Mutagenesis by exocyclic alkylamino purine adducts in Escherichia coli

Mutation Research 599 (2006) 1–10

Mutagenesis by exocyclic alkylamino purine adducts in Escherichia coli Dana C. Upton a , Xueying Wang a , Patrick Blans b , Fred W. Perrino a , James C. Fishbein b,∗∗ , Steven A. Akman a,∗ a b

Wake Forest University Health Sciences, Winston-Salem, NC, United States University of Maryland – Baltimore County, Baltimore, MD, United States

Received 8 September 2005; received in revised form 1 December 2005; accepted 28 December 2005 Available online 20 February 2006

Abstract Exocyclic alkylamino purine adducts, including N2 -ethyldeoxyguanosine, N2 -isopropyldeoxyguanosine, and N6 -isopropyldeoxyadenosine, occur as a consequence of reactions of DNA with toxins such as the ethanol metabolite acetaldehyde, diisopropylnitrosamine, and diisopropyltriazene. However, there are few data addressing the biological consequences of these adducts when present in DNA. Therefore, we assessed the mutagenicities of these single, chemically synthesized exocyclic amino adducts when placed site-specifically in the supF gene in the reporter plasmid pLSX and replicated in Escherichia coli, comparing the mutagenic potential of these exocyclic amino adducts to that of O6 -ethyldeoxyguanosine. Inclusion of deoxyuridines on the strand complementary to the adducts at 5 and 3 flanking positions resulted in mutant fractions of N2 -ethyldeoxyguanosine and N2 isopropyldeoxyguanosine-containing plasmid of 1.4 ± 0.5% and 5.7 ± 2.5%, respectively, both of which were significantly greater than control plasmid containing deoxyuridines but no adduct (p = 0.04 and 0.003, respectively). The mutagenicities of the three exocyclic alkylamino purine adducts tested were of smaller magnitude than O6 -ethyldeoxyguanosine (mutant fraction = 21.2 ± 1.2%, p = 0.00001) with the N6 -isopropyldeoxyadenosine being the least mutagenic (mutant fraction = 1.2 ± 0.5%, p = 0.13). The mutation spectrum generated by the N2 -ethyl and -isopropyldeoxyguanosine adducts included adduct site-targeted G:C → T:A transversions, adduct site single base deletions, and single base deletions three bases downstream from the adduct, which contrasted sharply with the mutation spectrum generated by the O6 -ethyldeoxyguanosine lesion of 95% adduct site-targeted transitions. We conclude that N2 -ethyl and -isopropyldeoxyguanosine are mutagenic adducts in E. coli whose mutation spectra differ markedly from that of O6 -ethyldeoxyguanosine. © 2006 Elsevier B.V. All rights reserved. Keywords: Mutagenesis; Alkyl DNA adducts; Exocyclic alkylamino purines

1. Introduction

∗ Corresponding author. Present address: Department of Cancer Biology, 1 Medical Center Boulevard, Winston-Salem, NC 27157, United States. Tel.: +1 336 716 0231; fax: +1 336 716 0255. ∗∗ Corresponding author. Tel.: +1 410 455 2190. E-mail addresses: [email protected] (J.C. Fishbein), [email protected] (S.A. Akman).

0027-5107/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.mrfmmm.2005.12.014

A considerable body of data inform us with regard to the mutagenicity and mutation spectra associated with the presence of bulky adducts, e.g. (+)-trans-antibenzo[a]pyrene diol epoxide, located at the N2 position of deoxyguanosine (dG) in DNA [1,2]. In contrast, there are few data addressing the mutagenicity of N2 alkyldG adducts, e.g. N2 -methyl, -ethyl, -n-propyl, or

2

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

-isopropyldG. These dG adducts have not been well studied because it had been thought that the exocyclic amino group of deoxyguanosine is an unfavorable site for alkylation by common alkylating agents [3]. Similarly, the N6 exocyclic amino group of deoxyadenosine (dA) was also thought to be a poor site for alkylation, and consequently there are few data addressing the mutagenicity of N6 -alkyldA adducts. The formation of N2 -ethyldG in DNA has been proposed to arise through oxidation of ethanol to acetaldehyde and biological reduction of the aldehyde adduct [4–6]. N2 -ethyldG has been identified in the DNA of ethanol-treated mice [4] and in human alcoholics [7] and can be detected in the urine of human volunteers who have abstained from alcohol for at least 1 week [6]. The association between ethanol consumption and risk for development of aerodigestive cancer is well-

documented [8–11]. The significance of N2 -ethyldG in the development of alcohol-related cancers is unknown, as there are few data regarding the biological consequences of this adduct in DNA. Terashima et al. have studied the properties of template N2 -ethyldG in vitro in reactions catalyzed by exonuclease-free DNA polymerase I from Escherichia coli [12]. Template N2 ethyldG stalled DNA polymerase I at or one base prior to the lesion, suggesting that this adduct might block translesion DNA synthesis in E. coli in vivo. However, when DNA polymerase I-catalyzed insertion across and extension beyond N2 -ethyldG did occur, insertion of dG was favored, suggesting that N2 -ethyldG might cause G:C → C:G transversions in vivo. Similarly, we have shown that N2 -ethyldG strongly blocked translesion synthesis catalyzed by the mammalian replicative DNA polymerase ␣ [13] and Choi and Guengerich showed a similar blockage of DNA synthesis by the viral DNA

