Proc. Natl. Acad. Sci. USA Vol. 77, No. 1, pp. 527-530, January 1980

"Superkiller" mutations suppress chromosomal mutations affecting double-stranded RNA killer replication in Saccharomyces cerevisiae (ski mutants/mak mutants) AKIo TOH-E* AND REED B. WICKNERt Laboratory of Biochemical Pharmacology, National Institute of Arthritis, Metabolism, and Digestive Diseases, National Institutes of Health, Bethesda, Maryland 20205 Communicated by G. Gilbert Ashwell, October 17,1979

ABSTRACT Saccharomyces cerevisiae strains carrying a MATERIALS AND METHODS 1.5 X 106-dalton double-stranded RNA in -like particles (killer plasmid) secrete a toxin that kills strains Strains. Some of the strains of Saccharomyces cerevsiae used not carrying this plasmid. At least 28 chromosomal (mak in this study are listed in Table 1. Description of the phenotype genes) are required to maintain or replicate this plasmid. Re- and genotype of killer strains was presented previously (21). cessive mutations in any of four other chromosomal genes (ski Curing of the killer plasmid is done by growing killer strains for superkiller) result in enhanced toxin production. We report at an elevated temperature (37°C) (23). Mitochondrial DNA that many ski- mak- double mutants are able to maintain the killer plasmid, indicating that the SKIproducts have an effect was eliminated from strains by streaking to single colonies on on plasmid replication. The skil-) mutation suppresses (by- YPAD medium containing ethidium bromide at 30 ug/ml passes) all mak mutations tested except makl6-l. A variant killer (24). plasmid is described that confers the superkiller phenotype and, Media. YPAD, YPG, SD, presporulation medium, sporula- like chromosomal ski mutations, makes several mak genes tion medium, MB medium, and various omission media were dispensable for plasmid replication. described (25). Genetic Techniques. General techniques for genetics Some strains of yeast kill other strains by secreting a protein (26) were followed. Cytoduction (cytoplasmic mixing) was toxin (1-4). The killer genome of Saccharomyces ceievisie is carried out with a karl (karyogamy) strain defective in nuclear a linear 1.5 X 106-dalton double-stranded (ds) RNA (M ds RNA fusion (27). On mating a karl strain with another strain, cyto- or P2) encapsulated in virus-like particles (5-11); this RNA plasmic mixing occurs, but the nuclei fail to fuse and separate codes for the killer toxin (12, 13). Most laboratory strains of yeast when the divides. Usually the recipient was p°, and the also have another species of virus-like particles containing a 3 donor of cytoplasm was p+. After mating donor and recipient X 106-dalton ds RNA (L or P1), which codes for the major coat in YPAD medium, donor nuclei were counter-selected by protein of the virus-like particles (14). plating the mating mixture on an appropriate selective plate. At least 28 nuclear genes are necessary to maintain the killer p+ strains having the genotype of recipient nuclei are cyto- plasmid (M ds RNA). These are maki and mak3-mak27 (refs. ductants. 5 and 15-17; unpublished results), petl8 (18, 19), and spe2 (20). RNA Extraction and Agarose . RNA Al mak mutants, including pet18 and spe2, lose M ds RNA but was extracted from intact cells (28). Electrophoresis was done retain L ds RNA. This result indicates that the mak genes are on agarose gels (1%) and photographs were taken after staining involved specifically in the replication or maintenance of M ds with ethidium bromide (21). RNA, although their exact functions are not known. Recessive mutations in any of four chromosomal genes (called RESULTS skil-ski4 for superkiller) result in increased production of toxin The recessive chromosomal ski mutations result