SEARCH

SEARCH BY CITATION

Keywords:

  • Egernia stokesii;
  • linkage;
  • multistep model;
  • mutations;
  • paternity;
  • skink;
  • tetranucleotide loci

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Seven tetranucleotide (AAAG) loci were analysed in a population (including 19 litters) of the Australian lizard Egernia stokesii. In an examination of 76 offspring we observed 13 mutations involving five loci. Two of our loci were highly mutable, with observed mutation rates of 2.7% and 4.2%, representing some of the highest mutations rates reported so far. A high proportion of mutations (46.2%) could not be assigned to changes involving only a single repeat, suggesting that mutations in at least two of the loci follow a multistep model. There was no significant bias of mutations leading to an increase or decrease in allele size; however, all multistep mutations involved a loss of repeats. These results add to increasing evidence casting doubt on microsatellite mutations being primarily single step mutations.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Microsatellite loci are now extensively used for studies requiring highly variable genetic markers such as paternity and kinship assessment, fine-scale population analyses, and genetic mapping. However, the processes which lead to this sometimes extreme genetic variability remain largely uncertain or speculative. Slipped strand mispairing (intrahelical replication slippage or SSM) is most often proposed as the molecular mechanism responsible for mutations occurring at microsatellite loci ( Levinson & Gutman, 1987) but this assumption has little solid empirical support ( Primmer et al., 1998 ). The role of other potential processes such as recombination events or unequal crossing-over (UCO) may also be important ( Levinson & Gutman, 1987).

Dinucleotide loci are the most common type of microsatellite loci isolated, perhaps due to their large number in most eukaryote genomes ( Queller et al., 1993 ). Although rarer, tri- and tetranucleotide loci are now being increasingly used, especially with the advent of enrichment techniques for microsatellite isolation (e.g. Gardner et al., 1999 ). Both tetra- and trinucleotide loci can provide greater readability and are less likely to suffer from stutter bands than the more commonly used dinucleotide loci ( Schlotterer & Tautz, 1992). Little is known about variations in mutation rates or whether there are any unifying patterns of mutations for particular motif types. Weber & Wong (1993), using in vivo mutation data from human microsatellite loci, suggested that tetranucleotide loci have higher mutation rates than both di- and trinucleotide loci. In a more recent and detailed study of allele size distributions in humans, Chakraborty et al. (1997 ) concluded the opposite. They found the most mutable loci are disease-causing trinucleotide loci followed by di, non-disease-causing tri-, and tetranucleotide loci in decreasing order of mutation rates. Similar results to those of Chakraborty et al. (1997 ) for the nondisease causing loci were also found by Schug et al. (1998 ) in a study of Drosophila microsatellite loci. Mutation rates are also reported to vary among loci of the same repeat type and among alleles at a particular locus ( Jin et al., 1996 ). However, the characteristics that make particular loci (or alleles) more mutable than others are not fully understood. It has been suggested that the number of repeats ( Primmer et al., 1996 ), the structure of repeat arrays ( Estoup et al., 1995 ), the base content of sequence flanking the repeat region ( Glenn et al., 1996 ) and heterozygote instability ( Amos et al., 1996 ) all influence mutation rates. Defining the characteristics that influence mutation rates is necessary for determining appropriate models of evolution of microsatellite loci. Data from the direct analysis of mutation events provide important information as much of the data for mutation models have been based on allelic distributions and these may not present an accurate picture of the underlying mutational processes.

Studies of germ-line microsatellite mutations, mainly in humans, found that mutations involving the gain or loss of a single repeat unit were much more frequent than multiple-repeat mutations (e.g. Weber & Wong, 1993). These findings contradict the assumptions of the infinite allele mutation model ( Kimura & Crow, 1964) where any allele arising by mutation is different from alleles already existing in the population. This led to the development of mutation models based on simple stepwise mutations ( Ohta & Kimura, 1973) to include allelic repeat scores (e.g. Valdes et al., 1993 ). However, these models did not take into account the existence of mutations of larger magnitude so Di Rienzo et al. (1994 ) developed a two-phase model which assumes that most mutational changes result in an increase or decrease of one repeat unit, but that less frequent larger jumps in repeat number also occur. More recent modifications of this model include allowance for rate variation between loci and the introduction of allele length ceilings ( Feldman et al., 1997 ).

