Essential for Viability in Saccharomyces Cerevisiae (Gene Cloning/Essential Gene/Yeast/Gene Inactivation/RAD3 Gene) LOUIE NAUMOVSKI and ERROL C

Proc. NatL Acad. Sci. USA Vol. 80, pp. 4818-4821, August 1983 Genetics

A DNA repair gene required for the incision of damaged DNA is essential for viability in Saccharomyces cerevisiae (gene cloning/essential gene/yeast/gene inactivation/RAD3 gene) LOUIE NAUMOVSKI AND ERROL C. FRIEDBERG Laboratory of Experimental Oncology, Department of Pathology, Stanford Universitv, Stanford, California 94305 Communicated by I. Robert Lehman, May 2, 1983 ABSTRACT A diploid strain (RAD3/RAD3) of Saccharomy- tational rearrangement in the chromosomal URA3 gene (ura3- ces cereviiae was transformed with an integrating plasmid con- 52) that precludes integration by homologous recombination at taining an internal fragment of the cloned yeast RAD3 gene. In- this site (6, 7). Surprisingly, no viable haploid integrants could tegration by homologous recombination inactivated one of the be isolated, even though we readily isolated integrants using diploid RAD3 genes, creating a recessive mutation. This mutation other plasmids. In this paper, we report that the RAD3 gene is inferred to be lethal in haploid cells since sporulation of diploid is essential and that mutations affecting its essential function(s) transformants segregated two viable and two inviable spores per result in lethality in haploid cells. This result is unexpected since tetrad, while integration of plasmids containing one or the other other DNA end of the RAD3 gene resulted in diploid transformants that seg- genes required for the specific incision of damaged regated normally-i.e., four viable spores in each tetrad. Evi- in yeast or in other organisms are not known to be essential. dence that integration of the internal fragment occurred specifically at one of the RAD3 genes in the diploid is provided by MATERIALS AND METHODS DNA-DNA hybridizations. In addition, transformation of a diploid strain heterozygous for the RAD3 gene (RAD3/rad3-2) (car- Yeast and Bacterial Strains. The yeast strains used were rying a rad3 mutation that does not affect the viability of haploid YM197 (MATa/MATa ura3-52/ura3-52 ade2-101/ade2-101 cells) results in the rad- phenotype in half of the transformants, lys2-801/lys2-801 SUC2/SUC2 GAL/GAL RAD/RAD) ob- indicating that the RAD3 gene was inactivated in these cells. tained from Mark Johnston (Department of Biochemistry, Stanford University) and LN3-2-I10 (MATa/MATa ura3-52/ The excision repair of base damage requires the action of an ura3-52 trpl-289/TRP his3-832/HIS ade2/ade2 rad3-2/RAD). endonuclease that specifically recognizes the presence of dam- The latter strain was constructed by mating strains SX46a (MATa age in DNA and catalyzes the incision (nicking) of the duplex ura3-52 trpl-289 his3-832 ade2) (obtained from Jasper Rine, adjacent to such sites (1). Such an activity has not been isolated University of California, Berkeley) with strain LN3-2-43 (MATa or characterized from any eukaryotic source; however, it has ura3-52 ade2 rad3-2) (this laboratory) and isolating the appro- been established that in the yeast Saccharomyces cerevisiae at priate diploids. (The rad3-2 mutation does not affect the via- least five genes (designated RAD1, RAD2, RAD3, RAD4, and bility of cells in the absence of DNA damage.) Tetrad analysis RADIO) are required for the incision of UV-irradiated DNA in of strain LN3-2-I10 verified that it contained the markers in- vivo (2, 3). Nonconditional mutants defective in these genes are dicated. Eschertichia coli strain HB101 was used for propaga- highly sensitive to UV radiation and UV-mimetic chemicals and tion of the plasmids. are totally defective in DNA incision and in pyrimidine dimer Culture Media. Yeast minimal medium was supplemented excision (2-4). However, these mutants are not pleiotropic in with the necessary