Purification and Characterization of polκ, a DNA Polymerase Encoded by the Human DINB1 Gene*

The Escherichia coli dinB gene encodes DNA polymerase (pol) IV, a protein involved in increasing spontaneous mutations in vivo. The protein-coding region ofDINB1, the human ortholog of DNA pol IV, was fused to glutathione S-transferase and expressed in insect cells. The purified fusion protein was shown to be a template-directed DNA polymerase that we propose to designate polκ. Human polκ lacks detectable 3′ → 5′ proofreading exonuclease activity and is not stimulated by recombinant human proliferating cell nuclear antigen in vitro. Between pH 6.5 and 8.5, human polκ possesses optimal activity at 37 °C over the pH range 6.5–7.5, and is insensitive to inhibition by aphidicolin, dideoxynucleotides, or NaCl up to 50 mm. Either Mg2+ or Mn2+ can satisfy a metal cofactor requirement for polκ activity, with Mg2+ being preferred. Human polκ is unable to bypass a cisplatin adduct in the template. However, polκ shows limited bypass of an 2-acetylaminofluorene lesion and can incorporate dCTP or dTTP across from this lesion, suggesting that the bypass is potentially mutagenic. These results are consistent with a model in which polκ acts as a specialized DNA polymerase whose possible role is to facilitate the replication of templates containing abnormal bases, or possessing structurally aberrant replication forks that inhibit normal DNA synthesis.

The Escherichia coli dinB gene encodes DNA polymerase (pol) IV, a protein involved in increasing spontaneous mutations in vivo. The protein-coding region of DINB1, the human ortholog of DNA pol IV, was fused to glutathione S-transferase and expressed in insect cells. The purified fusion protein was shown to be a templatedirected DNA polymerase that we propose to designate pol. Human pol lacks detectable 3 3 5 proofreading exonuclease activity and is not stimulated by recombinant human proliferating cell nuclear antigen in vitro. Between pH 6.5 and 8.5, human pol possesses optimal activity at 37°C over the pH range 6.5-7.5, and is insensitive to inhibition by aphidicolin, dideoxynucleotides, or NaCl up to 50 mM. Either Mg 2؉ or Mn 2؉ can satisfy a metal cofactor requirement for pol activity, with Mg 2؉ being preferred. Human pol is unable to bypass a cisplatin adduct in the template. However, pol shows limited bypass of an 2-acetylaminofluorene lesion and can incorporate dCTP or dTTP across from this lesion, suggesting that the bypass is potentially mutagenic. These results are consistent with a model in which pol acts as a specialized DNA polymerase whose possible role is to facilitate the replication of templates containing abnormal bases, or possessing structurally aberrant replication forks that inhibit normal DNA synthesis.
We previously reported the cloning and characterization of the human DINB1 and mouse Dinb1 genes, mammalian orthologs of the Escherichia coli dinB gene (1) and members of the UmuC/DinB superfamily of DNA polymerases (2). Expression of the E. coli dinB gene is tightly regulated by the SOS system (3). Following exposure of E. coli cells to DNA-damaging agents such as ultraviolet (UV) radiation, induction of dinB results in enhanced spontaneous (untargeted) mutagenesis of phage DNA introduced into the bacteria subsequent to irradiation (4). Increased spontaneous mutagenesis is also observed following overexpression of dinB in cells transfected with FЈlac plasmids, with the most prevalent mutations detected being Ϫ1 frameshifts (5). Recombinant E. coli DinB protein carrying a 6-histidine tag was purified and shown to be a DNA polymerase, designated DNA pol IV of E. coli, which is devoid of detectable exonuclease activity (6). Consistent with its apparent ability to generate frameshift mutations in vivo, DNA pol IV is able to extend a misaligned primer-template in vitro, resulting in a Ϫ1 frameshift mutation (6). More recently, DNA pol IV has been shown to be unable to efficiently bypass an abasic site, thymine dimer, or 6-4 photoproduct in vitro (7). Based on these observations, it has been suggested that DNA pol IV is a specialized enzyme whose role is to negotiate sites of stalled or arrested DNA replication caused by structurally abnormal replication forks, such as those caused by slippage at repeated sequences (2,6,7).
