Mutagenic and non-mutagenic bypass of DNA lesions by Drosophila DNA polymerases dpol and dpol *

cDNA sequences were identified and isolated that encode Drosophila homologues of human Rad30A and Rad30B called drad30A and drad30B. Here we show that the C-terminal-truncated forms of the drad30A and drad30B gene products, designated dpoletaDeltaC and dpoliotaDeltaC, respectively, exhibit DNA polymerase activity. dpoletaDeltaC and dpoliotaDeltaC efficiently bypass a cis-syn-cyclobutane thymine-thymine (TT) dimer in a mostly error-free manner. dpoletaDeltaC shows limited ability to bypass a 6-4-photoproduct ((6-4)PP) at thymine-thymine (TT-(6-4)PP) or at thymine-cytosine (TC-(6-4)PP) in an error-prone manner. dpoliotaDeltaC scarcely bypasses these lesions. Thus, the fidelity of translesion synthesis depends on the identity of the lesion and on the polymerase. The human XPV gene product, hpoleta, bypasses cis-syn-cyclobutane thymine-thymine dimer efficiently in a mostly error-free manner but does not bypass TT-(6-4)PP, whereas Escherichia coli DNA polymerase V (UmuD'(2)C complex) bypasses both lesions, especially TT-(6-4)PP, in an error-prone manner (Tang, M., Pham, P., Shen, X., Taylor, J. S., O'Donnell, M., Woodgate, R., and Goodman, M. F. (2000) Nature 404, 1014-1018). Both dpoletaDeltaC and DNA polymerase V preferentially incorporate GA opposite TT-(6-4)PP. The chemical structure of the lesions and the similarity in the nucleotides incorporated suggest that structural information in the altered bases contribute to nucleotide selection during incorporation opposite these lesions by these polymerases.


