the presence of the 8-kDa domain and strongly favored by the presence of a 5Ј-phosphate group at the end of the gap (42,48). To test whether ASFV pol X, which lacks the 8-kDa domain, is able to conduct processive synthesis on gapped molecules, we constructed 6 nucleotide-gapped DNA molecules bearing or lacking a 5Ј-phosphate group at the end of the gap (see "Materials and Methods"). As a control, we used the corresponding template-primer structure, i.e. lacking the downstream oligonucleotide. As shown in Fig. 6, ASFV pol X elongated the control template-primer structure producing a variety of elongation intermediates characteristic of a distributive pattern of DNA synthesis (panel A). On the 6 nucleotide-gapped structure lacking the 5Ј-phosphate (panel B), the pattern of elongation intermediates between positions 1 and 5 of the gap, obtained at different times, was very similar to panel A. As expected, the presence of a downstream oligonucleotide blocked DNA elongation, as deduced from the large accumulation of the band corresponding to the 6 nucleotide-extended primer. As it has also been described for pol ␤, ASFV pol X displayed a limited strand-displacement capacity, allowing further elongation (7 nucleotide-extended) of the primer strand. On the 6 nucleotidegapped structure bearing a 5Ј-phosphate (panel C), the elonga-tion pattern corresponding to the first two products was very similar to those shown in panels A and B; however, the rapid appearance of the band, corresponding to the fully repaired DNA gap, suggests that some recognition of the 5Ј-phosphate occurs at a 3-nucleotide distance from the primer, thus allowing a more processive gap-filling reaction. The presence of the 5Ј-phosphate group did not affect the appearance of stranddisplacement products.
Repairing a Single Nucleotide Gap-To analyze whether ASFV pol X is able to catalyze a minimal polymerization reaction (1 nucleotide addition) when the template information is also minimal (a single base), we prepared the 1 nucleotidegapped structure shown in Fig. 7A, as described under "Materials and Methods." When the complementary nucleotide (dATP) was provided, ASFV pol X was able to fill in the single nucleotide gap, generating a labeled 16 mer as the main product (Fig. 7B, lane 2); on the contrary, the reaction was very inefficient in the presence of a noncomplementary nucleotide (lane 3). As it occurred also during polymerization of a template-primer structure, ASFV pol X showed no discrimination against the dideoxy form of the complementary substrate (ddATP) in this minimal gap-filling reaction (lane 4).
When the reaction was carried out in the presence of T4 DNA ligase, the 16-mer product obtained with dATP (not detected) was immediately ligated to the downstream 5Ј-P oligonucleotide (12-mer), generating a 28-mer final (repaired) product (lane 6). This result agrees with the finding that different DNA ligases can substitute for each other in cell-free BER systems and that such back-up reactions may also occur in vivo (49). As a control, lane 5 shows that ASFV pol X is a strict requirement for this repair reaction. On the other hand, the simultaneous presence of ASFV pol X and DNA ligase did not prevent (inhibit) the insertion of ddATP at the 1-nucleotide gap (lane 8), although, as expected, the lack of an OH group at the 3Ј-end of the dAMP-extended 16-mer product precluded its ligation to the downstream oligonucleotide. DISCUSSION The entire DNA sequence of ASFV strain BA71V has been reported (31), revealing 151 open reading frames encoding proteins longer than 60 amino acids. Although ASFV genes are usually distant homologs of the corresponding genes from other organisms (50 -52), sequence similarity studies have proven useful for predicting the functional role of many ASFV proteins. As we characterize here, one of these predicted proteins is a pol ␤-type DNA polymerase most likely involved in viral DNA repair. As the smallest of the known DNA polymerases, ASFV pol X (20 kDa) could be envisioned as a good model for the understanding of the structural/functional basis of templated nucleotide selection. FIG. 3. Polymerization by ASFV pol X requires a matched templateprimer structure. The assay was carried out in the presence of 10 M dNTPs and 10 mM MgCl 2 as described under "Materials and Methods," using 3.2 nM of either a single-stranded oligonucleotide (SP1), or a matched (SP1/SP1cϩ6), or a mismatched (SP1p/SP1cϩ6) templateprimer structure as a substrate. After incubation for the indicated times at 30°C in the presence of the indicated amounts of ASFV pol X, 5Ј-3Ј extension of the 5Јlabeled (*) strand was analyzed by electrophoresis in 8 M urea, 20% polyacrylamide gels and autoradiography.

