In Vitro selection of sequence contexts which enhance bypass of abasic sites and tetrahydrofuran by T4 DNA polymerase holoenzyme1

Article No. jmbi.1998.2520 available online at http://www.idealibrary.com on

J. Mol. Biol. (1999) 286, 1045±1057

In Vitro Selection of Sequence Contexts Which Enhance Bypass of Abasic Sites and Tetrahydrofuran by T4 DNA Polymerase Holoenzyme Zafer Hatahet1*, Meixia Zhou1, Linda J. Reha-Krantz3, Hiroshi Ide4 Scott W. Morrical2 and Susan S. Wallace1 1

Department of Microbiology and Molecular Genetics and the Markey Center for Molecular Genetics and 2Department of Biochemistry, University of Vermont, Burlington VT 05405 3 Department of Biological Sciences, University of Alberta Edmonton, Alberta, Canada T6G 2E9 4

The Graduate Department of Gene Science, Hiroshima University, Kagamiyama Higashi-Hiroshima 739, Japan

The in¯uence of sequence context on the ability of DNA polymerase to bypass sites of base loss was addressed using an in vitro selection system. Oligonucleotides containing either an aldehydic abasic site or tetrahydrofuran surrounded by four randomized bases on both the 50 and 30 sides were used as templates for synthesis by phage T4 DNA polymerase holoenzyme pro®cient or de®cient in the 30 ! 50 proofreading exonuclease activity. Successful bypass products were puri®ed, subcloned and the sequences of approximately 100 subclones were determined for each of the four polymerase/lesion combinations tested. Between 7 and 19 % of the bypass products contained deletions of one to three nucleotides in the randomized region. In bypass products not containing deletions, biases for and against certain nucleotides were readily noticeable across the entire randomized region. Template strands from successful bypass products of abasic sites had a high frequency of T in most of the randomized positions, while those from bypass products of tetrahydrofuran had a high frequency of G at the positions immediately to the 30 and 50 side of the lesion. Consensus sequences were shared by successful bypass products of the same lesion but not between bypass products of the two lesions. The consensus sequence for ef®cient bypass of tetrahydrofuran was over-represented in several frames relative to the lesion. T4 DNA polymerase inserted A opposite abasic sites 63 % of the time in the presence of proofreading and 79 % of the time in its absence, followed by G > T > C, while the insertion of A opposite tetrahydrofuran ranged between 93 % and 100 % in the presence and absence of proofreading, respectively. Finally, sequence context in¯uenced the choice of nucleotide inserted opposite abasic sites and consensus sequences which favored the incorporation of nucleotides other than A were de®ned. # 1999 Academic Press

*Corresponding author

Keywords: sequence context; translesion DNA synthesis; T4 DNA polymerase; AP sites; mutagenesis

Introduction Interactions between replicative DNA polymerases and oxidative DNA lesions play a major Present addresses: Z. Hatahet, Department of Biochemistry, University of Texas Health Center at Tyler, Tyler, TX 75708, USA; M. Zhou, Center for Immunology, Washington University School of Medicine, St. Louis, MO 63110, USA. Abbreviations used: AP, abasic site. E-mail address of the corresponding author: [email protected] 0022-2836/99/091045±13 $30.00/0

role in determining the lesion's effect on the integrity of the cell. Unrepaired DNA lesions generally have one of two effects on DNA polymerases: they either block replication, effectively killing the cell, or they allow bypass, a potentially mutagenic process depending on the nucleotide which is incorporated opposite the lesion. A large body of in vitro and in vivo experiments from several laboratories has attempted to identify the factors involved in translesion synthesis and their relative roles. Such factors include lesion structure, properties of the DNA polymerase, as well as the lesion's sequence # 1999 Academic Press

1046 context (for a review, see Hatahet & Wallace, 1997). As predicted, lesion structure largely determines the outcome of translesion DNA synthesis. For instance, abasic (AP) sites, ring fragmentation products such as urea, and ring open structures such as b-ureidoisobutyric acid and formamidopyrimidines strongly block in vitro DNA synthesis (Sagher & Strauss, 1985; Ide et al., 1985, 1991; Tudek et al., 1992). Products that distort DNA such as thymine glycol also block DNA synthesis (Ide et al., 1985; Clark & Beardsley, 1987; Basu et al., 1989). However, other ring saturation and oxidation products such as dihydrothymine, dihydrouracil, uracil glycol, 5hydroxycytosine, 5-hydroxyuracil, 8-oxoguanine, and 8-oxoadenine are very poor blocks (Ide et al., 1991; Purmal et al., 1994, 1998; Shibutani et al., 1991, 1993; Liu & Doetsch, 1998). When bypass takes place, lesion structure also plays a major role in the choice of the incoming nucleotide. For example, dihydrothymine codes for A (Maccabee et al., 1994), a non-mutagenic pair, while 8-oxoguanine and 5-hydroxycytosine code for a mutagenic A (Shibutani et al., 1991; Purmal et al., 1994), as well as their non-mutagenic pairs, C for the former and G for the latter. On the other hand, 5-hydroxyuracil and uracil glycol, two oxidation products of cytosine (Wagner et al., 1992), code predominantly for a mutagenic A (Purmal et al., 1994, 1998). Properties of the DNA polymerase such as the proofreading 30 ! 50 exonuclease activity and processivity also play a role in translesion synthesis. In general, high processivity enhances lesion bypass, while the proofreading exonuclease activity impedes it (Shibutani et al., 1997; PazElizur et al., 1996, 1997; Mozzherin et al., 1997; Rajagopalan et al., 1992). However, the ability of a DNA polymerase to bypass a particular lesion and the nucleotide incorporated in the process seem to be intrinsic properties of the particular polymerase as well as the bypassed lesion (Shibutani et al., 1991). Perhaps the most dramatic factor in¯uencing translesion synthesis is sequence context. We and others have shown that several ``blocking lesions'', including thymine glycol, abasic sites, urea, and b-ureidoisobutyric acid, were completely bypassed within certain sequence contexts (Evans et al., 1993; Maccabee et al., 1994; Hayes & LeClerc, 1986; Belguise-Valladier & Fuchs, 1995). More recently, we demonstrated sequence context effects on the pairing properties of several oxidative lesions (Hatahet et al., 1998; Purmal et al., 1994). Since a strong correlation exists between mutagenesis/oncogenesis and the processing of oxidative DNA damage (Ames, 1989a,b), we set out to systematically address the in¯uence of sequence context on DNA polymerase bypass of strongly blocking lesions. To achieve this aim, we studied the bypass of AP sites or tetrahydrofuran (furan) positioned in the center of 65,536 different

Bypass of AP Sites by T4 DNA Pol Holoenzyme

nanonucleotide sequences (the product of four random nucleotides each on the 30 and 50 side of the lesion, see Materials and Methods) by phage T4 DNA polymerase holoenzyme pro®cient (holoenzyme) or de®cient (holoenzyme exoÿ) in the 30 ! 50 exonuclease activity. We opted to use AP sites (Figure 1) as model blocking lesions for several reasons. The steady state level of AP sites in cellular DNA is one of the highest for any lesion (Ames, 1989a,b; Lindahl, 1993), and a large body of in vitro and in vivo studies on the processing of AP sites exists in the literature (for reviews, see (Wallace, 1988, 1997)). In addition, bypass of AP sites is mutagenic, since they are predominantly produced in vivo by depurination of G and A (Lindahl & Nyberg, 1972) and they predominantly code for A (Hatahet & Wallace, 1997). Furan (Figure 1), a synthetic molecule used frequently as an AP site analog (Efrati et al., 1997; Shibutani et al., 1997; Cai et al., 1993; Randall et al., 1987; Kalnik et al., 1988; Takeshita et al., 1987), was also used. We opted to use T4 DNA polymerase holoenzyme since its biochemistry is well understood (Young et al., 1992) and it shares structural homology with eukaryotic DNA polymerases, including the human polymerase dPCNA complex (Braithwaite & Ito, 1993). Holoenzyme de®cient in exonucleolytic proofreading (exoÿ) was used in order to shed some light on the mechanism of lesion bypass (see Discussion) as well as the contribution of proofreading to this process.

