• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. May 15, 2007; 104(20): 8403–8408.
Published online May 7, 2007. doi:  10.1073/pnas.0610902104
PMCID: PMC1895962
From the Cover
Genetics

Evidence for mutation showers

Abstract

Mutants in the Big Blue transgenic mouse system show spontaneous clustered multiple mutations with unexpectedly high frequency, consistent with chronocoordinate events. We tested the prediction that the multiple mutations seen within the lacI mutation target sometimes occur in the context of chronocoordinate multiple mutations spanning multiple kilobases (mutation showers). Additional sequencing of mutants was performed in regions immediately flanking the lacI region (total of 10.7 kb). Nineteen additional mutations were found outside the lacI region (“ectomutations”) from 10 mutants containing two or more lacI mutations, whereas only one ectomutation was found in 130 mutants with a single mutation (P < 0.0001). The mutation showers had an average of approximately one mutation per 3 kb. Four mutants showed closely spaced double mutations in the new sequence, and analysis of the spacing between these mutations revealed significant clustering (P = 0.0098). To determine the extent of the mutation showers, regions (8.5 kb total) remote from the lacI region (≈16–17 kb away) were sequenced. Only two additional ectomutations were found in these remote regions, consistent with mutation showers that generally do not extend more than ≈30 kb. We conclude that mutation showers exist and that they constitute at least 0.2% and possibly 1% or more of mutational events observed in this system. The existence of mutation showers has implications for oncogenesis and evolution, raising the possibilities of “cancer in an instant” and “introns as sponges to reduce the deleterious impact of mutation showers.”

Keywords: Big Blue mouse, lacI transgene, multiple mutations, mutation clusters, transient hypermutability

Cancer is characterized by an accumulation of somatic mutations. The five to seven causative mutations estimated to accumulate in many cancers cannot solely be explained by the frequency of multiple independent mutational events (1). The accumulation of mutations in certain cancers may reflect a mutator phenotype, i.e., a high frequency of mutations in tumors due to more frequent genetic damage or compromised DNA repair (14).

Previously, we reported that in the lacI region spontaneous multiple mutations are more frequent (2.4 × 10−7) than expected for multiple independent mutational events (2 × 10−9) (5, 6). Approximately 1% of all lacI mutant plaques were doublets (two nonadjacent mutations within lacI) and multiplets (three or more nonadjacent mutations within lacI) (5, 6). Doublets were much more frequent than multiplets, and doublets were observed in almost all tissues examined. Singlets (one mutation within lacI), doublets, and multiplets from normal tissues have similar patterns of mutation, indicating a universal error-prone mechanism and no unique mutagen signature. In human and a variety of other organisms, the frequency of multiple mutations is higher than expected for random processes, consistent with a transient hypermutability (7, 8).

These multiple mutations showed close spacing fitting an exponential distribution with a median mutation spacing of 120 bp (5, 6). The spacing between doublet mutations did not fit a quasiuniform distribution (reflecting the finite target and the relative overabundance of mutations within the lacI DNA-binding domain) for randomly spaced events as would be expected for mutations occurring during different cell cycles or randomly within the gene. The data suggest chronocoordinate mutation events. Close spacing of multiple mutations has also been reported for an endogenous human gene (7, 8).

The close spacing of multiple mutations was unexpected and not predicted by the mutator phenotype hypothesis. A mutator phenotype in its simplest form predicts a higher frequency of independent events with multiple randomly spaced mutations transmitted to their progeny as a result of deleterious somatic mutations affecting the integrity of genome maintenance. Transient hypermutable conditions might be mediated by nucleotide pool imbalances and/or error-prone conformations of polymerases (911). Transient hypermutability may be caused by transcriptional or translational errors in the generation of a normal replicative DNA polymerase or any protein involved in replication fidelity (8).

Multiple mutations are enhanced in the lacI gene in p53+/− and p53−/− mice (6). Thus, multiple mutations may contribute to carcinogenesis. Mouse models of Li Fraumeni Syndrome develop cancer early at a very high frequency (12), whereas mice with mismatch repair deficiency have a high mutation load, a high frequency of cancer, and a mutator phenotype within cancer (2, 13, 14). Li Fraumeni mice and homozygous p53 knockout mice have tumors exhibiting a generally normal mutation frequency and pattern (15, 16). How might p53-deficient mice develop cancer at a high frequency if their overall mutation load is unchanged? Chromosomal instability in these mice explains at least part of this phenomenon (17). p53-deficient mice develop cancer in the context of elevated multiple mutations but unchanged total mutation load (16). What is not known is the extent of the multiplicity of mutations within each cell. This multiplicity might be envisioned as mutation showers occurring over the landscape of the genome within a single cell cycle. Could the multiple mutations observed in lacI be in the context of mutation showers? Are the multiple mutation events observed in lacI highly localized or do they extend over multiple kilobases?

