Title: Ssd1p of Saccharomyces cerevisiae Associates with RNA
Abstract: The SSD1 gene has been isolated as a single copy suppressor of many mutants, such as sit4,slk1/bck1, pde2, and rpc31, in the yeast Saccharomyces cerevisiae. Ssd1p has domains showing weak but significant homology with RNase II-related proteins, Cyt4p, Dss1p, VacB, and RNase II, which are involved in the modification of RNA. We found that Ssd1p had the ability to bind RNA, preferably poly(rA), as well as single-stranded DNA. Interestingly, the most conserved domain among the RNase II-related proteins was not necessary for interaction with RNA. Indirect immunofluorescence staining with anti-Ssd1p antibody revealed that Ssd1p was detected mainly in the cytoplasm. Furthermore, sucrose gradient sedimentation analysis demonstrated that Ssd1p was not cofractionated with polyribosomes, suggesting that Ssd1p is not particularly bound to a translationally active subpopulation of mRNA in the cytoplasm. The SSD1 gene has been isolated as a single copy suppressor of many mutants, such as sit4,slk1/bck1, pde2, and rpc31, in the yeast Saccharomyces cerevisiae. Ssd1p has domains showing weak but significant homology with RNase II-related proteins, Cyt4p, Dss1p, VacB, and RNase II, which are involved in the modification of RNA. We found that Ssd1p had the ability to bind RNA, preferably poly(rA), as well as single-stranded DNA. Interestingly, the most conserved domain among the RNase II-related proteins was not necessary for interaction with RNA. Indirect immunofluorescence staining with anti-Ssd1p antibody revealed that Ssd1p was detected mainly in the cytoplasm. Furthermore, sucrose gradient sedimentation analysis demonstrated that Ssd1p was not cofractionated with polyribosomes, suggesting that Ssd1p is not particularly bound to a translationally active subpopulation of mRNA in the cytoplasm. Cellular RNAs do not exist as a free form but as an RNA-protein complex. The proteins that directly associate with RNA are thought to play important roles in the regulation of gene expression at the post-transcriptional level (1Atwater J.A. Wisdom R. VermaI I.M. Annu. Rev. Genet. 1990; 24: 519-541Crossref PubMed Scopus (212) Google Scholar, 2Bandziulis R.J. Swanson M.S. Dreyfuss G. Genes & Dev. 1989; 3: 431-437Crossref PubMed Scopus (490) Google Scholar). In eukaryotic cells, proteins that bind to RNA polymerase II transcripts include both heterogeneous nuclear RNA-binding proteins and cytoplasmic mRNA-binding proteins. Heterogeneous nuclear RNA-binding proteins bind pre-mRNAs and are associated with them during the processing events required for the formation of mature mRNA (1Atwater J.A. Wisdom R. VermaI I.M. Annu. Rev. Genet. 1990; 24: 519-541Crossref PubMed Scopus (212) Google Scholar). Once mRNAs are transported to the cytoplasm, they form cytoplasmic mRNA-binding protein complexes (2Bandziulis R.J. Swanson M.S. Dreyfuss G. Genes & Dev. 1989; 3: 431-437Crossref PubMed Scopus (490) Google Scholar). Cytoplasmic mRNA-binding proteins seem to regulate translation, localization, or stability of mRNA (3Hargrove J.L. Hulsey M.G. Beale E.G. BioEssays. 1991; 13: 667-674Crossref PubMed Scopus (73) Google Scholar). At present, many RNA-binding proteins have been isolated and characterized (4Sachs A.B. Cell. 1993; 74: 413-421Abstract Full Text PDF PubMed Scopus (774) Google Scholar, 5Burd C.G. Dreyfuss G. Science. 1994; 265: 615-621Crossref PubMed Scopus (1734) Google Scholar), but their functions have not been fully understood. In Saccharomyces cerevisiae, the SSD1 gene has been first characterized to suppress the sit4 mutation defective in a protein phosphatase subunit (6Sutton A. Immanuel D. Arndt K.T. Mol. Cell. Biol. 1991; 11: 2133-2148Crossref PubMed Scopus (272) Google Scholar). Not only in this case, but also in many other cases, SSD1 has been isolated as a single copy suppressor of mutation defective in RPC31encoding a subunit of RNA polymerase III (7Stettler S. Chiannilkulchai N. Denmat S.H. Lalo D. Lacroute F. Sentenac A. Thuriaux P. Mol. & Gen. Genet. 1993; 239: 169-176Crossref PubMed Scopus (80) Google Scholar), in PDE2encoding the cyclic AMP phosphodiesterase (8Wilson R.B. Brenner A.A. White T.B. Engler M.J. Gaughran J.P. Tatchell K. Mol. Cell. Biol. 1991; 11: 3369-3373Crossref PubMed Scopus (50) Google Scholar), in BCK1encoding mitogen-activated protein kinase kinase kinase (9Costigan C. Gehrung S. Snyder M. Mol. Cell. Biol. 1992; 12: 1162-1178Crossref PubMed Scopus (202) Google Scholar), inMPK1 encoding mitogen-activated protein kinase (10Lee K.S. Irie K. Gotoh Y. Watanabe Y. Araki H. Nishida E. Matsumoto K. Levin D.E. Mol. Cell. Biol. 1993; 13: 3067-3075Crossref PubMed Scopus (312) Google Scholar), or in G1 cyclin (11Cvrckova F. Nasmyth K. EMBO J. 1993; 12: 5277-5286Crossref PubMed Scopus (141) Google Scholar). These reports indicate that SSD1is involved in many systems. Sutton et al. also reported that there are two alleles of the SSD1 gene; one is calledssd1-d (dead) and the other is called SSD1-V(viable). They described that SSD1-V could suppress the double mutations of ssd1-d and sit4 (6Sutton A. Immanuel D. Arndt K.T. Mol. Cell. Biol. 1991; 11: 2133-2148Crossref PubMed Scopus (272) Google Scholar). We have also isolated the SSD1 gene as the MCS1 gene involved in stable maintenance of the minichromosome (12Uesono Y. Fujita A. Toh-e A. Kikuchi Y. Gene ( Amst. ). 