Human DINB1 cDNA is predicted to encode a polypeptide with a molecular mass of 99 kDa, which shares extensive amino acid sequence homology with E. coli DNA pol IV, including proposed catalytic domains (1,8). As is the case for the E. coli dinB gene (5), overexpression of the mouse Dinb1 cDNA in murine cells is associated with an approximately 10-fold increase in spontaneous mutations (8). These observations suggested that the protein encoded by the human DINB1 gene might also function as a DNA polymerase. In the present study we have used a baculovirus expression system to express and purify a GST 1 -human DinB1 fusion protein from insect cells and show that the purified protein is a template-directed DNA polymerase in vitro. The enzyme has no detectable 3Ј 3 5Ј exonuclease activity and has optimal DNA polymerase activity at 37°C and pH 6.5-7.5 (over the range 6.5-8.5) in the presence of Mg 2ϩ cations. The enzyme is insensitive to inhibition by aphidicolin or dideoxynucleotides, and retains optimal activity in the presence of NaCl at levels Յ50 mM. We designate this polymerase activity DNA pol. Human pol is unable to bypass a cisplatin adduct in vitro, but has limited ability to bypass an AAF lesion in an error-prone fashion under the same conditions.

EXPERIMENTAL PROCEDURES
Media and Biochemical Reagents-Insect cell TMN-FH media was purchased from Pharmingen. The Klenow fragments of E. coli DNA polymerase I (exo ϩ and exo Ϫ ) were obtained from New England Biolabs. Aphidicolin was from Sigma. Dideoxynucleotides were from U. S. Biochemicals. Glutathione-Sepharose was from Amersham Pharmacia Biotech. The protease inhibitor mixture was purchased from Roche Molecular Biochemicals. N-Acetoxy-2-acetylaminofluorene (AAAF) was obtained from the National Cancer Institute. cis-Diamminedichloroplatinum(II) was purchased from Aldrich.
Expression of Wild-type and Mutant GST/pol-The human DINB1 open reading frame was amplified by high-fidelity polymerase chain reaction using HeLa cell cDNA as template with primers HDinB5Ј (5Ј-GTGGATCCGCCATGGATAGCACAAAGGAGAAGTG-3Ј) and HDinB3Ј-His 6 (5Ј-ATGGATCCGCGGTCGACTAATGGTGGTGATGAT-GGTGCTTAAAAAATATATCAAGGGTATG-3Ј) that introduce BamHI restriction sites (underlined) on both the 5Ј and 3Ј ends of the amplified fragment as well as six histidine residues on the 3Ј end. The polymerase chain reaction product was cloned into pGEM-T Easy (Promega) to generate pHDINB1-6His and sequenced to confirm the integrity of the coding region. The 2.6-kilobase BamHI fragment containing the human DINB1 coding region was then cloned into the same site of pAcG2T (Pharmingen) to generate an in-frame fusion with the glutathione Stransferase gene, generating plasmid pAcG2T/HDINB1-6His.
These plasmids were co-transfected into SF9 cells with BaculoGold DNA using a BaculoGold transfection kit (Pharmingen). Expression of both wild-type and mutant GST/pol was assayed by immunoblotting with anti-GST antisera. Two rounds of amplification produced a high titer stock of recombinant virus expressing GST/pol. The multiplicity of infection yielding optimal expression of full-length fusion proteins was determined empirically.