The Big Blue transgenic mouse mutation detection assay is an ideal system to test the hypothesis of transient hypermutability resulting in clusters of mutations over a target size of up to 45 kb. The assay uses transgenic mice that harbor chromosomally integrated λ-bacteriophage containing the lacI gene as a neutral mouse mutation target. The mutation target is flanked by additional neutral mutation target sequence and additional remote neutral sequence is available for analysis of the spectrum of multiple mutations. The Big Blue transgenic mouse mutation detection assay is a well validated system for detection and analysis of spontaneous and induced mutations in individual tissues (1821). In fact, the cloned plaque nature of the Big Blue assay allows detection of the presence of mutation showers, whereas these would be hard to detect in most other systems. The assay has been in use for >15 years, and an extensive database of 5,303 mutations exists for analysis of spontaneous mutations (5, 6, 16, 18, 2232).

Herein, we provide a detailed analysis of mutations in the context of single and multiple mutations in the regions flanking and remote to the lacI gene of Big Blue mutants. Regions outside of lacI are sequenced to test the hypothesis that the mutants with more than one mutation in lacI represent mutation showers. The study is designed to investigate the existence and signature of transient hypermutability producing multiple mutations. By sequencing regions flanking (10.9 kb) and remote (8.6 kb) to the lacI gene, additional mutations are detected at a frequency of 14% in 65 mutants with more than one nontandem mutation in lacI but only at a frequency of 0.8% in 130 mutants with a single mutation lacI. These observations suggest that transient hypermutability, perhaps mediated by error-prone conditions, exists.

Results

Ectomutations Exist and Are Enriched in Flanking Regions.

Flanking and remote regions outside of the normally sequenced lacI transgene target were sequenced in Big Blue mutants with previously identified mutations in lacI.

Flanking regions were sequenced in 65 mutants with two or more mutations previously identified in lacI (“domuplets,” see Terminology in Materials and Methods) and in 130 mutants with a single mutation identified in lacI (“singlets”) (Fig. 1). Seventeen ectomutations were identified in the flanking regions [upstream flanking region (UF) and downstream flanking region (DF); 10.7 kb per plaque) in a total of nine (14%) of the domuplet mutants (Table 1 and Fig. 2). The frequency of ectomutations in domuplet mutants is 24 per megabase sequenced (17/10.7 kb/65), which is 240-fold greater than the 10−7 expected frequency per base pair (3 × 10−5 mutation frequency in lacI / ≈300-bp target size) (6). One ectomutation was identified in a flanking region of one of the 130 singlet plaques, for a frequency of 0.7 ectomutations per megabase sequenced (Table 1). The mutation showers averaged approximately one mutation per 3 kb.

Fig. 1.
Big Blue λLIZ shuttle vector [45,530 bp; (37)]. lacI, lacI gene region (1.4 kb; bp 20347–21706), UR, 2.8 kb located 17.4 kb from the lacI region; bp 178-2942); UF, 3.6 kb; bp 16741–20346, DF, 7.1 kb; bp 21707–28847; DR, ...
Fig. 2.
The distribution of spontaneous mutations in the Big Blue λLIZ shuttle vector and regions flanking and remote to the lacI region. The shaded regions represent the regions not sequenced. These shaded regions are truncated, representing 13.8 kb ...
Table 1.
lacI spontaneous mutants analyzed and the number of mutations identified in new regions analyzed

Ectomutations Are Observed Less Frequently in Remote Regions.

