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

Meiotic drivers are selfish genetic loci that 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. 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% gametes when heterozygous. We also find three new wtf genes that suppress drive, including two that also act as autonomous drivers. Additionally, we find that wtf genes do not underlie a weak (64%) transmission bias caused by a locus or loci 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. Article Summary During gametogenesis, the two gene copies at a given locus, known as alleles, are each transmitted to 50% of the gametes (e.g. sperm). However, some alleles cheat so that they are found in more than the expected 50% of gametes, often at the expense of fertility. This selfish behavior is known as meiotic drive. Some members of the wtf gene family in the fission yeast, Schizosaccharomyces pombe, kill the gametes (spores) that do not inherit them, resulting in meiotic drive favoring the wtf allele. Other wtf genes act as suppressors of drive. However, the wtf gene family is diverse and only a small subset of the genes has been characterized. Here we analyze the functions of other members of this gene family and found eight new drivers as well as three new suppressors of drive. Surprisingly, we find that drive is relatively insensitive to changes in wtf gene sequence as highly diverged wtf genes execute gamete killing with similar efficiency. Finally, we also find that the expression and localization of some Wtf proteins are distinct from those of known drivers and suppressors, suggesting that these proteins may have non-meiotic drive functions.


Abstract 22
Meiotic drivers are selfish genetic loci that force their transmission into more than 50% 23 of the viable gametes made by heterozygotes. Meiotic drivers are known to cause 24 infertility in a diverse range of eukaryotes and are predicted to affect the evolution of 25 genome structure and meiosis. The wtf gene family of Schizosaccharomyces pombe 26 includes both meiotic drivers and drive suppressors and thus offers a tractable model 27 organism to study drive systems. Currently, only a handful of wtf genes have been 28 functionally characterized and those genes only partially reflect the diversity of the wtf 29 gene family. In this work, we functionally test 22 additional wtf genes. We identify eight 30 new drivers that share between 30-90% amino acid identity with previously 31 characterized drivers. Despite the vast divergence between these genes, they generally 32 drive into >85% gametes when heterozygous. We also find three new wtf genes that 33 suppress drive, including two that also act as autonomous drivers. Additionally, we find 34 that wtf genes do not underlie a weak (64%) transmission bias caused by a locus or loci 35 on chromosome 1. Finally, we find that some Wtf proteins have expression or 36 localization patterns that are distinct from the poison and antidote proteins encoded by 37 drivers and suppressors, suggesting some wtf genes may have non-meiotic drive 38 functions. Overall, this work expands our understanding of the wtf gene family and the 39 burden selfish driver genes impose on S. pombe. 40

Introduction 58
During meiosis, diploid cells divide to produce haploid gametes (e.g. sperm). This 59 process is generally fair in that each parental allele of a gene is represented at an equal 60 ratio in the gametes (1). However, many eukaryotic genomes contain 'selfish' elements 61 that bias their own transmission into the viable gametes generated by a heterozygote 62 (2-4). These loci are known as meiotic drivers and they can directly and indirectly 63 reduce fitness through a number of mechanisms (reviewed in (5, 6)). 64 diploids to undergo meiosis and then assayed the presence of each wtf gene of interest 137 in the viable spore population using the linked selectable markers. Genes capable of 138 autonomous drive are expected to be overrepresented in the viable progeny (>50%). 139 140 Most genes within class 1 have six exons, including all previously described wtf drivers. 141 Class 1 also includes two 5-exon wtf genes ( Figure 1A, Supplemental Figure 1). We 142 tested ten 6-exon and two 5-exon wtf genes from class 1 (Figure 2A, diploids 1-16). We 143 found that seven of the 6-exon genes and one 5-exon gene exhibited significant drive in 144 at least one strain background ( Figure 2A). These genes are: Sk wtf9, Sk wtf19, Sk 145 wtf30, Sk wtf33, Sp wtf19, FY29033 wtf36, FY29033 wtf18, and FY29033 wtf35. 146 Interestingly, these genes are incredibly diverse and share as little as 30% pairwise 147 amino acid identity (Supplemental Figure 2A). Moreover, all of the drivers except one 148 (Sp wtf19) drive into >90% of the progeny when tested under the same conditions (16, 149 18). These results demonstrate that a remarkably wide range of proteins can execute 150 spore-killing similarly well. Interestingly, the 5-exon FY29033 wtf35 meiotic driver lacks 151 sequences homologous to what is exon 4 in the other known 6-exon drivers. In addition, 152 the gene lacks the 7-amino acid repeat that is found in between 1-4 copies in the last 153 exon of all other drivers (21). The fact that FY29033 wtf35 can still drive suggests these 154 sequences can be dispensable for drive. 155

156
The four class 1 genes that did not exhibit drive in our tests were Sk wtf27, Sk wtf29, Sp 157 wtf23, and CBS5557 wtf23 (Figure 2A, diploids 12-16). CBS5557 wtf23 completely 158 lacks a specific 11-amino acid repeat within exon three that can be found in 2-7 copies 159 in all characterized drivers except FY29033 wtf35, which contains only four amino acids 160 of the repeat. We found no sequence features in Sk wtf27, Sk wtf29, or Sp wtf23 that 161 distinguish these genes from the confirmed drivers. While it is possible our results 162 reflect that these four genes are not capable of driving, it is also possible that these 163 genes are suppressed in the background in which we tested them. We previously saw 164 background-specific suppression of the Sp wtf13 driver, and similar suppression 165 explains why we observed drive of FY29033 wtf18 in the Sk, but not in the Sp 166 background (see below) (18). Overall, our analyses are consistent with class 1 167 containing a wide diversity of autonomous meiotic drive genes. 168