1994; 143: 135-138Crossref PubMed Scopus (42) Google Scholar). TheSSD1/MCS1 gene product was detected as a ∼160-kDa protein in certain wild type strains bearing SSD1-V, such as KA31 or RAY-3A, whereas a protein of this size was not detected in another wild type strain bearing ssd1-d, such as YPH499 (7Stettler S. Chiannilkulchai N. Denmat S.H. Lalo D. Lacroute F. Sentenac A. Thuriaux P. Mol. & Gen. Genet. 1993; 239: 169-176Crossref PubMed Scopus (80) Google Scholar). These findings indicate that SSD1-V is simply a wild type gene andssd1-d is a defective gene. However, the functions ofSSD1 have not yet been clarified. In recent years, it has been reported that SSD1 has a weak but significant similarity with dis3 + ofSchizosaccharomyces pombe (6Sutton A. Immanuel D. Arndt K.T. Mol. Cell. Biol. 1991; 11: 2133-2148Crossref PubMed Scopus (272) Google Scholar, 13Kinoshita N. Goebl M. Yanagida M. Mol. Cell. Biol. 1991; 11: 5839-5847Crossref PubMed Scopus (68) Google Scholar), DSS1 ofS. cerevisiae (14Dmochowska A. Golik P. Stepien P.P. Curr. Genet. 1995; 28: 108-112Crossref PubMed Scopus (56) Google Scholar), vacB of Shigella flexneri (15Tobe T. Sasakawa C. Okada N. Honma Y. Yoshikawa M. J. Bacteriol. 1992; 174: 6359-6367Crossref PubMed Google Scholar), cyt4 of Neurospora crassa(16Turcq B. Dobinson K.F. Serizawa N. Lambowitz A.M. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 1676-1680Crossref PubMed Scopus (20) Google Scholar), zam of Synechocytosis PCC 6803 (17Beuf L. Bedu S. Cami B. Joset F. Plant Mol. Biol. 1995; 27: 779-788Crossref PubMed Scopus (5) Google Scholar), andrnb of Escherichia coli (18Zilhao R. Camelo L. Arraiano C.M. Mol. Microbiol. 1993; 8: 43-51Crossref PubMed Scopus (38) Google Scholar). Some of these genes are known, or implied, to be involved in the modification of RNAs: 1)cyt4 is required for the mitochondrial rRNA splicing and processing reaction; 2) DSS1 is a multicopy suppressor of the disruptant of SUV3 encoding a putative RNA helicase-like protein; 3) the vacB mutation reduces the level of the virulence antigens, IpaB, IpaC, IpaD, and VirG, at the post-transcriptional level; and 4) the RNase II encoded byrnb has a 3′-to-5′ exoribonuclease activity. However, there have been no reports describing direct interaction with RNA in these gene products, except for RNase II of E. coli. Here we report the biochemical characterization and cellular localization of the Ssd1 protein. RAY-3A (MATa ura3 leu2 trp1 his3) and YRM1H (RAY-3A ssd1Δ::HIS3) were used for the ribonuclease assay, nucleotide binding studies, metabolic labeling with [32P]orthophosphate, and sucrose gradient fractionation of cell extract. KA31–2A (MATa ura3 leu2 trp1 his3) (19Irie K. Takase M. Lee K.S. Levin D.E. Araki H. Matsumoto K. Oshima Y. Mol. Cell. Biol. 1993; 13: 3076-3083Crossref PubMed Scopus (259) Google Scholar) and YKM1H (KA31–2Assd1Δ::HIS3) were used for testing growth rates and for indirect immunofluorescence microscopy. The SSD1disruption was performed by using pYK907 plasmid as described previously (12Uesono Y. Fujita A. Toh-e A. Kikuchi Y. Gene ( Amst. ). 1994; 143: 135-138Crossref PubMed Scopus (42) Google Scholar). Culture media, including YPD (1% yeast extract, 2% peptone, and 2% glucose) and synthetic minimal SD (0.7% yeast nitrogen base without amino acid and 2% glucose) with amino acid supplements, were prepared according to Rose et al. (20Rose M.D. Winston F. Hieter P. Methods in Yeast Genetics: Alaboratory Course Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY1990Google Scholar). SRaf contained 0.7% Difco yeast nitrogen base without amino acids, 2% raffinose, and appropriate supplements. Plasmid pFK4 was constructed by inserting the 6-kilobase BamHI fragment bearing theSSD1 gene derived from the original clone, pFK2CU (12Uesono Y. Fujita A. Toh-e A. Kikuchi Y. Gene ( Amst. ). 