Mechanisms of dCMP transferase reactions catalyzed by mouse Rev1 protein.

The Rev1 protein, a member of a large family of translesion DNA polymerases, catalyzes a dCMP transfer reaction. Recombinant mouse Rev1 protein was found to insert a dCMP residue opposite guanine, adenine, thymine, cytosine, uracil, and an apurinic/apyrimidinic site and to have weak ability for transfer to a mismatched terminus. The mismatch-extension ability was strongly enhanced by a guanine residue on the template near the mismatched terminus; this was not the case with an apurinic/apyrimidinic site and the other template nucleotides. Kinetic analysis of the dCMP transferase reaction provided evidence for high affinity for dCTP with template G but not the other templates, whereas the template nucleotide did not much affect the V(max) value. Furthermore, it could be established that the mouse Rev1 protein inserts dGMP and dTMP residues opposite template guanine at a V(max) similar to that for dCMP.

The Rev1 protein, a member of a large family of translesion DNA polymerases, catalyzes a dCMP transfer reaction. Recombinant mouse Rev1 protein was found to insert a dCMP residue opposite guanine, adenine, thymine, cytosine, uracil, and an apurinic/apyrimidinic site and to have weak ability for transfer to a mismatched terminus. The mismatch-extension ability was strongly enhanced by a guanine residue on the template near the mismatched terminus; this was not the case with an apurinic/apyrimidinic site and the other template nucleotides. Kinetic analysis of the dCMP transferase reaction provided evidence for high affinity for dCTP with template G but not the other templates, whereas the template nucleotide did not much affect the V max value. Furthermore, it could be established that the mouse Rev1 protein inserts dGMP and dTMP residues opposite template guanine at a V max similar to that for dCMP.
In yeast Saccharomyces cerevisiae, the REV1 gene is required for damage-induced and spontaneous mutagenesis (1)(2)(3)(4)(5)(6)(7). A defect in the REV1 gene has in fact been found to decrease the translesion replication of apurinic/apyrimidinic (AP) 1 sites, T-T (6 -4) UV photoproducts, and N-2-acetylaminofluorenemodified guanine (8,9). The encoded protein, containing a BRCA1 C terminus (BRCT) domain at its N terminus, possesses deoxycytidyl transferase activity (10 -12) inserting dCMP residues opposite templates G, A, U, and AP sites (12). The activity of Rev1 protein could be important for the bypass of AP sites in yeast (9) because cytidine is preferentially inserted opposite these lesions in vivo (9,13). However, a number of observations have suggested that the Rev1 protein may possesses a second function. First, when the REV1 gene is required for the bypass of a T-T (6 -4) UV photoproduct, dCMP incorporation occurs only very rarely in vivo (9). Second, translesion DNA synthesis and mutagenesis are greatly reduced in a rev1 mutant, rev1-1, with a BRCT domain alteration that does not affect the deoxycytidyl transferase activity in vitro (9,11). Third, during bypass of N-2-acetylaminofluorene-modified guanine, the REV1 gene is needed only for non-slipped translesion DNA synthesis, suggesting that the uncharacterized Rev1 activity is UmuDC-like in nature (8). Fourth, methyl methanesulfonate-induced mutagenesis was shown to be normal in a site-directed mutant lacking deoxycytidyl transferase activity (14). Fifth, ϩ1 frameshift mutations accompanying base substitutions are dependent on the REV1 gene (15).
Recently, cloning of a human homologue of the REV1 gene (16,17) revealed good conservation from yeast to humans. The human REV1 gene encodes a deoxycytidyl transferase, similar to the Rev1 protein of the yeast, S. cerevisiae (17,18), with activity localized to the central domain that is conserved in the UmuC superfamily (18). Mutations in conserved residues but not the BRCT domain completely abolish the transferase activity (18).
