Differential mutation of transgenic and endogenous loci in vivo

Mutation Research 454 (2000) 1–10

Differential mutation of transgenic and endogenous loci in vivo Lidia Cosentino, John A. Heddle∗ Department of Biology, York University, 4700 Keele Street, Toronto, Ont., Canada M3J 1P3 Received 18 June 1999; received in revised form 6 June 2000; accepted 20 June 2000

Abstract Although chemicals usually induce very similar frequencies of mutations in transgenes and endogenous genes in vivo when given acutely, chronic exposure to N-ethyl-N-nitrosourea (ENU) produced a more complex pattern in which the endogenous locus was spared many mutations. Here, we demonstrate that the effect is neither ENU-specific nor locus-specific, and thus, may be important in the extrapolations of risk assessment and in understanding mutational mechanisms. During chronic mutagen exposure, mutations at the transgene accumulate linearly with time, i.e. in direct proportion to the dose received. In contrast, mutations at the endogenous gene are much less frequent than those of the transgene early in the exposure period and the accumulation is not linear with time, but rather accelerates as the exposure continues. Previous comparisons involved the endogenous Dlb-1 locus and the lacI transgene from the Big BlueTM Mouse in the small intestine. These experiments involved the Dlb-1 locus and the lacZ transgene from the MutaTM Mouse in the small intestine and the hprt locus and the lacZ transgene in splenocytes. Comparisons were made in both tissues after acute and chronic exposures to ENU, the original mutagen, and in the small intestine after exposures to benzo(a)pyrene. All comparisons showed that during chronic exposures mutations at the transgene accumulate linearly with the increasing duration of exposure, whereas induced mutations of the endogenous gene initially accumulate at a slower rate. Thus, the difference in mutational response observed during low chronic treatment is not unique to a particular transgene, endogenous gene, tissue, or mutagen used, but may be a general phenomenon of such genes. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Somatic mutation; DNA repair; Dlb-1; hprt; lacZ

1. Introduction The measurement of somatic mutation has been revolutionized by the development of mice carrying a recoverable test locus, especially the commercially available MutaTM Mouse and Big BlueTM Mouse. These loci provide a window for the study of sponta∗ Corresponding author. Tel.: +1-416-736-2100/extn 33053; fax: +1-416-736-5698 E-mail addresses: [email protected] (L. Cosentino), [email protected] (J.A. Heddle).

neous and induced mutations in vivo. This window, however, is only valid if these loci accurately reflect the mutation rates at endogenous loci. Under these conditions, if these loci represent the genome as a whole, then the information obtained can be extrapolated to the human situation. The transgenic loci are not functional in the mouse and are presumed to be unexpressed, as they have neither mammalian promoter sequences nor polyA addition sites. Mutations at these loci should, therefore, be genetically neutral, conferring neither advantage nor disadvantage upon the cells that contain them. Accordingly, (a) somatic

0027-5107/00/$ – see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 7 - 5 1 0 7 ( 0 0 ) 0 0 1 2 5 - 1

2

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

tissues should simply accumulate mutations during chronic exposure and (b) the mutant frequency should remain constant after exposure and adequate manifestation time, predictions that have been tested in several tissues and largely fulfilled [1–3]. Two endogenous loci that are available for quantitative in vivo comparisons and analysis of mutations are each restricted to a tissue in the mouse hprt to splenocytes and Dlb-1 to the epithelium of the small intestine. Comparisons of these loci (Dlb-1, hprt) to transgenic loci (lacI, lacZ) after an acute treatment with ENU have shown a remarkable concordance of induced mutant frequencies in each tissue [4,5]. It was a surprise, therefore, when the endogenous Dlb-1 gene showed a different pattern of mutation in response to an extended subacute exposure regimen [6]. The existence of this effect that occurs under low dose conditions has important implications for extrapolations from acute exposures in the determination of risk. Under the subacute regime, mice that were treated daily with low doses of ENU, corresponding to 1 and 0.3% of the LD50, for up to 120 days, accumulated lacI mutations linearly but Dlb-1 mutations non-linearly. The initial accumulation of Dlb-1 mutations was slower than lacI mutations (<30 days). This was followed by a gradual increase in the Dlb-1 mutation rate [6]. The resulting curve can be arbitrarily divided into two phases. In the initial phase, the Dlb-1 mutation rate is substantially lower than the lacI mutation rate whereas in the later phase, the Dlb-1 mutation rate exceeds that of the transgene, in spite of the constant rate of exposure to the mutagen [6]. Clearly this effect cannot be explained by physiological changes in the animal that might alter the effective DNA exposure to the chemical, for this would have affected both loci equally. Indeed, animals that had been exposed chronically for 90 days, and were thus, in the phase of rapid accumulation of Dlb-1 mutations proved to be no more sensitive to a challenge with a high dose of ENU than previously unexposed animals or than animals exposed chronically for shorter times [6]. This confirmed that the overall physiology of the animal and the cells was not changed so as to sensitize the cells or alter the effective dose. Equally, the results are not an artifact of selection since Dlb-1 mutation frequencies were stable after exposure ceased, confirming the neutrality of Dlb-1 mutations in this tissue [6]. Here we address the magnitude and the generality of this phenomenon. The