Purification of GST/pol-Approximately 1 ϫ 10 8 virus-infected Sf9 cells were harvested 3 days after infection and lysed in 20 ml of Lysis Buffer I (1% Triton X-100, 10 mM Tris-HCl, pH 7.5, 10 mM Na 2 HPO 4 , pH 7.5, 1 mM EDTA, 5 MM ␤-mercaptoethanol, 1 ϫ protease inhibitors) by incubation on ice for 10 min. Insoluble material was removed by centrifugation to give the cytoplasmic extract. The pellet was resuspended in 20 ml of Lysis Buffer I containing 500 mM NaCl, and incubated on ice for 10 min. Insoluble material was removed by centrifugation to generate nuclear extract. The nuclear extract was diluted 2-fold and bound in batch to 500 l of glutathione-agarose for 2 h at 4°C. The resin was harvested by centrifugation and most of the supernatant removed. The resin was resuspended in the remaining supernatant and transferred to a 10-ml disposable column (Bio-Rad) to collect the resin by gravity. The resin was washed with 5 ml of Lysis Buffer I containing 250 mM NaCl, followed by 5 ml of Wash Buffer II (10% glycerol, 100 mM NaCl, 20 mM Tris-HCl, pH 7.5, 0.01% IPEPAL-630, 5 mM ␤-mercaptoethanol, 1 ϫ protease inhibitors). Bound protein was eluted with 3.5 ml of Wash Buffer II containing 10 mM reduced glutathione, and collected in a total of 10 fractions of 350 l each. GST/pol-containing fractions (determined by SDS-PAGE and immunoblotting) were aliquoted, frozen in liquid nitrogen, and stored at Ϫ80°C. GST/pol DNA polymerase activity was stable to multiple rounds of freezing and thawing. The mutant fusion proteins were expressed and purified by the same procedure.
Preparation of AAF-DNA Template-A synthetic DNA oligo of the sequence 5Ј-TCCTTCTGTCTCTT-3Ј (site of modification underlined) was purified by denaturing PAGE and desalted using a Sep-Pak C 18 cartridge (Waters, manufacturer's instructions). 0.38 mol of purified DNA was dissolved in TE buffer (45 mM Tris-HCl, 1 mM Na 2 EDTA, pH 8.0) to a concentration of 0.2 mM and incubated with AAAF (1 mM, 10% ethanol, 37°C, 12 h). Adducted DNA was admixed with an equal volume 8 M urea and purified by denaturing PAGE. The AAF-modified oligo has retarded gel mobility relative to the unmodified oligo, allowing efficient removal of unmodified DNA. DNA was eluted and concentrated using a Sep-Pak C 18 cartridge.
The presence of the AAF-modified deoxyguanosine residue was confirmed by electrospray mass spectrometry (ESI-MS) and by HPLC analysis of enzymatically digested DNA. HPLC separation of the enzymatic hydrolysate returned peaks corresponding to deoxycytidine, thymidine, and a strongly retained substance that co-eluted with the product of the reaction between deoxyguanosine and AAAF. HPLC analysis was conducted similarly to a previous report (9). The ESI-MS of each of these co-eluted substances was shown to correspond to N-(de- , the expected lesion. The AAF modification was assumed to be at C-8 of deoxyguanosine because AAAF is known to modify DNA preferentially at this position (10 -12). No appreciable peaks were observed in the HPLC assay that could potentially be attributed to modification at other positions of deoxyguanosine or at other bases on the DNA oligo.
7.6 pmol of the AAF-modified oligo was admixed with 1.1 molar equivalents of each of the two flanking DNA sequences 5Ј-TTTC-CCAGTCACGACG-3Ј and 5Ј-TCAGTGAATTCGAGCTCGGAGCC-3Ј and approximately 0.9 equivalents of the splinting DNA oligo, 5Ј-CACT-GAAAGAGACAGAAGGACGTCGT-3Ј. DNAs were heated briefly (90°C, 1 min) and annealed by cooling to room temperature on the benchtop (5 min). Annealed DNAs were ligated with T4 DNA ligase (1 ϫ in manufacturer's buffer, 16°C, 12 h), ethanol precipitated, and purified by denaturing PAGE. The 53-mer product band (5Ј-TTTCCCAGT-CACGACGTCCTTCTGTCTCTTTCAGTGAATTCGAGCTCGGAGCC-3Ј, site of AAF modification underlined) was identified by mobility relative to marker oligos and to mock-treated reactions in which one or more of the splicing components had been left out of the mixture.