169
The class 2 wtf genes are either known or predicted antidote-only drive suppressors 170 ( Figure 1B). Like the class 1 genes, class 2 contains both 5-exon and 6-exon genes, 171 although only 6-exon genes have been previously tested (16-18). Interestingly, class 2 172 also contains Sp wtf21, which was previously reported to be an essential gene because 173 a heterozygous mutant (Sp wtf21/wtf21∆) transmitted only the wild-type Sp wtf21 allele 174 to viable gametes (24). 175 176 We first tested four class 2 genes (Sk wtf13, Sk wtf23, Sk wtf35, and FY29033 wtf1) 177 using the same approach described above for class 1 genes. As expected for genes 178 predicted to produce only antidotes, we found that none of the genes tested exhibited 179 drive in an ectopic strain background ( Figure 2B, diploids 17-20). We also revisited the 180 idea that Sp wtf21 is an essential gene (24). We found that we could generate a deletion 181 of Sp wtf21 in a haploid strain, indicating the gene is not essential in that strain 182 background. Moreover, we generated an Sp wtf21/wtf21∆ heterozygote and did not 183 observe drive ( Figure 2B, diploid 21). 184 185 Finally, we tested if we could observe autonomous meiotic drive by the class 3 wtf 186 genes (wtf7, wtf11, wtf14, and wtf15) in Sp. Unlike the other wtf genes (class 1 and 187 class 2), each natural isolate has a clear ortholog of each class 3 wtf gene (21). We 188 therefore had to use a different strategy to test if these genes could drive. Instead of 189 introducing the genes into an ectopic strain background, we assayed whether these 190 genes could drive when heterozygous at their endogenous loci. We deleted wtf7 and 191 wtf11 individually, and wtf14 and wtf15 together as they are adjacent to each other. In 192 diploids heterozygous for any of these wtf gene deletions, the wild-type alleles were 193 transmitted to ~50% of the viable spores (Supplemental Figure 3A). This indicates that 194 these genes cannot drive or that they are suppressed in the Sp background. We also 195 assayed a homozygous deletion strain lacking all of the class 3 genes. We observed no 196 fertility defects, indicating these genes are not required for sexual reproduction 197 (Supplemental Figure 3B). We also observed no growth defects in haploids lacking the 198 class 3 genes (Supplemental Figure 3C). but we wanted to test if other wtf genes could also act as drive suppressors (18). Within 203 drivers and their known suppressors, the Wtf antidote proteins share high levels of amino 204 acid identity with the poisons they neutralize. This similarity appears to be particularly 205 important within the C-terminus (18). We used this knowledge to guide our search for 206 other drive suppressors. We found that the residues encoded in the last three exons of 207 the class 2 gene FY29033 wtf1 share >99% identity to those in the FY29033 wtf35 208 driver ( Figure 3A). We therefore predicted that FY29033 wtf1 could be a suppressor of 209 the FY29033 wtf35 driver. To test this, we made diploids heterozygous for both alleles 210 (FY29033 wtf35/FY29033 wtf1) and assayed transmission of each allele. FY29033 211 wtf35 was no longer able to drive in the presence of FY29033 wtf1 in the Sp 212 background ( Figure 3C, compare diploid 10 to 24). Additionally, FY29033 wtf1 was able 213 to rescue the fertility defect caused by FY29033 wtf35 ( Figure 3C). These results 214 demonstrate that FY29033 wtf1 is a suppressor of FY29033 wtf35. 215

216
To broaden our search for suppressors, we examined why FY29033 wtf18 drives in Sk, 217 but not in Sp. Specifically, we looked for wtf genes in Sp that could work as a 218 suppressor of FY29033 wtf18. We noticed that the C-terminal region (216 amino acids) 219 of FY29033 wtf18 is identical to that of the Sp wtf13 meiotic driver ( Figure 3B). This 220 suggested that the Sp Wtf13 antidote could potentially neutralize the FY29033 Wtf18 poison 221 and vice versa. To test this, we generated FY29033 wtf18::kanMX4/Sp wtf13::hphMX6 222 heterozygotes in an Sk strain background and assayed their phenotypes. We observed 223 that drive of both genes was suppressed in the heterozygote relative to hemizygotes 224 containing only one of the drivers (Figure 3, compare diploids 8 and 25 to diploid 26). 225 We still observed drive of the FY29033 wtf18 allele in the FY29033 wtf18/Sp wtf13 226 heterozygous diploid, suggesting the Wtf13 antidote is only partially effective against the 227 Wtf18 poison . Additionally, the level of drive of FY29033 wtf18 we observed is sufficient to 228 explain the decrease in fertility of the diploid (64.4% of wild-type, Figure 3C, diploid 26). 229 These results show that wtf drivers can also function as suppressors of each other. We previously detected a weak allele transmission bias that favored Sk chromosome 1 234 in the viable progeny of Sp/Sk hybrid diploids (25). We reexamined this observation 235 here with hybrid diploids that are heterozygous for Sp and Sk copies of chromosome 1, 236 but are homozygous for Sk chromosomes 2 and 3. These hybrids were also unable to 237 initiate meiotic recombination due to deletion of rec12 (SPO11 homolog), which is 238 required for programmed meiotic double-stranded break formation. The rec12 deletion 239 ensured chromosome-wide linkage on chromosome 1, allowing us to monitor 240 transmission of the unmapped locus into the spore progeny (Supplemental Figure 4). 241 Consistent with our previous observations, we observed 64.3% of the viable spores 242 generated by the hybrids inherited Sk chromosome 1. There is only one wtf locus, wtf1, 243 on both Sp and Sk chromosome 1. We found that the transmission of Sk chromosome 1 244 was not significantly different after deleting Sk wtf1 (Supplemental Figure 4B). This 245 result is consistent with the presence of a non-wtf driver on Sk chromosome 1. 246 However, it is also possible that Dobzhansky-Muller incompatibilities between Sp 247 chromosome 1 and Sk chromosomes 2 and/or 3 explain the transmission bias (26). 248 249 High amino acid identity is crucial for Wtf poison and antidote specificity 250 We next wanted to use our expanded knowledge of drivers and suppressors to further 251 refine our understanding of how similar antidote proteins must be to the poisons they 252 neutralize. All known suppressors have >97% identity in the amino acids encoded in the 253 last two exons to the drivers they suppress, but it is unclear if less similar Wtf proteins 254 could also work. We reasoned that genes that drive in Sp must not be suppressed by 255 any endogenous Sp genes. We therefore compared the similarity of genes that drive in 256 Sp to every Sp wtf gene that failed to suppress them (i.e. transcribed wtf genes) (22). 257 We found that the endogenous Sp Wtf proteins and the ectopic drivers share between 258 43-86% pairwise identity within the residues encoded in the final two exons (43-69 259 amino acids) (Supplemental Figure 5A and 5B). This analysis is consistent with the 260 notion that high amino acid identity within the C-terminal region of the Wtf antidote and 261 poison proteins (87-100%) is important for specificity (Supplemental Figure 5A and 5C). We previously analyzed the sequences of the class 3 wtf genes across four isolates of 265 S. pombe (Sp, Sk, FY29033, and CBS5557) and found that wtf7, wtf11, and wtf15 show 266 signatures of positive selection (21). This suggests that selection is favoring novelty and 267 led us to speculate that these genes could be involved in meiotic drive. Given that we 268 failed to observe drive of these class 3 genes (Supplemental Figure 3A), we thought 269 they could perhaps be modifying drive of other wtf genes. All class 3 wtf genes are 270 linked to drive loci so we predicted that they could facilitate drive, as they would also 271 gain a transmission advantage from a linked driver (8). 272 We tested this idea by generating diploids hemizygous for the Sk wtf4 meiotic driver (Sk 274 wtf4::kanMX4/ade6+) and lacking all four of the class 3 wtf genes. Sk wtf4 was found in 275 nearly 100% of the viable spores generated by this quadruple deletion mutant 276 (Supplemental Figure 3D, compare diploid 30 to 31). These results show that wtf7, 277 wtf11, wtf14, and wtf15 are not required to facilitate drive of wtf meiotic drivers. 278 279

