I am an evolutionary biologist interested in sex chromosome evolution, sexual selection and speciation.
Caenorhabditis
Please check the work of students
Salt conditioning in hermaphrodite and gonochoristic species
Hybrid incompatibility in between-species crosses
Rana temporaria
Within population variation in Y chromosome differentiation level
No evidence that Y-chromosome differentiation affects male fitness in a Swiss population of common frogs
Male frogs in the same population with Y chromosomes of varying differentiation level from the X have no significant difference in morphometrics, amplexus success and fathering success.
The standard theory of sex chromosome evolution predicts that they will accumulate mutations that are beneficial for one sex and deleterious to the other. Such genes are termed sexually antagonistic.
The reason sex chromosomes attract sexually antagonistic genes is that they spend the majority of their time in one sex. For example, the Y is only found in males, and the X spends twice the time in females (XX) compared to males (XY) in mammals. By localising on sex chromosomes, sexually antagonistic genes can therefore maximise the positive effects to the sex in which they occur most often and minimise the negative effects the same alleles have in the opposite sex.
The common frog presents a rare system to test this idea, because populations can be found in which Y chromosomes of varying differentiation to the X exist. We worked on a population with 3 major Y-haplotypes: one which is fully differentiated across its length, one which is only differentiated close to a candidate sex-determining gene, and one which is indistinguishable from the X (XX males). These haplotypes represent different portions of the genome that spend their time in males; the absence of differentiation indicates regular recombination. According to the standard sex chromosome evolution theory, one might expect the fully differentiated haplotype to have accumulated more male beneficial alleles than the other haplotypes found in males.
To do this, we measured frogs in their natural environment during breeding season. It is a short 2-3 week window in which they all travel and meet in a pond at high latitude. They start breeding as soon as some of the pond is not frozen so they can enter it. My first encounter with the field site was in negative temperatures and wind. I was lucky I was not the one to enter the pond in these conditions to catch the frogs.
We looked for biological differences between the Y haplotypes by measuring male fitness. We did this directly, by comparing the proportion of offspring fathered by males with the fully differentiated Y compared to the proportion of free-swimming males and, similarly, the proportion of males with a fully differentiated Y that were found in amplexus with females, compared to free-swimming males. We also measured male size to check whether different Y haplotypes are associated with body size differences.
One can never prove a negative result, and it is possible that larger sampling or a different measurement would identify a difference between the haplotypes. For example, maybe the system is unstable, and variation is only possible from gene flow which introduces lost variation from nearby populations. Or perhaps there are performance differences between the haplotypes in different environments, such as different temperatures, or in their development time, which our study was not particularly good at capturing as it would require sampling over many years.
However, there is a good reason for the standard sex chromosome evolution model not to apply to frogs, which is that sex-chromosome recombination is different than what the model assumes. Frog recombination depends on phenotypic sex and is extremely limited in males to all chromosome ends, including the sex chromosomes. Consequently, a male-determining mutation will immediately restrict recombination to almost the whole chromosome carrying the male-determining allele. This would allow it to capture a male-beneficial mutation and is likely the reason for the spread of a new male-determining mutation. However, it would make it difficult for additional male-beneficial alleles to accumulate, because they can mostly arrive by mutation. Rare recombination events are possible through rare sex-reversal cases when XY individuals develop as females and thus have a female-like chromosome-wide recombination pattern which allows recombination on the Y. The rare sex-reversed individuals, which have been observed in nature, represent a fountain of youth for the Y chromosome, and are likely the source of the semi-differentiated Y haplotypes. Nevertheless, the commonly experiences lack of recombination throughout most of the Y length means that the deleterious effects of no recombination, which are responsible for sex chromosome degeneration according to the standard model, are immediately present in young sex chromosomes in frogs.
It will be interesting to model the evolutionary dynamics of the different forces affecting sex chromosome evolution in systems with achiasmatic meiosis, or meiosis that is extremely restricted to chromosome ends, as in frogs. This type of meiotic recombination is also found in many fishes and Drosophila. It is possible that the sex chromosomes of these species are primarily affected by deleterious mutations over a larger parameter space than they are affected by sexually antagonistic mutations. In this light, it is perhaps not surprising that different frog Y haplotypes do not have obvious differences in fitness.
The published papers is available here.