difference in mutagenic response detected between the endogenous and transgenic loci can be explained by either (a) differential damage or (b) differential repair of the loci under these conditions. Non-random DNA damage could contribute to the effect since transcriptionally active DNA sequences have been shown to be more susceptible to damage than are inactive sequences [7–10]. This difference is, of course, in exactly the wrong direction to explain the reduced mutagenesis at the transcribed endogenous locus. These differential susceptibilities have been attributed to a difference in their chromatin organization. If chromatin structure is a contributing factor, it is surprising that it has no influence on the mutant frequencies observed at high acute doses, but would affect these frequencies only at low chronic doses. Another factor that may be involved in the different mutational response is the rate of DNA repair, as Dlb-1 is transcribed and the transgene is not. Transcription-coupled repair (TCR) has been well documented and characterized and may influence mutation rates characterized. It has been shown that DNA repair occurs faster in transcriptionally expressed genes when compared to non-expressed genes and more rapidly in the transcribed strand than in the non-transcribed strand [11–14]. This differential rate of DNA repair has been termed preferential repair. Such an effect may explain the lack of Dlb-1 mutations relative to lacZ mutations under low dose chronic exposure. The data at hand show that the different mutational response first detected between Dlb-1 and lacI after low chronic mutagen exposure is not exclusive to one locus, one tissue or one mutagen, but is a more general result. 2. Materials and methods 2.1. Animals Non-transgenic SWR mice (homozygous for Dlb-1a ) were mated with the lacZ MutaTM Mouse (homozygous for Dlb-1b ) to obtain an F1 generation that were hemizygous for the lacZ transgene and heterozygous at the Dlb-1 locus (Dlb-1b /Dlb-1a ). F1 animals used were 3–4 months old at the start of all experiments. An independent Animal Care Committee approved all experimental protocols in advance.

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

2.2. Chemical treatments 2.2.1. ENU treatment: Dlb-1 versus lacZ The ENU (CAS no. 759-73-9) test solution, at a concentration of 94 ␮g/ml, was made by dissolving ENU in 10% dimethyl-sulphoxide and 90% distilled water adjusted to pH 4. The stability of ENU was tested spectrophotometrically. At pH 4 ENU was stable for 1 week and was shown to break down completely within an hour at pH 8. Each treatment group contained six animals while each control group consisted of four animals. All ENU-treated animals presented in this paper were exposed to ENU orally (drinking water). Every 3–4 days, when water bottles were replaced, an individual vial of frozen concentrate was diluted with distilled water (pH 4) to a concentration of 94 ␮g/ml. In this experiment animals started treatment at the same time and were sacrificed 10 days after the end of treatment. 2.2.2. ENU treatment: hprt versus lacZ Four animals were randomly assigned to each treatment group while each control group contained three animals. It is well known that the induction of hprt mutations are heavily age dependent [18]. The complex nature of T-cell maturation and movement in the body also influences the mutant frequency at the time of sampling. Studying mutations at hprt is further complicated by the fact that the mutant frequency varies in a complex fashion with time after treatment, making valid comparisons difficult. Unfortunately it is not known if the lacI mutant frequency follows the same pattern with time found for hprt in these cells. For this reason, we designed the experiment to include two treatment protocols to minimize the effect of age and selection on the data. In the first set (Trial I), the start of treatment was staggered but animals were all sacrificed on the same day. The second set (Trial II), the reverse occurred, animals began treatment on the same day but were sacrificed on different days. Animals were treated daily and sacrificed either at 3, 7 or 15 weeks post-treatment. 2.2.3. B(a)P treatment: Dlb-1 versus lacZ B(a)P was selected because its mutagenic effect depends on metabolic activation. Additionally, B(a)P induces bulky DNA adducts which are primarily removed by the nucleotide excision repair pathway,