Proteins in the UmuC superfamily, except for the Rev1 protein, are novel DNA polymerases, capable of replicating damaged DNA (19). Only for the Rev1 protein has no polymerase activity been detected so far. It has been reported that its transferase activity is limited to the insertion of a dCMP residue (12,17). Although the enzyme is capable of incorporating a dCMP residue not only opposite G but also A, U, and AP sites (12,17), it is not clear how the Rev1 protein plays a role in mutagenesis.
In the present study, we cloned and characterized the Rev1 gene of the mouse, an animal commonly used for models of human disease. We found that the mouse Rev1 protein transfers a dCMP residue not only opposite template G but also A, T, C, U, and AP sites and extends a mismatched terminus by the addition of a dCMP residue. We further showed this mismatchextension ability to be strongly enhanced by the presence of a guanine residue (but not an AP site) on the template near the mismatched terminus. Kinetic analysis of the dCMP transferase reaction provided evidence for the high affinity of dCTP with template G. Furthermore, the mouse Rev1 protein could be shown to insert dGMP and dTMP residues opposite template guanine and AP sites. The same activity was also detected with recombinant human REV1 protein but not an inactive mutant protein.

EXPERIMENTAL PROCEDURES
Animals-Mice of the C57BL/6N and C3H/He strains were purchased from Charles River Laboratories Inc., Atsugi, Japan. All experiments followed the guidelines of the Animal Experimental Facility Committee of Hiroshima University.
Cloning of Mouse Rev1 cDNA and Construction of an Expression Plasmid-Poly(A) ϩ mRNA was isolated from the liver of an inbred mouse, strain C3H/He, using a poly(A) tract mRNA isolation system (Promega). Double-stranded cDNA was synthesized using a cDNA synthesis kit (Takara), and PCR primers corresponding to the human REV1 gene were used to amplify fragments of mouse Rev1 cDNA. The amplified fragments were cloned and sequenced. From the sequence information, several primer sets were designed and tested. Consequently, mouse Rev1 cDNA was successfully amplified as three overlapping cDNA fragments using Pyrobest TM DNA polymerase (Takara). A promoter proximal fragment (fragment I) was amplified with primers (5Ј-GAAGCTCCCATATGAGGCGA-3Ј and 5Ј-CTGAAGTTGAGCTGTT-TGGC-3Ј) at 55°C for annealing with 30 cycles. A fragment of the central region (fragment II) was amplified with primers (5Ј-AATCCT-GTGTGCAAACCTGAG-3Ј and 5Ј-TTTCGTCTCTGCAAGGATGTC-3Ј),  and a promoter distal fragment (fragment III) was amplified with  primers (5Ј-TGCGATGAAGCACTGATTGAC-3Ј and 5Ј-TCAGGT-CACTTTCAGTGTGCT-3Ј) under the same conditions. These fragments were cloned into a pCR 2.1-TOPO vector (Invitrogen). Several independently isolated clones were sequenced, and mutant clones with different sequences from the others were rejected. A plasmid containing wild-type fragment I was digested with NdeI and BamHI, a plasmid containing fragment II was digested with BamHI and AatII, and a plasmid containing fragment III was digested with AatII and KpnI and assembled into the pCR 2.1-TOPO vector. To make a bacterial expression plasmid, the human REV1 fragment of pBADREV1 (18) was replaced with the mouse Rev1 fragment. The resulting plasmid, pBAD-mRev1, encodes full-length mouse Rev1 protein tagged with hexahistidine residues at its N terminus (h6-mRev1). The tagged sequence is identical with that of pET15b (Novagen). Gene expression was induced by arabinose. The sequence data for the mouse Rev1 cDNA have been submitted to the DDBJ/EMBL/GenBank TM data bases under accession number AB057418.
Northern Blot Analysis of Mouse Rev1 Gene Expression in Various Mouse Tissues-A Northern blot membrane with 2 g of poly(A) ϩ mRNAs from tissues of C57BL/6N ϫ C3H/He mice was hybridized with a 32 P-labeled probe generated by PCR using primers 5Ј-TCCCAGATT-GACCAGTCTGTT-3Ј and 5Ј-TCAGGTCACTTTCAGTGTGCT-3Ј in Ex-pressHyb TM Hybridization Solution (CLONTECH) at 65°C and washed with 0.1 ϫ SSC and 0.1% SDS at the same temperature. Signals were visualized by autoradiography at Ϫ80°C.