in enhanced activity (21). Two lines of evidence suggest that ski genes may killer activity; thus, the SKI products might have a negative have some role in the replication of M ds RNA. (i) A particular regulatory function at some stage in the replication or expression mutant killer plasmid, called [KIL-sd], is unable to replicate in of the killer genome. The MAK products are each necessary SKI+ cells, but replicates normally in ski- strains (22) and (ii) for maintenance or replication of the killer genome (M ds ski2, ski3, and ski4 mutants seem to have more M ds RNA than RNA). Therefore, the phenotype of mak- ski- double mutants wild type (21). should give information about the functional relationship, if In this communication we show that ski mutations bypass the any, of MAK and SKI products. functions of some mak genes. These data strongly support the skil-1 Suppresses maklO-l. A skil-I strain (AT95) was idea that ski genes participate in the replication of M ds RNA. crossed with a maklO-l strain (M291). Asci containing more The bypass pattern suggests that there may be more than one than two killer clones appeared frequently (Table 2, cross pathway for the replication of M ds RNA. We also describe a superkiller plasmid variant [KIL-b], which, like the chromo- Abbreviations: [KIL-sdj, ski-dependent killer plasmid; [KIL-bi, su- somal ski - mutations, bypasses the requirement for many of perkiller bypass plasmid; ds, double-stranded; K+, killer phenotype; the mak genes. k-, nonkiller phenotype; R+, resistant to killer toxin; R-, sensitive to killer toxin; [KIL-o], absence of killer plasmid, [KIL-k], presence of wild-type killer plasmid. The publication costs of this article were defrayed in part by page * Present address: Department of Fermentation Technology, Faculty charge payment. This article must therefore be hereby marked "ad- of Engineering, Osaka University, Yamada-Kami Suita-Shi, Osaka, vertisement" in accordance with 18 U. S. C. §1734 solely to indicate Japan 565. this fact. t To whom reprint requests should be addressed. 527 Downloaded by guest on October 1, 2021 528 Genetics: Toh-e and Wickner Proc. Natl. Acad. Sci. USA 77 (1980)

Table 1. Strains Table 3. Confirmation of genotype of spore clones from Killer cross W109 Strain phenotype Genotype Killer phenotype AT95 K++R+ a his7 skil-l AT257 K++R+ a his7skil-1 Spore Killer of diploid with: Assigned AT202 K++R+ a iys2 his7 ski2-3 Tetrad clone phenotype skil maklO genotype AT17 K++R+ a Iys2 tyrl his7ski2-1 W109-16 A K+ K++ K- skil maklO AT206 K++R+ a argl ski3-1 B K+ K+ K+ SKI MAK AT21 K++R+ a argl thrl ski4-1 C K+ K++ K- skil maklO 299 K-R- a iysl mak3-1 D K+ K+ K+ SKI MAK 737 K-R- a thrl lys2 mak4-1 W109-17 A 918 K+ K+ K+ SKI MAK K-R- a his6 adel arg4 mak6-1 B K+ K+ K+ SKI MAK 490 K-R- a Ieu2 met5 mak7-1 C K+ K++ K- skil maklO M291 K-R- a ura3 ilv3 canl maklO-l D K+ K++ K- skil maklO M292 K-R- a ura3 ilv3 cani maklO-l W489-4A K-R- a argl thrl Iys2 maklO-2 Superkiller strains (K++) give a clear killing zone at 300C and a AT34 K-R- aleu2ural cdc16makil wider zone at 20'C. Normal killers (K+) give little or no killing zone P28-24C K++R+ a [KIL-bi at 30'C and a smaller zone at 20'C. AT159 K+R+ a his4 karl-i [KIL-ki dissected. The SKI genes are not linked to any of the MAK AT171 K++R+ a his4 karl-i [KIL-b] genes in these crosses. Therefore, the criteria for the suppression 2403-20A K+-++R+ a his7 makl7skil of a mak mutation by a ski mutation was that the segregation 2404-7C K+-++R+ a his7 makl3 skil pattern with respect to the killer trait be 3K+:1K- on the av- 2405-1A K+-++R+ a his7 makl2 skil erage. The pattern of suppression is summarized in Table 4. The 1105 K-R- a adel mak16-1 ski) mutation suppresses all mak mutations tested