INTRODUCTION
DNA is frequently damaged by environmental and endogenous genotoxic agents.
Although various mechanisms exist to ensure that the majority of DNA damage is recognized and repaired and the integrity of the DNA is faithfully restored, some DNA lesions escape repair and persist in the cell. Unrepaired DNA lesions can block the progress of the replication machinery. To help resolve this problem, cells have specialized polymerases that carry out translesion DNA synthesis (TLS), which is a mechanism that permits nucleotides to be incorporated opposite lesions. After TLS bypasses the DNA damage, replication can continue with the normal replication machinery downstream of the site of the damage.
Recently, a new family of DNA polymerases was identified called the UmuC/DinB/Rev1/Rad30 superfamily (1 -3). Several of these polymerases participate in TLS (4)(5)(6)(7)(8)(9)(10)(11) and members of this protein family exhibit lower fidelity and processivity than replicative DNA polymerases (10, 12 -15). The lower fidelity enables these polymerases to carry out TLS and the lower processivity facilitates dissociation of these enzymes after bypass of a lesion; this is important because it allows the normal replication machinery, with high fidelity and processivity, to preferentially carry out DNA synthesis distal to the lesion.
However, TLS is mutagenic when the base inserted opposite the DNA lesion is different than the base normally inserted opposite an undamaged base at that site. The chemical structure of the lesion is an important determinant for mutagenic or nonmutagenic bypass during TLS. Irradiation of DNA with UV light produces a variety of photoproducts that can cause mutations. cis-syn cyclobutane pyrimidine dimer (CPD) and  photoproducts ((6-4)PP) are the two major classes of UV-induced DNA photoproducts. In DNA irradiated with UV, CPD is the most abundant lesion and it is approximately 3-5-fold more abundant than PP. However, the mutagenic properties of these two lesions are different, and the relative contributions of the two lesions to the by guest on March 24, 2020 http://www.jbc.org/ Downloaded from Ishikawa et al. 4 mutagenesis are not simply proportional to their relative abundance. The mutagenic specificities of CPD and (6-4)PP have been studied using well characterized site-specific UV photoproducts in DNA substrates. The mutation frequency obtained either with a site-specific TT-CPD or a TC-CPD is quite low (17 -18), and in comparison, PP is more mutagenic. The PP at the 5'-TT site was highly mutagenic, and most of the mutations induced by this lesion are T to C transitions at the 3'-T (19). However, this type of mutation is not very common, because the (6-4)PP at the TT site is relatively rare.
On the other hand, the TC-(6-4)PP is less mutagenic than the TT-(6-4)PP but more mutagenic than the TT-CPD. Of the mutations at this site, 80% were C to T transitions at the 3' base (20). Since TC to TT is frequently observed in DNA exposed to UV, and (6-4)PP forms most frequently in the TC sequence, the TC-(6-4)PP could be a candidate for a strongly premutagenic lesion.
Human cells have three TLS polymerases that belong to the UmuC/DinB/Rev1/Rad30 superfamily: polη (encoded by the XPV/RAD30A gene, (21,22)), polι (encoded by RAD30B, (10,23)) and polκ (encoded by DINB1, (24)). In addition, the human protein hREV1 is a homologue of yeast REV1 (25). Polη, polι and polκ bypass UV-induced damage with different efficiencies. Polη bypasses TT-CPD efficiently and inserts AA opposite the lesion; however, it adds only one base opposite the 3'-T of TT-(6-4)PP (5). Polι bypasses TT-CPD and TT-(6-4)PP with low efficiency, adding one or a few bases opposite these lesions (26,27). Polκ stops DNA synthesis and is blocked one base before TT-CPD and TT-(6-4)PP (9). Thus, in human cells none of these polymerases bypass PP. In contrast, E. coli DNA polymerase V (UmuD' 2 C, (28)) efficiently bypasses both lesions, inserting AA or GA opposite TT-CPD or TT-(6-4)PP, respectively (8). Thus, the chemical structure and the properties of The goal of this work was to identify and characterize the Drosophila homologue(s) of the UmuC/DinB/Rev1/Rad30 superfamily. Two cDNAs were isolated that encode Drosophila Rad30A and Rad30B, called drad30A and drad30B. Truncated forms of dRAD30A and dRAD30B proteins, designated dpolη∆C and dpolι∆C, respectively, have been purified and their properties studied, including the mechanism of UV-induced mutagenesis and the products of lesion bypass in reactions with these enzymes. The bypass templates used in this study include TC-(6-4)PP, TT-CPD or TT-(6-4)PP; a template with TC-(6-4)PP was studied because of the biological relevance this lesion. The results indicate that both polymerases bypass the TT-CPD in an error-free manner. Furthermore, dpolη∆C bypasses TT-(6-4)PP and TC-(6-4)PP in a highly error-prone manner.

Isolation of Drosophila rad30A and rad30B cDNAs
To identify the genes encoding the UmuC/DinB/Rev1/Rad30 superfamily proteins in Drosophila, we searched the database of expressed sequence tag (GenBank dbEST) for Drosophila cDNAs that share homology with the proteins of this family. Several Drosophila cDNA sequences that share significant homology with the proteins of the superfamily were identified. Three such clones, SD05329, LD09220, and GH11153, were obtained from Research Genetics (Huntsville, AL) , and the entire insert of each clone was sequenced with an automated DNA sequence analyzer (Applied Biosystems PRISM310 Rad30B, and Rev1 proteins, respectively. Thus, the clones were designated as drad30A, drad30B, and drev1, respectively. The cDNAs did not contain in-frame stop codons. Thus, the 5' ends of each clone were amplified by 5' RACE (5' rapid amplification of cDNA ends). The sequence determination of the amplified 5' ends revealed the presence of the in-frame stop codon. Screening of the dbEST indicated that Drosophila seems to lack a true DinB ortholog, which was confirmed by searching the recently completed sequence of the Drosophila genome. The drad30A and drad30B cDNAs each hybridized to Drosophila polytene chromosomes at a single site corresponding to bands 3L-79BC and 3R-84EF, respectively.

