The wtf4 meiotic driver utilizes controlled protein aggregation to generate selective cell death

Meiotic drivers are parasitic loci that force their own transmission into greater than half of the offspring of a heterozygote. Many drivers have been identified, but their molecular mechanisms are largely unknown. The wtf4 gene is a meiotic driver in Schizosaccharomyces pombe that uses a poison-antidote mechanism to selectively kill meiotic products (spores) that do not inherit wtf4. Here, we show that the Wtf4 proteins can function outside of gametogenesis and in a distantly related species, Saccharomyces cerevisiae. The Wtf4poison protein forms dispersed, toxic aggregates. The Wtf4antidote can co-assemble with the Wtf4poison and promote its trafficking to vacuoles. We show that neutralization of the Wtf4poison requires both co-assembly with the Wtf4antidote and aggregate trafficking, as mutations that disrupt either of these processes result in cell death in the presence of the Wtf4 proteins. This work reveals that wtf parasites can exploit protein aggregate management pathways to selectively destroy spores.

Meiotic drivers are selfish DNA sequences that break the traditional rules of sexual 44 reproduction. Whereas most alleles have a 50% chance of being transmitted into a given 45 offspring, meiotic drivers can manipulate gametogenesis to bias their own transmission into 46 most or even all of an individual's offspring (Burt and Trivers, 2006;Lindholm et al., 2016). This 47 makes meiotic drive a powerful evolutionary force (Sandler et al., 1957). Meiotic drivers are 48 widespread in eukaryotes and the evolutionary pressures they exert are thought to shape major 49 facets of gametogenesis including recombination landscapes and chromosome structure (Crow, 50 1991;Dyer et al., 2007;Larracuente and Presgraves, 2012;Schimenti, 2000;Pardo-Manuel de 51 Villena and Sapienza, 2001;Hammer et al., 1989;C. Grey et al., 2018).

53
Harnessing and wielding the evolutionary power of meiotic drive has the potential to greatly 54 benefit humanity. Engineered drive systems, known as 'gene drives,' are being developed to 55 spread genetic traits in populations (Lindholm et al., 2016;Burt, 2014;Gantz et al., 2015;Esvelt 56 et al., 2014;Burt and Crisanti, 2018). For example, gene drives could be used to spread 57 disease resistance alleles in crops. Alternatively, gene drives can be used to suppress human 58 disease vectors, such as mosquitoes, or to limit their ability to transmit diseases (Lindholm et 59 al., 2016;Burt, 2014;Gantz et al., 2015;Esvelt et al., 2014, reviewed in Burt andCrisanti, 60 2018). While there are many challenges involved in designing effective gene drives, natural 61 meiotic drivers could serve as useful models or components for these systems (Lindholm et al., 62 (Ohira et al., 2017). We then integrated the wtf4 poison -GFP allele at the ura4 locus and the 165 wtf4 antidote -mCherry allele at the lys4 locus of the same haploid strain. Next, we observed the 166 localization of the Wtf proteins relative to vacuole (visualized using the CellTracker Blue CMAC 167 lumen stain) or the ER (using Sec63-YFP) following β-estradiol induction. Similar to our 168 observations in spores, we saw that the Wtf4 poison -GFP and Wtf4 antidote -mCherry proteins largely

176
We also attempted to assay the localization of the Wtf4 antidote and poison proteins individually 177 to test if the localization of the Wtf4 poison was altered in the presence of the Wtf4 antidote , as we 178 observed in spores ( Figure 1C). We found that the localization of the Wtf4 antidote -mCherry to the  (Matsuyama et al., 2010). We failed, however, to 183 generate cells carrying the wtf4 poison -GFP allele without the wtf4 antidote -mCherry allele by 184 transformation or by crossing the strain carrying both wtf4 poison -GFP and wtf4 antidote -mCherry to a 185 wild-type strain (Figure1-figure supplement 3A). This is likely due to leaky expression of the 186 wtf4 poison -GFP from the inducible promoter even without addition of β-estradiol. Overall, our 187 results suggest that the Wtf4 poison protein is toxic in vegetative cells, but the antidote is still  In the process of trying to understand the Wtf4 proteins' localization patterns, we assayed the 194 localization of the Wtf4 proteins relative to the nucleus. For this experiment, we imaged asci 195 produced by wtf4-GFP/ade6 + heterozygotes also carrying a tagged histone allele, hht1-RFP 196 (Tomita and Cooper, 2017). Although we did not observe colocalization of Wtf4 proteins and the 197 condensed ( Figure 1G (younger ascus), see methods). Additionally, in 11 out of 38 asci, one or 199 both of the nuclei in the wtf4spores were disrupted and the nuclear contents were dispersed 200 throughout the spores ( Figure 1G (older ascus)). To address the timing of these nuclear 201 phenotypes, we imaged diploids undergoing gametogenesis using time-lapse microscopy. We 202 saw that all four nuclei tended to look similar shortly after the second meiotic division. As spores 203 matured, however, we observed nuclear condensation sometimes followed by fragmentation in 204 the spores that did not inherit wtf4 (i.e. in spores lacking the enriched GFP expression and Wtf4 antidote , however, limited our ability to explore their mechanisms of action in this system. We 213 therefore tested if the Wtf4 proteins functioned in the budding yeast Saccharomyces cerevisiae.