Fig. 1. Chemical structures of guanine, O6 -ethylguanine, N2 -ethylguanine, N2 -isopropylguanine, adenine, and N6 -isopropyladenine.

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

polymerases bacteriophage T7 exonuclease− and HIV-1 reverse transcriptase [14]. Recently, Blans and Fishbein have demonstrated that isopropyl diazonium ions, in contrast to methyl and ethyl diazonium ions, readily react with the exocyclic amino groups of purines [15,16]. This and other sec-diazonium ions are derived from a number of carcinogenic nitrosamines, including those found in tobacco smoke [17,18]. In this regard, Blans and Fishbein observed that N2 -isopropyldG and N6 isopropyldA (Fig. 1) are abundantly formed by reaction of the carcinogens diisopropyltriazene [15] or N-(1hydroxyethyl)isopropylnitrosamine [16] with dG and dA [15], and DNA [16]. Thus, it must be considered that N2 -isopropyldG and N6 -isopropyldA may play a role in determining the biological consequences of exposure to propylating carcinogens. The mutagenicity of isopropylating agents has not been well studied. However, rearrangement of n-propylDNA adducts to isopropyl adducts is not uncommon [19]; thus, an unknown, but possibly substantial, fraction of the mutations occurring after exposure of DNA to npropylating agents may be a consequence of isopropyl DNA adducts. Exposure of target DNA to propylating agents such as N-propyl-N -nitro-N-nitrosoguanidine [20] and N-propyl-N-nitrosourea [21] cause primarily G:C → A:T transitions in E. coli. The human Y-family translesion synthesis DNA polymerases ␩ and ␫ efficiently bypass the N2 -ethylG, N2 isopropyldG and N6 -isopropyldA lesions in DNA templates during in vitro DNA synthesis [13,22], whereas the template N2 -ethylG and N2 -isopropyldG strongly block DNA polymerase ␣-catalyzed translesional synthesis. The blockade of DNA synthesis in vitro catalyzed by the human replicative B-family DNA polymerase, and not the Y-family DNA polymerases might suggest that the N2 -ethyl and -isopropyldG adducts affect faithful DNA replication in vivo. In contrast, the template properties of N6 -isopropyldA are similar to the normal nucleotide dA with regard to DNA synthesis in vitro catalyzed by the replicative DNA polymerase ␣ and the translesion synthesis DNA polymerases ␩ and ␫ [22] suggesting a possible lower mutagenic potential for the N6 -isopropyldA adduct. Considering the occurrence and potential pathophysiological importance of N2 -ethyldG, as well as the observation of N2 -isopropyldG and N6 -isopropyldA formed in the presence of carcinogenic alkylating agents, we sought to determine whether these exocyclic alkylamino purine adducts inserted site-specifically in a mutation reporting shuttle vector are mutagenic DNA adducts. Our data indicate that the mutagenicity of N2 -isopropyldG

3

exceeds that of N2 -ethyldG, which, in turn, is slightly more than that of N6 -isopropyldA when present in DNA replicating in E. coli. However, the magnitude of the mutant fraction associated with these exocyclic alkylamino adducts is smaller than that associated with the mutagenic DNA adduct O6 -ethyldG. Furthermore, the mutation spectra of N2 -ethyl and -isopropyldG differ from that of O6 -ethyldG. Whereas O6 -ethyldG causes primarily G:C → A:T transitions in E. coli, N2 -ethyl and -isopropyldG cause G:C → T:A transversions and frameshift deletions. 2. Materials and methods 2.1. Synthesis of oligonucleotides containing exocyclic alkylamino DNA adducts The methods for synthesis, purification, and chemical analysis of the phosphoramidites of O6 -ethyldG, N2 -ethyldG, N2 isopropyldG, and N6 -isopropyldA, as well as those for the synthesis and purification of oligodeoxynucleotides containing these adducts have been previously published [13,22]. Purine base analysis of the oligodeoxynucleotides was performed by acid hydrolysis and quantitation of the natural and adducted bases by HPLC/MS by methods previously reported [13,22]. In all cases the ratio of adducted to natural base was within less than 4% of theoretical. MALDI-TOF MS analysis of each oligodeoxynucleotide gave a molecular ion with a mass within 0.005% of theoretical. 2.2. Construction of double-stranded adduct-containing oligodeoxynucleotides Oligodeoxynucleotides were 5 -phosphorylated using 200 pmols of oligodeoxynucleotide and 20 units of T4 DNA polynucleotide kinase in 60 mM Tris–HCl, pH 7.6, 15 mM MgCl2 , 7.5 mM dithiothreitol, 2.5% PEG-8000, and 5 mM ATP. After incubation for 2 h at 37 ◦ C, the reaction was terminated at 70 ◦ C for 10 min. Phosphorylated oligodeoxynucleotides were then annealed by incubation of 200 pmol of each oligodeoxynucleotide in buffer (1 M NaCl, 1 M Hepes) at 65 ◦ C for 15 min followed by a decrease of −1 ◦ C/min to 25 ◦ C. 2.3. Construction of adduct-containing plasmids All restriction enzymes used were obtained from New England Biolabs (Ipswich, MA) and were reacted in the supplied buffers. Mutation studies were carried out using plasmid pLSX (Fig. 2), a derivative of pZ189 [23]. pLSX contains ␤-lactamase for ampicillin resistance, the pBR322 origin of replication for replication in E. coli, and the supF tRNA gene. Double-stranded pLSX DNA was prepared in E. coli strain JM109 cultured in Luria-Bertani broth supplemented

