Novel requirements for HAP2-mediated gamete fusion in Tetrahymena

The ancestral gamete fusion protein, HAP2, catalyzes sperm-egg fusion in a broad range of taxa dating to the last eukaryotic common ancestor. Remarkably, HAP2 orthologs are structurally related to the class II fusogens of modern-day viruses, and recent studies make clear that these proteins utilize similar mechanisms to achieve membrane merger. To identify factors that may regulate HAP2 activity, we screened mutants of the ciliate Tetrahymena thermophila for behaviors that mimic Δhap2 knockout phenotypes in this species. Using this approach, we identified two new genes, GFU1 and GFU2, whose products are necessary for the formation of membrane pores during fertilization and show that the product of a third gene, namely ZFR1, may be involved in pore maintenance and/or expansion. Finally, we propose a model that explains cooperativity between the fusion machinery on apposed membranes of mating cells and accounts for successful fertilization in T. thermophila’s multiple mating type system.

regulate HAP2 activity, we screened mutants of the ciliate Tetrahymena thermophila for 48 behaviors that mimic Dhap2 knockout phenotypes in this species. Using this approach, we 49 identified two new genes, GFU1 and GFU2, whose products are necessary for the formation of 50 membrane pores during fertilization and show that the product of a third gene, namely ZFR1, 51 may be involved in pore maintenance and/or expansion. Finally, we propose a model that 52 explains cooperativity between the fusion machinery on apposed membranes of mating cells 53 and accounts for successful fertilization in T. thermophila's multiple mating type system. 54 55

Introduction. 56
The conserved transmembrane protein, HAP2 (also known as GCS1 1 ), drives gamete 57 fusion in a vast number of species 2,3 . At the same time, HAP2 orthologs are structurally similar 58 to viral class II fusion proteins 4-6 and catalyze membrane merger by mechanisms that mimic 59 those used by Dengue, Zika, and related viruses for host cell entry. This begins with the 60 activation of HAP2 protomers at the plasma membrane of one cell (typically, the male gamete) 61 leading to insertion of hydrophobic "fusion loops" into the lipid bilayer of the partner cell 62 (usually the female gamete) 7-11 . Activation is followed by trimerization of HAP2 ectodomains 63 and conformational fold-back to deform membranes and bring them close enough to allow 64 them fuse 8-10,12 . 65 While dynamic rearrangements in HAP2 structure generate the forces necessary for 66 gamete merger, mechanistic details around the activation of pre-fusion protomers, the possible 67 transition of hemi-fusion intermediates to full fusion pores, and the expansion of pores to form 68 a single contiguous membrane are only just beginning to emerge. These steps regulate both 69 cell-cell and virus-host cell fusion in other contexts 13-15 and may be critically important for the 70 precise spatiotemporal control of fertilization. 71 In flowering plants and green algae, timely activation of the HAP2 machinery appears to 72 be regulated, in part, by accessory proteins that sequester the fusogen and/or constrain it in a 73 prefusion conformation prior to membrane merger. In the unicellular alga, Chlamydomonas 74 reinhardtii, the lineage-specific adhesion protein, MAR1, associates either directly or indirectly 75 with HAP2 in minus ("male") gametes and is required for its correct expression and localization 76 on a single, apical membrane protrusion known as the minus mating structure 16,17 . Interaction 77 of MAR1 with the broadly conserved GEX2 family member, FUS1, on plus ("female") gametes 78 releases HAP2 from an inactive state allowing it to drive gamete fusion at the site where mating 79 structures of plus and minus gametes adhere 9, 16 . 