Biochemical Analysis of ASFV pol X
ASFV pol X has been characterized as a DNA-dependent DNA polymerase based on in situ gel analysis and in vitro evaluation of the purified protein. Taken together, the biochemical analysis of ASFV pol X demonstrated its functional relationship to ␤-type enzymes according to: 1) template-directed DNA synthesis, 2) distributivity during primer extension, 3) strong sensitivity to ddNTPs, 4) poor sensitivity to inhibitors of family B DNA polymerases, and 5) efficient filling of a single nucleotide gap. On the other hand, significant differences with pol ␤ are: 1) low processivity during gap filling synthesis by ASFV pol X, likely related to the lack of the 8-kDa domain of pol ␤ and 2) unaltered base discrimination during Mn ϩ2 -activated DNA synthesis.
Whereas Mg ϩ2 is probably the divalent metal ion utilized by most polymerases for catalysis in vivo (13), it has been suggested that Mn ϩ2 displays its mutagenic effect on polymerases by promoting greater reactivity than Mg ϩ2 at the catalytic site, thereby allowing the nucleotidyl transfer reaction to occur with little or no regard to template instructions. A structural basis for these differences in metal activation was recently obtained by crystallographic studies of pol ␤ complexed with DNA and dATP, showing that the side chain position of Asp 192 , one of the three critical carboxylate residues at the polymerization active site, varied as a function of the different metal activator used to obtain the crystals (47). Despite the limited sensitivity of our in vitro misincorporation assay, it appears that fidelity of DNA synthesis by ASFV pol X relies on a proper selection of the complementary nucleotide (insertion fidelity), that is not mark-edly affected by the metal ion used (Mg 2ϩ versus Mn 2ϩ ) as activator.
It is tempting to speculate on the existence of a direct relationship between the high distributivity of DNA synthesis displayed by ASFV pol X and its lack of discrimination against ddNTPs, both common features of the pol X family. It is conceivable that during evolution of DNA polymerases, a proper checking of the 3Ј-OH group of the nucleotide constituted an evolutionary advantage mainly for those polymerases involved in processive DNA synthesis, since the 3Ј-OH group of an incoming dNTP would be critical only for the next polymerization reaction, once dNMP is incorporated to the primer strand. On the other hand, it cannot be ruled out that, during DNA repair, some concerted action or interaction between a pol X enzyme and a specific DNA ligase involved in DNA repair could improve discrimination against ddNTPs, since the 3Ј-OH group is critical for the "next reaction," i.e. ligation reaction. Recently, a specific interaction of pol ␤ and DNA ligase I has been described in a multiprotein BER complex from bovine testis (53).

Three-dimensional Structure Prediction
Modular Organization-The resolution of the three-dimensional structure of rat pol ␤ (9, 28) (see Fig. 8A) demonstrated that its 335 amino acids are folded in two distinct domains, separated by a protease-sensitive region: a specific N-terminal 8-kDa domain (shown in green in Fig. 8A), implicated in template binding and deoxyribose phosphate excision, and a Cterminal 31-kDa domain, containing the polymerase catalytic FIG. 5. Nucleotide specificity of ASFV pol X. A, scheme of the templateprimer structure used to analyze nucleotide specificity. B, the assay was carried out as described under "Materials and Methods," using 3.2 nM of the templateprimer shown in A, in the presence of the indicated concentration of either dATP, ddATP, or ATP as a substrate, and using either Mg 2ϩ or Mn 2ϩ as activating metal ions. A control reaction in the absence of nucleotides (c) was also carried out. After incubation for 30 min at 30°C in the presence of 200 ng of ASFV pol X, extension of the 5Ј-labeled (*) strand was analyzed by electrophoresis in 8 M urea, 20% polyacrylamide gels and autoradiography.
FIG. 6. Processivity during a 6-bp gap-filling reaction. The polymerization assay was as described under "Materials and Methods," using 3.2 nM of either a template-primer structure (A), a 6-bp gapped DNA (B), or a 6-bp gapped/5Јphosphate DNA (C). After incubation for the indicated times at 30°C, in the presence of 10 mM MgCl 2 , 10 M dNTPs and 200 ng of ASFV pol X, the primer-extended products were analyzed by electrophoresis in 8 M urea, 20% polyacrylamide gels and autoradiography.