nutrients (except uracil) as appropriate for their phenotype and are viable as haploid cells in the absence selective growth of strains. YPD medium, prepared as de- of DNA damage by the agents mentioned above (4). scribed (5), was used for nonselective growth. E. coli HB101 Studies in our laboratory have resulted in the isolation of a was grown on L medium with ampicillin (50 ,g/ml) as appro- series of recombinant plasmids from a yeast genomic library, priate. each of which contains a distinct RAD gene required for the Preparation of DNA. Plasmid DNA was purified by cen- incision of DNA during excision repair in yeast (ref. 5; unpub- trifugation through cesium chloride/ethidium bromide gra- lished data). We have previously reported the molecular clon- dients as described by Davis et al. (8), except that the ethidium ing of the RAD3 gene of S. cerevisiae and established by dele- bromide concentration was 200 ,ug/ml. Rapidly prepared ly- tion mapping that a 1.5-kilobase (kb) region flanked by BamHI sates of E. coli were made as described (5). and EcoRI sites is situated internal to the ends of the gene (5). Plasmid Construction. An integrating plasmid (pNF3200) During the course of other experiments, we constructed an in- containing a 1.5-kb internal fragment of the RAD3 gene was tegrating plasmid containing the 1.5-kb BamHI/EcoRI frag- constructed as follows. Plasmid pNF3207, containing a 4.5-kb ment. Since this plasmid does not contain either end of the RAD3 insert that includes the RAD3 gene (1) (see Fig. 2) was digested gene, its integration into the RAD3 site in haploid wild-type to completion with BamHI and EcoRI. The reaction mixture cells by homologous recombination should have generated rad- was electrophoresed through a 0.7% agarose gel and the 1.5-kb transformants containing two incomplete copies of the RAD3 fragment was isolated by using Nal and glass beads (9). The pu- gene (see Fig. 1). No RADW transformants were expected even rified fragment (100 ng) was mixed with the integrating plasmid though the plasmid carries the yeast URA3 gene as a selectable YIp5 (250 ng), which had been cleaved with EcoRI and BamHI marker, because the cells used for transformation have a mu- and also treated with calf alkaline phosphatase to prevent self- ligation. The mixture (20 ,ul) contained 20 mM Tris HCl (pH 10 mM 1 mM and 10 mM dithiothreitol. T4 The publication costs of this article were defrayed in part by page charge 7.5), MgSO4, ATP, payment. This article must therefore be hereby marked "advertise- ment" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviation: kb, kilobase(s). 4818 Downloaded by guest on October 2, 2021 Genetics: Naumovski and Friedberg Proc. Natl. Acad. Sci. USA 80 (1983) 4819 DNA ligase (Bethesda Research Laboratories) (0.1 unit) was were computed from the OD6eo readings after 12 hr of growth added and the mixture was incubated at 5°C for 16 hr. in exponential phase. Plasmids pNF3204 and pNF3205 (see Fig. 2) were constructed similarly by isolating the relevant DNA fragments from RESULTS gels after digestion of pNF3207 with EcoRI. To construct pNF3208 and pNF3209 (see Fig. 2), plasmids pNF3207 and To show that the RAD3 gene is essential, we used an experi- pNF3206 (containing the 4.5-kb fragment in opposite orien- mental approach reported by Shortle et aL (12) in which lethal tation) were digested to completion with BamHI. The larger mutations can be detected by making them recessive. In our DNA fragment obtained in each case was isolated and self-li- experiments, this was achieved by transforming a diploid (RAD3/ gated. In all cases, DNA was isolated from E. coli HB101 trans- RAD3) strain with an integrating plasmid containing the BamHI/ formants and plasmid constructions were verified by restriction EcoRI internal fragment of the RAD3 gene (5). During inte- enzyme analysis of the DNA. gration by homologous recombination, one of