Localization of class 3 Sk Wtf proteins in Sp 280
We next decided to take a more agnostic approach to look for possible functions of 281 Wtf7, Wtf11, Wtf14, and Wtf15 by investigating their expression and localization in cells. 282 We analyzed a published data set from a proteomics meiotic time course study (27) and 283 found that peptides from Wtf11, Wtf14, and Wtf15, but not Wtf7, were all detected 284 during meiosis in three replicate experiments (Supplemental Figure 6). Wtf11 and Wtf15 285 were detected in one or more timepoints taken after the first meiotic division. 286 Interestingly, the levels of Wtf14 remained steady throughout meiosis (Supplemental 287 Figure 6). These patterns are both unlike those of class 1 and class 2 Wtf proteins 288 which increase in abundance as meiosis progresses (18,27). 289

290
We also tagged each of the class 3 genes under the control of their endogenous 291 promoters (cloned from the Sk isolate) with GFP at the C-terminus and integrated them 292 at the ade6 locus of Sp. We then imaged haploid and diploid cells during logarithmic cell 293 growth and stationary phase. We also imaged cells undergoing meiosis and tetrads. 294 In contrast to the proteomics data, we were not able to detect any GFP fluorescence 296 under any conditions in cells containing the wtf11-GFP allele (Supplemental Figure 7) 297 (27). The reasons for this are unclear, but it could be due to our tag generating a null 298 allele. We also did not detect Wtf7-GFP or Wtf15-GFP in vegetatively growing haploids 299 or diploids (Supplemental Figure 8). However, we did observe GFP fluorescence in 300 tetrads produced by diploids heterozygous for wtf7-GFP and tetrads produced by wtf15-301 GFP heterozygotes (Figure 4 and Supplemental Figure 9). This late expression of Wtf15 302 is consistent with the proteomics data set (Wtf7 was not detected by proteomics) (27). In 303 tetrads generated by diploids with one tagged copy of Wtf7 or Wtf15, we observed that 304 the signal was greatly enriched in two of the four spores ( Figure 4). We speculate that 305 the two spores with bright GFP signal are those that inherited the tagged allele. This 306 would suggest that wtf7-GFP and wtf15-GFP are both expressed after spore 307 individualization, similar to the antidotes of wtf drivers and suppressors (16, 18). Wtf15-308 GFP exhibited a diffuse localization pattern that largely filled the spores ( Figure 4B and 309 Supplemental Figure 9B). The localization pattern of Wtf7-GFP varied in different 310 spores. In some instances, Wtf7-GFP made a ring-like structure next to Nsp1-mCherry 311 (nucleoporin marker) ( Figure 4A and Supplemental Figure 9A). However, we also 312 observed other cases where Wtf7-GFP seemed to be clustered within the spore (  Overall, we identified eight new wtf genes capable of meiotic drive. The S. pombe 331 isolates with assembled wtf sequences contain between 4-14 predicted drivers (17, 21). 332 Our current work shows that many of these predicted drive genes are able to drive in 333 the right genetic context. This work expands the rapidly growing number of bona fide 334 drive genes identified in recent years (4, 16-18, 31-41 In S. pombe, one is compelled to ask how a genome can contain so many genes that 349 act to destroy gametes. The costs of the selfish wtf genes could be at least partially 350 offset because each genome also harbors multiple suppressors of drive. We speculate 351 that we failed to observe drive of Sp wtf23, Sk wtf27, and Sk wtf29 in this study due to 352 the presence of suppressors. Consistent with this idea, we identified three new wtf 353 genes in this work that act to suppress drive. We also found that some wtf genes can 354 simultaneously promote their own drive while suppressing the drive of other wtf genes. 355 Finally, our new findings reinforce the idea that Wtf antidotes must be highly similar to 356 the poisons they neutralize (18). Similarity between drivers and suppressors has also 357 been observed in other systems and this similarity could potentially serve as a general 358 guide to identify suppressors (34, 35, 48). 359 360 In many ways, the landscape of meiotic drive genes in S. pombe is similar to that in 361 Podospora. This filamentous fungus also contains multiple meiotic drive loci, including 362 the het-s drive system in P. anserina, the best understood driver in any system at the 363 molecular level (43, 54). In addition, Podospora contains its own multi-gene family of 364 The wtf genes also show some marked differences from the Spok gene family. Firstly, 375 the same sequence appears to encode both the poison and antidote functions for a 376 given Spok gene. In addition, the driving wtf genes are diverse in amino acid identity 377 (30-90%), whereas the Spok genes are more conserved (>94% DNA sequence identity) 378 within P. anserina. This could be due to the predicted enzymatic (nuclease and kinase) 379 function of Spok proteins, which may constrain their divergence (34, 35). 380 381 Interestingly, although both wtf and Spok spore killers evolved independently and use 382 different genes to enact drive, the parallels suggest that there could be recurrent 383 themes shared by different drive systems. These characteristics could perhaps be 384 exploited to aid in the discovery and characterization of novel natural drive systems, 385 especially in organisms with low genetic tractability that show evidence of meiotic drive. 386 Additionally, these shared themes could help us understand how artificial gene drive 387 systems may evolve in populations. 388