To examine the extent of mutation showers, 8.5 kb of sequence remote to lacI was sequenced [2.8 upstream remote region (UR) and 5.7 downstream remote region (DR); Fig. 1]. Two ectomutations were identified in the DR region of domuplet mutants, and no mutations were identified in the UR region (Fig. 2). One mutation in DR was identified in a domuplet mutant that does not have ectomutations in the flanking regions (UF and DF), and the other DR mutation was identified in a domuplet mutant with two ectomutations in UF and one ectomutation in DF (#10 and #6, respectively) [Fig. 2, and supporting information (SI) Tables 3 and 4]. In total, 10 domuplet mutants (15%) and 1 singlet mutant (0.8%) contain ectomutations (Table 1). The lower mutation frequency in the remote regions was significant (P = 0.004). Because 10.7 kb (3.6 + 7.1) was sequenced in the flanking regions and 8.5 kb (2.8 + 5.7) was sequenced in the remote regions, the expected probabilities of mutations in the flanking and remote regions are 0.557 and 0.443, respectively, assuming the null hypothesis that mutations are equally likely in the flanking and remote regions. By using a χ2 goodness-of-fit test to compare these expected probabilities to the observed counts of 17 and 2, respectively, the P value is 0.004. This difference is not likely to be due to differences in the fraction of the targets in the flanking and remote regions that disrupt the λ-lytic cycle because proteins in both regions are generally quite tolerant to missense changes (SI Text).

Ectomutations Are ≈40-Fold More Frequent in Domuplet Mutants Than in Singlet Mutants.

A total of 10 of 65 domuplet mutants (15.4%) have a total of 19 ectomutations, but only 1 of 130 control singlet mutants (0.8%) has an ectomutation (Fisher's exact test, P = 0.0001). The relative frequency of ectomutations in domuplet mutants (19 of 65) and singlet mutants (1 of 130) is 38. The 11 plaques came from nine mice. Mutants #4 and #11 (mouse 100.75) had a mutation shower in thymus and another in spleen with nonoverlapping mutations. Mutants #7 and #10 came from a thymic tumor (mouse 100.108), which was our most extensively sequenced tumor (169 independent mutations), and had two mutation showers without overlap of mutations. Within the constraints of sample size, a comparison of tissue of origin for the mutation showers and total mutations ascertained did not demonstrate a significant tissue preference (data not shown).

Ectomutations Do Not Show a Mutagen Signature Compared with the Pattern of Spontaneous Mutations.

Among domuplet mutants, the patterns of (i) ectomutations alone; (ii) lacI mutations in mutant plaques with ectomutations, and (iii) all lacI mutations in mutant plaques without ectomutations are similar, with no evidence of a specific mutation signature for ectomutations (Table 2). When comparing the aforementioned patterns with the pattern of singlets, differences are seen as expected (P = 0.002), because domuplet mutants are enriched for p53-deficient tumors which have been shown to have a different mutation pattern: a decrease in mutations at CpG and an increase in non-CpG mutations at adenine and thymine (6). When p53-deficient mice are excluded from these analyses, no statistical differences are observed. Ectomutations from domuplet mutants from p53+/+ mice had a pattern similar to that of singlet mutations.

Table 2.
Pattern of mutations in domuplets, singlets, ectomutations, and the overall pattern of spontaneous mutations in Big Blue

Some Ectomutations Occur in Pairs; Overall the Spacing Between Ectomutations Is Nonrandom.

Among the 11 mutants with ectomutations, a single ectomutation was found in four doublets, one multiplet, and one singlet. Two ectomutations were found in two doublets, three were found in two doublets, and four were found in one doublet (SI Tables 3 and 4). Four of the doublet mutants with ectomutations (#2, #3, #6, and #9) contained closely spaced mutations with the spacing between clustered ectomutations of 310 bp or less (spacing = 4, 113, 76, and 308 bp, respectively) (SI Tables 3 and 4). These four mutants with ectomutations actually had pairs of clustered mutations; the spacing within the lacI region was 177, 2, 391, and 401 bp, respectively.

To test for significant clustering of ectomutations, a simulation of 10,000 iterations of 10 mutants with a total of 18 random ectomutations in UF and DF was performed. Each mutant was given the same number of ectomutations in UF and DF as observed, but with random positions. The separations between the simulated ectomutations within the same flanking region or between the ectomutation that is nearest to lacI and the nearest mutation in lacI were compared with the separations in the observed data. There were six observed separations of 500 bp or fewer and four of 200 bp or fewer. From the simulation, the probability of six or more separations of 500 bp or fewer is 0.014 (141 of 10,000) and the probability of four or more separations of 200 bp or fewer is 0.0098 (98 of 10,000).

Mutation Showers Tend to Extend over a Range <30 kb.

The distribution of mutations observed in the flanking regions in mutation showers is not significantly different from a uniform distribution (SI Fig. 3, P = 0.23 by exact χ2 goodness-of-fit test in StatXact), but when the remote regions are also considered, the distribution is significantly different from uniform (P = 0.0088). This analysis indicates that mutation showers extend beyond the flanking region (more than ± 7.2 kb from lacI), but have mostly dissipated by the remote regions (more than ± 16.6 kb from lacI).