the chromosomal Transformation of Yeast Cells. Yeast cultures were grown RAD3 genes is inactivated (Fig. 1). [Integration does not occur to an OD6co of =0.5 and transformed with plasmid DNA as de- at the ura3-52 site (6, 7).] If this mutation is lethal, subsequent scribed (5). All plasmids carry the yeast URA3 gene, which was sporulation of the diploid will segregate two viable spores (con- used as a selectable marker for transformation. taining the wild-type RAD3 gene) from each tetrad, while the Spore Analysis. Transformed diploids were purified on min- other two spores should be nonviable. The results of such an imal medium agar plates. They were then grown on YPD for analysis carried out on four independently isolated diploid 18-24 hr and transferred to sporulation plates. After 4-5 days, transformants are shown in Fig. 2. In every case, viability seg- tetrads were dissected using a micromanipulator and the phe- regated primarily as 2+ :20. Al the viable spores were ura- (Fig. notypes of the spores were analyzed by standard procedures 2), indicating that the URA3 gene on the integrating plasmid (10). was segregated to the nonviable spores and is thus tightly linked DNADNA Hybridization. DNA was isolated from yeast cul- to the lethal alteration in the genome. A control experiment tures by using glass beads and phenol as described (5), except with an integrating plasmid containing an internal fragment from that cultures were grown to an OD600 of 1-5 and then treated a nonessential yeast gene (GAL)) showed that, in the majority with glusulase for 30-60 min, suspended in lysis buffer and of the tetrads examined, all four members of the tetrad were phenol, and Vortex mixed with glass beads for 10 sec. Hybrid- viable (Fig. 2). In every complete tetrad, two of the viable spores izations were carried out by the Southern blotting technique were GAL' ura-, while the other two were gal- URA' (data (11) as described by Davis et al. (8). not shown). Effect of Gene Dosage on Generation Times. Diploid trans- To confirm that the 1.5-kb DNA insert present in the in- formants containing one, two, or more than two copies of the tegrating plasmid pNF3200 (Fig. 2) is indeed an internal frag- RAD3 gene were constructed as follows. Strain YM197 (RAD3 ment-i.e., contains neither end of the RAD3 gene-we con- ura3-52/RAD3 ura3-52) was transformed with either pNF3200 structed other integrating plasmids with various deletions of (see Fig. 2) (to inactivate one copy of the RAD3 gene), YIp5 the 4.5-kb DNA fragment previously shown to contain the RAD3 containing an internal fragment of the yeast GAL) gene (to leave structural gene (5). Integration of plasmids containing the 1.5- both RAD3 genes intact), or one or two copies of pNF3207 (see kb internal region that also include one or the other end of the Fig. 2) (to generate at least three or four RAD3 genes, re- RAD3 gene should result in the retention of one intact copy of spectively). The ploidy of the RAD3 gene in each transformed the RAD3 gene when integrated by homologous recombina- strain was inferred from tetrad analysis of the URA3 and/or tion. Sporulation of diploids transformed with such plasmids or RAD3 markers. Each of these transformants was grown in YPD a plasmid containing the entire RAD3 gene resulted in tetrads to an OD6w of =0.2. They were then diluted to an OD6co of with four viable spores (Fig. 2). In all cases the URA3 marker 0.002 and grown in YPD for a further 12 hr. Generation times present on the integrating plasmid segregated 2+ :2- (Fig. 2),