80 A somewhat analogous situation has been described in the model plant species, 81 Arabidopsis thaliana, where HAP2 is initially sequestered on intracellular vesicles of male 82 gametes through associations with the plant-specific proteins, DMP8/9, until sperm and egg 83 come into close proximity 18  Unlike sexually dichotomous species where the protein is expressed primarily in male 98 gametes 17,27,28 , HAP2 is produced by all seven mating types of T. thermophila and is required in 99 both cells of a mating pair for efficient pore formation to occur 4,25 . While complementary 100 mating types lacking HAP2 can recognize and adhere to one another, the absence of the fusion 101 protein from one cell of a mating pair results in a steep decline in the number of pairs capable 102 of forming fusion pores, while its absence from both cells completely abrogates pore 103 formation 4 . In the absence of fusion pores, mating pairs fail to exchange meiotic pronuclei and, 104 when left undisturbed, separate prematurely without generating progeny 25 . Furthermore, 105 compared to wild-type cells, pairs formed by Dhap2 deletion strains come apart readily when 106 agitated suggesting that junctional pores help to "rivet" these highly motile cells together long 107 enough to allow them to complete fertilization 25,29 . 108 Based on these findings, it seemed reasonable that unstable pairing between mutant 109 mating types might offer a useful screen to identify additional gene products involved in 110 membrane pore formation during fertilization. Using this approach, we identified two 111 hypothetical genes designated GFU1 and GFU2 (Gamete FUsion 1 and 2) that are necessary for 112 HAP2-mediated pore formation in mating T. thermophila. In addition, we show that a third 113 gene, referred to as ZFR1 30 , while necessary for pair stability, is not required for pore opening. 114 The implications of these findings are discussed in terms of possible mechanistic roles for GFU1, 115 GFU2, and ZFR1. Finally, direct observations of HAP2 mutant cell lines provide strong evidence 116 of cooperative interactions between the fusion machinery on apposed cells. A model that 117 explains cooperativity and accounts for the expression of HAP2 in all seven mating types of T. with roughly the same kinetics as WT X WT cultures over the first 6 hr of mating (Fig. 1E). 143 Similarly, mixtures of Dgfu2 deletion strains reached nearly WT levels of pairing (albeit with 144 slower kinetics) over the same time frame (Fig. 1F). In both cases, however, the pairs formed 145 between complementary Dgfu1 X Dgfu1 or Dgfu2 X Dgfu2 deletion strains prematurely exited 146 the sexual cycle and came apart several hours before WT X WT pairs (Fig. 1E,F). Notably, Dgfu1 147 and Dgfu2 deletion strains behaved differently from each other when mated to WT cells. As 148 shown in Figure 1, pairs formed by WT X Dgfu1 cells had an intermediate phenotype in which 149 many, but not all pairs, came apart prematurely before 12 hr (Fig. 1E), while WT X Dgfu2 cells 150 showed essentially the same kinetics of pairing and cell separation as WT X WT crosses (Fig. 1F). 151 To measure the relative strength of cell-cell adhesion, mating pairs were vortexed at a 152 fixed speed and examined microscopically for pair stability at varying intervals. As expected, WT 153 X WT pairs were highly stable, whereas a majority of pairs formed between Dgfu1 X Dgfu1 ( Fig.   154 1G) or Dgfu2 X Dgfu2 (Fig. 1H) deletion strains came apart relatively quickly after vortexing.