1994; 143: 135-138Crossref PubMed Scopus (42) Google Scholar), into pUC13. For overexpression of the SSD1 gene, a plasmid pSSD1.0 was made as follows. pYES2 (Invitrogen), a high copy number plasmid carrying the URA3 marker and the inducibleGAL1 promotor, was digested with HindIII, and the DNA ends were made flush with a Klenow fragment of DNA polymerase I, digested with BamHI, and then ligated with the 5.2-kilobaseHpaI-BamHI fragment bearing the SSD1gene derived from pFK4. Translational initiation of Ssd1.0p is started at the ATG at position 52 in the open reading frame of SSD1. Therefore, Ssd1.0p expressed from pSSD1.0 plasmid encodes the protein lacking 17 amino acids of the N terminus. pSSD1.1 was constructed as follows. After pSSD1.0 was digested with XhoI andHindIII, the DNA ends were made flush and ligated. pSSD1.2 was constructed by digestion of pSSD1.0 with XbaI, and the resulting large fragment was recircularized. Plasmid pFK1CU or plasmid pFK5EU contains SSD1 on YCUp4 or YEUp3 vector (constructed by Fujita), respectively (12Uesono Y. Fujita A. Toh-e A. Kikuchi Y. Gene ( Amst. ). 1994; 143: 135-138Crossref PubMed Scopus (42) Google Scholar). YRM1H cells carrying pYES2, pSSD1.0, pSSD1.1, or pSSD1.2 were grown in SRaf-Ura to early log phase and transferred to SRaf-Ura containing 0.5% galactose followed by incubation at 30 °C for 4 h. For the preparation of cellular extract, 5 × 107 cells were harvested by centrifugation and washed with ice-cold lysis buffer (100 mm Tris-HCl, pH 7.5, 200 mm NaCl, 1 mm EDTA, 5% glycerol, 0.5 mm dithiothreitol, 1% aprotinin, and 5 μg/ml each of leupeptin, pepstatin A, and antipain). Cells were resuspended in 150 μl of lysis buffer and lysed by shaking four times with an equal amount of glass beads for 30-s intervals. An additional 200 μl of lysis buffer was added, and mixtures were shaken again for 30 s. Extract was pipetted out and centrifuged at 12,000 × g for 15 min to remove cell debris. Immunoprecipitation of each Ssd1p derivative, gel electrophoresis, and immunoblotting experiments were performed as described previously (6Sutton A. Immanuel D. Arndt K.T. Mol. Cell. Biol. 1991; 11: 2133-2148Crossref PubMed Scopus (272) Google Scholar). An RNA binding experiment was carried out by incubating the electroblotted proteins with 32P-labeled RNA as described previously (21Lee F.-J.S. Moss J. J. Biol. Chem. 1993; 268: 15080-15087Abstract Full Text PDF PubMed Google Scholar). Radiolabeled total RNA were prepared from yeast cells by using PUREscriptTM RNA isolation kits (Gentra system Co., Ltd), and treated with RNase-free DNase (Sigma). Yeast cells of YRM1H carrying pFK1CU or YCUp4 were grown in SD-Ura to midlog phase at 25 °C. Early log phase cells of YRM1H carrying pSSD1.0 grown in SRaf-Ura were transferred to SRaf-Ura containing 0.5% galactose or 2% glucose and incubated for 4 h at 25 °C. Cellular extracts (200 μl) were prepared from approximately 5 × 107 cells as described above, and immunoprecipitations were performed using 20 μl of anti-Ssd1p antibody, as described previously (6Sutton A. Immanuel D. Arndt K.T. Mol. Cell. Biol. 1991; 11: 2133-2148Crossref PubMed Scopus (272) Google Scholar). Immunoprecipitants were washed twice with buffer for measurement of exoribonuclease and resuspended in 75 μl of the same buffer, and exoribonuclease assays were carried out. The assays of exoribonuclease were based on the release of acid-soluble radioactivity from [3H]poly(rA) or 32P-labeled total RNA according to the procedure as described previously (27Noguchi E. Hayashi N. Azuma Y. Seki T. Nakamura M. Nakashima N. Yanagida M. He X. Mueller U. Sazer S. Nishimoto T. EMBO J. 1996; 15: 5595-5605Crossref PubMed Scopus (78) Google Scholar, 28Piedade J. Zilhãbo R. Arraiano C.M. FEMS Microbiol. Lett. 1995; 127: 187-194Crossref PubMed Google Scholar, 29Stevens A. J. Biol. Chem. 1980; 255: 3080-3085Abstract Full Text PDF PubMed Google Scholar). For preparation of yeast lysates, RAY-3A cells were grown in YPD to midlog phase, and early log phase cells of YRM1H carrying pSSD1.0, pSSD1.1, or pSSD1.2 were grown in SRaf-Ura containing 0.5% galactose for 4 h at 25 °C. Approximately 1 × 108 cells were lysed by shaking with glass beads in 300 μl of binding buffer (10 mm Tris (pH 7.4), 1.5 mm MgCl2, 100 mm NaCl, 0.5% Triton X-100, 1 mmphenylmethylsulfonyl fluoride, 1% aprotinin, and 5 μg/ml each of leupeptin, pepstatin A, and antipain), an additional 200 μl of binding buffer was added, and cellular extracts were prepared by glass beads shearing as described above. Using this extract, single-stranded DNA (ssDNA), double-stranded DNA (dsDNA) chromatography, and ribohomopolymer (Sigma) binding assay, were performed as described previously (22Matunis M.J. Matunis E.L. Dreyfuss G. Mol. Cell. Biol. 1993; 13: 6114-6123Crossref PubMed Scopus (38) Google Scholar). Indirect immunofluorescence microscopy of yeast cells was performed with a modification of previously published procedures (23Pringle J.R. Preston R.A. Adams A.E.M. Stearns T. Drubin D.G. Haarer B.K. Jones E.W. Methods Enzymol. 