MBE Advance Access originally published online on February 22, 2006
Molecular Biology and Evolution 2006 23(5):973-980; doi:10.1093/molbev/msj112
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Proceedings of the SMBE Tri-National Young Investigators' Workshop 2005 |
Genome-Wide Associations Between Hybrid Sterility QTL and Marker Transmission Ratio Distortion
* Center for Population Biology, University of California, Davis; and Tomato Genetics Resource Center, University of California, Davis
E-mail: lmoyle{at}indiana.edu.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Marker transmission ratio distortion (TRD) in genetic mapping populations is frequently ascribed to selection against allelic combinations that cause hybrid incompatibility. Accordingly, genomic regions of TRD should be nonrandomly associated (colocated) with loci that underlie hybrid incompatibility. To directly test this hypothesis, we evaluated the genome-wide qualitative and quantitative agreement between chromosomal regions exhibiting marker TRD and those known to contain hybrid incompatibility quantitative trait locus (QTL). Incompatibility data came from a near-isogenic line (NIL) analysis of pollen and seed sterility in a cross between two Solanum (formerly Lycopersicon) species. We assessed (1) whether these incompatibility loci are colocated with markers that show significant TRD in two earlier generations preceding these introgression lines and (2) whether the magnitude of marker distortion quantitatively matches the estimated strength of selection against each incompatibility locus. We found evidence that TRD regions are chromosomally colocated with hybrid incompatibility loci more frequently than is expected by chance: pollen sterility QTLs were most closely associated with distorted heterozygote frequencies in later-generation backcrosses. Nonetheless, there was no evidence for an association between TRD and seed sterility and little evidence of a quantitative association between the magnitude of marker TRD and the fitness effects of heterospecific alleles at each chromosomal location. We propose and test a model (the "dance partner" model) to explain several cases where regions of TRD are not associated with hybrid incompatibility loci. Under this model, some NILs containing greater than one heterospecific introgression may not express hybrid incompatibility phenotypes because they carry both appropriate genetic dance partners required for a fully functional interaction. Accordingly, negative interactions expressed in earlier backcross generations are masked in these double-introgression NILs. Based on this model, we identify the location of several new putative pairwise interactors underlying hybrid incompatibility in this species cross.
Key Words: hybrid incompatibility • segregation • marker distortion • speciation • QTL mapping • NIL • Lycopersicon
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Transmission ratio distortion (TRD) describes the significant deviation of allele (or genotype) frequencies from simple Mendelian expectations within a known pedigree (Pardo-Manuel de Villena et al. 2000
). Marker TRD is rampant in contemporary quantitative trait locus (QTL) studies: from 10% to 90% of genetic markers have shown significant TRD in dozens of studies across a diverse range of organisms including wildflowers (e.g., Hall and Willis 2005), insects (e.g., Tan et al. 2001; Solignac et al. 2004; Wang and Porter 2004), marine invertebrates (e.g., Li and Guo 2004), fish (e.g., Kocher et al. 1998), trees (e.g., Yin et al. 2002; Myburg et al. 2004), and many crop plant crosses (Jenczewski et al. 1997; Rieseberg, Baird, and Gardner 2000 and references therein). There are several potential causes of these departures from Mendelian expectations. First, active processes that influence the segregation of alleles immediately prior to, during, and immediately following meiosis (e.g., direct gamete competition, meiotic drive, and male killing) can produce distorted ratios at affected loci. These processes are undoubtedly responsible for several well-studied cases of individual loci showing transmission distortion (e.g., Pardo-Manuel de Villena et al. 2000; Fishman and Willis 2005; Orr and Irving 2005). However, these mechanisms would have to be a ubiquitous force acting throughout the genome in order to generally account for the high frequency of genome-wide marker distortion—a hypothesis that is inconsistent with the currently available data (Hall and Willis 2005). Alternatively, TRD, especially as observed in distant crosses (i.e., between subspecies or species), could be due to inadvertent selection against dysfunctional heterospecific allelic combinations. Hybrid incompatibility is known to be frequently due to disrupted genetic interactions among loci that have diverged in isolated parental lineages (Orr and Turelli 2001). When hybrids between these lineages are created, these negative interactions can cause inviability and/or sterility in particular recombinant genotypes, such that they are removed by natural selection from hybrid populations. This nonrandom elimination of specific allelic combinations is the explanation most commonly invoked to explain high rates of marker TRD in QTL mapping populations (e.g., Fishman and Willis 2001; Harushima et al. 2001; Hall and Willis 2005).