389
The potential functions of class 3 wtf genes remain largely unknown 390 Our final goal in this work was to test the potential functions of the class 3 wtf genes: 391 wtf7, wtf11, wtf14, and wtf15. We observed no context in which we could detect Wtf11-392 GFP, but the proteomics data indicates it is expressed late in meiosis (27). We found 393 that none of the class 3 wtf genes are required for vegetative growth. Additionally, our 394 results showed that wtf7, wtf11, wtf14, and wtf15 cannot drive at their endogenous loci, 395 nor are they required to enable the drive of other wtf genes. However, we did find that 396 wtf7, wtf14, and wtf15 are expressed in a spore-specific manner, similar to the antidotes 397 encoded by wtf drivers and suppressors. The functional significance of the class 3 wtf 398 genes is unclear as none of these genes are required for fertility. Finally, we observed 399 that Wtf14-GFP is localized to the ER during vegetative growth and in spores. This is 400 the first observed expression of a Wtf protein during vegetative growth, demonstrating 401 the potential for this gene to function outside of gametogenesis. 402 403

Building a wtf meiotic driver 404
Our studies have elucidated that the 'rules' underlying the construction of wtf meiotic 405 drivers are curiously lax. Firstly, a surprisingly wide breadth of sequences are equally 406 capable of enacting drive. The sequence encoded in exon 1, found in antidote 407 messages, is the most conserved (68-100% amino acid identity), suggesting this region 408 may have specific interacting partners. The rest of the Wtf proteins is strikingly diverse. 409 For example, the Sk Wtf9 and FY29033 Wtf35 proteins share <30% amino acid identity 410 (Supplemental Figure 2B). However, both genes can drive into >90% of the functional 411 gametes generated by a heterozygote. 412

413
In addition, our new analysis of the FY29033 wtf35 driver revealed that the 7-amino acid 414 repeat sequences encoded in the last exon of the characterized 6-exon wtf drivers are 415 dispensable for drive. This is surprising because we previously found that a matching 416 numbers of repeats in the Wtf antidote and Wtf poison proteins was important for poison-417 antidote specificity (18). The fact that FY29033 wtf35 drives without these repeats 418 suggests that they may function as spacers between more functionally important 419 flanking domains. The important feature could be that having the same distance 420 between the flanking domains helps to create compatible poison and antidote protein 421 pairs. 422

423
Overall, the incredible diversity of wtf drivers could indicate that the toxicity of the Wtf 424 poisons is not due to a specific enzymatic activity of the proteins or the targeting of a 425 shared interactor. Rather, general shared features of Wtf proteins, such as the multiple 426 predicted transmembrane domains, may be important. 427

428
The biggest constraint within wtf drive systems seems to be that antidotes must be very 429 similar to the poisons they suppress. Within complete drivers, the absolute identity 430 between the poison and antidote is guaranteed as they are both encoded on the same 431 sequence. When an antidote is encoded by a locus distinct from the driver, it can still be 432 effective at neutralizing a given poison. However, these sequences need to be highly 433 similar. Although it is not yet clear exactly how much difference can be tolerated, the 434 similarity at the C-terminus of the protein is particularly important. All known antidotes 435 are >96% similar in amino acid sequence to the poisons they neutralize. Future studies 436 of the mechanisms used by the Wtf proteins will be guided by and hopefully reveal the 437 molecular basis of these features. 438 439

Viable spore yield assay and allele transmission 441
To assay fertility and allele transmission, we began by generating heterozygous diploids 442 as described in (25). We grew the haploid parental strains in YEL (0.5% yeast extract, 443 3% dextrose, and 250 mg/L adenine, histidine, leucine, lysine, and uracil) to saturation 444 at 32°C. Using the saturated cultures, we added 300 microliters of each strain to an 445 Eppendorf tube and vortexed it to mix the cells. We spun these cells down and 446 resuspended the cell pellet in sterile ddH2O. We then plated 100 microliters of the cell 447 mixture onto SPA (1% dextrose, 7.3mM KH2PO4, vitamins, and agar) or SPAS (SPA + 448 45mg/L adenine, histidine, leucine, lysine, and uracil) plates and incubated the plates at 449 25°C for ~16 hours. This allowed the haploids to mate and form diploids. Because some 450 of the strains we used are homothallic (meaning they can switch mating type), we 451 selected for heterozygous diploids. Most of our haploid parental strains had 452 complementary auxotrophic markers which allowed us to select for heterozygous 453 diploids by streaking the mated cell mix on minimal media. We then restreaked colonies 454 that grew on minimal media to further isolate single colonies on minimal media. 455 However, some of the heterozygous diploids were adenine auxotrophs. To select those 456 diploids, we did the same procedure using minimal media supplemented with 45 mg/L 457 of adenine. 458 Next, we cultured the heterozygous diploids in YEL overnight to saturation in a 32°C 460 shaker. We then plated 50 microliters of our saturated culture onto SPA to allow the 461 diploids to sporulate (for three days at 25°C). We also diluted the cultures and 462 subsequently plated the cells onto YEA (YEL + agar), and let them grow for three days 463 at 32°C. These plates were then replica-plated to minimal media (+ adenine for adenine 464 auxotrophs) as well as any other selective media to further confirm we isolated 465 heterozygous diploids. From the YEA plates, we also calculated the colony-forming 466 units (CFU) to determine the concentration of viable diploids in the YEL culture. We next 467 scraped the cells off of the three day old SPA plates into 500 microliters of sterile ddH2O 468 and treated them with five microliters of glusulase (Sigma-Aldrich) for four hours at 32°C 469 to shed the ascal membrane and wall to release the spores (55). We then killed any 470 remaining vegetative cells by adding 500 microliters of 60% ethanol for 10 minutes at 471 room temperature. Next, we washed the spores in ddH2O and resuspended them in 500 472 microliters of sterile ddH2O. We diluted the spores and plated them onto YEA and let 473 them grow into colonies at 32°C for three to five days to determine the CFU. 474 Additionally, we picked colonies from the YEA plate onto a YEA master plate and grew 475 the plate at 32°C for ~24 hours. We then replica-plated the master plate to yeast 476 nitrogen-based plates with a specific dropout of either adenine, histidine, uracil, lysine, 477 leucine, and plates containing various drugs (G418, Hygromycin B, and Nourseothricin) 478 to determine the genotype of each spore and thus assay allele transmission. 479 480

