Dramatically diverse Schizosaccharomyces pombe wtf meiotic drivers all display high gamete-killing efficiency

Summary

Meiotic drivers are selfish alleles that can force their transmission into more than 50% of the viable gametes made by heterozygotes. Meiotic drivers are known to cause infertility in a diverse range of eukaryotes and are predicted to affect the evolution of genome structure and meiosis. The wtf gene family of Schizosaccharomyces pombe includes both meiotic drivers and drive suppressors and thus offers a tractable model organism to study drive systems. Currently, only a handful of wtf genes have been functionally characterized and those genes only partially reflect the diversity of the wtf gene family. In this work, we functionally test 22 additional wtf genes for meiotic drive phenotypes. We identify eight new drivers that share between 30–90% amino acid identity with previously characterized drivers. Despite the vast divergence between these genes, they generally drive into >85% of gametes when heterozygous. We also identify three wtf genes that suppress other wtf drivers, including two that also act as autonomous drivers. Additionally, we find that wtf genes do not underlie a weak (64% allele transmission) meiotic driver on chromosome 1. Finally, we find that some Wtf proteins have expression or localization patterns that are distinct from the poison and antidote proteins encoded by drivers and suppressors, suggesting some wtf genes may have non-meiotic drive functions. Overall, this work expands our understanding of the wtf gene family and the burden selfish driver genes impose on S. pombe.

Keywords:

Genetic loci – Meiosis – Schizosaccharomyces pombe – Fungal spores – Polymerase chain reaction – Toxins – Suppressor genes – Antidotes

Introduction

During meiosis, diploid cells divide to produce haploid gametes (e.g. sperm). This process is generally fair in that each parental allele of a gene is represented equally in the gametes [1]. However, many eukaryotic genomes contain ‘selfish’ elements that bias their own transmission such that they are overrepresented in the viable gametes [2–4]. These loci are known as meiotic drivers and they can both directly and indirectly reduce fitness through a number of mechanisms (reviewed in [5, 6]).

Meiotic drivers can be in conflict with unlinked genes that do not gain a transmission advantage but must bear the fitness burdens often imposed by these selfish elements [7–10]. This genetic conflict is thought to favor the emergence of unlinked allele variants that can suppress the effects of drivers. In turn, variants of drivers that can evade this suppression will have a selective advantage [4, 8, 11]. The conflict between drivers and suppressors is predicted to lead to the rapid evolution of both sets of genes. This is analogous to the genetic arms race observed between viruses and host immune systems [12, 13]. Arms races caused by meiotic drivers are predicted to affect the evolution of gametogenesis and genome structure [14, 15].

Empirical analyses of meiotic drive/suppressor systems have traditionally been limited by the complexity of most known drive systems. However, the wtf (with tf) gene family of the fission yeast Schizosaccharomyces pombe offers an opportunity to study meiotic drive in a highly tractable model system. Some wtf genes are meiotic drivers and others act as suppressors of drive [16–18]. Characterized wtf drivers kill the meiotic products (spores) that do not inherit them from a wtf+/wtf- heterozygous diploid by producing two proteins, Wtf^poison and Wtf^antidote. These proteins are encoded on largely overlapping transcripts, but the antidote message includes a 5’ exon (exon 1) not found in the poison message (Fig 1A). The poison protein localizes throughout the spore sac (ascus) and acts on every spore in an indiscriminate fashion. The antidote protein, however, is highly enriched within the wtf+ spores, rescuing them from destruction by the poison [16, 17]. Similarly, the known drive-suppressing wtf genes produce a protein that mimics the antidote of a driver, rescuing the spores that inherit the suppressor from the driver’s poison [18].

The wtf family contains 25 genes (including pseudogenes) in the S. pombe reference strain, L972 (referred to here as ‘Sp’) [19, 20]. This gene family is rapidly evolving and other S. pombe isolates in which the wtf genes have been assembled contain up to 38 members [17, 21]. Currently, only a handful of these genes have been functionally characterized to be either drivers or suppressors [16–18]. Although the characterized genes have provided mechanistic insight, they do not represent the extensive diversity within the wtf gene family. It is currently unclear if the sequence diversity of wtf genes corresponds with phenotypic diversity or if some wtf genes could have a function outside of drive.

To aid in the functional characterization of the wtf gene family, we previously sorted them into three classes. Class 1 genes are predicted to be meiotic drivers (Fig 1A). These genes contain two potential transcriptional and translational start sites that could be used to encode distinct poison and antidote proteins. Class 2 genes are predicted drive suppressors that appear to encode only Wtf^antidote proteins (Fig 1B). The class 1 and class 2 genes are rapidly evolving, largely due to extensive ectopic gene conversion that has shuffled sequence variation and muddled orthology within the family [17, 21]. Class 3 includes four diverse genes (wtf7, wtf11, wtf14, and wtf15) that have no predicted function, but are grouped together only because they are unlike all other wtf genes (Fig 1C). These genes also have two potential translational start sites, but unlike class 1 genes, they produce only one major transcript type [22]. Finally, the class 3 wtf genes have not undergone ectopic gene conversion, so distinct S. pombe isolates have clear orthologs of each class 3 gene [21].

More extensive characterization of wtf genes is required to understand the functional breadth of these genes and their potential evolutionary impact on S. pombe. Here we use a combination of genetics and cell biology approaches to functionally analyze 22 previously untested wtf genes. We find that a tremendous breadth of Wtf protein sequences can build highly efficient poison-antidote drivers and drive suppressors. Moreover, we discover that some wtf genes can simultaneously act as drivers and as suppressors of other drive loci. Surprisingly, we also find that wtf genes do not underlie a previously identified but unmapped locus on chromosome 1 that causes an allele transmission bias into viable spores. Finally, we discover that the potential cellular roles of wtf family members extend beyond gametogenesis. This work demonstrates that the landscape of drive systems in S. pombe is even richer than previously anticipated and underscores the potential of this simple model system to reveal unexpected insights about meiotic drive.