One corollary of this explanation is that loci causing hybrid incompatibility should be located at or near regions of TRD observed in hybrid populations. Indeed, in some cases, patterns of TRD alone have been used to infer the genetic basis of hybrid incompatibility among species (e.g., Li et al. 1997; Harushima et al. 2001; Myburg et al. 2004). However, the genome-wide empirical correspondence between hybrid incompatibility loci and marker transmission distortion has never been directly evaluated. In this paper, we test the hypothesis that genomic regions that display TRD are associated with hybrid incompatibility QTL by evaluating (1) the qualitative agreement (frequency of genomic colocation) between chromosomal regions showing significant TRD and chromosomal regions where hybrid incompatibility QTLs have been mapped and (2) the quantitative association between the magnitude of TRD at specific markers and the magnitude of hybrid incompatibility associated with these locations. These relationships were evaluated using data from an interspecific cross between two closely related plant species, Solanum lycopersicum (formerly Lycopersicon esculentum—the cultivated tomato) and Solanum habrochaites (formerly Lycopersicon hirsutum—a wild congener). Based on the results of these tests, we propose and evaluate two mechanisms to explain cases where genomic regions of strong TRD are not associated with known hybrid incompatibility QTL. Combining data on genome-wide patterns of TRD and hybrid incompatibility in this way allows us to better evaluate and understand the genetic basis and complexity of loci underlying barriers to gene flow between species.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
All data on TRD and hybrid incompatibility used in our comparison came from later-generation individuals that descended from a single cross between a S. habrochaites (SH) pollen parent and a S. lycopersicum (SL) ovule parent (see Bernacchi et al. 1998
; Monforte and Tanksley 2000). These species are closely related, diploid, and chromosomally colinear and are therefore likely separated primarily by genic factors rather than by large-scale changes in chromosome organization (Quiros 1991). Genotype data used to assess TRD were kindly provided by A. Monforte; these raw data were the basis of a previously published study of allele frequencies at 96 marker locations (roughly evenly distributed throughout the genome) in 111 BC2 individuals and in each of 111 BC2S3 lines (described below) descended from these BC2 individuals (Monforte and Tanksley 2000). Data on hybrid incompatibility QTL were taken from a previously published analysis of hybrid pollen and seed sterility in 71 near-isogenic lines (NILs) descended from the same BC2S3 lines above (Moyle and Graham 2005). All BC2, BC2S3, and NILs were generated by Monforte and Tanksley (2000), who provided a detailed description of their construction. Briefly, from the original cross, a single F1 individual was backcrossed as the male parent to S. lycopersicum to generate a population of BC1 individuals that were backcrossed again (as female parents) to S. lycopersicum to produce a large BC2 population (Monforte and Tanksley 2000). From this population, 111 BC2 individuals were selfed through single-seed descent for three generations to produce 111 BC2S3 descendent lines (Monforte and Tanksley 2000). To generate a population of NILs, the BC2S3 population was then subjected to two rounds of marker-assisted selection to minimize the number of SH introgressed segments found within each line while maximizing genome-wide SH representation. This generated a set of 99 NILs, each of which carries one or more marker-characterized SH chromosomal introgressions in homozygous form in an isogenic SL background (Monforte and Tanksley 2000). Of these 99 NILs, 71 were used in a replicated randomized phenotyping experiment (N = 240) that evaluated hybrid incompatibility in terms of both pollen and seed fertility (Moyle and Graham 2005, see below).
TRD Data
Using the raw genotype data from Monforte and Tanksley (2000) for 96 markers in both the BC2 and BC2S3 populations, we generated several measures of TRD. For the BC2 generation, we calculated the observed number of S. habrochaites alleles (
) and the deviation of this from the expected number, given an expected frequency of 0.125 in a BC2 generation. The significance of this deviation was determined at each marker locus using a 
2 test and an arbitrary but conservative statistical cutoff (
= 0.005). For the BC2S3 generation, TRD was evaluated in two different ways. First (as in the BC2 population), we determined the observed number of H alleles (
) and whether they deviated significantly from expected values. Second, we determined the number of observed S. lycopersicum/hirsutum heterozygotes (
) in this population and evaluated whether this number deviated from the expected frequency. (Note that it is redundant to assess heterozygote frequencies in the BC2 population as all H alleles are in heterozygous form in this population.) In both cases, expected frequencies in the BC2S3 were conditioned on genotypic and allelic frequencies observed in the BC2 generation, using equations outlined by Monforte and Tanksley (2000); deviations from expectations were evaluated using 2 tests as above at each locus. If TRD is caused by the expression of hybrid incompatibility, then significant underrepresentation of H alleles in either generation indicates failure of allele transmission at these or closely linked loci in the foreign genetic background of the other species (expected to comprise on average 0.875 of the total genome in both BC2's and BC2S3's). Distorted genotype frequencies indicate differences in the relative fitness of specific allelic combinations at a locus. Overrepresentation of heterozygotes in the BC2S3 population, for example, could be due to overdominance at a locus or selection against homozygote genotypes (see Results).