Discussion

Among 19.2 kb of additional sequence analyzed, 19 ectomutations are identified among 10 domuplet plaques and one ectomutation is identified among one singlet plaque. The ectomutations do not show a different pattern of mutation types relative to mutations in lacI. Ectomutations are observed more frequently in the sequence flanking lacI than in the sequence remote to lacI and the mutations tend to cluster. The data are consistent with the following model: chronocoordinate mutations, tending to diminish within 30-kb region, constitute at least 0.2% of mutational events and may well constitute 1% or more of mutational events (see below). Some mutation showers will be missed by the assay because a gene essential to the assay (disrupting the λ-lytic cycle or lacZα) is disrupted by one of the mutations within the shower. If “heavy showers” existed, the great majority of them would be missed.

The finding that multiple mutations cluster is consistent with transient error-prone conditions that may be mediated by nucleotide pool imbalances that induce a wide variety of mutation types, such as microdeletions, microinsertions, and microindels (911) and increase mutation frequency overall. Transient nucleotide pool imbalances may cause multiple mutation showers throughout the genome, as might a mistranslated hypermutagenic normally high-fidelity polymerase, a transient overrecruitment of error-prone polymerases, or a transient hypermutagenic conformation of a normally high-fidelity polymerase. Each of these mechanisms may cause isolated mutation showers.

Caveats.

The Big Blue assay uses the lacI transgene within the concatameric λ-phage as a reporter gene. The assay has been extensively validated (1821), and the literature demonstrates similarities between human mutations and Big Blue mouse mutations (17, 24, 29). Also, the pattern of mutations from Big Blue mice is significantly different from the pattern of mutations observed in Escherichia coli, consistent with minimal contamination of E. coli derived mutations (25). Analyses of other types of uncommon mutations (e.g., indels) have demonstrated similarities with humans but not with E. coli (29).

It is possible that a subset of mice was exposed to some unknown mutagen, which caused the multiple mutations, but the similarity of the pattern with singlets, when the factor of p53 sufficiency or deficiency is controlled, makes that unlikely, as does the uniform treatment of the mice in the institutional animal facility and exposure to a lifetime constant diet. Herein, within the constraints of small sample size, mutation showers were not obviously clustered in individual mice, despite intensive mutation analysis of multiple tissues in certain mice. The frequency of mutation showers may be underestimated because approximately half of the λLIZ genome (Fig. 1) was not sequenced and for other reasons as well (see below).

Mutation Showers Account for at Least 0.2% and Possibly >1% of Mutational Events.

Eleven mutants with ectomutations were found (0.2%) among 5,303 independent mutants. All 65 domuplet plaques were analyzed, whereas only 2.5% (130 of 5,238) of singlet plaques were analyzed. If the sample of singlets is representative, 40 singlet plaques with ectomutations are expected in the total collection of singlet plaques in addition to the 10 domuplet plaques, for a total of 50.

The frequency of mutation showers may also be estimated by using the spacing of mutations within mutants with ectomutations. A “thought experiment” can be performed. In this thought experiment, plaques with mutation showers are sequenced in 1.4-kb regions, analogous to the length of the originally sequenced lacI region. Mutations found are then recorded as either a singlet or a domuplet. Two extreme possibilities provide perspective on the calculation. Suppose that mutation showers always contain pairs of closely spaced mutations. Then, essentially all 1.4-kb regions sequenced from plaques with mutation showers would contain doublets if they contained any mutations at all, implying that all of the mutation showers would have been identified as doublets in the lacI region that was sequenced in the Big Blue assay. Thus, 10 of 5,303 mutation events is the best estimate (0.2%). In contrast, suppose that only 5% of the mutations within mutation showers are closely spaced. Then, only 5% of the mutation showers in the 5,303 independent events would have been identified by doublets within the lacI region sequence in the Big Blue assay. Because 10 mutation showers (0.2% of total plaques) were identified by comprehensive sequencing of all doublets, the number of mutation showers expected in the singlets is 200, for a total frequency of 210 per 5,303 events (4%).

If the actual 11 plaques with mutation showers were provided for sequencing, four doublets would be found (see the ectomutations in #2, #3, #6, and #9), and 12 singlets would have been found. Therefore, the ratio of domuplets to singlets would be 1:3. If these mutation showers are representative, the 65 domuplets originally identified in lacI yielded the mutation showers observed as domuplets within 1.4 kb (or 10). This predicts another 30 mutation showers observed as singlets for a total of 40 (0.8%).

