Mutagenesis Advance Access first published online on May 25, 2008
This version published online on June 2, 2008
Mutagenesis, doi:10.1093/mutage/gen026
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Frequency and spectrum of ENU-induced mutation in the 
X174 transgene in mouse splenic lymphocytes and their significance to spontaneous transgenic rodent mutation frequencies
1Division of Genetic and Reproductive Toxicology, National Center for Toxicological Research, 3900 NCTR Road, HFT-120, Jefferson, AR 72079, USA 2Department of Epidemiology, College of Public Health, University of Arkansas for Medical, 4301 West Markham, #820 Little Rock, AR 72205, USA
A perceived disadvantage of transgenic rodent mutation assays is that spontaneous mutant frequencies are high compared to those of endogenous genes and may consequently reduce sensitivity to induced mutation. We have previously argued that unrepaired G:T mismatches from spontaneous deamination of 5-methylcytosine at CpG sites could be converted to apparent in vivo mutations in the bacterial recovery systems because of rapid, random, mismatch repair in Escherichia coli. In this study, we have measured mutation frequencies in spleen of male mice induced by N-ethyl-N-nitrosourea (ENU) using the 
X174 transgene, which is not subject to mismatch repair in E.coli, using single-burst analysis, a unique method to identify in vivo mutation. In order to compare our results to those using the lacI and cII transgenes, we converted all mutant frequencies to base pair substitution (bps) mutation frequencies per nucleotide based on mutant spectra from this study and published literature. We found this frequency in control spleen to be similar for lacI (3.8 ± 0.7 x 10–8) and X174 (3.1 ± 1.2 x 10–8) at 6 weeks of age. We found a strong age dependence for spontaneous lacI mutation that extrapolated to a value at conception (1.8 ± 0.9 x 10–8) that was not significantly different from the human germ line bps mutation frequency per nucleotide of 1.7 ± 0.2 x 10–8. These two transgenes provided similar mutational responses to 40 mg/kg ENU, 7- to 9-fold. In contrast, the cII target gene in the same tissue produces both spontaneous and induced mutation frequencies 
10 times higher, for unknown reasons. We conclude that the spontaneous mutant frequencies measured by the lacI and X174 transgenes in this moderately dividing tissue accurately measure in vivo mutation frequencies at early ages. For these two transgenes, seemingly high mutant frequencies may reflect the expected accumulation of somatic mutation with age.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionFundingSupplementary dataReferences |
|---|
Almost 20 years ago, several transgenic rodent mutation assays were developed that provided a much-anticipated ability to measure mutant frequencies in any somatic tissue large enough to provide the necessary amount of DNA (1
). The merit of these assays lies in the fact that it is not necessary to culture mammalian cells under selective conditions in order to identify mutant cells because the mutational target is a bacterial gene (or bacteriophage gene) integrated into the rodent genome within a recoverable vector. The rodent DNA is extracted from the tissue and, by various methods dependent on the vector, introduced into laboratory bacteria where the mutant phenotype is expressed in the natural host. However, a perceived disadvantage of these systems is that they are considered to have relatively high spontaneous mutant frequencies compared to endogenous genes, Hprt being the gene used most frequently for comparison (1, sections 1.4.2.2. and 3.6.6).
Early in the development of the transgenic systems, the possibility was considered that recovered mutations might originate in the recovery bacteria, and considerable work was done to establish that most recovered mutations originate in vivo, based on mutant frequency variation between animals, mutant spectra of jackpot, non-jackpot and Escherichia coli mutations, and the absence of mosaic plaques (2). It has been demonstrated that N-ethyl-N-nitrosourea (ENU) adducts can produce mutations in the lacI target (3–6), although not so for lacZ (7–9), cII (6) or X174 (10).
In addition, the possibility that unrepaired photoproducts might produce mutation in bacterial systems has been considered. Cruz-Munoz et al. (9) found that bacteriophage treated with 50–1620 J/m2 of ultraviolet C (UVC) produced lacI mutant plaques in E.coli that were all mosaic, but the same dosages produced no mutant plaques in the lacZ target. However, in mouse primary embryonic cell culture, Bielas and Heddle (11) found that UVC-induced damage to mouse DNA did not produce mutant lacI plaques in E.coli using a much lower dose (0.125 J/m2).
It has been repeatedly argued that positive selection assays (lacZ, cII and gpt genes and Spi– selection of deletions) prevent recovery of DNA damage in the rodent cell as mutations fixed in bacteria because the wild-type phenotype in the same bacterial cell with a mutant phage molecule would exert its selective pressure and eliminate recovery of the mutant (1,3,6). However, using positive selection for the lacZ gene in mouse epidermis after ultraviolet B (UVB) irradiation, Ikehata and Ono (12) note an immediate rise in mutant frequency, which declines before rising again. Since the same initial rise was seen with irradiated phage DNA (data not shown), the authors conclude that the early rise represents ex vivo mutation (fixed in bacteria from photoproducts) and the decline in mutant frequency reflects error-free repair of DNA photoproducts. This is an example of the recovery of ex vivo mutations with a positive selection assay.
An additional objection to this argument is that the expression of the wild-type phenotype requires achievement of a certain level of the appropriate protein for the selective effect, as for β-galactosidase (product of the lacZ gene) in producing the toxic metabolite, UDP-gal (13). The bacterial cell may survive through several phage DNA replications before selective pressure occurs.
It is widely claimed that the decision between lysis and lysogeny (cII selection) is made before any DNA replication, e.g. Willems and Benthem (14) ‘... ex vivo mutations are never a problem because of the immediate commitment to lysogeny/lysis following infection. The decision for lysogeny/lysis is already made before any DNA replication takes place (Watson et al., 1998)’. In turn, Watson et al. (15) refer to the ‘immediate commitment to lysogeny or lysis following infection’ and cite Jakubczak et al. (16). These authors state that ‘Upon infection, the lytic-lysogenic decision is made before any DNA replication occurs (Herskowitz and Hagen, 1980)’. In contrast, however, Herskowitz and Hagen [(17), p. 405] state that ‘Early gene expression and the early mode of DNA replication [of phage lambda] are explicitly programmed and constitute the uncommitted [to either the lysogenic or lytic] phase of growth’. Therefore, some DNA replication does take place before the cII protein builds up to the amount required to commit the bacterial cell to lysogeny (which prevents formation of a plaque).