1989; 31: 357-435Google Scholar). KA31–2A cells carrying pFK5EU, a multicopy plasmid carrying the SSD1 gene, were grown to early log phase in SD-Ura, and formaldehyde was directly added to a final concentration of 5%. After incubation for 1 h at room temperature, cells were washed twice with 100 mmKH2PO4, pH 7.5, and resuspended in 1 ml of buffer S (100 mm KH2PO4, pH 7.5, 1.2 m sorbitol) containing 20 μg of zymolyase 100T (Seikagaku) and 0.1% 2-mercaptoethanol and incubated for 30 min at 30 °C. Cells were rinsed twice with buffer S and twice with phosphate-buffered saline containing 3% bovine serum albumin. Cells were resuspended in 200 μl of the same buffer containing 2 μl of anti-Ssd1p antiserum (100:1) and incubated for 4 h at 30 °C. Cells were applied to polylysine-coated glass slides for 0.5 h and washed five times with phosphate-buffered saline containing 0.1% bovine serum albumin. Cells were incubated with the same buffer containing fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (100:1) for 1 h at 30 °C. Samples were washed five times with phosphate-buffered saline and added with 0.5 μg/ml 4′,6-diamidino-2-phenylindole. For the preparation of low salt extract, approximately 3 × 108 early log phase cells were lysed by vortexing with glass beads in 400 μl of standard extraction buffer A (20 mm Tris-HCl (pH 7.4), 2 mm MgCl2, 10 mm NaCl, 1 mm dithiothreitol, 1 mm phenylmethylsulfonyl fluoride, 1% aprotinin, and 5 μg/ml each of leupeptin, pepstatin A, and antipain). Cell debris were removed by centrifugation for 10 min at 5,000 × g. Then one half (150 μl) of the supernatant was incubated with RNase A (10 μg/ml) plus micrococcal nuclease (200 μg/ml) at 30 °C for 30 min, and the other half (150 μl) was incubated under the same conditions without enzyme (24Liang S. Hitomi M. Hu Y. Liu Y. Tartakoff A.M. Mol. Cell. Biol. 1996; 16: 5139-5146Crossref PubMed Scopus (81) Google Scholar). Each sample was layered onto 12 ml of a continuous 10–30% sucrose gradient and centrifuged at 35,000 rpm in a Beckman SW-40 rotor for 18 h at 4 °C. For a polysome preparation, 1.5 × 108 early log phase cells were treated with cycloheximide and were processed as described previously (25Sachs A.B. Davis R.W. Cell. 1989; 58: 857-867Abstract Full Text PDF PubMed Scopus (392) Google Scholar). For the preparation of extract that contains decaying polysome, 1.5 × 108 midlog phase cells were collected, and cell extracts were prepared by glass bead shearing (26Kraig E. Haber J. J. Bacteriol. 1980; 144: 1098-1112Crossref PubMed Google Scholar). Samples were fractionated through a continuous 15–50% sucrose gradient by centrifugation at 40,000 rpm in a Beckman SW-40 rotor for 2.5 h at 4 °C. A 254 of the gradient fraction was monitored using the Pharmacia FPLC system and Buchler Auto Densi-Flow IIC. It has been reported that the Ssd1 protein shows weak similarity with the Cyt4 protein, which is related to a mitochondrial RNA splicing and processing factor of N. crassa (16Turcq B. Dobinson K.F. Serizawa N. Lambowitz A.M. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 1676-1680Crossref PubMed Scopus (20) Google Scholar). Another report demonstrated the conserved domains in several proteins involving not only Cyt4p but also RNase II, a 3′ to 5′ exoribonuclease, encoded by rnb of E. coli (14Dmochowska A. Golik P. Stepien P.P. Curr. Genet. 1995; 28: 108-112Crossref PubMed Scopus (56) Google Scholar). Therefore, we compared the amino acid sequence of Ssd1p with those of other proteins. Computer search analysis using the GENETYX program (Software Development Co., Ltd.) revealed that Ssd1p has similarities with SpDis3p of S. pombe (23.1% identity in 565 amino acids ofSsd1p), ScDis3p of S. cerevisiae (24.2% identity in 505 amino acids) (27Noguchi E. Hayashi N. Azuma Y. Seki T. Nakamura M. Nakashima N. Yanagida M. He X. Mueller U. Sazer S. Nishimoto T. EMBO J. 1996; 15: 5595-5605Crossref PubMed Scopus (78) Google Scholar), Dss1p of S. cerevisiae (12.4% identity in 290 amino acids), Cyt4p of N. crassa (21.0% identity in 372 amino acids), VacB of Shigella flexneri(20.1% identity in 387 amino acids), and Rnb of E. coli(20.9% identity in 373 amino acids). The homology search demonstrated that similarities are restricted to the C-terminal half of Ssd1p and that three conserved domains exist in this region (Fig.1 A). The first domain of Ssd1p was located around a region from 689 to 785 (domain 1; D1), the second was a region from 880 to 910 (domain 2; D3), and the last was a region from 983 to 1014 (domain 3; D3) in Ssd1p. Of these domains, domain 3 is the most highly conserved among these proteins (Fig. 1 B). In addition to the proteins shown in Fig. 1, the protein predicted from the sequence of F48E8.6of Caenorhabditis elegans (BLAST network service) also has these domains. These findings indicate that domains 1, 2, and 3 are conserved from prokaryote to eukaryote. To test whether these conserved domains are necessary for the function of SSD1 or not, we constructed