Mimulus guttatus
Quantitative effects of gene expression (eQTL) and the effects of chromosomal inversions
This project was supported by a National Science Foundation grant to John Kelly.
Chromosomal inversions are known to have strong phenotypic effects, for example in determining life histories, in many organisms. However the quantitative importance of chromosomal rearrangements, which include inversions and translocations, is not known, because only rearrangements of known effects have been historically studied.
We aim to advance our understanding of the role of chromosomal rearrangements in evolution by studying genome-wide inversions at the level of the individual, the population and the species.
To do so, we are sequencing and assembling multiple genomes at varying degrees of phylogenetic distance from each other in order to characterise the accumulation of chromosomal rearrangements over evolutionary time. We are simultaneously obtaining full transcriptome data from the same individuals to quantify the chromosomal rearrangement effects in gene expression.
The organism of choice is the Mimulus species complex, because it harbours known inversions with phenotypic effects. In addition, Mimulus species have small genomes and high rates of divergence between populations and species, while retaining genetic compatibility, making them ideal to study chromosomal rearrangements.
Silene
The projects were initiated by Lynda Delph.
QTL for traits under sex specific selection in between population crosses
The level of sexual dimorphism depends on the environment where Silene latifolia grows. We are looking for QTL based on a cross between plants from dry and wet environments, with a special interest to QTL that localise on the sex chromosomes.
I am analysing ≈30 traits from a large F2 cross of plants from Spain and Croatia. Some of the traits are involved in both adaptation to a dry/wet environment and reproductive fitness, in an antagonistic way. For example, fast growth and flowering is generally beneficial for males, but deleterious in a dry environment. Consequently, it is sexually antagonistic (has different fitness optima for the two sexes) in the less stressful Croatian environment.
Such sexually antagonistic traits are predicted to occur on the recombining part of the sex chromosome, because they are expected to benefit from a non-random association with sex, while remaining capable to responding to environmental change.
We are constructing a genetic map and performing QTL analysis to test this prediction.
RNAseq of experimentally evolved plants to male and female traits
Lines with large and small flowers (good for females and males, respecively) have been generated by artificial selection. Gene expression differences between those lines should reveal the first genes that respond to such sexually antagonistic selection.
Plants from the same population have been selected for 4 generations on flower size. Both large and small flower size lines have been generated.
Flower size has different optima yet differs between the sexes, with large flowers being beneficial to females. We are investigating the changes in gene expression associated with selection in one sex at a time.
We are particularly interested in the genomic location of genes that respond to sex-specific selection.
YY Silene latifolia are inviable
No YY plants are produced when the Y is wild type, confirming the Y is significantly degenerated.
The Y chromosome of many organisms occurs as a shadow of its former self, in a degenerated state. Silene latifolia is one of the main plant species that has provided extensive evidence that the Y chromosome degenerates. Under the microscope the Y looks larger than its previously identical partner and is enriched in transposable elements and other repeats, which are also visible cytogenetically. Sequencing studies have found that about 50% of the genes on the Y have become non-functional by mutations and there have probably been two separate events that stopped the recombination between the X and the Y, that have resulted in characteristic patterns of strata of different levels of X and Y divergence.
Such events are expected to have biological consequences. One way to assess them is to generate and observe individuals with a YY genotype. This has been previously done using mutated Y chromosomes, and the effect depended on the Y variant used.
At the end of the last field season, Lynda Delph came to my office holding a dry fruit from an otherwise normal male (1/1151 flowers was female). We quickly became very excited because the offspring of this fruit come from a XY x XY cross. 25% of them should have the YY genotype, allowing us to assess the ability of a wild type Y chromosome to replace the function of the X in plant development.
We used PCR to score DNA sequences that were previously found to only occur on the X or the Y chromosome to assign the offspring of our cross into the XX, XY and YY genotypes. We did not see any YY individuals, and this was supported statistically i.e., it is very unlikely we would not see any given our sample size. Our data confirm that YY offspring do not develop. This suggests that there are genes on the X that are required for generating a normal plant, i.e. that the Y chromosome is too degenerated to support normal development.
Furthermore, it is likely that this developmental failure affects the production of ovules, because the rare fruit that we used had very few ovules, and a normal proportion of aborted seeds.
Read the full paper here.