Results

Class 1 contains autonomous meiotic drivers

To test the ability of class 1 and class 2 wtf genes to drive autonomously, we cloned untested genes representing unique subclasses from four isolates that largely reflect the range of S. pombe diversity: Sp, S. kambucha (Sk), FY29033, and CBS5557 (S1 Fig) [23, 24]. We then assayed the ability of these wtf genes to drive in one or more ectopic strain backgrounds (e.g. testing Sk wtf genes in Sp). We used ectopic strain backgrounds because drive occurs in heterozygotes and it is simpler to insert genes into S. pombe than it is to delete the repetitive wtf genes at their endogenous loci. Additionally, drivers may be less likely to be suppressed in an ectopic background [25].

To do these analyses, we first linked each wtf gene under the control of its endogenous promoter to a selectable drug resistance marker (kanMX4 or hphMX6) and integrated them into the ade6 locus (of either Sp and/or Sk). We then crossed each haploid carrying a wtf gene of interest to a wild-type strain from the same background to generate hemizygous (wtf::drug^R/ade6+) diploid strains. We induced the hemizygous diploids to undergo meiosis and then assayed the presence of each wtf gene of interest in the viable spore population using the linked selectable markers. In the absence of suppressors, genes capable of autonomous drive are expected to be overrepresented in the viable progeny (>50%).

Most genes within class 1 have six exons, including all previously described wtf drivers. Class 1 also includes two 5-exon wtf genes (S1 Fig) [21]. We tested ten 6-exon and two 5-exon wtf genes from class 1 (Fig 2A, diploids 1–16). We found that seven of the 6-exon genes and one 5-exon gene exhibited significant drive in at least one strain background (Fig 2A). These genes are: Sk wtf9, Sk wtf19, Sk wtf30, Sk wtf33, Sp wtf19, FY29033 wtf36, FY29033 wtf18, and FY29033 wtf35. Interestingly, these genes are incredibly diverse and share as little as 30% pairwise amino acid identity (S2A Fig). Moreover, all of the drivers except one (Sp wtf19) drive into >90% of the progeny when tested under the same conditions [16, 18]. These results demonstrate that a remarkably wide range of proteins can execute spore-killing similarly well.

Surprisingly, the 5-exon FY29033 wtf35 meiotic driver lacks sequences homologous to what is exon 4 in the other known 6-exon drivers. In addition, the gene lacks the 7-amino acid repeat that is found in between 1 and 4 copies in the last exon of all other drivers [21]. Despite these differences, the localization of the FY29033 Wtf35 proteins was indistinguishable from that observed in other drivers (S3 Fig, S1 Table, diploid 39). These observations indicate that the aforementioned sequences can be dispensable for Wtf protein localization and drive.

The four class 1 genes that did not exhibit drive in our tests were Sk wtf27, Sk wtf29, Sp wtf23, and CBS5557 wtf23 (Fig 2A, diploids 12–16). CBS5557 wtf23 completely lacks a specific 11-amino acid repeat within exon three that can be found in 2–7 copies in all characterized drivers except FY29033 wtf35, which contains only four amino acids of the repeat [21]. We found no coding sequence features in Sk wtf27, Sk wtf29, or Sp wtf23 that distinguish these genes from the confirmed drivers. However, it is possible that these four genes are not capable of driving or they are not transcribed. It is also possible that these genes are suppressed in the background in which we tested them. We previously saw background-specific suppression of the Sp wtf13 driver, and similar suppression explains why we observed drive of FY29033 wtf18 in the Sk, but not in the Sp background (see below) [18]. Overall, our analyses are consistent with class 1 containing a wide diversity of autonomous meiotic drive genes.

The class 2 wtf genes are either known or predicted antidote-only drive suppressors (Fig 1B, S1 Fig). Like the class 1 genes, class 2 contains both 5-exon and 6-exon genes, although only 6-exon genes have been previously tested [16–18]. Interestingly, class 2 also contains Sp wtf21, which was previously reported to be an essential gene because a heterozygous mutant (Sp wtf21/wtf21Δ) transmitted only the wild-type Sp wtf21 allele to viable gametes [26].

We first tested four class 2 genes (Sk wtf13, Sk wtf23, Sk wtf35, and FY29033 wtf1) using the same approach described above for class 1 genes. As expected for genes predicted to produce only antidotes, we found that none of the genes tested exhibited drive in an ectopic strain background (Fig 2B, diploids 17–20). We also revisited the idea that Sp wtf21 is an essential gene [26]. We found that we could generate a deletion of Sp wtf21 in a haploid strain, indicating the gene is not essential in that strain background. Moreover, we did not observe drive in an Sp wtf21/wtf21Δ heterozygote (Fig 2B, diploid 21).

Finally, we tested if we could observe autonomous meiotic drive by the class 3 wtf genes (wtf7, wtf11, wtf14, and wtf15) in Sp. Unlike the other wtf genes (class 1 and class 2), each natural isolate has a clear ortholog of each class 3 wtf gene [21]. We therefore had to use a different strategy to test if these genes could drive. Instead of introducing the genes into an ectopic strain background, we assayed whether these genes could drive when heterozygous at their endogenous loci. We deleted wtf7 and wtf11 individually, and wtf14 and wtf15 together as they are adjacent to each other. In diploids heterozygous for any of these wtf gene deletions, the wild-type alleles were transmitted to ~50% of the viable spores (S4A Fig). This indicates that these genes cannot drive or that they are suppressed in the Sp background. We also assayed a homozygous deletion strain lacking all of the class 3 genes. We observed no fertility defects, indicating these genes are not required for sexual reproduction (S4B Fig). We also observed no growth defects in haploids lacking some or all of the class 3 genes (S4C Fig).

wtf genes in both class 1 and class 2 can act as suppressors of drive

Previous work identified alleles of wtf18 in Sp and Sk as suppressors of meiotic drive, but we wanted to test if other wtf genes could also act as drive suppressors [18]. Within drivers and their known suppressors, the Wtf^antidote proteins share high levels of amino acid identity with the poisons they neutralize. This similarity may be particularly important within the C-terminus [18]. We used this knowledge to guide our search for other drive suppressors. We found that the DNA sequence in the last three exons of the class 2 gene FY29033 wtf1 shares >99% identity to the sequence in the FY29033 wtf35 driver (Fig 3A). We therefore predicted that FY29033 wtf1 could be a suppressor of the FY29033 wtf35 driver. To test this, we made FY29033 wtf35/FY29033 wtf1 diploids in an Sp background and assayed transmission of both genes into progeny. FY29033 wtf35 was no longer able to drive in the presence of FY29033 wtf1 (Fig 3C, compare diploid 10 to 24). Additionally, FY29033 wtf1 was able to rescue the fertility defect caused by FY29033 wtf35 (Fig 3C). These results demonstrate that FY29033 wtf1 is a suppressor of FY29033 wtf35.