several SSD1 derivatives (Fig.2 A). SSD1.0, encoding a protein lacking 17 amino acids of the N terminus of Ssd1p expressed under theGAL1 promotor on a multicopy vector (see "Materials and Methods"), could complement the temperature sensitivity of YKM1H, thessd1 disruptant of the KA31–2A strain; probably, it was expressed from the GAL1 promotor on SRaf medium containing glucose. This result indicated that SSD1.0 encoded a functional protein (Fig. 2 B, left panel, 37 °C). In contrast to this, the gene lacking both domains 2 and 3 (SSD1.1) or all of domains 1, 2, and 3 (SSD1.2) could not complement the temperature sensitivity of YKM1H (Fig. 2 B, left panel, 37 °C). To our surprise, overexpression ofSSD1.0 in the existence of galactose inhibited the growth of YKM1H at 25 °C, while SSD1.1 or SSD1.2 did not (Fig. 2 B, right panel). Thus, the SSD1 gene requires the region including domains 2 and 3 for its function. In the case of dis3 + of S. pombe, a similar result indicating that the most conserved region is necessary for its activity has been described previously (13Kinoshita N. Goebl M. Yanagida M. Mol. Cell. Biol. 1991; 11: 5839-5847Crossref PubMed Scopus (68) Google Scholar). The growth-inhibited cells did not show any characteristic morphology. This growth inhibition was more remarkable in the strain KA31 or W303 than in another wild type strain, RAY-3A (data not shown). Since the sequence of Ssd1p has similarity with that of the RNase II of E. coli as described above, we tested whether Ssd1p had an exoribonuclease activity. The immunoprecipitants were prepared by using anti-Ssd1p antibody from the lysates of the wild type, the ssd1 disruptant, and theSSD1-overexpressing cells. Using these immunoprecipitants, we examined an ability of Ssd1p to degrade 3H-labeled poly(rA) or 32P-labeled total RNA extracted from yeast cells under the following several assay conditions of RNase II ofE. coli (28Piedade J. Zilhãbo R. Arraiano C.M. FEMS Microbiol. Lett. 1995; 127: 187-194Crossref PubMed Google Scholar), 5′ to 3′ exoribonuclease (Xrn1p) of S. cerevisiae (29Stevens A. J. Biol. Chem. 1980; 255: 3080-3085Abstract Full Text PDF PubMed Google Scholar), or mitochondrial 3′ to 5′ exoribonuclease ofS. cerevisiae (30Min J. Heuertz R.M. Zassenhaus H.P. J. Biol. Chem. 1993; 268: 7350-7357Abstract Full Text PDF PubMed Google Scholar). However, we were unable to detect any exoribonuclease activity with these immunoprecipitants. We next examined whether Ssd1p could bind RNA in vitro. Ssd1.0p and derivatives from it were immunoprecipitated from extracts of yeast cells overexpressing each SSD1 derivative using anti-Ssd1p antibody and electroblotted onto a nitrocellulose filter after SDS-polyacrylamide gel electrophoresis. The molecular masses of Ssd1.0p, Ssd1.1p, and Ssd1.2p were about 160, 120, and 95 kDa, respectively, when detected by immunoblotting analysis (Fig.3, left panel). The molecular weights of Ssd1.0p, Ssd1.1p, and Ssd1.2p deduced from amino acid sequences were 140, 95, and 76 kDa, respectively, suggesting that all of these proteins are modified and that the modified regions may reside in the N-terminal half of Ssd1p, which does not show any significant similarity with other proteins. The filter was incubated with32P-labeled total RNA extracted from yeast cells. Only Ssd1.1p of 120 kDa lacking both D2 and D3 could bind to32P-labeled RNA, while Ssd1.0p and Ssd1.2p could not (Fig.3, right panel). This result indicates that Ssd1.1p can associate with RNA directly without any binding proteins and that the region containing D1 is necessary for RNA binding. The reason why Ssd1.0p did not bind to RNA is unclear at this moment. We cannot exclude a possibility that native Ssd1.0p and Ssd1.2p can bind RNA; therefore, it is important to test whether the native Ssd1p has an ability to bind RNA. To detect an ability of native Ssd1p to bind polynucleotide, we carried out DNA-cellulose chromatography. The same amount of extracts from the yeast cells overexpressing Ssd1.0p or its derivatives was mixed with dsDNA- and ssDNA-cellulose in 0.1 or 1.0 m NaCl, and proteins were eluted with 2 m NaCl. Ssd1.0p was able to bind to ssDNA at 0.1 m NaCl (Fig. 4 A, lane 1) but not to dsDNA efficiently (Fig. 4 A, lane 7), whereas Ssd1.2p lacking all conserved domains weakly bound ssDNA but not dsDNA (Fig.4 A, lanes 3, 4, 9, and10). Thus, Ssd1.0p has an ability to bind single-stranded DNA, and the conserved regions seem to be necessary for interaction. However, Ssd1.1p lacking the most highly conserved region, domain 3, could bind ssDNA more efficiently than Ssd1.0p (Fig. 4 A,lanes 1 and 5). These results indicate that Ssd1p can bind ssDNA without the most highly conserved region, as seen in the case of Fig. 3, and that a region, other than the conserved regions, may bind ssDNA. Interestingly, Ssd1.1p