The estimates of 0.8–1% are likely to be underestimates for several reasons. Because less than half of λLIZ was sequenced in this experiment, more mutants with ectomutations are expected if the entire 45-kb λLIZ were to be sequenced. In addition, one ectomutation that disrupted function of the λLIZ lytic genes would eliminate the plaque from detection.

Although most single base substitutions would be tolerated, nonsense mutations, some missense changes, and the great majority of deletions, insertions, and indels that produce protein truncation would have prevented plaque formation. One such ectomutation within a shower is sufficient for the mutation event to have been missed. In the coding region of selected λ-proteins, the fraction of base substitutions resulting in nonsense mutations is 4% relative to 73% and 22% causing missense and silent changes, respectively. More ectomutations increases the likelihood of disrupting the function of lytic genes or lacZα. Thus, it is possible, perhaps even likely, that a few percent of mutation events in lacI occur in the context of mutation showers. If there were a class of mutation showers with a substantially higher frequency of mutations, virtually all would have been missed because of a high likelihood that at least one of the mutations would disrupt the lytic cycle of the λLIZ genome.

Ectomutations Are Found Less Frequently in Regions 15–20 kb from lacI.

Observed mutation frequency is reduced in the remote region with respect to the flanking region. Mutation frequency can be reduced because the mutation showers have ceased 15–20 kb from the lacI region, or because the target size for observable mutations is much smaller in the remote regions. The second possibility is unlikely for several reasons (SI Text): (i) Analysis of λ-phage protein evolution led to the conclusion that λ-phage proteins essential for lytic growth are generally quite tolerant of missense changes; (ii) genes essential for the Big Blue assay are found in flanking regions with a high frequency of ectomutations, and (iii) a total of 10 missense and 3 silent mutations (ratio 3.3:1) are found in essential genes in the mutants with ectomutations, in marked contrast to the great predominance of silent mutations that occur in proteins. We therefore conclude that most mutation showers span less than the 30 kb separating the two remote regions.

Possible Implications for Evolution.

Multiple mutations may cause compensatory amino acid changes (e.g., two amino acid substitutions that might individually disrupt protein structure may compensate for one another). These may obscure the relationship between evolutionary conservation and the critical nature of an amino acid for protein function. This is of particular clinical significance because evolutionary conservation is a useful “in vivo functional test” for assessing the likelihood that a missense change found in a patient is likely to be deleterious.

The observed mutation showers often will affect one or a few genes in mammalian genomes, because they tend to diminish within 30 kb. Therefore, most mammalian genes range from 20 kb to 1 Mb with 90+% of the sequence within introns. Approximately 90% of the mutations within a mutation shower generally would not have functional consequences. Thus, the introns serve as a “sponge” to absorb many of the mutation showers without damage to protein function. In general, spontaneous mutation processes seem to be similar in eukaryotes (18, 33). If mutation showers occur in fungi and other organisms with short or minimal introns, it is possible that multiple contiguous genes can be affected. Novel interactions can occur if genes within pathways happen to be nearby in the genome.

Cancer in an Instant?

Might there be scattered mutation showers throughout the genome that occur, perhaps by nucleotide pool imbalances during replication or another cellular metabolic process? This is a critical unanswered question. If scattered mutation showers occur, multiple genes could be inactivated, leading to cancer in an instant. Scattered mutation showers may contribute to the surprising number of somatic mutations recently observed on sequencing of the protein in coding regions of 13,023 genes in breast and colon cancer; an average of 93 somatic mutations were found per tumor (39).

Materials and Methods

Terminology.

Herein we use the following definitions:

Mutant.

The blue circular mutant plaque cored in the Big Blue assay, regardless of the number of mutations detected.

Singlet.

A mutant with a single mutation affecting lacI expression, and hence allowing β-galactosidase expression.

The lacI region.

Mutants were originally detected because of mutations that compromised the function of the lacI repressor and allowed expression of the lacZ gene. The original sequencing window to detect the lacI mutations extends from base pair 20347 to 21706 (base numbering of λLIZ; see below) and includes the lacI promoter and coding sequence, followed by the lacZ operator sequence.

Ectoregion.

Regions sequenced outside of the lacI region. For this analysis, the ectoregions are referred to by their location. DF, UF, DR, and UR (Fig. 1).

Chronocoordinate mutations.

Multiple mutations occurring within the same cell cycle and in rapid succession (6).