Viable spore yield calculations 481
To determine the fertility of selected stable heterozygous diploids, we calculated the 482 viable spore yield (number of viable spores recovered from SPA/ the number of viable 483 diploid cells plated on SPA) (55). The number of viable cells used in these calculations 484 was determined using the CFU counts described above. 485

486
To determine the viable spore yield of homothallic haploids (i.e. SZY2254 and 487 SZY3529), we first grew the strains in YEL cultures overnight at 32 C with shaking. We 488 then spread 100 microliters of each saturated culture onto SPA plates and left the plates 489 at 25 C for three days. This step allowed the haploid cells to mate and subsequently 490 sporulate. Using the same culture, we also performed serial dilutions and plated them 491 onto YEA to quantify the number of haploids in the original culture. After three days, we 492 scraped the spores off of the SPA plates and treated them as described above. Next, 493 we performed a series of dilutions and plated the spores onto YEA plates to quantify the 494 number of viable spores. We then incubated these plates at 32 C for five days. To 495 determine the fertility, we calculated the viable spore yield (number of viable spores 496 recovered from SPA/ number of haploids plated on SPA). 497

Strain construction: ade6-integrating vectors 499
All strain names and genotypes are presented in Supplemental Table 4. To assay allele 500 transmission of different wtf genes, we used ade6-integrating vectors containing 501 kanMX4 or hphMX6 drug resistance markers (56, 57). pSZB188 (empty vector with 502 kanMX4 resistance) was published in (16). pSZB386 (empty vector with hphMX6 503 resistance) was published in (18). pSZB387 is identical to pSZB386. These vectors 504 have a KpnI site within a mutant ade6-targeting cassette that we cut to linearize the 505 plasmid. We then introduced plasmids into yeast using a standard lithium acetate 506 protocol. Proper integration of these vectors at ade6 yields an Ade-phenotype (red 507 colonies). To make the ade6-integrating vectors with the wtf transgenes, we cloned the 508 wtf alleles using the oligos, DNA templates, restriction enzymes, and target sites 509 described in Supplemental Table 5. 510

511
We could not directly amplify FY29033 wtf1 alone from the genome, due to repetitive 512 nature of the wtf genes. Instead, we first amplified both FY29033 wtf1 + wtf36 tandem 513 genes together from the FY29033 strain using oligos 1346 and 1348. From that PCR 514 product, we amplified the FY29033 wtf1 allele using oligos 1352 and 1592. We then 515 digested this fragment with SacI and subsequently ligated this allele into the SacI site of 516 pSZB387 to make pSZB879. The plasmids are described in Supplemental Table 6. 517 518 Deleting Sk wtf1, Sp wtf7, Sp wtf11, Sp wtf14 + wtf15 519 To generate an Sk wtf1 deletion cassette, we first amplified the regions (~750bp) 520 upstream and downstream of the wtf1 locus using oligo pairs 645+656 and 2158+646, 521 respectively. The upstream region includes the Tf1 transposon flanking the left side of 522 the Sk wtf1 locus. For these PCR reactions, we used SZY661 as a template. We then 523 amplified the hphMX6 casette from pAG32 with oligos 657 and 2159 (57). We stitched 524 all three fragments together using overlap PCR and transformed this deletion cassette 525 into SZY298 to generate SZY3829. We confirmed the deletion using a series of PCR 526 reactions: two oligo pairs with one oligo external to and one oligo within the deletion 527 cassette (660+AO638 and AO1112+661) and a pair of oligos with one oligo outside of 528 the deletion cassette and one oligo internal to the Sk wtf1 locus (660+2287). 529

530
To generate an Sp wtf7 deletion cassette, we first began by amplifying the regions (~1 531 kb) upstream and downstream of the gene with oligos 1565 + 1566 and oligos 1569 + 532 1570, respectively. We then amplified the drug cassettes, either hphMX6 or natMX4, 533 with oligos 1567 and 1568 using pAG32 or pAG25 as templates, respectively (57). 534 These oligos contained tails with homology to the upstream and downstream sequence 535 of Sp wtf7. We then stitched these separate fragments together using overlap PCR and 536 subsequently transformed it into yeast using the standard lithium acetate protocol. We 537 deleted Sp wtf7 in the SZY44 strain to generate SZY2309, and deleting Sp wtf7 in 538 SZY643 generated strains SZY2310 and SZY2336. We confirmed these deletions using 539 a series of PCR reactions: oligo pairs with one oligo outside of the deletion cassette and 540 one oligo internal to the gene (1571+1586 and 1572+1585), and 2 oligo pairs in which 541 one oligo was external to and one oligo was within the deletion cassette (1571+AO638 542 and 1572+AO1112). Additionally, we further confirmed deletion of the gene using oligos 543 internal to Sp wtf7 (2152 and 2153). 544 To generate an Sp wtf11 deletion cassette, we began by amplifying the regions (~500 546 bp) upstream and downstream of the gene using oligos 1667+1669 and oligos 547 1670+1672, respectively. Next, we amplified the hphMX6 drug cassette with oligos 1668 548 and 1671 using pAG32 as a template (57). We put these fragments together using 549 overlap PCR and transformed it into SZY2336 to generate strains SZY2854 and 550 SZY2855. This deletion was confirmed with two pairs of oligos with one oligo outside of 551 the deletion cassette and one oligo internal to wtf11 (1705+1707 and 1706+1707) and 2 552 pairs of oligos in which one oligo was external to the deletion cassette and one oligo 553 was internal to the hphMX6 cassette (1705+AO638 and 1706+1842). We further 554 confirmed the deletion of the gene with a PCR reaction using oligos internal to Sp wtf11 555 (2154 and 2155). 556