In this paper we present mutation data from 19 live-born litters, involving 76 offspring, from an ongoing study of mating structure and relatedness in the Australian lizard Egernia stokesii. We also investigate several mechanisms that have been proposed to affect mutations at microsatellite loci. We found a high proportion of multistep mutations at our loci and suggest that the commonly used assumption of the nearest allele being the source of the mutation may not be valid. One tetranucleotide locus was found to have a mutation rate higher than any previously reported rates, but despite this we did not find any bias towards expansion at this or at our other loci.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Samples

Lizards were sampled from Camel Hill and adjacent ridges in the Warruwarldunha Range (31°54′S; 138°25′E) near Hawker, in the lower Flinders Ranges of South Australia. Over the period 1994–98, 152 individuals were captured in unbaited elliot traps or by removal from their rock crevice refuges. The sample included most of the population on Camel Hill (G.A.D., personal observation). Blood was taken from the caudal vein before each lizard was released at the point of capture. The blood was transferred immediately into tubes containing 50% v/v ethanol/0.85% saline solution and stored at room temperature until use.

Litters of young are produced in February each year ( Duffield & Bull, 1996). In February of 1997 and 1998, 17 gravid females were captured from the study area and were held in cages until litters (totalling 66 offspring) were born. Litter size ranged from 1 to 7. Blood was collected from the juveniles. Another two litters were collected from Camel Hill females that had died before giving birth in 1994 and 1998 (10 offspring from mothers 65 and 32). These litters had been preserved in 70% ethanol.

Microsatellite analysis

DNA was extracted directly from 0.1–0.5 mL of blood from each lizard using a modified CHELEX 100 technique ( Walsh et al., 1991 ), or from skin preserved in 70% ethanol using a standard phenol/chloroform extraction procedure ( Sambrook et al., 1989 ). Samples were screened for six tetranucleotide microsatellite loci (EST1, 2, 3, 4, 8, 12) isolated from E. stokesii as described in Gardner et al. (1999 ) and for one tetranucleotide locus (Tr3.2) isolated from Tiliqua rugosa ( Cooper et al., 1997 ). Another polymorphic locus (EST9; Gardner et al., 1999 ) was not used because of a high frequency of null alleles at this locus (unpublished data). Amplification with the primers originally reported for EST2 ( Gardner et al., 1999 ) was supplemented with a second set of primers (EST2b) (forward primer 5′tactagaacaaattagaagcactg3′; reverse primer 5′ttccactgagctagcatgacta3′). The sizes of the alleles using these additional primers were converted back to those at the EST2 locus amplified with the original primers.

PCR reactions typically used 1.5 μL of chelex DNA and were performed using a hotstart approach in a total volume of 20 μL with 1× Amplitaq Gold buffer (Perkin Elmer); 4 m M MgCl2; 0.2 m M of each dNTP; 2 pmol of each primer; and 0.25 units of Amplitaq Gold. The reaction conditions were: one cycle of 9 min at 95 °C, 45 s at the appropriate annealing temperature 52–60 °C ( Cooper et al., 1997 ; Gardner et al., 1999 ), 2 min at 72 °C; 34 cycles of 45 s at 94 °C, 45 s at appropriate annealing temperature and 1 min at 72 °C; followed by one cycle of 5 min at 72 °C. Genotyping was performed on an ABI 377 (Applied Biosystems) with a fluorescent dye incorporated into one of the primers of each locus. Fluorescent labelling of the loci was as follows: TET (Tetrachloro Fluorescein) – EST1, EST3 and Tr3.2; FAM (Fluorescein) – EST2, EST8 and EST12; HEX (Hexachloro Fluorescein) – EST4, EST2b. All loci were amplified in separate reactions except EST3 and EST8 which were coamplified. Products from amplifications of EST3 + 8, EST4 and Tr3.2 were mixed together in equal proportions and analysed in a single lane. Likewise, products of the loci EST1, EST2, EST2b and EST12 were multiplexed for electrophoresis.

A standard ladder sample marked with GS2500ROX (ABI) fluorescent dye was run with each sample allowing accurate sizing of alleles and comparison between gels. Results were analysed with GENESCAN software (Applied Biosystems).

Genotype data from the Camel Hill population (considering only adults from the 95/96 season) were tested for conformation to Hardy–Weinberg equilibrium at each locus to ensure Mendelian inheritance, and to investigate the possible existence of null alleles. To do this we used a probability test, the exact HW test of Haldane (1954), Weir (1990) and Guo & Thompson (1992), using a Markov chain method in GENEPOP 3.1 ( Raymond & Rousset, 1995a, b). Any loci found to be significant were tested again with the more powerful U-test ( Rousset & Raymond, 1995), for a deficiency of heterozygotes, as implemented in GENEPOP 3.1. The CERVUS program ( Marshall et al., 1998 ) was used to estimate the probability of null alleles at each locus and to provide allele frequency and heterozygosity data. To be confident that genotypes at all loci were segregating independently, we tested for linkage disequilibrium between each pair of loci again using genotype data from adults of season 95/96. Loci showing linkage disequilibrium were further tested for possible linkage using lod scores generated from genotype data for parents and progeny using the program CRI-MAP v2.4 ( Lander & Green, 1987). Where multiple tests were performed, Bonferroni procedures ( Miller, 1980) were used in order to reduce the possibility of a significant value due to chance.

