Analysis at the sequence level of mutations induced by the ultimate carcinogen N-acetoxy-N-2-acetylaminofluorene.

The covalent binding of an ultimate carcinogen to the DNA bases or phosphate groups creates a premutational lesion that in vivo is processed by the repair, replication and recombination enzymes, and eventually may be converted into a mutation. Being interested in the way that an initial premutational event is converted into a stable heritable mutation, we have sequenced stable mutations in a gene that has formed covalent adducts in vitro with N-acetoxy-N-2-acetylaminofluorene (N-AcO-AAF, a model for the ultimate metabolite of the rat liver carcinogen 2-acetylaminofluorene, AAF). In vivo studies have shown the mutagenicity of AAF and its derivatives in both bacterial and eukaryotic systems. N-AcO-AAF reacts in vitro with DNA leading mainly to the formation of a guanine adduct, N-2-(deoxyguanosin-8-yl)-acetylaminofluorene (80%) and to at least three minor adducts. Studies by our group showed that binding of N-AcO-AAF to DNA resulted in a local distortion of the DNA helix around the C-8 adduct (the insertion-denaturation model). We describe here the analysis of forward mutations induced in the tetracycline-resistance gene of pBR322 by directing the chemical reaction of the carcinogen to a small restriction fragment (BamHI-SalI) inside the antibiotic-resistance gene. Mutants are selected for ampicillin (Ap) resistance and tetracycline (Tc) sensitivity. The plasmid DNA of such mutants was analyzed for sequence changes in the fragment where the AAF binding had been directed. We show here that the mutations are mainly frameshifts involving GC base pairs and that certain base pairs (hotspots) are affected at high frequencies.


Introduction
An important step in the carcinogenic process is thought to be the initial attack of the DNA molecule by a so-called ultimate carcinogen. In fact, more than 90% of the carcinogens tested are mutagens in bacterial systems (1). The premutational event is the covalent binding of the ultimate carcinogen to the DNA bases or phosphate groups. The chemical structure of the adducts formed, and to a lesser extent the structural changes induced in the DNA double helix in the neighborhood of the adducts, has been extensively studied during the last ten years. However the crucial question is "How will the different repair, replication and recombination enzymes handle these chemically modified bases?" In other words, since the end point of this initial step is a mutation, "How is this initial premutational event converted into a stable and heritable *Institut de Biologie Moleculaire et Cellulaire du CNRS, 15 rue Rene Descartes, 67084 Strasbourg Cedex, France mutation?" N-Acetoxy-N-2-acetylaminofluorene (N-AcO-AAF) is a model ultimate metabolite of the strong rat liver carcinogen 2-acetylaminofluorene (AAF). In vivo studies have shown the mutagenicity of AAF and its derivatives in both bacterial (2,3) and eukaryotic systems (4). N-AcO-AAF reacts in vitro with DNA leading mainly to the formation of a guanine adduct (5), N-24deoxyguanosin-8-yl)acetylaminofluorene (80%) and also to at least three minor adducts (N. Schwartz, R. P. P. Fuchs and M. P. Daune, unpublished results), one of which is characterized as 34deoxyguanosin-N2-yl)acetylaminofluorene (6).
Studies from our group led to the general conclusion that binding of N-AcO-AAF to DNA resulted in a local distortion of the DNA helix around the C-8 adduct (7)(8)(9). We have called this structural alteration the insertion-denaturation model (10). A similar model has been proposed by other investigators (11).
In this paper we describe the analysis of forward mutations induced in the tetracycline-resistance gene of pBR322 by directing the chemical reaction of the carcinogen to a small restriction fragment (BamHI, Sail) inside the antibiotic-resistance gene. A preliminary report of this work has recently been published (12).
Samples of pBR322 reacted with N-AcO-AAF to various extends (ranging from 0 to 2.5% of modified bases) were digested with BamHI, and SalI restriction enzymes (Boehringer, Mannheim). The large fragment (16S fragment) and the small fragment (6S fragment) were separated and purified either by velocity sedimentation on sucrose Ultraviolet Irradiation of the Cells prior to Transformation In some cases, the E. coli cells were ultravioletirradiated prior to the transformation procedure. This treatment was used to induce the cellular SOS response. The cells were ultraviolet-irradiated as a suspension in 0.O1M MgSO4 with a germicidal lamp (15 W, Phillips) at a dose giving about 50% survival (i.e., 60 J/m2 for the wild type strain, AB 1157; 6 J/m2 for the uvrA strain, AB 1886). The cells were then incubated in LB medium for 30 min at 370C to allow expression of the SOS function.

E. coli Transformation and Selection of the Ampicillin-Resistant (ApR)and
Tetracycline-Sensitive (TcS) Clones The E. coli was made competent for transformation by the classical CaCl2 treatment procedure (16).
The different ligation mixtures were diluted by a factor of 100 in lOmM Tris, lOmM CaCl2 and 10mM MgCl2 (pH 7) and used to transform the competent cell suspension by mixing one volume of the DNA solution with two volumes of the concentrated E. coli suspension. Following the transformation procedure, the cells were spread on LB plates containing Ampicillin (50 jAg/mL) and incubated at 370C overnight. The clones were then replated on LB plates containing tetracycline (20 ,ug/mL). Clones which grew on Ap but not on Tc were scored as ApR, Tcs mutants. Such individual mutant clones were then grown further in LB medium plus Ampicillin for preparation of the plasmid DNA contained in these clones. Plasmid DNA was purified either on a small scale (10 mL of culture) by an adaptation of the method of Clewell and Helinski (17) or on a larger scale (11 culture) by a NaCl/SDS lysis procedure followed by a CsCl/Ethidium bromide centrifugation step (18).