Overproduction of the dRAD30A and dRAD30B proteins
The expression vector used in these studies was pGEX-4T (Pharmacia Biotech). To overexpress the Drosophila dRAD30A and dRAD30B proteins, each gene was cloned in-frame with the glutathione S-transferase gene. A chimeric GST-drad30A gene was constructed by recloning the 3 kilobase pair (kb) EcoRI/XhoI DNA fragment from the EST clone SD05329 in EcoRI/XhoI digested pGEX-4T-1, thus generating plasmid pGEX-dRad30A. The Carboxyl (C-) terminal deletion mutant of the drad30A gene was constructed by recloning the 1640 base pair (bp) EcoRI/Eco52I DNA fragment, which contains the N-terminal 545 residues, in EcoRI/NotI digested pGEX-4T-1, thus generating pGEX-dRad30A∆C. The chimeric GST-drad30B gene was constructed as follows. EcoRI restriction site was generated just upstream of the start codon by PCR using the primer oligonucleotides 30B5'Eco, 5'-CGGAATTCATGGACTTCGCTAGCGTAC-3' and 3 kb EcoRI /XhoI DNA fragment from the EST clone LD09220 was then cloned in-frame with GST in the EcoRI/XhoI

Purification of the dRAD30A and dRAD30B protein
To purify the GST tagged dRAD30A or dRAD30B, E. coli cells were resuspended in PBS and disrupted using a sonicator before centrifugation at 10,000 x g. 10 mM Mg-ATP and 5 mg/mL casein were added to the extract, which was then incubated at room temperature for 20 min. The extract was then passed over a 5 mL glutathione-Sepharose antibody was added to the purified protein in all experiments in this paper. However, after addition of pol I antibody a slight 3'-exonuclease activity was still remaining.

DNA polymerase assays
The 30mer oligomer (non-damaged TT) with the sequence 5'-CTCGTCAGCATCTTCATCATACAGTCAGTG-3' was used as the nondamaged template. The same 30mer oligomers containing the CPD (29) and the (6-4)PP (30) at the underlined sites were synthesized as described. The 30mer oligomers, in which the underlined TT sequence of the "non-damaged TT" oligomers was changed to TC, were also used as nondamaged templates (designated "non-damaged TC") and the 30mer oligomers containing (6-4)PP at the TC site were synthesized as described (31). Various lengths (16-18 mer) of oligomers with sequences complementary to the 30mer template were synthesized and used as primers. Each primer was labeled at the 5'-end using T4 polynucleotide kinase and [γ-32 P] ATP and was annealed with the template at a molar ratio glycerol, 40 nM primer-template, and the indicated amount of enzyme. After incubation at 37 o C for 15 min, the reactions were terminated by the addition of 10 µL formamide followed by boiling. The products were subjected to 20% polyacrylamide/7M urea gel electrophoresis followed by autoradiography.

Detection of correctly replicated products opposite lesions.
Two kinds of oligomers were used: 49mers bearing the centrally located (6-4)PP or CPD at the TT site (32) and 30mers bearing the centrally located (6-4)PP at the TC site (31).
The substrate sequence is as follows (the introduced thymine dimer is underlined): damage containing strands, respectively, were synthesized and used as primers. After labeling with γ 32 P-ATP, the primer was annealed with the template and was used as the substrate for DNA synthesis. DNA synthesis was carried out as described in the DNA polymerase assay. After DNA synthesis, the replication products were denatured by incubation at 95 o C for 5 min, and then were annealed with a ten-fold excess of the 49mer which did not contain enzyme, was used as the background for bypass products (Counts-bp-bg). The amounts of the bypass products susceptible to each enzyme were determined by measuring the radioactivity of the 19 base or 14 base fragments (Countdig). The radioactivity of the same area of the undigested reaction was measured and used as the background of the digested products (Counts-dig-bg). The "% restriction enzyme sensitive bypass products" in the Table was calculated as (Count-dig)-(Counts-digbg)/(Counts-bp)-(Counts-bp-bg)x100.