214
To do this, we cloned the coding sequences of wtf4 poison -GFP and wtf4 antidote -mCherry under the 215 control of β-estradiol inducible promoters on separate plasmids (Ottoz et al., 2014). We then 216 introduced these plasmids into S. cerevisiae individually and together. We found that cells 217 carrying the wtf4 poison -GFP plasmid were largely inviable when Wtf4 poison -GFP expression was 218 induced, indicating the poison is also toxic to S. cerevisiae ( Figure

236
Wtf4 antidote -mCherry, on the other hand, generally localized to one or two large amorphous 237 regions adjacent to the vacuole ( Figure 2D). When co-expressed, Wtf4 poison -GFP and Wtf4 antidote -238 mCherry co-localized to this region next to the vacuole ( Figure 2E). In some cells, a faint circle 239 of Wtf4 poison -GFP could also be seen (likely ER localization); however, the majority colocalized

247
Because the Wtf4 proteins colocalize, we wondered if they physically interact. We tested this 248 using acceptor photobleaching Fluorescence Resonance Energy Transfer (FRET, Sekar et al., 249 2003, Figure 2F) in cells expressing both Wtf4 poison -GFP and Wtf4 antidote -mCherry proteins. This 250 process involves bleaching the fluorescence of a tagged protein (the acceptor) and looking for a 251 corresponding increase in fluorescence of another tagged protein (the donor). If an increase in 252 fluorescence of the donor is observed, the proteins are said to be physically interacting, as they 253 are in close enough proximity (less than 10 nanometers) to transfer energy to each other (Sekar 254 et al., 2003). When we bleached Wtf4 antidote -mCherry, we saw a corresponding increase in 255 Wtf4 poison -GFP emission, supporting the idea that the two proteins physically interact ( Figure

258
The Wtf4 proteins localize as puncta of varying sizes, so we hypothesized that the proteins 259 assemble into aggregates. To explore the nature of the Wtf4 protein assemblies, we utilized the 260 recently developed Distributed Amphifluoric FRET (DAmFRET) assay (Khan et al., 2018). This  Figure 1A). All of the known active Wtf antidote proteins are highly 278 similar to the Wtf poison they neutralize (Bravo Núñez et al., 2020). In addition, mutations that 279 disrupt the similarity between a given Wtf antidote and Wtf poison can eliminate the ability of the 280 Wtf antidote to neutralize the Wtf poison (Hu et al., 2017;Bravo Núñez et al., 2018). Here, we tested 281 the mechanism underlying that requirement using Wtf4 proteins. Given that each Wtf4 protein 282 self-assembles, we hypothesized that homotypic interactions between Wtf4 poison and Wtf4 antidote 283 mediated their co-assembly and neutralization of the poison. To test this idea, we mutated 284 sequences at the C-termini of the inducible wtf4 poison -GFP and wtf4 antidote -mCherry alleles in the 285 S. cerevisiae plasmids described above. Specifically, we targeted our mutagenesis to a seven 286 amino acid repeat sequence (IGNAFRG) that is found in many members of the wtf gene family 287 (Eickbush et al., 2019). We previously showed that a mismatched number of these repeats 288 between a Wtf poison and antidote proteins is enough to disrupt their specificity (Bravo Núñez 289 et al., 2018). The wild-type S. kambucha wtf4 allele contains ~1.5 repeat units ( Figure 3A). To 290 make the mutants, we inserted 18 additional codons into the repeat region of wtf4 to make a 291 total of four repeats. We denote these repeat insertion mutants with an * ( Figure 3A).

293
As expected, the Wtf4 poison *-GFP protein is functional (i.e. toxic) in S. cerevisiae and localizes 294 similarly to the tagged wild-type Wtf4 poison -GFP ( Figure 3B

295
Wtf4 poison *-GFP is neutralized by the matching Wtf4 antidote *-mCherry protein, and the two mutant 296 proteins colocalized in vacuole-associated assemblies, just like the tagged wild-type proteins in 297 S. cerevisiae ( Figure 3B-C). Wtf4 antidote *-mCherry protein on its own also resembled the wild- We next used transmission electron microscopy (TEM) to analyze the environment of Wtf 308 proteins within the vacuole-associated aggregates. Similar to our observations made using 309 fluorescence microscopy, we found using immuno-gold labeling that Wtf4-GFP largely clustered 310 near the vacuole in cells also expressing untagged Wtf4 antidote ( Figure 4A). These images also 311 revealed that the Wtf4 protein aggregates appeared within a cluster of lightly staining organelles

317
To look at these Wtf4 aggregate-associated organelles at higher resolution, we used TEM with a 318 sample preparation method that better maintains cellular morphology (see methods). We found 319 that the organelles were in fact a mix of lipid droplets and large vesicles with bilayer membranes  (Soper et al., 2008). The α-synuclein vesicles, however, appear smaller and more numerous 330 than the Wtf4-associated vesicles. To test if the increase in vesicles and lipid droplets was a 331 common feature of aggregation prone proteins, we expressed (using the β-estradiol system) a