155
When mixed with WT cells, Dgfu1 and Dgfu2 deletion strains again behaved differently from 156 each other. Whereas WT X Dgfu1 pairs were unstable, WT X Dgfu2 pairs were indistinguishable 157 from WT pairs in resisting mechanical shear (Fig. 1G,H). Taken together, these results indicate 158 that GFU1 (like HAP2) is required in both cells of a mating pair for stable pairing to occur, 159 whereas GFU2 confers pair stability when expressed in only one cell of a mating pair. 160

Requirements for GFU1 and GFU2 in membrane fusion. 161
To determine whether pair instability was correlated with an inability of Dgfu1 or Dgfu2 162 deletion strains to form fusion pores, we used a previously described flow cytometry assay to 163 measure exchange of labeled cytosolic proteins between mating cells as a proxy for membrane 164 pore formation during fertilization 4 . In these assays, starved cells of different mating types were 165 labeled either red or green with CellTrace TM Far Red (CTFR) or carboxyfluorescein succinimidyl 166 ester (CFSE), respectively, then combined and fixed either immediately after mixing or 18-24 hr 167 post-mixing after mating pairs had separated as postzygotic progeny ( Fig. 2A,B). 168 Figure 2 shows representative flow cytometry plots of mating cultures containing 1:1 169 mixtures of WT X WT, WT X Dgfu, or Dgfu X Dgfu strains. When fixed immediately after mixing 170 (0 hr), two intensely labeled cell populations were always present, one bright green (CFSE hi ) and 171 the other bright red (CTFR hi ) representing the labeled cells of different mating types before 172 membrane fusion (Fig. 2C). In WT X WT cultures, following fusion and pair separation, these 173 initial populations were almost entirely replaced by two new populations of labeled cells, one 174 more green than red and the other more red than green (Fig. 2D). These new populations 175 (delineated the "Mid" gate in Fig. 2D), represent single cells that had paired, formed fusion 176 pores, and exchanged cytosolic dyes before separating as newly formed progeny. Additionally, 177 a small population of CFSE hi /CTFR hi cells was almost always present (delineated the "Pairs" gate 178 in Fig. 2D) representing cells that continued to pair even at late time points 4 . 179 Unlike WT X WT crosses, matings between Dgfu1 X Dgfu1 and Dgfu2 X Dgfu2 deletion In the case of WT X Dgfu crosses, data from flow cytometry fusion assays were 185 consistent with the pair stability results described above. Only a small percentage (5-15%) of 186 cells were capable of fusion in WT XDgfu1 crosses (Fig. 2G,I), whereas in WT X Dgfu2 crosses 187 the percentage of cells capable of fusion was essentially the same as for WT X WT matings (Fig.  188 2H,I). These results reinforce the idea that pore formation and pair stability are closely linked 189 and demonstrate a bilateral versus a unilateral requirement for the GFU1 and GFU2 gene 190 products, respectively, for efficient membrane fusion between mating pairs. 191

Possible Functional Roles of GFU1 and GFU2. 192
At the level of primary sequence, GFU1 is predicted to be a 557 amino acid protein with 193 no obvious transmembrane domains (Fig. S2E). GFU2, on the other hand, is almost certainly a 194 transmembrane protein with a predicted N-terminal signal peptide (residues 1-17), a 146 amino 195 acid ectodomain, a single transmembrane helix (residues 164-186), and a 42 amino acid C-196 terminal cytosolic region that is enriched in positively charged amino acids (estimated pI = 197 11.46) (Fig. S2F). 198 Using BLAST alignment tools, no homologs outside the ciliate taxon were detected for 199 either GFU1 or GFU2 at the level of primary amino acid sequence. To gain insight into their 200 potential functions we used AlphaFold2 33 , along with other protein structure modelling 201 algorithms to deduce 3-dimensional structures for both proteins. Based on this analysis, GFU1 202 was predicted to be almost entirely alpha-helical with its long antiparallel hairpins bearing a 203 close resemblance to F-BAR (Fes/CIP4 homology-Bin/Amphiphysin/Rys) domains on proteins 204 that bind and regulate curved membranes 34,35 (Fig. 3A,B). In the case of GFU2, other than the 205 transmembrane helix, weak structural homologies with other proteins were confined to a 206 region containing three sets of anti-parallel b-sheets within the ectodomain (Fig. 3C). These 207 similarities were not considered to be functionally significant. 