Isolation of Drosophila drad30A and drad30B cDNAs
Two Drosophila cDNA clones were identified in the GenBank dbEST database which had significant homology to human Rad30A and Rad30B protein, and were designated drad30A and drad30B. DNA sequence analysis of the full length cDNAs revealed that drad30A encodes a protein of 885 aa with 31% identity to the human XPV/RAD30A protein (hpolη) and drad30B encodes a protein of 737 aa with 30% identity to the human RAD30B protein (hpolι). Previous studies have shown that the proteins in the UmuC/DinB/Rev1/Rad30 superfamily are most highly conserved in their N-terminal portion which includes five discrete conserved regions (1,33). As shown in The amino acid sequences were aligned for all the members of the UmuC/DinB/Rev1/Rad30 protein family, and the alignment was used to generate a phylogenetic tree ( Figure 1C). The tree reveals several distinct subgroups of proteins; the validity of these subgroups are convincingly supported by the bootstrap test. Since the Rad30B subfamily is equally distant from the RAD30A and DinB subfamilies, the superfamily can be classified into five subgroups; UmuC, DinB, Rev1, Rad30A, and Rad30B. The Drosophila genes drad30A and drad30B belong to the Rad30A and Rad30B subfamilies, respectively. A Drosophila homologue of human REV1 was also identified, which belongs to the REV1 subfamily as shown in Figure 1C; however, no Drosophila homologue of the human DINB1 protein (hpolκ) was identified in the cDNA by guest on March 24, 2020 http://www.jbc.org/ Downloaded from libraries that were searched. In addition, the sequence of the Drosophila genome (34) does not include a homologue of polκ.

Overproduction and purification of dRAD30A and dRAD30B proteins
The proteins encoded by drad30A and drad30B (dRAD30A and dRAD30B, respectively) were purified as recombinant fusion proteins with a GST tag at the Nterminus. Initially, the entire coding sequences of the drad30A and drad30B genes were fused with the GST gene; however, these constructs produced a very low yield of the desired recombinant proteins in E. coli. However, when a mutation was introduced in each gene that creates a stop codon in the C-terminal region, the resulting C-terminally truncated proteins were expressed at a higher level in E. coli. These truncated forms of the drad30A and drad30B gene products are similar in size to the truncated forms of hpolη and hpolκ proteins and include all of the N-terminal conserved regions of the UmuC/DinB/Rev1/Rad30 superfamily; thus, they were expected to be active polymerases despite the missing C-terminal region (5,9,14,21). The truncated forms of dRAD30A (dRAD30A∆C) and dRAD30B(dRAD30B∆C) include the N-terminal 545 residues and 445 residues of each polypeptide, respectively. They were purified by GST-Sepharose affinity chromatography (see Materials and Methods) and characterized enzymatically as described below (Figure 2).

DNA polymerase activities of dRAD30A C and dRAD30B C
The dRAD30A∆C and dRAD30B∆C proteins were assayed for DNA polymerase activity using a primer extension assay with a 5'-endlabeled 16mer primer annealed to a 30mer template. As shown in Figures 2B and 2C, both dRAD30A∆C (Figure 2B Figure   1B). In the mutants generated for this study, these residues were changed to alanine. The mutant proteins displayed the same chromatographic properties as the wild type protein, but they lacked DNA polymerase activity ( Figures 2B and 2C, lanes 2-4). Therefore, the observed DNA polymerase activity is intrinsic to the dRAD30A and dRAD30B proteins.
We therefore propose that dRAD30A be renamed dpolη and dRAD30B be renamed dpolι. In the following text, dpolη∆C will be used to refer to the truncated protein dRAD30A∆C, and dpolι∆C will be used to refer to dRAD30B∆C.
The nucleotide selectivity of dpolη∆C and dpolι∆C was analyzed in DNA synthesis reactions using a non-damaged DNA template in the presence of only one deoxyribonucleotide (i.e., dA, dG, dC or T). Incorporation of the correct complementary nucleotides and to less extents incorrect nucleotides opposite each template occurred (data not shown), suggesting that both dpolη∆C and dpol ι∆C have relatively low fidelity.
Translesion synthesis by dpol C and dpol C by guest on March 24, 2020 http://www.jbc.org/ Downloaded from To determine whether dpolη∆C and dpolι∆C bypass DNA lesions, polymerase assays were carried out using templates with TT-CPD or TT-(6-4)PP lesions. These templates were described in previous studies of hpolη (5 7), hpolι (26) and hpolκ (9). In addition, a template with TC-(6-4)PP in the same sequence context as the above two templates was constructed and used in this study. A 16mer primer was annealed to the template-containing lesion which terminates just before the lesion site (Figure 3). dpolη∆C and dpolι∆C efficiently bypass TT-CPD ( Figure 3A) and the efficiencies of DNA synthesis on the lesion-containing template were almost the same as on the nondamaged template. dpolη∆C showed some ability to bypass TT-(6-4)PP ( Figure 3B,  (7), hpolη did not bypass TT-(6-4)PP, arresting DNA synthesis after addition of one base opposite the 3'-T of the lesion. Therefore, the ability to bypass TTand TC-(6-4)PP is a remarkable feature of dpolη∆C , which distinguishes it from other RAD30-like polymerases.