4

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

Fig. 2. (A) Schematic of the reporting plasmid pLSX containing a single exocyclic amino purine adduct. pLSX has an 82 nucleotide “stuffer fragment” containing an XbaI site placed in the supF tRNA gene that is removed by cleavage with XhoI and BspEI. Double-stranded oligodeoxynucleotides containing a single exocyclic amino purine adduct are ligated into the XhoI-BspEI site in the plasmid, recreating the wild type supF gene with the adduct in place in the acceptor stem of the tRNA. (B) The sequences of the double-stranded oligodeoxynucleotides ligated into the XhoI–BspEI site of the supF tRNA gene are listed. X indicates the adducted deoxyguanosine and Y indicates the adducted deoxyadenosine.

with 75 ␮g/mL ampicillin. Plasmid DNA was isolated using the GenElute HP Plasmid Maxiprep Kit (Sigma–Aldrich, St. Louis, MO) according to manufacturer’s instructions. Plasmid DNA was then digested with 5 units/␮g of XhoI and BspEI, and the resultant doubly digested plasmid was separated from the 82 bp stuffer fragment and purified by agarose gel electrophoresis (QIAquick Gel Extraction Kit, Qiagen, Valencia, CA). Ten picomoles of double-stranded adduct-containing oligodeoxynucleotides and 1.5 pmols of gel-purified XhoI/BspEI-cleaved pLSX were ligated with 1600 units of

T4 DNA ligase at 16 ◦ C for overnight. The adduct-containing constructs were recovered by ethanol precipitation, and then used to transform E. coli strain MBM7070. 2.4. Transformation of reporter bacteria Thirty-five microliters of electrocompetent E. coli strain MBM7070 cells were transformed with 1 ␮L DNA in an 2 mm gapped electroporation cuvette pulsed at 2.5 kV, 2.5 ␮Fd, 200
. Transformed E. coli were recovered in SOC media (20 g/L tryptone, 5 g/L yeast extract, 0.5 g/L NaCl,

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

5

Medicine DNA Sequencing Laboratory, Winston-Salem, NC and Macrogen, Seoul, South Korea).

20 mM glucose, 10 mM MgCl2 , 2.5 mM KCl) and incubated at 37 ◦ C for 20 min. Cells were then plated onto SGalTM /LB agar plates (12 g/L agar, 10 g/L tryptone, 10 g/L NaCl, 5 g/L yeast extract, 0.5 g/L ferric ammonium citrate, 0.3 g/L 3,4-cyclohexenoesculetin-␤-d-galactopyranoside (S-GalTM ), 0.03 g/L isopropyl-1-thio-␤-d-galactoside (IPTG) supplemented with 75 ␮g/mL ampicillin. Colonies were scored as black (wild-type supF) or white (mutant supF). Mutant colonies were confirmed by secondary streaking. The mutant fraction was determined by the number of white colonies divided by the total number of colonies. Plasmid DNA was harvested from mutant colony minipreparations (QIAprep Spin Miniprep Kit, Qiagen, Valencia, CA), after which the supF gene was sequenced by automated DNA sequencing (Wake Forest University School of

3. Results 3.1. Mutagenicity of N2 -ethyldG, N2 -isopropyldG, and N6 -isopropyldA in double-stranded plasmids containing thymidine or deoxyuridine in the strand complementary to the adduct: comparison to O6 -ethyldG The mutagenicity of the exocyclic alkylamino purines N2 -ethyldG, N2 -isopropyldG, and N6 -isopropyldA was assessed when double-stranded oligodeoxynucleotides