For each measure of TRD, a transmission ratio distortion locus (TRDL) was assigned to each genomic location when a contiguous run of significantly distorted marker loci was detected, that is, a contiguous section of chromosome along which all markers showed significant distortion in the same direction. Only one TRDL was assigned for a single contiguous run, even if an entire chromosome showed significant marker distortion in the same direction. Overall, we identified 9, 14, and 4 TRDLs for 
and 
respectively (table 1). In addition, in order to evaluate the quantitative association between TRD and incompatibility, the relative magnitude of TRD ([expected–observed] frequency) was calculated at each marker for each measure of TRD.
|
Hybrid Incompatibility Data
Hybrid incompatibility phenotypes and associated QTLs used here were those previously determined in a study of 71 NILs of S. habrochaites in a S. lycopersicum background (Moyle and Graham 2005
), descended from the same later-generation recombinants genotyped for TRD (see above). In this NIL study, a total of 85% of the S. habrochaites genome was represented by introgressions in the S. lycopersicum background. The following fitness traits from Moyle and Graham (2005) were used in our analyses: pollen number (PN: the total number of pollen grains per flower), pollen fertility (PF: the proportion of fertile pollen grains per flower), and seed set (SSS: the total number of seeds set per self-pollination). In our prior analyses, we detected four QTLs for PN, eight for PF, and five for SSS (Moyle and Graham 2005). (Note that only SSS QTLs that were independent of PF were evaluated here; see Moyle and Graham 2005.) In addition, to evaluate quantitative associations, we needed to assign an individual fitness estimate (measured in terms of pollen production [PN], average PF, and average seed fertility [SSS]) to each genomic marker. To do so, we calculated the mean trait value associated with an H allele at each marker by averaging the trait values of all NIL genotypes known to carry an H allele at that locus, for each of PN, PF, and SSS traits.
Analyses
Qualitative Associations
To assess the degree of association between genomic regions with significant TRD in the BC2 and BC2S3 populations and genomic regions in which hybrid incompatibility factors had been mapped in the NIL population, we evaluated the probability (P) that these two kinds of QTLs were genomically associated (found at the same or overlapping chromosomal locations) more frequently than expected by chance. This was estimated using the hypergeometric probability distribution function (see Paterson 2002) according to which
 |
). In our analysis, an interval was defined as 35 cM—approximately the average size of individual introgressions in the NIL study (Moyle and Graham 2005). Given this, and a whole genome size of 1,260 cM in tomato, n = 36. When P is less than 0.05, it is considered significantly unlikely that the observed number of QTL matches would occur by chance. Tests were individually performed to assess the qualitative genomic association between TRD in each of the BC2 and BC2S3 populations and each of the QTLs for PN, PF, and SSS. Note that because hybrid individuals were used as both males (in the F1, to generate the BC1 population) and females (in the BC1, to generate the BC2 population) (see above), there was the opportunity for both male and female gamete selection to influence observed marker distortion in both BC2 and BC2S3 populations.
Quantitative Associations
To determine the quantitative association between the magnitude of transmission distortion and the magnitude of hybrid incompatibility associated with specific genomic locations, we assessed the strength and significance of a regression between the relative fitness (in terms of PN, PV, and SSS) associated with each marker location and the magnitude of transmission distortion observed at each marker. Relative fitness was estimated by subtracting the average absolute fitness of the H allele at each marker location from the observed average PN, PF, and SSS of the pure S. lycopersicum parent. Values less than zero, therefore, indicate marker loci associated with reduced fitness (hybrid incompatibility) in comparison to the parental S. lycopersicum genotype and values above zero indicate average increases in fitness (heterosis) in comparison to S. lycopersicum. (Note that these deviations need not be statistically significant in any individual case.) This analysis therefore assesses whether the strength of the distortion observed at a specific maker locus is on average proportional to the estimated strength of selection against (or for) the genomic location where it is located. This relationship was assessed for TRD in each generation (i.e., BC2 and BC2S3) against fitness measured in terms of PN, PF, and SSS.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
In several cases, we detected significant genome-wide qualitative associations between TRDLs and QTLs for PN and PF (table 2). In particular, regions in the BC2S3 generation with distorted heterozygote frequencies (i.e.,
) were genomically associated more frequently than expected by chance with QTL for both PN and PF (table 2). Of the four regions with significantly distorted BC2S3 heterozygote frequencies, two were found in the same genomic location as PN QTL and three overlapped with PF QTL (table 1). All the significant TRDLs for showed a significant excess of heterozygotes; in two cases (CT140, CT195), this appears to be associated with selection against (i.e., a deficit of) homozygote S. habrochaites genotypes at these locations, and in two cases (TG607, TG620) with a deficit of homozygote S. lycopersicum genotypes (data not shown). In addition, PN was significantly associated with distorted allele frequencies in the BC2 population () (table 2). In each case, the PN QTL colocalized with TRDL where H alleles were significantly underrepresented (table 1), consistent with selection against S. habrochaites alleles at these locations. In contrast to pollen incompatibility traits, although some individual TRDL appeared to colocalize with seed sterility QTL (table 1), no genome-wide significant association was detected between seed sterility QTL and TRDL in either generation (table 2).