None of the assurances that only mutations fixed in vivo are recovered in the bacterial recovery of transgene mutations directly addresses the possibility that a G:T mismatch in mouse DNA, which results frequently from the spontaneous deamination of 5-methylcytosine or any other unrepaired mismatch, will be randomly repaired in E.coli to match one strand or the other before the first round of viral (or plasmid) DNA replication (18). Dohet et al. (19) showed that mismatches in the lambda cI gene in which the lambda phage DNA did not contain methylated adenines as the cue for bacterial mismatch repair are randomly repaired before the first round of DNA replication because the resulting plaques are pure rather than mosaic. For a G:T mismatch, about half of the resulting plaques are mutations to A:T. Bacteriophage lambda DNA derived from transgenic mice would have cytosine methylation and not the adenine methylation required for directed mismatch repair in E.coli (20).
The issue of when phage DNA replication occurs relative to mutation fixation is important because if bacterial mismatch repair randomly converted mismatches to either correct (wild type, not recovered) or mutant matches before the first round of lambda DNA replication, none of the methods presumed to preclude the recovery of ex vivo mutation would function. For lacI, the plaques would be pure, not mosaic, and for any selective assay, the mutation could be fixed before the selective mechanism took effect. Since a major portion of spontaneous mutation in transgenes consists of G:C 
A:T mutation at CpG sites (1), the question as to whether these mutations were fixed in bacteria from unrepaired deaminated 5-methylcytosine needs to be answered.
The X174 transgenic mutation assay has a unique method of identifying in vivo mutants, single-burst analysis, which is based on the number of progeny phage produced from a single, electroporated bacterium (21). Homoduplex phages have large burst sizes, 180 for both wild-type [average, 182.5 (22)] and UVB-induced mutants [average, 182.3, based on data of Valentine et al. (23)]. In addition, X174 does not contain the GATC tetranucleotide (24), which is the bacterial recognition sequence for initiating bacterial mismatch repair (25). Therefore, for X174 replicating in E.coli, there is no mismatch repair. Consequently, a mismatch entering the cell by electroporation would not be swiftly and randomly converted to a fixed mutation 50% of the time, but rather would produce equal amounts of wild-type and mutant progeny. Selection for X174 mutant recovery is applied after replication in this single cell and single-burst analysis would result in a mutant burst on average half the size of a homoduplex mutant. Whereas other systems might recover 50% of G:T mismatches as genuine mutations, the X174 single-burst analysis would recover all of them, each associated with a quality factor (burst size, estimated by plaque count per aliquot). An examination of plaque counts per aliquot for spontaneous mutation could identify mismatches fixed in E.coli.
We, therefore, undertook to evaluate both spontaneous and ENU-induced mutant frequencies in splenic lymphocytes of the Malling mouse (X174) by single-burst analysis from an experiment that was designed to mimic a definitive study with Big Blue® mice using the lacI transgene (26). We showed recently in transgenic cell culture that the X174 target gene, analysed by single-burst analysis of the forward mutational assay, responded to UVB with the same induced mutant frequency per target site as the cII transgene while having a spontaneous mutant frequency only one-tenth that of cII (23). The goal of this study was to determine whether single-burst analysis would produce comparable results in vivo.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionFundingSupplementary dataReferences |
|---|
Animals and ENU exposure
B6C3F1 male mice transgenic for
X174 were treated at 6 weeks of age with a single intra-peritoneal dose of 40 mg/kg ENU using dimethyl sulphoxide as the solvent and spleens were collected 3 weeks after dosing. These tissues were collected as part of a larger experiment that included four doses and two sampling times after dosing (18). Weights at dosing for solvent control animals varied from 9.7–13.7 g (11.7 ± 1.4) and for ENU-treated animals from 9.7–13.3 g (11.3 ± 1.3). All work with animals was conducted in an Association for Assessment and Accreditation of Animal Laboratory Care-accredited facility with Institutional Animal Care and Use Committee approval. Eight animals were in each treatment group initially; however, one animal in the ENU group died after intra-peritoneal injection and another sample in the ENU group was lost during workup. Therefore, the control group had eight animals and the ENU-treated group, six. Splenic lymphocytes were recovered and processed as described in Valentine et al. (18) for determining Hprt mutant frequencies. Excess cells from this assay were frozen for X174 transgene recovery.
Mutant recovery and sequencing
Transgene and mutant recovery was performed as described in Valentine et al. (18) using 2 µg DNA and two 48-well plates for 96 aliquots per electroporation. To summarize, the electroporated bacteria were diluted in rich medium, divided into aliquots and incubated for 1 h with shaking at 37°C. Lysozyme was added to each well and the lysates incubated with shaking for an additional hour after which chloroform was added to dissolve bacterial membranes and kill surviving cells. Small aliquots were removed from four selected wells of each plate for determination of total recovery and all samples were stored at –70°C. Dilutions of the small aliquots were replica plated on agar plates with the permissive E.coli strain CQ2; the remainder of each lysate from the 96 wells was plated on a separate agar plate with the selective strain of E.coli, gro89pLS1D. DNA sequencing also was performed as described in Valentine et al. (18) using additional primers for gene f as described in Valentine et al. (23). The sequencing primer for the 3' end of gene f was omitted in the latter reference and was 1510F: 5'-GACTGCTCCGCTTCCTCCTGAGA-3'.
Calculation of mutant frequencies
Mutant frequencies for X174 single-burst analysis were calculated by summing all mutant plaque counts from only those aliquots with 
90 mutant plaques and for which a single mutation accounted for 50% of the mutants sequenced from that aliquot. When a mutation occurred singly and also with a second mutation within the same aliquot, all plaques with that mutation were counted for meeting the 50% criterion. Occasionally, mutants that were sequenced only in the 5' region of gene a with no mutation identified were assumed to have the same mutation as other plaques in the same aliquot sequenced outside this region that also had no mutation identified (plaque counts of 325 and 196) or were assumed to be the same as a mutation identified outside of 5' gene a (plaque counts of 458, 196, 189 and 153). For the aliquot with 325 total mutant plaques, the six mutants for which no mutation was found in the sequenced regions were assumed to have the same mutation, which was 50% of the total sequenced plaques.