Figure 1. Chemical structure of AP site and furan.

Bypass of AP Sites by T4 DNA Pol Holoenzyme

1047

Results Selection of sequence contexts for efficient bypass of AP sites and furan by T4 DNA polymerase holoenzyme This project was based on the premise that certain sequence contexts modulate the presentation of a blocking lesion to a replicating DNA polymerase such that nucleotide insertion and extension opposite the lesion is greatly facilitated. As a consequence, we predicted that AP sites and furan, generally considered to be strong blocking lesions, would be bypassed at a high rate within a small subset of the 65,536 different sequence contexts used in this study, and at a very low rate in the remaining contexts. To select for the sequences which promote ef®cient bypass, the reactions were controled in two ways. Heparin was used in large molar excess (500X, see Materials and Methods) of the DNA to prevent rebinding of polymerase molecules which dissociate from the template upon encountering the lesion. In addition, the extent of bypass as a function of reaction time was monitored using 8 M urea PAGE of radiolabeled nascent strands and quantitation by a molecular imager. Bypass products were selected for subcloning and sequencing from reactions where the fulllength nascent strand (46mer) represented only 2 to 4 % of the molecules blocked by the lesion (25mer, see Figure 2). The sequences of 110, 95, 114 and 121 independent products of AP bypass by holoenzyme, AP bypass by holoenzyme exoÿ, furan bypass by holoenzyme, and furan bypass by holoenzyme exoÿ, respectively, were determined (data not shown). Depending on the enzyme/template combination used, 7 to 19 % of the samples contained deletions of one or two nucleotides in the randomized region of the original template (and one sample contained a deletion of three nucleotides in products of furan bypass by holoenzyme exoÿ), while no deletions were observed in the constant region (data not shown). The sequences which contained deletions were rich in nucleotide repeats and/or inverted repeats, consistent with bypass through a misalignment mechanism (Kunkel, 1990). Unfortunately, the random nature of the template precluded de®nitive positioning of the deleted nucleotide(s) relative to the lesion, which in turn precluded the de®nition of sequence contexts which enhanced lesion bypass through a misalignment/deletion mechanism. Although such contexts are of interest to us, an approach which is different from the one described in this report is clearly needed to address them, and they will be the subject of another study. To determine whether the selected samples represented related sequences, nucleotide frequencies at each of the randomized positions of independent clones not containing deletions were calculated. For each lesion/polymerase combination, statistically signi®cant biases for and against certain

Figure 2. Analysis of bypass products during in vitro selection. Samples of the primer extension products of the four polymerase/template combinations used in this study were separated on 12 % polyacrylamide, 8 M urea denaturing gels, and analyzed by phosphorimaging. (a), (b) Different transformations of the same image, with the positions of the unextended primers (16mer), lesionblocked primers (25-26mer), and bypass products (46mer) pointed out.

nucleotides were readily noticeable at several positions (Table 1). For instance, in holoenzyme AP bypass products, T was signi®cantly over-represented as the 50 nearest neighbor of the lesion, while C was over-represented as the 30 nearest

1048

Bypass of AP Sites by T4 DNA Pol Holoenzyme

Table 1. Nucleotide frequencies (%) at randomized positions in the template strands of successful bypass products AP holoenzyme 50 ...CAGAGT A G C T w2 P

(nˆ99) N 27.3 8.1 28.3 36.4 17.3 <0.001

N 15.2 25.3 24.2 34.3 7.4 <0.1

N 28.3 23.2 22.2 26.3 0.9 >0.5

N 17.2 25.3 18.2 39.4 12.6 <0.01

(N) 13.1 5.1 19.2 62.6 79.5 <0.001

N 31.3 19.2 35.4 14.1 11.9 <0.01

N 19.2 30.3 20.2 29.3 4.1 <0.25

N 19.2 27.3 20.2 33.3 5.3 <0.25

N 26.3 20.2 20.2 32.3 4.1 <0.5

TGACAG...30

AP holoenzyme 50 ...CAGAGT A G C T w2 P

exoÿ (nˆ87) N 31.0 6.9 29.9 32.2 17.6 <0.001

N 13.8 23.0 27.6 35.6 10.0 <0.025

N 35.6 24.1 24.1 16.1 7.8 <0.1

N 26.4 17.2 31.0 25.3 4.0 <0.5

(N) 3.4 1.1 16.1 79.3 162.5 <0.001

N 32.2 21.8 20.7 25.3 3.2 <0.5

N 28.7 16.1 14.9 40.2 17.1 <0.001

N 28.7 11.5 25.3 34.5 11.5 <0.01

N 18.4 20.7 25.3 34.5 6.1 <0.25

TGACAG...30

N 24.2 29.7 15.4 29.7 5.5 <0.25

N 20.9 31.9 36.3 9.9 16.8 <0.001

N 11.0 45.1 20.9 22.0 25.0 <0.001

(N) 5.5 0.0 1.1 93.4 250.2 <0.001

N 29.7 45.1 14.3 11.0 29.4 <0.001

N 23.1 31.9 15.4 28.6 6.2 0.1

N 19.8 27.5 20.9 30.8 3.3 <0.5

N 20.9 23.1 25.3 30.8 2.2 >0.5

TGACAG...30

N 22.3 37.5 18.8 21.4 8.6 <0.05

N 14.3 35.7 26.8 23.2 9.4 <0.025

(N) 0.0 0.0 0.0 100.0 300.0 <0.001

N 27.7 40.2 8.9 23.2 20.0 <0.001

N 27.7 25.0 13.4 32.1 7.7 <0.1

N 16.1 15.2 32.1 36.6 14.5 <0.005

N 22.3 22.3 20.5 33.9 4.6 <0.25

TGACAG...30

Furan holoenzyme (nˆ91) 50 ...CAGAGT N A 34.1 G 8.8 C 28.6 T 26.4 w2 14.4 P <0.005

Furan holoenzyme exoÿ (nˆ112) 50 ...CAGAGT N N A 23.2 25.0 G 21.4 24.1 C 19.6 20.5 T 34.8 29.5 w2 5.6 1.6 P <0.25 >0.5

The uppermost rows indicate the positions of the randomized nucleotides in the original template strand. The nucleotide in parentheses represents the position of the lesion. w2 and P values are for deviation from random distribution of A, G, C or T at each position.

