Periodic Reporting for period 3 - GUPPYSEX (Evolutionary genetics of guppy sex chromosomes)
Periodo di rendicontazione: 2019-08-01 al 2021-01-31
The question addressed: the hypothesis that sexually antagonistic polymorphisms on the guppy sex chromosomes have selected for reduced genetic recombination between the two members of this chromosome pair, and whether such selection is causing ongoing changes in recombination.
The hypothesis to be tested is widely believed to be true, but empirical data have largely been lacking, as this is a difficult hypothesis to test. The project aims to contribute such data in a species that is often cited as supporting the hypothesis. In doing the work, it is contributing to development of approaches for detecting regions of sex linkage in genomes, generating the first detailed genetic map of the study species, and providing tests of high-throughput genotyping approaches for use in a non-model organism, with the potential to aid studies of other such organisms in which questions of biological interest have been beyond the reach of existing approaches.
Specific objectives, as in the original project.
1) Ascertain and sequence X- and Y-linked genes in guppies.
2) Establish a dense genetic map of the sex chromosome pair.
3) Estimate the age of the guppy sex chromosome system is (in Poecilia reticulata), using X-Y gene pairs.
4) Use population genetic data to distinguish between fully sex-linked genes and PAR genes closely linked to the fully sex-linked region, and to test whether there is ongoing evolution of the recombination rate between PAR genes with the MSY within P. reticulata.
5) Study the genetic control of male coloration phenotypes, including mapping genes to the sex chromosomes, including the pseudo-autosomal region (PAR), and to the autosomes, distinguishing fully Y-linked genes from PAR genes, and estimate allele frequencies in natural populations with low and high predation rates.
6) Ascertain X-linked genes and estimate the proportion that have retained Y-linked copies, to estimate the extent of genetic degeneration.
The hypothesis to be tested is widely believed to be true, but empirical data have largely been lacking, as this is a difficult hypothesis to test. The project aims to contribute such data in a species that is often cited as supporting the hypothesis. In doing the work, it is contributing to development of approaches for detecting regions of sex linkage in genomes, generating the first detailed genetic map of the study species, and providing tests of high-throughput genotyping approaches for use in a non-model organism, with the potential to aid studies of other such organisms in which questions of biological interest have been beyond the reach of existing approaches.
Specific objectives, as in the original project.
1) Ascertain and sequence X- and Y-linked genes in guppies.
2) Establish a dense genetic map of the sex chromosome pair.
3) Estimate the age of the guppy sex chromosome system is (in Poecilia reticulata), using X-Y gene pairs.
4) Use population genetic data to distinguish between fully sex-linked genes and PAR genes closely linked to the fully sex-linked region, and to test whether there is ongoing evolution of the recombination rate between PAR genes with the MSY within P. reticulata.
5) Study the genetic control of male coloration phenotypes, including mapping genes to the sex chromosomes, including the pseudo-autosomal region (PAR), and to the autosomes, distinguishing fully Y-linked genes from PAR genes, and estimate allele frequencies in natural populations with low and high predation rates.
6) Ascertain X-linked genes and estimate the proportion that have retained Y-linked copies, to estimate the extent of genetic degeneration.
Scientific Report
The report below outlines the experiments and analysis so far conducted, and their results, roughly in the order of the objectives of the proposal. As explained in detail below, we have made substantial progress on objectives 1, 2, 4 and 5.
Objective 1: identifying the sex-linked region of the Poecilia reticulata (guppy) genome, and ascertaining and sequencing X- and Y-linked genes.
Based on previous published work on the species, the fully sex-linked region was expected to be on chromosome 12 (the guppy has 23 acrocentric chromosome pairs). The Edinburgh group therefore first searched for male-specific sequence variants in a candidate region where a male-specific heterochromatin has been identified cytologically, near the tip of the P. reticulata assembly of chromosome 12. We used primers based on the complete genome sequence of a female individual from the Guanapo river, Caroni drainage, Trinidad (1). PCR amplification of sequences located in this region used a sample of 10 males and 6 females from a captive population derived from a high-predation population collected from the Aripo river (also in the Caroni drainage) and maintained in a large population by project COIs (Wilson, Croft) at the University of Exeter. High-predation populations are expected to have high genetic diversity (based on previously published studies on Trinidadian guppies), and should provide abundant markers for genetic mapping (objective 2 of the project). The sex chromosomes of such populations are thought to have less recombination than those found in up-river sites with lower predation, maximizing the chances of detecting sex linkage. The sexes of all the fish were ascertained at maturity.
Unexpectedly, these PCR experiments detected no variants with genotypes suggesting complete sex linkage in any chromosome 12 genes. This suggests that fully sex-linked variants may be restricted to a small genome region. Our second aim was to estimate a genetic map of the sex chromosome pair using densely spaced genetic markers. To obtain markers for this goal, as well as to test the hypothesis that the guppy fully sex-linked region might be very small, we obtained paired end high coverage genome sequences from the same set of males and females as described above. The sequencing and initial bioinformatic analysis, including mapping the reads to the published female guppy reference genome mentioned above, and generating VCF files with variants that passed stringent controls on quality and depth of coverage, were done by Edinburgh Genomics. The mean insert size of the sequences was 468 bp, and they at least 97% mapped to the reference genome in all 16 individuals. Coverage is high for most sites (not just on the basis of the genome-wide mean, which can obscure the presence of low-coverage sites); the lower 99th percentile values averaged 13, and exceeded 9 for all individuals. Overall, therefore, the aim of obtaining high coverage results was achieved, yielding sequence data for a sample of 22 X and 10 Y chromosomes.