Table 1 Mutant fractions associated with adducts Construct

Mutant colonies

Wild type colonies

Mutant fraction (%)

Mean mutant fraction ± S.D. (%)

p-Value

27 11 16

4973 2751 3946

0.5 0.4 0.4

0.4 ± 0.2

O6 -EthyldG

8 3 5

1457 699 2000

0.5 0.4 0.3

0.4 ± 0.1

0.33a

N2 -EthyldG

23 15 125

2384 2055 11640

1.0 0.7 1.1

0.9 ± 0.2

0.09a

N2 -IsopropyldG

30 36 12

2878 2377 2326

1.0 1.5 0.5

1.0 ± 0.5

0.11a

N6 -IsopropyldA

132 68 64

8340 6150 7702

1.6 1.1 0.83

1.2 ± 0.5

0.07a

0.9 0.7 0.3

0.6 ± 0.4

0.34a

dG

dG/dU in complementary strand

41 16 11

4526 2193 3389

O6 -EthyldG/dU in complementary strand

676 1318 349

3006 6458 1725

22.5 20.4 20.2

21.2 ± 1.2

0.00001b

N2 -EthyldG/dU in complementary strand

15 19 26

1251 854 2234

1.2 2.2 1.2

1.4 ± 0.5

0.04b

N2 -IsopropyldG/dU in complementary strand

117 145 137

1831 2251 2629

6.0 6.1 5.0

5.7 ± 2.5

0.003b

N6 -IsopropyldA/dU in complementary strand

35 14 4

2910 866 580

1.2 1.6 0.68

1.2 ± 0.5

0.13b

Mutant fractions induced by exocyclic alkylamino purine adducts in pLSX with/without dU in the strand complementary to the adduct. Each experiment was conducted with newly synthesized constructs. a Compared to control (dG) plasmid. b Compared to control (dG) plasmid containing dU on the complementary strand.

6

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

containing these single, chemically synthesized adducts were placed site-specifically in the supF gene of the reporter plasmid pLSX (Fig. 2A). Adducts were placed site-specifically in the double-stranded plasmid at a site in the acceptor stem of supF tRNA known to cause loss of tRNA function upon base substitution or deletion [23,24]. In these constructs, N2 -ethyldG, N2 isopropyldG, and N6 -isopropyldA did not cause statistically significant enhancements of the mutant fraction of pLSX replicated in E. coli (Table 1; Fig. 3). Additionally, the well known mutagenic DNA adduct O6 -ethyldG also did not cause a statistically significant enhancement of the mutant fraction (Table 1; Fig. 3). The potential mutagenicity of these adducts was also studied in double-stranded plasmids in which deoxyuridines were placed on the complementary strand at 5 and 3 positions flanking the adduct (Fig. 2B). The presence of deoxyuridine did not affect the mutant fraction of control (dG-containing) constructs replicated in E. coli (Table 1; Fig. 3) (p = 0.34). O6 -EthyldG enhanced the mutant fraction of constructs containing deoxyuridines in the complementary strand by 40-fold (Table 1; Fig. 3) (p = 0.00001). The exocyclic alkylamino adduct N2 -isopropyldG caused an 10-fold increase of the mutant fraction in comparison to the deoxyuridinecontaining control construct (Table 1; Fig. 3) (p = 0.003). N2 -EthyldG caused a much smaller, but still statistically significant, 2.5-fold increase in the mutant fraction above that of the deoxyuridine-containing control construct (Table 1; Fig. 3) (p = 0.04). However, N6 -isopropyldA caused only a marginal, statistically insignificant enhancement of the mutant fraction of deoxyuridine-containing constructs replicated in E. coli (p = 0.13).

Table 2 Mutation spectrum of O6 -ethyldG Mutation

Number

Total (%)

G:C → A:T at site of adduct −G at site of adduct Total

19

95

1

5

20

100.0

95% Confidence interval 73–>99 <1–27

Mutation spectrum in Escherichia coli associated with the presence of O6 -ethyldG in constructs containing deoxyuridine on the complementary strand.

3.2. Mutation spectra Mutants recovered after replication of adducted, deoxyuridine-containing constructs were sequenced. One hundred percent of 15 mutants recovered from deoxyuridine-containing constructs without adduct proved to be residual background plasmid pLSX in which the 82 bp stuffer fragment had not been cleaved out. The mutation spectrum induced by O6 -ethyldG is consistent with what has been reported for this adduct [25–30], principally comprised of adduct site-targeted transitions (Table 2). In contrast, adduct site-targeted transitions were not observed after replication of N2 ethyldG-containing plasmid. This adduct caused adduct site-targeted G:C → T:A transversions and single base deletions, as well as a substantial proportion of a single base deletion, −G at d(pGGG) located 3–5 nucleotides downstream of the adduct (Table 3). The mutation spectrum induced by N2 -isopropyldG in constructs containing deoxyuridine in the complementary strand included approximately equal proportions of adduct site-targeted

Fig. 3. Mutant fractions of plasmids containing single O6 -ethyldG, N2 -ethyldG, N2 -isopropyldG, or N6 -isopropyldA without (white bars) or with (gray bars) deoxyuridine in the strand complementary to the adduct.