Detection of multiple paternity in litters

Multiple paternity in litters was inferred by the presence of three or more nonmaternal alleles in at least two loci. A criterion of two loci was used to avoid falsely designating a litter to be multiply sired as a result of mutation of microsatellite alleles (e.g. Weber & Wong, 1993), the presence of null alleles ( Pemberton et al., 1995 ) or short allele dominance ( Wattier et al., 1998 ).

Mutation detection

Mismatches between known mother and offspring alleles were considered mutations if they occurred only at a single locus. Some mismatches could have resulted from the transmission of null alleles ( Pemberton et al., 1995 ). To decrease the possibility of incorrectly assigning a mismatch due to null alleles, we considered a mutation to have occurred only if either the mother or the offspring were heterozygous at the mismatched locus. We tentatively assigned male parents to litters by matching inferred genotypes from offspring genotypes to potential sires living close to the female (G.A.D., unpublished work). Alternatively we inferred the paternal genotypes from the progeny in a litter. From the paternal and maternal genotypes we identified mutations in the offspring presumed to be paternal in origin. We calculated the female and male mutation rate per locus by dividing the number of mutations observed by the number of offspring that were typed at that locus. The overall mutation rate per locus was calculated by dividing the number of mutations observed by twice the number of offspring, to account for each meiotic event. For any particular locus, homozygous offspring that mismatched homozygous parents were considered to have inherited a null allele and were removed from calculations. Thus our calculated mutation rates may be underestimates of the real rates. Offspring and mothers were run on the same gels to negate possible typing errors. To allow for small sample sizes when performing χ2 tests on mutational data we used the program CHIRXC ( Zaykin & Pudovkin, 1993) with 1000 randomizations of the data.

Tests of mutation models from population data

We tested the applicability of the one-step mutation model as compared with a multistep model for our population data (Camel Hill adults and subadults from the 95/96 season) using the program MISAT ( Nielsen, 1997). This program uses a Markov chain recursion method to estimate the likelihood function under different models of microsatellite evolution. With the one-step model as the null hypothesis, a likelihood ratio test was used to compare the value –2log λ (where λ is the ratio of the maximum likelihood values under the two models) to a χ2 distribution with one degree of freedom. We also estimated θ=4Neμ (four times the effective population size times the mutation rate) with MISAT.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Microsatellite analyses

All seven loci were highly polymorphic (Table 1 and Fig. 1). Among the 152 individuals sampled from Camel Hill, the number of alleles per locus ranged from 7 to 24, and observed heterozygosity ranged from 79.3% to 91.5% (Table 1). Fractional repeat allele sizes, alleles not differing in sizes by multiples of 4 bp, were found at two of the loci (Tr3.2 and EST2 see Fig. 1). We observed heterozygotes between alleles differing by only two base pairs in several individual samples at both loci (data not shown). The analysis of the EST2 locus using an alternative primer set also confirmed the fractional repeat allele sizes.

Table 1.   Loci summary statistics for all individuals from Camel Hill (including juveniles). Thumbnail image of
image

Figure 1.  Allele frequencies at seven microsatellite loci in Egernia stokesii from Camel Hill. Includes adults and subadults from the 95/96 season. Scale may vary for different loci.

Download figure to PowerPoint

Two loci, EST3 and EST4, were found to show deviations from Hardy–Weinberg equilibrium using the probability test in GENEPOP (< 0.0001) considering only adults from the 95/96 season. We tested both of these loci for deviations due to an excess of homozygotes and found the EST3 locus (< 0.0001) and the EST4 locus (P=0.0241) to have significantly more homozygotes than expected. Deviations at these loci were still significant after adjusting the significance level (to α=0.025) with Bonferroni procedures. Considering that the five other loci show Hardy–Weinberg expectations, the deviations at EST3 and EST4 suggest that null alleles and/or allelic dropout may be evident at these loci. Population subdivision may also be responsible for the observed deviations. However, examination of the litter data showed seven offspring with different homozygous genotypes at the EST3 locus compared with their mother or father’s genotypes (data not shown). These mismatches are most likely due to the presence of null alleles. Four parental to offspring mismatches at the EST4 locus could probably be explained by short allele dominance (allelic dropout). They involved homozygosity in the either offspring or father and the amplification of a large allele in the corresponding individual. The larger allele was observed to be of smaller intensity at most loci (data not shown). Although we found some mismatches as indicated above, estimates of null allele frequencies by the CERVUS program were low for both EST3 and EST4 (Table 1).