neighbor. Biases for G as the nearest neighbor of furan were even more striking in bypass products of the holoenzyme (P < 0.001). The possibility that these biases were generated during chemical synthesis of the templates was ruled out by the following observations. No biases were seen in the sequences of >200 independent clones which were either not subjected to selection (data not shown) or subjected to selection of a different nature (Hatahet et al., 1998). More convincingly, distinct biases were observed when either form of the polymerase were used. For instance, C was under-represented as the 50 nearest neighbor of AP sites in bypass products of holoenzyme, while it was overrepresented at the same position in bypass products of holoenzyme exoÿ (Table 1). Finally, enzyme kinetics was used to con®rm the ef®cacy of sequence contexts which enhanced lesion bypass (see below). Consensus sequences for efficient bypass of AP sites and furan by T4 DNA polymerase holoenzyme To de®ne speci®c contexts which enhanced lesion bypass, we searched the selected samples for all possible di-, tri- or tetranucleotide permutations

which were statistically over-represented (P < 0.05, see Materials and Methods). Gaps were accommodated to avoid bias for contiguous contexts (i.e. in addition to searching for all permutations of XYZ, all permutations of XNYZ, XYNZ, XNNYZ, etc. were searched). For each enzyme/template combination, several over-represented sequence contexts were obtained and are summarized in Table 2. Many over-represented sequences were not included because they were not statistically different from those reported. For instance, 23 clones of AP bypass products by holoenzyme shared the statistically over-represented dinucleotide sequence 50 (N)NGY (w2 ˆ 9.12, P < 0.005, the nucleotide in parentheses denotes the position of the original lesion). However, 20 of those clones shared the sequence 50 (N)MGY (Table 2), indicating that the former sequence was over-represented simply because it contained the latter. Similarly, furan bypass products by both holoenzyme and holoenzyme exoÿ shared over-represented 50 G(N) and 50 G(N)R which are not statistically different from the sequence 50 G(N)RK (Table 2). It is critical to note that in contexts containing a degenerate nucleotide (other than N which we used exclusively to indicate positions not involved in de®ning a consensus), every unique sequence which ®t the

1049

Bypass of AP Sites by T4 DNA Pol Holoenzyme Table 2. Over-represented sequence contexts in successful bypass products Oa

Consensus AP holoenzyme (n ˆ 99)

AP holoenzyme exoÿ (n ˆ 87) Furan holoenzyme (n ˆ 91) Furan holoenzyme exoÿ (n ˆ 112)

a b

b

5' (N)MGY 5' T(N)CGY 5' TAN(N) 5'YTNN(N) 5' (N)NTA 5' TRN(N) 5'YTNN(N) 5' G(N)RK 5' KSG(N)RK 5' (N)RKT 5' G(N)RK 5' KGS(N)RK

20 9 13 23 15 22 23 25 19 22 23 9

Ea

w2

6.2 0.8 6.2 12.4 5.4 10.9 10.9 5.7 1.4 7 7 1.8

30.82 87.5 7.5 9.12 16.82 11.38 13.52 65.58 217 32.14 36.57 30

P <0.001 <0.001 <0.01 <0.005 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

O ˆ observed, E ˆ expected. Following IUPAC-IUB nomenclature, R ˆ A or G; Y ˆ C or T; K ˆ G or T; M ˆ A or C; S ˆ G or C; N ˆ any nucleotide.

consensus was over-represented in the sample pool. For example, in the context 50 G(N)RK, each of the sequences 50 G(N)AG, 50 G(N)AT, 50 G(N)GG, and 50 G(N)GT was signi®cantly overrepresented.

(i.e. 50 YYY(AP), 50 YY(AP)Y, 50 Y(AP)YY or 50 (AP)YYY), and 19 % of holoenzyme products and 23 % of holoenzyme exoÿ products contained lesions which were part of a string of ®ve pyrimidine bases (data not shown).

Sequences for efficient bypass of AP sites

Sequences for efficient bypass of furan

The most signi®cant consensus shared by holoenzyme AP bypass products was 50 (N)MGY (20 samples, Table 2). Even more statistically overrepresented were nine of the latter 20 samples with the consensus 50 T(N)CGY (Table 2). The former consensus was likely selected on its own merit since 11 samples with the sequence 50 (N)MGY and a nucleotide other than T as the 50 nearest neighbor were still signi®cantly over-represented (w2 ˆ 5.14, P < 0.025). We suggest that the presence of MGY on the 30 side of the AP site affected ef®cient bypass, while the addition of T on the 50 side further augmented the bypass ef®ciency. Two sequences on the 50 side of the lesion, 50 TAN(N) and 50 YTNN(N), were also over-represented and largely independent of the consensus on the 30 side. Interestingly, loss of exonucleolytic proofreading altered sequence preference on the 30 side of the lesion. Holoenzyme exoÿ AP bypass products shared over-represented sequences which were only two nucleotides long. Although over-represented trinucleotide sequences were detected, they were not signi®cantly different from those reported in Table 2. The sequence on the 30 side of the lesion, 50 (N)NTA, shared no homology with those of the holoenzyme bypass products. On the other hand, the two over-represented sequences on the 50 side of the lesions, 50 TRN(N) and 50 YTNN(N) were identical with those found in holoenzyme bypass products (of the 22 50 TRN(N) sequences, 12 were 50 TAN(N)). Also common between holoenzyme and holoenzyme exoÿ AP bypass products were sequences rich in runs of pyrimidine bases. A total of 28 % of the holoenzyme bypass products and 33 % of the holoenzyme exoÿ products represented lesions which were part of a string of four consecutive pyrimidine bases

A much stronger consensus was detected within furan bypass products. A highly over-represented 25 samples contained the context 50 G(N)RK (Table 2) and 19 of the latter contained the sequence 50 KSG(N)RK. Unlike the two overlapping AP bypass consensus sequences (see above), 50 G(N)RK might not be statistically independent of 50 KSG(N)RK. However, 50 G(N)RK was highly over-represented in holoenzyme exoÿ bypass products (Table 2), suggesting that it represented the core of the consensus required for ef®cient bypass of furan. Here again, it appears that the addition of the KS sequence on the 50 end further enhanced the bypass ef®ciency. Indeed, a closely related sequence 50 KGS(N)RK was over-represented in holoenzyme exoÿ furan bypass products. A very interesting phenomenon was observed when non-contiguous strings of furan bypass products by either form of the polymerase were searched. The sequence 50 GNK was over-represented in several frames on the 30 side or straddling the lesion (i.e. 50 (F)NGNK, 50 (F)GNK, 50 G(F)NK, see Table 3). Every trinucleotide permutation of this sequence (i.e. GAG, GCG, GGG, etc.) was statistically over-represented with the exception of 50 GTT in holoenzyme products and 50 GAT in holoenzyme exoÿ products. Indeed, the latter trinucleotide strings were signi®cantly under-represented in almost every frame where the other permutations of GNK were over-represented. The reverse of this consensus, 50 KNG, was also overrepresented on the 50 side of furan in bypass products of both enzymes, and again, the sequences 50 TTG(N) and 50 TAG(N) were under-represented in holoenzyme and holoenzyme exoÿ AP bypass products, respectively. In all, 69 % of the holoenzyme bypass products contained the 50 GNK consensus in at least one frame and 36 % of the products con-