To broaden our search for suppressors, we examined why FY29033 wtf18 showed drive in Sk, but not in Sp. Specifically, we looked for wtf genes in Sp that could work as a suppressor of FY29033 wtf18. We noticed that the last 776 base pairs of FY29033 wtf18 were identical to the Sp wtf13 meiotic driver (Fig 3B). This suggested that the Sp Wtf13^antidote could potentially neutralize the FY29033 Wtf18^poison and vice versa. To test this, we generated FY29033 wtf18::kanMX4/Sp wtf13::hphMX6 heterozygotes in an Sk strain background and assayed their phenotypes. We observed that drive of both genes was suppressed in the heterozygote relative to hemizygotes containing only one of the drivers (Fig 3C, compare diploids 8 and 26 to diploid 27). These results show that wtf drivers can also function as suppressors of each other.

wtf genes are not responsible for meiotic drive of Sk chromosome 1

We previously detected a weak allele transmission bias that favored Sk chromosome 1 in the viable progeny of Sp/Sk hybrid diploids [27]. We reexamined this observation here with hybrid diploids that are heterozygous for Sp and Sk copies of chromosome 1, but are homozygous for Sk chromosomes 2 and 3. These hybrids were also unable to initiate meiotic recombination due to deletion of rec12 (SPO11 homolog), which is required for programmed meiotic double-stranded break formation. The rec12 deletion ensured chromosome-wide linkage on chromosome 1, allowing us to monitor transmission of the unmapped locus into the spore progeny. Consistent with our previous observations, we observed 64.3% of the viable spores generated by the hybrids inherited Sk chromosome 1 (S5A Fig).

The observed transmission bias is consistent with the presence of a driver on Sk chromosome 1. However, it is also possible that Dobzhansky-Muller incompatibilities between Sp chromosome 1 and Sk chromosomes 2 and/or 3 explain the transmission bias [28]. To distinguish these possibilities, we tested a different rec12Δ hybrid diploid that was heterozygous for Sp and Sk chromosome 1, but homozygous for Sp chromosomes 2 and 3. We observed that 70.5% of the viable spores generated by this hybrid inherited Sk chromosome 1 (S5B Fig). This result cannot be explained by incompatibilities because a fully compatible genotype (Sp chromosomes 1–3) is underrepresented. Together, our results indicate that meiotic drive, not genetic incompatibilities underlie this transmission bias.

We postulated that a wtf gene could underlie this drive phenotype. There is only one wtf locus, wtf1, on both Sp and Sk chromosome 1. Surprisingly, the coding sequences for the Sp and Sk alleles are identical and both are predicted to be pseudogenes due to an early in-frame stop codon. This suggested that wtf genes do not underlie the drive of Sk chromosome 1. To directly test this idea, we deleted Sk wtf1 from the first hybrid described above. We found that the transmission of Sk chromosome 1 was not significantly different after deleting Sk wtf1 (S5A Fig), indicating that Sk chromosome 1 encodes a non-wtf meiotic driver.

Class 3 wtf genes are not required for drive of Sk wtf4

We previously analyzed the sequences of the class 3 wtf genes across 57 isolates of S. pombe and found that wtf7, wtf11, and wtf15 show signatures of positive selection [21]. This suggests that selection is favoring novelty and led us to speculate that these genes could be involved in meiotic drive. Given that we failed to observe drive of these class 3 genes (S4A Fig), we thought they could perhaps be modifying drive of other wtf genes. All class 3 wtf genes are linked to drive loci so we predicted that they could facilitate drive, as they would also gain a transmission advantage from a linked driver [8].

We tested this idea by generating diploids hemizygous for the Sk wtf4 meiotic driver (Sk wtf4::kanMX4/ade6+) and lacking all four of the class 3 wtf genes. Sk wtf4 was found in nearly 100% of the viable spores generated by this quadruple deletion mutant (S4D Fig, compare diploid 32 to 33). These results show that wtf7, wtf11, wtf14, and wtf15 are not required to facilitate drive of the Sk wtf4 driver.

Localization of class 3 Sk Wtf proteins in Sp

We next decided to take a more agnostic approach to look for possible functions of Wtf7, Wtf11, Wtf14, and Wtf15 proteins by investigating their expression and localization in cells. We analyzed a published data set from a proteomics meiotic time course study [29] and found that peptides from Wtf11, Wtf14, and Wtf15, but not Wtf7, were all detected during meiosis in at least two replicate experiments (S6 Fig). Wtf11 and Wtf15 were detected in one or more time points taken after the first meiotic division. Interestingly, the levels of Wtf14 remained steady throughout meiosis (S6 Fig). These patterns are both unlike those of class 1 and class 2 Wtf proteins which increase in abundance as meiosis progresses [18, 29].

To assay the localization of the proteins within cells, we tagged each of the class 3 wtf genes (cloned from Sk) with GFP at the C-terminus and integrated them at the ade6 locus of Sp. The tagged genes were under the control of their endogenous promoters. We then separately imaged haploid and diploid cells grown in rich media during logarithmic cell growth. We also imaged the haploid and diploid cells again in stationary phase. Finally, we imaged diploid cells undergoing meiosis and the products of meiosis (tetrads).

