CORD-19:fd3fc2c49f5cc27e4262261d0c9045911d65cb6e JSONTXT 8 Projects

Restriction Factor in Yeasts 2 3 Abstract In eukaryotes, the degradation of cellular mRNAs is accomplished by Xrn1p and 24 the cytoplasmic exosome. Because viral RNAs often lack canonical caps or poly-A tails, 25 they can also be vulnerable to degradation by these host exonucleases. Yeast lack 26 sophisticated mechanisms of innate and adaptive immunity, but do use RNA 27 degradation as an antiviral defense mechanism. One model is that the RNA of yeast 28 viruses is subject to degradation simply as a side effect of the intrinsic exonuclease 29 activity of proteins involved in RNA metabolism. Contrary to this model, we find a highly 30 refined, species-specific relationship between Xrn1p and the "L-A" totiviruses of different 31 Saccharomyces yeast species. We show that the gene XRN1 has evolved rapidly under 32 positive natural selection in Saccharomyces yeast, resulting in high levels of Xrn1p 33 protein sequence divergence from one yeast species to the next. We also show that 34 these sequence differences translate to differential interactions with the L-A virus, where 35 Xrn1p from S. cerevisiae is most efficient at controlling the L-A virus that chronically 36 infects S. cerevisiae, and Xrn1p from S. kudriavzevii is most efficient at controlling the L-37 A-like virus that we have discovered within S. kudriavzevii. All Xrn1p orthologs are 38 equivalent in their interaction with another virus-like parasite, the Ty1 retrotransposon. To differentiate between models where Xrn1p restricts L-A through a passive 158 mechanism that is incidental to its inherant exonuclease activity, or through an active 159 mechanism where Xrn1p evolves to optimally suppress L-A replication, we first looked 160 for evidence of positive selection (dN/dS > 1) within the genes encoding the major 161 components of the SKI complex, the exosome, and Xrn1p ( Fig 1B) . Importantly, cerevisiae, S. paradoxus, S. mikatae, S. kudriavzevii, S. arboricolus, and S. bayanus) 171 [44] [45] [46] and created a multiple sequence alignment. We then analyzed each alignment 172 for evidence of codons with dN/dS > 1 using four commonly employed tests for positive 173 selection [47, 48] . We see some evidence for positive selection of specific codon sites in 174 several of these genes, however, only XRN1 and the exosome subunit gene RRP40 175 passed all four tests ( Fig 1C; Table S1 ). Other genes are determined to be under surprising for a critical and conserved gene involved in RNA quality control, but 203 consistent with the signatures of positive selection which suggest that certain parts of 204 this protein are highly divergent between species. We next used a functional and quantitative assay to confirm the species-specific 229 effects of XRN1 on virus replication. This assay exploits the dsRNA "killer virus" (also 230 known as M virus). The killer virus is a satellite RNA of L-A and is totally dependent on 231 L-A proteins for replication. It uses L-A-encoded proteins to encapsidate and replicate 232 its genome, and to synthesize and cap its RNA transcripts [12] . The killer virus encodes 233 only a single protein, a secreted toxin referred to as the killer toxin [19, 55, 56] . The result 234 is that "killer yeast" colonies, i.e. those infected with both L-A and the killer virus, kill 235 neighboring cells via the diffusion of toxin into the surrounding medium ( Fig 2C) . has been shown previously that Xrn1p can inhibit the expression of the killer phenotype 239 by degrading uncapped killer virus RNAs [14, 57] . Therefore, we use the presence and 240 size of kill zones produced by killer yeasts as a quantitative measurement of killer virus 241 RNA production in the presence of each Xrn1p ortholog. A strain of S. cerevisiae lacking XRN1, but harboring both the L-A and killer virus 244 (xrn1 L-A + Killer + ), was complemented with each XRN1 ortholog. Clonal isolates from 245 each complemented strain were grown to mid-log phase, and 6 x 10 5 cells were spotted 246 onto an agar plate seeded with a lawn of toxin-sensitive