3

whereas many of the alkylation-products induced by ENU are removed by the O6 -alkylguanine DNA repair protein. In the initial experiment (Trial I) five animals were assigned to either a treatment group or control group. In the repeat experiment (Trial II) six animals were assigned to each treatment group while four animals were assigned to each control group. The B(a)P (CAS no. 50-32-8) was thoroughly mixed with crushed lab chow and made into food pellets with the addition of water. Mice were fed a dose of 40 ppm B(a)P in the diet for up to 8 weeks and sacrificed 2 weeks post-treatment. 2.3. Dlb-1 mutation analysis Whole mounts of the small intestine were prepared as previously described [15]. Briefly, the small intestine was divided into its three sections: duodenum, jejunum, and ileum, and flushed clean with phosphate-buffered saline (PBS). The jejunum was used for this assay while the remaining sections of the intestine were used for the transgenic assay. After the jejunum was flushed, inflated, and cut along the mesenteric side, it was placed on a microscope slide, villi side up. After slides were fixed in 10% formal saline for a least 1 h, they were rinsed with PBS and incubated overnight in 20 mM DL-dithiothreitol (Sigma) dissolved in 20% ethanol, 80% 150 mM Tris (pH 8.2) to remove mucus. The slides were stained with the D. biflorus agglutinin-peroxidase conjugate (Sigma) at 5 ␮g/ml in the PBS/albumin (fraction V). The peroxidase was developed by using 3,30 -diaminobenzidine (Sigma) solution for 45 min. The slides were rinsed twice with PBS and stored in 10% formal saline until analyzed. The slides were scored with a dissecting microscope at 50× magnification. The Dlb-1b /Dlb-1a epithelial cells stain dark brown; a clone of mutant cells (Dlb-1− /Dlb-1a ) which have no lectin-binding ability, appear as an unstained (white) vertical ribbon on the villus. The number of villi scored per animals was estimated from duplicate counts of the number of villi in the first and last field. Each field contained approximately 200 villi, to yield an average of 10,000 villi per animal. Since there are about 10 stem cells per villus [16] roughly 1,00,000 stem cells were analyzed per animal.

4

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

2.4. hprt Mutation analysis The protocol used for isolating splenic lymphocytes for hprt analysis was carried out as previously described [4]. Briefly, lymphocytes were isolated by crushing spleens individually in RPMI-1640 Mouse Media. The cell suspension was then layered onto 4 ml of lympholyte M (Cedarlane Laboratories) for collection, washed, resuspended in medium, cultured, and plated in 96-multiwell plates at high densities with 6-thioguanine (1 ␮g/ml) to detect thioguanine resistant cells. An aliquot of each sample was stored at −80◦ C for DNA isolation for lacZ analysis (as described below). The plates were incubated in a 6% CO2 humidified incubator. Plating efficiency plates (PE) were scored on day 8 while selection plates were scored on day 10. The number of clone-forming units per well on the PE and 6TG plates was calculated by observing the fraction of negative wells and applying Poisson statistics.

packaging extract (Strategene, La Jolla, CA), under conditions recommended by Hazleton Research Products Inc. Packaged phage was absorbed with E. coli C lacZ− galE− bacteria strain. The remaining phage/bacteria mixture was mixed with approximately 8 ml freshly prepared top selection agar supplemented with 0.3% phenyl-B-D-galactopyranoside (P-gal, Sigma). P-gal is used as a selective agent for bacteria infected with phage containing non-functional lacZ genes. P-gal in the selection plates prevents the formation of wild type plaques, thus, any plaques observed represent lacZ− mutants [17]. The titre plates were determined by plating a small aliquot of the phage solution on LB top agar without P-gal. Plaques were scored on both titre and selective plates after overnight incubation at 37◦ C. The number of plaques recovered from each animal varied, with a mean of approximately 400,000 plaques.