An unmodified 53-mer oligonucleotide of the above sequence was used as a wild-type control template. The unmodified and AAF-modified templates were annealed to either a gel-purified running start oligo, P1-PF-RS, 5Ј-GGCTCCGAGCTCGAATTCAC-3Ј or a gel-purified standing start oligo, P2-alk-SS, 5Ј-GGCTCCGAGCTCGAATTCACTGAAA-GAGA-3Ј as described previously.
Preparation of Platinated DNA Template-A synthetic DNA oligo of the sequence 5Ј-TCCTTCGGTCTCTT-3Ј (site of modification underlined) was purified by denaturing PAGE and desalted using a Sep-Pak C 18 cartridge. An aqueous solution of cis-diamminedichloroplatinum(II) (1 mg/ml) was freshly prepared and preincubated in the dark (25°C, 12 h). 0.043 mol of purified DNA was dissolved in water to a concentration of 0.5 mM, admixed with 1 molar equivalent of cisplatin from the preincubated solution and incubated (5 h, 37°C, wrapped in foil). Crude cisplatin-modified DNA was gel-purified. The presence of the cisplatin adduct was confirmed by matrix-assisted laser desorption ionizationmass spectrometry and was presumed to be a 1,2-intrastrand d(GpG) cross-link based on previous reports that cisplatin modifies N-7 of purines (reviewed in Ref. 13). A 5Ј end-labeled aliquot of the sample confirmed that the modified oligonucleotide migrated as a single band on a denaturing gel and that its mobility was distinct from unmodified DNA and a cisplatin diadduct. The purified cisplatin-modified 14-mer was ligated to flanking sequences as described above for AAF-modified DNA except that the splinting DNA in ligation reactions was composed of the sequence, 5Ј-CACTGAAAGAGACCGAAGGACGTCGT-3Ј. The 53-mer product (5Ј-TTTCCCAGTCACGACGTCCTTCGGTCTCTTTCA-GTGAATTCGAGCTCGGAGCC-3Ј, site of cisplatin modification underlined) was purified by denaturing PAGE and identified similarly to the AAF-template.
An unmodified 53-mer oligonucleotide of the above sequence was used as a wild-type control template. The unmodified and cisplatinmodified templates were annealed to the running start oligo, P1-PF-RS, as described previously.
DNA Polymerase Assays-Standard polymerase reactions (10 l) were performed in 50 mM Tris-HCl (pH 7.0), 5 mM MgCl 2 , 1 mM dithiothreitol, 10 mM NaCl, 1% glycerol with 100 M dNTPs, 2 nM GST/pol, and 5 nM primer-template for 5 min at 37°C unless indicated otherwise. Reactions were terminated by the addition of 1 l of 0.5 M EDTA, concentrated under vacuum, and resuspended in 5 l of loading dye (90% deionized formamide, 0.1 ϫ TBE, 0.03% bromphenol blue, 0.03% xylene cyanole FF). Following denaturation at 95°C for 2 min, products were resolved by electrophoresis on 12% polyacrylamide gels containing 8 M urea. Gels were dried under vacuum and exposed to film at room temperature.
PCNA Experiments-Human recombinant PCNA (provided by R. Wood, Imperial Cancer Research Fund, Clare Hall Laboratories, South Mimms, Hertfordshire EN6 3LD, United Kingdom) and calf thymus pol␦ (provided by U. Hü bscher, Department of Veterinary Biochemisty, University of Zurich-Irchel, CH-8057 Zurich, Switzerland) were pre-pared as described (14,15). DNA polymerase assays were performed using the standard conditions described above unless indicated otherwise.