We were not able to detect any GFP fluorescence under any conditions in cells containing the wtf11-GFP allele (S7 Fig) [29]. The reasons for this are unclear, given the proteomics data indicates the protein is expressed in meiosis. We also did not detect Wtf7-GFP or Wtf15-GFP in vegetatively growing haploids or diploids (S8 Fig). However, we did observe GFP fluorescence in tetrads produced by diploids heterozygous for wtf7-GFP and tetrads produced by wtf15-GFP heterozygotes (Fig 4 and S9 Fig). This late expression of Wtf15-GFP is consistent with the proteomics data set (Wtf7 was not detected by proteomics) [29]. In tetrads generated by diploids with one GFP-tagged copy of Wtf7 or Wtf15, we observed that the signal was greatly enriched in two of the four spores (Fig 4). We speculate that the two spores with bright GFP signal are those that inherited the tagged allele. This would suggest that wtf7-GFP and wtf15-GFP are both expressed after spore individualization, similar to the antidotes of wtf drivers and suppressors [16, 18]. Wtf15-GFP exhibited a diffuse localization pattern that largely filled the spores (Fig 4B and S9B Fig). The localization pattern of Wtf7-GFP varied in different spores. In some instances, Wtf7-GFP made a ring-like structure next to Nsp1-mCherry (nucleoporin marker) (Fig 4A and S9A Fig). However, we also observed other cases where Wtf7-GFP seemed to be clustered within the spore (Fig 4A).

Unlike all other Wtf proteins, we saw expression of Wtf14-GFP in vegetative cells. We saw that Wtf14-GFP colocalized with the endoplasmic reticulum (ER) marker pbip1::mCherry-AHDL in a punctate pattern in vegetative haploid and diploid cells (Fig 5A and 5B, S10 Fig) [30]. This localization became more pronounced in cells during stationary phase (i.e. after ~16 hours growth in rich media). These observations are consistent with previous work showing that Sp Wtf14-YFP expressed from a repressible promoter localizes to the ER [31] (Fig 5A and 5B, S10 Fig). We also saw ER localization of Wtf14-GFP in two of the four spores generated by diploids heterozygous for wtf14-GFP (Fig 5C). Similar to what has been observed in Saccharomyces cerevisiae, ER proteins seem to be generated de novo in spores [32]. This localization pattern likely reflects expression of wtf14-GFP in the two spores that inherited the tagged wtf14 allele.

Discussion

S. pombe genomes harbor a surprising number of active drive genes

Overall, we identified eight new wtf genes capable of meiotic drive. The S. pombe isolates with assembled wtf sequences contain between 4–14 predicted drivers [17, 21]. Our current work shows that many of these predicted drive genes are able to drive in the right genetic context. This work expands the rapidly growing number of bona fide drive genes identified in recent years [4, 16–18, 33–43].

Moreover, our results highlight that one species can house a surprising number of functional meiotic drive genes. This is not unique to S. pombe as several species are known to harbor multiple gamete-killing drivers. These species include plants, fungi, insects, and animals [16–18, 27, 34–38, 41, 42, 44–47]. For example, in Drosophila simulans, there are three genetically distinct X-linked male drive systems. Two of these systems are paralogous, whereas the third arose independently [46, 48–50]. Additionally, species can also house multiple copies of ‘true meiotic drive loci’ that bias chromosome segregation as well. Maize, for example, contains different numbers and sizes of heterochromatic regions (knobs) that have preferential segregation during female meiosis [51–53]. Similarly, drive of more than one centromere has been observed in species of Mus [54, 55].

In S. pombe, one is compelled to ask how a genome can contain so many genes that act to destroy gametes. The costs of the selfish wtf genes could be at least partially offset because each genome also harbors multiple suppressors of drive. It is possible that we failed to observe drive of Sp wtf23, Sk wtf27, and Sk wtf29 in this study due to the presence of suppressors. Consistent with this idea, we identified three new wtf genes in this work that act to suppress drive (FY29033 wtf1, Sp wtf13, and FY29033 wtf18). We also found that two of these wtf genes (Sp wtf13 and FY29033 wtf18) can simultaneously promote their own drive while suppressing the drive of other wtf genes.

An alternate hypothesis is that the cost of wtf drivers could be mitigated if the genes are rarely in a heterozygous state (in which they drive and decrease fertility). A population genetic study of linkage disequilibrium between polymorphisms concluded that S. pombe rarely outcrosses with genetically diverged partners [56]. If so, why would so many meiotic drivers flourish in a species that rarely outcrosses? The extraordinary rate of wtf evolution may offer a solution to that paradox. Changes in class 1 and class 2 wtf gene sequences accumulate very rapidly due to extensive ectopic gene conversion within the family and due to the expansion and contraction of two sets of repeated sequences within the coding sequences of the genes [17, 21]. Because of this, crosses between otherwise isogenic isolates can generate diploids that are heterozygous at wtf loci [18]. The rate of such changes may be slow enough that few wtf drivers are likely to be heterozygous when closely related isolates mate. This could limit the fitness burden of carrying multiple active drive genes, while also maintaining the selfish selective advantage enjoyed by active drivers (or their suppressors) [18].

Finally, our new findings are consistent with the idea that Wtf antidotes must be highly similar to the poisons they neutralize [18]. Similarity between drivers and suppressors has also been observed in other systems and this similarity could potentially serve as a general guide to identify suppressors [36, 37, 50].

In many ways, the landscape of meiotic drive genes in S. pombe is similar to that in Podospora. This filamentous fungus also contains multiple meiotic drive loci, including the het-s drive system in P. anserina, the best understood driver in any system at the molecular level [45, 57]. In addition, Podospora contains its own multi-gene family of single-gene poison-antidote meiotic drivers, the Spok spore killers [36, 37]. Like the wtf genes, the Spok genes are variable in number and location between isolates, leading to varying degrees of spore-killing phenotypes in crosses. Different isolates of Podospora contain various Spok genes (Spok1, Spok2, Spok3, and Spok4). Like wtf drivers, some Spok genes can both drive and suppress other drivers [36, 37]. Similar to FY29033 wtf18 and Sp wtf13 in which suppression is bidirectional, Spok1 and Spok4 confer mutual resistance to each other [37]. However, this case is not general among the Spok drivers. For example, the Spok1 gene is resistant to the Spok2 driver, but not vice versa [36].