3. Results 2.5. Transgenic mouse assays 2.5.1. Part I. Tissue collection The jejunum section of the small intestine was reserved for the Dlb-1 assay, while the remainder of the tissue was used for the lacZ assay. After the intestine was flushed with PBS and inverted, it was placed in 3 ml of KCl (75 mM)–EDTA (20 mM) solution and forced in and out of a 5 ml syringe. The cell suspension was placed in liquid nitrogen then stored at −80◦ C for future use. 2.5.2. Part II. DNA isolation Genomic DNA was purified from the small intestine cell suspension and T-cell samples with a proteinase K solution (2 mg/ml) for 3 h at 55◦ C, followed by phenol:chloroform (1:1) extraction and precipitation with ethanol. The precipitated DNA was spooled onto a hooked glass Pasteur pipet, air dried and dissolved in Tris–EDTA buffer. The concentration of DNA was determined spectrophotometrically at 260 nm. 2.5.3. Part III. DNA packaging 2.5.3.1. lacZ mutations. The ␭ phage shuttle vector, which contains the entire lacZ target gene was recovered by in vitro packaging with TranspackTM

The data collected in each study were tested for linearity using the MicroSoft Excel software statistical test for regression analysis. 3.1. ENU treatment: lacI versus lacZ In a previous study when F1 lacZ mice were treated with ENU orally (drinking water) for 30, 60 and 90 days a deficiency of Dlb-1 mutations was detected early on relative to lacZ mutations [23]. The chronic treatment protocol was much more effective at inducing mutations and revealed differences between lacZ and Dlb-1 within 30 days, as reported here. Fig. 1 clearly demonstrates that lacZ mutations accumulate linearly (F = 3.3; df = 1, 12; P = 0.74) but the accumulation of Dlb-1 mutations does not (F = 115.3; df = 1, 16; P < 0.05). Instead, Dlb-1 mutations accumulate slowly at first and then at an accelerated rate. In acute experiments, the frequencies of mutations at these two loci are essentially the same [5], but this was not observed at the early time points during chronic exposure. The ratio of lacZ:Dlb-1 induced mutant frequency is remarkable (Fig. 2a). The lacZ/Dlb-1 ratio may be as high as a 22-fold difference after 7 days of treatment. By subtracting the induced mutant frequency from the previous treatment group we can

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

5

Fig. 1. Mean mutant frequencies (±S.E.M.) induced after chronic exposure to ENU (94 ␮g/ml per day). (a) Dlb-1 locus; (b) lacZ transgene.

obtain the ratio induced in each interval (Fig. 2b). A small number of Dlb-1 mutations are induced each day early in the exposure, but with continued treatment, the difference between the loci gradually decreases until the numbers of induced mutations approach that of the transgene. This study confirms the observations of Shaver-Walker et al. (1995) in a different genetic background, a different transgene in a different chromosomal location, and with a different route of exposure. 3.2. ENU treatment: hprt versus Dlb-1 Since both of the previous studies involved a single endogenous locus, Dlb-1, we compared the lacZ and hprt locus in splenic T-cells. As Fig. 3 demonstrates, at 3 weeks after treatment, a non-linear increase in the mutant frequency is observed for hprt (Trial I: F = 32.3; df = 1, 10; P < 0.05; Trial II: F = 28.9; df = 1, 10; P < 0.05), but not for lacZ (Trial I: F = 0.09; df = 1, 9; P = 0.09; Trial II: F = 2.6; df = 1, 9; P = 0.18). In fact, this was seen at 7 and 15 weeks after treatment regardless of when mice started or stopped treatment (Trial I versus Trial II) (data not shown). In acute experiments, hprt and lacI comparisons have given mixed results depending on the mutagen used. Skopek et al. showed that the induction of mutations by ENU is very similar whereas B(a)P produces many more lacI mutations than hprt mutations [19]. In both experiments animals were 3 weeks old at the