D.C. Upton et al. / Mutation Research 599 (2006) 1–10 Table 3 Mutation spectrum of N2 -ethyldG Mutation

Number

Total (%)

95% Confidence interval

G:C → T:A at site of adduct −G at site of adduct Single base deletion downstream of adducta Single base insertion downstream of adductb Total

11

20

5–35

5 33

9 61

<1–24 46–76

5

9

<1–24

54

100.0

Mutation spectrum in Escherichia coli associated with the presence of N2 -ethyldG in constructs containing deoxyuridine on the complementary strand. a −G at d(pCXAAGGG). b +G at d(pCXAAGGG). Table 4 Mutation spectrum of N2 -isopropyldG Mutation G:C → T:A at site of adduct G:C → A:T at site of adduct −G at site of adduct Single base deletion downstream of adducta Otherb Total

Number

Total (%)

95% Confidence interval

8

25

7–43

4

13

<1–31

9 9

28 28

10–46 10–46

2

6

<1–24

32

100.0

Mutation spectrum in Escherichia coli associated with the presence of N2 -isopropyldG in constructs containing deoxyuridine on the complementary strand. a −G at d(pCXAAGGG) (6); −A at d(pCXAAGGG) (3). b G:C → A:T 3 bases downstream of adduct.

transversions, single base deletions and downstream deletions, with adduct site-targeted transitions being about two-fold less abundant (Table 4). 4. Discussion 4.1. Neither exocyclic alkylamino nor alkyloxy purine adducts were detectably mutagenic in double-stranded plasmid DNA Mutagenicity by N2 -ethyldG, N2 -isopropyldG, N6 isopropyldA, and O6 -ethyldG was not detectably different from dG when the adducts were placed in the supF tRNA mutation reporting gene of the double-stranded shuttle plasmid pLSX. The lack of observed muta-

7

genicity of well known mutagenic adducts O6 -ethyldG [25–30] and N-2-acetylaminofluorene (AAF) adducted DNA [31,32] in assay systems using double-stranded DNA plasmids is well documented [29,33] and has been interpreted as reflecting the adduct-induced arrest of replication of the adducted strand while the unadducted complementary strand is replicated normally. A similar explanation for the lack of observed adduct-mediated mutagenicity by the N2 -ethyldG and N2 -isopropyldG in the double-stranded plasmid construct used in our experiments is likely. This explanation is supported by the strong blocking effects observed during in vitro DNA synthesis of templates containing these two adducts during DNA synthesis reactions catalyzed by a variety of DNA polymerases [13,14,22,34]. Thus, it is possible that all of the exocyclic dG adducts tested in this study contribute to blocked replication of the adductcontaining strand in vivo, causing preferred replication of the non-adducted strand. However, we cannot exclude the possibility that effective adduct repair in this double-stranded DNA construct precludes detection of mutagenicity. It has been suggested that the high fidelity replication complex in E. coli might be assisted by the lower fidelity translesion synthesis DNA polymerases in the bypass of DNA adducts that hinder its progression at least during unidirectional replication of doublestranded plasmids [35]. Thus, blocked procession of the replicative DNA polymerases in E. coli by N2 -ethyldG and N2 -isopropyldG could stimulate one of the three SOS-controlled DNA polymerases (Pol II, IV and V) to engage in translesion synthesis. However, in the absence of SOS-induction, as was the case in the experiments performed here, minimal bypass of the lesion might be expected. For example, it has recently been shown that bypass of the AAF-dG adduct requires the induction of the SOS response and involves either Pol V or Pol II in an error-free or a frameshift pathway, respectively [36]. The E. coli translesion synthesis DNA polymerases II, IV and V have not yet been tested using the N2 -ethyldG and N2 -isopropyldG adducted DNA templates in vitro, but the related human Y-family DNA polymerases ␩ and ␫ do synthesize DNA past these lesions with efficiencies and accuracies opposite the lesion comparable to those observed for these polymerases opposite unadducted dG [13,22]. 4.2. Mutagenicity of N2 -ethyldG and N2 -isopropyldG in “gapped” duplexes Replacement of thymidines that flank the adducted position by deoxyuridines on the complementary strand