The wtf genes also show some marked differences from the Spok gene family. Firstly, the same sequence appears to encode both the poison and antidote functions for a given Spok gene. In addition, the driving wtf genes are diverse in amino acid identity (30–90%), whereas the Spok genes are more conserved (>94% DNA sequence identity) within P. anserina. This could be due to the predicted enzymatic (nuclease and kinase) functions of SPOK proteins, which may constrain their divergence [36, 37].

Interestingly, although both wtf and Spok spore killers evolved independently and use different genes to enact drive, the parallels suggest that there could be recurrent themes shared by different drive systems. These characteristics could perhaps be exploited to aid in the discovery and characterization of novel natural drive systems, especially in organisms with low genetic tractability that show evidence of meiotic drive. Additionally, these shared themes could help us understand how artificial gene drive systems may evolve in populations.

The potential functions of class 3 wtf genes remain largely unknown

Our final goal in this work was to test the potential functions of the class 3 wtf genes: wtf7, wtf11, wtf14, and wtf15. We observed no context in which we could detect Wtf11-GFP, but proteomics data indicate it is expressed late in meiosis [29]. We found that none of the class 3 wtf genes are required for vegetative growth. Additionally, our results showed that wtf7, wtf11, wtf14, and wtf15 cannot drive at their endogenous loci, nor are they required to enable the drive of a known wtf driver. However, we did find that wtf7, wtf14, and wtf15 are expressed in a spore-specific manner, similar to the antidotes encoded by wtf drivers and suppressors. The functional significance of the class 3 wtf genes is unclear as none of these genes are required for viability or fertility. Finally, we observed that Wtf14-GFP is localized to the ER during vegetative growth and in spores. This is the first observed expression of a Wtf protein during vegetative growth, demonstrating the potential for this gene to function outside of gametogenesis.

Building a wtf meiotic driver

Our studies have elucidated that the ‘rules’ underlying the construction of wtf meiotic drivers are curiously lax. Firstly, a surprisingly wide breadth of sequences are equally capable of enacting drive. The sequence encoded in exon 1, found in antidote messages, is the most conserved (68–100% amino acid identity amongst bona fide antidotes), suggesting this region may have specific interacting partners. The rest of the Wtf protein is strikingly diverse. For example, the Sk Wtf9 and FY29033 Wtf35 proteins share <30% amino acid identity (S2B Fig). However, both genes can drive into >90% of the functional gametes generated by a heterozygote.

In addition, our new analysis of the FY29033 wtf35 driver revealed that the 7-amino acid repeat sequences encoded in the last exon of the characterized 6-exon wtf drivers are dispensable for drive. This is surprising because we previously found that matching numbers of repeats in the Wtf^antidote and Wtf^poison proteins was important for poison-antidote specificity (for the Sp wtf13 driver and the Sp wtf18-2 suppressor) [18]. The fact that FY29033 wtf35 drives without these repeats suggests that they may function as spacers between more functionally important flanking domains. The important feature could be that having the same distance between the flanking domains helps to create compatible poison and antidote protein pairs.

It is curious that such diverse proteins act as equally effective Wtf poisons (or Wtf antidotes). It seems unlikely that the toxicity of the Wtf^poison proteins could be due to an enzymatic function, as it is unclear how an enzyme could maintain equal functionality while diverging so dramatically. In addition, it also seems unlikely that the toxicity relies on specific protein-protein interactions between the poison and some other factor, given the differences between the different poison proteins. We favor a model in which some other shared feature of the Wtf proteins, such as the multiple predicted transmembrane domains, may be important for their function. Future studies of the mechanisms used by the Wtf proteins will be guided by and hopefully reveal the molecular basis of these puzzling observations.

Materials and methods

Viable spore yield assay and allele transmission

To assay fertility and allele transmission, we began by generating heterozygous diploids as described in [27]. We grew the haploid parental strains in YEL (0.5% yeast extract, 3% dextrose, and 250 mg/L adenine, histidine, leucine, lysine, and uracil) to saturation at 32°C. Using the saturated cultures, we added 300 μl of each strain to an Eppendorf tube and vortexed it to mix the cells. We spun these cells down and resuspended the cell pellet in sterile ddH₂O. We then plated 100 μl of the cell mixture onto SPA (1% dextrose, 7.3 mM KH₂PO₄, vitamins, and agar) or SPAS (SPA + 45 mg/L adenine, histidine, leucine, lysine, and uracil) plates and incubated the plates at 25°C for ~16 hours. This allowed the haploids to mate and form diploids. Because some of the strains we used are homothallic (meaning they can switch mating type), we selected for heterozygous diploids. Most of our haploid parental strains had complementary auxotrophic markers which allowed us to select for heterozygous diploids by streaking the mated cell mix on minimal media. We then restreaked colonies that grew on minimal media to further isolate single colonies on minimal media. However, some of the heterozygous diploids were adenine auxotrophs. To select those diploids, we did the same procedure using minimal media supplemented with 45 mg/L of adenine.

Next, we cultured the heterozygous diploids in YEL overnight to saturation in a 32°C shaker. We then plated 50 μl of our saturated culture onto SPA to allow the diploids to sporulate (for three days at 25°C). We also diluted the cultures and subsequently plated the cells onto YEA (YEL + agar), and let them grow for three days at 32°C. These plates were then replica-plated to minimal media (+ adenine for adenine auxotrophs) as well as any other selective media to further confirm we isolated heterozygous diploids. From the YEA plates, we also calculated the colony-forming units (CFU) to determine the concentration of viable diploids in the YEL culture. We next scraped the cells off of the three day old SPA plates into 500 μl of sterile ddH₂O and treated them with 5 μl of glusulase (Sigma-Aldrich) for four hours at 32°C to shed the ascal membrane and wall to release the spores [58]. We then killed any remaining vegetative cells by adding 500 μl of 60% ethanol for 10 minutes at room temperature. Next, we washed the spores with ddH₂O and resuspended them in 500 μl of sterile ddH₂O. We diluted the spores and plated them onto YEA and let them grow into colonies at 32°C for three to five days to determine the CFU. Additionally, we picked colonies from the YEA plate onto a YEA master plate and grew the plate at 32°C for ~24 hours. We then replica-plated the master plate to yeast nitrogen-base plates with a specific dropout of either adenine, histidine, uracil, lysine, or leucine, and plates containing various drugs (G418, Hygromycin B, and Nourseothricin) to determine the genotype of each spore and thus assay allele transmission. Resistance to G418, Hygromycin B, or Nourseothricin is conferred by the kanMX4, hphMX6, or natMX4 cassettes, respectively. For each genotype reported, we generated at least two independent diploids via crossing for each genotype tested.