time of treatment and sacrificed 3 weeks later. Since the kinetics of the lymphoid system slows down with age, 3-week-old animals are typically used and sacrificed 3 weeks later, which maximizes the yield of hprt− mutants [4]. Assuming that the lacZ transgene responds similarly to ENU than the lacZ/hprt ratio would be very close to 1. Thus, according to Fig. 4, after 10 days of treatment there is roughly an 80-fold difference (chronic ratio versus acute ratio). Nevertheless, the same pattern is observed, the lacZ/hprt ratio decreases with longer ENU exposure. The data, thus, show that the different mutational response observed in the Dlb-1/lacZ comparison in the small intestine is also found in this hprt/lacZ comparison involving different loci, and a different cell type. 3.3. B(a)P treatment Since the previous studies were all conducted with ENU, we tested another mutagen, B(a)P, to investigate the generality of this phenomenon. In a preliminary experiment, animals were treated chronically with B(a)P for 1, 3 and 4 weeks. As seen in Fig. 5 (Trial I), a linear increase in the mutant frequency was seen for lacZ (F = 4.2; df = 1, 11; P = 0.18) but not for Dlb-1 (F = 41.4; df = 1, 12; P = 0.16). The non-linear increase in mutant frequency at the Dlb-1 locus after B(a)P treatment is reproducible. In a second experiment, animals were treated with B(a)P for 1, 2, 3, 4, 5, 6 and 8 weeks. Fig. 5 (Trial II) shows that muta-

6

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

suggesting that Dlb-1 and lacZ are repaired, or damaged at a differential rate. Thus, the data confirm the effect for a different mutagen.

4. Discussion

Fig. 2. Data obtained from chronic ENU exposure, animals were treated for up to 90 days. (a) The induced lacZ/Dlb-1 mutant frequency ratio observed after chronic ENU treatment; (b) the number of Dlb-1− and lacZ− mutants induced per day after continuous ENU exposure. Open symbols represent individual animals; closed symbols represent the means (±S.E.M.).

tions accumulate at Dlb-1 non-linearly (F = 120.2; df = 1, 38; P = 0.22) whereas lacZ mutants increase linearly (F = 3.6; df = 1, 19; P = 0.2), as expected. After acute treatment, B(a)P like ENU, induces similar mutant frequencies at these two loci [5]. Although the lacZ/Dlb-1 mutant frequency ratio is not as striking after treatment with B(a)P, only a three-fold increase over acute ratio, compared to a 22-fold difference with ENU, the same pattern is observed. That is, only after prolonged treatment does the lacZ/Dlb-1 mutant frequency ratio equal that of acute treatments (Fig. 6),

The mechanism that underlies the differential mutational response between endogenous and transgenic loci is uncharacterized. There are two aspects that require explanation, first the lack of endogenous mutations early in a low dose chronic exposure and, second, the increasing rate of mutation at the endogenous gene while the rate at the transgene is constant. A difference in mutant frequency at two loci is not normally a matter requiring explanation, as every locus has a characteristic mutation rate, induced or spontaneous, but mutations at these loci are almost equally frequent after acute exposure [4,5]. Previous results show that the difference in mutational response is not the result of selection, as mutations at Dlb-1, lacI, and lacZ are all neutral [1,2,6]. Nor can this be attributed to changes in the effective dose to the cell, as all loci in the cells would be affected proportionately. The transgenes accumulate mutations linearly as the exposure continues, which indicates that the difference lies in mutation at the endogenous locus in each case. There are, of course, many differences between the loci other than their transgenic and endogenous status that may account for the difference in mutational response. The Dlb-1, hprt, lacI and lacZ are on different chromosomes (11, X, 4, and 3, respectively), which suggests that location is not a factor. Furthermore, their sequences and size differ; lacI (1 Kb) is roughly one-third the length of lacZ (3 Kb), hprt contains about 1 Kb of coding sequence in about 30 Kb of DNA, and Dlb-1 is of unknown size. The transgenes are (a) heavily methylated, (b) bacterial in origin and embedded in viral DNA, (c) present in multiple tandem copies in the genome, and (d) lacking in mammalian promoters. Conversely, Dlb-1 and hprt are naturally present in the genome and are expressed in the small intestine and T-cells, respectively. Of these factors the one that might reasonably account for the difference between the loci is transcription, as TCR has been well documented, but this does not provide a full explanation of the data, as discussed below. Alternatively non-random DNA damage in

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

7

Fig. 3. Mean mutant frequencies (±S.E.M.) induced after chronic exposure to ENU (94 ␮g/ml per day) at the hprt locus and the lacZ transgene 3 weeks after treatment. Trial I group of animals were sacrificed on the same day and the start of ENU treatment was staggered. In Trial II group, the reverse occurred: all animals began treatment at the same time but were sacrificed at staggered intervals.