358
Given that the Wtf4 antidote and Wtf poison +Wtf4 antidote aggregates localize adjacent to the vacuole,

359
we hypothesized that they could be at the IPOD in S. cerevisiae. To test this idea, we looked for 360 the localization of the Wtf4 proteins relative to Rnq1-mCardinal and GFP-Atg8. Rnq1 localizes 361 to the IPOD and Atg8 is a component of the pre-autophagosomal structure that is adjacent to 362 the IPOD (Kaganovich et al., 2008;Tyedmers et al., 2010, Rothe et al., 2018. Consistent with 363 our hypothesis, we found that Wtf4 antidote -mCherry either colocalized or was adjacent to Rnq1-

369
Proteins in the IPOD tend to be insoluble (Kaganovich et al., 2008;Bagola et al., 2008). To test 370 if the Wtf4 antidote shared this property in S. cerevisiae, we used half punctum-Fluorescence

371
Recovery After Photobleaching (half-FRAP) (Khan et al., 2018;Zhang et al., 2015). This 372 analysis revealed that the Wtf4 antidote -mCherry aggregate has very low internal mobility and is 373 thus more solid-like than liquid-like ( Figure 5B). We were curious if the Wtf4 antidote behaved 374 similarly in its native context. To test this, we performed the half-FRAP assay on the Wtf4 antidote -

388
Amongst our hits, the only significantly enriched (FDR p< 0.05) gene ontology groups were 389 mitochondrial translation and organization (Figure 5-source data 1). We speculate this 390 enrichment is due to two known roles of mitochondria in managing protein aggregates. The first 391 is the Mitochondria As Guardian In Cytosol (MAGIC) mechanism by which mitochondria help 392 degrade protein aggregates (Ruan et al., 2017). The second is that mitochondria mitigate the 393 impact of toxic aggregates by promoting asymmetric aggregate segregation in mitosis (Zhou et 394 al., 2014).

396
We also identified genes involved in Cell Wall Integrity (CWI) pathways (POP2, MPT5, SLT2 397 and BCK1) as necessary for survival after induction of Wtf4 antidote and Wtf4 poison (Jin et al., 2015; 398 Li et al., 2016;Stewart et al.,2007). The CWI pathway is triggered by diverse stress stimuli 399 (Fuchs and Mylonakis, 2009) and can promote stress-response gene expression and nuclear 400 into the cytoplasm promotes mitochondrial hyper-fission, stress response gene activation, and 402 either apoptosis or repair of the stress-induced damage (Jin et al., 2015). Consistent with this,  (Rothe et al., 2018;He et al., 2006;.

417
Given our results, which suggest that Wtf4 protein localization is an important factor in mitigating 418 toxicity, we next imaged the localization of Wtf4 poison -GFP and Wtf4 antidote -mCherry in all of the 419 screen hits. We found that the localization of the Wtf4 poison -GFP and Wtf4 antidote -mCherry proteins 420 was disrupted in all 106 hits relative to wild type (where the proteins coalesce to the IPOD). In 424 2B-C). We noted that there were often cells with dispersed Wtf4 antidote -mCherry aggregates or 425 cells with dispersed Wtf4 poison -GFP aggregates, but rarely cells with both. We speculate this is 426 due to toxicity of distributed aggregates and cells expressing both aggregates at the same time 427 being destroyed quickly. Another common feature we saw throughout the screen hits was 428 Wtf4 antidote -mCherry signal in the vacuole. We also observed this vacuolar localization in the C-429 terminal mutants depicted in Figure 3D-E, so this appears to be a common feature of the

436
Because the Wtf4 antidote protein is quite similar to the Wtf4 poison and also assembles into 437 aggregates, we were curious if the Wtf4 antidote alone was toxic in the absence of any of our 438 screen hits. We therefore assayed the viability of the 106 deletion mutants when only Wtf4 antidote 439 was expressed. We saw that in approximately half (44/106) of the deletion stains, Wtf4 antidote

444
We also investigated one hit from our screen, VPS1, more thoroughly using our β-estradiol-445 inducible system (described above). VPS1 is a dynamin-like GTPase that is necessary for 446 trafficking of aggregates to the IPOD and/or other inclusions sites (Kumar et al., 2016;Kumar et 447 al., 2017;Hill et al., 2016, Marshall et al., 2016. In the absence of VPS1, we found that the

453
Together, these experiments indicate that the physical interaction between the Wtf4 poison and 454 Wtf4 antidote proteins is insufficient to neutralize the toxicity of Wtf4 poison protein aggregates.

455
Sequestering the aggregates to a vacuole-associated inclusion is also required. Interestingly,

456
we also observed enhanced toxicity of the Wtf4 antidote -mCherry protein in the absence of Vps1

481
aggregates were largely dispersed throughout the cytoplasm, with some ER localization in S.

482
cerevisiae. The assembly of Wtf4 proteins is reminiscent of another meiotic drive element, Het-483 s, which employs prion-like amyloid polymerization to convert Het-S proteins to a lethal form 484 (Dalstra et al.,2003;Riek and Saupe, 2016). We therefore evaluated whether Wtf4 poison proteins 485 exhibit prion activity in S. cerevisiae using DAmFRET (Khan et al., 2018). We found that 486 Wtf4 poison -mEos proteins assembled with themselves even at very low expression levels 487 (Figure2-figure supplement 3C). In fact, we were unable to detect cells that lacked self-488 assemblies, revealing that the toxic form of the protein is not appreciably supersaturated, as 489 would be required for Wtf4 antidote to detoxify it through a simple prion-like mechanism.