(i) The guppy Y chromosome is not genetically degenerated (objective 6), and may have evolved recently: Almost all sites in chromosome 12 sequences had similar coverage in both sexes, similar to the genome-wide coverage value. Two alleles are therefore present in both sexes for most sequences on the XY pair, implying that sequences have not been lost in the fully Y-linked region in a process of genetic degeneration like that in old-established sex chromosomes.
Objective 3, estimating the age of the guppy fully sex-linked region using divergence between X- and Y-linked sequences, has not yet been achieved, because (as described below), we have not been able to identify any extensive fully sex-linked region.
Our sequencing yielded large numbers of single nucleotide polymorphisms (SNPs), providing excellent genetic markers, as well as other sequence variants (insertions and deletions). The SNPs were used for genetic mapping and also for population genomic analyses aimed at detecting fully sex-linked variants and identifying fully Y-linked regions. Before describing the results, we first outline the collection of samples from natural populations, which were used in both kinds of analyses relating to Objectives 1 and 2, and also for analyses relating to Objectives 4 and 5., including studies of phenotypic traits (male colour and pattern).
Collection of guppy samples from natural populations in Trinidad (for objectives 2, 3, 4 and 5).
In February 2017, the Exeter COIs and post-doctoral associate collected fish samples from the wild in Trinidad. Samples were obtained from 15 sites, including paired high and low predation sites within specific rivers as planned. Live fish from 12 of these sites were collected and imported to the UK (with permissions from all relevant authorities). Breeding colonies of these 12 populations have been successfully established at the University of Exeter, Falmouth campus.
Objective 2: Genetic mapping
Cytological observations in domesticated guppies (2) show that most crossovers in pachytene of male meiosis are localised near the tips of the acrocentric chromosomes, but about 5% of meioses had crossovers in a region of the sex bivalent proximal to the heterochromatic region mentioned above. No data were available wild guppies, though previous genetic mapping in an inter-population cross demonstrated more terminal crossover locations in male than female meiosis (3). We have advanced understanding of guppy meiosis by now estimating genetic maps for male and female meiosis separately, for markers with known genome assembly locations, using within-population crosses to minimise the likelihood of different chromosome arrangements, such as inversions (which could often be heterozygous in inter-population crosses, impeding crossing over).
The project COIs at Exeter provided families from natural populations (see above). Genetic maps have been estimated in Edinburgh from 4 families (Table 1). The markers used were mostly microsatellites that were genotyped using primers for sequences flanking short repeats ascertained from the genome sequences described above. They were frequently informative in male meiosis (which is the most relevant for defining chromosome 12 regions showing partial sex-linkage, and comparing these regions between different natural populations, our objectives 4 and 5). Many markers were also informative in female meiosis. We have established that there are high numbers of alleles in two natural populations so far studied (see below). These loci thus have sufficient variability that mapping families made using multiple males and females can be used, with the parents of each progeny individual being determined from their genotypes. Such pooled matings yield larger progeny numbers, compared with isolated male-female pairs, which should enable us to detect recombination events even if they are moderately rare. The genetic mapping yielded two valuable results, described next.
(i) Chromosome 12 is the sex chromosome
First, a family made from fish from the Aripo river captive population described above (LAH in Table 1) confirmed that, despite the negative results from our tests for male-specific sequences (see above), chromosome 12 (LG12) is the sex chromosome. In male meiosis, almost all markers showed complete sex linkage, and only three pseudo-autosomal (or PAR) markers, showing partial sex linkage, have so far been found (see below). Sex linkage of chromosome 12 markers was confirmed in three other families from natural populations (Table 1). Chromosome 12 was already known to be the sex chromosome in guppy populations from other river drainages in Trinidad and Venezuela. Despite the modest family sizes (Table 1), due to the fact that this fish is live-bearing, our results from different rivers and drainages strongly suggest that the sex chromosome pair is probably the same in all guppy populations.
Table 1. Families with genetic mapping results.
Family name Source population Drainage Predation level Number of progeny Chromosomes mapped PAR markers mapped
LAH Aripo Caroni High 42 LG12, 1, 9, 18 3
GHP3 Guanapo Caroni High 29 LG12 1
QHB4 Quare Oropouche High 35 LG12 2
QLPB1 Quare Oropouche Low 59 LG12 3
(ii) Recombination in male meiosis is restricted to chromosome tips
Our mapping also showed that, in male meiosis, crossing over is restricted to small regions at the chromosome tips, supporting the cytological result that crossovers in proximal regions are rare. In contrast, crossover events occur throughout the chromosomes in female meiosis, with a highly linear relationship between LG12 physical and genetic map positions, and an overall recombination rate estimate of about 1.85 cM/Mb (centiMorgans per megabase) of DNA. The sex difference (heterochiasmy) is not confined to chromosome 12: in the LAH family (Table 1), we obtained similar results for three other chromosomes, though we currently have a detailed assessment of where recombination events occur only for LG12.
For the guppy XY pair, LG12, no marker proximal to 24.5 Mb in the 26.5 Mb female assembly recombined with the sex-determining locus in males in any of the families, whereas all three markers distal to 26 Mb recombined frequently in all families with informative markers (Table 1). Overall, our results suggest an extremely high recombination rate in the small terminal region, of about 26 cM/Mb, similar to the values estimated for the human PAR1 (4). The high terminal recombination rate is not due to an obligatory crossover event in males at the boundary of the PAR with the fully Y-linked region, since the terminal region markers differ significant in their recombination distances from the sex-determining locus.