557
To delete the tandem Sp wtf14 and wtf15 genes, we first made a deletion cassette via 558 PCR. To do this, we amplified the upstream region (~600 bp) of Sp wtf14 using oligos 559 1649 and 1651, and the downstream region (~1 kb) of Sp wtf15 with oligos 1655 and 560 1657. We also amplified the natMX4 drug cassette from pAG25 with oligos 1650 and 561 1656 and then stitched the three fragments together using overlap PCR (57). We then 562 transformed this construct into SZY643 to generate strains SZY2856 and SZY2857. We 563 confirmed the deletion of Sp wtf14 and Sp wtf15 via PCR using two oligo pairs in which 564 one oligo was external to the deletion cassette and one oligo was within the wtf14 and 565 wtf15 locus (1709+1710 and 1658+1711). We also performed a PCR with in which one 566 oligo was external to the deletion cassette and one was internal to the natMX4 cassette 567 (1709+AO638 and 1658+1843). Additionally, we did a PCR with oligos within the Sp 568 wtf14 + wtf15 locus (2156 and 2157) to confirm the absence of the genes. 569

570
To assay the effect of the class 3 wtf genes, we generated strains lacking wtf7, wtf11, 571 wtf14, and wtf15 genes in Sp. First, we exchanged the natMX4 gene marking the 572 wtf14+wtf15 deletion with the CaURA3MX cassette as described in (58). Briefly, we 573 amplified the CaURA3MX cassette from pFA6-mTurq2-URA3MX using oligos PR78 and 574 PR79 (59). We then transformed this fragment into SZY2856 to generate the yeast 575 strain SZY3448. We then generated the quadruple deletion mutant strain via crosses. 576 577

Spot assay 578
To determine if a mutant strain lacking class 3 wtf genes had growth defects, we first 579 cultured the strains in five ml of YEL at 32°C with shaking overnight. We then diluted the 580 cultures to an OD600 of 0.1 and grew them for six hours at 32°C with shaking. We then 581 did serial dilutions and spotted five microliters onto YEA plates and incubated them at 582 32°C for 2 days. 583 584 Deleting Sp wtf21 using CRISPR 585 We used the S. pombe CRISPR-Cas9 genome editing system from (60) to generate 586 the Sp wtf21Δ::kanMX4 mutation in SZY890. This system uses two plasmids. The first 587 one expresses a guide RNA to the target sequence and the second plasmid expresses 588 Cas9 (pMZ222). To generate the plasmid carrying a guide RNA targeting Sp 589 wtf21 (pSZB197), we annealed oligos 623 and 624 to each other and ligated them into 590 the CspCI site of pMZ283. Next, we transformed pMZ222 and pSZB197 591 into Sp (SZY643) along with a wtf21Δ::kanMX4 repair cassette (see below). We initially 592 selected for yeast that contained both plasmids (Leu+ Ura+) and subsequently selected 593 for G418-resistant colonies. We then screened for the desired mutants by PCR using 594 oligos 631 and 632 that flank wtf21, but are outside of the region amplified in the repair 595 cassette. We generated the repair cassette using PCR to build a fragment containing 596 the regions upstream and downstream of wtf21 flanking the kanMX4 gene (56). We 597 amplified the upstream and downstream regions with oligos 625+626 and 629+630, 598 respectively, using Sp genomic DNA as a template. We amplified the kanMX4 gene with 599 oligos 627+628. We then stitched the three fragments together using overlap PCR to 600 generate the repair cassette. 601 602 C-terminally GFP-tagged wtf alleles 603 We generated Sk wtf7, wtf11, and wtf15 GFP-tagged alleles using the following 604 strategy. We first amplified the Sk wtf alleles (wtf7, wtf11, and wtf15) with their 605 endogenous promoters using genomic DNA from SZY661 using oligo pairs 1359+1360 606 for wtf7, 1362+2210 for wtf11, and 991+1368 for wtf15. We then amplified yEGFP using 607 pKT127 (61) as a template with oligos 1361+634 for wtf7, 2211+634 for wtf11, and 608 1369+634 for wtf15. We then stitched the two respective fragments together using 609 overlap PCR and digested the fragments with SacI. We then cloned these fragments 610 into the SacI site of pSZB188 to generate pSZB691 (wtf7-GFP), pSZB1087 (wtf11-611 GFP), and pSZB698 (wtf15-GFP). 612 To generate Sk wtf14-GFP, we amplified wtf14 from pSZB378 using oligos 1365 and 614 1366. To amplify yEGFP, we used oligos 1367 and 634 and used pKT127 as a template 615 (61). We then used overlap PCR to stitch the two fragments together and digested the 616 fragment with SacI. We then cloned it into the SacI site of pSZB188 to generate 617 pSZB696. 618

Imaging GFP-tagged Wtf proteins 620
To determine the localization of Wtf7-GFP, Wtf11-GFP, Wtf14-GFP, and Wtf15-GFP To image the spore sacs, we used diploids that had sporulated on SPA plates at 25°C 631 for two days. To prepare cells for imaging, we first scraped the cells off of SPA plates 632 and mixed them with 3 microliters of lectin. We then plated them on glass slides and 633 covered with a glass coverslip. 634 For all experiments, we imaged the cells on an LSM-780 (Zeiss) AxioObserver confocal 636 microscope with a 40X C-Apochromat water-immersion objective (Zeiss, NA= 1.2) or a 637 40X LD C-Apochromat water-immersion objective (Zeiss, NA=1.1). We acquired images 638 of every field of cells in two ways. We acquired a channel mode image to obtain a 639 transmitted light image. For the images in Figure 5C, we acquired the fluorescence 640 images by exciting GFP at 488 nm and collecting its emission between a 500-553 641 bandpass filter, while we excited mCherry at 561 nm and collected its emission between 642 a 562-615 nm bandpass filter. For all other images, we acquired images in lambda 643 mode over the entire visible range, with GFP and mCherry excitation at 488 and 561 644 nm, respectively, to obtain the true fluorescence. To eliminate cross talk, we collected 645 the GFP and mCherry lambda images separately. To distinguish true GFP and mCherry 646 signal from autofluorescence, we linearly unmixed the lambda mode data for GFP and 647 mCherry using reference GFP and mCherry images and an in-house plug-in in ImageJ 648 (https://imagej.nih.gov/ij/). 649 650