490
Nevertheless, the sequence-dependent self-assembly of Wtf4 remains consistent with amyloid 491 polymerization. However, given its intimate association with vesicles, extensive testing would be 492 required to further evaluate the structural basis of Wtf4 activity.

494
The significance of the Wtf4 poison aggregation is not clear. We speculate that the aggregation 495 propensity is intimately tied to the toxicity of Wtf4 poison . We propose that distributed Wtf4 496 aggregates interact broadly with other proteins and disrupt their folding or localization.

497
Compounding effects of these hypothesized interactions could disrupt protein homeostasis or 498 cellular integrity, leading to cell death. This death may occur via a programmed cell death 499 pathway, as in both S. pombe gametogenesis and in vegetative S. cerevisiae, cells succumbing 500 to the Wtf4 poison exhibit nuclear condensation (followed by nuclear fragmentation in S. pombe).

501
cell survival upon expression of the Wtf4 proteins. Testing these ideas may be challenging, 503 especially if understanding Wtf4 poison toxicity proves to be as elusive as understanding the 504 intensely studied neurotoxic aggregating proteins TDP-43 and α-Synuclein (Johnson et. al, 505 2011;Cookson et. al, 2007).

521
When we disrupted the ability of S. cerevisiae cells to transport the Wtf4 antidote aggregates with 522 the vps1∆ mutation, we found that the Wtf4 antidote aggregates were distributed and more toxic 523 than in wild-type cells. This is consistent with the idea that a key feature of Wtf4 protein toxicity 524 relies in the aggregates being widely dispersed in the cytoplasm. When Wtf4 poison and Wtf4 antidote 525 are found together in wild-type cells, the proteins co-assemble into aggregates. The co-526 assembled aggregates then behave similarly to the Wtf4 antidote aggregates and are trafficked into 527 the vacuole (in S. pombe cells) or to the IPOD adjacent to the vacuole (in S. cerevisiae cells) 528 where they cause limited toxicity. Also, like the Wtf4 antidote aggregates, the toxicity of the 529 Wtf4 poison +Wtf4 antidote co-assembled aggregates is greatly enhanced if aggregate transport to the 530 vacuole is disrupted by mutations (e.g. vps1∆).

532
Together, our observations suggest a mechanistic model for wtf4 function. In this model, wtf4 533 exploits protein aggregation control pathways to induce selective cell death. The Wtf4 poison forms 534 distributed toxic aggregates and the Wtf4 antidote co-assembles with the Wtf4 poison and neutralizes 535 of any other meiotic driver described to date (Grognet et al., 2014;Didion et al., 2015;Long et 537 al., 2008;Dawe et al., 2018;Rhoades et al., 2019;Dalstra et al., 2005;Hammond et al., 2012; 538 Vogan et al., 2019, Chen et al., 2008Akera et al., 2017;Bauer et al., 2012;Pieper et al., 2018; 539 Herrmann et al., 1999;Shen et al.,2017;Yu, et al., 2018;Bauer et al., 2007;Wu et al., 1988;540 Xie, et al., 2019;Kruger et al., 2019;Lin et al., 2018), but there are very few mechanistically 541 characterized gamete-killing drive systems (reviewed in Bravo Núñez et al., 2018). This study focused on the wtf4 meiotic driver. There is, however, an incredibly diverse array of 551 wtf genes that cause meiotic drive. For example, the poison protein encoded by wtf35 (from the 552 FY29033 isolate) shares less than 23% amino acid identity with Wtf4 poison (Bravo Núñez et al., 553 2020). Despite that extreme divergence, both genes cause essentially the same phenotype: 554 drive of the gene into > 90% of the progeny of a heterozygote (Bravo Núñez et al., 2020). The 555 conserved protein aggregation model offers an explanation for how such a diverse array of 556 proteins can cause the same phenotype. Under our model, the mechanism of the Wtf poison 557 proteins is dependent upon their aggregation propensity. Presumably, the evolution of a protein 558 that must self-aggregate could be less constrained than the evolution of a protein that must 559 maintain a specific enzymatic activity or interaction partner.

587
Not all of our screen hits, however, are in genes or pathways with annotated roles that clearly fit 588 our model. Some of the genes have no annotated functions. It is possible that at least some of 589 these genes are not directly involved in aggregate management, but the mutants are especially 590 sensitive to the stresses imposed by Wtf4 aggregates. It is also possible that some of the genes 591 do have roles in mitigating the effects of toxic aggregates. Indeed, in deletions of some genes 592 with unknown functions, we saw distributed Wtf4 aggregates, suggesting these unknown 593 proteins could play a role in sequestration of aggregates. Interestingly, other hits are in well-594 studied genes, such as multiple acetyltransferases and various kinetochore proteins. Future 595 analysis of these hits will be essential to refine or to potentially reject our current model.

597
Insight into protein cellular response to aggregates via studying meiotic drive 598 Studying how parasites manipulate their hosts can uncover unexpected insights on the host's 599 biology. For example, studies of the mouse t-haplotype meiotic driver revealed that gene 600 expression in spermatids can create sperm-autonomous phenotypes, even though spermatids 601 are connected by intercellular bridges (Herrmann et al., 1999). Under our model, a fine line 602 exists between protein aggregates that cells can manage (i.e. Wtf4 antidote ) and lethal aggregates 603 exploited this feature for its own selfish advantage. Future studies can now exploit the Wtf4 605 proteins to learn about protein aggregate toxicity and cellular aggregate management 606 strategies.