site (54,55). The catalytic domain showed also the common fold of three distinct subdomains (fingers (yellow), palm (red), and thumb (magenta), as colored in Figs. 1 and 8A), characteristic of different polymerase superfamilies, as shown in the Klenow fragment of pol I (5, 6), HIV-1 reverse transcriptase (8,56), T7 RNA polymerase (7), Taq DNA polymerase (10,57), Moloney murine leukemia virus reverse transcriptase (11), and Bacillus stearothermophilus DNA polymerase I (12). The primary sequence multiple alignment shown in Fig. 1, adjusted on the basis of the pol ␤ three-dimensional structure, allows the prediction of significant differences in the modular organization of domains and subdomains for the different members of the pol X family, but having in all cases the same linear arrangement (connectivity) along the primary sequence. Yeast pol IV and TdTs would contain an extra N-terminal domain followed by a pol ␤-type structure, whereas ASFV pol X could be considered as a lower limit, being mainly formed by the palm and thumb subdomains of pol ␤ (see Fig. 8A).
DNA Binding and Processivity-Interestingly, and despite its reduced size, ASFV pol X is still able to interact with different template-primer structures, catalyzing a distributive insertion of dNTPs according to template instructions. This capacity can be explained considering that four amino acid residues identified as DNA ligands in pol ␤ (9), mapping in the palm and thumb subdomains, could have a functional homolog in ASFV pol X as in the rest of pol X family members (see Figs. 1 and 8B); Lys 234 , located in the ␤-turn between strands 3 and 4 of the palm subdomain and interacting through the phosphates and base of the template strand, is invariant in all the sequences aligned, including ASFV pol X (Lys 85 ); Arg 254 , preceding ␤-strand 5 of the palm subdomain and hydrogen-bonded to the phosphate of the nucleotide forming the primer-terminus, is invariant in all sequences aligned in Fig. 1, with the exception of ASFV pol X. In the latter, a glutamine residue (Gln 98 ) could play an analogous function; Arg 283 , located in ␣-helix N of the thumb subdomain and H-bonded to the templating base, is invariant in all the sequences aligned, including ASFV pol X (Arg 127 ). Alteration of pol ␤ Arg 283 residue dramatically lowered fidelity (58, 59); Tyr 296 , located on a loop FIG. 7. Repairing a single-nucleotide gap with ASFV pol X. A, scheme of the 1-nucleotide gapped structure used as DNA substrate (3.2 nM). B, when indicated, 20 ng of ASFV pol X were used in the presence of 10 mM MgCl 2 , and either complementary (dATP (dA) or ddATP (ddA)), or noncomplementary (dGTP (dG)) nucleotides, at 10 M and 40 M, respectively. When indicated, T4 DNA ligase (0.5 unit) was simultaneously added with ASFV pol X. After incubation for 10 min (without ligase) or 60 min (with ligase) at 30°C, the primer-extended products corresponding either to the filling-in reaction or to the complete repair reaction (filling-in ϩ ligation) were analyzed by electrophoresis in 8 M urea, 20% polyacrylamide gels and autoradiography.
FIG. 8. Three-dimensional structure prediction for ASFV pol X inferred from the crystal structure of rat pol ␤. A, left, domain structure of rat pol ␤ (9, 28), showing its four different subdomains colored accordingly to the alignment shown in Fig. 1: 8 kDa (green), fingers (yellow), palm (red), and thumb (magenta). Right, a pol ␤-type core structure, formed by only palm and thumb subdomains, is predicted for ASFV pol X on the basis of the primary sequence alignment shown in Fig. 1. The pol ␤ structural elements absent in ASFV pol X are not colored. B, top view of the palm and thumb subdomains of rat pol ␤ (colored as in panel A) complexed with DNA, showing the amino acid residues that are invariant or highly conserved in ASFV pol X and the rest of pol X DNA polymerases. The palm subdomain, which consists largely of a ␤-sheet, constitutes an appropriate surface to accommodate the substrate molecules (nucleotides and DNA) and to present the pair of metal ions adequately for catalysis. The thumb subdomain forms a single wall of the original cleft described for the complete pol ␤ enzyme. between ␤-strands 6 and 7 of the thumb subdomain and interacting with the backbone phosphates of the template strand, aligns with Tyr 140 of ASFV pol X. In TdTs and yeast pol IV, this tyrosine is always substituted for a histidine residue. Among these generally conserved residues, Lys 234 and Arg 283 of pol ␤ were proposed to break up the water structure in the minor groove of the template-primer upon complex formation, as the first step in the B-DNA to A-DNA transition at the pol ␤ active site (60).