Our genetic mapping information has assisted the genome assembly being done by the guppy genome sequencing group (at the Max Planck Institutes for Developmental Biology and Evolutionary Biology, respectively in Tübingen and Plön, Germany, and the University of Sussex, UK). We revealed that one sequence assembled on chromosome 9 shows complete sex linkage in male meiosis, and co-segregates with an X-linked marker in female meiosis. We also mapped candidate sex linked sequences that were unplaced in the published assembly, confirming their sex linkage. Finally, we confirmed a suspected assembly error in the middle of the X chromosome that was detected in assembling the genome of a male from the Guanapo river. The new order of sequences (inverted across a region of 8 Mb) corresponds with the genetic map order of the X chromosomes of families from the Aripo and Quare rivers, demonstrating that this is not an inversion polymorphism involving a difference between the X in the Guanapo female and male.
(iii) Cross-chromosome comparison of recombination rates in male and female meiosis
A major part of Objective 4 of our project was to test whether there is ongoing evolution of the recombination rate between PAR genes with the fully Y-linked region (also called the “male-specific Y-linked region”, or MSY) within P. reticulata. Two genetic mapping approaches are potentially informative, comparing the genetic maps of the XY pair in meiosis of males from different populations, and comparisons between the sexes, using genetic maps of the XY pair and autosomes; if recombination rates of the former are unexpectedly low in male meiosis, this would suggest a specific reduction affecting the sex chromosome pair. We implemented both these approaches, and, as described above, the first approach does not suggest any difference between the populations.
(iv) Comparing cytological and genetic mapping results for guppy male meiosis, and making new cytological observations
The previous cytological work estimated that 95% of crossovers in males occur mostly in the terminal 15% of chromosome 12, and the other 5% of bivalents proximal to a small non-recombining region; crossovers are therefore expected in a region between 45 and 50% of the physical distance from the chromosome tip (or perhaps a smaller distance if lengths of synaptonemal complex over-estimate the physical distances). We compared these predictions with our genetic mapping results. Our genetic map results agree with these conclusions, but our mapping involving 165 progeny of males recently or immediately derived from wild natural populations (Table 1) has not detected more proximal recombination, though planned further families will increase this number. With a recombination rate of 5%, we should have observed about 8 crossovers in our current families.
To further test for proximal recombination, we organised a visit from Dr. Lisachov from his home institution in Sovosibirsk to the Falmouth campus of the University of Exeter, to perform experiments and train Dr. Lenny Yong, the post-doctoral researcher in Falmouth, to analyse MLH1 foci in wild guppies. They confirmed that most crossovers were in terminal regions of the chromosomes. Dr. Yong will continue this work in order to obtain larger numbers of cells and test whether more proximal foci also occur.
Objective 4: Population genomic analyses
(i) Confirmation that chromosome 12 is the sex chromosome
To test whether there is ongoing evolution of the recombination rate between genes showing partial sex-linkage (PAR, or pseudo-autosomal genes) and the MSY within P. reticulata, and distinguish between chromosome 12 genes showing complete and partial sex-linkage (PAR, or pseudo-autosomal genes), genetic map estimates are too coarse. We therefore also analysed our genome sequences obtained for Objective 2, and used the quantity FST to test for allele frequency differences between the sexes. With our sample size of males and females, the FST value expected for fully sex-linked sites is 0.27. In a recently evolved fully sex-linked region, sites will start to accumulate Y-linked substitutions, so that the mean value for all sites in the region will generally be lower than this, but genome-scale data can nevertheless potentially detect regions with concentrations of values higher than elsewhere in a genome.
We found genotype patterns indicating sex linkage largely on chromosome 12 (and very rarely on any of the other 22 guppy chromosomes). As this work was being planned, we learned that another group of biologists (Judith Mank’s group in London) were also doing population genomic analysis on guppy sex chromosomes, and they published results at the end of January 2017 (5). However, only two individuals of each sex were used to detect the sex-linked region, and the analyses used an unreliable quantity, SNP density (which depends strongly on gene density, as well as selective constraints on differences between Y- and X-linked sequences). Our analysis provides stronger evidence that LG12 is the sex chromosome pair. Wright et al. suggested that a 3 Mb region of LG12 forms a fully non-recombining “old stratum”, but this was not detected in our analysis. The average FST for LG12 sequences is only about 0.05 and the chromosome appears to have only a small fully sex-linked region, and mostly to be partially sex-linked. In total, only 14 SNPs had the genotype configuration expected for complete sex linkage. Given our sample of 10 males and 6 females, no individual SNP shows statistically significant evidence for complete sex-linkage, but the concentration of 14 such sites on chromosome 12, out of a total of 35 sites across all 23 chromosomes, is highly significant. The region with these 14 sites is not the region identified by Wright et al.
(ii) Evidence that
chromosome 12 recombines in males
Many LG12 SNPs show associations with the sex-determining locus, and FIS values indicate excess heterozygote frequencies in males. However, the genotypes of two males and one female suggest that recombination can occur between the X and the Y. Even excluding these three individuals, the FST values remain low across most of chromosome 12. We also directly detected recombination in our sequence data, by two tests. First, at biallelic sites, both homozygotes were often present, indicating either reciprocal recombination or gene conversion. Second, although our data are un-phased (as we know only that sequences from females are X-linked, but cannot directly assign variants in sequences from males to the X versus Y haplotypes carried by those males), we analysed the sub-set of sites at which no variants are present in the females. For such sites, the X-linked state is known, and potentially Y-linked variants can be inferred, and four-gamete tests can be applied; this test also detected recombination. Overall, therefore, we conclude that some recombination must occur, and that genotypes of many sites throughout most of chromosome 12 show partial, not complete, sex linkage. Nevertheless, because XY recombination is rare, due to crossing over being restricted to the chromosome tips in male meiosis, SA polymorphisms will be expected to accumulate predominantly in genome regions that are partially Y-linked, but recombine very rarely with the sex-determining region.