610
We confirmed all the vectors we generated (described below) via sequencing.

612
Generation of tagged wtf4 alleles for expression in S. pombe gametogenesis 613 Generation of a vector containing wtf4 antidote -mCherry expressed from the endogenous promoter:

622
Generation of a vector containing wtf4 antidote -GFP expressed from the endogenous promoter:

623
We amplified the upstream sequence and the beginning of the wtf4 allele from pSZB203 624 (Nuckolls et al., 2017) using oligos 620+736. We amplified the rest of the wtf4-GFP sequence 625 (with an ADH1 transcriptional terminator) from pSZB203 using oligos 735+634. Oligos 734 and 626 735 introduced mutations that interrupt the Wtf4 poison start site within intron 1. We then used 627 overlap PCR with oligos 620+634 to unite the two pieces. We digested the complete wtf4 antidote -628 GFP cassette with SacI site and cloned it into the SacI site of pSZB188 (Nuckolls et al., 2017) to 629 generate pSZB260.

631
Generation of a vector containing the predicted wtf4 antidote coding sequence expressed from the 632 endogenous promoter: We amplified the wtf4 coding sequence in three pieces. We amplified the 633 promoter with oligos 633+604 using SZY13 DNA as a template. We amplified the coding 634 sequence from a gBlock DNA fragment (Integrated DNA Technologies, Inc., Coralville) using 635 oligos 605+614. We amplified the sequence downstream of wtf4 using oligos 613+635 and 636 SZY13 genomic DNA as a template. We then stitched the three pieces together using overlap 637 PCR with oligos 633+635. We then digested the product with SacI and ligated the cassette into 638 SacI-digested pSZB188 (Nuckolls et al., 2017) 15 to generate pSZB199. Intron 5 was predicted 639 wrong, so there is a mutation at the C-terminus. Within this study, this plasmid was only used to 640 with the PCR oligos.

643
S. pombe Z3EV β-Estradiol inducible system 644 Z3EV promoter system is a titratable inducible promoter system (Ohira et al., 2017). The system 645 requires the Z3EV transcription factor and a Z3EV-responsive promoter (Z3EVpr). β-estradiol 646 induces nuclear import of the Z3EV protein; therefore, genes placed immediately downstream of 647 Z3EVpr in a strain expressing Z3EV become expressed upon β-estradiol addition to the media.

649
Background strain construction: To integrate the Z3EV transcription factor at the leu1 locus of S. 650 pombe, we digested plasmid pFS461 (Addgene #89064, Ohira et al., 2017) with XhoI and 651 transformed it into the yeast strain SZY643 (selecting for Leu+) via standard lithium acetate 652 protocol (Gietz, et al., 1995). This generated the yeast strain SZY2690, into which we 653 transformed all of the proteins with Z3EV promoters (see below).

655
Generation of a strain that expresses wtf4 antidote -mCherry under the control of a β-estradiol 656 inducible promoter: We amplified the Z3EVpr from pFS478 (Addgene #89066, Ohira et al., 2017) 657 using oligos 1734+1735. We then amplified the wtf4 antidote -mCherry sequence (with an ADH1 658 transcriptional terminator) from pSZB891 (described above) using oligos 1738+634. We used 659 overlap PCR to add the Z3EV promoter piece to the wtf4 antidote -mCherry piece using oligos 660 1738+634. We then digested this cassette with SacI and ligated it into the SacI site of pSZB322 661 (Bravo Núñez et. al, 2018), a lys4 integrating vector with a hphMX6 cassette, to create 662 pSZB892. We cut pSZB892 with KpnI and integrated into the lys4 locus of SZY2690 to create 663 SZY2740.

665
Generation of a strain that expresses wtf4 antidote -mCherry and wtf4 poison -GFP under the control of 666 β-estradiol inducible promoters: To create an estradiol inducible Wtf4 poison -GFP vector, we 667 amplified the Z3EVpr on pSZB892 (see above) using oligos 1734+2068. We amplified the 668 Wtf4 poison -GFP (with an ADH1 transcriptional terminator) from pSZB203 (Nuckolls et al., 2017) 669 using oligos 2069+634. We then completed overlap PCR (using oligos 1734+634) on the two 670 pieces. We then digested the completed wtf4 poison -GFP cassette with SacI and ligated it into the 671 SacI site of pSZB331 (see above) to create pSZB975. We cut pSZB975 with KpnI and 672 integrated into the ura4 locus of SZY2740 to generate SZY2888.

674
For the above transformations, we used high-efficiency, lithium acetate transformation protocol 675 (Gietz, et al., 1995) to integrate the vectors, selecting first for drug resistance and then Hygromycin B (to select against pop-outs of the lys4 and ura4 integrating plasmids described 682 above). The next day, we diluted 1 ml of each saturated culture into 4 mls of fresh 683 YEL+G418+HYG media. We then added β-estradiol (from VWR, #AAAL03801-03) to a final 684 concentration of 100 nM and shook the cultures at 32°C for four hours. We then used these 685 induced cultures for imaging (see below for microscopy details).