Human DinB1
Protein Is a DNA Polymerase, pol-To determine whether the product of the human DINB1 gene is a DNA polymerase, we expressed and purified recombinant human DinB1 protein. Expression in both E. coli and the yeast Schizosaccaromyces pombe consistently resulted in low yields and/or degraded protein (data not shown). However, we were able to express full-length hDinB1 protein fused to GST in insect cells using a baculovirus expression system. The recombinant GST/hDinB1 protein was purified to near physical homogeneity from nuclear extracts by affinity chromatography on glutathione-agarose (Fig. 1A, lane 1). The purified GST/ hDinB1 fraction contained primarily full-length fusion protein; however, some degradation products, including free GST, were observed and confirmed by immunoblotting with anti-GST antisera (data not shown).
To test for DNA polymerase activity, various 5Ј-32 P-end labeled oligonucleotide primers were annealed to a 44-nucleotide template and used as substrates. In the presence of dNTPs and Mg 2ϩ , the Klenow fragment of E. coli DNA polymerase I efficiently extended the primer to generate the expected 44-nucleotide product (Fig. 1B, lane 4). Purified GST/hDinB1 protein also extended the primer, demonstrating an intrinsic DNA polymerase activity (Fig. 1B, lanes 2 and 3). We therefore propose to rename the human DinB1 protein as DNA polymerase (pol) and the gene encoding it, POLK, in accordance with standard nomenclature for eukaryotic DNA polymerases (16,17). This designation has been approved by the Human Genome Organization nomenclature committee.
As a control, a GST/hDinB1 mutant protein in which the conserved amino acid residues Asp 198 and Glu 199 were changed to alanine, was purified by the same procedure (Fig. 1A, lane 2) and shown to be devoid of detectable DNA polymerase activity (Fig. 1C, lanes 3-6), indicating that the observed polymerase activity is intrinsic to the human DinB1 protein. In addition a truncated GST/hDinB1 fusion protein lacking 360 amino acids at the C terminus (GST/hDinB1⌬C) did not demonstrate DNA polymerase activity, indicating that sequences within this less highly conserved portion of the protein are required for activity (data not shown).
We performed a series of experiments to determine the optimal conditions for pol DNA polymerase activity in vitro. As shown in Fig. 2A, between pH 6.5 and 8.5 GST/pol was most active over the range 6.5-7.5 ( Fig. 2A, lanes 1-5), with reactions carried out at 37°C (Fig. 2A, lanes 6 -9). To investigate the effect of ionic strength on DNA synthesis, increasing amounts of NaCl were added to the reactions ( Fig. 2A, lanes  11-16). GST/pol activity was relatively insensitive to NaCl concentration up to 50 mM, but was significantly inhibited at salt concentrations of 100 mM or higher. As expected, a metal cofactor was required for activity (Fig. 2B). Either Mg 2ϩ or Mn 2ϩ could be utilized, with the former being preferred (Fig.  2B, compare lanes 2 and 3). Based on these observations, all standard DNA polymerase assays using GST/pol were performed at pH 7.0 and 37°C in the presence of Mg 2ϩ .
The range of incomplete extension products produced by GST/pol in the experiments described above suggested that human pol is endowed with limited processivity, as has also been observed for the E. coli DinB protein (6, 7). We therefore examined whether purified human PCNA, a sliding clamp known to stimulate the processivity of the replicative DNA polymerases pol␦ and pol⑀ (18 -20), increases the extent of DNA synthesis by GST/pol. The ability of PCNA to stimulate the activity of pol␦ on short oligonucleotide-derived primer-templates has been observed previously (21). As shown in Fig. 3, addition of recombinant human PCNA had no detectable effect on GST/pol activity (Fig. 3, lanes 1-4) but could readily stimulate synthesis of full-length products by purified pol␦ (Fig. 3,  lanes 5-7) on a slightly longer template. As was observed on the shorter template, no effect of PCNA on pol activity was observed on the longer template nor did the PCNA possess polymerase activity of its own (data not shown). In contrast to pol, E. coli pol IV is stimulated 3000-fold by addition of the bacterial sliding clamp ␤,␥ complex (7). A comprehensive study of the effects on replication factors RPA, RFC, and PCNA on pol is currently underway.