DNA Sequence Analysis of the Mutants
Plasmids were digested with BamHI and SalI restriction enzymes and IP end-labeled at their 5' extremities with T4 DNA kinase (Boehringer, Mannheim). Strand separation and sequencing were performed according to the method of Maxam and Gilbert (19).

Results
The strategy that was used to obtain mutants in the tetracycline resistance gene, within the small restriction fragment (BamHI, Sal), is outlined in Figure 1. This restriction fragment (275 base pairs long: 6S fragment) modified to various extents with N-AcO-AAF[3H-ring] was reinserted by in vitro ligation into the nonreacted large (BamHI, Sail) restriction fragment (16S fragment). This large fragment contains both the gene coding for the 13lactamase (Ap resistance gene) and the origin of replication. The ligation mixture was used to transform CaCl2-treated E. coli recipient cells. Mutants are selected for Ampicillin (Ap) resistance and tetracycline (Tc) sensitivity.

Frequency of Obtention and Restriction
Enzyme Analysis of the ApR TcS Mutants Mutation frequencies were calculated as the ratio of ApR TcS clones/ApR clones. The mutation frequency in the control experiment in which the 16S fragment was ligated to a nonmodified 6S fragment was 0.4% when no ultraviolet treatment was applied to the bacteria prior to the transformation step. This frequency was similar (0.6%) when the ultraviolet treatment was applied. When analyzed by gel electrophoresis, the plasmid DNAs isolated from such clones were always shorter in length than the original pBR322. In general, the size reduction ranged from 0.2 to 0.8 kb. The restriction analysis pattern showed that these mutant DNAs had retained the unique Eco RI site but that in general they had lost both the BamHI SalI restriction sites. We call these mutants class I mutants and suggest that they mainly arise from the dimerization of the 16S fragment. Such dimers, which have the ApR TcS phenotype, are then converted to smaller plasmids (monomers) through in vivo recombination. Work is in progress to lower this mutation background by using an alkaline phosphatase-treated 16S fragment. Class I mutants were easily recognized and excluded from the pool of mutants to be sequenced.
When 6S-AAF fragments ligated to the nonmodified 16S fragment are used to transform E. coli one finds a decrease in the transformation efficiency with increasing levels of bound AAF residues. (Fig.  2). The extent of this AAF-dependent inactivation of transformation is strongly related to the general repair genotype of the recipient cell (R. P. P. Fuchs and E. Seeberg, manuscript in preparation). One also finds a corresponding increase in the mutation frequency, provided the cells are exposed to UV prior to the transformation step. The mutant DNAs isolated from such experiments fall into two classes when analyzed by gel electrophoresis: class I mutants defined as in the control experiment, and class II mutants, exhibiting the original size of pBR322 and retaining both BamHI and SalI restriction sites. The frequency of class II mutants increases with the level of AAF modification and reaches about 3% at the highest level tested (Fig. 2). It should be stressed that only the class II mutation frequency is a function of -AAF modification and dependent on ultraviolet irradiation of the host cell (for the conditions, see legend to Fig. 1).

Sequence Analysis of Nine pBR322 ApR TcS Mutants
Nine class II mutant plasmids were isolated from either the wild type E. coli strain, AB 1157, or the corresponding uvrA mutant strain, AB 1886. The double BamHI/SalI digested DNA was -P-endlabeled at the 5' extremities and sequenced according to the Maxam and Gilbert technique (19).
The sequence of the wild type 6S fragment of pBR322 was found to be identical to the sequence published by Sutcliffe (20). In all of the nine class II mutants we found a mutation located within the 6S fragment. All of the mutants showed a deletion of either a single GC base pair or a doublet of adjacent GC-CG base pairs. Two of these mutants (numbers 35 and 36) also had a second mutation (Table 1).
Two hot spot sequences for mutagenesis were found within the collection of the nine mutants.
Hot Spot Sequence 1. AS shown in Table 1, four out of the nine mutants (mutants number 4, 30, 41 and 45) show a deletion of a single G residue at position 520 or 521. (Numbering starts clockwise from the unique Eco RI site). It should be noted that the four mutants have arisen under quite different conditions (i.e., in two different strains, with or without ultraviolet induction of the SOS functions, with very different AAF modification levels).
Hot Spot Sequence 2. Mutants 34, as shown in GGCGCC, sequence that is in common to all four mutants 34, 36, 33, 35. This given sequence, that is found three times within the 6S fragment, can therefore be considered as a mutational hot spot. Such a GC deletion in an alternating GC sequence was shown in vivo to be a hot spot for reversion of the mutation, his D 3052, in Salmonellk by the mutagen 2-nitrosofluorene (21). It is striking to find that the same type of mutation occurs in both a reversion and a forward mutation assay.

Conclusion
The reason why such particular sequences are hot spots for mutagenesis is not clear. Whether or not such sequences are hot spots for the AAF binding reaction itself is presently under investigation. Alternatively, it is more likely that the processing of the premutational lesions is strongly sequencedependent. It is noteworthy that at both hot spots the sequence is such that one can draw a short "hairpin-type" secondary structure (Fig. 4). Such hairpins are too short to be stable by itself but might be highly stabilized by the conformational change that -AAF introduces when bound to C-8 of guanine. As stated by the insertion-denaturation model proposed by Fuchs and co-workers (7,10), there is a local denaturation of the helix around the  Sutcliffe (20). The bases in-  guanine-AAF adduct that might favor the hairpin structure shown in Figure 4. Due to the multicopy state and to the recessivity of the mutations that are scored in our system, the conversion of the premutagenic lesion into a stable mutation most likely occurs simultaneously in both strands prior to replication.  The molecular mechanism by which the mutation is being fixed remains to be elucidated.