|
In terms of quantitative associations, we detected few weak correlations between degree of marker distortion and estimated relative fitness at each marker location. The strongest (and only significant) association was detected between relative fitness for PN and magnitude of deviation in heterozygote frequencies in the BC2S3 population (fig. 1 and table 3); however, even in this case, the amount of variation explained was modest (<8%). Heterozygote distortion was also marginally associated with the other two hybrid incompatibility traits—PF and self SSS (table 3). For all these relationships, a positive slope indicated that reduced relative fitness (i.e., hybrid incompatibility) at a marker location was on average weakly associated with an overrepresentation of heterozygotes, consistent with observed qualitative patterns of distortion. Overall, these generally weak or absent quantitative associations might be due to the fact that only a modest number of QTLs underlie hybrid incompatibility in this species cross and that none of these individually causes complete hybrid sterility. For the vast majority of markers, therefore, there is only a small and nonsignificant deviation from the fitness of the pure parental S. lycopersicum genotype and relatively little variation with which to detect a significant association. In the few cases where there is a weak or marginal association, several extreme values (at very low and high relative fitness) appear to be responsible for the relationship (e.g., fig. 1).
|
|
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
The marker TRD that is commonly observed in recombinant hybrid populations is frequently attributed to the nonrandom elimination of linked hybrid incompatibility factors. Here we have provided evidence that genome-wide regions of TRD are in fact associated with hybrid incompatibility QTLs that influence pollen production and sterility, more frequently than expected by chance. It is worth noting that most significant associations between TRD and hybrid incompatibility traits were revealed in the BC2S3 generation. In comparison to the BC2 population, this population is the one in which homozygous regions of both S. lycopersicum and S. habrochaites alleles can be formed (via the preceding three generations of selfing). In many previous studies, patterns of hybrid incompatibility suggest that underlying QTLs are frequently recessive; that is, hybrid incompatibility is most strongly expressed when homozygous regions from both parental species are able to interact epistatically (the "dominance theory," Turelli and Orr 2000
). Our results are consistent with the expression of hybrid incompatibility QTL in the generation that allows these recessive interactions to be expressed.
Nonetheless, given the common expectation that marker TRD is caused by proximity to hybrid incompatibility loci, a better correspondence between hybrid incompatibility and TRD might have been expected. In particular, at least 11 (of 22) chromosomal locations that show TRD in one or both of the earlier generations are not apparently associated with any hybrid incompatibility factors identified in the NIL study. If TRD is predominantly due to the expression of hybrid incompatibilities, what might explain this disagreement? TRD in these genomic regions might be due to the action of selection at developmental stages apart from pollen and seed fertility, including seed germination and seedling, juvenile, and adult mortality. Although this explanation cannot be definitively excluded, we observed little evidence for systematic fitness differences between genotypes at these other stages (L. C. Moyle and E. B. Graham, unpublished data). In contrast, a previous analysis of the BC1 generation of this interspecific cross identified a number of QTLs associated with reduced crossing success (proportion of pollinated flowers that produced fruit with seeds) in both self-fertilizations and backcrosses with tomato (Bernacchi and Tanksley 1997). Several of these QTLs appear to be colocated with TRDLs in the later generations analyzed here (table 1), suggesting that deleterious interspecific interactions acting during pollination and/or immediately following fertilization (but not detected as reduced seed production in the NIL study) might be responsible for marker distortion at some of the remaining TRDLs.