8

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

reveals the mutagenic potential of N2 -isopropyldG and N2 -ethyldG. The use of deoxyuridine-containing constructs was based on previous work of Pauly et al. [27], who demonstrated enhanced mutant fractions in constructs containing certain O6 -alkyldG adducts when such constructs contained nearby deoxyuridines in the complementary strand. These investigators inferred that the repair of the deoxyuridines creates a gap in the strand complementary to the adduct, thereby forcing DNA synthesis across the template adduct. Our data confirm the observations of Pauly et al. with regard to O6 -ethyldG; the mutagenicity of O6 -ethyldG in pLSX was markedly enhanced by the presence of nearby deoxyuridines in the complementary strand. N2 -ethyldG and N2 -isopropyldG are about one tenth and one quarter as potent mutagens, respectively, as O6 -ethyldG. Some genetic data using the gapped DNA duplexes suggests that the E. coli DNA polymerase III, and not DNA pols I or II, is responsible for copying adducted and unadducted gapped duplex DNA substrates in vivo [27]. However, what role the translesion bypass polymerases might play in assisting DNA polymerase III in adduct bypass in copying doublestranded gapped duplex DNAs, and perhaps more importantly in adducted gapped duplexes, remains to be determined. 4.3. The N2 -ethyldG and N2 -isopropyldG-mediated mutation spectra contrast with the O6 -ethyldG-mediated mutation spectrum Adduct site-targeted N2 -ethyldG-mediated mutations in E. coli consisted of G:C → T:A transversions and single base deletions; N2 -isopropyldG-mediated mutations were divided between transitions, transversions, and one- or two-base deletions. These mutation spectra contrast sharply with the mutation spectrum produced by O6 -ethyldG, which was comprised almost exclusively (95%) of adduct site-targeted transitions. The high prevalence of deletions caused by N2 -ethyldG and N2 -isopropyldG suggests that these adducts constitute strong blocks to DNA synthesis. The in vitro work of Terashima et al. [12] indicates that N2 -ethyldG stalls DNA polymerase I at or one base prior to the lesion, and when bypass of the N2 -ethyldG does occur by this polymerase misinsertion of dG is favored. In contrast, G:C → C:G transversions were not observed in our experiments after replication of either N2 -ethyldG or N2 -isopropyldG in E. coli, suggesting that DNA polymerase I is not likely the enzyme responsible for translesion synthesis across N2 -alkylamino purine adducts in vivo.

4.4. N2 -ethyldG and -isopropyldG caused a single base deletion to occur at the site of repair of dU on the complementary strand Interestingly, a high prevalence of −G deletions occurring at d(pGGG) located 3–5 nucleotides 3 of the N2 -ethyldG and -isopropyldG adducts was also observed in the deoxyuridine-containing constructs. This mutation did not result simply as a byproduct of repair of the deoxyuridines as it was not observed in deoxyuridinecontaining constructs that did not also contain N2 ethyldG or -isopropyldG adducts. The cause of this prevalent mutation is unclear. Repair DNA synthesis of the sequence d(pCCCUUCGAAGU) across from the adduct is initiated from the dC complementary to the 5 dG of d(pGGG). Thus, it is possible that the presence of the N2 -ethyldG or -isopropyldG causes structural distortion of the template strand, promoting slippage of the nascent strand during initiation of the repair DNA synthesis event. Attempts to initiate repair synthesis by the DNA polymerase III might be stalled by the N2 -ethyldG or -isopropyldG adducts leading to a futile insertion–excision cycle, as has been reported to occur by this enzyme at sites of AAF adducts [36]. 4.5. N6 -isopropyldA was not mutagenic in double-stranded plasmids with deoxyuridine in the complementary strand In contrast to N2 -isopropyldG, N6 -isopropyldAmediated mutagenesis was not detected in E. coli, even in constructs containing deoxyuridine in the complementary strand. The contrasting mutagenicity of N2 isopropyldG and N6 -isopropyldA suggest that these two isopropylamino purine adducts interact quite differently with DNA polymerases. The in vitro polymerization data of Perrino et al. [22] provide a precedent for differing DNA polymerase interactions by these two isopropyl adducts. At least with respect to the replicative human DNA polymerase ␣, template N2 -isopropyldG constitutes a strong block to replication, whereas template N6 -isopropyldA is efficiently and accurately bypassed. Our data suggest that N6 -isopropyldA is also accurately bypassed by the polymerase in E. coli that is engaged in gap filling. 4.6. Implication of these observations for human health The observations that N2 -ethyldG occurs in human DNA in vivo after alcohol consumption [4,7], and that