In addition to these DNA binding "core" residues, resolution of the crystal structure of pol ␤ complexed with different DNA substrates allowed the identification of two structurally similar, but functionally distinct, DNA binding motifs, named helix-hairpin-helix (HhH); the HhH motif present in the 8-kDa domain (formed by helices C and D; see Fig. 1) binds the 5Ј-phosphate end of a DNA gap (42), allowing the removal of a 5Ј-deoxyribose phosphate intermediate prior to the gap-filling step during the BER reaction (16,61). The HhH motif in the fingers subdomain (formed by helices F and G; see Fig. 1) helps to position the primer strand of the DNA substrate in the polymerase active site, and its coordinated metal ion was proposed to act as a "processivity cofactor" (61). A comparative analysis of the structurally solved HhH motifs pointed out their role in non-sequence-specific recognition of DNA, a feature that appears to be common in different proteins involved in BER processes (62). On the basis of the amino acid sequence alignment shown in Fig. 1, ASFV pol X lacks the first 141 amino acids of pol ␤, consequently lacking both HhH motifs. Accordingly, ASFV pol X retains a very limited ability to carry out processive synthesis, but only on very short (3 nucleotide) 5Јphosphate-gapped DNA molecules, suggesting that, in the absence of both the specific 8-kDa domain and most of the fingers subdomain, the small ASFV pol X requires a closer distance (shorter gap size) between the 3Ј-primer-terminus and the 5Јphosphate end of the gap for a processive DNA synthesis.
Nucleotide Binding and Catalysis-Despite some controversies on the relative orientation of different polymerase structures with respect to their bound DNA substrates (9,63,64), perhaps the most valid argument to support the idea of a generally conserved polymerization active site and reaction mechanism in different classes of nucleic acid-synthesizing enzymes is the finding of a similar three-dimensional arrangement of three acidic residues contained in the generally conserved motifs A and C (64 -67). These residues, always located in the palm subdomain, form a carboxylate triad implicated in the arrangement of a pair of divalent metal ions in a precise geometry to catalyze the nucleotidyl transfer reaction. As described for other DNA polymerases, mutation of these residues in pol ␤ (Asp 190 , Asp 192 , and Asp 256 ) resulted in severe loss of catalytic activity (68,69). As shown in Fig. 1, these three aspartates, invariant in all the sequences aligned, correspond to residues Asp 49 , Asp 51 , and Asp 100 of ASFV pol X.
Much of the nucleotide binding site observed in the ternary complex of pol ␤ (see Fig. 8B) is made up of the 3Ј-terminus of the primer strand and the template overhang. Additionally, the two metal ions in the pol ␤ active site bind to all three phosphates (␣, ␤, and ␥) of the incoming nucleotide. However, two pol ␤ residues interact by van der Waals contacts with the sugar moiety (C2Ј and C3Ј carbons) of the nucleotide, Phe 272 and Gly 274 , located at the end of ␣-helix M, and predicted to participate in nucleotide selectivity of DNA over RNA. This glycine residue is invariant in all the sequences aligned in Fig.  1, including ASFV pol X (Gly 118 ). The phenylalanine is conserved in ASFV pol X (Phe 116 ), whereas it is substituted either by a tyrosine in yeast pol IV or by a tryptophan in TdTs. Moreover, pol ␤ residue Arg 183 , located in ␣-helix K of the palm subdomain, and H-bonded to the negatively charged phosphate moiety of the nucleotide, is invariant in all the sequences aligned in Fig. 1, including ASFV pol X (Arg 42 ).
ASFV pol X Likely Has a Moveable Thumb-It has been shown that the thumb subdomain of pol ␤ undergoes a conformational change triggered upon nucleotide binding (60). It appears that the cis-peptide bond found between Gly 274 and Ser 275 (between ␣-helices M and N of the thumb subdomain) is critical for such a change from the more open conformation of the binary complex (enzyme-DNA) to the more closed conformation of the ternary complex (enzyme-DNA-dNTP). It has also been noticed that closing of the thumb not only positions the dNTP, but also the templating base (via interaction with the invariant Arg 283 at ␣-helix N) for an efficient and faithful nucleotide incorporation (58). It has also been proposed that a change back to the open conformation after nucleotide addition could be critical for facilitating the product-off step of catalysis, rate-limiting for the distributive mode of synthesis required for DNA repair (60). In addition to numerous hydrophobic interactions lining the palm-thumb interface, a salt bridge between pol ␤ residues Arg 182 (located in ␣-helix K) and Glu 316 (located in ␣-helix O) stabilizes the open thumb position. Interestingly, a substitution of Glu 316 to Lys in rat pol ␤ destroys the ability for complementing both pol I-type repair and replication of DNA in E. coli cells (70). On the contrary, the closed thumb position is stabilized by a hydrogen bond between Gly 179 and Phe 272 . As shown in Fig. 1, these four residues and those forming the cis-peptide bond are invariant or highly conserved in all members of the pol X family, being Arg 41 , Glu 156 , Gly 38 , Phe 116 , Gly 118 , and Pro 119 , the corresponding ASFV pol X residues. The conservation of the critical residues referred above and the fact that 15 out of 15 in vivo selected mutants of pol ␤ map in the thumb subdomain (70), emphasizes its functional importance and suggests that a mobile thumb subdomain analogous to that of pol ␤ is operating also in the small ASFV pol X and the rest of polymerases from the pol X family. It is also worth noting that pol ␤ residues Tyr 327 , Pro 330 , and Arg 333 , located at the C terminus of the thumb, are invariant in all the sequences aligned (see Fig. 1). Based on the spatial proximity of Tyr 327 and Arg 333 to the Arg 182 -Glu 316 pair (see Fig. 8B), it is tempting to speculate that these C-terminal residues could contribute to the dynamics of the thumb movement.