Purification of the Human and Mouse Rev1 Protein-BL21(DE3) harboring pBADmRev1 was grown in 500 ml of SB medium (20) supplemented with ampicillin (100 mg/ml) at 15°C with aeration until the culture reached an A 600 value of 0.6. L(ϩ)-arabinose was added to 1%, and the incubation was continued for 10 h. The cells were harvested, and cell lysate was prepared as described previously (18). Subsequent column chromatography was carried out at 4°C using a SMART system (Amersham Biosciences, Inc.). Two-ml aliquots of lysate were applied at 0.1 ml/min to a 1-ml HiTrap chelating column (Amersham Biosciences, Inc.), which had been flushed with 2 ml of 0.1 M NiSO 4 and then equilibrated with buffer A (50 mM HEPES-NaOH, pH 7.5, 1 M NaCl, 10% glycerol, 10 mM ␤-mercaptoethanol) containing 10 mM imidazole, 5 mM ATP, and 10 mM MgCl 2 . The column was washed with 10 ml of equilibration buffer and then with 12 ml of buffer A containing 100 mM imidazole, 5 mM ATP, and 10 mM MgCl 2 . The h6-mRev1 was eluted with buffer A containing 300 mM imidazole, 5 mM ATP, and 10 mM MgCl 2 . Five hundred l of the peak fraction of the h6-mRev1 was concentrated using an Ultrafree-0.5 centrifugal filter device, Biomax-10 (Millipore), and then applied at 0.01 ml/min to a Superdex 200 PC 3.2/30 column (Amersham Biosciences, Inc.), equilibrated with buffer A, and 40-l fractions were collected. Protein concentrations were determined by the Bio-Rad protein assay using bovine serum albumin (BSA) (Bio-Rad) as a standard. The recombinant human REV1 protein and its mutant, D569A/E570A, were purified by nickel-chelating column and gel filtration chromatography as described (18).
Sucrose Density Gradient Sedimentation-Sucrose density gradient sedimentation was performed as described previously (21). The purified h6-mRev1 protein (4 g) was sedimented through 2 ml of 10 -40% sucrose gradient in buffer A by centrifugation at 55,000 rpm for 15 h in a TLS 55 rotor (Beckman) at 4°C. Fractions (70 l) were collected from the bottom of the tube and analyzed by SDS-polyacrylamide gel electrophoresis. Gel bands were stained with Colloidal Blue (NOVEX) and quantified using NIH image 1.60 software. The sedimentation coefficient was determined relative to those of standard proteins sedimented in parallel gradients.

Primary Structure of Mouse Rev1 Protein-The cloned mouse
Rev1 cDNA encodes a putative protein of 1249 amino acid residues with a calculated molecular mass of 137 kDa. The sequence alignment of the human and mouse Rev1 proteins is shown in Fig. 1. Comparison of the amino acid sequences of the two proteins revealed an overall amino acid identity of 84% and similarity of 90% with all of the motifs found in the human REV1 protein conserved in the mouse counterpart (Fig. 1). The BRCT domain, motif I, and motif VIII are specific to the Rev1 family. Motifs II-VII are conserved in polymerases of the UmuC superfamily. We (18) previously showed the minimum region required for deoxycytidyl transferase activity of the human REV1 protein ( Fig. 1, boxed region). This region was highly conserved with 88% identity and 94% similarity.
Expression of the Rev1 Gene in Mouse Tissues-Expression of the mouse Rev1 gene in various tissues was examined by Northern blot analysis (Fig. 2), and the mouse Rev1 mRNA was detected in all tissues examined. Expression of the Rev1 gene was relatively high in the heart, skeletal muscle, and testis.
