Sexual mating is key to evolutionary processes, allowing combination of beneficial alleles between individuals. Data across species suggest that organisms can exert selection of partners that in part depends on factors such as genetic distance and fitness of partners. Fitness is the ultimate evolutionary trait, and its inheritance as a quantitative trait is multi-genic. Revealing the factors that govern mate selection and fitness inheritance are critical for understanding the effects of sex on evolution. Yet, large data are limited, main conclusions are inconsistent, and bigger and consistent datasets are needed. Here, we developed a massive mating assay that enables the simultaneous quantification of mate choice among ~100 Saccharomyces cerevisiae natural strains. DNA barcode recombination allows sequencing of recombined barcode pairs to reveal mating frequencies among all strains, and fitness measurements of both parental haploids and offspring diploids. We find preferential mating with partners with which higher-fitness offspring emerge. Following inheritance of fitness, we find that offspring fitness in fermentable carbon source correlates mainly with parental fitness, while on non-fermentable carbon, offspring fitness is mainly an increasing function of the genetic distance between their parents. Our results suggest that mate choice and fitness inheritance depend upon environmental condition and strains genetics.
Article
A high throughput-mating assay between S. cerevisiae natural strains reveals a genetic basis for sexual selection and fitness inheritance
https://doi.org/10.21203/rs.3.rs-3127016/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Mate selection is wide spread in nature and exists in many species including fungi and yeast, plants, insects, fish, lizards, amphibian, birds and mammals [1], [2], [11]–[18], [3]–[10]. In some species mate selection was shown to be advantageous, producing highly fit offspring [19], [20]. Fitness is a quantitative trait, involving many loci, and as such it resembles other quantitative traits, such as height in human [21], [22] or body size in other animals [23], [24], in which the exact inheritance mode is not always clear [25]–[27]. Fitness is known to be affected by gene compatibility and incompatibility, that are responsible for two phenomena, hybrid vigor on the one hand, and genetic lethality on the other [28]–[36]. Avoidance of partners that share high genetic similarity, was shown in many organisms [37]–[41]. In contrast, mating with partners with low genetic similarity was reported to give rise to sterile or low fit offspring [41]–[44]. Thus, fitness should be maximized in an intermediate range of genetic [16]. Indeed, Wei and Zhang show that for yeast, plant and mice, there is an intermediate optimal genetic distance that maximized offspring fitness [45].
The budding yeast Saccharomyces cerevisiae, can grow either vegetatively as haploid or as diploid, and can also mate sexually. Following pheromone sensing, haploid cells (i.e., Mata or Matα) mate to generate a diploid cell that can then continue growing vegetatively (refer to here as “offspring”). Upon nitrogen and carbon starvation, diploid cells undergo sporulation and meiosis to form haploid spores. At the end of the process 4 spores are produced and are stored within the intact cytoplasm of the mother cells, the ascus [46]. Subsequently, when reproducing sexually, yeast are most often mate with their sibling haploid cells close to spore germination [47], [48]. Nonetheless, previous studies have shown that outcrossing and haplo-selfing does take place and that although it is relatively infrequent [49], it is sufficient to result in mixing of genomes through recombination [50]–[57]. It was inferred that environments that provide opportunities for outcrossing may not be uncommon; One such environment is vineyards, in which several different yeast strains have been isolated from the same vineyard, probably transported there by insects [58]). Further, studies in social wasps showed that an insects digestive track forms an environment in which yeast cells can meet and outcross [50] and a study of yeast from woodland showed a high rate of outcrossing between asci [51]. Analysis of diploid S. cerevisiae isolated strains showed extensive heterozygosity [50], [52]–[54], [59], potentially due to the advantageous nature of mating in evolution [57], especially in strains isolated from the human associated environments [54]. In addition, it was recently shown that repetitive cycles of hybridization generate diversity, providing raw material for speciation [60]. The observed levels and patterns of such heterozygosity were interpreted to result from outcrossing during their most recent sexual cycle.
Since outcrossing plays a key role in shaping the yeast population structure, a major question that remains unanswered is to what extent mate choice is exerted in yeasts. One criterion for mate choice was shown to exist in S. cerevisiae, as both mating types will mate with the strongest pheromone secreting partner, if given a choice [61], [62]. additionally, yeast can also sense Mata to Matα ratio and change their mating behavior to be advantageous [63]. Yet, whether additional factors affect mate choice remains unknown. To study mate choice and fitness inheritance in a high-throughput manner, on a collection of yeast strains with unprecedented magnitude [35], [45], we devised a novel mating-recording barcode strategy (based on the barcode fusion genetic system from [64]) and performed en mass mating assays among ~ 100 haploids from the S. cerevisiae natural isolates collection, gathered by Peter et al. [52]. This method enabled us to produce and track thousands of diploid offspring to investigate mate choice and fitness inheritance at scale.
A massively parallel barcode recombination system for tracking yeast sexual mating and fitness
To study mate choice and fitness inheritance, we have used the S. cerevisiae natural strains collection [52]. From those, ~ 100 strains were chosen that vary in growth fitness and in pairwise genetic distance (GD) (Figure S1 and see supplementary note1). To allow high-throughput experiments, these natural strains were genetically engineered to be used as haploid parental strains and as such, to allow recording of all their mating pairings. In short, the HO locus [65], [66] was knocked out by a construct containing an antibiotic resistance cassettes, a fluorescence marker, and a barcode recombination element, based on the Cre-lox recombinase (barcode fusion genetics, BFG) [64] (and similarly to [67]) (Figure S2A, supplementary note1). This cassette allows recombination of haploid parental unique barcodes (from the two mating partners, Mata and Matα) into a fused barcode on the same DNA segment that can be sequenced (Fig. 1A) (Figure S2, supplementary note1). To perform high-throughput experiments of both mate selection assay and fitness inheritance assay, all parental haploid strains from the two mating types were mixed and allowed to mate en masse. After mating, diploid cells were retained and haploids that did not mate were eliminated (see Materials & Methods). Offspring diploids were used to quantify mate selection of their parents (Fig. 1C) and subject to a competition assay to measure their relative fitness [68] (Fig. 1D).
To understand if environmental conditions affects mate selection or fitness inheritance, both a sexual selection mating assay and competition experiments were performed in a fermentable carbon source – glucose, and non-fermentable carbon source – glycerol. In addition, we also experimented with a condition- in which haploid parental strains were grown on glucose but were switched to glycerol just prior to mating to reveal if strains select mates according to the current condition or according to recent previous conditions that may affect their fitness and “attractiveness”.
Figure 1.
Sexual selection mating assay reveals pair-specific avoidance
We aimed to accurately quantify the frequency of each potential offspring. Our initial pool consisted of 89 parents from the Mata mating type and 46 parents from the Matα mating type, resulting in a potential total of 4,094 offspring combinations. However, our findings revealed that while 2,668 combinations (~ 65%) of offspring were detected in at least one repeat and under at least one of the conditions examined, the rest were never observed. In Fig. 2A, we present a matrix of all pairwise offspring combinations color coded according to the number of experiments in which they were found. The color gradient ranges from white, indicating no offspring in any condition, to black, representing offspring in all conditions tested. The clustering of haploid strains revealed two distinct behaviors of parental pairs with no offspring: one that consists of parents who showed no, or nearly no, mating activity with any partner across any condition, and the second is which consists of pairs of parents that were generally active across many partners, but they specifically avoided one another. Further, certain strains exhibited condition-specific absence of mating activity, wherein they lacked partners only under one specific condition (Figure S3). There are 6 Matα strains, and 1 Mata strain that have no offspring with any partners, in all conditions. Their hypothetical offspring (n = 574) were removed from further analysis.
Selected offspring in each condition are made either more than, less than, or as expected from their parents’ over-all mating activity
To quantify mate selection, recombined barcodes from diploid offspring were sequenced following en masse mating. A striking observation is that the abundance of the observed offspring spans close to five orders of magnitude (Fig. 2B). We thus hypothesized that this observation represents enhanced or diminished mating selection and/or mating compatibility between specific pairs. We thus compared and contrasted the observed number of offspring from each parental pair to an expectation under a null model in which mating features no attraction or repulsion between parents. This null model was constructed as follows: In each condition, for each pair of parents, we measured the overall mating activity for each of the Mata and Matα parents, as the sum of number of offspring it had across all potential mating partners in the experiment. Given two parental strains of opposing mating types, the product of their mating activity and their initial frequencies, represents the expected number of offspring from each pair under the null model (as described in the equation below):
$$expected \# offspring \left(i,j\right)=$$
\([\sum parent\left(i\right)*initial frequency parent \left(i\right)\left]*\right[\sum parent\left(j\right)*initial frequency parent \left(j\right)]\) ).