Significant linkage disequilibrium, after applying Bonferroni corrections (new α=0.002), was found between loci EST3 and EST4 (P=0.0007), EST4 and EST8 (P=0.0017) and between Tr3.2 and EST12 (< 0.0001). However, examination of the pedigree data from mothers, inferred fathers and offspring genotypes found close linkage only between Tr3.2 and EST12 with a recombination fraction (θ) of 0.04, and a maximum lod score of ~15.93 (combined results of recombination in males and females). The other loci combinations EST3 and EST4 (θ=0.5); and EST4 and EST8 (θ=0.48) had much higher recombination fractions suggesting that population subdivision may be responsible for the linkage disequilibrium reported using GENEPOP. We therefore considered only Tr3.2 and EST12 to be linked and all other loci to be independent data points. The close linkage between Tr3.2 and EST12 loci enabled us to infer the most likely progenitor allele for mutations occurring at these two loci.

Detection of multiple paternity in litters

Multiple paternity was detected, by the presence of at least three nonmaternal alleles at two or more loci, in four of the 16 litters (25%) which had at least two offspring. We were able to assign fathers for all litters from Camel Hill as this area was extensively collected and all samples were typed (C.M.B. & G.A.D., unpublished work). Paternal genotypes were inferred for the litters from adjacent populations that had not been extensively sampled. Our establishment of the level of multiple paternity in the litters allowed us to infer mutations while taking into account the possibility of multiple paternity.

Mutations

We identified eight mismatches in 76 offspring that were most likely due to mutations of maternal alleles (Table 2a). The mutations identified were not due to experimental artefacts, such as allelic dropout, as most were from heterozygous mothers. Of the two mutations that were from homozygous mothers, one was a single positive repeat change and the other was a change to an allelic state four repeats less than the mothers allele. Additionally all new alleles were one of two alleles in heterozygous offspring with the other allele in the offspring matching an allele in the father (data not shown). Thus allelic dropout or null alleles could not explain any of the observed changes. We assumed that the mutation was derived from the maternal allele that required the least changes in repeat units. This assumption cannot be formally proven but is consistent with the common practice of other studies (e.g. Primmer et al., 1998 ; Crozier et al., 1999 ). Four mutations were consistent with being a single-step mutation, three gaining and one losing a single four-base pair repeat unit (Table 2a). None of the remaining maternal mutation events could be explained by the loss or gain of a single repeat unit: one involved a loss of two repeat units, another involved a loss of three repeat units and two were losses of four repeat units.

Table 2.   Observed maternal and paternal allele mutations. Thumbnail image of

Five paternal mutations could also be inferred ( Table 2b). We cannot entirely exclude the possibility of false assignment of paternity although this is unlikely due to the high discrimination power of the loci. We also cannot distinguish an inferred mutation from an extra pair fertilization (EPF) by a closely related male that differed at a single locus. However, we left out any possible mutations that could also be explained by EPF. Three of the paternal mutation events probably involved a single repeat (two gains and one loss) and two involved the loss of two repeats. One of these mutations (EST12 in offspring 2034, Table 2b) was considered more likely to have originated from allele 320 and not the closer allele 316, due to information about the linkage phase between this locus and Tr3.2. Two other offspring in this litter had inherited the same paternal allele 214 at the Tr3.2 locus and this allele was linked to the 320 allele at the EST12 locus for these individuals.

All of the observed mutations involved the increase or decrease of multiples of four bases. This indicates that the mutation events probably resulted from changes in the number of tandem repeats rather than insertions or deletions in the flanking region. One possible exception is the mutation observed in offspring 2045 at the EST2 locus (Table 2a). The closest allele to the new mutation differs by 10 base pairs but we considered the second allele, 16 base pairs different from the new mutation, is the more likely originating allele. Mutations that could be assigned to single tandem repeat changes accounted for 53.8% of all mutations.

The overall mutation rate of each locus ranged from 0 to 4.2% for the seven loci examined (Table 3). Mutation rates differed significantly among the loci (χ2=16.76, d.f.=6, P=0.013) (Tables 2a,b and 3). Although the most polymorphic locus (EST2) had the most mutation events, there was no significant correlation between mutation rate and number of alleles at a locus (R=0.683, P=0.091). There was also no significant correlation between percentage GC content of locus flanking regions (determined from the original cloned alleles, see Cooper et al., 1997 ; Gardner et al., 1999 ) and number of alleles (R=–0.646, P=0.117) or observed mutation rate (R =–0.464, P=0.294). However, we did observe that the most polymorphic and mutable locus (EST2 51%) had the lowest percentage GC and the least polymorphic locus (EST8 96%) had the highest percentage GC content (Table 1).