Viable spore yield calculations

To determine the fertility of selected stable heterozygous diploids, we calculated the viable spore yield (number of viable spores recovered from SPA/ the number of viable diploid cells plated on SPA). This is a slight variation of the method describe by Smith, as we frequently start with diploid cells rather than haploids [58]. The number of viable cells used in these calculations was determined using the CFU counts described above. For each genotype tested, we generated at least two independent diploids via crossing.

To determine the viable spore yield of homothallic haploids (i.e. SZY2254 and SZY3529), we first grew the strains in YEL cultures overnight at 32°C with shaking. We then spread 100 μl of each saturated culture onto SPA plates and left the plates at 25°C for three days. This step allowed the haploid cells to mate and subsequently sporulate. Using the same culture, we also performed serial dilutions and plated them onto YEA to quantify the number of haploids in the original culture. After three days, we scraped the spores off of the SPA plates and treated them as described above. Next, we performed a series of dilutions and plated the spores onto YEA plates to quantify the number of viable spores. We then incubated these plates at 32°C for five days. To determine the fertility, we calculated the viable spore yield (number of viable spores recovered from SPA/ number of haploids plated on SPA).

Strain construction: ade6-integrating vectors

All strain names and genotypes are presented in S4 Table. To assay allele transmission of different wtf genes, we used ade6-integrating vectors containing kanMX4 or hphMX6 drug resistance markers [59, 60]. pSZB188 (empty vector with kanMX4 resistance) was published in [16]. pSZB386 (empty vector with hphMX6 resistance) was published in [18]. pSZB387 is identical to pSZB386. We also generated a ura4-integrating vector linked to kanMX4. We ordered a gBlock from IDT (Coralville, IA) with a mutant ura4 allele. This mutant allele contained 5’ and 3’ deletions. We also deleted 50 bp from the middle of the gene which we replaced with a KpnI site. We then blunt-cloned this gBlock into the PvuII site of pFA6 to generate pSZB331. All of the vectors described above have a KpnI site within the mutant ade6-targeting cassette or the ura4-targeting cassette that we cut to linearize the plasmid.

Additionally, we made an ade6-integrating vector linked to natMX4. Similar to how pSZB188 and pSZB386 were constructed, we amplified the 5’ and 3’ side of the ade6 gene with oligos 588+589 and 590+591, respectively. We then stitched the two fragments together using overlap PCR. One of the oligos introduced a KpnI site in the middle of the gene. We then digested the fragment with BamHI and XhoI and ligated it into the BamHI and SalI sites of pAG25, to generate pSZB722. Due to the presence of KpnI sites within the natmX4 cassette, we mutated the KpnI site at ade6 to a StuI site. To do this, we used the QuickChange II XL Site-Directed Mutagenesis Kit from Agilent Technologies with pSZB722 as a template and oligos 2165 and 2166 to make pSZB849. We used StuI to linearize the plasmid. We then introduced plasmids into yeast using a variation of the standard lithium acetate protocol [61]. The variation to the protocol is that we do not sonicate the cells. Proper integration of these vectors at ade6 yields an Ade- phenotype (red colonies). To make the ade6-integrating vectors with the wtf transgenes, we cloned the wtf alleles using the oligos, DNA templates, restriction enzymes, and target sites described in S5 Table. For each gene, we cloned at least 500 bp upstream and 300 bp downstream of each gene in an effort to include regulatory sequences. The promoter regions of the class 1 and class 2 wtf genes are highly similar and 288 bp of upstream and 325 bp of downstream sequences were sufficient to ensure expression of previously tested drivers and suppressors [16–18]. We verified each plasmid by Sanger sequencing the entire gene and the associated flanking sequences.

We could not directly amplify FY29033 wtf1 alone from the genome, due to the repetitive nature of the wtf genes. Instead, we first amplified both FY29033 wtf1 + wtf36 tandem genes together from the FY29033 strain using oligos 1346 and 1348. From that PCR product, we amplified the FY29033 wtf1 allele using oligos 1352 and 1592. We then digested this fragment with SacI and subsequently ligated this allele into the SacI site of pSZB387 to make pSZB879. We verified the clone by sequencing the gene and flanking sequences. The plasmids are described in S6 Table.

Deleting Sk wtf1, Sp wtf7, Sp wtf11, and Sp wtf14 + wtf15

To generate an Sk wtf1 deletion cassette, we first amplified the regions (~750 bp) upstream and downstream of the wtf1 locus using oligo pairs 645+656 and 2158+646, respectively. The upstream region includes the Tf1 transposon flanking the left side of the Sk wtf1 locus. For these PCR reactions, we used SZY661 as a template. We then amplified the hphMX6 cassette from pAG32 with oligos 657 and 2159 [60]. We stitched all three fragments together using overlap PCR and transformed this deletion cassette into SZY298 to generate SZY3829. We confirmed the deletion using a series of PCR reactions: two oligo pairs with one oligo external to and one oligo within the deletion cassette (660+AO638 and AO1112+661) and a pair of oligos with one oligo outside of the deletion cassette and one oligo internal to the Sk wtf1 locus (660+2287).