1050

Bypass of AP Sites by T4 DNA Pol Holoenzyme

Table 3. Over-representation of the context 5' GNK in furan bypass products Consensus Furan holoenzyme (n ˆ 91)

Furan holoenzyme exoÿ (n ˆ 112)

O

5' (F)NGNK 5' (F)GNK 5' G(F)NK 5' SN(F)G 5'KVG(F) 5' (F)GNK 5' G(F)NK 5'KNG(F)

22 26 28 32 28 29 29 24

E

w2

11.4 11.4 11.4 11.4 8.5 14 14 14

9.92 18.8 24.3 37.4 44.43 16.07 16.07 7.14

P <0.005 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.01

Nucleotide nomenclature as described in the footnotes to Table 2; V ˆ A, C or G.

tained it in only one frame, while 55 % and 29 % of the holoenzyme exoÿ products contained 50 GNK in at least one or only one frame, respectively.

sequence 50 G(N)C was under-represented in holoenzyme exoÿ bypass products of both AP sites and furan.

Consensus sequences for inefficient bypass of AP sites and furan by T4 DNA polymerase holoenzyme

Influence of sequence context on the nucleotide incorporated opposite the lesion

Although this project was designed primarily to de®ne sequence contexts which enhanced lesion bypass, those which impeded bypass were inferred from the data by searching for statistically underrepresented sequences. The size of our sample pools limited the consensus sequences length to dinucleotides, i.e. in a pool of 100 random octanucleotides, any trinucleotide would be expected to appear in approximately two samples, while a dinucleotide would be expected to appear in approximately six samples. Therefore, a dinucleotide sequence which has an observed value of one is signi®cantly under-represented, while a trinucleotide sequence with an observed value of zero is not. In spite of this limitation, consensus sequence contexts for inef®cient bypass of both lesions were deduced and are presented in Table 4. Some correlation was observed between bypass products of the same lesion when either version of T4 DNA polymerase were used. For instance, 50 GC and 50 TG were under-represented in both holoenzyme and holoenzyme exoÿ AP bypass products, albeit at different frames relative to the lesion (Table 4). Similarly, none of the furan bypass products contained the sequence 50 (N)CC (part of the 50 (N)YC and 50 (N)CS consensus for holoenzyme and holoenzyme exoÿ, respectively). Finally, the

The choice of nucleotide incorporated opposite the lesion varied for all four enzyme/template combinations used (Table 1). In general, both versions of the polymerase inserted A opposite furan more frequently than opposite AP sites, and the proofreading de®cient polymerase inserted A more frequently opposite both lesions than the proofreading pro®cient polymerase, (Note that nucleotide frequencies reported in Table 1 are for template strands. Therefore, a frequency of 63 % T at the lesion position in holoenzyme AP bypass products indicates that the polymerase inserted A opposite the lesion 63 % of the time.) The second most frequent nucleotide inserted opposite AP sites was G, followed by T and then C, both in the presence and absence of proofreading. T and G were incorporated opposite furan in the presence of proofreading, albeit at a lower frequency than opposite AP sites, and A was exclusively incorporated opposite furan by Holoenzyme exoÿ. To address sequence context in¯uence on the choice of nucleotide inserted opposite the lesion, over-represented sequences were searched in bypass products with a nucleotide other than T at the lesion position (Table 5). The nearly exclusive incorporation of A opposite furan limited the search to AP bypass products. Out of 19 samples where the holoenzyme apparently inserted G

Table 4. Under-represented dinucleotide sequences in successful bypass products AP holoenzyme (n ˆ 99) AP holoenzyme exoÿ (n ˆ 87) Furan holoenzyme (n ˆ 91) Furan holoenzyme exoÿ (n ˆ 112)

Consensus

O

E

w2

5' (N)GC 5' (N)TG 5' G(N)C 5'TG(N) 5' (N)YC 5' W(N)C 5' (N)CS 5' S(N)C 5' (N)AC 5' A(N)T

2 2 1 1 2 3 1 2 1 1

6.2 6.2 5.4 5.4 11.4 11.4 14 14 7.0 7.0

2.83 2.83 3.62 3.62 7.73 6.17 12.07 10.28 5.14 5.14

Nucleotide nomenclature as described in the footnotes to Table 2; W ˆ A or T.

P <0.1 <0.1 <0.1 <0.1 <0.01 <0.025 <0.001 <0.005 <0.025 <0.025

1051

Bypass of AP Sites by T4 DNA Pol Holoenzyme

Table 5. Over-represented sequences in bypass products where a nucleotide other than A was inserted opposite the AP sites Sample sizea

Consensus

O

E

w2

9 7 5

1.2 0.3 0.8

51.40 155.2 21.58

<0.001 <0.001 <0.001

7

0.9

42.88

<0.001

Holoenzyme G inserted

19

T inserted

13

5' T(C)C 5'YNT(C)CNBY 5' R(A)CS

14

5'MNNM(C)NWNK

Holoenzyme exoÿ G Inserted

P

Nucleotide nomenclature as described in the legend to Table 1, B ˆ C, G or T. a Sample size re¯ects the total number of molecules where the indicated nucleotide was inserted opposite the lesion.

opposite AP, nine contained the sequence 50 T(C)C on the template strand, and seven of those contained the sequence 50 YNT(C)CNBY. The limited sample size made it dif®cult to conclude whether or not the shorter sequence was suf®cient to direct incorporation of G opposite the AP site or if the longer sequence was needed. However, both sequences were highly signi®cant relative to the number of samples with C at the lesion site. Moreover, 50 T(N)C (i.e. T and C as the lesion's two nearest neighbors regardless of the nucleotide recovered at the position of the lesion) appeared only 17 times in 99 bypass products, and the sequence 50 YNT(N)CNBY appeared 11 times, indicating that in each case where they appeared, G was preferentially inserted opposite the lesion (9/ 17 and 7/11 are signi®cantly over-represented, w2 5.3 and 6.6, respectively). In samples where the polymerase inserted T opposite the lesion, ®ve out of 13 shared the sequence 50 R(A)CS. Here again, the sequence 50 R(N)CS (i.e. regardless of the nucleotide recovered at the position of the lesion) appeared only seven times in the total pool of holoenzyme AP bypass products, suggesting that it directed the incorporation of T opposite the lesion. Finally, in holoenzyme exoÿ AP bypass pro-

ducts, the sequence 50 MNNM(C)NWNK (seven samples) was over-represented in products where G was incorporated opposite the lesion (14 samples) and G was inserted opposite the lesion at high frequency in all samples containing this sequence (7/14). Verification of the in vitro selection scheme To demonstrate the effectiveness of the selection assay as well as our analysis of the data, we chose several representative sequence contexts which were predicted to either enhance or impede lesion bypass (Figure 3). We then synthesized oligonucleotide templates containing AP sites or furan in these contexts and determined the steady-state kinetic constants of single nucleotide incorporation opposite the lesion using wild-type T4 DNA polymerase. The oligonucleotide templates were identical except for the speci®c sequence contexts under question (Figure 3). The data are summarized in Table 6 and two representative Michaelis-Menten plots are presented in Figure 4. The ®rst comparison was made between contexts which were predicted to enhance lesion bypass (50 TAT(AP)CGCT and 50 (AP)AGT satisfying the consensus 50

Figure 3. Sequences contexts predicted by in vitro selection to effect the ef®ciency of lesion bypass, and veri®ed by steady-state kinetic analysis.