Multiple mutations.

A general term encompassing all situations with two or more mutations in lacI occurring in a single mutant.

Tandem base mutation (TBM).

A mutant plaque with substitutions at adjacent nucleotides in lacI (26).

Doublet.

A mutant with two nonadjacent mutations within lacI identified in a mutant. Doublets are generally chronocoordinate events (5, 6).

Multiplet.

A mutant with three or more mutations in lacI identified in a mutant, excluding the situation in which all mutations are adjacent. Multiplets can include a mix of TBM and non-TBM mutations.

Domuplet.

A mutant that is either a doublet or a multiplet. Approximately 1% of lacI mutant plaques are domuplets (6).

Ectomutation.

Mutation identified outside of the lacI region.

Mutation shower.

Chronocoordinate multiple mutations that span multiple kilobases.

The Big Blue Transgenic Mouse Mutation Detection Assay.

Details regarding the Big Blue transgenic mouse mutation detection system and mutation analysis were reported previously in the Stratagene instruction manual, August 15, 1992 (34).

Animals and Tissues.

Details regarding animal care and tissue collection were reported previously (28). Normal tissues (liver, cerebellum, male germ line, thymus, forebrain, adipose tissue, heart, bone marrow, whole brain, kidney, mammary epithelium, and spleen) were collected from Big Blue mice (95% C57BL/6J genetic background at all ages and a few C3B6F1 mice at 14 and 25 mo. of age). The mutations examined in these mice were termed spontaneous in that the mice were not intentionally exposed to any known mutagen. Normal and tumor tissues were also collected from C57BL p53-deficient animals (16, 35).

Mutants Assessed for Additional Mutations.

In the past 15 years, our laboratory identified 5,303 independent mutations in the Big Blue lacI gene that affect lacI repressor function from mice not treated with a known mutagen. Among these mutations, 65 domuplet mutants were identified. Herein, additional regions of the Big Blue λLIZ construct were sequenced to identify and analyze the spectrum of additional mutations. Additional sequence was examined in regions either flanking (adjacent to lacI) or remote (16–17 kb away) with reference to the lacI gene. In addition to the 65 domuplets, 130 singlet mutants were included in the analysis to compare the frequency and spacing of additional mutations among three classes of mutants (singlets, doublets, and multiplets). These reference mutants were collected in parallel to the 65 domuplets. All mutations were confirmed by reamplification and resequencing.

PCR Amplification of Additional Mutation Target Sequences.

The primers for PCR and sequencing are shown in SI Table 5. The Expand High Fidelity PCR system (Roche, www.roche.com, Basel, Switzerland) was used to amplify each segment of 3 kb. The PCR mixtures contained a total volume of 30 μl: 1× Expand High Fidelity buffer #3 (Roche), 2.5 mM MgCl2, 200 μM dNTP, 0.25 μM of each pair of primers, 1 unit of Expand High Fidelity enzyme (Roche), and 50 ng of DNA. The cycling entailed denaturation at 94°C for 20 sec, annealing at 58°C for 30 sec, and elongation at 72°C for 3 min for 35 cycles. The PCR products were analyzed after electrophoresis through a 1% agarose gel, purified with SAP plus Exo 1 (Roche), and sequenced.

λLIZ Construct and Sequencing Strategy.

The genetic map of the Big Blue λLIZ construct is restriction enzyme-based (Fig. 1) (36). The size of the wild-type λ-phage genome is 48,502 nt (National Center for Biotechnology Information, NC_001416). In the λLIZ construct, the sequence containing lacZα, the lacZ operator, the lacI gene and the ampicillin resistance marker replace the middle region of the λ-genome, thus making the size of the λLIZ construct 45,530 nucleotides. The λLIZ shuttle vector of the Big Blue mouse is derived from wild-type λ-phage (48,502 bp) by a deletion of three nonconsecutive segments (5,109 bp, 2,323 bp, and 17 bp) and insertion of the LIZ plasmid (4,475 bp) containing the lacI gene promoter and coding sequence, operator sequence, and α portion of the lacZ gene in the reversed orientation (SI Fig. 4) (37). Two germ-line mutations were found in the sequencing of the additional regions (SI Text).

In the standard Big Blue mutation detection assay, a 1.4-kb region containing the lacI gene promoter and coding sequence and the lacZ operator region (20,347 bp to 21,706 bp) is analyzed to identify the mutation responsible for expression of lacZ producing the blue circular mutant plaque.