Another plausible alternative for unexplained TRDL is that some interactions responsible for TRD in earlier BC generations no longer operate in the descendent NIL population that was used to evaluate hybrid sterility. At least two mechanisms could account for this. First, complex genetic interactions might contribute to the expression of some hybrid incompatibility traits (and therefore result in TRD) in the BC populations, and these complex interactions might not be detectable in NIL population. This is because BC and NIL populations differ systematically in the number and kind of possible interactions between parental genomes: recombinant individuals in early BC populations can experience epistatic interactions among multiple factors from each of the parental genomes, whereas epistasis in each NIL genotype is limited to interactions between the few genomic regions represented from one species interacting with the second species' genetic background. One prediction of this "complex epistasis" mechanism is that the more factors (introgressed regions) from one species that are present in the other species' genetic background, the more deleterious interactions that can occur and the stronger the expected expression of hybrid incompatibility. This prediction is testable using data from our previous NIL analysis because one-third of the examined NIL genotypes contained either two or three introgressed regions from SH into SL, rather than a single SH genomic introgression. In comparison to "single" NILs, these "double-" and "triple-" introgression NILs can experience interactions among more than one chromosomal region. When we compare mean expression of hybrid incompatibility among NILs subdivided into those with one, two, and three introgressions from SH, however, we find no statistically significant effect of the number of introgressed chromosomal regions on the fitness of NIL genotypes (i.e., on the expression of hybrid incompatibility; table 4). Note that because NILs with more than one introgression have both more opportunity for interactions between chromosomal regions and a greater absolute amount of SH genome represented in the SL genetic background, we simultaneously included a covariate of total introgression length in centimorgans in our analysis (see Moyle and Graham 2005). Although somewhat limited (e.g., by only addressing the effects of one vs. few introgressions), this test provides little evidence that the absence of multi-interaction epistasis (interactions among three or more loci) in the NIL population is primarily responsible for the lack of correspondence between TRD and hybrid incompatibility QTL.
|
An alternative (but not necessarily mutually exclusive) explanation is that some interactions that were expressed in earlier recombinant BC populations are masked in NILs due to cointrogression of genomic regions that would otherwise interact to cause hybrid incompatibility. In particular, some double- and triple-introgression NILs might not express hybrid incompatibility because they contain both conspecific genetic components—genetic "dance partners"—required for a fully functional interaction. This dance partner model generates several predictions concerning the marker TRD at these double- and triple-introgression locations. First, TRD observed in the earlier generations should, on average, occur more frequently at markers with H alleles that are found exclusively in double- or triple-introgression NILs, in comparison to markers at which H alleles are represented in single-introgression NILs. This is because some proportion of these double- or triple-introgression NILs include markers involved in sufficiently deleterious interactions that they are only found in NIL populations when cointrogressed with their appropriate conspecific partner. To test this, we compared the number of times markers were significantly distorted in each of the BC2 and BC2S3 generations between two groups: markers that were found exclusively in double- or triple-introgression NILs versus markers found in at least one single-introgression NIL. Consistent with our prediction, we found that markers found only in multi-introgression NILs were distorted more often in the BC2 population than markers found in at least one single-introgression NIL (P = 0.012; Fisher's exact test). This difference was not found in the BC2S3 population (P = not significant for both
and 
; Fisher's exact test). A second similar prediction of the dance partner model is that markers found exclusively in double- and triple-introgression NILs should show greater average magnitudes of distortion in prior BC generations than markers found in single-introgression NILs. We compared the average magnitude of TRD in these two groups of markers and found mean TRD to be significantly larger in the double- and triple-introgression marker group than in the single-introgression markers in the BC2 population (F = 8.282, P = 0.005; one-way analysis of variance [ANOVA]; fig. 2), though not in the BC2S3 (F = 2.617, P = 0.11 for and F = 1.167, P = 0.28 for ; one-way ANOVAs). Both tests support the predictions of the dance partner model in the BC2 population and suggest that this hypothesis is a plausible explanation for the failure to find hybrid incompatibility phenotypes associated with some regions that show substantial TRD in this earlier generation.
|
One implication of this model is that it should be possible to identify the interacting partners responsible for hybrid incompatibility by combining information on patterns of TRD in early recombinant generations with information on hybrid incompatibility phenotypes and patterns of cointrogression in later near-isogenic generations. These pairs of genomic loci should show TRD in earlier BC2 and/or BC2S3 generations, co-occurrence in double or triple NILs, but no significant hybrid incompatibility phenotype in the NIL population. We examined patterns of TRD in markers found in double- and triple-introgression lines to evaluate whether such pairs could be identified. In two cases, we found evidence for pairwise interacting partners from S. habrochaites: one pair involves loci on chromosomes 3 and 6 and the second involves loci on chromosomes 10 and 11. For each pair, the markers showing the most exaggerated segregation distortion at these locations are candidate regions for loci involved in hybrid incompatibility between S. habrochaites and S. lycopersicum.