A main goal of our project was to test for associations of SNPs with sex in regions other than the sex-determining locus, as these can arise if a locus with a sexually antagonistic (SA) polymorphism is present, such as the male coloration factors known in guppy populations. Two regions on chromosome 12 indeed showed particularly pronounced genetic differentiation between sequences of females and males (unusually high FST and high frequencies of heterozygotes specifically in males). We are doing further analyses to test whether SA polymorphism can explain our results.
(iii) Natural population data
The results just described are from a captive population, and a major part of the work in the coming months will be directed towards getting data from natural populations. As explained above, the samples are already in hand. However, genotyping to test whether SNPs in these populations also show associations with the sex-determining locus remains to be done. We have already genotyped microsatellite markers from samples from high and low predation sites in the Aripo river, and these do not suggest any such associations (though they do support the belief, based on limited previous studies, that low-predation populations in up-river sites have undergone severe bottlenecks).
Objective 5: Studying the genetic control of male coloration phenotypes
(i) Description of male coloration phenotypes
The project plans included mapping male coloration factors to the sex chromosomes, including the pseudo-autosomal region (PAR), and to the autosomes, distinguishing fully Y-linked genes from PAR genes, and estimating allele frequencies in natural populations with low and high predation rates.
Captive bred individuals from the natural population parents (housed under common environmental conditions) have been phenotyped as planned, and male coloration phenotypes quantified using a customized software developed by sensory ecologists at Exeter. The individuals studied include samples from the high and low predation sites described above, which will allow us to test whether males from the latter are significantly more colourful than from the former. Several families have been generated by crosses between individuals, and, as described above, have been used in genetic mapping with molecular markers. These families will be used to test which traits show complete Y-linkage, and which do not. Y-linkage is likely for some traits in our material (as in the classical papers on such traits in guppies), since the parental males differ, and progeny with complete sets of several traits resembling the parental males are seen. Genotyping with our molecular markers will allow us to classify the linkage of the coloration factors into the categories outlined above.
References
1. Künstner A, et al. (2017) The genome of the Trinidadian guppy, Poecilia reticulata, and variation in the Guanapo population. PLOS ONE 11(12):e0169087.
2. Lisachov A, Zadesenets K, Rubtsov N, & Borodin P (2015) Sex chromosome synapsis and recombination in male guppies. Zebrafish 12(2):174-180.
3. Tripathi N, Hoffmann M, Weigel D, & Dreyer C (2009) Linkage analysis reveals the independent origin of Poeciliid sex chromosomes and a case of atypical sex inheritance in the guppy (Poecilia reticulata). Genetics 182:365–374.
4. Flaquer A, Fischer C, & Wienker T (2009) A new sex-specific genetic map of the human pseudoautosomal regions (PAR1 and PAR2). Hum. Hered. 68(3):192-200.
5. Wright A, et al. (2017) Convergent recombination suppression suggests a role of sexual conflict in guppy sex chromosome formation Nature Communications 8:14251.
The report below outlines the experiments and analysis so far conducted, and their results, roughly in the order of the objectives of the proposal. As explained in detail below, we have made substantial progress on objectives 1, 2, 4 and 5.
Objective 1: identifying the sex-linked region of the Poecilia reticulata (guppy) genome, and ascertaining and sequencing X- and Y-linked genes.
Based on previous published work on the species, the fully sex-linked region was expected to be on chromosome 12 (the guppy has 23 acrocentric chromosome pairs). The Edinburgh group therefore first searched for male-specific sequence variants in a candidate region where a male-specific heterochromatin has been identified cytologically, near the tip of the P. reticulata assembly of chromosome 12. We used primers based on the complete genome sequence of a female individual from the Guanapo river, Caroni drainage, Trinidad (1). PCR amplification of sequences located in this region used a sample of 10 males and 6 females from a captive population derived from a high-predation population collected from the Aripo river (also in the Caroni drainage) and maintained in a large population by project COIs (Wilson, Croft) at the University of Exeter. High-predation populations are expected to have high genetic diversity (based on previously published studies on Trinidadian guppies), and should provide abundant markers for genetic mapping (objective 2 of the project). The sex chromosomes of such populations are thought to have less recombination than those found in up-river sites with lower predation, maximizing the chances of detecting sex linkage. The sexes of all the fish were ascertained at maturity.
Unexpectedly, these PCR experiments detected no variants with genotypes suggesting complete sex linkage in any chromosome 12 genes. This suggests that fully sex-linked variants may be restricted to a small genome region. Our second aim was to estimate a genetic map of the sex chromosome pair using densely spaced genetic markers. To obtain markers for this goal, as well as to test the hypothesis that the guppy fully sex-linked region might be very small, we obtained paired end high coverage genome sequences from the same set of males and females as described above. The sequencing and initial bioinformatic analysis, including mapping the reads to the published female guppy reference genome mentioned above, and generating VCF files with variants that passed stringent controls on quality and depth of coverage, were done by Edinburgh Genomics. The mean insert size of the sequences was 468 bp, and they at least 97% mapped to the reference genome in all 16 individuals. Coverage is high for most sites (not just on the basis of the genome-wide mean, which can obscure the presence of low-coverage sites); the lower 99th percentile values averaged 13, and exceeded 9 for all individuals. Overall, therefore, the aim of obtaining high coverage results was achieved, yielding sequence data for a sample of 22 X and 10 Y chromosomes.
(i) The guppy Y chromosome is not genetically degenerated (objective 6), and may have evolved recently: Almost all sites in chromosome 12 sequences had similar coverage in both sexes, similar to the genome-wide coverage value. Two alleles are therefore present in both sexes for most sequences on the XY pair, implying that sequences have not been lost in the fully Y-linked region in a process of genetic degeneration like that in old-established sex chromosomes.
Objective 3, estimating the age of the guppy fully sex-linked region using divergence between X- and Y-linked sequences, has not yet been achieved, because (as described below), we have not been able to identify any extensive fully sex-linked region.