yeast. After several days' 247 incubation at room temperature, kill zones around these culture spots were measured 248 and the total area calculated. The transformation of xrn1 L-A + Killer + with S. cerevisiae 249 XRN1 produced an average kill zone that covered 0.68 cm 2 (n = 14). However, 250 transformation with XRN1 from S. mikatae, S. bayanus, or S. kudriavzevii produced 251 significantly larger kill zones covering 0.92 cm 2 (n = 11), 0.96 cm 2 (n = 17) and 0.97 cm 2 252 (n = 17), respectively. The kill zone produced by xrn1 L-A + Killer + yeast expressing S. 253 cerevisiae XRN1 was significantly smaller than those produced by yeast expressing any 254 of the other XRN1 ortholog (Tukey-Kramer test, p<0.05) (Figs 2D). The smaller kill 255 zones in the strain expressing S. cerevisiae XRN1 are consistent with lower levels of 256 killer and L-A derived RNAs. In summary, this assay also supports a species-specific 257 restriction phenotype for XRN1. It has been observed that over-expression of XRN1 can cure S. cerevisiae of the 260 L-A virus, presumably by degrading viral RNA so effectively that the virus is driven to 261 extinction [8, 28] . Therefore, we developed a third assay to test the ability of XRN1 262 orthologs to control L-A, in this case by assessing their ability to cure S. cerevisiae of the 263 All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint virus. Plasmids expressing HA-tagged and untagged Xrn1p were transformed into a 264 killer strain of S. cerevisiae with its genomic copy of XRN1 intact. This was followed by 265 the analysis of more than 100 purified clones for virus curing, that is, the absence of the 266 killer phenotype as indicated by the loss of a kill zone when plated on a lawn of sensitive 267 yeast. Importantly, the introduction of an empty plasmid fails to produce any cured 268 clones (n = 103) (Figs 3A and 3B). Provision of an additional copy of S. cerevisiae XRN1 269 cured 49% of clones (n = 159) (Figs 3A and 3B). Cured clones remained cured (i.e. non-270 killers) when purified and tested again for their ability to kill sensitive yeasts (n = 20). Over-expression of XRN1 from S. mikatae, S. kudriavzevii, and S. bayanus was unable 272 to efficiently cure the killer phenotype, resulting in only 12% (n = 129), 8% (n = 120), and 273 9% (n = 123) cured clones, respectively ( Fig 3A, blue bars). The loss of L-A from cured 274 strains was also verified by RT-PCR. We detected no L-A or killer RNAs within the four 275 cured clones analyzed ( Fig 3C) . These data show that XRN1 from all Saccharomyces 276 species have the ability to cure the killer phenotype, however, XRN1 from S. mikatae, S. kudriavzevii, and S. bayanus is considerably less efficient than S. cerevisiae XRN1. Taken together, we show that viral restriction by XRN1 is species-specific. These data 279 contradict a model where viral restriction is merely incidental to the RNA quality control 280 functions of XRN1, but is rather something that can be refined through sequence 281 evolution in XRN1. We next tested the presumption that the XRN1 orthologs are functionally 296 equivalent for cellular processes when expressed within S. cerevisiae. We first 297 confirmed that XRN1 orthologs successfully complemented the severe growth defect of 298 S. cerevisiae xrn1, by measuring the doubling time of S. cerevisiae xrn1 with or 299 without a complementing XRN1-containing plasmid ( Fig 4A) and MEME (green). See Table S1 for a list of all sites. The seven residues that To define the importance of the two regions that we identified as containing 375 signatures of positive selection, we replaced portions of S. kudriavzevii XRN1 with the 376 equivalent portions of S. cerevisiae XRN1, and assayed for a region of S. cerevisiae 377 XRN1 that would convey the ability to cure the killer phenotype. We found that an XRN1 378 chimera encoding the last 775 amino acids from S. cerevisiae (Sc-775) was sufficient to 379 cure 56% of clones analyzed, and this was very similar to S. cerevisiae XRN1 (57%) (Fig 380 5D ). Conversely, when the last 777 amino acids from S. kudriavzevii (Sk-777) were 