To rule out potential biases by barcode fusion genetic (BFG) efficiency and mating efficiency we have performed two validation experiments.I the first, we measured mating efficiency of individual pairs of strains mated in isolation from all others. In the second, we measured the BFG efficiency of individual pairs using barcode deep sequencing (see Materials & Methods and Supplementary note 2). Mating efficiency and BFG efficiency of each pair is correlated to the product of BFG efficiency and mating efficiency of the parents (respectively) (Figure S4, S5 and supplementary note2). Thus, the null model that uses the product of the overall offspring of each parent eliminate BFG and mating efficiency biases. As described earlier, the mate selection assay was done in three conditions, glucose or glycerol as carbon source, or mating in glycerol after growth in glucose, in two repeats each. Figure 2B shows the observed number of offspring of each type vs. the expected number under the null model in each repeat in each of the three conditions. We discern three mating groups defined by observed-to-expected offspring abundance. The over-represented mating group consists of offspring that are found to be 3–4 orders of magnitude over-represented relative to the null model (Fig. 2B, blue dots). In the middle, close to the x = y line, is the mating group of offspring that are observed close to their expected level of abundance, thus reflecting no negative or positive mating selection (Fig. 2B, orange dots). The lower mating group consists of offspring that are absent, despite the fact that under the null model many of them should have been observed (Fig. 2B, green dots). Here too, the 754 offspring that originates from parents that have no offspring in any condition are not shown in this plot. This figure shows that yeast exercise mate selection; the over represented group consists of partners that mate preferentially with one another, while the absent group reveals pair-specific lack of compatibility or selective avoidance.
To verify that our results indicate a mate selection, and are not due to mating efficiency nor BFG efficiency, we used the pairwise mating and BFG experiments. We found that mating efficiency and BFG efficiency between individual pairs of strains was not correlated to the mating cluster affiliations (Figure S6, S7 and supplementary note2). This result indicates that belonging to the three mate groups does not trivially represent mating or BFG efficiencies between individual strains, but rather points to mate selection.
To investigate the potential influence of growth conditions on mate choice, we clustered the six experiments representing three distinct conditions, with two repeats for each condition. The clustering was based on the correlation between the appearance of offspring in the six experiments. Notably, the two repeats of the glucose condition clustered together; the two repeats of glycerol were co-clustered with the two repeats of the glucose to glycerol carbon source switching condition (Fig. 3C). These results indicate that mate choice vary in different carbon sources. In addition, it shows that yeast can quickly change their mate choice upon environmental change and select partners based on the prevailing environmental mating conditions rather than being influenced solely by the conditions experienced in recent generations
High fitness offspring are made more often in the over-represented group in fermentable medium
We next turned to ask what characterizes the offspring in the over-represented and in the absent groups. Here too, the 754 hypothetical offspring from parents that have no offspring in any condition were removed from this analysis. Starting with parental haploids fitness, we partitioned parents (either Mata or Matα) into four bins based on their relative fitness as measured in the parents’ competition (see below and Materials & Methods). We examined the representation of each of the four fitness bins in each of the three mating groups. We find that Matα parents that were extinct during the parents’ competition are enriched in the over-represented mating group. We did not detect consistent trends for Mata parents in the different mating groups based on their fitness (Figure S8).
We next examined whether the genetic distance between parents differs between the three mating groups. We partitioned parental pairs into the four bins of genetic distances between them based on data presented in Figure S1B. In the majority of the experiments, we find that parental pairs in the second shortest genetic distance (0.05–0.5 SNPs per 1,000 base pairs) are enriched among the over-represented mating group. This result shows that mating selection is maximized in this intermediate genetic distance regime (Fig. 3A). Considering the variation in heterozygosity among the natural strains, we hypothesized that low or high levels of heterozygosity is due to a potential genetically encoded tendency against or for outcrossing, a tendency that could be featured in the current mating assay too. Consequently, we sought to examine the potential correlation between the level of heterozygosity of the original strains, and the current GD-based mate choice. We find that for original strains with low heterozygosity (< 0.08 heterozygous SNPs from all SNPs), the second lowest GD is enriched in the over represented mate group (Figure S9). However, for original strains with high heterozygosity, no clear and significant trend emerged. These results suggest that yeast that were not outcrossed in their past natural history (low heterozygosity) tend to prefer genetically similar mates and thus reveal a genetic determinant for in- or out-crossing.
As it was shown in other organisms, mate choice may lead to high fitness offspring [69], [70]. In order to explore this phenomenon in our system, we partitioned offspring diploids into four bins based on their relative fitness as measured in the offspring competition. Specifically, we observed a notable enrichment of the highest fitness offspring in the over-represented mating group across the two repeats of the glucose condition. Conversely, these offspring were significantly depleted from the under-represented mating group in glucose (Fig. 3B). In glycerol, the over represented group is enriched with very low fitness offspring, as indicated by their extinction during the competition experiment. This result suggests that in glucose, but not in glycerol, yeast mate more with partners that would maximize the fitness of their offspring (Fig. 3B).
Figure 4.
Parental fitness and genetic distance govern offspring fitness
As seen before, in fermentable media, yeast select mating partners such that the fitness of the offspring is increased (Fig. 3B). In addition, Figure S8 indicates that they mate less with very low fitness partner. As such, we wanted to examine the modes of fitness inheritance in yeast. In particular, we ask how offspring fitness correlates with parental fitness and with the genetic distance between the parents. To measure fitness, we have used a competition assay. Figure S10 shows a distribution of the relative fitness of all parents, and of all offspring that had calculable fitness in each of the conditions (see Materials & Methods). Strains that were extinct during the competition or that were absent throughout the entire competition were assigned a low-fitness tag. The relative fitness of diploid offspring strains is mildly correlated between glucose and glycerol (Figure S11A). We also found a significant correlation between the relative fitness of the Mata and Matα parents of the same strain on media containing glucose as carbon source (Figure S11B).
Offspring fitness correlates positively with parental fitness on glucose but less so on glycerol
We then wished to examine if the fitness of offspring correlates with the fitness of their parents under each growth condition. We begin by examining the correlation between offspring fitness and the mean fitness of their parents. Figure 4A (and Figure S12A) shows a very different behavior between the fermentable and non-fermentable carbon sources. In glucose, parental fitness correlates positively, moderately but significantly with offspring fitness. In glycerol there is no such correlation between the fitness of the parents and that of their offspring. As an alternative to the mean parental fitness, we also characterized each pair of parents by their maximal or minimal fitness and computed the correlation between each of these parental measurements to offspring fitness (Figure S12B,C). In agreement with the mean parental fitness, we find that in glucose offspring fitness correlates positively and significantly with the maximal or minimal parental fitness, but in glycerol such correlations were much lower. We also examined whether offspring fitness correlated better with the fitness of the Mata or Matα parent and found that the fitness of each of the two parental mating types correlated similarly with the offspring fitness, and similarly to the correlation seen with the mean parental fitness, yet here too, only on glucose (Figure S12D,E). Thus, according to each parental fitness definition, we find that in glucose there exists a positive correlation with offspring fitness while in glycerol we see little or no such correlation.
High content of minor allele correlates with low fitness under a dominance genetic model, especially in glycerol
It is known that minor alleles (alleles that are rare in the population) are linked to low fitness (e.g., genetic disorders) in many genetic traits [71], [72]. As such, we asked if minor allele content of strains, summed over all SNPs in the genome, affect their fitness. To that end, we used the genetic annotation from Peter et al. [52] and estimated the content of minor allele homozygosity of each parental strain, and of each offspring (see Materials & Methods) under three models; (i) dominance, in which the heterozygous is as fit as the major allele homozygous, (ii) recessive, in which the heterozygous is as fit as the minor allele homozygous, and (iii) co-dominance in which the heterozygous has an intermediate fitness of the major allele and minor allele homozygous. For the dominance model only, there is a significant negative correlation between offspring fitness and the estimation of homozygous minor allele content (Fig. 4B). In contrast, in the co-dominance and recessive models, there is no correlation between offspring fitness and minor allele estimation (Figure S13). In addition, by and large, no correlation is seen between parental haploid minor allele content and their fitness (Figure S14).
An optimum, at high parental genetic distances, maximizes offspring fitness, especially in glycerol
Each offspring is defined by the genetic distance (GD) between its two parents. An interesting question is whether offspring fitness is dependent upon parental GD. Figure 4C shows the correlation between offspring fitness and their parental GD. In glycerol, offspring fitness increases rather sharply with parental GD until it reaches a maximum at a distance of about 1.1 to 1.4 SNPs per kb and then it starts to decline at higher GDs (Figure S15). The lowest offspring fitness is observed among genetically close or identical parents, an observation that may be explained by homozygosity of recessive deleterious traits. In contrast, the high offspring fitness at larger GDs may represent cases of beneficial heterozygosity in this condition. Yet, as parental genetic distance become even larger, the extent of heterozygosity may become maladaptive.
In glucose, the dependency of offspring fitness on parental GD is much less pronounced (Fig. 4C). Yet, in this condition too, offspring fitness is maximized around a similar parental GD of about 1.2 to 1.4 SNPs per kb (Figure S15). Since non-fermentable carbon sources require mitochondrial respiration, we repeated all calculations of GD but with mitochondrial functional genes only. We observe similar trends (Figure S16A), namely that in glycerol high parental GD, measured based on mitochondrial genes, correlates positively with offspring fitness. In addition, we note that GDs calculated based on the entire genome, or based on mito-nuclear genes only, are very highly correlated (Pearson R = 0.9, P-value ~ = 0) (Figure S16B).