Our sequencing yielded large numbers of single nucleotide polymorphisms (SNPs), providing excellent genetic markers, as well as other sequence variants (insertions and deletions). The SNPs were used for genetic mapping and also for population genomic analyses aimed at detecting fully sex-linked variants and identifying fully Y-linked regions. Before describing the results, we first outline the collection of samples from natural populations, which were used in both kinds of analyses relating to Objectives 1 and 2, and also for analyses relating to Objectives 4 and 5., including studies of phenotypic traits (male colour and pattern).
Collection of guppy samples from natural populations in Trinidad (for objectives 2, 3, 4 and 5).
In February 2017, the Exeter COIs and post-doctoral associate collected fish samples from the wild in Trinidad. Samples were obtained from 15 sites, including paired high and low predation sites within specific rivers as planned. Live fish from 12 of these sites were collected and imported to the UK (with permissions from all relevant authorities). Breeding colonies of these 12 populations have been successfully established at the University of Exeter, Falmouth campus.
Objective 2: Genetic mapping
Cytological observations in domesticated guppies (2) show that most crossovers in pachytene of male meiosis are localised near the tips of the acrocentric chromosomes, but about 5% of meioses had crossovers in a region of the sex bivalent proximal to the heterochromatic region mentioned above. No data were available wild guppies, though previous genetic mapping in an inter-population cross demonstrated more terminal crossover locations in male than female meiosis (3). We have advanced understanding of guppy meiosis by now estimating genetic maps for male and female meiosis separately, for markers with known genome assembly locations, using within-population crosses to minimise the likelihood of different chromosome arrangements, such as inversions (which could often be heterozygous in inter-population crosses, impeding crossing over).
The project COIs at Exeter provided families from natural populations (see above). Genetic maps have been estimated in Edinburgh from 4 families (Table 1). The markers used were mostly microsatellites that were genotyped using primers for sequences flanking short repeats ascertained from the genome sequences described above. They were frequently informative in male meiosis (which is the most relevant for defining chromosome 12 regions showing partial sex-linkage, and comparing these regions between different natural populations, our objectives 4 and 5). Many markers were also informative in female meiosis. We have established that there are high numbers of alleles in two natural populations so far studied (see below). These loci thus have sufficient variability that mapping families made using multiple males and females can be used, with the parents of each progeny individual being determined from their genotypes. Such pooled matings yield larger progeny numbers, compared with isolated male-female pairs, which should enable us to detect recombination events even if they are moderately rare. The genetic mapping yielded two valuable results, described next.
(i) Chromosome 12 is the sex chromosome
First, a family made from fish from the Aripo river captive population described above (LAH in Table 1) confirmed that, despite the negative results from our tests for male-specific sequences (see above), chromosome 12 (LG12) is the sex chromosome. In male meiosis, almost all markers showed complete sex linkage, and only three pseudo-autosomal (or PAR) markers, showing partial sex linkage, have so far been found (see below). Sex linkage of chromosome 12 markers was confirmed in three other families from natural populations (Table 1). Chromosome 12 was already known to be the sex chromosome in guppy populations from other river drainages in Trinidad and Venezuela. Despite the modest family sizes (Table 1), due to the fact that this fish is live-bearing, our results from different rivers and drainages strongly suggest that the sex chromosome pair is probably the same in all guppy populations.
Table 1. Families with genetic mapping results.
Family name Source population Drainage Predation level Number of progeny Chromosomes mapped PAR markers mapped
LAH Aripo Caroni High 42 LG12, 1, 9, 18 3
GHP3 Guanapo Caroni High 29 LG12 1
QHB4 Quare Oropouche High 35 LG12 2
QLPB1 Quare Oropouche Low 59 LG12 3
(ii) Recombination in male meiosis is restricted to chromosome tips
Our mapping also showed that, in male meiosis, crossing over is restricted to small regions at the chromosome tips, supporting the cytological result that crossovers in proximal regions are rare. In contrast, crossover events occur throughout the chromosomes in female meiosis, with a highly linear relationship between LG12 physical and genetic map positions, and an overall recombination rate estimate of about 1.85 cM/Mb (centiMorgans per megabase) of DNA. The sex difference (heterochiasmy) is not confined to chromosome 12: in the LAH family (Table 1), we obtained similar results for three other chromosomes, though we currently have a detailed assessment of where recombination events occur only for LG12.
For the guppy XY pair, LG12, no marker proximal to 24.5 Mb in the 26.5 Mb female assembly recombined with the sex-determining locus in males in any of the families, whereas all three markers distal to 26 Mb recombined frequently in all families with informative markers (Table 1). Overall, our results suggest an extremely high recombination rate in the small terminal region, of about 26 cM/Mb, similar to the values estimated for the human PAR1 (4). The high terminal recombination rate is not due to an obligatory crossover event in males at the boundary of the PAR with the fully Y-linked region, since the terminal region markers differ significant in their recombination distances from the sex-determining locus.
Our genetic mapping information has assisted the genome assembly being done by the guppy genome sequencing group (at the Max Planck Institutes for Developmental Biology and Evolutionary Biology, respectively in Tübingen and Plön, Germany, and the University of Sussex, UK). We revealed that one sequence assembled on chromosome 9 shows complete sex linkage in male meiosis, and co-segregates with an X-linked marker in female meiosis. We also mapped candidate sex linked sequences that were unplaced in the published assembly, confirming their sex linkage. Finally, we confirmed a suspected assembly error in the middle of the X chromosome that was detected in assembling the genome of a male from the Guanapo river. The new order of sequences (inverted across a region of 8 Mb) corresponds with the genetic map order of the X chromosomes of families from the Aripo and Quare rivers, demonstrating that this is not an inversion polymorphism involving a difference between the X in the Guanapo female and male.