Herein, four additional regions adjacent to the lacI gene are sequenced. First, 10.7 kb of lacI flanking sequence was analyzed. Region UF is the 3.6-kb segment upstream flanking the lacI gene, and region DF is the 7.1-kb segment downstream flanking the lacI gene (Fig. 1). To examine the extent of mutation showers, 8.5 kb of sequence remote to lacI was then analyzed. Region UR is the 2.8 kb segment upstream and remote from the lacI gene at the beginning of the λLIZ construct. Region DR is the 5.7-kb segment remote downstream from the lacI gene near the end of the λLIZ construct. Therefore, the total length of the new sequence analysis window is 19.2 kb. The sequences were given numbers by assuming that the unsequenced segments corresponded to the λ-phage sequence (GenBank NC_001416). For each region, the sequencing primers are designed such that two primers have an overlapping region of at least 50 nucleotides to ensure all of the nucleotides are sequenced. All new mutations detected were confirmed by sequencing in the reverse direction. Sequence discrepancies between wild-type λ-phage and the λLIZ construct (37) in Big Blue mice are provided in SI Table 6.

Statistical Analyses.

To analyze whether the spacing of ectomutations is random or nonrandom in the lacI and flanking regions, the expected distributions for random mutations were estimated based on simulated mutation data. The separations used in the analysis include the separations between ectomutations within the same flanking region and the separation between the ectomutation that is nearest to lacI and the nearest mutation in lacI. For each of the 10,000 iterations of the simulation, 18 ectomutations were randomly selected from the target and assigned to the 10 plaques in the same proportions as in the observed data.

Analysis of Ectomutation Pattern.

Mutation pattern was defined as the set of counts of five types of base substitutions (transitions at CpG, two types of transitions not at CpG, transversions at CpG, and transversions not at CpG) and deletions and insertions combined, for a total of six categories of mutation types. Mutation patterns were tested for significant differences by analysis as unordered r × c contingency tables using the Fisher–Freeman–Halton test implemented by the StatXact statistical analysis software package (CYTEL Software, Cambridge, MA). This test is an extension of the traditional Fisher's exact test for 2 × 2 contingency tables and was first proposed by Freeman and Halton (38).

Supplementary Material

Supporting Information:

Acknowledgments

We thank Sami Nasrawi, Asanga Halangoda, Chaniga Chitaphan, and Jenny Ngo for assistance in collecting mutation data. This work was partially supported by National Institutes of Health Grant R01 AG 19784 and U.S. Army Grant DAMD 17-01-1-02-06.

Abbreviations

DF
downstream flanking region
DR
downstream remote region
UF
upstream flanking region
UR
upstream remote region.

Footnotes

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

See Commentary on page 8203.

This article contains supporting information online at www.pnas.org/cgi/content/full/0610902104/DC1.