208 To determine the intracellular location of GFU1, we replaced the endogenous gene with 209 chimeric constructs encoding either mCherry-or HA-epitope tagged versions of GFU1 at their C-210 termini. In both instances, transgenic cell lines paired normally with WT cells and with 211 complementary mating types expressing the same GFU1 tagged proteins ( Fig. 3D-F)). 212 Furthermore, consistent with its requirement in membrane fusion, GFU1 chimera localized 213 along the entire length of the conjugation junction punctuated by small gaps (Fig. 3D,E). When 214 GFU1:mCherry was co-expressed with an inducible, C-terminal GFP-tagged version of HAP2, the 215 two proteins showed overlapping and non-overlapping patterns of staining (Fig. 3F). To date, 216 we have been unable to generate tagged versions of GFU2 due to the high AT-content of its 217 corresponding cDNA. GFU2 with Tetrahymena HAP2 resulted in a nearly 2-fold increase in activity while no effect was 228 seen with the Arabidopsis HAP2 ortholog suggesting that species-specific interactions between 229 ciliate proteins may occur (Fig. S3). 230

Cooperativity between the fusion machinery on apposed membranes. 231
While the interplay between HAP2 and the newly identified GFU1 and GFU2 proteins 232 remains unclear, both GFU1 and HAP2 are required in both cells of a mating pair for efficient 233 pore formation to occur (Fig. 2I). This raises the interesting possibility that the fusion machinery 234 on apposed membranes of mating cells somehow interacts, an argument bolstered by the fact 235 that the 75-85% decline in fusion efficiency observed when HAP2 is absent from one cell of a 236 mating pair cannot be compensated for by massive overexpression of the native fusion protein 237 in the wild-type partner 4,20 . 238 To address this question further, we decided to revisit the fusogenic capacity of cells 239 carrying a mutation to a highly conserved arginine residue (R 164 ), which likely stabilizes the 240 HAP2 fusion loop(s) through formation of a salt bridge with an equally conserved glutamic acid 241 residue (E 108 ) present in virtually all HAP2 orthologs 5 . In previous studies using Tetrahymena, 242 substitution of an alanine residue for the conserved arginine at position 164 had no effect on 243 the ability of mutant cells to fuse with WT strains 4 . This was surprising since the same mutation 244 abrogates the fusogenic capacity of HAP2 in species where the protein is present on only one 245 gamete membrane 5,8 . 246 To determine whether R 164 A mutant protein was fully functional or whether the WT 247 protein on the apposed membrane was able to spare an otherwise defective gene product, we 248 examined the ability of R 164 A X R 164 A strains to form fusion pores using flow cytometry. Previously, Xu et al. found that ZFR1, a zinc-finger protein that localizes specifically to 275 the conjugation junction is required for T. thermophila mating types to form mechanically 276 stable pairs 30 . Structure prediction algorithms suggest that ZFR1 may be an E3 ubiquitin ligase 277 (Fig. S4), and as such, could alter the half-lives of proteins involved in stable pairing including 278 those required for membrane fusion. Given the link between pair stability and membrane pore 279 formation described here, we predicted that cells lacking ZFR1 would be unable to fuse. To test 280 this, we deleted the ZFR1 coding sequence in mating types carrying different drug resistance 281 markers in their gametic micronuclei and examined the characteristics of mating, as well as 282 overall fertility, in crosses between Dzfr1 X Dzfr1, WT X Dzfr1, and WT X WT strains.

283