so far except Several different specificities of killers ofSaccharomyces have been for mak16-1. Mutations in ski2, skiW, and ski4 do not suppress found, called K1, K2, etc. (3,4); they are distinguished by the strains mak)6, mak3, pet18, or maklO, but the rest of the mak muta- they can kill. All killer strains used in this study belong to the K1 tions tested can be suppressed by any of the ski mutations. specificity. K+-++ indicates strains whose killer phenotype was in- Translational suppressors are allele-specific and locus nonspe- termediate between K+ and K++. cific. The independently isolated mak)O mutations, mak)O-) W109), indicating that a suppressor of mak)O-1 was involved and mak)O-2, are both suppressed by the skil)- mutation in this cross. To test whether the suppressor of makIO-1 was (Table 2), and it is unlikely that so many mak mutations in skil-), we determined the genotype of spore clones of two asci different genes would be suppressible by the same translational from cross W109 by complementation with standard ski-) and suppressor (Table 4). Finally, translational (tRNA) suppressors mak)O-1 strains (Table 3). Of the 8 K+ clones, 16A, 16C, 17C, are generally dominant, whereas the suppressor activity of the and 17D were found to be both maklO-1 and skilc-I. When each ski- mutations is recessive (e.g., Table 3). spore clone of these two asci was crossed with a wild-type Suppressors of mak mutations may be isolated directly as [KIL-o] strain of appropriate mating type and meiotic spores killer sectors in mak spore clones from a mak-/ + [KIL-k] were dissected, K- clones frequently segregated from the diploid (29, 30). By using this method, we have isolated one such crosses involving 16A, 16C, 17C, and 17D, but crosses involving suppressor of mak5 that also suppressed mak), mak6, mak7, 16B, 16D, 17A, and 17B gave only K+ clones. This result is and mak8. This suppressor was a recessive chromosomal consistent with the assigned genotype shown in Table 3. All mutation conferring the superkiller phenotype. It did not segregants grew on glycerol-rich medium (p+ phenotype), so complement ski2 mutations and did not segregate relative to none of the killer clones are due to bypass of mak)O-) by the ski2-3 and is thus another ski2 mutation. This provides further p- state (29). The mak)O-2 mutation, which was isolated in- evidence that it is the ski mutation of these strains that is the dependently, was also suppressed by skil)- (Table 2). One of suppressor of the mak mutations. Finally, the ski strains used the ski) mak)O segregants (W462-2D) was crossed with the ski) in Table 4 are independent mutant isolates derived by muta- strain, /AT257, and meiotic tetrads were dissected. As expected, genesis of strain A364A and crossing with strain AN33. Neither almost all segregants were K+. Although the ski) mak)O strains A363A nor AN33 carry suppressors of mak mutations. can maintain the killer plasmid, K- mitotic segregants appear Double-stranded RNA was extracted from the following skil in such strains more often than in wild-type killers. mak double mutants: ski) mak8, ski) pet18, skil mak)O, ski) ski mak Double Mutants Other than skil maklO. Each ski makI2, ski) makB3, and ski) mak) . Analysis of these RNA strain was crossed with various mak strains as shown in Table samples by agarose gel electrophoresis revealed that all these 4, and the resulting diploids were sporulated and tetrads were strains had M ds RNA. Thus, the skil)- mutation is bypassing these mak- mutations by allowing the replication of M ds RNA Table 2. skil mutation bypasses maklO mutation in the double mutant. Segregation of killer Several ski makx- maky triple mutant combinations were phenotype also found to be killers (Table 5). Diploid 3K+: 2K+: Normally the L and M ds are found encapsulated in Cross Parents genotype 4K+:0 1K- 2K- intracellular virus-like particles. Encapsulation of L and M ds RNAs in each of the ski mutants was tested by sedimentation AT95 a skil + W109 - - 2 8 17 of crude extracts of each strain under conditions in which free M291 a + maklO-l ds RNA remains in the supernatant whereas virus-like particles L M RNA AT257 a skil + pellet. Essentially all of both and ds were found in W462 0 6 6 the pellet in the wild type and mutants in each of the four ski W444-3A a + maklO-i genes. W513 AT257 a skil + 1 5 5 Mutant Plasmid [KIL-bJ. Crosses between a mak)l strain W489-4A a + maklO-2 and a particular killer strain from our collection (P-28-24C) Downloaded by guest on October 1, 2021 Genetics: Toh-e and Wickner Proc. Natl. Acad. Sci. USA 77 (1980) 529

Table 4. Suppression of mak mutations by ski mutations Segre- mak mutation gation 16 3 10 petl8 1 4 5 6 7 8 11 12 13 14 15 17 18 19 20 21 22 24 26 27 2K+:2K- 16 5 17 5 1 3 2 4 2 1 4 0 4 1 0 4 3 1 4 2 skil 3K+:1K- 0 4 8 6 9 7 6 4 4 9 10 13 11 4 13 3 6 5 3 10 4K+:0 0 2 2 0 0 7 4 1 6 14 7 10 2 6 9 4 3 6 4 8 2K+:2K- 12 8 22 13 6 3 3 0 ski2 3K+:1K- 2 0 1 1 9 10 6 6 4K+:0 0 0 1 0 5 2 2 4 2K+:2K- 30 17 14 12 1 1 0 4 1 1 ski3 3K+:1K- 0 0 3 0 1 9 6 2 8 4 4K+:0 0 0 0 0 2 1 6 4 2 7 2K+:2K- 48 13 12 10 2 2 4 3 3 2 2 0 2 0 2 2 2 ski4 3K+:1K- 0 1 0 1 8 14 4 22 9 8 5 8 3 2 2 3 6 4K+:0 0 0 0 0 4 2 4 18 5 1 5 3 0 2 4 0 5 2K+:2K- 12 22 21 0 0 0 12 0 24 11 [KIL-b] 3K+:1K- 0 0 0 1 0 2 0 0 0 0 4K+:0 0 0 0 7 17 15 0 12 0 1 The difference between K++ and K+ segregants in these crosses was not clear, so the results are expressed as K+ (includes K++ and K+) or

gave 4 K+ segregants in all 12 asci examined. The killer speci- toductants was crossed with a makil strain (a makll cdc16 ficity of P-28-24C was the same as that of A364A, namely, K1 ural, AT34), and meiotic tetrads were dissected (Table 6). The specificity. The killer of A364A and that of P-28-24C cross containing [KIL-k] from A364A segregated 2K+:2K- in were each cytoduced into a p0 karl [KIL-o] strain. The strains 15 asci tested. On the other hand, the cross containing the killer harboring the killer plasmid that came from P-28-24C showed plasmid from P-28-24C gave 4K+:0 asci (15 asci out of 17 dis- a superkiller phenotype when the killing activity of these cy- sected). In each cross, cdcl6, which is tightly linked to makll, toductants was tested at 300C, whereas strains carrying the segregated 2+:2-. This result indicates that the killer plasmid A364A killer plasmid were not superkillers. Each of these cy- contained in P-28-24C can be maintained in the absence of the makll function. This killer plasmid was designated [KIL-b] (b Table 5. ski- makx maky triple mutants are killers stands for bypass). The dependence of the maintenance of Segregation [KIL-b] on other mak genes was tested by crossing the [KIL-b] Cross Diploid genotype* 4K+:O 3K+:1K- 2K+:2K- strain with several mak- strains. The results (Table 4) show that [KIL-b] can be maintained by mak7, mak4, and makl 7, but not W54st ski4-1 makl7-1 + 12 0 0 by mak3, maklO, makl2, makl6, mak2l, mak26, or petl8 ski4-1 + mak4-1 strains. ski4-1 makl7-1 + W546 ski4-1 +a makl51 11 0 0 DISCUSSION ski4-1l+amakl5-+ The enhanced killing activity shown by the ski mutants could W547 sk41mk71 +9 2 1 be due to an increased copy number of the killer genome, or ski4-1 + makl2-1 to a higher rate of expression (per genome) of the structural ski4-1 makl2-1 11 1 0 gene for killer toxin, or to the production of a more stable toxin. ski4-1 + Because the toxin is coded by the M ds RNA (12, 13), toxin skil-1 makl3-1 + stability should not be altered by the ski mutations, which