References

1. Loeb LA, Loeb KR, Anderson JP. Proc Natl Acad Sci USA. 2003;100:776–781. [PMC free article] [PubMed]
2. Mark SC, Sandercock LE, Luchman HA, Baross A, Edelmann W, Jirik FR. Oncogene. 2002;21:7126–7130. [PubMed]
3. Sommer SS. Hum Mutat. 1994;3:166–169. [PubMed]
4. Venkatesan RN, Bielas JH, Loeb LA. DNA Repair (Amsterdam) 2006;5:294–302. [PubMed]
5. Buettner VL, Hill KA, Scaringe WA, Sommer SS. Mutat Res. 2000;452:219–229. [PubMed]
6. Hill KA, Wang J, Farwell KD, Scaringe WA, Sommer SS. Mutat Res. 2004;554:223–240. [PubMed]
7. Colgin LM, Hackmann AF, Emond MJ, Monnat RJ., Jr Proc Natl Acad Sci USA. 2002;99:1437–1442. [PMC free article] [PubMed]
8. Drake JW, Bebenek A, Kissling GE, Peddada S. Proc Natl Acad Sci USA. 2005;102:12849–12854. [PMC free article] [PubMed]
9. Bebenek K, Roberts JD, Kunkel TA. J Biol Chem. 1992;267:3589–3596. [PubMed]
10. Dare E, Zhang LH, Jenssen D, Bianchi V. J Mol Biol. 1995;252:514–521. [PubMed]
11. Song S, Wheeler LJ, Mathews CK. J Biol Chem. 2003;278:43893–43896. [PubMed]
12. Donehower LA, Harvey M, Slagle BL, McArthur MJ, Montgomery CA, Jr, Butel JS, Bradley A. Nature. 1992;356:215–221. [PubMed]
13. Stuart GR, Glickman BW. Genetics. 2000;155:1359–1367. [PMC free article] [PubMed]
14. Hegan DC, Narayanan L, Jirik FR, Edelmann W, Liskay RM, Glazer PM. Carcinogenesis. 2006 [PMC free article] [PubMed]
15. Buettner VL, Hill KA, Nishino H, Schaid DJ, Frisk CS, Sommer SS. Oncogene. 1996;13:2407–2413. [PubMed]
16. Hill KA, Buettner VL, Heidt A, Chen LL, Li W, Gonzalez KD, Wang JC, Scaringe WA, Sommer SS. Carcinogenesis. 2006;27:1860–1866. [PubMed]
17. Venkatachalam S, Shi YP, Jones SN, Vogel H, Bradley A, Pinkel D, Donehower LA. EMBO J. 1998;17:4657–4667. [PMC free article] [PubMed]
18. Hill KA, Buettner VL, Glickman BW, Sommer SS. Mutat Res. 1999;436:11–19. [PubMed]
19. Nishino H, Buettner VL, Sommer SS. Mutat Res. 1996;372:97–105. [PubMed]
20. Piegorsch WW, Margolin BH, Shelby MD, Johnson A, French JE, Tennant RW, Tindall KR. Environ Mol Mutagen. 1995;25:231–245. [PubMed]
21. Thybaud V, Dean S, Nohmi T, de Boer J, Douglas GR, Glickman BW, Gorelick NJ, Heddle JA, Heflich RH, Lambert I, et al. Mutat Res. 2003;540:141–151. [PubMed]
22. Buettner VL, Nishino H, Haavik J, Knöll A, Hill KA, Sommer SS. Mutat Res. 1997;379:13–20. [PubMed]
23. Buettner VL, Hill KA, Halangoda A, Sommer SS. Environ Mol Mutagen. 1999;33:320–324. [PubMed]
24. Halangoda A, Still JG, Hill KA, Sommer SS. Environ Mol Mutagen. 2001;37:311–323. [PubMed]
25. Hill KA, Nishino H, Buettner VL, Halangoda A, Li W, Sommer SS. Mutat Res. 1999;425:47–54. [PubMed]
26. Hill KA, Wang J, Farwell KD, Sommer SS. Mutat Res. 2003;534:173–186. [PubMed]
27. Hill KA, Buettner VL, Halangoda A, Kunishige M, Moore SR, Longmate J, Scaringe WA, Sommer SS. Environ Mol Mutagen. 2004;43:110–120. [PubMed]
28. Hill KA, Halangoda A, Heinmoeller PW, Gonzalez K, Chitaphan C, Longmate J, Scaringe WA, Wang JC, Sommer SS. Environ Mol Mutagen. 2005;45:442–454. [PubMed]
29. Hill KA, Gonzalez KD, Scaringe WA, Wang JC, Sommer SS. Hum Mutat. 2006;27:55–61. [PubMed]
30. Moore SR, Hill KA, Heinmoller PW, Halangoda A, Kunishige M, Buettner VL, Graham KS, Sommer SS. Environ Mol Mutagen. 1999;34:195–200. [PubMed]
31. Nishino H, Buettner VL, Haavik J, Schaid DJ, Sommer SS. Environ Mol Mutagen. 1996;28:299–312. [PubMed]
32. You YH, Halangoda A, Buettner V, Hill KA, Sommer SS, Pfeifer G. Mutat Res. 1998;420:55–65. [PubMed]
33. Kunz BA, Ramachandran K, Vonarx EJ. Genetics. 1998;148:1491–1505. [PMC free article] [PubMed]
34. Stratagene Instruction Manual. La Jolla, CA: Stratagene Cloning Systems; 1992.
35. Donehower LA. Semin Cancer Biol. 1996;7:269–278. [PubMed]
36. Dycaico MJ, Provost GS, Kretz PL, Ransom SL, Moores JC, Short JM. Mutat Res. 1994;307:461–478. [PubMed]
37. Kohler SW, Provost GS, Fieck A, Kretz PL, Bullock WO, Putman DL, Sorge JA, Short JM. Environ Mol Mutagen. 1991;18:316–321. [PubMed]
38. Freeman GH, Halton JH. Biometrika. 1951;38:141–149. [PubMed]
38. Sjoblom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD, Mandeiker D, Leary RJ, Ptak J, Sillman N, et al. Science. 2006;314:268–274. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...