relationship to chromatin structure may contribute to the difference. There is now evidence that chromatin structure can affect the production, distribution and repair of damage in DNA [7–10]. While a high degree of genome compactness could result in a resistance to damage, it also implies that the genome will be less accessible to repair enzymes, thus, resulting in less efficient repair. In contrast, transcriptionally active genes, with a decondensed chromatin structure, may show a higher sensitivity to various chemical

agents but may have the best efficiency to repair the damage. If chromatin structure is a factor influencing the mutant frequencies during chronic treatment it is not evident (a) why its influence would change with time in the case of the endogenous loci at which the mutant frequency increases non-linearly, and (b) why it would not affect the mutant frequencies at high acute doses. Furthermore, since the mutant frequency changes with time for the endogenous locus then it would be necessary that the chromatin structure

Fig. 4. Ratios of induced lacZ:hprt mutant frequencies after daily treatment with ENU (0.8 M/day). Open symbols represent individual animals; closed symbols represents the means.

8

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

Fig. 5. Mean mutant frequencies (±S.E.M.) observed after daily treatment with B(a)P (40 ppm) in two separate studies; Trial I and Trial II. (a) Dlb-1 locus; (b) lacZ transgene.

would also. Certainly this explanation alone does not account for the differences observed. The deficiency of mutations at the endogenous locus early on may involve DNA repair since it is known that expressed loci are preferentially repaired over non-expressed DNA, i.e. transgenes [11–14]. Evidence shows that during TCR in expressed genes the non-transcribed strand is repaired at the global

genomic rate [11,12]. Assuming the global rate applies to the transgene, TCR would reduce the mutation rate by no more than a factor of 2 relative to a transcribed gene even with perfect repair of the transcribed strand. In contrast, the lacZ/Dlb-1 ratio reaches a 22-fold difference (Fig. 2a) after 7 days of treatment and continues to decline thereafter with continued mutagen exposure. Thus, the effect is too

Fig. 6. Ratios of induced lacZ:Dlb-1 mutant freuencies after daily treatment with B(a)P (40 ppm) in two separate studies; (A) Trial I and (B) Trial II. Open symbols represent individual animals; closed symbols represents the means; straight line represents ratio after acute treatment [23].

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

large to be explained by classical TCR alone. TCR could be involved if (a) the global rate is not uniform, for which there is some evidence, in which case, the non-transcribed strand may be repaired faster than some parts of the genome, such as heavily methylated transgenes. Alternatively, the non-transcribed strand may also be preferentially repaired at low doses as used here, which are lower than the high doses typically used in repair studies. Furthermore, the general measurement in TCR studies is the rate of removal of cyclobutane pyrimidine dimers or alkylation damage and not mutations. If repair is involved in this effect it is of considerable importance with respect to induced mutation. The slower accumulation of mutations at the endogenous verses transgenic loci could be a result of a difference in repair efficiency at low damage levels. It is possible that a steady state level of a particular pro-mutagenic adduct is escaping preferential repair and is slowly building up in the population, which would account for the rapid increase in the mutant frequency at the endogenous locus. If, however, this were occurring, then the mutant frequency would continue to rise after the cessation of treatment but it does not [6], ruling out this possibility. Additionally, another contributing factor could be the activation of a different mutagenic pathway operating at the endogenous loci but not at the transgenic loci. A deletion or recombination-producing pathway would fit, since transgenic loci would not be expected to respond to such pathways. If the difference in the frequency between the loci were in fact all deletion mutants (which would be a surprising result from ENU and B(a)P which primarily induce base changes) this does not explain the non-linearity component of the curve. Additionally, X-rays have been shown to induce a higher mutant frequency at Dlb-1 than lacI [2]. This difference, however, can be explained by the nature of mutant selection. Obviously, if a deletion occurs at one end of the vector, then a viable phage will not be recovered. Thus, it does not seem likely that a deletion-producing pathway could account for the difference in mutational response detected between the loci. The recombination-producing pathway also fails to explain the data since hprt is located on the X-chromosome but the mutant frequency increases non-linearly in male mice equally with females (data not shown).