are 2537 23 1 0 chromosomal (21). In fact, the toxin produced by ski mutants skil-1 + makl2-1 has the same rate of heat inactivation as the wild-type toxin skil-1 makl3-1 + 2538 24 0 0 (unpublished results). skil-1 + makl7-1 As shown in Table 4, the ski mutations bypass many mak skil-1 makl7-1 + mutations. Similarly, the polyamine requirement for killer 2539 24 0 0 plasmid replication is bypassed by each of the ski mutations skil-1 + makl2-1 (20). Thus, the SKI genes are likely to be involved in the inhi- skil-1 makl3-1 2540 16 8 0 bition of some aspects of replication of M ds RNA. skil-1 makl3-1 If mak mutations act by decreasing a plasmid copy-number * The presence of the indicated mak mutations in the parents of control to less than one per cell, while ski mutations simply each cross was confirmed by mating the K+ putative skih makx increase copy number, suppression of mak mutations by ski strain with a SKI+ makx strain and showing that the diploids mutations would be expected. This explanation would predict formed were K-, as described for makl0 in Table 3. In each cross which has little, if any, effect on copy number (21), (except 2540), the two mak mutations are unlinked to each other. that skil-l, For this reason, the most frequent ascus type, 4K+:0, should include should be ineffective in suppressing mak mutations; but skil- ski makx maky triple mutants. These are killers in each case. The suppresses all mak mutations tested except makl6-1, whereas presence of occasional K- segregants (e.g., the 8 K- segregants ski2, ski3, and ski4 mutations, which substantially increase among 96 total segregants in cross 2540) may be due to either un- cellular M ds RNA (21), are, in addition, unable to suppress defined background effects on the suppression or to the relative mak3, maklO, and petl8 mutations (Table 4). Furthermore, instability of the killer plasmid in ski- mak- killers in comparison mutants were with wild-type killers. skil-i makx maky or ski4-1 makx maky triple t Spores of cross W545 were germinated and grown at 30°C because killers if the skil- makx, skil- maky, ski4-1 makx, and ski4-1 mak4-1 is temperature-sensitive for its mak phenotype. maky double mutants were killers (Table 5). Downloaded by guest on October 1, 2021 530 Genetics: Toh-e and Wickner Proc. Natl. Acad. Sci. USA 77 (1980)

Table 6. Independence of maintenance of [KIL-bi on makil Asci Segregation of killer cdcl6 Cross Diploid genotype dissected 4K+:0 3K+:lK- 2K+:2K- 2ts:2+ makl cdcl6 [KIL-ol W169 -a 15 0 0 15 15* a + + [KIL-k1 a makil cdcl6 W10 [KIL-ol 17 15 2 0 17 a + + [KIL-bi * All temperature-sensitive clones were mak -. The bypass pattern shown in Table 5 can be most easily ex- Another scheme to explain the bypass of a mak- mutation plained by assuming there is a second pathway for the repli- by a ski- mutation is shown in Fig. 1 Lower. Here, MAK cation of M ds RNA. According to this hypothesis, the normal products act by inhibiting the SKI products which, in turn, pathway involves the various mak products in the order shown: inhibit M ds RNA replication. first makl6; then mak3, petl8, and makl0; then makl2, The [KIL-bi plasmid confers both the superkiller phenotype mak2l, and mak26; and finally maki, mak4, mak,5 mak6, and suppression of some mak mutations and so may be con- mak7, mak8, makll, and spe2. Substitute enzymes for part of sidered an alteration of a plasmid-coded ski gene. However, this pathway exist (alternate pathway), but the SKI products the [KIL-b] plasmid's relationship to [KIL-k] is not clear, and prevent the killer plasmid from gaining access to this second may involve more than one difference. pathway (Fig. 1 Upper). A skil mutation bypasses all but The presence of a second pathway