pol Is a Template-directed DNA Polymerase Lacking 3Ј 3 5Ј Proofreading Exonuclease Activity-To demonstrate that GST/ pol is a template-directed DNA polymerase we performed polymerase assays in the presence of single deoxyribonucleoside triphosphates (dNTPs) on 4 different primer-templates, each designed to test for the correct incorporation of a particular dNTP. As shown in Fig. 4A, under the single set of conditions tested GST/pol preferentially incorporated the correct nucleotide on each template. However, in all cases significant levels of misincorporation were also observed. For example, on the 27/44 primer-template GST/pol primarily catalyzed the accurate incorporation of dGTP as the first nucleotide, but also supported misincorporation of dATP and to a lesser extent dCTP (Fig. 4A, lanes 7-11). It can also be observed from Fig. 4A that the level of GST/pol activity on the 24/44 substrate was significantly and consistently lower than on the other primertemplates. The cause of this phenomenon is presently unclear, but is most likely related to the immediate sequence context and is apparently exacerbated by decreasing the dNTP concentration from 100 M to 10 M (compare Fig. 3A, lane 2, and Fig.  4A, lane 6). Poor extension of this primer-template by E. coli pol IV was also observed by Wagner et al. (6).
Given the detectable levels of nucleotide misincorporation observed in Fig. 4A, we tested GST/pol for 3Ј 3 5Ј proofreading exonuclease activity. Using a substrate in which the 3Ј  1, 7, 13, and 19). B, GST/pol, Klenow (exo Ϫ ), and Klenow (exo ϩ ) were assayed on the 3Ј mispaired 31/44 primer-template in the absence (lanes 2-4)  nucleotide of the primer was not base paired with the template, no shortening of the primer by GST/pol or Klenow (exo Ϫ ) enzyme was observed in the absence of dNTPs (Fig. 4B, lanes 2  and 3). In contrast, Klenow (exo ϩ ) enzyme readily cleaved the primer (Fig. 4B, lane 4). In the presence of dNTPs, the primer could be efficiently extended by Klenow (exo ϩ ) enzyme only following cleavage of the mispaired base (Fig. 4B, lane 8). Limited extension by GST/pol was also observed from the 3Ј mispaired primer (Fig. 4B, lane 6). The low level of primer extended by Klenow (exo Ϫ ) enzyme yielded a product 45 nucleotides in length due to incorporation of an additional dATP in a template-independent fashion (22). This nucleotide is normally removed by the 3Ј 3 5Ј exonuclease activity of Klenow (exo ϩ ) enzyme. The detectable levels of misincorporation together with the observed lack of a proofreading exonuclease activity suggest that pol is endowed with a low level of fidelity during synthesis of DNA.
To complete our preliminary characterization of GST/pol we tested the sensitivity of the enzyme to aphidicolin and dideoxynucleoside triphosphates (ddNTPs). Aphidicolin is an inhibitor of eukaryotic DNA polymerases ␣, ␦, and ⑀ while pol␤ and -␥ are sensitive to ddTTP (23). In the presense of 25 M dNTPs GST/pol activity was not inhibited by either aphidicolin (50 ng/l) or up to an 8-fold molar excess of ddNTPs (data not shown). The lack of sensitivity of pol to aphidicolin and ddNTPs is similar to that observed for human pol (24).