INTEGRATION BY HOMOLOGOUS I RECOMBINATION 8 511 1 1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Oo 00 *______TWO VIABLE TWO SPORES INVIABLE SPORES FIG. 1. General scheme for inactivation ofone ofthe RAD3 genes in a homozygous (RAD3/RAD3) diploid strain. Integration of a plasmid containing any internal fragment of the RAD3 gene generates two incomplete copies of the gene. Downloaded by guest on October 2, 2021 4820 Genetics: Naumovski and Friedberg Proc. Nad Acad. Sci. USA 80 (1983) PLASMID PHENOTYPE IN FRAGMENT INTEGRATED TETRAD ANALYSIS REPLICATING PLASMID SEGREGATION OF VIABILITY 3: j3 I 1.7kb 1.5kb 11.3kb 40 3~1° 24:2° 1-' 04° gWRI ~HI1 EoRl LORI pNF3200 rod- - 0 0 45 5 0 0:2 pNF3204 rad 7 2 0 0 0 2:2 pNF3205 rad- - 7 1 0 0 0 2;2 pNF3208 rad- 6 1 1 0 0 2:2 pWF3209 rad 7 2 0 0 0 2:2 pNF3207 RAD 9 0 0 0 0 2:2 4.5kb Untransformed 6 3 1 0 0 pNF20 GAL I internal frogment 5 1 1 0 0

FIG. 2. Genetic analysis of a diploid (RAD3/RAD3) strain transformed with integrating plasmids containing various DNA fragments. The diploid strain YM197 was transformed with plasmid YIp5 containingthe indicated fiagments. Diploids were isolated and sporulated, and tetrads were examined for viability and for segregation ofthe URA3 marker carried on the integrating plasmid. The tetrads in which viability segregated 1+ :30 (pNF3200) or 3+:1° are presumed to be incomplete. A total of 50 tetrads derived from four diploids transformed with pNF3200 and 12-20 tetrads from transformants with the control plasmids were examined for segregation ofthe URA3 gene. The data presented here are from complete tetrads only. Plasmid pNF20 contains an internal fragment ofthe yeast GALl gene (obtained from Mark Johnston, Department of Biochemistry, Stanford University).

indicating that integration of these plasmids was not associated the expected hybridization bands. Similarly, DNA from a cul- with lethality. ture of ura- viable haploid spores that resulted from the trans- DNA hybridization using the Southern blotting technique formed diploid showed only a single band representative of the provided physical evidence that homologous integration ac- wild-type RAD3 sequence. companied the transformations described above. DNA isolated To obtain further evidence that plasmid pNF3200 integrated from the untransformed diploid strain and from URA+ trans- into the RAD3 gene, we transformed a heterozygous diploid formants was cleaved with Sst I [which cuts DNA outside of the strain (RAD3/rad3-2) with this plasmid and examined the RAD RAD3 gene (5)] and then probed with 'P-labeled pBR322 DNA phenotype of the resulting transformants. The rad3-2 mutation containing the 1.5-kb BamHI/EcoRI fragment. The results (Fig. confers UV sensitivity to haploid cells but does not affect their 3) show a single hybridization band of 13 kb in the digest of the viability (5). Hence, if integration occurs at the RAD3 gene and untransformed diploid DNA. Cleaved DNA from the diploids destroys its ability to function in DNA repair, diploids should transformed with the integrating plasmid pNF3200 (6.7 kb) become phenotypically rad--i.e., sensitive to UV radiation. showed two hybridization bands. One corresponds in size to the Of the diploid transformants examined, two were RADW (pre- wild-type chromosomal fragment containing the RAD3 gene sumably due to integration of the plasmid into the rad3 gene) (13 kb) and the other has the expected size of a chromosomal (Table 1) and two were rad- (presumably due to integration of fragment containing the plasmid integrated at RAD3 (20 kb). the plasmid into the RADW gene) (Table 1). As expected from Analysis of integrants containing control plasmids also showed the results presented in Fig. 2, when these diploids were sporulated, tetrads segregated viable spores as 2+:2° (data not shown). Thus, deletion mutations in the RAD3 gene abolished both DNA 2 3 4 5 6 7 8 X marker repair capacity and haploid cell viability. Transformation of the heterozygous diploid strain with integrating plasmids containing fragments that included one or the other end of the RAD3 gene (Fig. 2) yielded only RADW transformants (Table 1). -23.7kb_ Tetrads derived by sporulation of the homozygous (RAD3/ RAD3) diploid strain transformed with the plasmid containing 9.5kb