could bind dsDNA as well as ssDNA (Fig. 4 A, lane 11), suggesting that the region including both domains 2 and 3 is necessary for specific binding to single-stranded polynucleotides. To further characterize the RNA binding property of Ssd1p, we examined whether Ssd1p or its derivatives was bound to four ribohomopolymers. This binding assay has been successful in distinguishing the specificities of a variety of RNA-binding proteins (22Matunis M.J. Matunis E.L. Dreyfuss G. Mol. Cell. Biol. 1993; 13: 6114-6123Crossref PubMed Scopus (38) Google Scholar). Extracts from the wild type cells were mixed with poly(rA)-, poly(rU)-, poly(rG)-, and poly(rC)-agarose in 0.1 or 1.0 m NaCl. The ribohomopolymer-binding proteins were eluted with SDS sample buffer. Basically, Ssd1p bound efficiently to the all kinds of ribohomopolymers in 0.1 m NaCl, whereas it bound to poly(rA) more efficiently than to the other polynucleotides under a high salt condition (Fig. 4 B, top). The same result was obtained, using the extract of cells overproducing Ssd1.0p (Fig.4 B, middle). Ssd1.1p lacking both domains 2 and 3 seemed to bind efficiently with poly(rU) in 1.0 m NaCl (Fig. 4 B, bottom). This finding suggests that the poly(rA)-specific binding of Ssd1p may depend on the region including both domains 2 and 3. It is important to determine the localization of Ssd1p because the nucleic acid-binding proteins exert their functions in their respective compartments. Indirect immunofluorescence microscopy using anti-Ssd1p antibody was carried out to determine the subcellular localization of Ssd1p. To facilitate detection of the immunofluorescence, the SSD1 gene was cloned on a multicopy YEUp3 vector (pFK5EU) to overexpressSSD1 in cells. The affinity-purified anti-Ssd1p antibody revealed an intense signal in cytoplasm and a weak signal in the nucleus in the ssd1 disruptant cells (YKM1H) carrying pFK5EU. The signal was always observed in the cytoplasm, irrespective of budded or unbudded cells (Fig. 5, A andC), indicating that Ssd1p mainly stays in the cytoplasm throughout the cell cycle. As a reference, signals were not observed in the ssd1 disruptant cells carrying YEUp3 vector (Fig.5 E). These observations suggest that Ssd1p may associate with RNA in the cytoplasm but not DNA or RNA in the nucleus; therefore, Ssd1p may play some roles in the stability or turnover of cytoplasmic RNA rather than its maturation. To know whether Ssd1p can associate with cellular RNA or not, low salt extracts prepared from the wild type cells were fractionated on a 10–30% continuous sucrose density gradient, and each fraction was analyzed by immunoblotting using anti-Ssd1p antibody. Ssd1p showed a broad distribution in fractions 7–19 (Fig. 6, upper panel). However, after treatment of extracts with RNase A and micrococcal nuclease, Ssd1p was mainly sedimented in the upper fractions 7 and 8, where the monomeric Ssd1 protein (160 kDa) was expected to sediment (Fig. 6, lower panel). The fact that Ssd1p in the extract treated with nuclease also sedimented in the faster sedimenting fractions indicates that some population of Ssd1p may form complexes with certain protein or nucleic acids. Thus, we presumed that Ssd1p associated with RNA in vivo. Cytoplasmic mRNAs exist either in translationally active form or translationally inactive form. In mammalian cells, the histone mRNA degradation occurs at the 3′ terminus and appears to be catalyzed by a polyribosome-associated 3′ to 5′ exoribonuclease. In fact, Caruccioet al. (31Caruccio N. Ross J. J. Biol. Chem. 1994; 269: 31814-31821Abstract Full Text PDF PubMed Google Scholar) has purified a polyribosome-associated 3′-to-5′ exoribonuclease in human cells. In yeast, it has been reported that some of RNA-binding proteins, like Pab1p and Pub2p, are associated with actively translating mRNAs, while Pub1p is not (32Anderson J.T. Paddy M.R. Swanson M.S. Mol. Cell. Biol. 1993; 13: 6102-6113Crossref PubMed Scopus (72) Google Scholar). To test whether Ssd1p was associated with polyribosomal mRNAs, cellular RNAs were fractionated by sucrose gradient centrifugation. The distribution of Ssd1p in the gradient was determined by immunoblot analysis. In cycloheximide-treated extract in which polyribosomes accumulated, Ssd1p was distributed from 80 S to the top fraction. A small amount of the protein was distributed in polyribosome fractions (Fig. 7 A). A similar distribution pattern of Ssd1p was obtained when extract without cycloheximide treatment was analyzed (Fig. 7 B). This result indicates that Ssd1p is not stably associated with actively translated mRNA. mRNA stability influences gene expression by affecting mRNA abundance and the rate at which mRNA accumulates or disappears when the transcription rate changes. In S. cerevisiae, it has been known that there are two major pathways for mRNA degradation. One is the so-called deadenylation-dependent pathway, and the other is the deadenylation-independent pathway (33Muhlrad D. Paker R. Nature. 