381 used to replace the same region within S. cerevisiae XRN1, only 9% of clones were 382 cured ( Fig S3) . This focused our construction of further chimeras to the second half of 383 the protein, which also contains all of the codons under positive selection and has less 384 amino acid conservation between S. cerevisiae and S. kudriavzevii (82% protein identity, amino acids of S. cerevisiae Xrn1p were unable to convey efficient L-A restriction to S. kudriavzevii Xrn1p (Fig S3) . For this reason, we focused further chimeric analysis on the 390 region encompassing the D1-D3 domains (Fig 5A-C) , as defined previously [59] . We 391 swapped into S. kudriavzevii Xrn1p the D2-D3, D1, or D1-D3 domains of S. cerevisiae 392 Xrn1p, and saw increasing rescue of the ability to cure the L-A virus ( Fig 5D) . All 393 chimeric XRN1 genes were functionally equivalent with respect to their cellular functions, 394 as all were able to establish normal growth and benomyl resistance in S. cerevisiae 395 xrn1 ( Fig S4) . Therefore, the species-specificity domain maps predominantly to D1, 396 with contribution from the neighboring D2 and D3 domains as well. Together, our data 397 suggest that the exonuclease activity of Xrn1p is important for virus restriction and is 398 preserved across species, but that evolution has tailored a novel virus interaction The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint encoded L-A Gag was immunoprecipitated in the presence of Xrn1p-HA from 443 either S. cerevisiae or S. kudriavzevii using beads with (+Ig) or without (-Ig) anti-444 Gag antibody present. Adh1p was used in all panels as a loading control to 445 ensure equal input of total protein and the specificity of immunoprecipitation. Confirmation of the species-specific restriction of Xrn1p using a newly 448 described totivirus from S. kudriavzevii. We next wished to test our findings against other related yeast viruses. Indeed, 451 the S. cerevisiae totivirus L-A-lus has been shown to have limited susceptibility to XRN1 452 from a different strain of S. cerevisiae [28] . We also wanted to test viruses of other residue required for cap-snatching (H154), a -1 frameshift region, packaging signal, and in S. cerevisiae (Fig 3) . We did not observe any virus curing by any orthologs of Xrn1p, 506 but believe that this could be because the high-copy plasmids that we used in this 507 experiment in S. cerevisiae are unable to drive Xrn1p expression in S. kudriavzevii high 508 enough to actually cure the virus. However, we have observed previously that Xrn1p can 509 reduce the abundance of totivirus RNAs (Fig 2A) , so we further analyzed the XRN1-510 transformed clones of S. kudriavzevii for changes in SkV-L-A1 RNA levels using reverse 511 transcriptase quantitative PCR (RT-qPCR). Total RNA was extracted from clones of S. kudriavzevii and converted to cDNA using random hexamer priming. cDNA samples 513 were amplified using primers designed to specifically target SkV-L-A1 GAG and the 514 cellular gene TAF10. The empty vector control was used as the calibrator sample, and 515 TAF10 expression was used as the normalizer to calculate the relative amount of SkV-L-516 A1 RNAs present within each XRN1 expressing S. kudriavzevii cell line using the 517 comparative CT method [65] . We found that expression of XRN1 from S. kudriavzevii (n 518 = 10) reduced the relative levels of SkV-L-A1 RNAs by 40% (Fig 7E) , even though this Xrn1p was expressed at the lowest levels ( Fig 7D) . This is in contrast to XRN1 from S. 520 mikatae (n = 9) and S. bayanus (n = 8) that only showed a 13% increase or 15% 521 decrease in SkV-L-A1 RNAs, respectively. S. cerevisiae XRN1 was able to reduce SkV-522 L-A1 RNAs by 27% and is noteworthy due to the close evolutionary relationship between 523 SkV-L-A1 and other L-A-like viruses from S. cerevisiae ( Fig S6) . These data suggest 524 that Xrn1p is a species-specific restriction factor in different Saccharomyces yeasts, and 525 that coevolution of totiviruses and yeasts has specifically tailored the potency of Xrn1p to 526 control the replication of resident viruses within the same species. 