Purification of Recombinant Mouse Rev1 Protein and Physicochemical Properties-To purify the mouse Rev1 protein, we expressed a recombinant protein tagged with hexa-histidine at its N terminus in E. coli cells. The tagged Rev1 protein (h6-mRev1) was purified by affinity chromatography on a nickelchelating column, and the fraction containing h6-mRev1 was applied to a gel filtration column. The h6-mRev1 protein eluted with an apparent molecular mass of 330 kDa with a Stokes' radius of 58 Å (Table I). As shown in Fig. 3, analysis by SDS-PAGE revealed a full-length h6-mRev1 protein of 139 kDa, and smaller forms were also detected specifically when the h6-mRev1 protein was induced in E. coli cells, indicating that these bands are degradation products (data not shown). The properties were found to be identical with those of the human REV1 protein (18). In this preparation, neither DNA polymerase activity nor deoxyribonuclease activity was detected ( Fig. 4A and data not shown). We used this fraction for further physicochemical and biochemical characterization.
To determine the molecular mass of the purified h6-mRev1 protein in solution, it was analyzed by sucrose gradient sedimentation. The determined sedimentation coefficient was 5.7 S. Employing the method described by Siegel and Monty (22), we calculated the molecular mass at 138 kDa (Table I). These results suggested that the h6-mRev1 protein was asymmetrically shaped, existing as a monomer in solution.
Ability of the Mouse Rev1 Protein to Transfer dGMP and dTMP Residues-Using six different primer-templates, we examined the substrate specificity of the transferase activity of the h6-mRev1 protein in the presence of 100 M dGTP, dATP, dTTP, or dCTP (Fig. 4). In this experiment, the respective primer-templates differed only at the template nucleotide immediately downstream from the annealed primer. When a G template was incubated with the h6-mRev1 protein and each of the dNTPs, we surprisingly detected a one-base-extended product in the presence of not only dCTP but also dGTP and dTTP (Fig. 4A, panel a). We also found an ability to insert dGMP and dTMP residues opposite the template AP site (Fig. 4A, panel f).
Although the efficiency was very low, significant activity was confirmed by reactions with higher concentrations of the h6-mRev1 protein (data not shown). We could not detect dGMP insertion opposite template C or dTMP insertion opposite template A (Fig. 4A, panels b and d), and the results clearly indicated that activity from contaminating bacterial DNA polymerases was less than the detectable level. The mobility of the reaction products and defined oligonucleotide markers was compared (Fig. 4B). Those with dGTP, dTTP, and dCTP migrated exactly like the respective markers (Fig. 4B). These results indicate that the h6-mRev1 protein has a potential to transfer dGMP and dTMP residues, albeit with only a fifth of the activity with dCMP residues (Fig. 4C). In the presence of all four dNTPs, the dCMP transfer reaction predominated over the dGMP and dTMP transfer reactions (Fig. 4A, panel a, lane 6).
Only the ability to transfer dCMP has been reported for the human REV1 protein (17,18). However, when dGMP and dTMP transferase activities of the human REV1 protein were examined (Fig. 4D), it incorporated not only dCMP but also dGTP and dTTP residues opposite template G. The activity was completely eliminated with an inactive mutant protein, D569A/ E570A (18) (Fig. 4D).