1052

Bypass of AP Sites by T4 DNA Pol Holoenzyme

Table 6. Steady-state kinetics of nucleotide incorporation opposite AP sites and furan by T4 DNA polymerase in selected sequence contexts Sequence context

dNTP incorporated

Range[dNTP] (mM)

Km (mMdNTP)

Vmax (nMDNA/minute)

Vmax/Km (nMDNA/minute mMdNTP)

5'YNT(AP)CNBY

dATP dGTP dATP dGTP dATP dATP dATP

0.047-2.0 0.125-6.0 0.047-2.0 0.093-8.0 0.0234-10.0 0.047-2.0 0.234-10.0

0.269(0.025) 0.234(0.039) 0.124(0.033) NDa 6.299(1.19) 0.336(0.025) 1.743(0.38)

3.8(0.114) 0.47(0.023) 2.7(0.192) NDa 1.1(0.112) 6.2(0.161) 4.7(0.625)

14.1 2.0 21.8

5' 5' 5' 5' a

(AP)MGY G(AP)C G(F)RK (F)YC

0.17 18.5 2.7

Not detected.

YNT(AP)CNBY and 50 (AP)MGY, respectively) and a context predicted to impede lesion bypass (50 G(AP)C, Figure 3). Con®rming our predictions, Vmax/Km values for DNA synthesis on the 50 YNT(AP)CNBY and 50 (AP)MGY templates were, respectively, 83 and 128-fold higher than those for the 50 G(AP)C template (Table 6). The higher ef®ciency was primarily derived from lower Km values. Next, we examined the prediction that the context 50 YNT(AP)CNBY would direct the incorporation of G opposite the lesion at a higher ef®ciency than other bypass contexts. Here again, our

predictions were borne out by the data. Although the context 50 (AP)MGY supported the incorporation of A opposite the lesion with a slightly higher ef®ciency than the context 50 YNT(AP)CNBY, no incorporation of G could be detected in the former sequence even at extended reaction times (20 minutes) and in the presence of high concentrations of dGTP (up to 8.0 mM). Finally, we compared the ef®ciency of incorporating A opposite furan in the contexts 50 KSG(F)RK and 50 (N)YC which were predicted to support higher and lower bypass ef®ciency, respectively. Indeed, a seven-fold higher

Figure 4(a) (legend opposite)

Bypass of AP Sites by T4 DNA Pol Holoenzyme

1053

Figure 4. Steady-state kinetics of single nucleotide incorporation opposite an AP site in the context 50 YNT(AP)CNBY. (a) dATP incorporation. (b) dGTP incorporation. Reaction initial rates (insets) were determined at the indicated dNTP concentrations, plotted (main graphs) as a function of these concentrations, followed by non-linear least-square ®tting to a hyperbolic function (see Materials and Methods). At concentrations between 3.0 and 8.0 mM dGTP, reaction poisoning was observed proportionately to the nucleotide concentration (data not shown).

reaction ef®ciency was observed in the former sequence as compared to the latter in spite of fact that the two templates were identical on the 50 side of the lesion (Figure 3).

Discussion A detailed model of the mechanism of natural DNA synthesis which is based on recent kinetic and crystallographic studies of several polymerases (Johnson, 1993; Pelletier et al., 1996; Sawaya et al., 1997; Wang et al., 1997; Kiefer et al., 1997) demonstrates that DNA polymerase ®delity is equally the result of accurate incorporation of the correct nucleotide and very slow extension of mispaired nucleotides. This model can be adapted, with minor modi®cation, to synthesis on lesion-containing templates. First, the chemical structure of a lesion (and therefore its pairing properties) is predicted to be the major determinant of choosing the ``correct'' incoming nucleotide. On the other hand, it is strongly suggested here and in previous stu-

dies (Evans et al., 1993; Maccabee et al., 1994; Cai et al., 1993; Efrati et al., 1997; Randall et al., 1987; Goodman, 1997; Miller & Grollman, 1997) that stacking interactions between the lesion and neighboring nucleotides on both the 30 and 50 sides modulate the presentation of the lesion such that the ``correct pair'' for the same lesion may be different in different sequence contexts. Secondly, incorporation of the best ®tting nucleotide opposite the lesion might result in a poorly extendible primer terminus, while incorporation of a poorer ®tting nucleotide opposite the lesion might produce an extendible primer terminus as has already been demonstrated for 8-oxoguanine (Shibutani et al., 1991). Several predictions can be made based on the modi®ed model. Common sequence contexts should emerge within pools of bypass products of the same lesion and polymerase. Similarly, AP sites and furan bypass products should display common sequence contexts (if the two lesions were structurally equivalent). However, common

1054 sequence contexts would not necessarily be found between bypass products of the same lesion in the presence and absence of proofreading. For example, a particular context might support the incorporation of A opposite AP sites at a much higher ef®ciency than G, but the primer terminus resulting from G misincorporation might be extended at a higher frequency than the primer terminating with A. In the absence of proofreading, the higher probability of incorporating A would lock the product in an non-extendible status and full lesion bypass would fail. In the presence of proofreading, on the other hand, cycling between incorporation and editing of the terminal nucleotide would enrich for events where the extendible nucleotide is incorporated and therefore enrich for full bypass events. The prediction of common sequence contexts between bypass products of the same lesion was largely borne out by the data (Tables 2 and 3). How the neighboring nucleotides on the 30 side and nucleotides straddling the lesion modulated lesion presentation can be readily explained by invoking the in¯uence of base stacking interactions on lesion presentation. This argument lends less support to consensus sequences found entirely on the 50 side of the lesion. We suggest two explanations to resolve this issue. It is possible that the 50 sequences enhanced bypass in molecules where the best ®tting nucleotide opposite the lesion resulted in a primer terminus blocked at one or two nucleotides to the 50 side of the lesion. We and others have observed that extension of the nucleotide opposite the lesion proceeded at a much lower kinetic rate than DNA synthesis on undamaged templates (data not shown; Miller & Grollman, 1997). An alternative explanation is that the 50 nucleotides may alter the presentation of the lesion in the tight binding polymearse-dNTP-template ternary complex proposed in the model of natural DNA synthesis (Johnson, 1993). Studies showing that T4 DNA polymerase covers a minimum of seven nucleotides to the 30 of the primer terminus (Gopalakrishnan & Benkovic, 1994) lend support to this suggestion. The furan bypass consensus sequence 50 GNK which seems to modulate lesion bypass when present in several frames relative to the lesion, and particularly in the reverse orientation when on the 50 side of the lesion (i.e. 50 KNG(F)), lends further support to the latter suggestion. In other words, a particular sequence context might create a micro DNA structure in which a lesion's presentation approximates a natural base within the con®nes of a polymerase active site. Conversely, other sequence contexts are predicted to alter lesion presentation such as to exaggerate the difference between the lesion and a natural base and would be under-represented in bypass products. Although considered in many previous studies as close structural analogs (Cai et al., 1993; Randall et al., 1987; Kalnik et al., 1988; Pinz et al., 1995; Takeshita et al., 1987), AP sites and furan