Dominance and Co-dominance models explains the mode of inheritance in glycerol and glucose, respectively
We hypothesized that an optional explanation for the differences in fitness inheritance in the two carbon sources may be due to different genes that are involved in affecting fitness. To test this hypothesis, we constructed a simple theoretical model of fitness inheritance that implements different inheritance modes, in similarity to [45]. In particular, we examined cases of over dominance (i.e., when the heterozygote (“Aa”) is more fit than the homozygote of the better allele, “AA”), dominance (i.e., when the heterozygote is as fit as the homozygote of the better allele), recessive (i.e., when the heterozygote is fit as the homozygote of the worse allele, “aa”), and co-dominance (i.e., the heterozygote has a fitness equal to the mean fitness of the two homozygote types). Two of those models, the dominance and the co-dominance fit the fitness inheritance behavior observed in the glycerol and the glucose, respectively; both regarding parental fitness and genetics distance (Figure S17). In particular, the co-dominance model shows a positive correlation between offspring fitness and parents’ fitness (measured either as mean, min, or max parental fitness), and also no correlation between offspring fitness and parents’ GD, as observed in glucose. On the other hand, the dominance model shows an increase in offspring fitness with parents’ mean fitness at the low parental fitness range, and then a plateau of no correlation; but a positive correlation between offspring fitness and parents’ GD, as observed in glycerol. The fit between the model and the experimental results suggests a distinct mode of inheritance in each of the conditions.
Offspring fitness can be predicted using a multiple linear regression model
As shown so far, offspring fitness is correlated to parental fitness, minor allele homozygosity content and parental GD (although to different extents in the two carbon sources). Therefore, we next asked if offspring fitness can be predicted based on those three parameters using a linear model (multiple linear regression, see materials & methods). We compared the correlation between the observed offspring fitness to the fitness of the offspring as predicted by our model (Fig. 4D). We succeeded to predict offspring fitness with high correlation in glycerol and moderately yet significantly in glucose (Pearson r of 0.73 and 0.23 respectively). In agreement with the above findings, the model coefficients show that GD is mostly important in predicting fitness in glycerol, while the homozygosity minor allele content and parental mean fitness contributes more to the prediction in glucose as carbon source. These results indicate that in glycerol about 50% of the variance in offspring fitness is explained by parental GD.
The fermentable carbon source mode of inheritance is valid upon a non-metabolic environmental stress
To test if the differences between glucose and glycerol are rooted in the metabolic nature of the conditions or in one of them being more stressful than the other for each strain, we have added a non-metabolic stressful condition, of an osmotic stress (1.2M Sorbitol, yet still using glucose as a carbon source, see Materials & Methods). Modes of fitness inheritance in this osmotic stress resembled clearly the glucose conditions. In particular, in similarity to the glucose condition, under osmotic stress, offspring fitness is correlated positively to fitness of their parents, but not to GD (Figure S18A, B). In addition, offspring fitness in this condition correlated with minor allele counts similarly to the glucose condition (Figure S18C). Lastly, MLR that was performed on the fitness measurements in this condition, shows that it also resembles glucose in respect to the correlation between the model results and the real fitness (Pearson R = 0.32) and to the model coefficient weights (Figure S18D). Taken together these results suggest that it is the metabolic condition, and not the amount of stress, that govern mode of fitness inheritance.
We have shown, for the first time, that yeast choose mating partners (Fig. 2B). In both conditions tested in this paper, yeast choose mating partners that are genetically similar to them, yet not the same (Fig. 3A). Additionally, we observed a correlation in which natural strains with low heterozygosity prefer mating with partners with low genetic distance (Figure S9). This finding suggests an inherent tendency in yeast to engage in in- or outcrossing. Notably, previous research on haplo-selfing in yeast has demonstrated variation among different naturally isolated strains regarding their inclination towards intra-ascus mating versus haplo-selfing [55]. Apart from the GD, in the glucose condition yeast mate more with high fitness partners, while in glycerol they mate more with low fitness partners (Figure S8). At the same time, offspring that are being made more than expected (over-represented) tend to have high fitness, suggesting that mate choice, in glucose increase offspring fitness (Fig. 3B).
We investigated the influence of parental characteristics on mate choice under different growth conditions, mainly different carbon sources. Our high-throughput analysis of offspring fitness revealed contrasting patterns in the two conditions. In glycerol, offspring fitness was predominantly correlated with GD of their parents, while in glucose, parental fitness impacts offspring fitness, as indicated in (Fig. 5A, C). We further asked if different genes, and thus, inheritance mode, may explain this phenomenon. Indeed, while a simplified model using a dominance inheritance mode gave results similar to the glycerol condition, a co-dominance model has agreed more with the results in glucose (Figure S17).
Collectively, our findings suggest that in glucose, yeast has the ability to choose optimally their partners. As fitness is inherited in this condition, choosing high fitness parents results in high fitness offspring, as we see. In contrast, in the glycerol condition, the fitness of offspring is not optimized by selecting high-fitness parents, and the observed mate choice patterns do not align with this objective. It is plausible that choosing partners with high GD could have been advantageous in the glycerol condition. However, given a tendency towards outcrossing that we demonstrate, perhaps by selecting only low heterozygous original strains (Figure S1 and Supplementary note1), collection exhibits a bias towards selecting closely related partners.
Sexual reproduction is conserved throughout the eukaryotic tree of life, and thus is thought to have evolved once, in the last eukaryotic common ancestor [73], [74]. Fungi have diverged from the animal kingdom 1.3 billion years ago [75], [76]. In similarity to many animals exercise of mate selection [1], [2], [11]–[16], [3]–[10], our results show that S. cerevisiae features mate selection too. This suggests that mate selection might be evolutionarily important for sexual reproduction. An interesting question is whether sexual mating appeared in evolution before, or concomitantly with the capacity for mate selection. If the primordial mating system has not yet gained mate selection capabilities, then mating could be conceived to be advantageous even without mate selection. If, in contrast, mate selection has been acquired concomitantly with sexual mating, then this could indicate that sexual mating can be significantly more beneficial if accompanied by mate selection. Our results support the notion that mate selection is deeply rooted in sexual life. Two important questions that warrant further investigations are: what might be potential cellular and molecular mechanisms that facilitate mate selection in yeast, and why is it that in glycerol mate selection is not optimal in the sense of it not maximizing offspring fitness and is it possible that mate selection in this condition optimizes another feature.
Strains and media
Strains in this project are natural isolates taken from Peter et al. [52]. Table S1 shows the list of strains used in this work.
The following media types were used:
SD Glu– 20g/L glucose, 6.7g/L yeast nitrogen base, 1.5g/L amino acid mix and 2% Glucose (according to [72]).
SD Gly– 20g/L glucose, 6.7g/L yeast nitrogen base, 1.5g/L amino acid mix and 2% Glycerol (according to [72]).
SD Glu + 1.2M Sorbitol - 20g/L glucose, 6.7g/L yeast nitrogen base, 1.5g/L amino acid mix and 2% Glycerol (according to [72]) + 1.2M Sorbitol
YPD - 10g/L yeast extract, 20g/L peptone, 20g/L glucose
YPA - 10g/L yeast extract, 20g/L peptone, 20g/L Potassium Acetate
SPO media - 2.5g/l yeast extract, 15 g/l potassium acetate
Antibiotic concentrations and initials:
Hygromycine B (Hyg) - 300mg/L,
Nourseothricin (NAT) – 100mg/L,
Kanamycin (G418) – 200mg/L
Zeocin (Zeo) – 200 mg/L
Plasmids and design construction
Strains were engineered by integrating into their genome a compound design that included the following parts; (Figure S2) (i) genome homology region: 350bp and 790bp homology to the HO locus on both ends of the construct. The specific sequences of homology were chosen to have the least amount of SNPs between most strains, using multiple sequence alignment and BLAST search on strains’ genomes published in Peter et al. [115]. Homology region was amplified from the genome of BY4741 strain (upstream homology region, chrIV:46062...46271, downstream homology region, chrIV: 47982...48682, (ii) Barcode Fusion Genetic (BFG) system (Figure S2B). This region includes (1) the barcodes region flanked with lox sites that was synthesized and cloned into the rest of the construct by Twist based on Fredrick Roth lab design, as well as (2) the Cre enzyme and rtTA inducer (Figure S2A). This region was kindly given to us by the lab of Fredrick Roth [119], (iii) Constitutive markers and fluorescence proteins (kindly given to us by Naama Barkai’s lab). Two combinations of constitutive fluorescence and antibiotic resistance was used, either a yeGFP gene under the control of the TDH3 promoter and Hygromycine resistance cassette with a TEF1 promoter, or a mCherry gene under the control of TEF2 promoter and Nourseothricin resistance cassette with a TEF1 promoter. (iv) Haploid selecting markers. Zeocin or Geneticin (G418) resistance cassettes under the control of the ste2 Mata specific promoter or ste3 Matα specific promoter, respectively. A synthetic terminator was added to both resistance cassettes [77]. Sections (iv) was synthetized by GenScript company.