Analysis of meiotic proteomic data 651
We used the data set collected by Krapp et al and determined the relative protein levels 652 following the method described in (27). We considered all Wtf proteins that were 653 detected in at least one timepoint in at least one of the three replicate experiments. We 654 plotted the mean of the quantified Wtf protein for each timepoint that had at least two 655 replicates, regardless of the number of replicates at that specific timepoint. When a Wtf 656 protein was detected in only a single replicate, we plotted that value for that timepoint. 657

Alignments 659
To determine the DNA and amino acid sequence identity shared by wtf genes and 660 proteins, we aligned the sequences using Geneious Prime (https://www.geneious.com). 661 We used the Geneious alignment tool and performed a global alignment with free end 662 gap using the default parameters. For DNA sequence alignments, the parameters we 663  and Sk chromosomes are shown in red. * indicates a p-value<0.02 (G-test). We 861 genotyped at least 200 haploid offspring for each diploid. We tested some of the wtf 862 transgenes using multiple independent strains (i.e. one in which the transgene was 863 marked with kanMX4 and one marked with hphMX6) and we present the combined 864 data. The complete raw data are presented in Supplemental Table 1. 865 [fertility]). We genotyped more than 200 haploid spores for each diploid. Spores that 877 inherited both markers at ade6 (Ade+ G418 R , Ade+ HYG R , or G418 R HYG R ) were 878 excluded from the analyses as they are likely aneuploid or diploid. We present the 879 complete raw data in Supplemental Tables 1 and 2. 880  shown. We found no significant difference between diploid 30 and diploid 31 using a G-927 test. For (A) and (D), we genotyped more than 200 spore progeny for each diploid and 928 the complete raw data are presented in Supplemental Table 1 The complete raw data is presented in Supplemental Table 3.  represents the diploid number, which matches the numbers in Figure 2, Figure 3, and 992 Supplemental Figure 3. In columns C2-C5, the strain number (SZY) and relevant 993 genotype of the haploid parent strains used to determine the allele transmission at the 994 drive locus (ade6 or wtf locus) are shown. Sp, Sk, CBS5557, and FY29033 alleles are 995 labeled in blue, red, yellow, and green, respectively. Columns C6-C8 indicate which 996 phenotypes were followed at the control locus (ura4) and the number of progeny that 997 showed the indicated phenotypes. Columns C9 and C10 indicate the phenotypes that 998 were followed at the drive loci (ade6 or wtf locus) and the number of haploid progeny 999 that exhibited the indicated phenotypes. Some of the progeny inherited both markers 1000 from the parent strains at the ade6 locus. The number of the progeny that inherited both 1001 markers is presented in column C11 and the percentage of the progeny with this 1002 phenotype is shown in column C12. These progeny were excluded from the data 1003 presented in Figure 2, 3, and Supplemental Figure 3. Column C13 shows the fraction of 1004 the haploid progeny that inherited the genotype of allele 1. Column C14 shows the 1005 fraction of the haploid progeny that inherited the genotype of allele 2. Column C15 1006 shows the total progeny assayed excluding the progeny that inherited both genetic 1007 markers at ade6. Column C16 shows the total progeny. Column C17 shows the total 1008  number shows the SZY numbers of the haploid parent strains. We present all the viable 1018 spore yield values from independent assays. We normalized diploids 10, 20, and 24 to control diploid 22. We normalized diploids 8, 25, and 26 to diploid 23. We previously

Supplemental Table 4. Yeast strains used in this study. 1042
The strain SZY643 contains the wtf18-2 allele (18). 1043 Supplemental Table 6. Plasmids used in this study. 1053 Supplemental Table 7. Oligos used in this study. 1055  Phenotypes of wtf+/wtfdiploids. Allele transmission of (A) class 1 wtf gene and (B) class 2 wtf gene heterozygotes. We used the kanMX4 or hphMX6 drug resistance markers (drug R ) linked to the wtf allele of interest and the ade6 marker to follow allele transmission. For diploids 1-20, we excluded the spores that had inherited both ade6+ and drug R markers as these are likely aneuploid or diploid. For diploid 21, allele transmission was assayed using only the drug marker. Diploids 1-4,7,9,10,12,13,15, and 17-20 were compared to control diploid 22; diploids 5,6,8,11,14, and 16 were compared to the control diploid 23. Sp chromosomes are depicted in blue and Sk chromosomes are shown in red. * indicates a p-value<0.02 (G-test). We genotyped at least 200 haploid offspring for each diploid. We tested some of the wtf transgenes using multiple independent strains (i.e. one in which the transgene was marked with kanMX4 and one marked with hphMX6) and we present the combined data. The complete raw data are presented in Supplemental Table 1.  and Sk (red) diploids with the indicated transgenes integrated at ade6. Allele transmission was determined by following the genetic markers (ade6 and drug resistance) linked to each allele. Diploids 10, 20, and 24 were compared to control diploid 22, while diploids 8, 25, and 26 were compared to control diploid 23. The data for diploid 25 was previously published in (18). We normalized the fertility values to the control diploid and reported them as a percentage. * indicates p-value of <0.05 (G-test [allele transmission] and Wilcoxon test [fertility]). We genotyped more than 200 haploid spores for each diploid. Spores that inherited both markers at ade6 (Ade+ G418 R , Ade+ HYG R , or G418 R HYG R ) were excluded from the analyses as they are likely aneuploid or diploid. We present the complete raw data in Supplemental Tables 1 and 2.  Wtf35. Using this method, the percent identity between these two proteins is 24%.  Allele transmission into spores from diploids heterozygous for class 3 wtf genes is shown. No significant differences in the allele transmission were found using a G-test when compared to the control ura4 locus. (B) Fertility of wild-type and mutant backgrounds (normalized to wild-type). No significant differences in the viable spore yield were found using a Wilcoxon test. The complete raw data are presented in Supplemental Table 2. (C) Serial dilutions of strains with the denoted genotype were spotted onto YEA plates. The slight growth advantage of the quadruple mutant is likely due to the fact that the mutant is a uracil prototroph while the others are uracil auxotrophs. All other auxotrophies are matched between the strains. (D) Allele transmission of the Sk wtf4 meiotic driver in a wild-type and mutant background is shown. We found no significant difference between diploid 30 and diploid 31 using a Gtest. For (A) and (D), we genotyped more than 200 spore progeny for each diploid and the complete raw data are presented in Supplemental  (B) Allele transmission of Sk chromosome 1 in rec12diploids. We genotyped more than 400 haploid spore progeny from each diploid. * indicates a p-value<0.05 (G-test). The complete raw data is presented in Supplemental Table 3.