Table 3.   Mutation rates per locus. Thumbnail image of

We found no evidence for a variable mutation rate among different alleles at a locus as no allele was found to have mutated more than once. One litter had two offspring (2073 & 2074), with mutations from alternate paternal alleles (see Table 2b). As we have sampled all potential fathers near the mother of this litter we are confident that no additional male was responsible for paternity in these offspring. Additionally, we found one offspring (2034) with mutations originating from both its mother and father but those occurred at two different loci (see Table 2a,b).

Our data, with eight out of 13 observed mutations involving contraction of allele size (Table 2a,b & Fig. 2), showed no significant trend for mutations to lead to either an increase or decrease in allele size (χ2=0.337, d.f.=1, P=0.706). However, it is notable that all six mutations of greater than a single step involved contractions. All but three mutations were to alleles pre-existing in the population. Two of the exceptions were at the EST2 locus in offspring 809 and 2074. Offspring 2074 was from the ridges adjacent to Camel Hill that we have not sampled extensively so this allele may be present in other individuals at that site. The allele 340 at locus EST12 in offspring 803 represents the other unique mutation.

image

Figure 2.  Magnitude and direction of mutations observed at all loci in relation to the presumed progenitor allele. The smallest and largest repeats we observed (inferred from flanking region of each original cloned allele) were 13 and 41.5, respectively.

Download figure to PowerPoint

Heterozygote instability is where individuals with larger size differences between alleles at a particular locus are more likely to mutate than those with smaller size differences ( Amos et al., 1996 ). We tested if mutations occurred in the parent with the smaller or larger allele span (i.e. smallest or largest difference in allele size within a single locus) and found no significant difference. Eight occurred in the large span heterozygote parent compared with five in the small span parent (χ2=0.337, d.f.=1, P=0.706). We performed an additional test for heterozygote instability by comparing the allele spans in parents where mutations were observed in offspring to those parents whose offspring did not show mutations. Although there was a trend for the allele span in parents with observed mutations to be larger (n=12 mean= 19.83 vs. n=217 mean = 13.55), this was not significant (t=1.46; P=0.085).

Tests of mutation models from population data

For the two loci with fractional repeats, we considered that alleles differing by four base pairs at a locus had a more common evolutionary history than the other suite of alleles, also differing from each other by repeat lengths of four. Therefore, we split the analysis of Tr3.2 (suite A = alleles 176–232; B = alleles 190–234) and EST2 (suite A = alleles 176–232; B = alleles 190–234) into two sections per locus. Each suite was treated as a single ‘locus’ for the likelihood analysis. The strict one-step model was rejected in four instances (Table 4). One suite of alleles in each of Tr3.2 and EST2 and the loci EST3 and EST4 appear to follow a multistep model of evolution. Two of these loci, EST3 and EST4 were found to have a reduced level of heterozygotes due to the nonamplification of alleles. This probably violates assumptions of the underlying coalescence model ( Nielsen & Palsbøll et al., 1999 ) and we therefore cannot be confident with the results for EST3 and EST4. The rejection of the one-step model for locus EST2 is in agreement with our observations of mutations at this locus. However, the failure to reject the one-step model for EST12 is in contrast to the observed mutations of which half were multistep.

Table 4.   Results of the test on one-step mutation model, estimated proportion of multistep mutations and θ. Thumbnail image of

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The mutation rates we found at the tetranucleotide loci EST2 (4.2%) and EST12 (2.7%) are among the highest mutation rates reported so far for nondisease-causing microsatellite loci of any type. We have probably underestimated mutation rates for our loci by excluding all cases that were ambiguous as to either the direction or the parent these occurred in, or if they could have represented an EPF. Additionally, we probably underestimated male mutations by inferring the paternal genotypes in many litters. Reliable comparison of mutation rates between the sexes was not possible. Our results confirm the high mutability of some tetranucleotide loci. One other study reported comparable mutation rates (3.6%) at a tetranucleotide locus ( Primmer et al., 1996 , 1998). The average for our seven loci (1.3%) was high compared to other studies (e.g. average of 12 tetranucleotide loci = 0.21% Weber & Wong, 1993). Chakraborty et al. (1997 ) suggested that tetranucleotide loci have lower mutation rates than di- or trinucleotide repeat loci. To our knowledge, no studies of dinucleotide loci have found equivalent mutation rates to those of Primmer et al. (1996 ) or the present study. Perhaps high mutability at some tetranucleotide loci is quite common and indeed Chakraborty et al. (1997 ) suggested that the initial findings of higher mutability at tetranucleotide loci by Weber & Wong (1993) and colleagues, was due to a few highly mutable loci. This could indicate that mutability may be reliant on other characteristics not specific to any particular motif type.