To generate an Sp wtf7 deletion cassette, we first began by amplifying the regions (~1 kb) upstream and downstream of the gene with oligos 1565 + 1566 and oligos 1569 + 1570, respectively. We then amplified the drug cassettes, either hphMX6 or natMX4, with oligos 1567 and 1568 using pAG32 or pAG25 as templates, respectively [60]. These oligos contained tails with homology to the upstream and downstream sequence of Sp wtf7. We then stitched these separate fragments together using overlap PCR and subsequently transformed it into yeast using the standard lithium acetate protocol [61]. We deleted Sp wtf7 in the SZY44 strain to generate SZY2309, and deleted Sp wtf7 in SZY643 to generate strains SZY2310 and SZY2336. We confirmed these deletions using a series of PCR reactions: oligo pairs with one oligo outside of the deletion cassette and one oligo internal to the gene (1571+1586 and 1572+1585), and 2 oligo pairs in which one oligo was external to and one oligo was within the deletion cassette (1571+AO638 and 1572+AO1112). Additionally, we further confirmed deletion of the gene using oligos internal to Sp wtf7 (2152 and 2153).

To generate an Sp wtf11 deletion cassette, we began by amplifying the regions (~500 bp) upstream and downstream of the gene using oligos 1667+1669 and oligos 1670+1672, respectively. Next, we amplified the hphMX6 drug cassette with oligos 1668 and 1671 using pAG32 as a template [60]. We put these fragments together using overlap PCR and transformed it into SZY2336 to generate strains SZY2854 and SZY2855. This deletion was confirmed with two pairs of oligos with one oligo outside of the deletion cassette and one oligo internal to wtf11 (1705+1707 and 1706+1707) and 2 pairs of oligos in which one oligo was external to the deletion cassette and one oligo was internal to the hphMX6 cassette (1705+AO638 and 1706+1842). We further confirmed the deletion of the gene with a PCR reaction using oligos internal to Sp wtf11 (2154 and 2155).

To delete the tandem Sp wtf14 and wtf15 genes, we first made a deletion cassette via PCR. To do this, we amplified the upstream region (~600 bp) of Sp wtf14 using oligos 1649 and 1651, and the downstream region (~1 kb) of Sp wtf15 with oligos 1655 and 1657. We also amplified the natMX4 drug cassette from pAG25 with oligos 1650 and 1656 and then stitched the three fragments together using overlap PCR [60]. We then transformed this construct into SZY643 to generate strains SZY2856 and SZY2857. We confirmed the deletion of Sp wtf14 and Sp wtf15 via PCR using two oligo pairs in which one oligo was external to the deletion cassette and one oligo was within the wtf14 and wtf15 locus (1709+1710 and 1658+1711). We also did a PCR with in which one oligo was external to the deletion cassette and one was internal to the natMX4 cassette (1709+AO638 and 1658+1843). Additionally, we did a PCR with oligos within the Sp wtf14 + wtf15 locus (2156 and 2157) to confirm the absence of the genes.

To assay the effect of the class 3 wtf genes, we generated strains lacking wtf7, wtf11, wtf14, and wtf15 genes in Sp. First, we exchanged the natMX4 gene marking the wtf14+wtf15 deletion with either kanMX4 or CaURA3MX cassette as described in [62]. Briefly, we amplified the kanMX4 (from pFA6) or CaURA3MX (from pFA6-mTurq2-URA3MX) cassettes using oligos PR78 and PR79 [59, 63]. We then transformed the kanMX4 fragment into SZY2856 to generate the yeast strain SZY3444, and the CaURA3MX fragment into SZY2856 to generate SZY3448. We then generated quadruple deletion mutant strains via crosses.

Spot assay

To determine if a mutant strain lacking class 3 wtf genes had growth defects, we first cultured the strains in 5 ml of YEL at 32°C with shaking overnight. We then diluted the cultures to an OD600 of 1. We carried out serial dilutions and spotted 5 μl onto YEA plates and incubated them at 32°C for 2 days, after which the plates were imaged.

Deleting Sp wtf21 using CRISPR

We used the S. pombe CRISPR-Cas9 genome editing system from [64] to generate the Sp wtf21Δ::kanMX4 mutation in SZY890. This system uses two plasmids. The first one expresses a guide RNA to the target sequence and the second plasmid expresses Cas9 (pMZ222). To generate the plasmid carrying a guide RNA targeting Sp wtf21 (pSZB197), we annealed oligos 623 and 624 to each other and ligated them into the CspCI site of pMZ283. Next, we transformed pMZ222 and pSZB197 into Sp (SZY643) along with a wtf21Δ::kanMX4 repair cassette (see below). We initially selected for yeast that contained both plasmids (Leu+ Ura+) and subsequently selected for G418-resistant colonies. We then screened for the desired mutants by PCR using oligos 631 and 632 that flank wtf21, but are outside of the region amplified in the repair cassette. We generated the repair cassette using PCR to build a fragment containing the regions upstream and downstream of wtf21 flanking the kanMX4 gene [59]. We amplified the upstream and downstream regions with oligos 625+626 and 629+630, respectively, using Sp genomic DNA as a template. We amplified the kanMX4 gene with oligos 627+628. We then stitched the three fragments together using overlap PCR to generate the repair cassette.

We confirmed the Sp wtf21Δ::kanMX4 deletion via PCR. We used two oligo pairs in which one oligo was external to the deletion cassette and one oligo was within the wtf21 locus (631+1154 and 1557+632). We also did two sets of PCRs where one oligo was external to the deletion cassette and one was internal to the kanMX4 cassette (631+AO638 and AO1112+632). In addition, we further confirmed the deletion via Sanger sequencing. Finally, we did a PCR with oligos within the Sp wtf21 locus (1635+1636 and 1635+1637) to confirm the absence of the gene.

C-terminally GFP-tagged wtf alleles

We generated Sk wtf7, wtf11, and wtf15 GFP-tagged alleles using the following strategy. We first amplified the Sk wtf alleles (wtf7, wtf11, and wtf15) with their endogenous promoters using genomic DNA from SZY661 using oligo pairs 1359+1360 for wtf7, 1362+2210 for wtf11, and 991+1368 for wtf15. We then amplified yEGFP using pKT127 [65] as a template with oligos 1361+634 for wtf7, 2211+634 for wtf11, and 1369+634 for wtf15. We then stitched the two respective fragments together using overlap PCR and digested the fragments with SacI. We then cloned these fragments into the SacI site of pSZB188 to generate pSZB691 (wtf7-GFP), pSZB1087 (wtf11-GFP), and pSZB698 (wtf15-GFP).