All of the above parts, except for (ii) were assembled together (in the order described in Figure S2) using restriction free methods and cloned into pET28a (Novagen #69864-3) plasmid by the cloning unit in Weizmann institute, generating one backbone plasmid to generate matA strains and another for Matα. The barcode fusion genetics library section (ii1) was then cloned into the Mata and Matα plasmids by Twist Bioscience Company.
Figure S2A, top panel shows the construct that was transformed to create Mata cells, while the bottom panel shows the construct that was transformed to create Matα cells (from now on referred to as Mata construct and Matα construct, respectively).
BFG construct and primers
The BFG system (shown in Figure S2B) is composed of barcodes flanked by lox sites (based on [64]). While in the Mata construct the order is loxP-BC1-lox2272-BC2, in Matα construct the order is opposite BC1-loxP-BC2-lox2272. The other main component is the Tet-on system that is composed of an rtTA inducer under constitutive promoter and an Cre enzyme under pTet promoter. The rtTA inducer is only active, and can mediate Cre transcription when tetracycline presents.
Following activation of Cre enzyme, a recombination event takes place, recombining the barcode regions of the two parents and results in two fused fragments, one on each chromosome; BC1(matA)-loxP-BC1(Matα)-lox2272 and loxP-BC2(matA)-lox2272-BC2(Matα)
In addition to the barcodes and lox sites, these regions contain unique sequences of ~25nt that are either shared between the parents or are unique to each construct. Those regions can thus be used as primers for amplifying either Mata barcodes construct only (primers H and K), Matα barcodes construct only (primers B and F), fused barcodes only (primers B and I or primers E and K), or all of the above (primers A and G).
Primers used for amplifying Mata construct only, Matα construct only, fused barcode only or all, will be termed Mata primers, Matα primers, fused primers or general primers, respectively. Primers sequence can be found in Table 1.
Table 1. list of primers of the BFG system used in this paper
|
Name |
Sequence (5'->3') |
|
A |
CCTCATAAGCAGCAATCAATTCTATCTATACTTTAAA |
|
B |
TAACCCTTAGAACCGAGAGTGTG |
|
E |
CTCCAGGGTTAGGCAGATG |
|
F |
CAGCGGGATAGTGCGATTG |
|
G |
GGCCGTTACTTACTTAGAGCTT |
|
H |
CCCGTAATGTGCTCGTATGG |
|
I |
GTTATCAGAGGTATGCGAGTTAG |
|
K |
CCTCAGTCGCTCAGTCAAG |
Strain construction
Transformation
Constructs (from the section “plasmid and design construction”) were amplified using the following primers; F: GGTGAAAACCTGTACTTCCAGGG, R: ATGCTAGTTATTGCTCAGCGGT. PCR was done using KAPA HiFi HotStart ReadyMix (Roche, KK2602) enzyme according to manufacturer instructions with the following details: primer annealing Tm of 64°C, elongation of 5 minutes, 30 cycles. PCR products were then transformed into the chosen strains as follows: For each transformation, 5 reactions of PCR were made (to increase complexity of the library). All reactions were ran on agarose gel (0.8% agarose) and size was verified (7kb).
To transform the cells, LiAc protocol [78] was adapted to allow high throughput transformation of many strains in a 96-well plate. In short, 48 strains were taken out of the -80 and inoculated in a 96-well plate in a checkerboard manner to avoid cross contamination between strains. Cells were grown overnight at 30°C with shaking (1200rpm), in a shaker incubator. Then, each strain was diluted 1:20000 in a 50ml tube, (0.5ul of culture into 10ml YPD) and grown for 16 hours at 30°C while shaking. Following the growth phase, four random strains were counted to estimate cell concentration. Cultures usually reached the late log stage (budded yeast, ~7E7cells/ml); each culture was diluted 1:50 into fresh YPD and allowed to grow for another couple of hours.
Cells were harvested by centrifugation (4000rpm for 5min) and washed twice; first with 5ml DDW then with 1ml LiAc 100uM. After the second wash, remaining liquid was vacuumed, and pellet was re-suspended in 60ul DDW. Two 96-well plates for transformation, in a checkerboard manner (one for Mata construct and one for Matα construct) were made with 25ul of PCR product of the relevant construct. 30ul of each strain were suspended into each of the two transformation plates. Transformation mix (100ul PEG 50%, 15ul LiAc 1M, 4ul of 10mg/ml salmon sperm (Sigma Aldrich, D9156-1ML) were added to each well. Plates were incubated for 40 minutes at 42°C. Plates were centrifuges (3000rpm for 3 minutes), and liquid was vacuumed with a multi pipette vacuum adaptor. 150ul of YPD was added to each well and plates were incubated overnight at 30°C. The following morning, each well was plated on YPD agar containing the relevant antibiotic (Hyg for Mata construct plate, and Nat for Matα construct plate). Agar plates were incubated in 30°C for a couple of days until the appearance of colonies.
After colonies appeared, four colonies from each strain were picked, inoculated into liquid SD in 96-well plate (to verify fluorescence) and patched on YPD agar plate with the corresponding antibiotic (for continuing). 96-well plates were grown overnight at 30°C and diluted 1:50 into fresh SD. Plates were FACS analyzed to verify the correct fluorescence marker (cells with Mata construct had yeGFP while Matα construct corresponds to mCherry). One positive colony per strain was chosen to continue.
Sporulation and random spore analysis
Reagents for random spore analysis:
To generate engineered strains, positive diploid colonies (obtained as described in the section “transformation”) were inoculated into a 24-well plate with 1.2ml YPA and grown overnight at 30°C with shaking (1200rpm). Plates were centrifuged (4000rpm, 3 minutes) and washed with DDW, spin down and vacuumed. Pellets were resuspended in 1.2ml SPO media and incubated for 4-5 days in 25°C while shaking constantly. After 4 days, a couple of random cultures were observed under the microscope to verify sporulation, and score sporulation efficiency. In case of low sporulation efficiency plates were incubated an extra day.
50ul of each sporulated culture was transferred into a 96-well plate for random spore analysis as follows:
Cultures were centrifuge and pellets were re-suspended with 50ul TE buffer (Tris 10mM EDTA 1mM, pH 8.0) + 2.5ul β-Mercaptoethanol (dilute 7.5ul into 1ml DDW) and incubated on bench for 10 minutes. Plates were centrifuged and sup was discarded, pellets were then washed with 150ul DDW twice. Pellets were re-suspended in 50ul β-Glucoronidase (Sigma- cat no. G7017-5ML dilute 1:2) and incubated for two hours in 37°C, while shaking. Plates were centrifuged and pellets were washed with 200ul Triton-X100, 0.5%, this step was repeated twice. Plates were centrifuged, and pellets were re-suspended with 120ul DDW and plated on YPD agar plates, with corresponding antibiotic for haploid of the correct mating type (Mata construct were plated on Zeocin containing plates, while Matα construct cells were plated on G418). Plates were incubated in 30°C for 48h until colonies appeared.
From each strain, 2 colonies were picked, inoculated into liquid SD in 96-well plate and patched on YPD agar plate with the corresponding antibiotic. 96-well plates were grown overnight at 30°C and diluted 1:50 into fresh SD. Plates were FACS analyzed to verify correct fluorescence marker (cells with Mata construct had GFP while Matα construct corresponds to mCherry).
Positive colonies (1-2 colonies) were continued to Sanger sequencing to recover barcode sequences.
One correct haploid colony per strain was frozen in a 96-well plate in -80°C.
En masse mating
All verified haploid strains were taken out from the -80°C into YPD media with corresponding antibiotic (Hyg or NAT for Mata or Matα respectively) in a 96-well plate using pinners. Strains were grown overnight (30°C, while shaking) and then diluted 1:1000 for another overnight incubation in 30°C. Strains were diluted 1:50 into either SD-Glu or SD-Gly and grown for another couple of hours in 30°C to reach mid-log phase. OD was measured to all wells by plate reader (infinite 200, Tecan). All Mata strains and all Matα strains were mixed (separately for Mata and Matα), based on the measured OD, such that they will have equal representation in the mix. Mixes were centrifuged and re-suspended in 0.2X volume to create a 5-fold increase in cell concentration (~1E8 cells/ml). En masse mating was conducted by mixing 50ul (~5E6 cells total) of each mating type mix in a 1ml medium (either SD-Glu or SD-Gly) with doxycycline (final concentration of 10ug/ml). Mating was executed for 20 hours in 25°C without shaking. One condition was done where strains were diluted 1:50 into SD-Glu and grown to reach mid-log, but then switched to SD-Gly after centrifugation, thus mating took place in SD-Gly. Three repetitions of en masse mating were done per media.
Mate choice
Mating was performed as described in en masse mating section. Mating was done in three conditions; cells growth and mating on SD-Glu, cells growth and mating on SD-Gly or cells growth on SD-Glu followed by mating in SD-Gly (termed Glu switch to Gly experiment).