1994; 370: 578-581Crossref PubMed Scopus (329) Google Scholar, 34Beelman C.A. Stevens A. Caponigro G. LaGrandeur T.E. Hatfield L. Fortner D.M. Paker R. Nature. 1996; 382: 642-646Crossref PubMed Scopus (277) Google Scholar). Both degradation pathways require XRN1 encoding a 5′ to 3′ exoribonuclease (35Kenna M. Stevens A. McCammon M. Douglas M.G. Mol. Cell. Biol. 1993; 13: 341-350Crossref PubMed Scopus (94) Google Scholar, 36Larimer F.W. Stevens A. Gene ( Amst. ). 1990; 95: 85-90Crossref PubMed Scopus (97) Google Scholar, 37Masison D.C. Blanc A. Ribas J.C. Carrol K. Sonenberg N. Wickner R.B. Mol. Cell. Biol. 1995; 15: 2763-2771Crossref PubMed Scopus (107) Google Scholar). Subsequently, the study using the xrn1 mutant showed the existence of 3′ to 5′ exoribonuclease (33Muhlrad D. Paker R. Nature. 1994; 370: 578-581Crossref PubMed Scopus (329) Google Scholar). In mitochondria of S. cerevisiae, an NTP-dependent 3′ to 5′ exoribonuclease has also been isolated and characterized (30Min J. Heuertz R.M. Zassenhaus H.P. J. Biol. Chem. 1993; 268: 7350-7357Abstract Full Text PDF PubMed Google Scholar, 38Min J. Zassenhaus H.P. J. Bacteriol. 1993; 175: 6245-6253Crossref PubMed Google Scholar). However, the gene encoding 3′ to 5′ exoribonuclease of cytoplasm has not been identified yet. There are no reports identifying a gene encoding a 3′ to 5′ exoribonuclease, except for rnb of E. coli. The SGD (Saccharomyces Genome Data base) project revealed that three Saccharomyces genes, DSS1,YOL021C/ScDIS3, and SSD1, encode domains showing weak similarities with RNase II of E. coli. These gene products may be candidates of the 3′ to 5′ exoribonuclease of S. cerevisiae. However, we could not detect exoribonuclease activity of Ssd1p using poly(rA) or total RNA as a substrate. Highly purified preparations of mitochondrial 3′ to 5′ exoribonuclease of S. cerevisiae have shown to form a complex and require NTP-dependent RNA helicase for its activity (30Min J. Heuertz R.M. Zassenhaus H.P. J. Biol. Chem. 1993; 268: 7350-7357Abstract Full Text PDF PubMed Google Scholar, 39Margossian S.P. Li H. Zassenhaus H.P. Butow R.A. Cell. 1996; 84: 199-209Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). It contains three major polypeptides estimated to be 75, 90, and 110 kDa, which is predicted to be the gene product of DSS1 (29Stevens A. J. Biol. Chem. 1980; 255: 3080-3085Abstract Full Text PDF PubMed Google Scholar). In addition, the Dis3 protein of S. pombe has also reported to be part of a 250–350-kDa oligomer (13Kinoshita N. Goebl M. Yanagida M. Mol. Cell. Biol. 1991; 11: 5839-5847Crossref PubMed Scopus (68) Google Scholar). Thus, Ssd1p may also need another component for an exoribonuclease activity. As shown in Fig. 6, most of Ssd1p was in monomeric fraction after treatment with RNase, and a small fractions of Ssd1p remained as stable complexes. Ssd1p may form a stable complex not only with proteins but also with certain RNA molecules, and formation of such a complex may be necessary for expressing exoribonuclease activity. Post-translational modification of Ssd1p also may contribute to the expression of exoribonuclease activity; Ssd1p is phosphorylated in vivo (12Uesono Y. Fujita A. Toh-e A. Kikuchi Y. Gene ( Amst. ). 1994; 143: 135-138Crossref PubMed Scopus (42) Google Scholar), suggesting that the yet unidentified ribonuclease activity of Ssd1p may be tightly regulated by a certain protein kinase. Ssd1.1p, lacking the highly conserved region domain 3, could bind RNA more efficiently than Ssd1.0p as shown in Fig. 3, therefore, domain 3 seems to inhibit the association of Ssd1p with RNA. If Ssd1p is exoribonuclease, domain 3 may be necessary for its exoribonuclease activity rather than RNA binding, because the SSD1 gene requires domain 3 for its entire function as shown in Fig. 2. The major known RNA-binding motifs (5Burd C.G. Dreyfuss G. Science. 1994; 265: 615-621Crossref PubMed Scopus (1734) Google Scholar) are not seen in the sequence of Ssd1p. Interestingly, the region necessary for interaction with RNA seems to be a region having domain 1 showing a weak similarity but not a region having highly conserved domain 3. Therefore, the region containing domain 1 of Ssd1p may be a new RNA binding motif. The DSS1 gene has been isolated as a multicopy suppressor of the disruptant of SUV3, encoding putative RNA helicase, involved in mitochondrial RNA metabolism. It has been reported that the amino terminus of Dss1p is predicted to have a mitochondrial targeting sequence and that the dss1 disruptant is viable but does not grow in a glycerol medium (14Dmochowska A. Golik P. Stepien P.P. Curr. Genet. 