Sp Wtf18
Sp

Color Key
Sp Wtf18-2 CPGALKRMPKFIRNGIASFLGGLGNAFGGIGNAFGGIGNAIGRIGNAFRGANDNNDIPLGEMDVESEV FY29033 Wtf36 CPGALKRMPKFIRNGIASFLEGIGN----IGNAIGRIGNAIGRIGNAFRGANDNNDIPLGEMEVESEV ******************** * ** **** * ************************** ***** Sp Wtf18-2 CPGALKRMPKFIRNGIASFLGGLGNAFGGIGNAFGGIGNAIGRIGNAFRGANDNNDIPLGEMDVESEV Sp Wtf13 RPGALKRMPKFIGNGIASFLGGLGNAFGGIGNAFGGIGNAIGRIGNAFRGANDNNDIPLGEMDVESEV *********** ******************************************************* B Alignment of the C-terminal region of FY29033 Wtf36 and the closest non-suppressor C Alignment of the C-terminal region of Sp Wtf13 and its known suppressor Supplemental Figure 6. Class 3 Wtf proteins are present during meiosis. Analysis of Wtf protein levels during meiosis from the data of (27). Due to the high sequence similarity between Sp Wtf4 and Wtf13 proteins, the data points for these two proteins were merged. heterozygous diploids (right) during logarithmic phase and saturation are shown. We linearly unmixed these images and adjusted the brightness and contrast differently for each image and smoothed them using Gaussian Blur. The brightness and contrast were adjusted to observe the background. We verified the green autofluorescence via spectral imaging. The scale bar represents five microns. TL represents transmitted light.   Figure  4A of Sk Wtf7-GFP (cyan) and Nsp1-mCherry (magenta) is shown. We normalized the intensities of the green and red autofluorescence to the intensities of the GFP and mCherry signals, respectively. (B) Linear unmixing of the representative image shown in Figure 4B of Sk Wtf15-GFP (cyan) localization is shown. We normalized the intensity of the green autofluorescence to the intensity of the GFP channel. We adjusted the brightness and contrast differently for each image and smoothed them using Gaussian blur. The scale bar represents five microns and TL represents transmitted light. Linear unmixing of (A) haploids and (B) diploids containing Sk Wtf14-GFP (cyan) and mCherry-AHDL (magenta) from the representative images shown in Figure 5A and 5B (28). We adjusted the intensities of the autofluorescence images to their respective channels. We adjusted brightness and contrast differently for each image and smoothed them using Gaussian blur. TL represents transmitted light. Scale bar represents five microns.  Figure 3. In columns C2-C5, the strain number (SZY) and relevant genotype of the haploid parent strains used to determine the allele transmission at the drive locus (ade6 or wtf locus) are shown. Sp, Sk, CBS5557, and FY29033 alleles are labeled in blue, red, yellow, and green, respectively. Columns C6-C8 indicate which phenotypes were followed at the control locus (ura4) and the number of progeny that showed the indicated phenotypes. Columns C9 and C10 indicate the phenotypes that were followed at the drive loci (ade6 or wtf locus) and the number of haploid progeny that exhibited the indicated phenotypes. Some of the progeny inherited both markers from the parent strains at the ade6 locus. The number of the progeny that inherited both markers is presented in column C11 and the percentage of the progeny with this phenotype is shown in column C12. These progeny were excluded from the data presented in Figure 2, 3, and Supplemental Figure 3. Column C13 shows the fraction of the haploid progeny that inherited the genotype of allele 1. Column C14 shows the fraction of the haploid progeny that inherited the genotype of allele 2. Column C15 shows the total progeny assayed excluding the progeny that inherited both genetic markers at ade6. Column C16 shows the total progeny. Column C17 shows the total number of independent diploids assayed for each cross. The last column (C18) shows the p-value calculated by comparing diploids 1-4,7,9,10,12,13,15,17-20, and 24 to control diploid 22; diploids 5,6,8,11,14,16,25, and 26 to control diploid 23; diploids 21, 27, 28, and 29 to the control ura4 locus; and diploid 31 to diploid 30. We previously published the allele transmission data for diploid 25 (18).

C1
C2 C3  C4  C5  C6 C7  C8  C9  C10  C11  C12  C13  C14  C15  C16  C17  C18  Supplemental Table 2. Raw data for fertility from Figure 3 and Supplemental Figure 3. Each column represents the diploid assayed, which matches the diploid number in Figure 3 and Supplemental Figure 3. The first row underneath the diploid number shows the SZY numbers of the haploid parent strains. We present all the viable spore yield values from independent assays. We normalized diploids 10, 20, and 24 to control diploid 22. We normalized diploids 8, 25, and 26 to diploid 23. We previously published the viable spore yield data for diploid 25 (18). We normalized the average viable spore yield value from the h 90 mating of strain SZY3529 (Sp wtf7Δ, wtf11Δ, wtf14+wtf15Δ) to the average viable spore yield value of the wild-type control strain SZY2254. We calculated the p-value comparison using the Wilcoxon test.  Table 5. Table summary of plasmid constructions. Column 1 lists the wtf gene cloned into each vector. Column 2 denotes the isolate origin of each wtf gene in column 1. The DNA templates and oligos used in the PCR reactions to amplify the wtf alleles are shown in columns 3 and 4, respectively. We digested each of the amplified fragments with the enzymes reported in column 5 and then integrated into the target site listed in column 6. The number of each of the plasmids that we generated is reported in column 7. The description of each plasmid can be found in Supplemental