In summary, the domain organization of the different DNA polymerases forming family X indicates the existence of an evolutionarily conserved pol ␤-like catalytic core, which is naturally occurring as a minimal version in ASFV pol X. This core, formed by the palm and a mobile thumb subdomains, contains critical residues involved in catalysis, nucleotide binding, and DNA binding, thus having the capacity to form a catalytically competent ternary complex (see Fig. 8B). However, as it has been recently reported for family A DNA polymerases (71), RNA polymerases (72) and reverse transcriptases (73), it is conceivable that changes at the active site as small as a single amino acid residue could define individual parameters relative to nucleotide substrate specificity, catalytic rate, or nucleotide insertion fidelity. More evident differences as the presence of additional DNA binding subdomains (fingers) or domains , determine the enzyme processivity on physiologically relevant substrates and provide built-in nucleolytic activities to enhance the efficiency of nucleotide excision repair.
Putative Role of ASFV pol X in BER of Viral DNA DNA base damage generated by hydrolysis or oxidation can be repaired through a five-step pathway of enzymatic reactions (reviewed in Barnes et al. (74)), collectively referred to as BER: 1) hydrolytic cleavage of altered base-sugar bonds by glycosylases, 2) 5Ј-incision at abasic sites by an apurinic/apyrimidinic (AP) endonuclease, 3) removal of the 5Ј-terminal base-free deoxyribose phosphate, 4) short gap filling by a specific DNA polymerase, and 5) nick sealing by a DNA ligase. BER is considered to be a cellular defense mechanism repairing modified bases in DNA, but as shown in this report, ASFV could be endowed with such a defense mechanism, in which a very small pol X DNA polymerase would be responsible for a single-nucleotide gap-filling synthesis. Moreover, the analysis of the ASFV genome reveals that, in addition to ASFV pol X, the virus encodes for other enzymes to actively perform most of the stages of the BER pathway. Thus, ASFV protein pNP419L is a DNA ligase (52), and ASFV protein pE269R is a putative AP endonuclease of class II, able to perform hydrolysis of the phosphodiester bond at the 5Ј-side of AP sites, and potentially having also 3Ј-phosphatase and 3Ј-repair diesterase activities to remove the 5Ј-terminal deoxyribose phosphate (31). Thus, the single nucleotide gap generated by this type of AP endonuclease (74) could be filled in by ASFV pol X and sealed by ASFV DNA ligase. Therefore, ASFV pol X would not require an intrinsic nucleolytic activity, as that contained in the 8-kDa subdomain of pol ␤. Data bank searches of the ASFV genome failed to identify a gene potentially encoding a DNA glycosylase. It is worth noting that, although AP sites could be produced by a more or less specific DNA glycosylase action, they can also be introduced, at a significant rate, by nonenzymatic depurination in vivo (74).
Northern and Western blotting analyses of the expression of ASFV pol X gene in infected cells indicated that ASFV pol X is not an early protein. 2 Its appearance at the late stage of the viral infection, together with the fact that ASFV expresses Bcl2-like and IAP-like proteins to delay programmed cell death (75)(76)(77)(78), suggests that the ASFV repair machinery might be timed to repair damage to its DNA, produced in a late cellular response to the viral infection. Whereas the in vitro studies presented in this report provide indirect evidence of a role of ASFV pol X in viral DNA repair, more direct studies using recombinant viruses depleted of the corresponding pol X gene are required to demonstrate its participation in BER processes in vivo and its relevance for ASFV pathogenicity.