Mercurialis annua
Characterisation of homomorphic sex chromosomes in a species complex with polyploidy and dioecy - monoecy transitions
The Y chromosome of Mercurialis annua is homomorphic yet 1/3 is not recombining and is enriched in sex-biased genes, perhaps a sign of ongoing sexual antagonism.
Most of the times when you see a plant with separate male and female individuals, you are witnessing an example of a new evolution of sex chromosomes.
The study of different plant species is therefore a great opportunity to test the theoretical predictions about how sex chromosomes evolve and become different to each other. There are non-selective and selective reasons for this. The former involve their smaller population size compared to the autosomes, while the latter come down to the fact that each sex chromosome experiences a different genomic and ecological environment, which is specific to each sex. To be exact, there is 100% association of the Y chromosome and 33% association of the X chromosome with males (and 66% association with females, since they have two copies of the X).
Old sex chromosomes (of mammals, birds, Drosophila and some plants) have been found to be genetically degenerated. This is predicted to be a secondary consequence of the evolution of recombination suppression on the sex chromosomes, which can only be studied in young sex chromosomes, because the degeneration of old sex chromosomes masks most of the beneficial effects associated with each, that are hypothesised to have been the reason for the suppression of recombination.
In principle, the sex chromosomes are expected to stop recombining if this leads to the generation of combinations of genes that benefit the sex in which the sex chromosomes occur most frequently (the Y in male and the X in females). For example, a male-beneficial variant linked to the male-determining gene will achieve a higher frequency in the population relative to an alternative Y chromosome that sometimes recombines to lose this variant. Similarly, a Y chromosome carrying a male-beneficial gene variant that is simultaneously deleterious to females will outcompete other Y variants even faster, because the products of recombination will be outcompeted even faster. These genetic variants that have opposite fitness consequences for the different sexes are termed "sexually antagonistic".
In the short term the lack of recombination between sex chromosomes carrying sexually antagonistic genes maximises the fitness of each sex, so they are expected to accumulate on the sex chromosomes. However, the recombination of genetic information is beneficial over the long term, for example it allows the removal of deleterious mutations that occur within a stretch of genes. That is why the fate of all non-recombining sex chromosomes is to degenerate.
And this brings us to the recently published genome of the annual plant M. annua. We seem to have been very lucky in choosing M. annua as a study species because it shows signs of recombination suppression without much of the degeneration that is expected to follow. Its sex chromosomes may therefore be in the right stage where the reason for recombination suppression (sexually antagonistic selection) may be still present, and not masked by degeneration.
The M. annua genome project evolved over a long period of time and combined lots of different datasets. We report an assembly of the genome, a genetic map that assigned ≈10,000 genes on the expected number of chromosomes and a transcriptome that allowed to identify genes that are expressed at a higher level in one sex relative to the other (sex-biased genes). We also obtained various metrics associated with DNA sequence, such as divergence to other species or within-species diversity and tested whether the sex chromosomes stand out in any of these metrics. Having a genome allowed to produce a high throughput assay allowing to selectively sequence genes in many populations covering the species range.
Our genetic map and species-wide data independently confirm that 1/3 of the sex chromosomes do not recombine. This is surprising because the recombination suppression was recent: we found only one clear case of a non-functional Y copy and comparisons with the closely related species M. huetii showed that the recombination suppression happened after the species split.
Interestingly the sex chromosomes are enriched in sex-biased genes. Overall there are more genes with higher expression in females than males, but the few genes with higher expression in males compared to females have more extensive expression differences between the sexes. Assuming that sex bias in gene expression is associated with fitness, which would have caused the lower expression in the opposite sex, this simultaneous masculinisation and feminisation of the sex chromosome in terms of gene expression can be interpreted as the outcome of ongoing sexual conflict. It may have been associated with the reason behind the observed recombination suppression between the sex chromosomes.
This pattern is only a hint towards potential ongoing sexually antagonistic selection on the sex chromosomes, which needs to be followed up. The existence of a large region with suppressed recombination without much degeneration, and the availability of many closely related lineages for comparative genomic analysis makes M. annua a very promising species to understand the early events in sex chromosome evolution.
Read the full paper here.
Double Y dosage reduces fertility, but not inviability in Mercurialis annua
Plants with 2 Y chromosomes suffer infertility, but not inviability, perhaps because XX individuals that produced male flowers had an advantage in establishing populations in the past. Or YY plants have abnormal gene dosage.