Mutant frequencies were the ratio of the summed total number of mutant plaque-forming units (PFUs) from aliquots with 90 mutant PFUs divided by the total number of plaques recovered from non-selective plates (corrected for dilutions). This is a calculation of mutant frequencies by PFUs as opposed to by bursts, in which the total number of genomes recovered is the denominator and the total number of in vivo mutant bursts is the numerator (18).
Mutant frequencies from published literature references were recalculated from raw data to avoid rounding errors for the trend analysis. Only the report of Hill et al. (27) did not contain the original data, with only the mean (1.3 x 10–5) and standard error (SE) (0.4) for four measurements reported. Therefore, the four values of 1.9928, 1.9928, 0.6072 and 0.6072 x 10–5 were chosen arbitrarily to produce that mean and SE.
For the cII gene, the data of Monroe et al. (28) appear to contain an error in the mutant frequency for animal 0I: it is given as 8 x 10–6 in Table I even though the number of blue plaques is 3 and the total plaques screened is 260 400, an actual frequency of 11.5 x 10–6. We used the latter value calculated from the raw data.
|
For both cII and lacI calculations from the data of Zimmer et al. (29
), the raw data for each animal were combined among packaging reactions and the mean and variance were determined by animal. Since cI mutations were not distinguished from cII mutations by sequencing, the proportion of independent spontaneous base pair substitution (bps) mutants in the cII gene for the data of Zimmer et al. (29) and Wang et al. (30) was estimated from the data of Monroe et al. (28) (Table V). We know of no mutant spectrum data in ENU-treated spleen for the cII gene; based on the lacI data (Table V), we would expect ENU-induced mutations to have a greater proportion of independent and bps substitution than spontaneous mutations. In order to compare bps mutation frequencies per nucleotide at the cII target from ENU-treated animals to lacI and X174 (Table VI), we used the combined ratio conversion factor from spontaneous spectra (Table V) with the expectation that this produces a lower limit on the estimate of mutation frequency and that the frequency may be higher if appropriate spectral data were available.
Clonal correction
When the same mutation was found more than once in the same animal, only one occurrence was counted for clonal correction. For calculating mutation frequencies (clonally corrected), rather than randomly picking one of the aliquots and using its plaque count, the average number of plaques from all aliquots with that mutation was used as the plaque count for the clonally corrected, single occurrence. This average number was assigned to the first electroporation of Table III in which that mutation occurred.
Trend analysis
A linear regression was fit: bps mutation frequency per nucleotide versus age in weeks. Data from each of the target genes were assumed to have the same slope with age, but their intercepts were allowed to vary. The non-lacI assays were tested for their fit to the trend by comparing the intercept to that of the lacI trend. P-values were adjusted for multiple comparisons by the Tukey method. The statistical calculations were done using PROC GLM of SAS (31). SEs are reported throughout the manuscript.
Mutant spectra were compared for significant differences by the Monte Carlo algorithm of Cariello et al. (32).
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionFundingSupplementary dataReferences |
|---|
Single-burst analysis and selection of in vivo cut-off based on plaque counts per aliquot
Single-burst analysis is conducted by dividing a single electroporated sample into multiple aliquots (21
), in this study, 96 aliquots. Each aliquot is subsequently plated on a separate agar plate under selective conditions. The mutant plaque count for all aliquots with >80 mutant plaques is shown in Table I with the mutations identified from all except two of these aliquots. In addition, mutations from selected aliquots with fewer plaque counts are included in the table from both control and ENU-treated samples.
Inspection of sequencing results for the control samples (Table I) indicated that <90 mutant PFUs, the 4225AG mutation dominated the spectrum. This is the hotspot mutation for E.coli mutagenesis in this forward mutational assay for X174 [>50% of spontaneous E.coli mutations recovered are this mutation (33)]. The 4225AG mutation did not appear as a predominant mutation in any aliquot from control samples with a plaque count of >90, although it did occur twice among the ENU samples (plaque counts 325 and 123). A:T G:C transitions are characteristic of ENU-induced mutagenesis, but this mutation did not dominate the ENU-induced spectrum (Table II). Therefore, in order to avoid including mutations that might have had an origin in E.coli, we used 90 mutant PFUs as the cut-off for in vivo mutation.
|
Effect of plating density on distribution of mutant bursts
Our previous studies (18
,23) indicated that a plating density of 
4 x 105 total PFUs per aliquot (400 x 105 total PFUs per electroporation with 96 aliquots) was desirable in order to separate larger E.coli bursts into separate aliquots so that together they would not produce a total plaque count above the cut-off for in vivo bursts. We used 2 µg DNA to accomplish this goal, but occasionally total PFUs >400 x 105 were obtained (Table III). Nevertheless, independence of mutations was established by sequencing all (but one) the aliquots with 90 mutant plaques. The one aliquot (127) without sequencing was from an electroporation with 530 x 105 total mutant plaques, which had six aliquots with 90 mutant plaques; therefore, it was assumed to have a mutation that was not independent. All aliquots with mutant plaque counts 90 had a mutation that accounted for 50% of the total sequenced plaques. We found only one aliquot with two identified in vivo mutations, based on the assumption that all the mutants for which no mutation was found had the same mutation (from an ENU-treated sample, 325 PFUs). Overall, the low plating density used here was effective for heterozygous X174 transgenic animals for separating in vivo bursts and larger E.coli bursts, while yet detecting multiple mutations per electroporation.
|
Mutant and mutation spectra
Mutant and mutation spectra of in vivo mutation for control and ENU-treated mice are shown in Table II. From eight control animals, nine independent mutations (of 13) were found at four different identified sites (plus at least one but no more than six additional unidentified sites). From six ENU-treated animals, 39 independent mutations (of 49) were found at 29 different identified sites (plus at least one but no more than six additional unidentified sites). Twenty-eight per cent (11 of 39) of the mutations from ENU samples were not in the 5' end of gene a (nucleotides 3981–4480). Several mutations were identified in gene f, an additional gene known for recovery of mutations by this selection method (34
). Two other mutations (4887TC and 5359TG) were found in the 3' region of gene a, which is a large gene (nucleotides 3981-136, 1542 nucleotides in a circular genome of 5386 nucleotides). Not all mutants with no mutation in the 5' end of gene a were sequenced in the 3' region of gene a and their mutations remained unidentified. Thirty-three per cent (3 of 9) of control mutations occurred outside the 5' end of gene a (also unidentified). In contrast, although the E.coli spectrum contains 154 independent mutations, these occur at only 16 identified sites and only 2 are outside the 5' end of gene a [not identified, 1.3% (33)]. The full spectra for independent (identified) mutations from control and ENU samples from spleen were both highly different from the spontaneous E.coli spectrum (P < 10–6). However, the control and ENU spectra were not different from each other (P = 0.096), probably because of the relatively few spontaneous mutants.