(iii) Cross-chromosome comparison of recombination rates in male and female meiosis
A major part of Objective 4 of our project was to test whether there is ongoing evolution of the recombination rate between PAR genes with the fully Y-linked region (also called the “male-specific Y-linked region”, or MSY) within P. reticulata. Two genetic mapping approaches are potentially informative, comparing the genetic maps of the XY pair in meiosis of males from different populations, and comparisons between the sexes, using genetic maps of the XY pair and autosomes; if recombination rates of the former are unexpectedly low in male meiosis, this would suggest a specific reduction affecting the sex chromosome pair. We implemented both these approaches, and, as described above, the first approach does not suggest any difference between the populations.
(iv) Comparing cytological and genetic mapping results for guppy male meiosis, and making new cytological observations
The previous cytological work estimated that 95% of crossovers in males occur mostly in the terminal 15% of chromosome 12, and the other 5% of bivalents proximal to a small non-recombining region; crossovers are therefore expected in a region between 45 and 50% of the physical distance from the chromosome tip (or perhaps a smaller distance if lengths of synaptonemal complex over-estimate the physical distances). We compared these predictions with our genetic mapping results. Our genetic map results agree with these conclusions, but our mapping involving 165 progeny of males recently or immediately derived from wild natural populations (Table 1) has not detected more proximal recombination, though planned further families will increase this number. With a recombination rate of 5%, we should have observed about 8 crossovers in our current families.
To further test for proximal recombination, we organised a visit from Dr. Lisachov from his home institution in Sovosibirsk to the Falmouth campus of the University of Exeter, to perform experiments and train Dr. Lenny Yong, the post-doctoral researcher in Falmouth, to analyse MLH1 foci in wild guppies. They confirmed that most crossovers were in terminal regions of the chromosomes. Dr. Yong will continue this work in order to obtain larger numbers of cells and test whether more proximal foci also occur.
Objective 4: Population genomic analyses
(i) Confirmation that chromosome 12 is the sex chromosome
To test whether there is ongoing evolution of the recombination rate between genes showing partial sex-linkage (PAR, or pseudo-autosomal genes) and the MSY within P. reticulata, and distinguish between chromosome 12 genes showing complete and partial sex-linkage (PAR, or pseudo-autosomal genes), genetic map estimates are too coarse. We therefore also analysed our genome sequences obtained for Objective 2, and used the quantity FST to test for allele frequency differences between the sexes. With our sample size of males and females, the FST value expected for fully sex-linked sites is 0.27. In a recently evolved fully sex-linked region, sites will start to accumulate Y-linked substitutions, so that the mean value for all sites in the region will generally be lower than this, but genome-scale data can nevertheless potentially detect regions with concentrations of values higher than elsewhere in a genome.
We found genotype patterns indicating sex linkage largely on chromosome 12 (and very rarely on any of the other 22 guppy chromosomes). As this work was being planned, we learned that another group of biologists (Judith Mank’s group in London) were also doing population genomic analysis on guppy sex chromosomes, and they published results at the end of January 2017 (5). However, only two individuals of each sex were used to detect the sex-linked region, and the analyses used an unreliable quantity, SNP density (which depends strongly on gene density, as well as selective constraints on differences between Y- and X-linked sequences). Our analysis provides stronger evidence that LG12 is the sex chromosome pair. Wright et al. suggested that a 3 Mb region of LG12 forms a fully non-recombining “old stratum”, but this was not detected in our analysis. The average FST for LG12 sequences is only about 0.05 and the chromosome appears to have only a small fully sex-linked region, and mostly to be partially sex-linked. In total, only 14 SNPs had the genotype configuration expected for complete sex linkage. Given our sample of 10 males and 6 females, no individual SNP shows statistically significant evidence for complete sex-linkage, but the concentration of 14 such sites on chromosome 12, out of a total of 35 sites across all 23 chromosomes, is highly significant. The region with these 14 sites is not the region identified by Wright et al.
(ii) Evidence that
chromosome 12 recombines in males
Many LG12 SNPs show associations with the sex-determining locus, and FIS values indicate excess heterozygote frequencies in males. However, the genotypes of two males and one female suggest that recombination can occur between the X and the Y. Even excluding these three individuals, the FST values remain low across most of chromosome 12. We also directly detected recombination in our sequence data, by two tests. First, at biallelic sites, both homozygotes were often present, indicating either reciprocal recombination or gene conversion. Second, although our data are un-phased (as we know only that sequences from females are X-linked, but cannot directly assign variants in sequences from males to the X versus Y haplotypes carried by those males), we analysed the sub-set of sites at which no variants are present in the females. For such sites, the X-linked state is known, and potentially Y-linked variants can be inferred, and four-gamete tests can be applied; this test also detected recombination. Overall, therefore, we conclude that some recombination must occur, and that genotypes of many sites throughout most of chromosome 12 show partial, not complete, sex linkage. Nevertheless, because XY recombination is rare, due to crossing over being restricted to the chromosome tips in male meiosis, SA polymorphisms will be expected to accumulate predominantly in genome regions that are partially Y-linked, but recombine very rarely with the sex-determining region.
A main goal of our project was to test for associations of SNPs with sex in regions other than the sex-determining locus, as these can arise if a locus with a sexually antagonistic (SA) polymorphism is present, such as the male coloration factors known in guppy populations. Two regions on chromosome 12 indeed showed particularly pronounced genetic differentiation between sequences of females and males (unusually high FST and high frequencies of heterozygotes specifically in males). We are doing further analyses to test whether SA polymorphism can explain our results.