9

Another possible explanation is that the protein O6 alklyguanine alkyltransferase (O6 -AGT) is implicated in the effect, as it has been found to be the major repair process for O6 -alkylguanine and O4 -alkylthymine adducts induced by ENU. O6 -AGT has been shown to repair O6 -alklyguanine lesions more rapidly from active than from inactive genes [21]. It is not known if O6 -AGT shows strand-specific repair of transcriptionally active genes. Because O6 -alkyguanine lesions do not halt transcription, it has been suggested that the presence of O6 -AGT at active transcription sites may be associated with an additional TCR pathway [22]. Indeed, it appears that there are two distinct preferential repair pathways, one being the prominent transcription-coupled repair pathway typically associated with nucleotide excision repair (NER), and the other being a transcription-independent one, correlated with AGT. Because O6 -AGT is not rejuvenated after its reaction with an alkyl group, a depletion and or saturation of the repair protein was thought to be an important factor implicated in this effect. Since B(a)P-induced adducts are removed via the NER pathway and B(a)P induced a differential mutational response between the loci (Dlb-1 versus lacZ), it is unlikely that O6 -AGT alone is involved in the phenomenon. In a prior experiment with PhIP, which also produces bulky DNA adducts, a differential response between the endogenous Dlb-1 locus and the lacI transgenic locus was not observed. The reason for this difference is unknown [20]. Nevertheless, further tests elucidating the role of O6 -AGT are possible by using transgenic mice carrying extra copies of the bacterial methyltransferase gene and O6 -methylguanine-DNA methyltransferase knockout mice. These experiments confirm that the difference in mutational response (i.e. during chronic mutagen exposure mutations at the transgene accumulate linearly whereas mutations at the endogenous gene initially accumulate at a slower rate, resulting in a non-linear accumulation) first detected by Shaver-Walker et al. (1995) is not limited to one genetic background nor to one locus, one tissue or one mutagen, but is a more general event. The data reflects the importance of an uncharacterized mechanism, one that appears to be associated with endogenous genes. The existence of this mechanism that occurs specifically under low dose conditions has important implications for

10

L. Cosentino, J.A. Heddle / Mutation Research 454 (2000) 1–10

extrapolations from acute exposures in the determination of risk. Since the effect is observed at the hprt locus, it is now feasible to analyze changes in the mutation spectrum, which is not possible with Dlb-1 as it has not yet been cloned. Such analysis may provide a means to help define the mechanism that may be involved in this phenomenon.

[9]

[10]

[11]

Acknowledgements [12]

We are grateful to HRP Inc. for permission to cross the MutaTM Mouse to produce the F1 necessary for this study. We thank Cesare Urlando, Jennifer Moody, Roy Swiger, Grace Trentin, Naoko Shima, Beichen Sun, William Cruz, Dara Dickstein, Waseem Kalair, Jason Bielas and Anita Samardzic for help with these experiments. T.R. Skopek provided helpful advice on the hprt assay. This work was supported principally by a grant from the National Cancer Institute of Canada and Stratagene also assisted in this research.

References [1] L. Cosentino, J.A. Heddle, A test for neutrality of mutations of the lacZ transgene, Environ. Mol. Mutagen. 28 (1996) 313–316. [2] K.S. Tao, C. Urlando, J.A. Heddle, Comparison of somatic mutation in a transgenic versus host locus, Proc. Natl. Acad. Sci. U.S.A. 90 (1993) 10681–10685. [3] G.R. Douglas, J. Jiao, L.M. Gingerich, J.A. Soper, J.A. Gossen, Temporal and molecular characteristics of lacZ mutations in somatic tissues of transgenic mice, Environ. Mol. Mutagen. 28 (1996) 317–324. [4] T.R. Skopek, K.L. Kort, D.R. Marino, Relative sensitivity of the endogenous hprt gene and the lacI transgene in ENU-treated Big Blue B6C3F1 mice, Environ. Mol. Mutagen. 26 (1995) 9–15. [5] L. Cosentino, J.A. Heddle, A comparison of the effects of diverse mutagens at the lacZ and Dlb-1 loci in vivo, Mutagenenesis 14, 113–119. [6] P.M. Shaver-Walker, C. Urlando, K.S. Tao, X.B. Zhang, J.A. Heddle, Enhanced somatic mutation-rates induced in stemcells of mice by low chronic exposure to ethylnitrosourea, Proc. Natl. Acad. Sci. U.S.A. 92 (1995) 11470–11474. [7] S. Chiu, N. Oleinick, L. Friedman, P.J. Stambrook, Hypersensitivity of DNA in transcriptionally active chromatin to ionizing radiation, Biochem. Biophys. Acta 699 (1982) 15– 21. [8] N. Oleinick, S. Chiu, L. Friedman, Gamma radiation as a probe of chromatin structure: damage to and repair of active