The summary of the types of independent mutations recovered from control and treated animals is shown in Table IV compared to the independent, spontaneous mutation spectrum in E.coli. The control spectrum was notable for having 50% of the identified, independent mutations at a single CpG site (G:C A:T), whereas the E.coli spectrum has none in this category, and the ENU-treated spectrum had only 6%. The spectrum from ENU-treated animals was dominated by A:T G:C and G:C A:T. Next most common were A:T T:A and A:T C:G. In contrast to the full spectra, the difference between the control and ENU summary spectra was marginally significant (P = 0.045). The E.coli summary spectrum is also highly different from the control and ENU spleen summary spectra (P < 10–6). The E.coli spectrum for the X174 forward mutational assay is dominated by A:T G:C because that includes the E.coli hotspot, 4225AG.
|
Newly described target sites
Fifteen new target sites for the
X174 forward mutational assay were identified in this study, of both mouse and bacterial origin (Supplementary Table S1 is available at Mutagenesis Online). This brings the total number of identified target sites to 57, which are listed at the following public website: http://www.fda.gov/nctr/science/divisions/documents/genreptox_docs.htm).
One of these new mutations was a triple bps (2311GC and 2314, 2315GATC, plaque count 255), between genes f and g. Even though the mutation was of bacterial origin because it occurred in only one sequenced mutant, the two Gs were both at CpG sites. All three bps(s) were considered a single, CpG target site.
Mutant and mutation frequencies
Mutant and mutation (clonal correction) frequencies of in vivo mutation in X174 from splenic lymphocytes are shown in Table III. The mutation frequency in X174 for control samples was 0.18 ± 0.07 x 10–5 and for ENU-treated samples, 1.2 ± 0.3 x 10–5. The mutation frequency declined 28% from mutant frequency for controls and 20% for treated samples. The control Hprt mutant frequency was 0.41 ± 0.06 x 10–5 and the ENU-treated Hprt mutant frequency was 22.6 ± 2.3 x 10–5.
Conversion of published mutant frequencies to mutation frequencies for bps per nucleotide
In order to compare our data, which were corrected for clonal expansion to mutation frequency and included only bps, to that of published mutation studies using lacI, cII, and Hprt, we needed all data expressed as bps mutation frequencies per nucleotide. Mutant frequencies from published reports in mouse spleen were converted to an estimate of this unit by the application of conversion factors, which were based on the proportion of bps and independent mutations in published spectra and the total number of known target sites for each transgene (Table V).
Studies reporting mutant frequencies in spleen for the lacI transgene were chosen that included four or more mouse spleen samples per data point (26,28,29,35). Data from embryos were included (27) in order to determine the age-related trend since the other studies were done at varying ages. The mutation frequency for mouse embryo was converted to bps mutation frequency based on the mutational spectrum in that publication.
For cII, the mutant frequencies of Monroe et al. (28) are corrected for clonality; however, the listing of mutations is not. It includes all sequenced mutations identified by dose group, but not by animal. This makes the calculation of the proportion of total mutations that are bps ambiguous. Four different mutations that are not bps(s) occur 10 times in the spontaneous group, an insertion three times and a deletion five times. If all are independent, the ratio of bps is 95/105 = 0.904. If only four are independent, the ratio is 101/105 = 0.961. Based on the appearance of these mutations in other dose groups, we have assumed that seven non-bps mutations out of 10 were independent: 98/105 = 0.9333, which is effectively the mean of the possible range. Both cII and cI mutants were sequenced; only cII frequencies were used here.
In contrast, for cII mutant frequencies of Zimmer et al. (29) and Wang et al. (30), no sequencing was done to distinguish cII from cI. Therefore, three conversion factors for these studies were used for independence, bps, and cII mutation based on the data of Monroe et al. (28) (Table V).
Spontaneous mutations from the Hprt gene were considered independent because of their low frequency. A correction factor for Hprt bps was based on the report of Aidoo et al. (36) for rats (Table V) since the murine spectra of Walker et al. (35) contained few spontaneous mutants. This factor provided only an upper limit for bps since only large deletions were eliminated from the overall mutant frequency and all remaining were not necessarily bps. The spectra of Walker et al. (35) were used for ENU-treated mice for converting Hprt frequencies.
|
Based on these conversion factors (Table V), bps mutation frequencies per nucleotide in mouse spleen were calculated at several ages for lacI and cII and at 6 weeks for
X174 and Hprt; all were compared to the human germ line bps mutation frequency per nucleotide (Table VI).
|
Comparison of mutation frequencies per nucleotide between studies
A trend analysis of the spontaneous bps mutation frequency per nucleotide with age for the lacI transgene in spleen was calculated (Figure 1) and found to be significant (P = 7 x 10–4). This trend extrapolated to a bps mutation frequency per nucleotide at conception of 1.8 ± 0.9 x 10–8 and produced an increase of 0.28 x 10–8 per week to 12 weeks of age. Figure 1 also compares the lacI trend to
X174, Hprt, and the human germ line assuming that the human data correspond to the time of conception. Based on the appropriate age, neither the X174 transgene (P = 0.16) nor the human germ line (P = 0.94) bps mutation frequencies per nucleotide (1.7 ± 0.2) were significantly different from the lacI trend. However, the Hprt frequencies from both this study (P = 3 x 10–4) and that of Skopek et al. (26) were significantly lower than the lacI trend (P = 4 x 10–7). Even without age consideration, the Hprt frequency of Skopek et al. (26) was significantly lower than the human germ line frequency (t-test on Table VI, P = 0.018). (The Hprt data of this study were not significantly lower, P = 0.088.)
|
The response to ENU for the
X174 transgene in spleen was compared to that of lacI with the similar study design of Skopek et al. (26) based on the spontaneous age-related lacI trend. The response was not significantly different (P = 0.81). The fold increase for this X174 study was 6.9 and for lacI, 8.7 (Table VI).