6.7kb Table 1. RAD phenotype of heterozygous (RAD3/rad3-2) diploid transformants Transformants, no. Plasmid Total RADW rad- pNF3200 4 2 2 pNF3204 5 5 0 pNF3205 7 7 0 FIG. 3. DNA hybridization analysis of diploid integrants. DNA was pNF3208 5 5 0 cleaved with Sst I and electrophoresed through a 0.5% agarose gel. pNF3209 7 7 0 Transfer of DNA was to nitrocellulose, and 3 P-labeled pBR322 con- pNF3207 6 6 0 tainingtheBamfl/EcoRI 1.5-kb internal fiagmentfom the RAD3 gene was used as the probe. Lanes: 1, YM197 diploid strain (nonintegrant); A diploid strain heterozygous for the RAD3 gene (rad3-2/RAD3) was 2-7, YM197 containing, respectively, pNF3200, pNF3204, pNF3205, transformed with the plasmids indicated and transformants were scored pNF3208, pNF3209, and pNF3207; 8, ura- viable haploid from spor- for the RAD phenotype by replica plating and streak testing at various ulation of YM197 transformed with pNF3200. UV doses. Downloaded by guest on October 2, 2021 Genetics: Naumovski and Friedberg Proc. Natl. Acad. Sci. USA 80 (1983) 4821

the 1.5-kb RAD3 internal fragment were dissected and the fate expected since single-strand binding proteins and DNA ligase of the spores on agar plates was examined under a dissecting are presumably required for DNA replication and for genetic microscope. Budding of all spores from a tetrad was recogniz- recombination. However, proteins involved in the specific in- able within 5 hr after plating. By 7 hr, it was possible to dis- cision of DNA during excision repair have not previously been tinguish the viable from the inviable pair. The former each formed associated with essential functions in any organism. In addition, microcolonies consisting of four to six cells, while the latter di- as indicated above, our studies suggest that RAD3 has two dis- vided more slowly and consisted of two or three cells each. By tinct functions in yeast since viable haploid rad3 cells are ap- 24 hr after plating, the inviable spores had generated colonies parently totally defective in DNA incision (2, 3). Thus it is likely consisting of only 2-10 cells each, and then all further division that the role of RAD3 in DNA incision is different from its es- ceased. Some microcolonies were teased apart to examine the sential role. The essential function of RAD3 could provide a morphology of individual budding cells. Inviable haploid cells regulatory mechanism for slowing or even arresting DNA rep- appeared larger and had a thicker cell wall than did normal hap- lication during excision repair, if incision of damaged DNA loid cells, but no other consistent morphologic differences were renders the RAD3 gene product limiting for its essential role. noted. Such regulation could provide an explanation for the phenom- The effect of gene dosage on the growth of cells was ex- enon of cell division delay caused by DNA damage. amined by comparing the generation time of diploid cells con- The RADJ and RAD2 genes of S. cerevisiae have also been taining one, two, or more than two copies of the RAD3 gene. cloned (unpublished data). Haploid RADW cells transformed No significant differences were observed (data not shown). with integrating plasmids containing fragments from the interior of these genes are viable and phenotypically rad-. We DISCUSSION therefore conclude that the RADI and RAD2 genes are not essential and that not all genes required for incision of damaged We have previously studied two allelic mutants at the RAD3 DNA are essential genes. locus (rad3-1 and rad3-2), both of which grow normally unless exposed to DNA-damaging agents (5). It was thus initially sur- We thank Mark Johnston and Phil Hieter for their interest and advice prising that we were unable to isolate haploids transformed with in various phases of this study and David Clayton, David Korn, Roger a plasmid containing a cloned fragment from the interior of the Schultz, Eric Radany, and Elizabeth Yang for