[13] [14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

chromatin in the metaphase chromosome, Radiat. Res. 3 (1984) 629–641. T. Boulikas, Evolutionary consequences of nonrandom damage and repair of chromatin domains, J. Evol. 2 (1992) 156–180. D. Lopez-Larraza, N. Bianchi, DNA response to bleomycin in mammalian cells with variable degrees of chromatin condensation, Environ. Mol. Mutagen. 21 (1993) 258–264. V.A. Bohr, C.A. Smith, D.S. Okumoto, P.C. Hanawalt, DNA repair in an active gene: removal of pyrimidne dimers from the DHFR gene of CHO cells is much more efficient that in the genome overall, Cell 40 (1985) 359–369. I. Mellon, P.C. Hanawalt, Induction of the Escherichia coli lactose operon selectively increases repair of its transcribed DNA strand, Nature 342 (1989) 95–98. L. Ho, P.C. Hanawalt, Gene-specific DNA repair in terminally differentiating rat myoblasts, Mut. Res. 255 (1991) 123–141. L. Lommel, C. Carsell-Crumpton, P.C. Hanawalt, Preferential repair of the transcribed DNA strand in the dihydrofolate reductase gene throughout the cell cycle in UV-irradiated human cells, Mut. Res. 336 (1995) 181–192. D.J. Winton, M.A. Blount, B.A.J. Ponder, A clonal marker induced by mutation in mouse intestinal epithelium, Nature 333 (1988) 463–466. L. Cosentino, P.M. Shaver-Walker, J.A. Heddle, The relationship among stem cells crypts, and villi in the small intestine of mice as determined by mutation-tagging, Dev. Dyn. 207 (1996) 420–428. J.A. Gossen, A.C. Molijn, G.R. Douglas, J. Vijg, Application of galactose-sensitive E. coli strains as selective hosts for lacZ-plasmids, Nucleic Acids Res. 20 (1992) 3254. I.M. Jones, K. Burkhart-Schultz, C.L. Strout, T.L. Crippen, Factors that affect the frequency of thioguanine-resistant lymphocytes in mice following exposure to ethylnitrosourea, Environ. Mol. Mutagen. 9 (1987) 317–329. T.R. Skopek, K.L. Kort, D.R. Marino, L.V. Mittal, D.R. Umbenhauer, G.M. Laws, S.P. Adams, Mutagenic response of the endogenous hprt gene and lacI transgene in benzo(a)pyrene-treated Big Blue B6C3F1 mice, Environ. Mol. Mutagen. 28 (1996) 384. X.B. Zhang, J. Felton, J.D. Tucker, C. Urlando, J.A. Heddle, Intestinal mutagenicity of 2-amino-1-methyl-6-phenylimidazo [4,5-b]pyridine (PhIP) and amino carboline in transgenic mice, Carcinogenesis 17 (1996) 2259–2265. J. Thomale, K. Hochleitner, M.F. Rajewsky, Differential formation and repair of the mutagenic DNA alkylation product O6 -ethylguanine in transcribed and nontranscribed genes of the rat, J. Biol. Chem. 269 (1994) 1681–1686. R.B. Ali, A.K.C. Teo, H.K. Oh, L.S.H. Chuang, T.C. Ayi, B.F.I. Li, Implication of localization of human DNA repair enzyme O6 -methylguanine-DNA methyltransferase at active transcription sites in transcription-repair coupling of the mutagenic O6 -methylguanine lesion, Mol. Cell. Biol. 18 (1998) 1660–1669. L. Cosentino, J.A. Heddle, Effects of extended chronic exposures on endogenous and transgenic loci: implications for low dose extrapolations, Environ. Mol. Mutagen. 34 (1999) 208–215.