Sorting experiments were done in three consecutive days. Mata and Matα mixes were kept in 4°C and used for mating in all three days. 12 mating reactions were performed for each media type, in a 24-well plate. Mating was carried out for 20 hours in 25°C without shaking in all 3 media type. After 20 hours of mating, all mating reactions of the same media type were combined, and EDTA was added for a final concentration of 5mM.
Flow cytometry analysis and sorting were performed on a FACSAria Fusion instrument (BD Biosciences) equipped with a 405, 488, 561 and 640 nm lasers, using a 100 mm nozzle, controlled by BD FACS Diva software v8.0.1 (BD Biosciences), at The Weizmann Institute of Science Flow Cytometry Core Facility. Further analysis was performed using FlowJo software v10.2 (Tree Star). Cells were gated according to FSC and SSC to avoid debris and big aggregates. Another gate of high GFP and high mCherry was determined for sorting of offspring only.
Each day, the order of experiments to be sorted was changed to avoid bias. Cultures were kept in 4°C before and while sorting of other experiments took place.
Approximately 5E7 cells were sorted per experiment, resulting in ~1E6 offspring cells.
Pooled competition
For the pooled competitions= each of the en masse mating was continued to a competition assay to measure the fitness of each offspring. Competitions were done in a daily dilution (1:240) manner in either SD-Glu, SD-Glu + 1.2M sorbitol or SD-Gly, with both Hyg and NAT to select for diploids. In the first dilution, media also contained doxycycline (10ug/ml). Cultures were grown in 50ml tubes at 10ml volume in 30°C while shaking. Dilution was carried out every day (in SD-Glu) or every two days (SD-Glu + 1.2M sorbitol and SD-Gly) by transferring ~40 ul of the culture into 10ml of freash media.
Cells were frozen in 30% Glycerol and kept in -80˚c every dilution (8 generations per dilution, for a total of ~55 generations). All but the first dilutions were freezed leading to 6 time points per a competition experiment.
Competition experiments for parents (either Mata or Matα) were conducted similarly.
Frozen samples were used for library preparation using the following time points; offspring (diploids) – generations 16, 24, 32, 40, 48, 56; haploids (parents): generations 8,16,24,32,40.
Library preparation, sequencing and read analysis
For all experiments, library preparation was done using a similar protocol. Primer and DNA extraction methodology used in the different experiments vary, and are specified here below.
Libraries for sequencing the barcode region were constructed by designing Plate-Row-Column PCR methodology; in which a first PCR is done using primers targeting the barcode region, plate barcode and tails that match Illumina adapters (F: ACGACGCTCTTCCGATCTNNNNNBFGprimer, R: AGACGTGTGCTCTTCCGATCTNNNNNBFGprimer
Capital letters corresponds to Illumina adaptors, N correspond to plate index and BFGprimer is the primer shown in Table 1. First PCR was done in 25ul final volume with 2ul of template DNA (either genomic DNA or after cell blow up in 20mM NaOH and boiling for 15 minutes). PCR program: Tm of 60°C, elongation of 10 seconds, ~20 cycles. Usually, 4 PCR reactions were done per experiments and they were pooled together after the first PCR (to avoid PCR biases). 2ul of the first PCR was used as a template for the second PCR.
A second PCR with the following primers
(F: AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT,
R:CAAGCAGAAGACGGCATACGAGATNNNNNNNNGTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT. N corresponds to Illumina index for library multiplexing) was carried out to attach the adapters for the Illumina run. PCR was done in 25ul volume. PCR program: Tm of 62°C, elongation of 10 seconds, ~20 cycles. After second PCR libraries were cleaned using SPRI beads ~1.2X ratio) to eliminate unspecific bands and primer dimers.
Amplicons were sequenced using paired-end methodology, on the NovaSeq platform (Illumina) (see kits used in Table 2).
Table 2. DNA extraction and library preparation protocols for each of the experiments
After initial de-multiplexing by the Illumina platform, libraries were further separated based on the plate index using cutadapt [79]. All reads were further processed by cutadapt to leave only the barcode region. For alignment, a synthetic genome from all strains’ barcodes was made using bowtie2 (bowtie2-build command). Alignment was performed using bowtie2 as well. Read counts per variant were determined by in-house script.
Fitness estimation based on pooled competition
Fitness was derived by employing a Maximum-Likelihood (ML) algorithm on all read count measurements along the competition experiment per variant. Briefly, first, each variant fitness is estimated by using a simple loglinear regression over the first three time points. Based on these estimations, the initial relative frequencies of each variant, and a noise model that accounts for experimental errors [68], expected trajectory of each variant is estimated and compared to the measured trajectory. Next, small changes are made to fitness estimates, comparison is repeated, fitness is updated if they better fit the data (higher likelihood). This procedure is performed iteratively until fitness estimates are stable (maximized likelihood).
We measured relative fitness as vegetative growth rate of each of the haploids from each mating type at the presence of all others, and of each of the diploid offspring at the presence of all others. Measurement of the relative fitness of each strain in each such pool is performed by deep sequencing of the parental (for haploids) and recombined (for diploids) barcodes in pooled competitions, in similarity to previous studies [68], [80]. Relative fitness was measured and calculated seperatly for each condition. Fitness is derived
from the following equation: 
The frequency of a strain at time point t is a product of its initial frequency and its relative fitness, raised to a time-dependent power (f = frequency of a strain at a given time point, t = time, anc = strain in time point zero, s = fitness coefficient). Overall, we performed the competitions over 40 and 56 generations for the haploids and diploids respectively.
Filtering strains and re-assigned fitness
After initial fitness calculation, strains were filtered, and for some strains fitness was re-calculated (according to the criteria describes below). Fitness distribution was normally distributed (between -0.2 and 0.2) with an additional single peak (usually at -0.6 or -0.8 that was detected as variants that were extinct during the competition).
Offspring (diploid) fitness re-assignment
First, a goodness of fit (GoF) was calculated per variant as follows; PyFitSeq algorithm outputs the calculated read counts in each time point, given the initial read counts in the first time point and the calculated fitness. For each variant, GoF was determined as the Pearson coefficient between the calculated read counts and the real read counts data. Strains were filtered to have GoF>0.5 (~95% of the data in SD-Glu and SD-Gly, 85% of the data in SD-Glu + 1.2M sorbitol). In addition, variants were filtered to have at least 100 read counts in the first time point, or having more than 10 read counts in 3 additional time points (~75% of the data). Next, a fitness per variant per media was calculated as follows; (i) if the variant was not extinct in any of the repetitions per media, fitness of variant was determined to be the average of the repeats (ii) if variant was extinct in more than 60% of the repetitions per media, variant was determined to be “extinct” (arbitrary fitness of -1), (iii) if variant was extinct in less than 40% of the repetitions per media, the extinct value was removed, and fitness was determined to be the average fitness in the other repetitions (iv) if variant was extinct in 40% to 60% of the repetitions fitness was calculated for each variant in each repetition. In case this score was <-0.4, fitness value in that repetition was changed to “extinct”, then, if the variant was extinct in >60%, variant was determined to be “extinct” (arbitrary fitness of -1), (v) lastly, a few variants could still not be given a single value. Those variants were inspected manually, showed a genuine decrease in read counts even if not extinct, so they were determined as “extinct”, but given an arbitrary fitness value of -2 (see Table 3).
Parents (haploid) fitness re-assignment
First, strains were filtered to have at least 200 reads in the first time point (~80-90% of data). For filtered variants, fitness was calculated similarly to the offspring calculation; (i) if the variant was not extinct in any of the repetitions per media, fitness of variant was determined to be the average of the repeats (ii) if variant was extinct in more than 50% of the repetitions per media, variant was determined to be “extinct” (arbitrary fitness of -1), (iii) if variant was extinct in 25% of the repetitions per media, the extinct value was removed, and fitness was determined to be the average fitness in the other repetitions, (iv) lastly, a few variants could still not be given a single value. Those variants were inspected manually, showed a genuine decrease in read counts even if not extinct, so they were determined as “extinct”, and given an arbitrary fitness value of -1. One strain (Matα, AIM in SD-Gly) showed no read counts in all time points (except for the first one) and it was determined as “absent” (see Table 4).
Table 3. filtering offspring data based on Goodness of Fit (GoF) and appearance in different time points
|
media |
initial # variants |
GoF > 0.5 |
% from initial |
tp1 > 100, or found in at least 3 tp |
% from GoF > 0.5 |
% from initial |
|
Glucose + 1.2M Sorbitol |
2694 |
2264 |
84% |
1707 |
75% |
63% |
|
Glucose |
3201 |
3028 |
95% |
2344 |
77% |
73% |
|
Glycerol |
3082 |
3007 |
98% |
2261 |
75% |
73% |
TAble 4. filtering haploids data based on Goodness of Fit (GoF) and appearance in different time points
|
exp |
media |
initial # variants |
tp1 > 200 |
% from initial |
|
Mata |
S |
77 |
71 |
92% |
|
|
U |
77 |
75 |
97% |
|
|
Y |
76 |
61 |
80% |
|
Matα |
S |
45 |
42 |
93% |
|
|
U |
44 |
38 |
86% |
|
|
Y |
44 |
39 |
89% |
Homozygous minor allele content calculation
Relevant strains (104 strains that were used in this study) were selected from the .gvcf file [52]. For each parental strain, +1 was added to the minor allele content if original strain was homozygous for minor allele, or +0.5 was added if original strain was heterozygous. For offspring, three models were used; (i) in the dominance model, +1 was added to the minor allele content if the two original parental strains were homozygous for minor allele, +0.5 was added if one strain was heterozygous and the other was homozygous for minor allele, and +0.25 was added in both parental strains were heterozygous for the minor allele. (ii) co-dominance model: +1, +0.75 and +0.5 were added to the minor allele content respectively, (iii) recessive: +1, +1, +0.75 were added to the minor allele content respectively. Those numbers take into consideration both the chances of the offspring to be either homozygous or heterozygous for the minor allele, as well as the additive value of the genotype.