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

N2 -isopropyldG occurs in DNA exposed to carcinogenic nitrosamines and triazenes [15,16] suggest that these adducts may play a role in mediating human health hazards. In comparison to O6 -ethyldG, these N2 -alkyldG adducts are more weakly mutagenic in the site-specific assay employed in our studies. However, in contrast to O6 -ethyldG, their mutation spectra include a significant proportion of frameshift deletions. If N2 -ethyldG or N2 -isopropyldG is present during replication of coding portions of critical target genes, their propensity to cause frameshift deletions may encourage loss of function and/or rapid degradation of important proteins. Studies are currently underway to determine whether N2 -ethyldG or N2 -isopropyldG also induce frameshift deletions when replicated in human cells. Acknowledgments This work was supported by NIH grants R01 CA52881 and R01 CA88950. D.C.U. is supported by PHS grant ES-07331-02. References [1] D.M. Jerina, J.M. Sayer, S.K. Agarwal, H. Yagi, W. Levin, A.W. Wood, A.H. Conney, D. Preuss-Schwartz, W.N. Baird, M.A. Pigott, Reactivity and tumorigenicity of bay-region diol epoxides derived from polycyclic aromatic hydrocarbons, Adv. Exp. Med. Biol. 197 (1986) 11–30. [2] M. Moriya, S. Spiegel, A. Fernandes, S. Amin, T. Liu, N. Geacintov, A.P. Grollman, Fidelity of translesional synthesis past benzo[a]pyrene diol epoxide-2 -deoxyguanosine DNA adducts: marked effects of host cell, sequence context, and chirality, Biochemistry 35 (1996) 16646–16651. [3] A. Dipple, DNA adducts of chemical carcinogens, Carcinogenesis 16 (1995) 437–441. [4] J.L. Fang, C.E. Vaca, Development of a 32 P-postlabelling method for the analysis of adducts arising throughout the reaction of acetaldehyde with 2 -deoxyguanosine-3 -monophosphate and DNA, Carcinogenesis 16 (1995) 2177–2185. [5] C.E. Vaca, J.L. Fang, E.K. Schweda, Studies of the reaction of acetaldehyde with deoxynucleosides, Chem. Biol. Interact. 98 (1995) 51–67. [6] T. Matsuda, I. Terashima, Y. Matsumoto, H. Yabushita, S. Matsui, S. Shibutani, Effective utilization of N2 -ethyl-2 -deoxyguanosine triphosphate during DNA synthesis catalyzed by mammalian replicative DNA polymerases, Biochemistry 38 (1999) 929– 935. [7] J.L. Fang, C.E. Vaca, Detection of DNA adducts of acetaldehyde in peripheral white blood cells of alcohol abusers, Carcinogenesis 18 (1997) 627–632. [8] A.E. Rogers, M.W. Connor, Alcohol and cancer, Adv. Exp. Med. Biol. 206 (1986) 473–495. [9] Alcohol drinking. Epidemiological studies of cancer in humans, IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 44, 1988, pp. 153–250.

9

[10] R. Talamini, S. Franceschi, S. Barra, C. La Vecchia, The role of alcohol in oral and pharyngeal cancer in non-smokers, and of tobacco in non-drinkers, Int. J. Cancer 46 (1990) 391–393. [11] A.J. Garro, C.S. Lieber, Alcohol and cancer, Ann. Rev. Pharm. Tox. 30 (1990) 219–249. [12] I. Terashima, T. Matsuda, T.-W. Fang, N. Suzuki, J. Kobayashi, K. Kohda, S. Shibutani, Miscoding potential of the N2 -ethyl-d deoxyguanosine DNA adduct by the exonuclease-free Klenow fragment of Escherichia coli DNA polymerase I, Biochemistry 40 (2001) 4106–4114. [13] F.W. Perrino, P. Blans, S. Harvey, S. Gelhaus, C. McGrath, S.A. Akman, G.S. Jenkins, W.R. LaCourse, J.C. Fishbein, The N2 ethylguanine and the O6 -ethyl- and O6 -methylguanine lesions in DNA: contrasting responses from the bypass DNA polymerase ␩ and the replicative DNA polymerase ␣, Chem. Res. Toxicol. 16 (2003) 1616–1623. [14] Y.-J. Choi, F.P. Guengerich, Analysis of the effect of bulk at N2 alkylguanine DNA adducts on catalytic efficiency and fidelity of the processive DNA polymerases bacteriophage T7 exonucleaseand HIV-1 reverse transcriptase, J. Biol. Chem. 279 (2004) 19217–19229. [15] P. Blans, J.C. Fishbein, Predicted exocyclic amino group alkylation of 2 -deoxyadenosine and 2 -deoxyguanosine by the isopropyl cation, Chem. Res. Toxicol. 13 (2000) 431–435. [16] P. Blans, J.C. Fishbein, Determinants of selectivity in alkylation of nucleosides and DNA by secondary diazonium ions: evidence for, and consequences of, a preassociation mechanism, Chem. Res. Toxicol. 17 (2004) 1531–1539. [17] R.N. Loeppky, C.J. Michejda, 204th National Meeting of the American-Chemical-Society, Washington, DC, August 23–28, 1992. Preface, nitrosamines and related N-nitroso compounds R11–R12. 1994. [18] W. Lijinsky, Chemistry and Biology of N-Nitroso Compounds, Cambridge University Press, Cambridge, UK, 1992. [19] M.L. Todd, P.F. Schendel, Repair and mutagenesis in Escherichia coli K-12 after exposure to various alkyl-nitrosoguanidines, J. Bacteriol. 156 (1983) 6–12. [20] G.M. van der Vliet, M. Zielenska, M.W. Anderson, B.W. Glickman, The complexity of nitrosoguanidine mutagenesis increases with size: observations of the mutational specificity of N-propylN -nitro-N-nitrosoguanidine, Carcinogenesis 10 (1996) 949– 952. ´ [21] F. Ferrezuelo, M.-J. Prieto-Alamo, J. Jurado, C. Pueyo, Influence of DNA repair by (A)BC excinuclease and Ogt alkyltransferase on the distribution of mutations induced by n-propyl-N-nitrosourea in Escherichia coli, Env. Mol. Mutagenesis 31 (1998) 82–91. [22] F.W. Perrino, S. Harvey, P. Blans, S. Gelhaus, W.R. LaCourse, J.C. Fishbein, Polymerization past the N2 -isopropylguanine and the N6 -isopropyladenine DNA lesions with the translestion synthesis DNA polymerases ␩ and ␫ and the replication DNA polymerase ␣, Chem. Res. Toxicol. 18 (2005) 1451–1461. [23] M.M. Seidman, K. Dixon, A. Razzaque, R.J. Zagursky, M.L. Berman, A shuttle vector plasmid for studying carcinogeninduced point mutations in mammalian cells, Gene 38 (1985) 233–237. [24] K.A. Canella, M.M. Seidman, Mutation spectra in supF: approaches to elucidating sequence context effects, Mut. Res. 450 (2000) 61–73. [25] R.W. Chambers, Site-specific mutagenesis in cells with normal DNA repair systems: transitions produced from DNA carrying a single O6 -alkylguanine, Nucleic Acids Res. 19 (1991) 2485–2488.