527 528 All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint important for controlling the abundance of totivirus RNAs. We find that XRN1 and the 538 virus the best. The exact nature of the host-virus protein-protein interaction that is driving 539 this evolutionary arms race is not clear. To thwart XRN1, the totiviruses are known to 540 synthesize uncapped RNAs with an exposed 5' diphosphate, which is a suboptimal 541 substrate for Xrn1p-mediated decay [24] . Further, it has been shown that the totivirus 542 Gag protein has a cap-snatching activity that cleaves off caps from host mRNAs and 543 uses them to cap viral transcripts, protecting them from Xrn1p degradation [12, 14] . We 544 have found that Xrn1p interacts with L-A Gag, and that this interaction is not mediated by 545 the presence of single-stranded RNAs. What remains unknown is whether Xrn1p is 546 targeting Gag as part of the restriction mechanism, or whether Gag is targeting Xrn1p as 547 a counter defense. As we did not observe an obvious species-specific differences in the 548 interaction between Xrn1p and L-A Gag by coimmunoprecipitation, we cannot clearly 549 define the observed role of sequence variation in Xrn1p. This may be because of the 550 low sensitivity of our assay system, or because direct binding of Xrn1p by L-A Gag is 551 ubiquitous and that the rapid evolution of XRN1 results from another intriguing facet of 552 virus-host interaction and antagonism. However, we now know that the interaction 553 All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint between L-A and Xrn1p goes beyond the simple recognition of L-A RNA by Xrn1p. We 554 can speculate that Xrn1p may compete with Gag for access to uncapped viral RNAs as 555 they are extruded into the cytoplasm, or that interaction with unassembled Gag allows 556 the recruitment of Xrn1p to sites of virion assembly resulting in viral RNA degradation. Alternately, it is possible that the target of Xrn1p is simply L-A RNA, and that the The literature suggests that Xrn1p is a widely-utilized restriction factor against 564 viruses, as it has been reported to have activity against mammalian viruses [9,16], yeast 565 viruses [8, 24] , and plant viruses [53] . The potent 5'-3' exonuclease activity of Xrn1p has 566 resulted in viruses developing a rich diversity of strategies to protect their RNAs. For 567 instance, Hepatitis C virus recruits MiR-122 and Ago2 to its 5' UTR to protect its RNA 568 genome from Xrn1p degradation [7, 16] . The yeast single-stranded RNA narnavirus uses 569 a different strategy to protect its 5' terminus, folding its RNA to form a stem-loop 570 structure that prevents Xrn1p degradation [8] . In some cases, viruses even depend on 571 Xrn1p to digest viral RNA in a way that benefits viral replication, for example, preventing proteins that initiate endonucleolytic cleavage of host mRNAs, revealing exposed 5' 578 monophosphates that are substrates for Xrn1p degradation. This is thought to interfere 579 All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint "decoys", and viral RNA pseudoknots are all utilized to prevent Xrn1p-mediated viral 584 RNA destruction [7, 8, [12] [13] [14] [15] [16] [17] [18] (Fig 2) , virus loss (Fig 3) , or a reduction in viral RNA (Fig 2 and Fig 7) . Signatures of 624 positive selection that we have detected in Saccharomyces XRN1 are also consistent 625 with a host-virus equilibrium that is in constant flux due to the dynamics of a back-and-626 forth evolutionary conflict (Fig 2 and Fig 6) . There are several examples of mammalian housekeeping proteins engaged in 629 evolutionary arms races with viruses. (By "housekeeping" we refer to proteins making 630 critical contributions to host cellular processes, as opposed to proteins dedicated to 631 All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint assembled plasmids selected for by growth on complete medium (CM) -leucine. The 657 XRN1 open reading frame (HA-tagged and untagged) from S. mikatae, S. bayanus, or S. kudriavzevii was introduced into pPAR219 between the 5' and 3' UTRs from S. 