707
Generation of a vector containing wtf4 poison -GFP under the control of a β-estradiol inducible 708 promoter: We amplified the LexApr from on FRP1642 (Addgene #58442, Ottoz et al. 2014) 709 using oligos 1195+1240. We cloned the promoter into the KpnI/XhoI sites of pSZB464 (see 710 below) to create pSZB585.

742
Induction of Wtf4 proteins with β-estradiol: For imaging, we grew 5 mL saturated overnight 743 cultures in SC -His -Ura -Trp (without agar). The next day, we diluted 1 ml of the saturated 744 culture into 4 mls of media of the same type. We then added β-estradiol to a final concentration 745 of 500 nM and shook the cultures at 30°C to induce. Cells were induced for four hours and then 746 imaged at one or multiple timepoints, depending on the experiment. For spot assays, we diluted 747 saturated cultures to an OD of ~1, then serial diluted (10 0 ,10 -1 ,10 -2 ,10 -3 ,10 -4 ) in a 96-well plate.

754
We amplified the beginning of the wtf4 antidote coding sequence from pSZB388 (see below) using 755 oligos 1065+678 and amplified the rest of wtf4 antidote (and the CYC1 transcriptional terminator) 756 from pSZB392 (see below) using oligos 679+964. We then used overlap PCR with oligos 757 1065+964 to join the two pieces. We then digested the complete wtf4 antidote cassette with XhoI 758 and BamHI and ligated them into XhoI+BamHI-digested pDK20 (DasGupta et. al, 1998) to 759 generate pSZB497.

761 762
Generation of a vector containing wtf4 poison -GFP under the control of a galactose inducible 763 promoter: We first amplified wtf4 poison followed by a CYC1 terminator from pSZB388 (see below) 764 using oligos 963+964. We then digested the PCR product with XhoI and BamHI and ligated it 765 into XhoI+BamHI-digested pDK20 (DasGupta et. al, 1998). This created pSZB392, a URA3 766 integrating vector with wtf4 poison under the control of a GAL promoter. We then amplified 767 wtf4 poison (including the GAL promoter) from pSZB392 with oligos 1045+606 and amplified GFP 768 followed by an ADH1 transcriptional terminator from pSZB203 (Nuckolls et al., 2017) using 769 oligos 998+1040. We then stitched those two PCRs together using overlap PCR (amplifying 770 with oligos 1040+1045). Finally, we digested the PCR product with KpnI and BamHI and cloned 771 it into KpnI+BamHI-digested pRS316 (Sikorski et al, 1989) to generate pSZB464 and into 772 KpnI+BamHI-digested pRS314 (Sikorski et al, 1989) to generate pSZB463.

774
Generation of a vector containing wtf4 antidote -mCherry under the control of a galactose inducible 775 promoter: We amplified the wtf4 antidote sequence with the galactose inducible promoter from 776 pSZB497 (see above) using oligos 1929+997 and the wtf4 antidote -mCherry sequence (with a 777 CYC1 terminator) from pSZB708 (see above) using oligos 1072+964. We then used overlap 778 PCR using oligos 1929+964 to combine the two pieces. We then digested the complete 779 wtf4 antidote -mCherry cassette with BamHI and ligated it into BamHI-digested PRS315 (Sikorski et 780 al, 1989) 63 to generate pSZB1005.

782
Generation of a vector containing wtf4 poison coding sequence: We amplified the wtf4 poison coding 783 sequence from pSZB199 (see above) using oligos 916+926. We then digested the PCR product 784 generate pSZB388. Within this study, this plasmid was only used to build other plasmids.

787
Induction of Wtf4 proteins with galactose: For imaging, we grew 5 ml saturated overnight 788 cultures in SC media lacking appropriate amino acids for selection of the plasmids. The next 789 day, we pelleted the cultures, resuspended in YP raffinose media, and grew overnight. The next 790 day, we diluted 1 ml of the saturated raffinose culture into 4 mLs of SC galactose media lacking 791 amino acids for selection of plasmids. We then added β-estradiol to a final concentration of 500 792 nM and shook the cultures at 30°C for four hours to create induced samples. For spot assays, 793 we diluted saturated cultures to an OD of ~1, then serial diluted (10 0 ,10 -1 ,10 -2 ,10 -3 ,10 -4 ) in a 96-794 well plate. We then spotted 10 μl of each dilution onto both SC media (lacking amino acids 795 appropriate for selection of the plasmids) and SC galactose media lacking the same amino 796 acids. We grew the plates 2 to 3 days at 32°C and imaged them on a SpImager (S&P Robotics).

Construction of the ER marker in S. pombe 799
To create the Sec63-YFP strain, we PCR amplified the C-terminus of sec63 (using oligos 800 939+941) and the sequence downstream of sec63 (using oligos 945+946) using SZY643 as a 801 template. We also amplified a YFP-HIS3 cassette from pYM41 (Janke et al., 2004) using oligos 802 944+943. We then used overlap PCR (using oligos 939+943) to unite those three PCR 803 products. We then transformed this PCR product into GP1163 with standard lithium acetate 804 protocol (Gietz, et al., 1995) (selecting for His+) to integrate the tagged sec63-YFP at its 805 endogenous locus to generate SZY1277. We confirmed the strain via PCR using oligos 806 2037+2038.