Bypass of AP Sites by T4 DNA Pol Holoenzyme

(Figure 1) were clearly structurally different from the perspective of the DNA polymerase. No homology was found between bypass products of AP and furan, with the exception of the under-represented sequence (50 G(lesion)C) when holoenzyme exoÿ was used (Table 4). Bypass products of the two lesions varied greatly in terms of over and under-represented sequences, as well as nucleotides incorporated opposite the lesion. Large variations in the frequency of C and G at the 30 nearest neighbor of AP and furan in holoenzyme bypass products (Table 1) suggest a major difference in the way the two lesions in¯uence stacking interactions with bases at that position. Indeed, in the presence of proofreading, the sequence 50 T(N)C was over-represented in AP bypass products (Table 2) and under-represented in furan bypass products (part of 50 W(N)C, Table 4). The incorporation of G opposite AP sites at relatively high frequency (compared to furan) is in agreement with NMR studies showing both A and G to stack into a B-conformation helix opposite AP sites (Withka et al., 1991). On the other hand, the higher frequency of incorporating A opposite furan in the presence and absence of proofreading is in agreement with NMR studies showing A and G opposite furan to be predominantly intrahelical at low temperature, and G to be in a melted state at higher temperatures (yet well before denaturation of the helix; Cuniasse et al., 1990). It should be noted that AP sites used in this study were not reduced and, therefore, the deoxyribose should have been in equilibrium (Figure 1) between a furanoside and its open form (Bayley et al., 1961). The observed differences in the frequency of nucleotides inserted opposite an AP site and furan might be attributed either to the hydroxyl group at C1, or to the presence of an open ribose form. Finally, the model predicts that differences in the structures of polymerase active sites should lead to some differences in sequence contexts which enhance bypass of the same lesion when different DNA polymerases are used. For instance, the consensus sequence for bypass of furan presented here does not correlate with data presented for nearest neighbor effects on the bypass of furan by HIV reverse transcriptase (Cai et al., 1993), while it does correlate with similar studies using Drosophila DNA polymerase a (Randall et al., 1987). However, the latter studies were limited to nearest neighbor effects on one side of the lesion, precluding extensive comparisons with our data. Another contrasting observation was the small number of samples which suggested AP bypass through misalignment in the pool of products not containing deletions. Although the over-represented sequence 50 T(N)CGY was consistent with bypass through misalignment, the apparent preference for inserting G opposite AP sites within this sequence argues against such a mechanism.

1055

Bypass of AP Sites by T4 DNA Pol Holoenzyme

Materials and Methods Chemicals and enzymes dNTPs, ddNTPs, heparin, Sequenase v.2, T4 polynucleotide kinase, T7 gene 6 exonuclease, and restriction endonuclease EcoRV were from US Biochemical/Amersham; Taq DNA polymerase was from Boehringer Mannheim; uracil DNA N-glycosylase was from Epicentre Biotechnologies; 8 M urea PAGE was performed using Sequegel components from National Diagnostics; [g-32P]ATP was from DuPont-NEN; tetrahydrofuran phosphoramidite was prepared as described (Ide et al., 1992). Wild-type T4 gp43, gp43 de®cient in the 30 ! 50 proofreading exonuclease activity as a result of two amino acid substitutions, D112A ‡ E114A, T4 gp44, gp62, gp45, and gp32 were prepared as described (RehaKrantz & Nonay, 1993; Bittner et al., 1979; Morris et al., 1979). Oligonucleotides and DNA Two oligonucleotides (52mers) containing either a deoxyuridine or tetrahydrofuran surrounded by four randomized nucleotides on both the 30 and 50 sides were synthesized by standard phosphoramidite chemistry and used as templates for DNA synthesis (see below for the sequence). Deoxyuridine was converted to AP sites by treatment of the oligonucleotide with a molar excess of uracil DNA N-glycosylase in 10 mM Tris-HCl (pH 7.5), 50 mM NaCl at 37 C. To make sure that complete conversion of deoxyuridine to AP sites was achieved, a portion of the oligonucleotide was heated at 95 C for 30 minutes and then used as a template for PCR with primers straddling the lesion site. Since such heat treatment results in complete cleavage of AP sites, PCR should detect any dU-containing contaminant. Additional oligonucleotides were synthesized and used as primers for sequencing of the lesion bypass products. Cloning vector pBluescript SK‡ was obtained from Stratagene. In vitro selection reaction conditions A 5 pmol sample of the 52mer containing either an AP site or furan (marked X below) were annealed to a 50 end-labeled 16mer in 10 mM potassium phosphate, 16 mM Tris-HCl (pH 7.4), 65 mM NaCl, 1 mM 2-mercaptoethanol, 150 mM dithiothreiotol, 1 mM EDTA and 12 % (v/v) glycerol, resulting in the following primed template:

simultaneously to the DNA/holoenzyme complex, incubated for ®ve minutes at 37 C, followed by addition of Mg2‡ and incubated for 15 minutes. Under the latter conditions, no signi®cant primer extension was observed (data not shown). Subcloning and sequencing Primer strands which were extended beyond the lesion block by DNA polymerase, were puri®ed by 8 M urea PAGE. The primer was designed to give a full extension product of 46 nt (see sequence above), which was readily separable from the template strand (52 nt) and lesion-blocked products (25-26 nt). To identify the nucleotide sequence of individual bypass products, they were subcloned into pBluescript KS‡. A ligase and restriction endonuclease-free method of subcloning (Zhou & Hatahet, 1995) was used to avoid altering the composition of the selected molecules (i.e. eliminating molecules which contain restriction endonuclease recognition sites in the randomized sequence). Individual subclones were sequenced using the double-stranded DNA chain termination method and Sequenase v.2 following a protocol recommended by US Biochemical. Data analysis Sequences from successful bypass products were analyzed for common patterns using the relational database program Microsoft FoxPro. Infrequently, clones shared identical sequences across the entire randomized region. Since one cycle of colony plating was performed, it was not possible to determine whether duplicates were independent colonies or the result of colony contamination. As a result, duplicated sequences were entered only once during analysis. Sequences of the randomized region from individual clones were searched for all possible strings of two, three or four contiguous nucleotides, and statistically under and over-represented sequences, as determined by a w2 test and P values 40.05, were isolated. Strings of two and three nucleotides interrupted by (n) nucleotides were analyzed in a similar manner. T4 DNA polymerase steady-state kinetics Oligonucleotides (46mers, Figure 3) containing AP sites and furan in sequence contexts deduced from the selection assay were used as templates to determine the steady-state kinetics of single nucleotide incorporation