When left undisturbed, complementary mating types lacking ZFR1 recognized one 284 another and formed pairs with similar kinetics and slightly reduced efficiency compared to WT 285 X WT and WT X Dzfr1 cultures (Fig. 5A). Nevertheless, when mechanically agitated, Dzfr1 X 286 Dzfr1 pairs behaved more like Dhap2 knockout pairs, coming apart within seconds of vortexing 287 essentially as described by Xu et al. 30 (Fig. 5B). To determine whether these cells were able to 288 fuse, mating types were labelled with fluorescent dyes and subjected to flow cytometry assays 289 as described above. Consistent with their ability to form mechanically stable pairs, crosses 290 between WT and Dzfr1 strains showed nearly wild type levels of fusion (Fig. 5C,E). Surprisingly 291 however, Dzfr1 X Dzfr1 knockout pairs also showed high levels of fusion despite their unstable 292 pairing phenotype (Fig. 5D,E). 293 We then examined the ability of Dzfr1 X Dzfr1 knockout pairs to generate progeny by 294 crossing parental cell lines carrying either cycloheximide or 6-methylpurine drug resistance 295 markers in their germline micronuclei, isolating individual pairs, and determining the drug 296 resistance phenotypes of resulting synclones after refeeding (Table S1). Consistent with the 297 findings of Xu et al. 30 , we saw that only 22.5% of mating pairs gave rise to true progeny 298 (compared to >80% in standard crosses between WT cells 25 ). In short, while a majority of Dzfr1 299 deletion pairs were able to generate fusion pores, a relatively small percentage gave rise to 300 progeny. As argued below, alterations to pore size and/or the kinetics or pore opening could 301 affect the exchange of meiotic pronuclei across the junctional interface between cells and lead 302 to reduced fertility. 303 304 Discussion. 305 306 As shown here, Tetrahymena mating types carrying deletions in either GFU1 or GFU2 307 were able to recognize and adhere to their homologous deletion strains, but, as in the case of 308 Dhap2 knockouts, were unable to form fusion pores. The corresponding GFU1 and GFU2 gene 309 products can therefore be added to a list of lineage-specific factors, that together with MAR1 in 310 Chlamydomonas reinhardtii 16 , and the DMP8/9 family members of Arabidopsis thaliana 18,37,38 , 311 appear to regulate HAP2-mediated gamete merger during fertilization. 312 From a functional standpoint, MAR1 and the DMP8/9 family members play critical roles 313 in the localization and timely activation of HAP2 in algae and plants, respectively 16,18,37,38 . 314 Whether the same is true of the newly identified GFU1 and GFU2 proteins is unclear. GFU1 315 differs markedly from the algal and plant proteins based on its predicted structure and lacks 316 obvious transmembrane domains. GFU2, on the other hand, is a single-pass transmembrane 317 protein that could potentially interact with receptors on the apposed membranes of mating 318 cells. Indeed, given the reduced kinetics of pairing in crosses between complementary Dgfu2 319 deletion strains (Fig. 1F), GFU2 could act as an adhesion protein, and like MAR1 16 , participate in 320 activation of the HAP2 fusion machinery as a member of a hypothetical receptor-ligand pair or 321 signaling complex. 322 That said, it is unlikely that GFU2 interacts directly with HAP2 as appears be the case for 323 DMP8/9 18 , and possibly MAR1 16 as well. If such interactions occur, one would predict that the 324 absence of GFU2 in one mating type would have a significant impact on HAP2-mediated fusion 325 in T. thermophila, which is clearly not the case (Fig. 2H,I) Chlamydomonas suggests that trans-interactions between membranes at the tips of these 388 structures (rather than between the HAP2 protomers, as in Model 1) are essential for the 389 formation of fusion pores. In this context it is worth noting that actin-dependent invasive 390 protrusions enhance EFF1-dependent syncytia formation in cultured insect cells, which appears 391 to involve bilateral interactions as well 52 . Indeed, the involvement of microvilli and other 392 membrane protrusions in cell-cell fusion 53,54 , as well as virus-host-cell fusion has long been 393 argued 55,56 . 394 Based on these considerations, our second model (Model 2, Fig. 5) proposes that curved 395 membranes at the tips of membrane protrusions destabilize the lipid bilayer locally creating an 396 energetically favorable environment for fusion pore formation independent of HAP2-driven 397 conformational foldback. In this model, HAP2 would play a dual role in helping to shape 398 curvature at the tips of membrane protrusions while at the same time providing the motive 399 force necessary to drive membrane merger (Fig. S5B,C). Efficient pore formation would require 400 the presence of HAP2 in both membranes of a mating pair, not due to trans-interactions 401 between protomers, but to the ability of the fusogen to contribute to curvature on interacting 402 membranes (again, independent of its role in force generation). 