The continuous growth of plants is based on embryonic tissue they contain. Unlike animals, they make developmental pathway choices as they grow, through the differentiation of pluripotent cells into the different parts of the plant. This regeneration potential allows plants to grow from cuttings and reproduce clonally. When combined with the limited differentiation between the sexes, the regeneration ability of plants makes it possible for a normal male plant to produce female flowers, occasionally or in exceptional circumstances.
Mercurialis annua is an annual herb with male and female individuals and is an interesting plant model for studying sex chromosome evolution. While it has a XY sex chromosome system, the sex chromosomes are homomorphic and of limited differentiation, making them interesting for informing early sex-chromosome evolution theory. Males can be induced to produce female flowers by spraying them with a feminising hormone, or by repeatedly cutting the top of the plant. We used both approaches to generate female flowers and studied the phenotypic effects of having 2 Y chromosomes in the resulting offspring.
Since the female flowers originated in a XY individual, their progeny represent a XY x XY cross. We developed a molecular genetic assay to distinguish the X and Y chromosomes and genotyped the offspring. Both the genotyping and the sex ratio in the first generation suggested that YY individuals were perfectly viable. We also measured various phenotypes such as height and flower mass and did not find any major differences between XY and YY males.
We might expect the XY and YY males to differ because the Y chromosome is subject to opposing evolutionary forces. On one hand it is expected to accumulate male beneficial mutations, because the genes on the Y are always present in males. We might therefore expect a super-male phenotype from YY individuals. On the other hand, the Y has degenerated somewhat, which might result in lower fitness of YY individuals, if some genes on the Y are not working well, and a X copy of a gene is required to compensate for the degeneration of Y copies. In addition, there is an unusual gene dosage of YY plants whose ratios of Y: X or Y : Autosome genes have not been tested by selection, since these plants are very rare in nature. Like most untested changes to something that works, these dosage effects would likely work less well than the progenitor state, and result in lower fitness.
Our study of the sex ratio in the offspring of the XY and YY males and direct observations of pollen from the YY plants suggest that they are infertile, at least partially. It is surprising that fertility was affected before viability, because the Y is usually considered to be specialising in male function. For example genes on the Y improving pollen fertility are likely to be under strong selection to maintain good function, which would have selected against any mutations of genes on the Y that negatively affect pollen. Nevertheless, pollen infertility could have been caused by the dosage effects mentioned above.
We also came up with a selective explanation for our data, relating to the ability of plants to produce flowers of the opposite to their genotypic sex. This is termed sex inconstancy, and it has been found in field populations of M. annua. It is possible that the X has been under strong selection to maintain genes involved in pollen function in the recent past. For example it would be an advantage for XX individuals to occasionally produce functional male flowers, because it would allow long-distance dispersing XX individuals to establish a population on their own, without relying on long dispersing pollen for fertilisation.
In fact there is a hexaploid lineage of M. annua (in the Iberian Peninsula) in which some populations have hermaphrodite flowers, and some populations have males and hermaphrodites. This situation can be explained through a metapopulation model where hermaphrodite individuals can establish populations on their own by selfing. Once they build to a higher density they can be invaded by males, who outcompete hermaphrodites in male function. Diploid inconstant male individuals might have worked in a similar fashion in the past as they are functionally hermaphrodite even though they have separate male and female flowers. For example sex inconstancy might have conferred a fitness advantage at the end of the last ice age when new locations became available for colonisation. This would have generated strong selection to maintain pollen fertility function on the X.
More experiments will be required to understand the reasons behind the poor pollen production of YY individuals. Their initially unexpected infertility and viability has opened up new understanding in how the modular function of the plant body plan may have influenced the evolution of their sex chromosomes.
Drosophila pseudoobscura
The projects were a collaboration between the labs of Mike Ritchie, Rhonda Snook and Andy Cossins.
Experimental evolution under high and low sexual selection
Sexual selection leads to the evolution of traits of one sex based on their preference by the other. It is known to lead to rapid divergence and often results in extreme phenotypes. It can be a major force in the diversification of populations and generation of genetic diversity. The effects of sexual selection on phenotypic diversity have been the easiest to study while the effects on the underlying genetics have been elusive. By studying the genetic responses to sexual selection manipulation we hope to understand their relevance to the generation of genetic diversity and, ultimately, biodiversity.