5'GACTGGTCTGCAGAGTNNNNXNNNNTGACAGTCTGGAATTCGGAGCTTTTTT 3' CAGACCTTAAGCCTCG 5' To assemble the holoenzyme complex, 100 mM dNTPs, 1 mM ATP (®nal concentrations), 140 pmol gp44/62, and 100 pmol gp45 were added to the primed DNA and incubated for three minutes at 37 C. A 10 pmol sample of gp43 pro®cient or de®cient in proofreading was then added. Reactions were initiated by addition of 10 mM MgCl2 (®nal) and 2.5 nmol of heparin in a total volume of 65 ml, incubated for 15 minutes at 37 C, and terminated by the addition of an equal volume of formamide. Under these reaction conditions, approximately 2-4 % of the primer termini blocked by the lesion (25mer) were extended to the end of the template (46mers) as analyzed by a BioRad Molecular Imager GS250. The effectiveness of the heparin trap was veri®ed during reactions in which heparin and gp43 were added

opposite the lesion by wild-type gp43. Each template was primed with a 50 32P-labeled 16mer terminating one nucleotide 30 to the lesion (relative to the template strand). Single nucleotide incorporation rates were determined as a function of dATP or dGTP concentration at 22 C. A 2 nM sample of primed template in 25 mM Trisacetate (pH 7.5), 150 mM potassium acetate, 10 mM 2mercaptoethanol (all ®nal concentrations), and increasing concentrations of single dATP or dGTP, were mixed with 10 nM gp43 and 10 mM magnesium acetate in a total volume of 20 ml. Four samples of 5 ml each were removed at ®ve second intervals and the reactions terminated in 95 % formamide. Following product analysis by denaturing PAGE and quanti®cation by molecular imaging, the reactions' initial rates were determined and

1056 plotted as a function of substrate concentration. The resulting curve was ®tted to a hyperbolic function using the program SigmaPlot and the values of Km and Vmax were derived.

Acknowledgements We wish to thank our colleagues Dr Robert Melamede, and Dr Wah Kow for stimulating discussions; Dr Lynn Ripley for graciously reviewing the manuscript and providing valuable comments; and Ms Pam Vacek for assistance in the statistical analyses. This project was supported by NIH grants CA52040 (to S.S.W.) and CA72778 (to Z.H.).

References Ames, B. N. (1989a). Endogenous DNA damage as related to cancer and aging. Mutation Res. 214, 4146. Ames, B. N. (1989b). Endogenous oxidative DNA damage, aging, and cancer. Free Radical Res. Commun. 7, 121-128. Basu, A. K., Loechler, E. L., Leadon, S. A. & Essigmann, J. M. (1989). Genetic effects of thymine glycol: sitespeci®c mutagenesis and molecular modeling studies. Proc. Natl Acad. Sci. USA, 86, 7677-7681. Bayley, C. R., Brammer, K. W. & Jones, A. C. (1961). J. Chem. Soc., 1903-1905. Belguise-Valladier, P. & Fuchs, R. P. (1995). N-2-Amino¯uorene and N-2-acetylamino¯uorene adducts: the local sequence context of an adduct and its chemical structure determine its replication properties. J. Mol. Biol. 249, 903-913. Bittner, M., Burke, R. L. & Alberts, B. M. (1979). Puri®cation of the T4 gene 32 protein free from detectable deoxyribonuclease activities. J. Biol. Chem. 254, 9565-9572. Braithwaite, D. K. & Ito, J. (1993). Compilation, alignment, and phylogenetic relationships of DNA polymerases. Nucl. Acids Res. 21, 787-802. Cai, H., Bloom, L. B., Eritja, R. & Goodman, M. F. (1993). Kinetics of deoxyribonucleotide insertion and extension at abasic template lesions in different sequence contexts using HIV-1 reverse transcriptase. J. Biol. Chem. 268, 23567-23572. Clark, J. M. & Beardsley, G. P. (1987). Functional effects of cis-thymine glycol lesions on DNA synthesis in vitro. Biochemistry, 26, 5398-5403. Cuniasse, P., Fazakerley, G. V., Guschlbauer, W., Kaplan, B. E. & Sowers, L. C. (1990). The abasic site as a challenge to DNA polymerase. A nuclear magnetic resonance study of G, C and T opposite a model abasic site. J. Mol. Biol. 213, 303-314. Efrati, E., Tocco, G., Eritja, R., Wilson, S. H. & Goodman, M. F. (1997). Abasic translesion synthesis by DNA polymerase beta violates the ``A-rule''. Novel types of nucleotide incorporation by human DNA polymerase beta at an abasic lesion in different sequence contexts. J. Biol. Chem. 272, 2559-2569. Evans, J., Maccabee, M., Hatahet, Z., Courcelle, J., Bockrath, R., Ide, H. & Wallace, S. (1993). Thymine ring saturation and fragmentation products: lesion bypass, misinsertion and implications for mutagenesis. Mutat. Res. 299, 147-156.

Bypass of AP Sites by T4 DNA Pol Holoenzyme Goodman, M. F. (1997). Hydrogen bonding revisited: geometric selection as a principal determinant of DNA replication ®delity. Proc. Natl Acad. Sci. USA, 94, 10493-10495. Gopalakrishnan, V. & Benkovic, S. J. (1994). Spatial relationship between polymerase and exonuclease active sites of phage T4 DNA polymerase enzyme. J. Biol. Chem. 269, 21123-21126. Hatahet, Z. & Wallace, S. S. (1997). Translesion DNA Synthesis. In DNA Damage and Repair, DNA Repair in Prokaryotes and Lower Eukaryotes (Nickoloff, J. A. & Hoekstra, M. F., eds), vol. 1, p. 229, Humana Press Inc., Totowa, NJ. Hatahet, Z., Zhou, M., Reha-Krantz, L. J., Morrical, S. W. & Wallace, S. S. (1998). In search of a mutational hotspot. Proc. Natl Acad. Sci. USA, 95, 8556-8561. Hayes, R. C. & LeClerc, J. E. (1986). Sequence dependence for bypass of thymine glycols in DNA by DNA polymerase I. Nucl. Acids Res. 14, 1045-1061. Ide, H., Kow, Y. W. & Wallace, S. S. (1985). Thymine glycols and urea residues in M13 DNA constitute replicative blocks in vitro. Nucl. Acids Res. 13, 80358052. Ide, H., Petrullo, L. A., Hatahet, Z. & Wallace, S. S. (1991). Processing of DNA base damage by DNA polymerases. Dihydrothymine and beta-ureidoisobutyric acid as models for instructive and noninstructive lesions. J. Biol. Chem. 266, 1469-1477. Ide, H., Murayama, H., Murakami, A., Morii, T. & Makino, K. (1992). Effects of base damages on DNA replication-mechanism of preferential purine nucleotide insertion opposite abasic site in template DNA. Nucl. Acids Symp. Ser. 167-168. Johnson, K. A. (1993). Conformational coupling in DNA polymerase ®delity. Annu. Rev. Biochem. 62, 685-713. Kalnik, M. W., Chang, C. N., Grollman, A. P. & Patel, D. J. (1988). NMR studies of abasic sites in DNA duplexes: deoxyadenosine stacks into the helix opposite the cyclic analogue of 2-deoxyribose. Biochemistry, 27, 924-931. Kiefer, J. R., Mao, C., Hansen, C. J., Basehore, S. L., Hogrefe, H. H., Braman, J. C. & Beese, L. S. (1997). Crystal structure of a thermostable Bacillus DNA Ê resolution. polymerase I large fragment at 2.1 A Structure, 5, 95-108. Kunkel, T. A. (1990). Misalignment-mediated DNA synthesis errors. Biochemistry, 29, 8003-8011. Lindahl, T. (1993). Instability and decay of the primary structure of DNA. Nature, 362, 709-715. Lindahl, T. & Nyberg, B. (1972). Rate of depurination of native deoxyribonucleic acid. Biochemistry, 11, 36103618. Liu, J. & Doetsch, P. W. (1998). Escherichia coli RNA and DNA polymerase bypass of dihydrouracil: mutagenic potential via transcription and replication. Nucl. Acids Res. 26, 1707-1712. Maccabee, M., Evans, J. S., Glackin, M. P., Hatahet, Z. & Wallace, S. S. (1994). Pyrimidine ring fragmentation products. Effects of lesion structure and sequence context on mutagenesis. J. Mol. Biol. 236, 514-530. Miller, H. & Grollman, A. P. (1997). Kinetics of DNA polymerase I (Klenow fragment exo-) activity on damaged DNA templates: effect of proximal and distal template damage on DNA synthesis. Biochemistry, 36, 15336-15342. Morris, C.-F., Hama-Inaba, H., Mace, D., Sinha, N. K. & Alberts, B. (1979). Puri®cation of the gene 43, 44, 45 and 62 proteins of the bacteriophage T4 DNA replication apparatus. J. Biol. Chem. 254, 6787-6796.