Deoxycytidyl Transferase Activity and Ability to Extend the Mismatched Primer Terminus-When all of the templates were tested, the h6-mRev1 protein inserted a dCMP residue opposite not only the template AP site but also all bases examined (Fig.  4A, panels b-f, lane 5). The enzyme activity for insertion reactions opposite the template AP site was slightly higher than that opposite templates G, A, and U (1.2-1.5-fold) and six times higher than that opposite templates T and C (Fig. 4C). When G,   A, T, and U templates were incubated with h6-mRev1 and dCTP, a faint two base longer band was detected (Fig. 4A,  panels a-c and e), and it was concluded that the h6-mRev1 protein might possess the ability to add a dCMP residue to the mismatched 3Ј terminus resulting from the first insertion of a dCMP residue. We investigated this preliminary observation further by performing reactions using another set of primertemplates having mismatched primer-template termini (Fig. 5,  A and B) differing only in the attachment of a cytidine residue at the 3Ј terminus from the set of primer-templates described in Fig. 4. As illustrated in Fig. 5, A and B, the transferase extended C⅐A, C⅐T, and C⅐U (primer⅐template) mismatched termini by only one nucleotide (Fig. 5A, panels b, c, and e) but failed to extend C⅐C and C⅐AP (primer⅐template) termini (Fig.  5A, panels d and f). The ability did not arise from a potential of the enzyme to extend the 3Ј terminus of the single-stranded DNA (Fig. 5C). In the sequence context, it should be noted that the enzyme activity for extension of the mismatched primer termini was 10 times lower than that of the matched C⅐G primer (primer⅐template) terminus ( Fig. 5B and Table II).

A Template Guanine Residue near the Mismatched Base Pair Assists Primer Extension from the Mismatched 3Ј Terminus-
The mismatch-extending ability was examined using all 16 possible combinations of primer-template termini (Fig. 6). As shown in Fig. 6, extensions of all mismatched termini except for C⅐C were detected. The quantitative results are summarized in Table II. Interestingly, the 3Ј mismatched termini with a G template, G⅐G, A⅐G, and T⅐G (primer⅐template) tended to be extended at high efficiency ( Fig. 6 and Table II). Notably, the specific activity of extension of the T⅐G (primer⅐template) mismatch was close to that of the matched termini (36% of T⅐A and 90% of C⅐G (primer⅐template) matched termini). This property of the T⅐G (primer⅐template) mismatch is not due to a property of the base pairing of T and G because the enzyme activity for extension of the G⅐T (primer⅐template) mismatch was very low (Fig. 6 and Table II). While a fraction of the mismatched primer terminus could have been misaligned and the dCMP transferred from a dCTP pairing with template G, as shown in Fig.  8A, if this was the case, the template nucleotide immediately downstream from the new primer-terminus could have become A (Figs. 6 and 8A). Because template A is a better substrate than template C (Fig. 4), h6-mRev1 could extend one more nucleotide by the insertion of dCMP opposite the template A (Fig. 8A). Expectedly, two base longer bands were detected when the G template was used (Fig. 6). This result supported our explanation for the enhancement (see Fig. 8A).
We examined the potential sequence context effect on transferase activity using other sets of primer-templates (Fig. 7). With these, the template nucleotide immediately downstream from the annealed primer was cytidine (Fig. 7A). The enzyme activity for the insertion reaction of dCMP opposite template C was strongly affected by the surrounding sequence (Fig. 7, B  and C). The specific activity of insertion to set a substrates was 5-8 times less than that to set b (Fig. 7, D and E). With both sets of primer-templates shown in Fig. 7A, the enzyme activity of insertion of dCMP opposite template C was affected by a template nucleotide, the position of which is represented by X in Fig. 7A. The specific activity of insertion of dCMP was decreased in the order of template nucleotides of A Ͼ T Ͼ G Ͼ C at position X (Fig. 7, D and E). The C⅐C mismatch is the most inefficient substrate for the transferase reaction and was not extended by the Rev1 protein (Figs. 5A and Fig. 6). However, when a guanine residue was located on the template 5Ј to the mismatched base pair, the C⅐C mismatch terminus was efficiently extended (Fig. 7, B and C, panel a) to an ϳ15% extent (Fig. 7C, panel a). This mismatch extension was not detected when other nucleotide residues were located in this position (Fig. 7, B and C, panels b-d).