Observed mutation rates differed among loci for the study, with most mutations (10 of 13) occurring at two of the seven loci (EST2 and EST12). This variation in mutation rates appears to be commonly found with microsatellite loci (e.g. Weber & Wong, 1993; Fitzsimmons, 1998; Primmer & Ellegren, 1998; Crozier et al., 1999 ). Although slipped strand mispairing is considered to be the major mechanism by which new alleles are generated ( Levinson & Gutman, 1987) our understanding of what makes one microsatellite locus more mutable than others is still limited. Jin et al. (1996 ) have also suggested that mutation rates differ among alleles at a single locus. By sequencing alleles with variable mutation rates at a single locus the authors showed that mutations were more common in alleles with long uninterrupted repeat arrays. Alleles were less likely to mutate where nucleotide substitutions shortened the length of uninterrupted repeat arrays. This finding may also explain differences among loci. We did not find any alleles that mutated more than once in the present study. Additionally, because we have not sequenced more than the initial allele from which primers were developed, we are unable to comment on whether the alleles that mutated were interrupted.

Other workers have suggested that the mutation rate may be positively correlated with the number of repeats ( Primmer et al., 1996 ). We did not observe any mutations from progenitor alleles under 21.5 repeats in length (Table 2 & Fig. 2) which supports this argument. However, our findings do not shed any light on mutation rates above this number of repeats. Observed mutations are spread throughout the range of repeat lengths (>21.5) for our loci ( Fig. 2). We found no evidence that either mutation rate or level of allelic diversity was correlated with percentage GC content of flanking sequence as suggested by Glenn et al. (1996 ). However, the most variable locus (EST2) had the lowest GC content of its flanking region, and the least variable locus (EST8) had the highest percentage GC, indicating a possible role for this characteristic in mutation rates at some loci.

Heterozygote instability ( Amos et al., 1996 ), another mechanism proposed to explain microsatellite mutations, suggests that heterozygotes with larger differences in allele size (large span) would be expected to mutate more than heterozygotes with smaller size differences in their alleles (small span). Our data do not support this mechanism although we found a trend for span differences to be greater in parents of offspring where mutations were observed. Although we did not detect a significant relationship for this phenomena with the statistical tests we ran, this does not mean the relationship does not exist. However, we may not have been able to detect it due to our inevitably small sample sizes. This also applies for several other relationships we tested. As this heterozygote instability model requires some interallelic interaction, it is not consistent with a slipped-strand mispairing mode of mutation ( Primmer et al., 1998 ) but could provide evidence for recombination as an important mechanism in microsatellite mutations.

Unlike other workers (e.g. Weber & Wong, 1993; Amos et al., 1996 ; Primmer et al., 1996 ), we found that mutations were not biased towards expansions. Therefore, despite hypervariability at two of our loci, we found no evidence for tetranucleotide loci to be always subject to expansion bias as suggested by Primmer et al. (1998 ). Indeed there was a trend for contractions when only multistep mutations were considered, suggesting that the loci are not at mutation-drift equilibrium. Unless there is selection preventing this contraction it would eventually result in the loss of all repeats. However, we did not observe any mutations under 21.5 repeats at any loci, suggesting the possibility of a constraint on repeat loss.

We observed a high proportion of the total mutations (46.2%) that could not be explained by a single-step mutation. Fitzsimmons (1998) found an even higher proportion of mutations (74%) that could not have been single-step changes at five microsatellite loci in the green turtle. Results of de-novo mutations from our study and that of Fitzsimmons (1998) suggest that the mutational processes at these microsatellite loci follow a two-phase model more closely than a single stepwise model. Additionally, the large number of multistep mutations observed indicates mutational processes may be more varied than previously thought. This is highlighted by Colson & Goldstein (1999) reporting complex mutations at microsatellite loci in Drosophila. They found large deletions at some loci as well as deletions and insertions in the flanking regions sometimes mimicking repeat loss or gain. Palsbøll et al. (1999 ) also presented allele sequence data suggesting a complexity of mutational processes and a high proportion of multistep and partial step mutations. They suggested that single-strand slippage of partial repeats may provide a mechanism for counteracting the continuous expansion of microsatellite loci. The presence of fractional repeats at two of our loci (Tr3.2 & EST2) provide further evidence for mutational processes other than slipped-strand mispairing.