Lines of the fruit fly Drosophila pseudoobscura have been evolving in the lab under different strengths of sexual selection, controlled through sex ratio manipulation, for about 150 generations. Populations experiencing a male-biased sex ratio experience greater male competition and stronger female choice on male characteristics. Previous studies have identified various phenotypic changes that affect mating success, including both courtship behaviour, such as courtship intensity and courtship song, and reproductive physiology characters, such as accessory gland size and effects of the males on the remating interval of females.
The use of next generation sequencing allows to study the underlying genetics behind such phenotypes. We hope to obtain information on the types of changes associated with responses to sexual selection. Such information would answer exciting evolutionary questions such as:
Whether the same phenotypic response is based on the same underlying genetics when it evolves multiple times independently.
Whether the X chromosome has disproportionately more changes than other chromosomes, as suggested by theory.
Whether changes in gene expression are associated with changes in gene sequence.
Reproductive tissue mating response after experimental sexual selection
Experimentally elevated sexual selection increases male manipulation of the female post-mating response, but also makes females poised in readiness for mating and more resistant to male manipulation.
We used tissue-specific transcriptomes of reproductive tissues from both sexes. Working with specific tissues is beneficial because it minimises the effects of allometry from the observed response, it can detect gene regulation evolution even if, for example, testes have evolved to be larger under high sexual selection. One disadvantage of using specific tissues from a small insect is that there are fewer cells that provide the RNA, so the dissections required to obtain sufficient RNA took a whole week. We had 4 2-hour blocks of dissections per day.
Some dissections involved mated females that had to be prepared first. This made for a very busy week, so that the following week of (lots of) RNA extractions seemed relatively mild in comparison. With so focused physical work events blur and it is difficult to recall the experience. Nevertheless, I remember we were late for the xmas party because we wanted to finish within the year, and that we looked at an inverted microscope after tissue disruption with metal beads, to make sure we had successfully pulverised the very small reproductive tissues, including the hard spermathecae. I also have a vague memory of delivering the samples in ice in person at the Centre for Genomics Research in Liverpool, which prepared and sequenced the libraries.
All this work was before we started analysing the data, which took the most time. Here I will focus on two of our results, both of which involve interactions. But first, a brief note on the experimental design.
High sexual selection females do not require mating to activate some gene expression
Our experimental design allowed us to look for an interaction between the effect of sexual selection (the overall difference between high and low sexual selection females i.e. E vs M contrast) and the effect of mating (i.e. virgin vs mated contrast). Identifying interactions is quite a standard statistical practice, the main issue is interpreting the result and understanding the biology behind an interaction. One of the most interesting results of the paper is the strong interaction we found between mating status and sexual selection history.
To visualise the interaction, we plotted the difference in expression between virgin and mated females, separately (on the x and y axes) for the E and M females.
It is quite striking that most of the genes significant for the interaction are in one direction (green points): they are genes expressed in virgin high sexual selection females (E), but only expressed after mating in monogamous females (M).
The result supports previous phenotypic studies suggesting that the high sexual selection females are primed for mating. This might be expected, since they have evolved in an environment where they have the opportunity to mate with different males. The fact that they do not need mating to activate some genes that monogamous females only activate after mating suggests the possibility that there may be a cost to having them expressed as virgins. Their functions are related to the immune system and egg production which makes sense for the female reproductive tract and further support our interpretation that E females are poised for mating.
A similar analysis in ovaries found many genes with a significant main effect of the selection treatment, few genes changing expression based on mating, but no genes significant for an interaction, suggesting the ovariole genes of the two types of female respond the same way to mating. One explanation is that the 6 h after mating in which we sampled for RNA did not change much in terms of gene expression in the ovaries. So perhaps the main conclusion is that it was a very good call to separately analyse the two female reproductive tissues as they show very different results.
High sexual selection males manipulate the female mating response more - but high sexual selection females seem more resistant to this manipulation
We were able to study the effect of different types of male in the female mating response by analysing females of both selection treatments mated with males they had coevolved with and with males they had not. This is a complicated interaction so we plot separately the two types of female.