Bypass of AP Sites by T4 DNA Pol Holoenzyme Mozzherin, D. J., Shibutani, S., Tan, C. K., Downey, K. M. & Fisher, P. A. (1997). Proliferating cell nuclear antigen promotes DNA synthesis past template lesions by mammalian DNA polymerase delta. Proc. Natl Acad. Sci. USA, 94, 6126-6131. Paz-Elizur, T., Takeshita, M., Goodman, M., O'Donnell, M. & Livneh, Z. (1996). Mechanism of translesion DNA synthesis by DNA polymerase II. Comparison to DNA polymerases I and III core. J. Biol. Chem. 271, 24662-24669. Paz-Elizur, T., Takeshita, M. & Livneh, Z. (1997). Mechanism of bypass synthesis through an abasic site analog by DNA polymerase I. Biochemistry, 36, 1766-1773. Pelletier, H., Sawaya, M. R., Wol¯e, W., Wilson, S. H. & Kraut, J. (1996). Crystal structures of human DNA polymerase beta complexed with DNA: implications for catalytic mechanism, processivity, and ®delity. Biochemistry, 35, 12742-12761. Pinz, K. G., Shibutani, S. & Bogenhagen, D. F. (1995). Action of mitochondrial DNA polymerase gamma at sites of base loss or oxidative damage. J. Biol. Chem. 270, 9202-9206. Purmal, A. A., Kow, Y. W. & Wallace, S. S. (1994). Major oxidative products of cytosine, 5-hydroxycytosine and 5-hydroxyuracil, exhibit sequence context-dependent mispairing in vitro. Nucl. Acids Res. 22, 72-78. Purmal, A. A., Lampman, G. W., Bond, J. P., Hatahet, Z. & Wallace, S. S. (1998). Enzymatic processing of uracil glycol, a major oxidative product of DNA cytosine. J. Biol. Chem. 273, 10026-10035. Rajagopalan, M., Lu, C., Woodgate, R., O'Donnell, M., Goodman, M. F. & Echols, H. (1992). Activity of the puri®ed mutagenesis proteins UmuC, UmuD0 , and RecA in replicative bypass of an abasic DNA lesion by DNA polymerase III. Proc. Natl Acad. Sci. USA, 89, 10777-10781. Randall, S. K., Eritja, R., Kaplan, B. E., Petruska, J. & Goodman, M. F. (1987). Nucleotide insertion kinetics opposite abasic lesions in DNA. J. Biol. Chem. 262, 6864-6870. Reha-Krantz, L. J. & Nonay, R. L. (1993). Genetic and biochemical studies of bacteriophage T4 DNA polymerase 30 ! 50 -exonuclease activity. J. Biol. Chem. 268, 27100-27108. Sagher, D. & Strauss, B. (1985). Abasic sites from cytosine as termination signals for DNA synthesis. Nucl. Acids Res. 13, 4285-4298. Sawaya, M. R., Prasad, R., Wilson, S. H., Kraut, J. & Pelletier, H. (1997). Crystal structures of human

1057 DNA polymerase beta complexed with gapped and nicked DNA: evidence for an induced ®t mechanism. Biochemistry, 36, 11205-11215. Shibutani, S., Takeshita, M. & Grollman, A. P. (1991). Insertion of speci®c bases during DNA synthesis past the oxidation-damaged base 8-oxodG. Nature, 349, 431-434. Shibutani, S., Bodepudi, V., Johnson, F. & Grollman, A. P. (1993). Translesional synthesis on DNA templates containing 8-oxo-7,8-dihydrodeoxyadenosine. Biochemistry, 32, 4615-4621. Shibutani, S., Takeshita, M. & Grollman, A. P. (1997). Translesional synthesis on DNA templates containing a single abasic site. A mechanistic study of the ``A rule''. J. Biol. Chem. 272, 13916-13922. Takeshita, M., Chang, C. N., Johnson, F., Will, S. & Grollman, A. P. (1987). Oligodeoxynucleotides containing synthetic abasic sites. Model substrates for DNA polymerases and apurinic/apyrimidinic endonucleases. J. Biol. Chem. 262, 10171-10179. Tudek, B., Boiteux, S. & Laval, J. (1992). Biological properties of imidazole ring-opened N7-methylguanine in M13mp18 phage DNA. Nucl. Acids Res. 20, 30793084. Wagner, J. R., Hu, C. C. & Ames, B. N. (1992). Endogenous oxidative damage of deoxycytidine in DNA. Proc. Natl Acad. Sci. USA, 89, 3380-3384. Wallace, S. S. (1988). AP endonucleases and DNA glycosylases that recognize oxidative DNA damage. Environ. Mol. Mutagen. 12, 431-477. Wallace, S. S. (1997). Oxidative Damage to DNA and Its Repair. In Oxidative Stress and the Molecular Biology of Antioxidant Defenses, p. 49, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Wang, J., Sattar, A. K., Wang, C. C., Karam, J. D., Konigsberg, W. H. & Steitz, T. A. (1997). Crystal structure of a pol alpha family replication DNA polymerase from bacteriophage RB69. Cell, 89, 10871099. Withka, J. M., Wilde, J. A., Bolton, P. H., Mazumder, A. & Gerlt, J. A. (1991). Characterization of conformational features of DNA heteroduplexes containing aldehydic abasic sites. Biochemistry, 30, 99319940. Young, M. C., Reddy, M. K. & von Hippel, P. H. (1992). Structure and function of the bacteriophage T4 DNA polymerase holoenzyme. Biochemistry, 31, 8675-8690. Zhou, M. & Hatahet, Z. (1995). An improved ligase-free method for directional subcloning of PCR ampli®ed DNA. Nucl. Acids Res. 23, 1089-1090.

Edited by J. M. Miller (Received 17 April 1998; received in revised form 5 November 1998; accepted 17 December 1998)