To generate Sk wtf14-GFP, we amplified wtf14 from pSZB378 using oligos 1365 and 1366. To amplify yEGFP, we used oligos 1367 and 634 and used pKT127 as a template [65]. We then used overlap PCR to stitch the two fragments together and digested the fragment with SacI. We then cloned it into the SacI site of pSZB188 to generate pSZB696.

To generate FY29033 wtf35-GFP, we amplified wtf35 (from the upstream region to exon 5) from pSZB788 using oligos 1036 and 1349. We then digested this fragment with PvuII. We also ordered a gBlock from IDT (Coralville, IA) that contained 182 bp upstream of the stop codon of wtf35, a 24 bp linker (encoding GDGAGLIN), and the first 44 bp of yEGFP. To amplify yEGFP, we used oligos 2371 and 634 and used pKT127 as a template [65]. We then used overlap PCR to stitch the three fragments together and digested this new fragment with SacI. We then ligated it into the SacI site of pSZB188 to generate pSZB1176.

Imaging GFP-tagged Wtf proteins

To determine the localization of Wtf7-GFP, Wtf11-GFP, Wtf14-GFP, and Wtf15-GFP during haploid and diploid vegetative growth, we imaged cells during logarithmic and stationary phase. To image the cells during stationary phase, we first made a 5 ml YEL culture of each haploid and diploid strain and grew them for 16 hours in a 32°C shaker. We then imaged these cells as our stationary phase samples. To obtain cells in logarithmic phase, we used the YEL cultures described above and diluted them 1:10 in 5 ml cultures and incubated the tubes with shaking at 32°C for four hours. To image the cells, we spotted 3 μl of culture onto a glass slide pre-coated with lectin and covered with a glass coverslip to keep the cells in place for imaging.

To image the spore sacs, we used diploids that had sporulated on SPA plates at 25°C for two days. To prepare cells for imaging, we first scraped the cells off of SPA plates and mixed them with 3 μl of lectin. We then plated them on glass slides and covered with a glass coverslip.

For most experiments, we imaged the cells on an LSM-780 (Zeiss) AxioObserver confocal microscope with a 40X C-Apochromat water-immersion objective (Zeiss, NA = 1.2) or a 40X LD C-Apochromat water-immersion objective (Zeiss, NA = 1.1). We acquired images of every field of cells in two ways. We acquired a channel mode image to obtain a transmitted light image. For the images in Fig 5C, we acquired the fluorescence images by exciting GFP at 488 nm and collecting its emission between a 500–553 bandpass filter, while we excited mCherry at 561 nm and collected its emission between a 562–615 nm bandpass filter. For all other images, we acquired images in lambda mode over the entire visible range, with GFP and mCherry excitation at 488 and 561 nm, respectively, to obtain the true fluorescence. To eliminate cross talk, we collected the GFP and mCherry lambda images separately. To distinguish true GFP and mCherry signal from autofluorescence, we linearly unmixed the lambda mode data for GFP and mCherry using reference GFP and mCherry images and an in-house plug-in in ImageJ (https://imagej.nih.gov/ij/). When YFP was employed, we acquired the fluorescence by exciting YFP at 488 nm. We then collected the fluorescence spectrally as with GFP and mCherry above.

For S3A Fig, we used an LSM-700 (Zeiss) AxioObserver.Z1 confocal microscope with a 63x Plan Apochromat oil-immersion objective (Zeiss, NA = 1.4). The fluorescence was excited at 488 and 555 nm for GFP and mCherry, respectively. The GFP fluorescence was collected through a 400–577 nm bandpass filter while the mCherry was collected through a 577–700 nm bandpass filter in multitrack mode. The continuously variable secondary dichroic was set at 578 nm. The transmitted light was collected using the 488 nm excitation.

Analysis of meiotic proteomic data

We used the data set collected by Krapp et al and determined the relative protein levels following the method described in [29]. We considered all Wtf proteins that were detected in at least one timepoint in at least two of the three replicate experiments. We plotted the mean of the quantified Wtf protein for each timepoint that had at least two replicates, regardless of the number of replicates at that specific timepoint.

Alignments

To determine the DNA and amino acid sequence identity shared by wtf genes and proteins, we aligned the sequences using Geneious Prime (https://www.geneious.com). We used the Geneious aligner and performed a global alignment with free end gap using the default parameters. For DNA sequence alignments, the parameters we used were: cost matrix = 65% similarity, gap open penalty = 12, and gap extension penalty = 3. For protein sequence alignments we used Blosum62 as the matrix, gap open penalty = 12, gap extension penalty = 3, and refinement iterations = 2.

Supporting information

S1 Fig [tif]
Classification of the gene family based on their sequence and structure (21).

S2 Fig [a]
meiotic drivers are extremely diverse.

S3 Fig [a]
Localization of Wtf4-GFP, Wtf13-YFP, and FY29033 Wtf35-GFP.

S4 Fig [a]
Class 3 genes do not exhibit meiotic drive phenotypes.

S5 Fig [a]
The allele transmission bias favoring chromosome 1 is not caused by genes.

S6 Fig [tif]
Class 3 Wtf proteins are present during meiosis.

S7 Fig [green]
expression was not observed during vegetative growth or sporulation.

S8 Fig [a]
and expression was not observed in vegetative haploids or diploids.

S9 Fig [a]
Linear unmixing of Wtf7-GFP and Wtf15-GFP localization in spores.

S10 Fig [a]
Linear unmixing of Wtf14-GFP in haploid and diploid cells.

S1 Table [c1]
Raw data of allele transmission from Figs and , and .

S2 Table [pdf]
Raw data for fertility from and .

S3 Table [szy]
Raw data of allele transmission from .

S4 Table [pdf]
Yeast strains used in this study.

S5 Table [pdf]
Table summary of plasmid constructions.

S6 Table [pdf]
Plasmids used in this study.

S7 Table [pdf]
Oligos used in this study.