In contrast, the cII transgene in the same tissue and at the same age (28) has a 10 times higher spontaneous bps mutation frequency per nucleotide (Table VI, P < 10–8). Further, this high frequency does not increase with age from 6 to 30 weeks (29,30), Table VI. The response of cII to ENU has not been reported at the same dose of ENU as for lacI and X174; however, comparable fold increases are induced at doses two to three times as high (Table VI). Although the fold increase may not be as great as for the other transgenes, induced mutant frequencies for cII are nevertheless 10–70 times higher (depending on which studies are compared, Table VI) because of the high spontaneous frequencies.
The percentage of G:C A:T mutation at CpG sites for the spontaneous spectrum in cII of Monroe et al. (28) (Discussion) was calculated from the raw data (Table V, [28]) because Table 7 of that reference has some errors in tabulation. The total number of G:C A:T mutations is 77, not 83 and Table V incorrectly identifies some sites as CpG (95, 101, 160, 163 and 223), which are not (they are GpC, based on the sequence of the lambda cII gene and surrounding regions in the Big BlueTM manual supplied by Stratagene, La Jolla, CA) with numbering starting at the initiating ATG). Also, some CpG sites are not identified: 196 and 214. The total number of G:C A:T mutations at CpG sites by our count is 16, or 21% of the total G:C A:T mutations.
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionFundingSupplementary dataReferences |
|---|
Calculation of mutation frequencies by PFUs
We calculated mutation (and mutant) frequencies (Table III) by the method of PFUs rather than by bursts because previous work indicated the former method produced greater significance between treated and untreated samples (18
). However, the two methods produced similar frequencies for these data as shown by the following calculations by bursts for the data of this study: controls—nine independent mutant bursts/44.5 x 105 genomes (Table III) = 0.20 x 10–5 mutation frequency compared to 0.18 x 10–5 by PFUs; ENU animals—39 independent mutant bursts/34.8 x 105 genomes = 1.1 x 10–5 by bursts compared to 1.2 x 10–5 by PFUs.
Mutation spectra
Even though the difference between the control and ENU mutation summary spectra was of marginal statistical significance because of the small number of spontaneous mutations obtained, both spectra were characteristic of their type based on other transgenic systems (1): spontaneous (G:C A:T at CpG sites) and ENU-induced spectra (G:C A:T, A:T G:C and A:T T:A transitions and transversions).
In spite of the small number of CpG target sites known for the X174 forward mutation assay (three in single bps(s) and one in a triple bps), one-third of the nine independent spontaneous mutations were at the same CpG site, 4162GA, the most frequently recovered CpG site from mouse control samples in previous 
X174 studies using the forward assay (18,23,37). We are confident that these were all in vivo mutants because three of four of these mutations isolated from control mice were from aliquots with >300 mutant PFUs and almost all sequenced mutants contained the same mutation. The fourth mutant was from an aliquot with plaque count of 160, but was from one of the same animals with a larger burst and therefore was effectively removed during clonal corrections. The frequency of G:C A:T at CpG sites in control animals in this study (33%) is similar to the 38% found by Walker et al. (35) and is characteristic of spontaneous mutation from a variety of transgenic systems (1). Therefore, we conclude that this characteristic of spontaneous mutation spectra is, indeed, a reflection of in vivo mutation. This also agrees with the observation that human germ line mutation frequencies are 10 times higher at CpG sites than non-CpG sites (38).
We found no evidence that any of the recovered spontaneous mutations for lacI in spleen at early ages derive from mismatch repair in E.coli. However, we cannot rule out the possibility that the lower (not significantly) mutation frequency observed here with X174 at 6 weeks predicts a significantly lower mutation frequency by 12 weeks of age, particularly since the discovery of any additional X174 target sites would lower the mutation frequency per nucleotide for the present data [the spectrum is probably not saturated (23)]. If there are as many as 65 target sites for X174 (up from 57), then the mutation frequency per nucleotide at 6 weeks in this study would be 2.7 ± 1.0 x 10–8 significantly (although marginally) lower than the lacI trend (P = 0.048). A line through the X174 and human data points (Figure 1) would predict that the X174 mutation frequency might be two-thirds of lacI at 12 weeks. If such was the case, this could suggest some accumulation of unrepaired damage recovered by the lacI system that would account for, at most, one-third of the lacI mutations at this age. However, such a modest, hypothesized difference also could be explained as differences in mutation frequencies at the specific target sites of the two transgenes.
The intergenic region containing the triple mutant 2311GC and 2314, 2315GATC forms a hairpin structure that functions as a binding site to initiate assembly of the primosome, a complex that initiates the first round of DNA replication of single-stranded viral DNA [(39), Fig. 9]. Strikingly, these three substitutions eliminate two bulges in the hairpin creating perfect matches throughout the stem structure. A direct interaction between this DNA structure and the protein product of the rep gene of E.coli [which is mutated in the selective strain and provides the basis for recovery of X174 mutants (37)] is not known (39). However, we have shown with UVB mutagenesis (23) that this selection can retrieve a mutant in gene c; this gene product (gpC) is hypothesized (39) to interact with the RFII–gpA–rep complex for viral DNA synthesis and packaging into the phage head (this complex provides the basis for obtaining compensatory mutants in gene a to the mutant rep protein). Since gpC is also part of the complex that assembles at the hairpin, an indirect interaction between the rep protein and the hairpin may be indicated. Alternatively, an additional role of either the hairpin loop or the rep protein may exist.