Generation the IPOD marker for expression in S. cerevisiae 809
Generation of a vector containing RNQ1-mCardinal under the control of a β-estradiol inducible 810 promoter: We amplified the LexApr from pSZB708 using oligos 1835+1834, the RNQ1 811 sequence from pDK412 (Kryndushkin et. al, 2012) using oligos 1833+1832 and the mCardinal-812 CYC1 terminator from V08_mC (a gift from the Halfmann lab) using oligos 1831+964. We then 813 used overlap PCR (using oligos 1835+964) to stitch the three pieces together. We then digested 814 the cassette with BamHI and ligated it into BamHI-digested pRS315 (Sikorski et al, 1989) to 815 generate pSZB942.

831
688+1280 and pSZB203 as a template. We amplified the rest of wtf4 and the downstream 832 sequence using oligos 1281+686 and pSZB203 as a template. We used overlap PCR using 833 oligos 688+686 to join the two pieces. We then digested the PCR product with SacI and cloned 834 it into the SacI site of pSZB386 (Bravo Núñez et al., 2018) to generate pSZB647. Within this 835 study, this plasmid was only used to build other plasmids.

Generation of the wtf4 exon 6 mutant alleles for expression in S. cerevisiae 838
Generation of a vector containing wtf4 antidote *-mCherry under the control of a β-estradiol 839 inducible promoter: We amplified the beginning of wtf4 antidote (using oligos 1402+1021) from 840 pSZB700 (see above), the mutated section of wtf4 antidote * (using oligos 1072+997) from 841 pSZB647 (see above) and mCherry-CYC1 terminator (using oligos 998+964) from pSZB708 842 (see above). We then stitched the three pieces together (using 1402+964) to generate the 843 complete wtf4 antidote *-mCherry cassette and digested it with XhoI and BamHI. We also digested 844 the LexApr from pSZB708 using KpnI and XhoI. We then cloned those digested pieces into 845 KpnI-BamHI digested pRS314 to generate pSZB774.

864
Generation of vector containing wtf4 antidote -mEos3.1 under the control of a galactose inducible 865 promoter: We used Gibson assembly (New England Biolabs) using oligos rh1282+rh1283 to 866 insert a sequence that encodes the 45 amino acids of the codon-optimized wtf4 exon1 into the 867 Aar1-digested rhx1389 (see above) to create pSZB1120.

869
Generation of a vector containing wtf4 poison -mEos3.1 under the control of a β-estradiol inducible 870 promoter: We amplified wtf4 poison -mEos3.1 with a CYC1 terminator sequence (using oligos 871 1466+964) from rhx1389 (see above) and digested with BamHI. We then ligated the cassette 872 into the BamHI site of pSZB668 to generate pSZB732.

874
Generation of a vector containing wtf4 antidote -mEos3.1 under the control of a β-estradiol inducible 875 promoter: We amplified the wtf4 antidote -mEos3.1 with a CYC1 transcriptional terminator (using 876 oligos 1465+964) from pSZB1120 (see above) and digested with BamHI. We then ligated the 877 cassette into the BamHI site of pSZB670 (see above) to generate pSZB756.

879
Generation of a vector containing wtf4 antidote under the control of β-estradiol inducible promoters:

883
DAmFRET 884 We induced samples of SZY2072, SZY2070, SZY2159, and SZY2059, with β-estradiol as 885 described above. We then aliquoted these induced samples into a 96-well plate. We then 886 partially photoconverted the mEos3.1 protein by exposing the plate, while shaking at 800 RCF, 887 to 405 nm illumination for 25 mins using an OmniCure® S1000 fitted with a 320-500 nm (violet) 903 heterozygous diploids as previously described (Nuckolls et al., 2017). We placed the diploids on 904 sporulation agar (SPA, 1% glucose, 7.3 mM KH2PO4, vitamins, agar) for 2-3 days. We then 905 scraped the cells off of the SPA plates and onto slides for imaging.

907
For vegetatively growing samples ( Figure 1E  922 923 (Slaughter et al., 2013). Briefly, we drew a segmented line (width of two pixels) throughout the 925 spore, randomly covering as much of the spore as we could. We then used an in-house custom 926 written plugin for ImageJ (https://imagej.nih.gov/ij/) to generate a two-color line profile. We

931
To quantify nuclear size, we calculated the full width at half maximum of the fluorescence 932 intensity of RFP. We quantified 42 spores that inherited wtf4-GFP and 19 that did not, all from a 933 wtf4-GFP/ade6+ heterozygote after 2 days on SPA media. We excluded any nuclei that 934 appeared to have already fragmented. . We imaged using a Zeiss Observer.Z1 wide-field microscope with a Flash 4.0 using μManager software. We acquired the mCherry with BP 530-585 nm excitation 961 and LP 615 emission, using an FT 600 dichroic filter, acquiring images every 10 minutes.

Budding yeast microscopy 964
For all budding yeast images except for the two experiments described below, we induced

1008
In S. pombe: We placed SZY1142/SZY1049 heterozygous diploids on SPA plates for 2 days.

1009
We then scraped the sample off of the SPA plates into a 35 mm glass bottom poly-D-lysine 1010 coated dish (MatTek corporation) and carried out half-FRAP as above (n=10 spores), except 1011 that recovery images were then acquired for three minutes total time.