It is conceivable that due to the common practice of assuming that a mutation has occurred from the allele closest in size to the mutated allele, many studies of de-novo mutations are underestimating the number of multistep mutations that have occurred. Using information from two closely linked loci, we found one mutated allele that most likely originated from an allele not closest to it in size, highlighting the possible flaw in this procedure. Data from other linked loci would be useful to test this closest allele assumption. Workers (e.g. Primmer et al., 1996 ) have generally stated that the assumption is warranted given the regularity of the distribution of length changes. However, a recent study based on allele size variation ( Colson & Goldstein, 1999) suggested that great care should be taken to select appropriate loci for population and phylogenetic inference as only seven of 17 loci they studied followed a stepwise mutation model. Because at least 46.2% of our mutations were multistep, it seems likely that our mutation data have more in common with a two-phase model ( Di Rienzo et al., 1994 ). The results of likelihood analyses on our allele data also indicate that assumptions based on allelic distributions may not reflect the underlying mutational processes. At a locus where we observed a high proportion (50%) of multistep mutations (EST12), likelihood analysis failed to reject conformance to a one-step model. The failure of this model to predict mutational characteristics for our loci highlights the need for more information on mutational processes from observations of de-novo mutations rather than allelic distributions alone. Additionally, the inability of this and other models to take into account loci with fractional repeats requires a reassessment of the modelling approach to include more complex mutational processes.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Leanne Haigh, Kathy Saint and Jan Birrell for technical assistance, and Steve Donnellan and Warwick Grant for helpful discussions. Staff at the Institute of Medical and Veterinary Science in Adelaide are thanked for running the microsatellite gels and for their friendly assistance. The manuscript was improved by comments from two anonymous reviewers. Funding was provided by the Australian Research Council and the South Australian Museum.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Amos, W., Sawcer, S.J., Feakes, R.W., Rubinsztein, D.C. 1996. Microsatellites show mutational bias and heterozygote instability. Nature Genet. 13: 390 391.
  • 2
    Chakraborty, R., Kimmel, M., Stivers, D.N., Davison, L.J., Deka, R. 1997. Relative mutation rates at di-, tri-, and tetranucleotide microsatellite loci. P.N.A.S. 94: 1041 1046.
  • 3
    Colson, I. & Goldstein, D.B. 1999. Evidence for complex mutations at microsatellite loci in Drosophila. Genetics 152: 617 627.
  • 4
    Cooper, S.J.B., Bull., C.M., Gardner, M.G. 1997. Characterization of microsatellite loci from the socially monogamous lizard Tiliqua rugosa using a pcr-based isolation technique. Mol. Ecol. 6: 793 795.
  • 5
    Crozier, R.H., Kaufmann, B., Carew, M.E., Crozier, Y.C. 1999. Mutability of microsatellites developed for the ant Camponotus consobrinus. Mol. Ecol. 8: 271 276.
  • 6
    Di Rienzo, A., Peterson, A.C., Garza, J.C., Valdes, A.M., Slatkin, M., Freimer, N.B. 1994. Mutational processes of simple-sequence repeat loci in human populations. P.N.A.S. 91: 3166 3170.
  • 7
    Duffield, G.A. & Bull., C.M. 1996. Characteristics of the litter of the gidgee skink Egernia stokesii. Wildlife Res. 23: 337 342.
  • 8
    Estoup, A., Tailliez, C., Cornuet, J.M., Solignac, M. 1995. Size homoplasy and mutational processes of interrupted microsatellites in two bee species, Apis mellifera and Bombus terrestris (Apidae). Mol. Biol. Evol. 12: 1074 1084.
  • 9
    Feldman, M.W., Bergman, A., Pollock, D.D., Goldstein, D.B. 1997. Microsatellite genetic distances with range constraints – analytic description and problems of estimation. Genetics 145: 207 216.
  • 10
    Fitzsimmons, N.N. 1998. Single paternity of clutches and sperm storage in the promiscuous green turtle (Chelonia mydas) . Mol. Ecol. 7: 575 584.
  • 11
    Gardner, M.G., Cooper, S.J.B., Bull., C.M., Grant, W.N. 1999. Isolation of microsatellite loci from a social lizard, Egernia stokesii, using a modified enrichment procedure . J. Hered. 90: 301 304.
  • 12
    Glenn, T.C., Stephan, W., Dessauer, H.C., Braun, M.J. 1996. Allelic diversity in alligator microsatellite loci is negatively correlated with GC content of flanking sequences and evolutionary conservation of PCR amplifiability. Mol. Biol. Evol. 13: 1151 1154.