Comparison of bps per nucleotide mutation frequencies
The lacI mutation frequency trend with age was found to intercept the bps mutation frequency per nucleotide of the human germ line (1.7 x 10–8) when extrapolated to the time of conception. The human and mouse germ line mutation rates are considered to be similar (40). An additional 6 weeks of development after conception produced a similar additional mutational burden in spleen (6 weeks x 0.28 x 10–8/week = 1.68 x 10–8 mutations). Six weeks is approximately one generation because although sexual maturity in mouse occurs at 8–12 weeks, oogenesis in female mammals for the next generation is completed during gestation, which is 3 weeks for mouse. In view of this fact, the somatic mutation rate observed for the lacI transgene is comparable to the human (and presumably mouse) germ line mutation rate. This linear increase in mutation frequency continues until 12 weeks of age; after this time, the data of Hill et al. (27) indicate that a plateau occurs in mutation frequency in other tissues, corresponding to the cessation of rapid growth. A linear mutant increase is expected for a population of growing cells that have been cleansed of mutants at the start of growth (41,42), which is the case for germ cells since a whole mouse is derived from a one-cell zygote (a mutation in the zygote would produce a recognizably high mutation frequency related to the number of copies of transgene in the animal).
Therefore, we conclude that the spontaneous mutation frequencies detected by the lacI and X174 transgenes are not unexpectedly high. Rather, the relatively high observed mutant frequencies from mutagenesis studies may reflect primarily the accumulation of somatic mutation, for which no selection pressure acts either to remove or enhance. Measured Hprt mutation frequencies have been suggested to be underestimates of actual mutation frequencies because of selection pressure against Hprt mutants in vitro (43) and in vivo (44). Human HPRT mutant frequencies in cord blood are approximately 10-fold lower than the human germ line mutation rate, by the following calculations. The HPRT mutant frequency is 0.64 x 10–6 in 45 normal newborns (45) and 1.2 x 10–6 for 60 infants unexposed to 3'-azido-3'-dideoxythymidine (46). Since 35% of these are V-(D)-J recombinase-mediated deletions (47), the upper limit on the bps mutation frequency is 0.4–0.8 x 10–6. Dividing by 312 target sites produces a maximum 1–2.5 x 10–9 bps per nucleotide in the HPRT gene of umbilical cord blood of normal human newborns compared to 1.7 x 10–8 for the human germ line. This, and the fact that somatic Hprt mutation frequencies at 6 weeks of age in mice are similar to or lower than for the human germ line, contributes to the interpretation that a strong selective pressure exists against Hprt mutants in vivo.
Summary of conclusions
Consequently, because the trend of spontaneous, per nucleotide mutation frequencies with age for the lacI gene extrapolates at the time of conception to the average human germ line mutation rate, because this trend increases after birth at a comparable rate and because the demonstrated in vivo X174 spontaneous mutation frequency and spectrum was similar to lacI at 6 weeks of age, we conclude that the lacI gene accurately measures in vivo spontaneous bps mutation frequencies at early ages in spleen, a tissue with moderate turnover. In such a tissue, mismatched bases arising from spontaneous deamination of methylated cytosine may be efficiently repaired before tissue collection (3 weeks in this study) because of DNA synthesis and cell division (48). Our results do not rule out the possibility that some other tissue might not repair DNA promptly. It has been shown in heart that the mutant frequency induced by the alkylating mutagen ENU in the lacZ transgene increases significantly between 35 and 70 days after exposure following a plateau (9). This result suggests that DNA damage may be long lived and slowly repaired in this tissue, which has a low mitotic index (9).
Since the only mutation recovered as G:C A:T at a CpG site for the X174 forward mutation assay from control animals was found in this study to have an average burst size of 250 PFUs, an increase in the number of half-sized bursts of this mutation in another tissue relative to that in spleen should be recognizable. Therefore, a study with the X174 transgenic mutation assay in heart might determine whether unrepaired G:T mismatches contribute to recovery of spontaneous G:C C:T mutation in a slowly dividing tissue.
In contrast, cII bps spontaneous mutation frequencies per nucleotide are strikingly high compared to lacI or X174 (10 times higher). If the high mutant frequencies for cII are a consequence of ex vivo mutation, it must be through a mechanism common both to spontaneous and induced mutation and not based on spontaneous deamination of CpG sites. In fact, for the spontaneous cII spectrum (28), the proportion of G:C A:T mutation at CpG sites is only 21% (our calculations, see Results) compared to 77% for lacI (35) or 100% for X174 (Table I).
If the high spontaneous and induced frequencies in the cII gene are from mutation fixed in vivo, this target seems to have some inherent propensity to mutation. Bacteriophage genes have a higher per nucleotide mutation frequency than bacterial genes in their natural host [because of their smaller genome size (49); cII is a bacteriophage lambda gene and lacI is an E.coli gene]. However, the X174 viral transgene does have a relatively high mutant frequency in E.coli (33) and has a genome nine times smaller than lambda, but is shown in this study to have a mutation frequency as a transgene comparable to mammalian cells. The popularity of the cII transgene is increasing because of its selective assay, short DNA sequence and prolific production of mutations; however, the origin of its high mutation frequencies is not known.
|
Funding |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionFundingSupplementary dataReferences |
|---|
US Food and Drug Adminstration.
|
Supplementary data |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionFundingSupplementary dataReferences |
|---|
Supplementary Table S1 is available at Mutagenesis Online.
|
Acknowledgments |
|---|
We acknowledge the expertise, encouragement and intellectual stimulation of Dr Heinrich V. Malling, who inspired this work. The contents of this manuscript do not necessarily reflect the views and policies of the US Food and Drug Administration or any other agency of the US government. The mention of trade names or commercial products does not constitute endorsement or recommendation for use.