In combination, TRDL that showed significant associations with hybrid incompatibility QTL and the four additional regions identified under the dance partner model account for 85% of the regions that show significant TRD in the BC2 population and 50% of the regions significantly distorted in the BC2S3 population. What might account for those TRDLs that remain apparently unassociated with incompatibility regions? Several statistical or biological factors might account for these remaining unexplained TRDLs. First, not all distorted loci will necessarily have a biological basis; in any individual case, distortion could be due to experimental factors such as genotyping error, smaller sample sizes, and related statistical effects (Montagutelli, Turner, and Nadeau 1996). However, contiguous runs of loci distorted in the same direction suggest a genuine underlying biological basis for observed TRD (Hall and Willis 2005). In our study, only three of the "unmatched" TRDLs consist of a single distorted marker; while these three distorted markers could potentially result from nonbiological factors, the majority of the unexplained TRDLs are likely underpinned by real biological causes. Second, because different methods of measurement and statistical analysis are involved in their assessment, measurements of ratio distortion might be more sensitive than those of incompatibility effects, so that it may simply be easier to detect TRD than hybrid incompatibility loci. Accordingly, some unmatched TRDLs might be associated with incompatibility loci that have failed to make more stringent statistical cutoffs. Our data on hybrid incompatibility suggest that at least three unexplained BC2S3 TRDLs on chromosomes 10, 11, and 12 are likely explained by such weak effect, statistically nonsignificant, hybrid incompatibility loci.
Several biological rather than statistical factors may also account for the remaining unmatched TRDLs. For example, some TRDLs might be due to instances where selection acts against certain allelic combinations without producing evident consequences in terms of observed fitness. For example, if more ovules are fertilized than will be matured, differential abortion of specific genotypes can take place without a substantial reduction in seed count from the resulting fruits. (A similar argument has been proposed in the context of TRD in mice [Montagutelli, Turner, and Nadeau 1996].) Alternatively, although our initial analyses showed no evidence that epistasis among three or more loci could explain observed TRDLs, the inclusion of multi-introgression dance partner NILs (with accordingly suppressed hybrid incompatibility phenotypes) in this test may have weakened our ability to detect these complex epistasis effects. To assess this possibility, we reevaluated the prediction of the complex epistasis mechanism after having removed those NILs that contain putative dance partners. In this new analysis, we detected a significant effect of introgression number on hybrid pollen inviability, such that NILs with one, two, or three SH introgressions showed progressively more mean pollen sterility on average (F = 3.995, P = 0.023, introgression number effect on pollen sterility; ANOVA with introgression size included as covariate). No newly significant effects were detected for PN or SSS traits (data not shown). This result is consistent with the hypothesis that complex epistatic interactions among two or more introgressed regions from a single species may contribute to the expression of hybrid pollen sterility in earlier BC generations. In this context, it is also worth noting that the original BC1 generation of this interspecific cross showed evidence of interactions between QTL associated with reduced crossing success (Bernacchi and Tanksley 1997; see above); although these interactions were able to operate in the subsequent recombinant BC generations, based on their reported marker locations we confirmed that none of them could operate in the NIL population, likely contributing to the failure to detect their effect in this generation. Accordingly, for loci affecting hybrid sterility, there is some evidence that differences in the complexity of genetic interactions in BC versus NIL populations might also contribute to the lack of correspondence between TRDL and hybrid incompatibility QTL, once putative dance partner NILs have been taken into account.
Finally, some of the individual unexplained TRDLs might be the result of active distorting processes outlined at the outset of the paper, including differential gamete competitiveness and/or the action of meiotic drivers. One TRDL in particular, located at the end of chromosome 10, produces an overrepresentation of S. habrochaites alleles at this location; this TRDL could be due to competitive superiority of pollen carrying S. habrochaites alleles near this locus or a linked meiotic driver that favors S. habrochaites alleles. Conversely, several observed TRDLs might be involved in the expression of stylar incompatibility during pollination (Bernacchi and Tanksley 1997; Monforte and Tanksley 2000; table 1), causing underrepresentation of H alleles at these loci. Other remaining TRDLs might also be candidate regions for loci involved in such processes, including meiotic drivers that actively influence allelic segregation during female gametogenesis, as has recently been documented in other plant systems (Fishman and Willis 2005). Which of these statistical or biological factors might be responsible for the marker distortion observed at any specific TRDL remains to be investigated.