Steady-state Kinetics of the Transferase Reactions-We have determined kinetic parameters by steady-state gel kinetic assays (Table III). Because the reactions with mismatched templates seem to be complicated and a mixture of several pathways (see "Discussion"), we focused on the transferase reactions with matched templates. The assays were all carried out with a 5-min incubation because the time course of the reactions was linear until 10 min (data not shown). The V max for the dCMP insertion opposite template G, A, and U was 1.0 min Ϫ1 , and it was 0.83 and 0.85 min Ϫ1 opposite template T and C, respectively. With the template AP site, the V max was a little faster (1.3 min Ϫ1 ). These results indicate that the template nucleotide did not appreciably affect the velocity of the reaction. However, we found that the template nucleotide strongly affected the K m value for dCTP, revealing a specific nature for template G. The K m value with the template AP site (12 M) was nine times higher than that with template G (1.4 M) but still lower than that with the other templates. Because the velocity of the enzyme reaction is near to the V max at higher substrate concentration, values for the velocity at 100 M of dCTP concentration were calculated using equations obtained from the Lineweaver-Burk plot, giving 1.0 min Ϫ1 and 1.2 min Ϫ1 with template G and the AP site, respectively (Table III). These values agree with the results of Fig. 4, for which the experiments had been performed at 100 M of dNTP. Therefore, the higher activity of the Rev1 protein with the template AP site shown in Fig. 4 is due to the V max value under this condition.
The kinetic parameters for dGMP and dTMP insertion opposite template G were also determined (Table III). Affinity for dGTP and dTTP was very low, although the value of V max for dGTP insertion was found to be identical to that of dCTP insertion, whereas that for dTTP insertion was only half.

DISCUSSION
The Rev1 protein is unique in that while belonging to a large family of translesion DNA polymerases, its reported activity is restricted to the transfer of dCMP to primer termini. In this work, cloning of the mouse Rev1 gene and characterization of the transferase reaction revealed particular enzymatic properties.
We showed that the mouse Rev1 gene is expressed ubiquitously, as reported for the human REV1 gene (17,18). Why expression in the mouse heart, skeletal muscle, and thymus was relatively high compared with that in the corresponding tissues of humans (18) remains to be clarified. a 5Ј-32 P-labeled P13G, P13A, P13T, and P13C primers were each annealed with the templates: 30G, 30A, 30T, and 30C. The base pairs in the primer terminus next to the primer/template are indicated in parentheses.
b An appropriate amount of h6-mRev1 was incubated with the indicated primer-template and dCTP at 30°C for 30 min in 25 l of reaction solution. The reaction products were resolved in 20% polyacrylamide gels containing 8 M urea, and the band intensities of substrates and products were determined using a Bio-Imaging Analyzer BAS2000 to give amounts in picomoles.
c Specific activity of the T⅐G (primer⅐template) mismatch-extension reaction was compared with that for the C⅐G primer terminus. We (18) previously suggested, based on the results of gel filtration chromatography, that the human REV1 protein might be a dimer in solution. However, the present results with sucrose gradient sedimentation strongly suggest that the h6-mRev1 protein is an asymmetrically shaped molecule with a molecular mass of 138 kDa calculated from the Stokes' radius and a sedimentation coefficient very close to the value of 139 kDa calculated from the amino acid sequence. These data imply the monomeric status of the human REV1 protein in solution.
The present experiments revealed that the mouse Rev1 protein inserts dGMP and dTMP residues opposite template G and AP sites, whereas no incorporation was detected opposite templates C and A. The V max for dGMP insertion is identical to that for dCMP insertion, and that for dTMP insertion is only half. Therefore, the low efficiency of the reactions is due to the low  The incorporating dCMP residue is paired with template G, and the new 3Ј terminus of the resulting product may be extended by a second dCMP insertion opposite template (A). The Rev1 protein does not extend 3Ј terminus mispairing with a template AP site (B). When a dCMP residue is incorporated opposite template C and a guanine residue is located on the template immediately downstream from the mismatched 3Ј terminus, a second dCMP residue is efficiently incorporated, but other nucleotides and an AP site do not assist such mismatch extension (C and D). This pathway might not assist mismatch extension and is not prevalent (E). Dashes represent AP sites. Kinetic assays were performed for 5 min in 25-l reaction solutions using 71 fmol (10 ng) or 140 fmol (20 ng) of h6-mRev1 and 2.5 pmol of the primer-templates shown in Fig. 4A. dCTP, dGTP, and dTTP concentrations ranged from 1 to 2500 M. The K m and V max values were determined using equations obtained from a Lineweaver-Burk plot. Data from two to four independent experiments were plotted together, and the correlation efficients (R 2 ) of the straight lines were 0.9 for dCTP and dGTP and 0.8 for dTTP.