659 cerevisiae XRN1 using recombineering to produce pPAR225, pPAR226, and pPAR227, 660 respectively. As a negative control, NUP133 was cloned into the pPAR219 plasmid 661 backbone to produce pPAR221, which was used to allow growth of xrn1 on medium 662 lacking leucine without XRN1 complementation. The LEU2 gene was replaced by TRP1 663 using recombineering techniques to produce the plasmids pPAR326, pPAR327, 664 pPAR328, and pPAR329. Using PCR and recombineering, we also constructed The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint sec. The dsRNA was eluted from the column by the addition of 100 l of 0.15 mM EDTA 706 (pH 7.0) and incubation at 65 o C for 5 min before centrifugation at 16,000 x g for 30 sec. dsRNA that was extracted from 1 x 10 9 yeast cells using our rapid extraction of viral 710 dsRNA protocol was used as template for superscript two-step RT-PCR (Thermo 711 Fisher). cDNA was created using a primer specific for the negative strand L-A genomic The same primers were used to detect L-A and killer virus specific cDNAs using PCR 724 with Taq polymerase (New England Biolabs). We assayed Ty1 retrotransposition in S. cerevisiae xrn1, using the previously 728 described Ty1 retrotransposition reporter system [89] , and confirmed that XRN1 deletion 729 causes a dramatic reduction in Ty1 retrotransposition (~50-fold) [35] . To test the effect of 730 All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint volume of log-phase yeast cells as per manufacturers instructions or by bead beating as of the analysis, two models of codon frequencies (F61 and F3x4) and multiple seed 758 values for dN/dS () were used (Table S1) All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint YPD plates containing 15 g ml -1 of benomyl were prepared as described previously The curing of the killer phenotype was measured by transforming S. cerevisiae BJH006 804 with approximately 100 ng of plasmid encoding various XRN1 genes using the LiAc 805 method. The addition of 1000 ng or as little as 10 ng of plasmid had no affect on the 806 percentage of colonies cured using this assay. After 48 h of growth, colonies were 807 streaked out and grown for a further 48 h. Clonal isolates of killer yeasts were patched 808 onto a YPD "killer assay" plate (see kill zone measurement protocol) that were 809 previously inoculated with S. cerevisiae K12, and incubated at room temperature for 72 810 h. The presence or absence of a zone of inhibition was used to calculate the percentage 811 of killer yeast clones cured of the killer phenotype. Xrn1p structural modeling 814 PHYRE was used to create a template-based homology model of S. cerevisiae Xrn1p 815 using the solved structure of K. lactis Xrn1p as a template [58, 59] . The structure was 816 determined with an overall confidence of 100% (36% of aligned residues have a perfect 817 alignment confidence as determined by the PHYRE inspector tool), a total coverage of 818 81%, and an amino acid identity of 67% compared to K. lactis Xrn1p. PDB coordinates 819 for the modeled structure can be found in File S1. Structural diagrams were constructed 820 using MacPyMOL v7.2.3. Co-immunoprecipitation of Xrn1p and L-A Gag. Strains were grown in CM lacking the appropriate amino acids in order to retain the 824 relevant plasmids. For co-immunoprecipitations involving L-A Gag-V5 and Xrn1p-HA, 50 825 mL cultures (CM -tryptophan -leucine, 2% raffinose) were used to inoculate 500 mL 826 cultures (CM -tryptophan -leucine, 2% galactose) at OD600 ~0.1. Cells were harvested 827 All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. Schmitt MJ, Breinig F. Yeast viral killer toxins: lethality and self-protection. Nat All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. n/a n/a All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint 3' Gly Phe Arg Gly Leu Gly -1 frame 0 frame 3' L-A-lus ΔG -5.9 kCal/mol 3' SkV-L-A1 ΔG -2.1 kCal/mol A Figure All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint Doubling time (min) All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. . https://doi.org/10.1101/069799 doi: bioRxiv preprint