Calculating mitochondrial based GD
Gene list of 250 mito-nuclear genes were downloaded from the Saccharomyces genome database website (SGD) [81]. Eight additional mitochondrial genes were found in the 1,011 strains website. For each pair of strains, pairwise sequence alignment was performed using python Bio package (Bio.pairwise2.globalxx). For each gene, for each pair of strains #SNPs/total length of gene was calculated and saved. Final mitochondrial GD was calculated as the sum of all SNPs frequencies from all genes, for each pair of strains. all genes sequences were obtained from Peter et al. [52].
Modeling fitness inheritance
Two haploids genome containing a single allele were made (“A” or “a”). fitness was assigned to be 1 in case of the upper case allele and 0 in the case of the lower case allele. The 4 offspring diploid genomes are thus “AA”, “Aa”, “aA” and “aa”. GD was calculated based on the haploid genomes and is 0 when both parents have the same allele and 1 otherwise. Parental average fitness was calculated as the average of parents’ fitness. in addition, diploid offspring fitness was calculated based on the Dominance (heterozygous is as fit as the homozygous “A” variant) or the Co-dominance (heterozygous is half as fit as the homozygous “A” variant) (Table 5).
Table 5. Fitness and GD values of parents and offspring based on the Dominance and Co-dominance models
|
Mata |
Matα |
Mata fit |
Matα fit |
avg fit |
parents GD |
Dominance |
Co-dominance |
|
A |
A |
1 |
1 |
1 |
0 |
1 |
1 |
|
A |
a |
1 |
0 |
0.5 |
1 |
1 |
0.5 |
|
a |
A |
0 |
1 |
0.5 |
1 |
1 |
0.5 |
|
a |
a |
0 |
0 |
0 |
0 |
0 |
0 |
BFG efficiency
Chosen strains (Table S2) were used for measurement and calculation of BFG efficiency. Strains were taken out from -80°C and grown at 30°C overnight in YPD while shaking. Cells were diluted 1:1000 and grown at 30°C overnight in SD-Glu while shaking. OD was measured using plate reader (infinite 200, Tecan), and all strains were diluted to equal OD values. Matα strains were inoculated into the corresponding row of 96-well plates while that Mata strains were added to the columns of each plate and allowed to mate for 20 hours at 25°C without shaking in SD-Glu supplemented with 10ug/ml Doxycycline.. In both plates the Matα strains were inoculated into the corresponding row of the plates while that Mata strains were added to the columns of each plate and allowed to mate for 20 hours at 25°C without shaking. After 20 hours mating, cells were diluted 1:120 into fresh SD-Glu supplemented with 10ug/ml doxycycline, Hyg and NAT, and continued growing for one day. Cells were then diluted 1:120 into SD-Glu + Hyg + NAT (without Doxy) and grown for one day. After 5 of 1:120 daily dilutions days (~30 generations), plates were FACS analyzed to determine the offspring ratio in each culture. Offspring were scored using GFP/mCherry gating. In most wells offspring were found to reach a proportion of over 90% of the culture. One row and one column had no offspring, indicating a problem with that strain (data not shown). At this point, DNA was extracted by boiling the cells in 20mM NaOH for 15 minutes and library were constructed with the most upstream and downstream primers (Figure S2 primers A + G, and Table1 in Materials & Methods). BFG efficiency was calculated by dividing the number of reads with fused barcodes with number of reads with original barcodes for each pair.
Pairwise mating for determining mating efficiency
Relevant strains (see Table S2) were taken from -80°C and were grown for 48 hours at 30°C in YPD while shaking. Cells were diluted 1:1000 into SD-Glu or SD-Gly and incubated overnight in 30°C while shaking. Cells were diluted 1:50 and allowed growing for a couple of hours to reach mid-log. OD was measured using a plate reader (infinity 200, Tecan). Two strains (one Mata and one Matα strain) were inoculated into the same well such that their cell number is equal (~1E7 cells). Mating was executed for 20 hours in 25°C without shaking. All mating reactions were done in 3 repetitions.
Mating efficiency for each pair of strains was determined using FACS (Attune) using 96-well plate module. Cultures were diluted 1:200 before FACS into SD with 5mM EDTA. Each culture was divided into 3 populations based on GFP/mCherry ratio; high GFP and low mCherry were considered Mata cells, high mCherry and low GFP were the Matα cells, and high mCherry and GFP are the offspring. Mating efficiency (ME) was calculated based on the following formula: . Mating efficiency was calculated as the number of offspring divided by the minimum of Mata and Matα number of cells.
Acknowledgements
We thank the Pilpel lab for extensive feedback and discussions. We thank Naama Barkai for useful comments. We thank Frederick Roth lab and Dayag Sheykhkarimli from Toronto university for sharing the BFG system with us, and for advices in the cloning process.
All cloning, but adding the BFG into the construct, was done by Yoav Peled and the Structural Proteomics unit in Weizmann Institute of Science. FACS sorting was done with Tomer Meir Salame in the FACS unit at the Weizmann Institute of Science. Sequencing was performed at the Crown Institute of Genomics, G-INCPM at Weizmann Institute of Science.
We thank the Israel Science Foundation, the Yeda-Sela Foundation at the Weizmann Institute, and the Minerva Foundation for grant support. YP is the Ben- May Professorial Chair and a Kimmel Investigator.
- A. . Moore, “Sexual selection inNauphoeta cinerea: Inherited mating preference?,” Behav Genet, vol. 19, pp. 717–724, 1989.
- N. F. Silva and D. R. Goring, “Mechanisms of self-incompatibility in flowering plants,” Cell. Mol. Life Sci., vol. 58, no. 14, pp. 1988–2007, 2001, doi: 10.1007/PL00000832.
- D. J. Merrell, “Selective Mating in Drosophila Melanogaster,” Genetics, vol. 34, no. 4, p. 370, Jul. 1949, doi: 10.1093/GENETICS/34.4.370.
- J. Yanai and G. E. McClearn, “Assortative mating in mice. I. Female mating preference,” Behav. Genet., vol. 2, no. 2–3, pp. 173–183, Jun. 1972, doi: 10.1007/BF01065688/METRICS.
- P. A. Bedinger, A. K. Broz, A. Tovar-Mendez, and B. McClure, “Pollen-Pistil Interactions and Their Role in Mate Selection,” Plant Physiol., vol. 173, no. 1, pp. 79–90, Jan. 2017, doi: 10.1104/PP.16.01286.
- N. Burley, “Parental investment, mate choice, and mate quality.,” Proc. Natl. Acad. Sci. U. S. A., vol. 74, no. 8, p. 3476, 1977, doi: 10.1073/PNAS.74.8.3476.
- B. P. S. Nieuwenhuis and D. K. Aanen, “Sexual selection in fungi,” J. Evol. Biol., vol. 25, no. 12, pp. 2397–2411, Dec. 2012, doi: 10.1111/JEB.12017.
- M. J. Ryan, “Female Mate Choice in a Neotropical Frog,” Science (80-. )., vol. 209, no. 4455, pp. 523–525, Jul. 1980, doi: 10.1126/SCIENCE.209.4455.523.
- M. Itzkowitz, “Female mate choice in the pupfish, Cyprinodon variegatus,” Behav. Processes, vol. 3, no. 1, pp. 1–8, May 1978, doi: 10.1016/0376-6357(78)90025-6.
- F. A. Beach and B. J. Leboeuf, “Coital behavior in dogs. 1. Preferential mating in the bitch.,” Anim. Behav., vol. 15, no. 4, pp. 546–58, Oct. 1967.
- L. Brown, “Patterns of female choice in mottled sculpins (Cottidae, teleostei),” Anim. Behav., vol. 29, no. 2, pp. 375–382, May 1981, doi: 10.1016/S0003-3472(81)80096-6.
- M. Perrone, “Mate size and breeding success in a monogamous cichlid fish,” Environ. Biol. Fishes, vol. 3, no. 2, pp. 193–201, Jun. 1978, doi: 10.1007/BF00691943/METRICS.
- N. Burley and N. Moran, “The significance of age and reproductive experience in the mate preferences of feral pigeons, Columba livia,” Anim. Behav., vol. 27, no. PART 3, pp. 686–698, Aug. 1979, doi: 10.1016/0003-3472(79)90005-8.