b Velocity at 100 M substrate was calculated using the equations obtained from the Lineweaver-Burk plot. affinity of dGTP and dTTP but not due to the V max . These results suggest that the template G is not a true substrate for the transferase reaction and that activity of the Rev1 protein might be required for bypassing some DNA lesions that could increase the affinity for dGTP and dTTP. With the yeast Rev1 protein but not the human protein, a weak ability to incorporate a dGMP opposite template G has been detected (12). The fact that incorporation on dGMP and dTMP was detected for the mouse Rev1 protein might result from differences in the sequence context of the primer-templates used. Examination of the human REV1 protein (18) using our substrates was therefore conducted. We found that the properties of the human and mouse enzymes were essentially identical. It is possible that the nucleotide sequence of our primer-templates used in the primer extension assay may be serendipitously suitable for detecting such activity. Importantly, the activity was completely eliminated with the D569A/E570A, providing evidence that transferase activity is intrinsic to mammalian Rev1 proteins.
Our systematic analysis of the transferase activity of the mouse Rev1 protein using various primer-templates revealed a unique property. We showed that the mouse Rev1 protein inserts a dCMP opposite templates G, A, T, C, U, and AP sites. The yeast Rev1 protein has only weak ability to insert a dCMP opposite template C (12), and no such activity has been reported for the human REV1 protein, but we showed the reaction efficiency to be strongly affected by the surrounding nucleotide sequence so that a direct comparison of the results is difficult. We found that the Rev1 protein has a weak ability to extend the mismatched 3Ј terminus and that this was strongly enhanced by the presence of a guanine residue on the template near the primer terminus but not an AP site or the other nucleotides. Kinetic analysis provided evidence for a high affinity of dCTP with template G and low affinity with the other templates, whereas the template nucleotide did not appreciably affect the V max value. These properties might explain the specific nature of template G enhancement of mismatch extension. We propose a model to account for this enhancement (Fig. 8). In Fig. 8A, the mispaired 3Ј terminus with a guanine residue on the template could be misaligned and ligated to a dCMP paired with the guanine residue (Fig. 8A). Even though the Rev1 protein efficiently inserts dCMP opposite a template AP site, an AP site does not assist mismatch extension (Fig. 8B). When a guanine residue is located on the template 5Ј close to the mismatched base pair, the mismatched 3Ј terminus may be efficiently ligated to the dCMP residue paired with the guanine residue on the template (Fig. 8C). Again, a template AP site does not assist under this condition (Fig. 8D). Alternatively, the first insertion of dCMP might be the result of misalignment with template G as shown in Fig. 8E. However, we suspect that this process is not prevalent because a guanine base at this position did not enhance the dCMP insertion reaction (Fig. 7). Currently, we cannot distinguish the two reaction pathways (Fig. 8, C and E) and are not able to quantitate their respective efficiencies. Recently, Harfe and Jinks-Robertson (15) found complex mutations with frameshifts accompanied by base substitutions accumulating in a yeast strain defective for nucleo-tide excision repair. These events were dependent on the REV1 and REV3 genes and triggered by a misincorporation opposite template G, similar to the pathway shown in Fig. 8. The properties described in this study for the Rev1 protein might imply a possible direct role of this protein in inducing frameshift mutations.
The kinetic analysis revealed that the velocity of the transferase reaction was restricted by the concentration of dCTP but not template nucleotides. It is known that a second ribonucleotide reductase is induced by DNA damage in yeast and humans (23,24). It is very likely that the concentration of dNTPs is increased in the nucleus in response to DNA damage (25,26) and could affect the substrate specificity of the Rev1 protein.