pol Cannot Bypass a Cisplatin Adduct but Can Bypass an AAF Lesion in a Potentially Error-prone Manner-A number of the DNA polymerases in the UmuC/DinB superfamily have been implicated in DNA damage-induced mutagenesis and translesion synthesis (2). We asked whether GST/pol is able to bypass cisplatin (Fig. 5) or an AAF adduct (Fig. 6), lesions that have previously been shown to strongly block other DNA polymerases (reviewed in Ref. 25). GST/pol is unable to bypass d[GpG-N 7 (1)-N 7 (2)] cisplatin intrastrand cross-links, even when a 2-fold molar excess of enzyme is added to the primertemplate (Fig. 5, lanes 6 -9). Klenow (exo Ϫ ) enzyme is also unable to bypass this lesion (Fig. 5, lane 10), unless high concentrations of enzyme are added (data not shown). Both GST/pol and Klenow (exo Ϫ ) enzyme arrest one nucleotide prior to the lesion, suggesting that neither enzyme is able to efficiently insert nucleotides across from the damaged bases.
In contrast to the results observed with cisplatin, GST/pol does appear to have an intrinsic ability to bypass N-(deoxyguanosin-8-yl)-N-2-acetylaminofluorene (G-AAF) adducts (Fig. 6A, lanes 6 -9). The bypass appears to be relatively inefficient under the conditions used in these experiments, since relatively high enzyme concentrations are necessary to achieve bypass. Furthermore, even under conditions where bypass occurs, a large percentage of the primer extension products appear to terminate at the site of the AAF lesion. Since it is clear that the GST/pol enzyme is able to insert nucleotides across from the AAF lesion, we were interested in determining which nucleotides could be incorporated. As shown in Fig. 6B, GST/ pol primarily inserts dCTP across from an undamaged G residue (lanes 2-5). In contrast, GST/pol appears equally able to incorporate either dTTP or dCTP across from the G-AAF lesion (Fig. 6B, lanes 8 -11). A darker exposure of this experiment shows that dATP can also be incorporated across from the G-AAF lesion at much lower frequency (data not shown). These results suggest that pol may play a role in bypass of AAF lesions in vivo and that such bypass could potentially be mutagenic. DISCUSSION In this study we report that the product of the human DINB1 gene is a DNA polymerase, pol. While this work was in pro-gress, a GST/hDinB1 fusion protein purified from the yeast Saccharomyces cerevisiae was reported to have DNA polymerase activity (26). These authors designated human DinB1 protein as DNA polymerase (pol). In view of the fact that this name has been assigned to the product of another gene (27), we suggest that the DinB1 gene product be referred to as pol, a designation approved by the Human Genome Organization nomenclature committee.
The loss of DNA polymerase activity of pol associated with mutation of two highly conserved residues known to be required for the activity of both E. coli pol IV and yeast pol (6,28) eliminates the possibility that the activity of our purified GST/pol fraction is due to the presence of a contaminant or interacting protein. We can also conclude that fusion of pol to GST does not abolish its DNA polymerase activity. However, we cannot exclude the possibility that the GST domain alters the DNA polymerase activity of pol in a more subtle way. Pol represents the tenth eukaryotic polymerase reported, and the seventh member of the UmuC/DinB superfamily of proteins endowed with such activity. It is therefore likely that most if not all members of this superfamily are DNA polymerases or nucleotidyl transferases.
To demonstrate DNA polymerase activity we used a simple primer extension assay. Interestingly the largest extension product produced by pol is a single nucleotide shorter than the template. This phenomenon was consistently observed on templates used in the present studies regardless of the primer (e.g. Figs. 1B and 4B). Similar observations have been made using FIG. 5. GST/pol is unable to bypass a cisplatin adduct. Replication of 5 nM wild-type (lanes 2-4) or cisplatin-containing (lanes 7-9) primer-templates was tested using the indicated concentrations of GST/ pol. Reactions were performed for 10 min at 37°C. 1 nM Klenow (exo Ϫ ) enzyme was used as a control (lanes 5 and 10). The position of the cisplatin adduct is indicated at the right; the first G of the adduct is located at the 30 nucleotide (nt) position. The unextended running start primer (lanes 1 and 6) is 20 nucleotides and the full-length extension product is 53 nucleotides in length.