(iii) Natural population data
The results just described are from a captive population, and a major part of the work in the coming months will be directed towards getting data from natural populations. As explained above, the samples are already in hand. However, genotyping to test whether SNPs in these populations also show associations with the sex-determining locus remains to be done. We have already genotyped microsatellite markers from samples from high and low predation sites in the Aripo river, and these do not suggest any such associations (though they do support the belief, based on limited previous studies, that low-predation populations in up-river sites have undergone severe bottlenecks).
Objective 5: Studying the genetic control of male coloration phenotypes
(i) Description of male coloration phenotypes
The project plans included mapping male coloration factors to the sex chromosomes, including the pseudo-autosomal region (PAR), and to the autosomes, distinguishing fully Y-linked genes from PAR genes, and estimating allele frequencies in natural populations with low and high predation rates.
Captive bred individuals from the natural population parents (housed under common environmental conditions) have been phenotyped as planned, and male coloration phenotypes quantified using a customized software developed by sensory ecologists at Exeter. The individuals studied include samples from the high and low predation sites described above, which will allow us to test whether males from the latter are significantly more colourful than from the former. Several families have been generated by crosses between individuals, and, as described above, have been used in genetic mapping with molecular markers. These families will be used to test which traits show complete Y-linkage, and which do not. Y-linkage is likely for some traits in our material (as in the classical papers on such traits in guppies), since the parental males differ, and progeny with complete sets of several traits resembling the parental males are seen. Genotyping with our molecular markers will allow us to classify the linkage of the coloration factors into the categories outlined above.
References
1. Künstner A, et al. (2017) The genome of the Trinidadian guppy, Poecilia reticulata, and variation in the Guanapo population. PLOS ONE 11(12):e0169087.
2. Lisachov A, Zadesenets K, Rubtsov N, & Borodin P (2015) Sex chromosome synapsis and recombination in male guppies. Zebrafish 12(2):174-180.
3. Tripathi N, Hoffmann M, Weigel D, & Dreyer C (2009) Linkage analysis reveals the independent origin of Poeciliid sex chromosomes and a case of atypical sex inheritance in the guppy (Poecilia reticulata). Genetics 182:365–374.
4. Flaquer A, Fischer C, & Wienker T (2009) A new sex-specific genetic map of the human pseudoautosomal regions (PAR1 and PAR2). Hum. Hered. 68(3):192-200.
5. Wright A, et al. (2017) Convergent recombination suppression suggests a role of sexual conflict in guppy sex chromosome formation Nature Communications 8:14251.
Objective 2: Genetic mapping
Currently, we have genetic mapping results from only one low-predation population, because these fish mature slowly, but further families have been generated and the fish have either reached maturity, or will soon do so. Families from several more populations, including both high and low predation sites, are in hand and will be used in further genetic mapping. These will fulfil the objective of comparing genetic maps in male meiosis between populations with different predation levels (to test for the expected greater recombination rate in low- than high-predation populations).
With the current total number of progeny so far genotyped (165), the recombination rate estimate based on observing no recombinants in regions proximal to 24.5 Mb in the pooled progeny from the male parents of all families is 0.4 cM. In the next months, we will increase the progeny numbers, to better estimate this rate. The current upper 99% confidence interval is 2.8 cM, not inconsistent with the cytological estimate of 5% recombinants. It is therefore possible that some recombination occurs in parts of chromosome 12 some distance from its tip, in other words that the guppy XY pair has two PARs, one at the tip, and another one more proximal that rarely undergoes crossover events. A crossover rate of 5% in such a PAR2 could potentially explain the published results for male coloration factors, which suggest recombination rates of at most 10% with the male-determining gene. If so, this suggests that some of these factors are in genes in proximal locations.
We will also genetically map several chromosomes other than 12 in further families, to obtain detailed assessments of where recombination events occur on these chromosomes, and to add markers physically close to both ends.
We will also define the PAR in more detail. The marker at 25.3 Mb will be mapped in all families in which the male parents have informative genotypes (if not, other markers in this region will be developed and mapped). Further markers between this and 26 Mb will also be mapped. Our current results are based on comparing the overall total map lengths in the LAH family, which had the largest number of mapped markers, including markers located near each end of the chromosome 12 physical assembly. For this chromosome, the value in male meiosis was 65% of the value in female meiosis. Two autosomes (9 and 18) yielded data for female meiosis, but we have not yet succeeded in mapping terminal markers in the female parents of two of them, while no terminal marker has yet been mapped in male meiosis for chromosome 1. If we assume that the female map lengths are 50 cM, the male values for chromosomes 9 and 18 (Table 2) would be 80 and 84%, respectively, of the female lengths. We will map more markers to test further whether there is indeed a significantly larger reduction in length in males than female meiosis.
Objective 4: Population genomic analyses
In addition to the further genetic mapping outlined above, we plan to supplement these analyses by phasing the variants in sequences from males, to assign them to the X or Y haplotypes. We will test several different software packages that are currently available. Specifically, we plan to test approaches that take account of SNPs within individual reads (called “read-aware” approaches), rather than methods, such as Beagle, designed for human genome sequences. The human genome has very low diversity, so that reads rarely include more than a single SNP. Our sequences, however, indicate high diversity (even in the captive population so far studied), with SNP densities on all chromosomes of around 20 per kilobase, and somewhat higher on chromosome 12 (as expected if this chromosome includes some male-specific or male-associated variants). Initially, we plan to test two read-aware” approaches, implemented in the Hap-Cut and SHAPEIT2 programs, which can run such analyses in relatively short times, and have low phasing error rates.
Testing whether our empirical results can be explained in terms of the expected signals of SA polymorphisms will require further work, to specifically model neutral variants in an evolving sex chromosome. In the next months, we plan to initiate such modelling using a multi-locus simulator, SLiM2.