Electron Microscopy 1014
We made 50 mL saturated overnight cultures of SZY1821, SZY1952, SZY1954, and SZY2731 1015 in SC media lacking histidine, tryptophan, and uracil (to select for retention of the plasmids). The 1016 next day, we diluted 10 mLs of the saturated cultures into 90 mLs of the same media with 500 1017 nM β-estradiol. We shook these cultures for four hours at 30°C, reaching log phase. We then 1018 pelleted the yeast cells by filtering and carried out high pressure freezing with the Leica ICE 1019 system (Leica Biosystems). We further processed the frozen cell pellets by freeze substitution 1020 (FS) using acetone containing 0.2% uranyl acetate (UA) and 2% H2O was used as FS medium.

1025
For Immuno-EM, we used HM-20 resin. We cut 60 nm sections with a Leica Ultra microtome 1026 (Leica UC-6) and picked up onto a carbon-coated 150 mesh nickel grid. The grids were labeled 1027 with anti-GFP primary antibody (a gift from M. Rout, Rockefeller University, New York, NY) and 1028 12 nm colloidal gold goat anti-Rabbit secondary antibody (Jackson Immuno Research 1029 minutes. We acquired images using a FEI Tecnai Biotwin electron microscope. For non-1031 immuno-EM, we used Epon resin to better maintain morphology, but the rest of the procedure 1032 was the same. We analyzed the tomographs of at least 10 cells per condition.

1034
For quantification purposes, we also used completed array tomography. For array tomography,

1045
For model building, the segmentation was done based on intensity and known organelle 1046 structure with Microscopy Image Browser (Belevich et. al, 2016) and with IMOD. We used Amira 1047 (Thermo Fisher Scientific) software for model rendering and visualization.

1049
Further quantification of mitochondrial volumes was performed on selected cells after training a 1050 Unet (Ronneberger et al., 2015). Hand annotation of training data was performed in Fiji. A 1051 suite of internally developed Fiji plugins, macros and CherryPy scripts called DeepFiji (see 1052 below) sent training data to a pair of in-house NVIDIA Tesla-equipped deep learning machines 1053 running Tensorflow. Representative cells were selected, and segmented images inferred using 1054 the same macros and deep learning machines before being aligned using a StackReg variant.

1055
Mitochondrial volumes were quantified in Fiji using the 3D Segmentation tools.

DeepFiji training 1058
DeepFiji is a suite of macros and plugins in Fiji, Python, and CherryPy (a Python web 1059 framework) that enable end users on any machine with a reasonable amount of RAM to request 1060 deep learning training and inference on a remote deep learning box as long as both machines 1061 have access to a shared file system.

1063
First, a user selects example sub-images that span the realm of potential objects, background

1097
Once training is completed, and a reasonable iteration point is found in TensorBoard, the user 1098 can run Inferer.ijm in Fiji on their host machine to apply their model to a new dataset. Inferer will 1099 similarly parse images into sub-images, and contact a deep learning box to initiate processing. 1100 de-window their images. The output outline and mask channels are ranged from 0-1 and 1102 represent probabilities. Typically, thresholding pixel values above 0.5 in the mask channel will 1103 suffice for finding objects of interest. However, in cases with frequent object touching, one can 1104 subtract the outline probability from the mask.

1106
If consistent mistakes are found in the inferred data, the user can annotate them properly using protocol (Gietz, et al., 1995). The transformed collection was then spot inoculated on SC and 1120 SC galactose media lacking leucine and uracil. As a control, a strain expressing the galactose 1121 inducible poison and antidote in a wild-type background was used for comparison. We grew the 1122 plates for three days at 30°C, imaged them using the SpImager (S&P Robotics), and manually 1123 scored growth. For the antidote-only screen, we transformed the galactose driven Wtf4 antidote -1124 mCherry (pSZB1005) plasmid into the 106 hits from the first screen and scored them as 1125 mentioned above.

1127
Confirmation of hits: The initial screen identified 250 strains that grew poorly on inductive media.

1128
To confirm that this poor growth was due to the Wtf4 proteins and not due to the background 1129 strain being sick or a poor grower on galactose media in general, we completed a follow-up 1130 screen. We transformed the 250 strains we identified as "poor growers" with empty [URA3] and

1131
[LEU2] vectors and assayed the strains as above to identify those that grew poorly on galactose 1132 media independent of wtf4 gene expression. We found 106 strains that passed this secondary 1133 screen, which we then called hits. We imaged this 106 strains after a short galactose induction 1134 (~4) to ensure we saw Wtf4 protein 1135 1136 PANTHER overrepresentation Test (Thomas et al., 2003;Thomas et al., 2006). The 1138 background list we used for the analysis was the list of MATa deletion collection strains that we 1139 successfully transformed with our plasmids of interest (n= 4793). We used Fisher's Exact test 1140 and corrected with false discovery rate. We imaged the cells as described above for Gal-

1141
inductions, but we added 80 mg/L adenine to the inducing media to circumvent any potential 1142 autofluorescence introduced by the adenine auxotrophy.         predicted coiled-coil domain (depicted with an orange coil, Lupas, Dyke, and Stock, 1991).

1557
There are six predicted transmembrane domains (depicted by blue lines) (TMHMM model, 1558 Krogh et al., 2001) found throughout the amino acid sequences shared by both proteins. There