403 Such a scenario would explain the low level of fusion seen in the absence of HAP2 on 404 one membrane, as well as the wild-type levels of fusion seen in crosses between WT and R 164 A 405 mutant strains (Fig. 4C). In WT X Dhap2 crosses, the energy barrier to membrane merger would Given its structural resemblance to BAR-domains, the potential involvement of GFU1 in 418 this context should also be considered. In Model 2, GFU1 is shown contributing to curvature at 419 the tips of membrane projections through its interaction with the inner leaflet of the lipid 420 bilayer and, possibly, with HAP2 as well (Fig. S5C). In this scenario, GFU1 would be required in 421 both cells as part of larger protein complexes that would include the fusogen itself (Fig. S5D). Accordingly, Dzfr1 X Dzfr1 pairs may form fewer pores, smaller pores, or pores that 458 close prematurely relative to WT X WT pairs. Indeed, given its localization 30 and structural 459 resemblance to E3 ubiquitin ligases (Fig. S4)

Construction of T. thermophila deletion strains. 477
Different strategies were used to generate macronuclear gene disruptions of GFU1, 478 GFU2, and ZFR1. For GFU1, the "co-deletion" method was used, which relies on the inherent 479 genome editing properties of T. thermophila during macronuclear development 60 . Briefly, a 480 1562-bp target sequence covering roughly two thirds of the coding region between primers #1 481 and #2 (Table S2)  downstream of the predicted GFU2 coding sequence were amplified in separate PCR reactions 500 using primers #5-#8 (Table S2)  GFU2 gene by the CHX cassette was confirmed by qPCR using primers #9 and #10 (Table S2). vector containing the knockout construct was made by separately amplifying 5'-and 3'-flanking 512 regions of the ZFR1 gene via PCR using T. thermophila genomic DNA as the template and primer 513 pairs ZFR1 5'Flankfor/ZFR1 5'flankrevXhoI and ZFR1 3'FLANKFORSacI /3'FLANK REV, respectively 514 (Table S2). The resistance cassette was amplified from an existing Neo4 vector using the primer 515 pair Neo4for2XhoI/Neo4cyrevSacI (Table S2). The three PCR products were then cut with 516 restriction enzymes XhoI, SacI, or both as appropriate, gel purified and sequentially ligated in 517 reactions using T4 DNA ligase (New England Biolabs). The final ligation product was gel purified, 518 blunt-end cloned into pCR4Blunt-TOPO, and sequenced. The resulting plasmid was linearized 519 and introduced into the somatic macronucleus of T. thermophila using biolistic bombardment 61 . 520 Positive transformants were then selected by serial passage of clones in NEFF medium 521 containing increasing concentrations of paromomycin (up to 800µg/mL) 66 . Complete 522 replacement of the endogenous ZFR1 gene was verified by PCR amplification of genomic DNA 523 using primers spanning the Neo4 insertion site (ZFRrescueF1 and ZFRrescuerev, Table S2). 524 Deletion strains used in these studies showed only one PCR fragment of the size expected for 525 the knockout construct and were designated ZFR1KOinCU427cl.3 (MT = VI) and 526 ZFR1KOinCU428cl.51 (MT = VII). 527

Construction of T. thermophila strains expressing mutant HAP2 and tagged GFU1 proteins. 528
Cell lines carrying the T. thermophila HAP2 R164A mutant allele were constructed as 529 previously described 67 . Briefly, site-directed mutagenesis was carried out to alter the sequence 530 of the wild-type HAP2 cDNA using a Q5® Site-Directed Mutagenesis Kit (New England Biolabs). 531 The mutant cDNA was then amplified using primers N7_R64A_F and N7_R164A_R (Table S2)  Construction of mCherry-and HA-tagged GFU1 expressing strains was carried out using 543 a modified PCR-based C-terminal epitope tagging approach described by Kataoka et al. 68 . 