Despite these unexplained loci, it is reassuring that the majority of TRDLs appear to be explicable in terms of direct associations with hybrid incompatibility or indirect signatures of their effect as predicted under the dance partner model. Several additional implications follow if the dance partner model appropriately describes some of the introgression and fitness patterns observed in NIL populations. First, analyses of hybrid incompatibility in NIL populations will tend to underestimate the number of loci involved in hybrid incompatibility (Moyle and Graham 2005), in any case where cointrogression of interacting genetic partners masks expression of hybrid incompatibility phenotypes. Second, two or more homospecific chromosomal introgressions that cannot be dissociated via recombination during the development of NILs are potential dance partners. Accordingly, patterns of cointrogression might be used to identify interacting genetic loci that underlie hybrid incompatibility. Of course, it is important to be cautious when interpreting such data; for example, if the priority in generating backcross recombinant inbred lines is to represent all chromosomal regions from one species in the genetic background of another, then occasional double introgressions could be due to pragmatic logistical decisions against further crossing effort, rather than genuine underlying biological causes. In addition, our analyses show that NILs that contain more than one introgression will likely be a heterogeneous class that includes some lines with introgressions that interact "positively" to relieve hybrid incompatibility (i.e., dance partners) as well as other lines with introgressions that interact "negatively" to increase the expression of hybrid incompatibility for some kinds of traits (i.e., complex negative epistasis). Accordingly, not all cointrogressed regions will be genetic dance partners. However, as we have shown here, combining data on cointrogression with hybrid incompatibility analyses and TRD information provides an opportunity to identify new genomic locations of interacting hybrid incompatibility factors and to gain insight both into the coadaptation of loci within species and the genetic basis and complexity of reproductive isolation and speciation between lineages.
|
Conclusions |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
We detected evidence that regions of substantial marker TRD co-occur with QTL that underlie hybrid incompatibility, especially pollen sterility traits, between S. lycopersicum and S. habrochaites. This is the first direct genome-wide demonstration of this correspondence. Associations were strongest with distortion patterns in the population in which homozygous regions from both parental species can interact, consistent with the recessivity of hybrid incompatibility factors underlying these traits. We evaluated two alternative (but not mutually exclusive) models to explain cases where regions of substantial TRD in earlier BC populations do not appear to be associated with hybrid incompatibility QTL identified in a descendent NIL population. We demonstrated that some interactions that produce TRD in the earlier generations are likely masked in the NIL population due to (unintentional) cointrogression of conspecific dance partners that relieve otherwise incompatible interactions. Data on patterns of TRD in markers found exclusively in multi-introgression NILs support this hypothesis. Once these putative dance partners are accounted for, we also detected evidence to support a hypothesis that complex epistasis (i.e., negative interactions among more than two genomic regions from the two parental species) underlies some of the TRD observed in BC populations, at least with respect to hybrid pollen sterility. Overall, our analysis indicates that combining data from TRD patterns with direct assessments of hybrid incompatibility can be valuable in clarifying the genetic basis of hybrid incompatibility (e.g., dominance vs. recessivity of factors), in evaluating alternative models of the genetic complexity of hybrid incompatibility, and in identifying new putative hybrid incompatibility QTL that would otherwise be missed in relatively genetically simple NIL populations.
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
We thank A. Monforte for kindly providing raw genotype data from the backcross populations as well as M. Turelli, M. Hahn, and J. Willis for useful suggestions on data analysis and interpretation. L.C.M. is grateful to the organizers of the Society for Molecular Biology and Evolution 2005 Young Investigator's Workshop for providing a stimulating forum in which to discuss this research. L.C.M. was supported in part by a Center for Population Biology Postdoctoral Research Fellowship from U. C. Davis and by National Science Foundation DEB grant #0532097.
|
Footnotes |
|---|
1 Present address: Department of Biology, Indiana University
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Bernacchi, D., and S. Tanksley. 1997. Linkage analysis and a QTL study of sexual compatibility factors and floral traits. Genetics 147:861–877.[Abstract]
Bernacchi, D., T. Beck-Bunn, Y. Eshed, J. Lopez, V. Petiard, J. Uhlig, D. Zamir, and S. Tanksley. 1998. Advanced backcross QTL analysis in tomato. I. Identification of QTLs for traits of agronomic importance from Lycopersicon hirsutum. Theor. Appl. Genet. 97:381–397.[CrossRef][ISI]
Fishman, L., and J. H. Willis. 2001. Evidence for Dobzhansky-Muller incompatibilites contributing to the sterility of hybrids between Mimulus guttatus and M. nasutus. Evolution 55:1932–1942.[CrossRef][ISI][Medline]
———. 2005. A novel meiotic drive locus almost completely distorts segregation in Mimulus (monkeyflower) hybrids. Genetics 169:347–353.
Hall, M. C., and J. H. Willis. 2005. Transmission ratio distortion in intraspecific hybrids of Mimulus guttatus: implications for genetic divergence. Genetics 170:375–386.
Harushima, Y., M. Nakagahra, M. Yano, T. Sasaki, and N. Kurata. 2001. A genome-wide survey of reproductive barriers in an intraspecific hybrid. Genetics 159:883–892.