- R. . Thornhill, “Sexual selection and nuptial feeding behavior in Bittacus apicalis (Insecta: Mecoptera),” Am. Nat., 1976.
- C. Packer, “Inter-troop transfer and inbreeding avoidance in Papio anubis,” Anim. Behav., vol. 27, no. PART 1, pp. 1–36, Feb. 1979, doi: 10.1016/0003-3472(79)90126-X.
- P. Bateson, “Sexual imprinting and optimal outbreeding,” Nat. 1978 2735664, vol. 273, no. 5664, pp. 659–660, 1978, doi: 10.1038/273659a0.
- J. Y. Leu and A. W. Murray, “Experimental Evolution of Mating Discrimination in Budding Yeast,” Curr. Biol., vol. 16, no. 3, pp. 280–286, Feb. 2006, doi: 10.1016/J.CUB.2005.12.028.
- M. Olsson and T. Madsen, “Female choice on male quantitative traits in lizards - why is it so rare?,” Behav. Ecol. Sociobiol., vol. 36, no. 3, pp. 179–184, Mar. 1995, doi: 10.1007/BF00177794/METRICS.
- E. Hine, S. Lachish, M. Higgie, and M. W. Blows, “Positive genetic correlation between female preference and offspring fitness,” Proc. R. Soc. London. Ser. B Biol. Sci., vol. 269, no. 1506, pp. 2215–2219, Nov. 2002, doi: 10.1098/RSPB.2002.2149.
- J. D. Reynolds and M. R. Gross, “Female mate preference enhances offspring growth and reproduction in a fish, Poecilia reticulata,” Proc. R. Soc. London. Ser. B Biol. Sci., vol. 250, no. 1327, pp. 57–62, Oct. 1992, doi: 10.1098/RSPB.1992.0130.
- Y. S. Aulchenko et al., “Predicting human height by Victorian and genomic methods,” Eur. J. Hum. Genet. 2009 178, vol. 17, no. 8, pp. 1070–1075, Feb. 2009, doi: 10.1038/ejhg.2009.5.
- B. P. McEvoy and P. M. Visscher, “Genetics of human height,” Econ. Hum. Biol., vol. 7, no. 3, pp. 294–306, Dec. 2009, doi: 10.1016/J.EHB.2009.09.005.
- K. Larsson, “Inheritance of body size in the Barnacle Goose under different environmental conditions,” J. Evol. Biol., vol. 6, no. 2, pp. 195–208, Mar. 1993, doi: 10.1046/J.1420-9101.1993.6020195.X.
- C. K. Chai, “Analysis of Quantitative Inheritance of Body Size in Mice. I. Hybridization and Maternal Influence,” Genetics, vol. 41, no. 2, p. 157, Mar. 1956, doi: 10.1093/GENETICS/41.2.157.
- T. F. C. MacKay, E. A. Stone, and J. F. Ayroles, “The genetics of quantitative traits: Challenges and prospects,” Nature Reviews Genetics, vol. 10, no. 8. Nature Publishing Group, pp. 565–577, Aug-2009, doi: 10.1038/nrg2612.
- S. Treusch, F. W. Albert, J. S. Bloom, I. E. Kotenko, and L. Kruglyak, “Genetic Mapping of MAPK-Mediated Complex Traits Across S. cerevisiae,” PLOS Genet., vol. 11, no. 1, p. e1004913, 2015, doi: 10.1371/JOURNAL.PGEN.1004913.
- J. S. Bloom et al., “Rare variants contribute disproportionately to quantitative trait variation in yeast,” Elife, vol. 8, Oct. 2019, doi: 10.7554/ELIFE.49212.
- J. M. Rhode and M. B. Cruzan, “Contributions of Heterosis and Epistasis to Hybrid Fitness,” 2005.
- D. Hedgecock, D. J. McGoldrick, and B. L. Bayne, “Hybrid vigor in Pacific oysters: an experimental approach using crosses among inbred lines,” Aquaculture, vol. 137, no. 1–4, pp. 285–298, Dec. 1995, doi: 10.1016/0044-8486(95)01105-6.
- D. Hedgecock and J. P. Davis, “Heterosis for yield and crossbreeding of the Pacific oyster Crassostrea gigas,” Aquaculture, vol. 272, no. SUPPL. 1, pp. S17–S29, Jan. 2007, doi: 10.1016/j.aquaculture.2007.07.226.
- D. Bar-Zvi, O. Lupo, A. A. Levy, and N. Barkai, “Hybrid vigor: The best of both parents, or a genomic clash?,” Current Opinion in Systems Biology, vol. 6. Elsevier Ltd, pp. 22–27, 01-Dec-2017, doi: 10.1016/j.coisb.2017.08.004.
- R. H. Herbst et al., “Heterosis as a consequence of regulatory incompatibility,” BMC Biol., vol. 15, no. 1, p. 38, Dec. 2017, doi: 10.1186/s12915-017-0373-7.
- J. P. Bernardes, R. B. Stelkens, and D. Greig, “Heterosis in hybrids within and between yeast species,” J. Evol. Biol., vol. 30, no. 3, pp. 538–548, Mar. 2017, doi: 10.1111/jeb.13023.
- R. Shapira, T. Levy, S. Shaked, E. Fridman, and L. David, “Extensive heterosis in growth of yeast hybrids is explained by a combination of genetic models,” Heredity (Edinb)., vol. 113, 2014, doi: 10.1038/hdy.2014.33.
- R. Shapira and L. David, “Genes with a Combination of Over-Dominant and Epistatic Effects Underlie Heterosis in Growth of Saccharomyces cerevisiae at High Temperature,” Front. Genet., vol. 7, May 2016, doi: 10.3389/fgene.2016.00072.
- J. A. Zeh and D. W. Zeh, “The evolution of polyandry I: Intragenomic conflict and genetic incompatibility,” Proc. R. Soc. B Biol. Sci., vol. 263, no. 1377, pp. 1711–1717, 1996, doi: 10.1098/rspb.1996.0250.
- M. J. E. Charpentier, J. C. Crawford, M. Boulet, and C. M. Drea, “Message ‘scent’: lemurs detect the genetic relatedness and quality of conspecifics via olfactory cues,” Anim. Behav., vol. 80, no. 1, pp. 101–108, Jul. 2010, doi: 10.1016/J.ANBEHAV.2010.04.005.
- M. L. THOMAS and L. W. SIMMONS, “Crickets detect the genetic similarity of mating partners via cuticular hydrocarbons,” J. Evol. Biol., vol. 24, no. 8, pp. 1793–1800, Aug. 2011, doi: 10.1111/j.1420-9101.2011.02319.x.
- A. S. Hirao, “Kinship between parents reduces offspring fitness in a natural population of Rhododendron brachycarpum,” Ann. Bot., vol. 105, no. 4, pp. 637–646, Apr. 2010, doi: 10.1093/aob/mcq018.
- S. C. Roberts and L. M. Gosling, “Genetic similarity and quality interact in mate choice decisions by female mice,” Nat. Genet. 2003 351, vol. 35, no. 1, pp. 103–106, Aug. 2003, doi: 10.1038/ng1231.
- M. Lynch, “THE GENETIC INTERPRETATION OF INBREEDING DEPRESSION AND OUTBREEDING DEPRESSION,” Evolution (N. Y)., vol. 45, no. 3, pp. 622–629, May 1991, doi: 10.1111/j.1558-5646.1991.tb04333.x.
- K. M. Hufford and S. J. Mazer, “Plant ecotypes: genetic differentiation in the age of ecological restoration,” Trends Ecol. Evol., vol. 18, no. 3, pp. 147–155, Mar. 2003, doi: 10.1016/S0169-5347(03)00002-8.
- M. V. Price and N. M. Waser, “Pollen dispersal and optimal outcrossing in Delphinium nelsoni,” Nat. 1979 2775694, vol. 277, no. 5694, pp. 294–297, 1979, doi: 10.1038/277294a0.
- C. Pélabon, M. L. Carlson, T. E. Hansen, and W. S. Armbruster, “Effects of crossing distance on offspring fitness and developmental stability in Dalechampia scandens (Euphorbiaceae),” Am. J. Bot., vol. 92, no. 5, pp. 842–851, May 2005, doi: 10.3732/AJB.92.5.842.
- X. Wei and J. Zhang, “The optimal mating distance resulting from heterosis and genetic incompatibility,” Sci. Adv., vol. 4, no. 11, p. eaau5518, Nov. 2018, doi: 10.1126/sciadv.aau5518.
- A. M. Neiman, “Sporulation in the Budding Yeast Saccharomyces cerevisiae,” Genetics, vol. 189, no. 3, pp. 737–765, Nov. 2011, doi: 10.1534/GENETICS.111.127126.
- M. Knop, “Evolution of the hemiascomycete yeasts: on life styles and the importance of inbreeding,” BioEssays, vol. 28, no. 7, pp. 696–708, Jul. 2006, doi: 10.1002/bies.20435.