Nucleotide selectivity of dpol C and dpol C during incorporation opposite lesions
Polymerization reactions by dpolη∆C and dpolι∆C were performed with the lesioncontaining template-primer substrates described above in the presence of a single deoxynucleotide. The reaction products were analyzed and are shown in It should be noted that on the template containing no damage, both dpolη∆C and dpolι∆C incorporated not only A but also G opposite T ( Figure 4A, no damage).
Misincorporation of G opposite T is a common feature of the polη and polι family, and especially hpolι incorporates G opposite T more efficiently than A opposite T (10, 13, 27, 37). However, doplι∆C misincorporates G opposite T less often than it correctly incorporates A opposite T.

Influence of base pairing at the lesion on elongation past the lesion by dpol C and dpol C
The results shown in Figure 4 indicate that dpolη∆C and dpolι∆C can insert a variety of bases opposite the 3' base of TT-CPD, TT-(6-4)PP or TC-(6-4)PP. However, it is possible that the misincorporated base may not be extended as efficiency as the correctly incorporated base is, as shown in the case of hpolη for the bypass of various lesions (7).
To explore this question, primers were synthesized in which each of the four base would be correctly or incorrectly paired with the 3' or 5' base of the above three lesions. First, the ability of dpolη∆C to elongate such primers annealed to TT-CPD was tested and the  Figure 5. dpolη∆C elongated DNA chains more efficiently from the correctly base paired primers, the 17mer with a 3'-terminal A opposite 3' T of TT-CPD ( Figure 5B) and the 18mer with AA opposite the lesion ( Figure 5D). Similar results were obtained with dpolι∆C (data not shown).
Next, such primers were annealed to TT-(6-4)PP or TC-(6-4)PP to examine whether they could be extended by dpolη∆C. On a template with TT-(6-4)PP and a 17mer primer whose 3'-terminal nucleotide pairs with the 3'-T of the lesion, dpolη∆C elongated the DNA chain more efficiently from a primer with a 3'-terminal G (G17) than from a primer with a 3'-terminal A (A17) ( Figure 6A). This makes a sharp contrast with the above result that in the case of TT-CPD, dpolη∆C carried out the elongation more efficiently from the A17 primer than from the G17 primer ( Figure 5B Figure   7A). Almost all of the bypass products of TT-CPD by dpolη∆C or dpolι∆C were cleaved by MseI ( Figure 7B, lanes 9-12); more than 90% of the bypass products have the AA sequence (Table 1). On the other hand, only some of the bypass products of either TT- Damage-induced mutagenesis is an important biological process whose molecular mechanism is not yet understood, and which is the subject of much current research.
Recently, the proteins of the UmuC/DinB/Rev1/Rad30 superfamily have been isolated and characterized biochemically, and it was shown that these proteins have a nonprocessive DNA polymerase activity that can bypass DNA lesions. Thus, these enzymes are central to the process of damage-induced mutagenesis. In this study, Drosophila homologues of this polymerase superfamily, dpolη and dpolι, were identified and characterized with respect to their ability to bypass UV-induced DNA lesions.
S. cerevisiae and human polη are considered to bypass TT-CPD in a mostly error-free manner (7,15,21). Similarly, dpolη∆C and dpolι∆C incorporate AA opposite TT-CPD more efficiently than other nucleotides ( Figure 4A) and the elongation starts selectively from a primer ending in A opposite the lesion ( Figure 5B and D). These results suggest that dpolη∆C and dpolι∆C bypass TT-CPD in a error-free manner as in the case of hpolη. In addition, direct analysis of the products of lesion bypass also indicates that dpolη∆C and dpolι∆C carry out error-free bypass of TT-CPD ( Figure 7).