Comparing the number of significant points between the two graphs illustrates that E females do not change the expression of as many genes as M females do, regardless of mating partner. This also supports the suggestion that E females are primed for mating, since they activate genes prior to mating, compared to the M females.
Let us now focus on the two types of male (E or M). Notice that the same type of male has a different colour in the two plots, for example high sexual selection males affect the expression of red genes for a) but blue genes for c). Once you realise this, it is quite clear that E males cause more genes to change expression between virgin and mated M females (a - red genes, 155+103) than E females (c - blue genes 66+73).
The results lead to two intriguing interpretations:
- High sexual selection males manipulate the female postmating response more than low sexual selection males
- Simultaneously, high sexual selection females are more resistant to this manipulation than low sexual selection females.
For the full story, including results on virgin tissues (where male accessory glands and female reproductive tracts had more gene expression changes between E and M treatments than testes and ovaries) and comparisons with proteomic work, see the paper.
Drosophila montana
Selection analysis and QTL for traits associated with reproductive isolation within species in Drosophila montana
The project was a collaboration between the labs of Mike Ritchie, Anneli Hoikkala, Roger Butlin and Jon Slate.
Sexual selection leads to the evolution of traits of one sex based on their preference by the other. It is known to lead to rapid divergence and often results in extreme phenotypes. Sexual selection is a major force in the evolution of populations and generates genetic diversity. Its effects on phenotypic diversity have been the easiest to study while the effects on the underlying genetics have been elusive. In this project we hope to shed light to genetic responses to sexual selection and understand their relevance to the generation of genetic diversity and, ultimately, biodiversity.
Drosophila montana belongs to the virilis group of Drosophila. The male courtship song has been shown to be necessary for successful courtship as well as the target of sexual selection. We are performing a QTL study of parameters of the male song known to influence female choice, as well as female choice itself. We are using wild-caught populations to minimise the effects of keeping flies in the lab in the observed variation.
One long-term aim of the project is to determine whether intrapopulation variation in courtship characters is due to candidate genes, usually identified from interpopulation crosses and mutations studied in lab populations. Alternatively, some of this variation could be due to novel QTLs.
Another long-term aim of the project is to assess whether the same regions of the genome are involved in behavioural reproductive isolation a) in different populations of the same species and b) in closely related species. We hope that such comparative studies will increase our understanding on how repeatable evolution is and whether some outcomes can be predicted.
Podisma pedestris
Ribosomal DNA cytogenetics and a sex chromosome hybrid zone
My PhD was at Queen Mary University of London with Richard Nichols.
I worked on the Podisma pedestris hybrid zone, trying to understand the cause of hybrid inviability. Selection against hybrids is one of the two forces that maintain the hybrid zone (the other being dispersal of the pure races into the zone), therefore attributing a particular genomic sequence to it would be of great significance in understanding the zone dynamics.
Previous studies had suggested the involvement of ribosomal DNA in hybrid zone maintenance. I took a cytogenetics approach (fluorescent in-situ hybridisation) to directly visualise the rDNA loci in various grasshopper populations. These data were combined with data at the sequence level from Irene Keller, a postdoc in the lab.
Using a combination of different techniques applied to rDNA, our research looked into concerted evolution (Keller et al., 2006), biogeography (Veltsos et al., 2009) and nucleolar dominance. The unifying theme behind these studies is the large genome size of P. pedestris (about 6 times bigger than the human genome, or 100 times bigger the Drosophila melanogaster genome) which may be evolving differently than the better studied model organisms, which have smaller genomes.
I also studied the effect that the chromosomal fusion that distinguishes the two races may have in the hybrid zone. The idea is that the sexually antagonistic effects that quickly arise on a Neo-Y chromosome may generate some of the hybrid inviability in the hybrid zone. For example the Neo-Y can enter females in the hybrid zone, which is expected to select against them. This selection disproportionately affects the ancestral, unfused X chromosome. We employed simulations to show that under some selection regimes, selection on the Neo-Y can lead to the establishment of the whole XY system throughout the species range (Veltsos et al., 2008), leaving the genetic variation on the autosomes to neutrally admix. The model provides a way for new sex chromosome systems to establish throughout a species range without the species going through an extreme bottleneck.
Comparisons of the model predictions to real sex chromosome hybrid zones in nature can inform us on the evolutionary forces affecting young sex chromosomes.