- I. Herskowitz, “Life cycle of the budding yeast Saccharomyces cerevisiae.,” Microbiol. Rev., vol. 52, no. 4, pp. 536–53, Dec. 1988.
- D. M. Ruderfer, S. C. Pratt, H. S. Seidel, and L. Kruglyak, “Population genomic analysis of outcrossing and recombination in yeast,” Nat. Genet. 2006 389, vol. 38, no. 9, pp. 1077–1081, Aug. 2006, doi: 10.1038/ng1859.
- I. Stefanini, L. Dapporto, L. Berná, M. Polsinelli, S. Turillazzi, and D. Cavalieri, “Social wasps are a Saccharomyces mating nest,” Proc. Natl. Acad. Sci., vol. 113, no. 8, pp. 2247–2251, Feb. 2016, doi: 10.1073/pnas.1516453113.
- H. A. Murphy and C. W. Zeyl, “Yeast sex: surprisingly high rates of outcrossing between asci,” PLoS One, vol. 5, no. 5, May 2010, doi: 10.1371/JOURNAL.PONE.0010461.
- J. Peter et al., “Genome evolution across 1,011 Saccharomyces cerevisiae isolates,” Nature, vol. 556, no. 7701, pp. 339–344, Apr. 2018, doi: 10.1038/s41586-018-0030-5.
- C. T. Hittinger, “Saccharomyces diversity and evolution: a budding model genus,” Trends Genet., vol. 29, no. 5, pp. 309–317, May 2013, doi: 10.1016/j.tig.2013.01.002.
- P. M. Magwene, Ö. Kayikçi, J. A. Granek, J. M. Reininga, Z. Scholl, and D. Murray, “Outcrossing, mitotic recombination, and life-history trade-offs shape genome evolution in Saccharomyces cerevisiae,” Proc. Natl. Acad. Sci. U. S. A., vol. 108, no. 5, pp. 1987–1992, Feb. 2011, doi: 10.1073/PNAS.1012544108.
- A. W. McClure, K. C. Jacobs, T. R. Zyla, and D. J. Lew, “Mating in wild yeast: Delayed interest in sex after spore germination,” Mol. Biol. Cell, vol. 29, no. 26, pp. 3119–3127, Dec. 2018, doi: 10.1091/MBC.E18-08-0528/ASSET/IMAGES/LARGE/MBC-29-3119-G005.JPEG.
- I. J. Tsai, D. Bensasson, A. Burt, and V. Koufopanou, “Population genomics of the wild yeast Saccharomyces paradoxus: Quantifying the life cycle,” Proc. Natl. Acad. Sci. U. S. A., vol. 105, no. 12, pp. 4957–4962, 2008, doi: 10.1073/pnas.0707314105.
- M. J. McDonald, D. P. Rice, and M. M. Desai, “Sex speeds adaptation by altering the dynamics of molecular evolution,” Nature, vol. 531, no. 7593, pp. 233–236, Mar. 2016, doi: 10.1038/nature17143.
- M. S. Cappello, G. Bleve, F. Grieco, F. Dellaglio, and G. Zacheo, “Characterization of Saccharomyces cerevisiae strains isolated from must of grape grown in experimental vineyard,” J. Appl. Microbiol., vol. 97, no. 6, pp. 1274–1280, 2004, doi: 10.1111/J.1365-2672.2004.02412.X.
- B. Gallone et al., “Interspecific hybridization facilitates niche adaptation in beer yeast,” Nat. Ecol. Evol., vol. 3, no. 11, pp. 1562–1575, Nov. 2019, doi: 10.1038/S41559-019-0997-9.
- C. Eberlein et al., “Hybridization is a recurrent evolutionary stimulus in wild yeast speciation,” Nat. Commun., vol. 10, no. 1, 2019, doi: 10.1038/s41467-019-08809-7.
- C. L. Jackson and L. H. Hartwell, “Courtship in Saccharomyces cerevisiae: an early cell-cell interaction during mating.,” Mol. Cell. Biol., vol. 10, no. 5, pp. 2202–13, May 1990, doi: 10.1128/MCB.10.5.2202.
- C. L. Jackson and L. H. Hartwell, “Courtship in S. cerevisiae: Both cell types choose mating partners by responding to the strongest pheromone signal,” Cell, vol. 63, no. 5, pp. 1039–1051, Nov. 1990, doi: 10.1016/0092-8674(90)90507-B.
- A. Banderas, M. Koltai, A. Anders, and V. Sourjik, “Sensory input attenuation allows predictive sexual response in yeast,” Nat. Commun., vol. 7, p. 12590, Aug. 2016, doi: 10.1038/ncomms12590.
- N. Yachie et al., “Pooled-matrix protein interaction screens using Barcode Fusion Genetics.,” Mol. Syst. Biol., vol. 12, no. 4, p. 863, Apr. 2016, doi: 10.15252/MSB.20156660.
- J. E. Haber, “MATING-TYPE GENE SWITCHING IN SACCHAROMYCES CEREVISIAE,” 1998.
- J. E. Haber, “Mating-Type Genes and MAT Switching in Saccharomyces cerevisiae,” Genetics, vol. 191, no. 1, pp. 33–64, May 2012, doi: 10.1534/genetics.111.134577.
- A. N. Nguyen Ba et al., “Barcoded Bulk QTL mapping reveals highly polygenic and epistatic architecture of complex traits in yeast,” Elife, vol. 11, 2022, doi: 10.7554/ELIFE.73983.
- S. F. Levy, J. R. Blundell, S. Venkataram, D. A. Petrov, D. S. Fisher, and G. Sherlock, “Quantitative evolutionary dynamics using high-resolution lineage tracking,” Nature, vol. 519, no. 7542, pp. 181–186, Mar. 2015, doi: 10.1038/nature14279.
- L. Partridge, “Mate choice increases a component of offspring fitness in fruit flies,” Nat. 1980 2835744, vol. 283, no. 5744, pp. 290–291, 1980, doi: 10.1038/283290a0.
- L. W. Simmons, “Female choice contributes to offspring fitness in the field cricket, Gryllus bimaculatus (De Geer),” Behav. Ecol. Sociobiol., vol. 21, no. 5, pp. 313–321, Nov. 1987, doi: 10.1007/BF00299969/METRICS.
- T. Kido et al., “Are minor alleles more likely to be risk alleles?,” BMC Med. Genomics, vol. 11, no. 1, pp. 1–11, Jan. 2018, doi: 10.1186/S12920-018-0322-5/FIGURES/4.
- P. M. Adams et al., “Assessment of the genetic variance of late-onset Alzheimer’s disease,” Neurobiol. Aging, vol. 41, pp. 200.e13-200.e20, May 2016, doi: 10.1016/J.NEUROBIOLAGING.2016.02.024.
- J. Heitman, “Evolution of sexual reproduction: a view from the Fungal Kingdom supports an evolutionary epoch with sex before sexes,” doi: 10.1016/j.fbr.2015.08.002.
- V. L. Koumandou, B. Wickstead, M. L. Ginger, M. Van Der Giezen, J. B. Dacks, and M. C. Field, “Molecular paleontology and complexity in the last eukaryotic common ancestor,” http://dx.doi.org/10.3109/10409238.2013.821444, vol. 48, no. 4, pp. 373–396, Jul. 2013, doi: 10.3109/10409238.2013.821444.
- M. L. Berbee, T. Y. James, and C. Strullu-Derrien, “Early Diverging Fungi: Diversity and Impact at the Dawn of Terrestrial Life,” https://doi.org/10.1146/annurev-micro-030117-020324, vol. 71, pp. 41–60, Sep. 2017, doi: 10.1146/ANNUREV-MICRO-030117-020324.
- T. Y. James, J. E. Stajich, C. T. Hittinger, and A. Rokas, “Toward a Fully Resolved Fungal Tree of Life,” Annu. Rev. Microbiol., 2020, doi: 10.1146/annurev-micro-022020.
- K. A. Curran, N. J. Morse, K. A. Markham, A. M. Wagman, A. Gupta, and H. S. Alper, “Short Synthetic Terminators for Improved Heterologous Gene Expression in Yeast,” ACS Synth. Biol., vol. 4, no. 7, pp. 824–832, Jul. 2015, doi: 10.1021/SB5003357.
- R. D. Gietz and R. A. Woods, “Transformation of yeast by lithium acetate/single-stranded carrier DNA/polyethylene glycol method.,” Methods Enzymol., vol. 350, pp. 87–96, 2002.
- M. Martin, “Cutadapt removes adapter sequences from high-throughput sequencing reads,” EMBnet.journal, vol. 17, no. 1, pp. 10–12, May 2011, doi: 10.14806/EJ.17.1.200.
- I. Frumkin et al., “Gene Architectures that Minimize Cost of Gene Expression,” Mol. Cell, 2017, doi: 10.1016/j.molcel.2016.11.007.
- J. M. Cherry et al., “Saccharomyces Genome Database: The genomics resource of budding yeast,” Nucleic Acids Res., vol. 40, no. D1, Jan. 2012, doi: 10.1093/nar/gkr1029.
There is NO Competing Interest.
- Suptables.xlsx
Supplementary Tables 1-2