for the replication of M makl6, whereas ski2, skiS, and skiA mutations, which only allow ds RNA implies the existence of genes whose products are the plasmid to switch pathways later, are unable to suppress necessary for killer plasmid replication in a ski mak killer makS, makl0, or petl8. If mutants defective in different ski strain, but not needed in a SKI+ MAK + killer. This approach genes produced superkilling by completely unrelated effects, is formally analogous to that used in the genetic dissection of one might expect skix skiy double mutants to be stronger su- recombination in (31). mutants. However, perkillers than either skix or skiy single 1. Makower, M. & Bevan, E. A. (1963) Proc. Int. Congr. Genet. 11 double mutants showed the same degree of superkiller phe- 1,202. notype as single mutants, consistent with the model shown. 2. Bussey, H. (1972) Nature (London) New Bilo. 235,73-75. 3. Naumova, T. I. & Naumov, G. I. (1973) Genetika 9,85-90. 4. Young, T. W. & Yagiu, M. (1978) Antonie van Leeu-wenhoek J. M ds RNA Microbiol. Serol. 44,59-77. 5. Somers, J. M. & Bevan, E. A. (1969) Genet. Res. 13, 71-83. MAK16 M. Genet. Res. 14, 71-77. Normal SKI] 6. Bevan, E. A. & Somers, J. (1969) pathway / 7. Bevan, E. A., Herring, A. J. & Mitchell, D. J. (1973) Nature (London) 245, 81-86. MAK3 B. K. Biochem. Soc. Trans. PET 1\ Alternate 8. Buck, K. W., Lhoas, P. & Street, (1973) MAKIC09 pathway 1, 1141-1142. SKI2, SKI3, SKI4 9. Vodkin, M., Katterman, F. & Fink, G. R. (1974) J. Bacteriol. 117, 681-686. 10. Herring, A. J. & Bevan, E. A. (1974) J. Gen. Virol. 22, 387- MAK26 skp 394. A~A&I')I ffA -'Ikp 11. Adler, J., Wood, H. A. & Bozarth, R. F. (1976) J. Virol. 17, 472-476. 12. Bostian, K. A., Hopper, J. E., Rogers, D. T. & Tipper, D. J. (1979) Cell, in press. 13. Palfree, R. G. E. & Bussey, H. (1979) Eur. J. Biochem. 93, 487-493. 14. Hopper, J. E., Bostian, K. A., Rowe, L. B. & Tipper, D. J. (1977) J. BMot. Chem. 252, 9010-9017. Replicated M ds RNA 15. Wickner, R. B. (1974) Genetics 76,423-432. 16. Wickner, R. B. & Leibowitz, M. J. (1976) J. Mol. Biol. 105, 427-443. MAK 17. Wickner, R. B. (1978) Genetics 88,419-425. 18. Fink, G. R. & Styles, C. A. (1972) Proc. Natl. Acad. Sci. USA 69, 2846-2849. jneg 19. Leibowitz, M. J. & Wickner, R. B. (1978) Mol. Gen. Genet. 165, 115-121. 20. Cohn, M. S., Tabor, C. W., Tabor, H. & Wickner, R. B. (1978) J. Biol. Chem. 253,5225-5227. 21. Toh-e, A., Guerry, P. & Wickner, R. B. (1978) J. Bacteriol. 136, SKI 1002-1007. 22. Toh-e, A. & Wickner, R. B. (1979) Genetics 91, 673-682. 23. Wickner, R. B. (1974) J. Bacteriol. 117, 1356-1357. _neg 24. Goldring, E. S., Grossman, L. I., Krupnick, D., Cryer, D. R. & Marmur, J. (1970) J. Mol. Biol. 52, 323-335. 25. Wickner, R. B. & Leibowitz, M. J. (1976) Genetics 82, 429- 442. 26. Mortimer, R. K. & Hawthorne, D. C. (1975) in Methods in Cell , ed. Prescott, D. M. (Academic, New York), Vol. 11, pp. M ds RNA Replication 221-233. 27. Conde, J. & Fink, G. R. (1976) Proc. Natl. Acad. Sci. USA 73, FIG. 1. Two possible models to explain the bypass of mak- 3651-3655. mutations by ski- mutations. (Upper) Model supposes that there 28. Fried, H. M. & Fink, G. R. (1978) Proc. Natl. Acad. Sci. USA 75, exists an alternate pathway for killer plasmid replication, access to 4224-4228. which is normally blocked by the SKI gene products and the wild-type 29. Wickner, R. B. (1977) Genetics 87, 441-452. allele of the plasmid site altered in [KIL-bi. (Lower) Model (simpli- 30. Wickner, R. B. & Leibowitz, M. J. (1977) Genetics 87, 453- fied) assumes that MAK gene products act normally by negative 469. regulation of the production or action of SKI gene products. 31. Clark, A. J. (1973) Annu. Rev. Genet. 7,67-86. Downloaded by guest on October 1, 2021