Objective 5: Studying the genetic control of male coloration phenotypes
In the next phase of the project, we will complete the genotyping of progeny males for our microsatellite markers, and test whether Y-linked haplotypes can be inferred that carry specific sets of marker alleles and phenotypic trait factors. The generation of F1 and F2 offspring are ongoing on the Exeter campus, and will be used for further genetic mapping (QTL). In addition to the families made by crossing fish from the same population described above, several between-population families have been generated and are growing. The parents have been phenotyped, and progeny will be phenotyped in the next few months to look at the segregation of male phenotypes in the offspring. Progress has also been made with quantifying male coloration phenotypes using a customized software developed by sensory ecologists at Exeter.
We will also examine the progeny males for traits that do not co-segregate with the haplotypes based on markers showing complete sex-linkage, and distinguish between autosomal traits and partially sex-linked ones. This will be possible given that we have markers at different map distances from the boundary with the chromosome 12 region showing apparently complete sex-linkage. However, further genetic markers will need to be developed in the region close to the boundary, as the marker currently mapping closest to the boundary has a recombination rate of 15% or more in our families. Marker development will be done once the new assembly is complete, to ensure that we target the region to be examined. We anticipate that this will be done in autumn 2018 and the mapping in the following winter.
Currently, we have genetic mapping results from only one low-predation population, because these fish mature slowly, but further families have been generated and the fish have either reached maturity, or will soon do so. Families from several more populations, including both high and low predation sites, are in hand and will be used in further genetic mapping. These will fulfil the objective of comparing genetic maps in male meiosis between populations with different predation levels (to test for the expected greater recombination rate in low- than high-predation populations).
With the current total number of progeny so far genotyped (165), the recombination rate estimate based on observing no recombinants in regions proximal to 24.5 Mb in the pooled progeny from the male parents of all families is 0.4 cM. In the next months, we will increase the progeny numbers, to better estimate this rate. The current upper 99% confidence interval is 2.8 cM, not inconsistent with the cytological estimate of 5% recombinants. It is therefore possible that some recombination occurs in parts of chromosome 12 some distance from its tip, in other words that the guppy XY pair has two PARs, one at the tip, and another one more proximal that rarely undergoes crossover events. A crossover rate of 5% in such a PAR2 could potentially explain the published results for male coloration factors, which suggest recombination rates of at most 10% with the male-determining gene. If so, this suggests that some of these factors are in genes in proximal locations.
We will also genetically map several chromosomes other than 12 in further families, to obtain detailed assessments of where recombination events occur on these chromosomes, and to add markers physically close to both ends.
We will also define the PAR in more detail. The marker at 25.3 Mb will be mapped in all families in which the male parents have informative genotypes (if not, other markers in this region will be developed and mapped). Further markers between this and 26 Mb will also be mapped. Our current results are based on comparing the overall total map lengths in the LAH family, which had the largest number of mapped markers, including markers located near each end of the chromosome 12 physical assembly. For this chromosome, the value in male meiosis was 65% of the value in female meiosis. Two autosomes (9 and 18) yielded data for female meiosis, but we have not yet succeeded in mapping terminal markers in the female parents of two of them, while no terminal marker has yet been mapped in male meiosis for chromosome 1. If we assume that the female map lengths are 50 cM, the male values for chromosomes 9 and 18 (Table 2) would be 80 and 84%, respectively, of the female lengths. We will map more markers to test further whether there is indeed a significantly larger reduction in length in males than female meiosis.
Objective 4: Population genomic analyses
In addition to the further genetic mapping outlined above, we plan to supplement these analyses by phasing the variants in sequences from males, to assign them to the X or Y haplotypes. We will test several different software packages that are currently available. Specifically, we plan to test approaches that take account of SNPs within individual reads (called “read-aware” approaches), rather than methods, such as Beagle, designed for human genome sequences. The human genome has very low diversity, so that reads rarely include more than a single SNP. Our sequences, however, indicate high diversity (even in the captive population so far studied), with SNP densities on all chromosomes of around 20 per kilobase, and somewhat higher on chromosome 12 (as expected if this chromosome includes some male-specific or male-associated variants). Initially, we plan to test two read-aware” approaches, implemented in the Hap-Cut and SHAPEIT2 programs, which can run such analyses in relatively short times, and have low phasing error rates.
Testing whether our empirical results can be explained in terms of the expected signals of SA polymorphisms will require further work, to specifically model neutral variants in an evolving sex chromosome. In the next months, we plan to initiate such modelling using a multi-locus simulator, SLiM2.
Objective 5: Studying the genetic control of male coloration phenotypes
In the next phase of the project, we will complete the genotyping of progeny males for our microsatellite markers, and test whether Y-linked haplotypes can be inferred that carry specific sets of marker alleles and phenotypic trait factors. The generation of F1 and F2 offspring are ongoing on the Exeter campus, and will be used for further genetic mapping (QTL). In addition to the families made by crossing fish from the same population described above, several between-population families have been generated and are growing. The parents have been phenotyped, and progeny will be phenotyped in the next few months to look at the segregation of male phenotypes in the offspring. Progress has also been made with quantifying male coloration phenotypes using a customized software developed by sensory ecologists at Exeter.
We will also examine the progeny males for traits that do not co-segregate with the haplotypes based on markers showing complete sex-linkage, and distinguish between autosomal traits and partially sex-linked ones. This will be possible given that we have markers at different map distances from the boundary with the chromosome 12 region showing apparently complete sex-linkage. However, further genetic markers will need to be developed in the region close to the boundary, as the marker currently mapping closest to the boundary has a recombination rate of 15% or more in our families. Marker development will be done once the new assembly is complete, to ensure that we target the region to be examined. We anticipate that this will be done in autumn 2018 and the mapping in the following winter.