544 pBluescript TM plasmid DNA constructs containing codon-optimized sequences for mCherry or an 545 HA-epitope tag immediately upstream of a T. thermophila BTU1-3' termination sequence and 546 neo4 resistance cassette 65 were kindly provided by Dr. Kazufumi Mochizuki (IGH, CNRS-UM, 547 Montpellier, FR). Plasmid DNA was then used to amplify the region between the 5'-ends of the 548 coding sequences of the mCherry or HA-epitope tags and the 3'-end of the neo4 cassette in 549 separate PCR reactions using primers BamHI-mCherry_Fw1 and HindIII-Neo4-RV2 (for mCherry) 550 or BamHI-HA_FW1 and HindIII-Neo4-RV2 (for HA) (Table S2). T. thermophila genomic DNA was 551 then used as a template to amplify a roughly 1,000-bp segment immediately upstream of the 552 3'-stop codon in the macronuclear GFU1 gene in two separate PCR reactions using the same 553 forward primer (5'FW) and either of two linker primers designated 5'RVmCherry and 5'RV-HA, 554 respectively (Table S2). Lastly, a third segment extending 863-bp downstream of the 3'-end of 555 the GFU1 coding sequence was generated using the forward primer, 3'FW, and reverse primer, 556 3'-RV (Table S2) with T. thermophila genomic DNA as the template. The PCR products from each 557 of the reactions were then stitched together by overlap PCR. After agarose gel electrophoresis, 558 bands of the expected sizes of the full-length products were purified and amplified with the 559 5'RACE-Outer and 3'RACE-Outer primers (Table S2)  For visualization of HA-tagged GFU1, mating T. thermophila cells were centrifuged at 642 400 x g, resuspended in 20 mM HEPES buffer (pH 7.5) and fixed with IC Fixation Buffer as 643 described above for flow cytometry. Cells were then resuspended and washed three times in 1 644 mL of 1x Permeabilization Buffer in a 1.5 mL plastic microcentrifuge tube before being 645 resuspended in blocking solution (1x Permeabilization buffer containing 5% Goat Serum and 1% 646 BSA). The same blocking solution was used for all subsequent washes and antibody incubations. 647 After blocking for a minimum of 30 min at room temperature, cells were again washed and 648 then incubated at 4°C overnight in blocking solution containing a 1:100 dilution (100 μg/mL) of 649 primary antibody (Anti-HA, clone 3F10) (Catalog # 50-100-3325; Fisher Scientific). Following 650 incubation, cells were washed 3X in blocking solution before being incubated in the dark at RT 651 with a 1:400 dilution (10 μg/mL) of FITC-conjugated secondary goat Anti-Rat IgG (H+L) (Catalog 652 # 401414; Sigma-Aldrich). Following incubation, cells were washed 3 times and resuspended in 653 a small volume of blocking solution for imaging on a Leica SP5 confocal microscope. 654 For co-localization studies, the HAP2GFP522 strain described above containing an 655 mCherry-tagged version of GFU1 at the endogenous locus and an inducible GFP-tagged version 656 of HAP2 at the b-tubulin locus were induced with 0.1 µg/mL CdCl2, 2 hr prior to mating. 657 Following induction, transgenic cells were mixed at ratios of 1:1 with wild-type strain CU428.2. 658 At varying times thereafter, mating pairs were examined with an Olympus BX-50 fluorescence 659 microscope equipped with a Fluoview scanning laser confocal imaging system and images 660 captured with a DP-72 digital camera. 661 Protein structure modeling. 662 The 3D structure prediction of GFU1, GFU2, and ZFR1 were done with a local installation 663 Mating success in crosses between Dzfr1 deletion strains was determined by assaying 672 progeny cells for resistance to growth in either cycloheximide (CY), 6-methlypuriine (6MP), or 673 both drugs after mating. As indicated above (Construction of T. thermophila deletion strains), 674 ZFR1 was deleted in the parental cell lines CU427 and CU428 with complementary mating type 675 backgrounds (VI and VII) and carrying either a cycloheximide resistance (Cyr) or a 6-676 methylpurine resistance (Mpr) allele in their germline micronuclei, respectively. These deletion 677 strains are wild-type in their transcriptionally active macronuclei and are sensitive to either 678 drug. To determine mating success in crosses between deletion strains, cells were grown to 679 mid-log phase, starved in 10 mM Tris buffer (pH 7.4) and mated at a ratio of 1:1. At 6-9.5 hr 680 post-mixing, individual pairs were hand-isolated into hanging drops to establish "synclones," or 681 mixed populations of the karyonidal descendants from each exconjugant. Following transfer to 682 96-well plates, synclones were subjected to sequential drug testing in 25 µg/mL cycloheximide 683 followed by 25 µg/mL 6-methylpurine and examined for survival 36-48 hr after drug treatment 684 using a dissecting microscope. True "cross-fertilized" progeny are synclones that were resistant 685 to both drugs; "self-fertilizers" or "cytogamonts" were resistant to either CY or 6MP; while 686 "backouts" (parental cells that failed to successfully mate) were sensitive to both CY and 6MP. 687 688