their critical review of the RAD3 gene. The experiments presented here show that dele- manuscript. These studies were supported by Public Health Service tion mutations in the chromosomal RAD3 gene created by ho- Grant CA12428. L.N. is a trainee in the Stanford Medical Scientist mologous integration of such a plasmid are lethal. Thus, when Training Program supported by Public Health Service Training Grant integration occurred in a diploid strain homozygous for the RAD3 GM07365. gene, only two of the four spores were viable in each tetrad generated during sporulation. On the other hand, normal vi- 1. Hanawalt, P. C., Cooper, P. K., Ganesan, A. K. & Smith, C. A. ability of spores was observed when the integrating plasmid (1979) Annu. Rev. Biochem. 48, 783-836. 2. Reynolds, R. J. & Friedberg, E. C. (1981)J. Bacteriol. 146, 692- contained deletions that left one or the other end of the RAD3 704. gene intact, since integration of these plasmids by homologous 3. Wilcox, D. R. & Prakash, L. (1981)J. Bacteriol. 148, 618-623. recombination generates a wild-type copy of the gene in the 4. Haynes, R. H. & Kunz, B. A. (1981) in The Molecular Biology of affected chromosome. the Yeast Saccharomyces: Life Cycle and Inheritance, eds. Strat- Evidence that these integrations occurred at the chromo- hem, J. N., Jones, E. W. & Broach, J. R. (Cold Spring Harbor somal RAD3 gene comes both from DNADNA Laboratory, Cold Spring Harbor, NY), pp. 371-414. specifically hy- 5. Naumovski, L. & Friedberg, E. C. (1982)J. Bacteriol. 152, 323- bridization and from the demonstration that a RAD3/rad3-2 331. diploid strain transformed with the plasmid containing the in- 6. Scherer, S. & Davis, R. W. (1979) Proc. Natl. Acad. Sci. USA 76, ternal fragment yielded the expected equal distribution of UV- 4951-4955. resistant and UV-sensitive transformants. In addition, the loss 7. St. John, T. P., Scherer, S., McDowell, M. W. & Davis, R. W. of UV resistance due to inactivation of the RAD3 gene by in- (1981) J. Mol. Biol. 152, 317-334. tegration was accompanied by loss of viability of half the hap- 8. Davis, R. W., Botstein, D. & Roth, J. R. (1980) Advanced Bac- terial Genetics (Cold Spring Harbor Laboratory, Cold Spring loid spores from the rad- diploids, indicating that the RAD3 Harbor, NY). gene is required both for excision repair and for cell viability. 9. Vogelstein, B. & Gillespie, D. (1979) Proc. Nati Acad. Sci. USA An obvious question that emerges from these studies relates 76, 615-619. to the viability of previously isolated rad3 haploid mutants (4). 10. Mortimer, R. K. & Hawthorne, D. C. (1969) in The Yeasts, eds. These strains may contain mutations that are localized to a re- Rose, A. H. & Harrison, J. S. (Academic, New York), Vol. 1, pp. gion of the RAD3 gene that is not essential for viability but is 385-460. 11. Southern, E. M. (1975)J. Mol. Biol. 98, 503-517. required for DNA repair. Alternatively, point mutations at many 12. Shortle, D., Haber, J. E. & Botstein, D. (1982) Science 217, 371- locations in the RAD3 gene may not interfere with the essential 373. function(s) of the gene product. Studies on the RAD3 gene us- 13. Pringle, J. R. & Hartwell, L. H. (1981) in The Molecular Biology ing site-specific mutations should shed further light on these of the Yeast Saccharomyces: Life Cycle and Inheritance, eds. speculations. Strathern, J. N., Jones, E. W. & Broach, J. R. (Cold Spring Har- The CDC8 and CDC9 genes of S. cerevisiae are also essential bor Laboratory, Cold Spring Harbor, NY), pp. 97-142. 14. Arendes, J., Kim, K. C. & Sugino, A. (1983) Proc. Natl. Acad. Sci. (13) and are involved in excision repair of DNA (4). The former USA 80, 673-677. gene encodes a single-strand binding protein (14) and the latter, 15. Johnston, L. H. & Nasmyth, K. A. (1978) Nature (London) 174, DNA ligase (15). The essential nature of these genes is not un- 891-893. Downloaded by guest on October 2, 2021