Conflict of interest statement: None declared.
|
Notes |
|---|
* To whom correspondence should be addressed: Tel: +1 501 686 8849; Fax: +1 501 686 5845; Email: rdelongchamp{at}uams.edu
3 Present address: Cincinnati Children's Hospital Medical Center, Division of Allergy and Immunology, Cincinnati, OH 45040, USA 
4 Present address: University of Arkansas for Medical Sciences, School of Medicine, Little Rock, AR 72205, USA
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionFundingSupplementary dataReferences |
|---|
-
1. Lambert IB, Singer TM, Boucher SE, Douglas GR. Detailed review of transgenic rodent mutation assays. Mutat. Res. (2005) 590:1–280.[CrossRef][Web of Science][Medline]
2. Hill KA, Buettner VL, Glickman BW, Sommer SS. Spontaneous mutations in the Big Blue® transgenic system are primarily mouse derived. Mutat. Res. (1999) 436:11–19.[CrossRef][Web of Science][Medline]
3. Paashuis-Lew Y, Zhang XB, Heddle JA. On the origin of spontaneous somatic mutations and sectored plaques detected in transgenic mice. Mutat. Res. (1997) 373:277–284.[Web of Science][Medline]
4. Sui H, Suzuki M, Yamada M, Hara T, Kawakami K, Shibuya T, Nohmi T, Sofuni T. Effects of O6-alkylguanine-DNA alkyltransferase deficiency in Escherichia coli as the host for the detection of mutations in lacI transgenic mice. Environ. Mol. Mutagen. (1999) 34:221–226.[CrossRef][Web of Science][Medline]
5. Piegorsch WW, Margolin BH, Shelby MD, Johnson A, French JE, Tennant RW, Tindall KR. Study design and sample sizes for a lacI transgenic mouse mutation assay. Environ. Mol. Mutagen. (1995) 25:231–245.[Web of Science][Medline]
6. Swiger RR, Cosentino L, Shima N, Bielas JH, Cruz-Munoz W, Heddle JA. The cII locus in the MutaTMMouse system. Environ. Mol. Mutagen. (1999) 34:201–207.[CrossRef][Web of Science][Medline]
7. Gossen JA, de Leeuw WJF, Tan CHT, Zwarthoff EC, Berends F, Lohman PHM, Knook DL, Vijg J. Efficient rescue of integrated shuttle vectors from transgenic mice: a model for studying mutations in vivo. Proc. Natl Acad. Sci. USA (1989) 86:7971–7975.
8. Douglas GR, Jiao J, Gingerich JD, Gossen JA, Soper LM. Temporal and molecular characteristics of mutations induced by ethylnitrosourea in germ cells isolated from seminiferous tubules in spermatozoa of lacZ transgenic mice. Proc. Natl Acad. Sci. USA (1995) 92:7485–7489.
9. Cruz-Munoz W, Kalair W, Cosentino L, Heddle JA. ENU induces mutations in the heart of lacZ transgenic mice. Mutat. Res. (2000) 469:23–34.[Web of Science][Medline]
10. Burkhart JG, Burkhart BA, Sampson KS, Malling HV. ENU-induced mutagenesis at a single A:T base pair in transgenic mice containing X174. Mutat. Res. (1993) 292:69–81.[Web of Science][Medline]
11. Bielas JH, Heddle JA. Quiescent murine cells lack global genomic repair but are proficient in transcription-coupled repair. DNA Repair (2004) 3:711–717.[Medline]
12. Ikehata H, Ono T. Mutation induction with UVB in mouse skin epidermis is suppressed in acute high-dose exposure. Mutat. Res. (2002) 508:41–47.[Web of Science][Medline]
13. Reddy M. Positive selection system for identification of recombinants using 
-complementation plasmids. Biotechniques (2004) 37:948–952.[Web of Science][Medline]
14. Willems MI, van Benthem J. Mutagenicity of Chemicals in Genetically Modified Animals (2000) RIVM Report No. 650210, TNO Report No. V99.1097, pp. 1–47.
15. Watson DE, Cunningham ML, Tindall KR. Spontaneous and ENU-induced mutation spectra at the cII locus in Big Blue® rat2 embryonic fibroblasts. Mutagenesis (1998) 13:487–497.
16. Jakubczak JL, Merlino G, French JE, Muller WJ, Paul B. Analysis of genetic instability during mammary tumor progression using a novel selection-based assay for in vivo mutations in a bacteriophage 
transgene target. Proc. Natl Acad. Sci. USA (1996) 93:9073–9078.
17. Herskowitz I, Hagen D. The lysis-lysogeny decision of phage : explicit programming and responsiveness. Ann. Rev. Genet. (1980) 14:399–445.[CrossRef][Web of Science][Medline]
18. Valentine CR, Raney JL, Shaddock JG, Dobrovolsky VN, Delongchamp RR. In vivo mutation in gene A of splenic lymphocytes from X174 transgenic mice. Environ. Mol. Mutagen. (2004) 44:128–150.[CrossRef][Web of Science][Medline]
19. Dohet C, Wagner R, Radman M. Repair of single base-pair mismatches in Escherichia coli. Proc. Natl Acad. Sci. USA (1985) 82:503–505.
20. Hattman S. DNA methylation. In: The Enzymes—Boyer PD, ed. (1981) New York: Academic Press. 517–548. Vol. 14.
21. Malling HV, Delongchamp RR, Valentine CR. Three origins of X174am3 revertants in transgenic cell culture. Environ. Mol. Mutagen. (2003) 42:259–273.
22. Delongchamp RR, Valentine CR, Malling HV. Estimation of the average burst size of X174 am3, cs70 for use in mutation assays with transgenic mice. Environ. Mol. Mutagen. (2001) 37:356–360.[CrossRef][Web of Science][Medline]
23. Valentine CR, Rainey HF, Farrell JM, Delongchamp RR, Howard PC, Hass BS. Sensitivity of UVB-induced mutant detection in the X174 transgenic forward mutation assay. Genes Environ. (2007) 29:38–53.[CrossRef]
24. Sanger F, Coulson AR, Friedman T, et al. The nucleotide sequence of bacteriophage X174. J. Mol. Biol. (1978) 125:225–246.[CrossRef][Web of Science][Medline]
25. Marti TM, Kunz C, Fleck O. DNA mismatch repair and mutation avoidance pathways. J. Cell. Physiol. (2002) 191:28–41.[CrossRef][Web of Science][Medline]
26. Skopek TR, Kort KL, Marino DR. Relative sensitivity of the endogenous hprt gene and lacI transgene in ENU-treated Big BlueTM B6C3F1 Mice. Environ. Mol. Mutagen. (1995) 26:9–15.[Web of Science][Medline]
27. Hill KA, Buettner VL, Halangoda A, Kunishige M, Moore SR, Longmate J, Scaringe WA, Sommer SS. Spontaneous mutation in Big Blue® mice from fetus to old age: tissue-specific time courses of mutation frequency but similar mutation types. Environ. Mol. Mutagen. (2004) 43:110–120.[CrossRef][Web of Science][Medline]
28. Monroe JJ, Kort KL, Miller JE, Marino DR, Skopek TR. A comparative study of in vivo mutation assays: analysis of hprt, lacI, and cII/cI as mutational target for N-nitroso-N-methylurea and benzo[a]pyrene in Big BlueTM mice. Mutat. Res. (1998) 421:121–136.[Web of Science][Medline]
29. Zimmer DM, Harbach PR, Mattes WB, Aaron CS. Comparison of mutant frequencies at the transgenic lambda lacI and cII/cI loci in control and ENU-treated Big Blue® mice. Environ. Mol. Mutagen. (1999) 33:249–256.[CrossRef][Web of Science][Medline]
30. Wang J, Liu X, Heflich RH, Chen R. Time course of cII gene mutant manifestation time in the liver, spleen, and bone marrow of N-ethyl-N-nitrosourea-treated Big Blue transgenic mice. Toxicol. Sci. (2004) 82:124–128.