other templates of varying length and sequence composition 2 and were observed independently by Johnson et al. (26). The incomplete extension products produced by pol could be extended by the Klenow fragment of E. coli DNA polymerase I, demonstrating that failure to replicate precisely to the end of a linear DNA template is likely an intrinsic property of pol rather than the result of template slippage. 2 Conceivably the enzyme requires interactions with sequences downstream to those located at the active site in order to form a stable protein-DNA complex. The ability of pol to fill in short single-stranded gaps is currently being investigated.
In most of the GST/pol polymerase assays the reaction products consisted of a ladder of primer extension products, ranging from the addition of a single nucleotide to 1 nucleotide short of full-length. A similar size range of products was observed at high GST/pol dilutions where the primer-template was present in vast excess (29). These results suggest that the processivity of pol is variably low to moderate, but not entirely distributive as is the case for E. coli DNA pol IV (6, 7). Similarly, the apparent high rate of nucleotide misincorporation (Fig. 4A) and lack of 3Ј 3 5Ј proofreading exonuclease activity (Fig. 4B) suggest that pol possesses low fidelity. These conclusions are supported by the results of independent experiments that quantitatively measured the fidelity and processivity of pol in vitro (29).
As noted above, E. coli DNA pol IV is a specialized DNA polymerase that is regulated by the SOS system, suggesting a role in DNA replication and spontaneous mutagenesis in response to certain cellular stress conditions. The apparent low fidelity and moderate processivity of human DNA pol are consistent with a similar role in human cells. The demonstration that the human DINB1 gene encodes pol supports a model in which the role of this polymerase is to facilitate the replication of abnormal templates which inhibit the activity of replicative DNA polymerases pol␣, pol␦, and/or pol⑀.
Abnormal templates might contain particular types of DNA damage or possess aberrant structures. Efficient translesion synthesis by DNA pol IV on templates containing an abasic site, thymine dimer, or 6-4 photoproduct has not been observed (7). However, the bacterial protein can extend misaligned primer-templates resulting in single nucleotide deletions (6). Similarly, a human GST/pol fusion protein is unable to bypass abasic sites, cis-syn thymine-thymine dimers or 6-4 photoproducts in vitro (26).
In this study we have shown that human pol is also unable to bypass a cisplatin lesion. The enzyme is, however, able to inefficiently bypass an AAF-adduct and incorporates primarily either dTTP or dCTP across from the lesion. This result suggests that pol has the potential to bypass AAF by an errorprone mechanism and therefore that pol might play a role in targeted mutagenesis. However, the characterization of translesion synthesis using simple primer extension assays in vitro requires caution. A number of experimental parameters may influence the outcome, including reaction conditions, enzyme concentration, nucleotide concentration, and template sequence context (25). The physiological relevance of the bypass of AAF lesions by pol is uncertain since it is observed only at high enzyme concentrations, and even under these conditions a significant fraction of the enzyme is arrested at the site of the lesion. However, it remains a formal possibility that pol can be utilized in human cells for translesion synthesis of specific lesions (such as AAF or other types of DNA damage which have not yet been tested), as may be the case for pol (30).
In human cells pol may function in additional or alternative modes other than the replication of abnormal templates. For example, the enzyme may be required to replicate specific normal template regions in an abnormal (error-prone) manner, such as appears to be required during somatic hypermutation of immunoglobulin genes (31). More precise definition of the biological function(s) of this enzyme will require further detailed biochemical and genetic studies, including the phenotypic characterization of mammalian cells defective in this enzyme activity. are particularly grateful to Tom Kunkel and Haruo Ohmori for helpful discussions and communication of results prior to publication.