  • 13
    Guo, S.W. & Thompson, E.A. 1992. Performing the exact test of Hardy–Weinberg proportions for multiple alleles. Biometrics 48: 361 372.
  • 14
    Haldane, J.B.S. 1954. An exact test for randomness of mating. J. Genetics 52: 631 635.
  • 15
    Jin, L., Macaubas, C., Hallmayer, J., Kimura, A., Mignot, E. 1996. Mutation rate varies among alleles at a microsatellite locus – phylogenetic evidence. P.N.A.S. 93: 15285 15288.
  • 16
    Kimura, M. & Crow, J.F. 1964. The number of alleles that can be maintained in a finite population. Genetics 49: 725 738.
  • 17
    Lander, E. & Green, P. 1987. Construction of multilocus genetic linkage maps in humans. P.N.A.S. 84: 2363 2367.
  • 18
    Levinson, G. & Gutman, G.A. 1987. Slipped-strand mispairing: a major mechanism for DNA sequence evolution. Mol. Biol. Evol. 4: 203 221.
  • 19
    Marshall, T.C., Slate, J., Kruuk, L.E.B., Pemberton, J.M. 1998. Statistical confidence for likelihood-based paternity inference in natural populations. Mol. Ecol. 7: 639 655.
  • 20
    Miller, R.G. 1980. Simultaneous Statistical Inference. Springer Verlag, New York.
  • 21
    Nielsen, R. 1997. A likelihood approach to populations samples of microsatellite alleles. Genetics 146: 711 716.
  • 22
    Nielsen, R. & Palsbøll, P.J. 1999. Single-locus tests of microsatellite evolution: multi-step mutations and constraints on allele size. Mol. Phylog. Evol. 11: 477 484.
  • 23
    Ohta, T. & Kimura, M. 1973. A model of mutation appropriate to estimate the number of electrophoretically detectable alleles in a finite population. Genet. Res. 22: 201 204.
  • 24
    Palsbøll, P.J., Berube, M., Jorgensen, H. 1999. Multiple levels of single-strand slippage at cetacean tri- and tetranucleotide repeat microsatellite loci. Genetics 151: 285 296.
  • 25
    Pemberton, J.M., Slate, J., Bancroft, D.R., Barrett, J.A. 1995. Nonamplifying alleles at microsatellite loci: a caution for parentage and population studies. Mol. Ecol. 4: 249 252.
  • 26
    Primmer, C.R. & Ellegren, H. 1998. Patterns of molecular evolution in avian microsatellites. Mol. Biol. Evol. 15: 997 1008.
  • 27
    Primmer, C.R., Ellegren, H., Saino, N., Moller, A.P. 1996. Directional evolution in germline microsatellite mutations. Nature Genet. 13: 391 393.
  • 28
    Primmer, C.R., Saino, N., Møller, A.P., Ellegren, H. 1998. Unraveling the processes of microsatellite evolution through analysis of germ line mutations in barn swallows Hirundo rustica. Mol. Biol. Evol. 15: 1047 1054.
  • 29
    Queller, D.C., Strassmann, J.E., Hughes, C.R. 1993. Microsatellites and kinship. TREE 8: 285 288.
  • 30
    Raymond, M. & Rousset, F. 1995a. An exact test for population differentiation. Evolution 49: 1280 1283.
  • 31
    Raymond, M. & Rousset, F. 1995b. GENEPOP (Version 1.2): population genetics software for exact tests and ecumenicism. J. Hered. 86: 248 249.
  • 32
    Rousset, F. & Raymond, M. 1995. Testing heterozygote excess and deficiency. Genetics 140: 1413 1419.
  • 33
    Sambrook, J., Fritsch, E.F., Maniatis, T. 1989. Molecular Cloning: a Laboratory Manual, 2nd edn. Coldsprings Harbour Laboratory Press, New York.
  • 34
    Schlotterer, C. & Tautz, D. 1992. Slippage synthesis of simple sequence DNA. Nuc. Acids Res. 20: 211 215.
  • 35
    Schug, M.D., Hutter, C.M., Wetterstrand, K.A., Gaudette, M.S., Mackay, T.F.C., Aquadro, C.F. 1998. The mutation rates of di-, tri- and tetranucleotide repeats in Drosophila melanogaster. Mol. Biol. Evol. 15: 1751 1760.
  • 36
    Valdes, A.M., Slatkin, M., Freimer, N.B. 1993. Allele frequencies at microsatellite loci: The stepwise mutation model revisited. Genetics 133: 737 749.
  • 37
    Walsh, P.S., Metzger, D.A., Higuchi, R. 1991. Chelex 100 as a medium for simple extraction of DNA for PCR-based typing from forensic material. Biotech. 10: 506 513.
  • 38
    Wattier, R., Engel, C.R., Saumitoulaprade, P., Valero, M. 1998. Short allele dominance as a source of heterozygote deficiency at microsatellite loci – experimental evidence at the dinucleotide locus Gv1ct in Gracilaria gracilis (Rhodophyta). Mol. Ecol. 7: 1569 1573.
  • 39
    Weber, J.L. & Wong, C. 1993. Mutation of human short tandem repeats. Hum. Mol. Genet. 2: 1123 1128.
  • 40
    Weir, B.S. 1990. Genetic Data Analysis. Sinauer Publishers, Sunderland, MA.
  • 41
    Zaykin, D.V. & Pudovkin, A. 1993. Two programs to estimate significance of chi-square values using pseudo-probability tests. J. Hered. 84: 152152.