Thus, for these enzymes, TT-CPD maintains a structure that directs correct Watson-Crick base pairing, in spite of the base modification. In fact, Lawrence and co-workers have argued that the CPD must be an instructive lesion by virtue of its high coding specificity in E. coli (38) and structural studies have shown that the Watson-Crick base pair is still intact at the CPD site (39 -41). Therefore, the TLS polymerases bypass this lesion by incorporating the correct bases on this instructional lesion.
However, evidence also indicates that dpolη∆C does not carry out error-free bypass of (6-4)PP lesions. On templates with TT-(6-4)PP, dpolη∆C did not show  (19). These results indicate that the TT-(6-4) photoproduct can be classified as a mis-instructional lesion.
In spite of the biological relevance of this DNA lesion, the tertiary structures of oligonucleotide duplexes which include TC-(6-4)PP have not been well characterized. Fujiwara and Iwai (1997) showed that the duplex containing G opposite the 3'-C of TC-(6-4)PP is thermodynamically more stable than another duplex (42). Horsfall and Lawrence (1994) have studied the mutagenecity of TC-(6-4)PP in E. coli cells and found that 66% of the bypass products have the correct TC sequence, 28% of the mutants having the TT sequence (20). Our observation that both G and A are incorporated that dpolη∆C elongated the DNA chain equally well from the 18mer primer with A or G opposite the 5'-T (data not shown). It will be necessary to determine the sequence of the bypass products directly or perform steady-state kinetics to clarify this point.
The mutagenic properties of dpolη∆C and dpolι∆C fit well with the mutation spectrum obtained using a defined photoproduct introduced into SOS-induced E. coli (19,38). The mutation spectrum reported in those studies reflects the mutagenic properties of pol V (UmuD' 2 C complex). Thus, the mutagenic properties of dpolη∆C on TT-CPD and TT-(6-4)PP are very similar to those of pol V. Both enzymes bypass TT-CPD in an error-free manner and incorporate GA opposite TT-(6-4)PP cells ((8) and this study). The similarity in the nucleotides incorporated suggests that structural information in the altered bases contribute to nucleotide selection during incorporation opposite these lesions by these polymerases.
In the Rad30 protein family, dpolη∆C seems to be an unusual member of the Rad30 family in having the ability to bypass (6-4)PP, although we cannot rule out a possibility that having a N-terminal GST tag and a C-terminal truncation have affected the TLS specificity and activity and we also have to keep in mind that we cannot draw a quantitatively definitive conclusion since the observations present in this study are qualitative. It should also be noted that the DNA repair capacity of Drosophila cells is somewhat different from that of human cells. The most striking difference is that In this study, TT(6-4)PP was bypassed by dpolη∆C but not by dpolι∆C . The TLS activity of dpolη∆C and dpolι∆C was also examined on chemically modified damage, and all of the other lesions tested were bypassed by both enzymes (data not shown). Thus, at present we cannot assign any specific biological function to dpolι.

ACKNOWLEDGEMENT
We thank E. Ohashi for technical assistance and for kindly providing oligonucleotides used as primer and template and T. Ogi for the phylogenetic analysis to construct the tree ( Figure 1C).