Jenczewski, E., M. Gherardi, I. Bonnin, J. M. Prosperi, I. Olivieri and T. Huguet. 1997. Insight on segregation distortions in two intraspecific crosses between annual species of Medicago (Leguminosae). Theor. Appl. Genet. 94:682–691.[CrossRef][ISI]
Kocher, T. D., W. J. Lee, H. Sobolewska, D. Penman, and B. McAndrew. 1998. A genetic linkage map of a cichlid fish, the tilapia (Oreochromis niloticus). Genetics 148:1225–1232.
Li, L., and X. Guo. 2004. AFLP-based genetic linkage maps of the Pacific oyster Crassostrea gigas Thunberg. Mar. Biotechnol. 6:26–36.[CrossRef][Medline]
Li, Z. K., S. R. M. Pinson, A. H. Paterson, W. D. Park, and J. W. Stansel. 1997. Genetics of hybrid sterility and hybrid breakdown in an intersubspecific rice (Oryza sativa L) population. Genetics 145:1139–1148.[Abstract]
Monforte, A. J., and S. D. Tanksley. 2000. Development of a set of near isogenic and backcross recombinant inbred lines containing most of the Lycopersicon hirsutum genome in a L. esculentum genetic background: a tool for gene mapping and gene discovery. Genome 43:803–813.[Medline]
Montagutelli, X., R. Turner, and J. H. Nadeau. 1996. Epistatic control of non-Mendelian inheritance in mouse interspecific crosses. Genetics 143:1739–1752.[Abstract]
Moyle, L. C., and E. B. Graham. 2005. Genetics of hybrid incompatibility between Lycopersicon esculentum and L. hirsutum. Genetics 169:355–373.
Myburg, A. A., C. Vogl, A. R. Griffin, R. R. Sederoff, and R. W. Whetten. 2004. Genetics of postzygotic isolation in eucalyptus: whole-genome analysis of barriers to introgression in a wide interspecific cross of Eucalyptus grandis and E. globulus. Genetics 166:1405–1418.
Orr, H. A., and S. Irving. 2005. Segregation distortion in hybrids between the Bogota and USA subspecies of Drosophila pseudoobscura. Genetics 169:671–682.
Orr, H. A., and M. Turelli. 2001. The evolution of postzygotic isolation: accumulating Dobzhansky-Muller incompatibilities. Evolution 55:1085–1094.[CrossRef][ISI][Medline]
Pardo-Manuel de Villena, F., E. de La Casa-Esperon, T. L. Briscoe, and C. Sapienza. 2000. A genetic test to determine the origin of maternal transmission ratio distortion: meiotic drive at the mouse Om locus. Genetics 154:333–342.
Paterson, A. H. 2002. What has QTL mapping taught us about plant domestication? New Phytol. 154:591–608.[CrossRef]
Quiros, C. F. 1991. Lycopersicon cytogenetics Pp. 119–137 in P. K. Gupta and T. Tsuchiya, eds. Chromosome engineering in plants: genetics, breeding, evolution. Elsevier, Amsterdam.
Rieseberg, L. H., S. J. E. Baird, and K. A. Gardner. 2000. Hybridization, introgression, and linkage evolution. Plant Mol. Biol. 42:205–224.[CrossRef][ISI][Medline]
Solignac, M., D. Vautrin, E. Baudry, F. Mougel, A. Loiseau and J. M. Cornuet. 2004. A microsatellite-based linkage map of the honeybee, Apis mellifera. Genetics 167:253–262.
Tan, Y.-D., C. Wan, Y. Zhu, C. Lu, Z. Xiang H. W. Deng. 2001. An amplified fragment length polymorphism map of the silkworm. Genetics 157:1277–1284.
Turelli, M., and H. A. Orr. 2000. Dominance, epistasis and the genetics of postzygotic isolation. Genetics 154:1663–1679.
Wang, B., and A. H. Porter. 2004. An AFLP-based interspecific linkage map of sympatric, hybridizing Colias butterflies. Genetics 168:215–225.
Yin, T., X. Zhang, M. Huang, M. Wang, Q. Zhuge, S. Tu, L.-H. Zhu, and R. Wu. 2002. Molecular linkage maps of the Populus genome. Genome 45:541–555.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. C. Moyle and T. Nakazato Comparative Genetics of Hybrid Incompatibility: Sterility in Two Solanum Species Crosses Genetics, July 1, 2008; 179(3): 1437 - 1453. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Niehuis, A. K. Judson, and J. Gadau Cytonuclear Genic Incompatibilities Cause Increased Mortality in Male F2 Hybrids of Nasonia giraulti and N. vitripennis Genetics, January 1, 2008; 178(1): 413 - 426. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. C. Moyle Comparative Genetics of Potential Prezygotic and Postzygotic Isolating Barriers in a Lycopersicon Species Cross J. Hered., March 1, 2007; 98(2): 123 - 135. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||