31. SAS Institute Inc. SAS/STAT User's Guide, Version 8 (1999) Cary, NC: SAS Institute Inc. 3884.
32. Cariello NF, Piegorsch WW, Adams WT, Skopek TR. mutations. Carcinogenesis (1994) 15:2281–2285.
33. Raney JL, Delongchamp RR, Valentine CR. Spontaneous mutant frequency and mutation spectrum for gene A of X174 grown in E. coli. Environ. Mol. Mutagen. (2004) 44:119–127.[CrossRef][Web of Science][Medline]
34. Ekechukwu MC, Oberste DJ, Fane BA. Host and X174 mutations affecting the morphogenesis or stabilization of the 50S complex, a single-stranded DNA synthesizing intermediate. Genetics (1995) 140:1167–1174.[Abstract]
35. Walker VE, Gorelick NJ, Andrews JL, Craft TR, deBoer JG, Glickman BW, Skopek TR. Frequency and spectrum of ethylnitrosourea-induced mutation at the hprt and lacI loci in splenic lymphoctyes of exposed lacI transgenic mice. Cancer Res. (1996) 56:4654–4661.
36. Aidoo A, Mittelstaedt RA, Bishop ME, Lyn-Cook LE, Chen Y-J, Duffy P, Heflich RH. Effect of caloric restriction on Hprt lymphocyte mutation in aging rats. Mutat. Res. (2003) 527:57–66.[Web of Science][Medline]
37. Valentine CR, Montgomery BA, Miller SG, Delongchamp RR, Fane BA, Malling HV. Characterization of mutant spectra generated by a forward mutational assay for gene A of X174 from ENU-treated transgenic mouse embryonic cell line PX-2. Environ. Mol. Mutagen. (2002) 39:55–68.[CrossRef][Web of Science][Medline]
38. Kondrashov AS. Direct estimates of human per nucleotide mutation rates at 20 loci causing Mendelian diseases. Hum. Mutat. (2002) 21:12–27.[CrossRef][Web of Science]
39. Hayashi M. Biology of the bacteriophage X174. In: The Bacteriophages—Calendar R, ed. (1988) New York: Plenum Publishing Corporation. 1–71.
40. Drost JB, Lee WR. Biological basis of germline mutation: comparison of spontaneous germline mutation rates among Drosophila, mouse and human. Environ. Mol. Mutagen. (1995) 25:48–64.[CrossRef][Web of Science][Medline]
41. Rossman TG, Goncharaova EI, Nádas A. Modeling and measurement of the spontaneous mutation rate in mammalian cells. Mutat. Res. (1995) 328:21–30.[Web of Science][Medline]
42. Bachl J, Dessing M, Olsson C, von Borstel RC, Steinberg C. An experimental solution for the Luria-Delbruck fluctuation problem in measuring hypermutation rates. Proc. Natl Acad. Sci. USA (1999) 96:6847–6849.
43. Khaidakov M, Glickman BS. Possible factors leading to a misjudgement of mutant frequencies in HPRT assay. Mutat. Res. (1996) 354:9–14.[Web of Science][Medline]
44. Deubel W, Bassukas ID, Schlereth W, Lorenz R, Hempel K. Age-dependent selection against HPRT deficient T lymphocytes in the HPRT+/– heterozygous mouse. Mutat. Res. (1996) 351:67–77.[Web of Science][Medline]
45. McGinniss MJ, Falta MT, Sullivan LM, Albertini RJ. In vivo hprt mutant frequencies in T-cells of normal human newborns. Mutat. Res. (1990) 240:117–126.[CrossRef][Web of Science][Medline]
46. Meng Q, Olivero O, Fasco MJ, et al. Plasma and cellular markers of 3'-azido-3'-dideoxythymidine (AZT) metabolism as indicators of DNA damage in cord blood mononuclear cells from infants receiving prepartum NRTIs. Environ. Mol. Mutagen. (2007) 48:307–321.[CrossRef][Web of Science][Medline]
47. Keohavong P, Xi L, Day RD, Zhang L, Grant SG, Day BW, Ness RB, Bigbee WL. HPRT gene alterations in umbilical cord blood T-lymphocytes in newborns of mothers exposed to tobacco smoke during pregnancy. Mutat. Res. (2005) 572:156–166.[Web of Science][Medline]
48. Bielas JH, Heddle JA. Proliferation is necessary for both repair and mutation in transgenic mouse cells. Proc. Natl Acad. Sci. USA (2000) 97:11391–11396.
49. Drake JW. A constant rate of spontaneous mutation in DNA-based microbes. Proc. Natl Acad. Sci. USA (1991) 88:7160–7164.
50. Chen T, Aidoo A, Manjanatha M, et al. Comparison of mutant frequencies and types of mutations induced by thiotepa in the endogenous Hprt gene and transgenic lacI gene of Big Blue® rats. Mutat. Res. (1998) 403:199–214.[Web of Science][Medline]
Received on February 20, 2008; revised on April 18, 2008; accepted on April 21, 2008.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||