10

D.C. Upton et al. / Mutation Research 599 (2006) 1–10

[26] R.W. Chambers, Site-directed mutagenesis in single cells: transitions produced by DNA carrying a single O6 -alkylguanine residue, Mut. Res. 299 (1993) 123–133. [27] G.T. Pauly, S.H. Hughes, R.C. Moschel, Mutagenesis in Escherichia coli by three O6 -substituted guanines in doublestranded or gapped plasmids, Biochemistry 34 (1995) 8924–8930. [28] G.T. Pauly, S.H. Hughes, R.C. Moschel, Comparison of mutagenesis by O6 -methyl and O6 -ethylguanine and O4 -methylthymine in Escherichia coli using double-stranded and gapped plasmids, Carcinogenesis 19 (1998) 457–461. [29] G.T. Pauly, S.H. Hughes, R.C. Moschel, Response of repaircompetent and repair-deficient Escherichia coli to three O6 substituted guanines and involvement of methyl-directed mismatch repair in the processing of O6 -methylguanine residues, Biochemistry 33 (1994) 9169–9177. [30] L. Samson, J. Thomale, M.F. Rajewsky, Alternative pathways for the in vivo repair of O6 -alkylguanine and O4 -alkylthymine in Escherichia coli: the adaptive response and nucleotide excision repair, EMBO J. 7 (1988) 2261–2267. [31] N. Koffel-Schwartz, J.M. Verdier, M. Bichara, A.M. Freund, M.P. Daune, R.P. Fuchs, Carcinogen-induced mutation spec-

[32]

[33]

[34]

[35]

[36]

trum in wild-type, uvrA and umuC strains of Escherichia coli. Strain specificity and mutation-prone sequences, J. Mol. Biol. 177 (1984) 33–51. S. Shibutani, Mutagenic specificity of acetylaminofluorenederived DNA adducts in vitro, IARC Sci. Publ. 125 (1994) 457–462. N. Koffel-Schwartz, G. Maenhaut-Michel, R.P. Fuchs, Specific strand loss in N-2-acetylaminofluorene-modified DNA, J. Mol. Biol. 193 (1987) 651–659. I. Terashima, T. Matsuda, T.-W. Fang, N. Suzuki, J. Kobayashi, K. Kohda, S. Shibutani, Miscoding potential of the N2 -ethyl-2 deoxyguanosine DNA adduct by the exonuclease-free Klenow fragment of Escherichia coli DNA polymerase I, Biochemistry 40 (2001) 4106–4114. J. Wagner, H. Etienne, R. Janel-Bintz, R.P. Fuchs, Genetics of mutagenesis in E. coli: various combinations of translesion polymerases (Pol II, IV and V) deal with lesion/sequence context diversity, DNA Repair (Amst.) 1 (2002) 159–167. V. Pages, R. Janel-Bintz, R.P. Fuchs, Pol III proofreading activity prevents lesion bypass as evidenced by its molecular signature within E. coli cells, J. Mol. Biol. 352 (2005) 501–509.