1995; 28: 108-112Crossref PubMed Scopus (56) Google Scholar). From these findings, Dss1p seems to localize in mitochondria. On the other hand, the ScDis3p of S. cerevisiae shows a high identity with the SpDis3p of S. pombe throughout the length (27Noguchi E. Hayashi N. Azuma Y. Seki T. Nakamura M. Nakashima N. Yanagida M. He X. Mueller U. Sazer S. Nishimoto T. EMBO J. 1996; 15: 5595-5605Crossref PubMed Scopus (78) Google Scholar), while Ssd1p and Dss1p show low identities. In addition, Noguchi et al. (27Noguchi E. Hayashi N. Azuma Y. Seki T. Nakamura M. Nakashima N. Yanagida M. He X. Mueller U. Sazer S. Nishimoto T. EMBO J. 1996; 15: 5595-5605Crossref PubMed Scopus (78) Google Scholar) have also reported that ScDIS3 was able to rescue the dis3mutant of S. pombe, while Kinoshita et al. (13Kinoshita N. Goebl M. Yanagida M. Mol. Cell. Biol. 1991; 11: 5839-5847Crossref PubMed Scopus (68) Google Scholar) have reported that SSD1 could not. These findings indicate that ScDIS3, but not SSD1, is a counterpart ofdis3 + of S. pombe. Thedis3 + gene product of S. pombelocalizes mainly in the nucleus (13Kinoshita N. Goebl M. Yanagida M. Mol. Cell. Biol. 1991; 11: 5839-5847Crossref PubMed Scopus (68) Google Scholar), suggesting that theScDIS3 gene product localizes in nucleus. Intracellular localization of Dss1p and ScDis3p is in a clear contrast to that of Ssd1p, which was mainly localized in the cytoplasm as shown in Fig. 5. These observations suggest an interesting possibility that these three proteins of S. cerevisiae having similarity with RNase II ofE. coli, Dss1p, ScDis3p, and Ssd1p associate with mitochondrial RNA, nuclear RNA, and cytoplasmic RNA, respectively. The fact that Ssd1p was mainly localized in cytoplasm suggests that the target of Ssd1p could be RNA in cytoplasm including rRNA, mRNA, or others. It has been reported that the cyt4 mutant inN. crassa had several defects including maturation of rRNA in mitochondria (40Garriga G. Bertrand H. Lambowitz A.M. Cell. 1984; 36: 623-634Abstract Full Text PDF PubMed Scopus (72) Google Scholar). However, rRNA prepared from the ssd1disruptant did not show any remarkable difference in its maturation or concentration, in comparison with rRNA prepared from wild type cells (data not shown). In addition, Ssd1p was not cofractionated either with ribosomes or with polysomes, as shown in Fig. 7. Thus, the target of Ssd1p does not seem to be rRNA. We could not detect Ssd1p in UV-cross-linked polyadenylated RNA-RNPs (data not shown). Therefore, we have no evidence to indicate direct interaction of Ssd1p with mRNA. Immunoblotting analysis using anti-Ssd1p antibody recognizing its N-terminal region revealed that the Ssd1-d2 protein in W303 used as wild type strain was detected as an 83-kDa protein (data not shown). This protein may be a C-terminal truncated protein lacking all of the domain 1, 2, and 3 regions, suggesting that the Ssd1-d2 protein may also be a nonfunctional protein. The Ssd1-d protein of another wild type strain, YPH499, was also the same size as that of W303 (data not shown). These results suggest that a ssd1-d mutation is widely spread among many laboratories' strains. The ssd1-dmutation results in a subtle phenotype, such as caffeine-sensitive (6Sutton A. Immanuel D. Arndt K.T. Mol. Cell. Biol. 1991; 11: 2133-2148Crossref PubMed Scopus (272) Google Scholar) and leaky temperature-sensitive. However, when it is combined with another mutation such as sit4, the double mutation causes a severer temperature-sensitive phenotype (synthetic lethal) than either of the single mutations (6Sutton A. Immanuel D. Arndt K.T. Mol. Cell. Biol. 1991; 11: 2133-2148Crossref PubMed Scopus (272) Google Scholar, 7Stettler S. Chiannilkulchai N. Denmat S.H. Lalo D. Lacroute F. Sentenac A. Thuriaux P. Mol. & Gen. Genet. 1993; 239: 169-176Crossref PubMed Scopus (80) Google Scholar, 8Wilson R.B. Brenner A.A. White T.B. Engler M.J. Gaughran J.P. Tatchell K. Mol. Cell. Biol. 1991; 11: 3369-3373Crossref PubMed Scopus (50) Google Scholar, 9Costigan C. Gehrung S. Snyder M. Mol. Cell. Biol. 1992; 12: 1162-1178Crossref PubMed Scopus (202) Google Scholar, 10Lee K.S. Irie K. Gotoh Y. Watanabe Y. Araki H. Nishida E. Matsumoto K. Levin D.E. Mol. Cell. Biol. 1993; 13: 3067-3075Crossref PubMed Scopus (312) Google Scholar, 11Cvrckova F. Nasmyth K. EMBO J. 1993; 12: 5277-5286Crossref PubMed Scopus (141) Google Scholar). It is reasonable that theSSD1 gene has been obtained as a single copy suppressor in various screens. Mutations that show synthetic lethality withssd1-d seem to be involved in the transcriptional regulation of certain genes: 1) sit4 (41Fernandez-Sarabia M.J. Sutton A. Zhong T. Arndt K.T. Genes & Dev. 1992; 6: 2417-2428Crossref PubMed Scopus (115) Google Scholar), 2) bcy1 (42Tanaka K. Matsumoto K. Toh-e A. EMBO J. 1988; 7: 495-502Crossref PubMed Scopus (95) Google Scholar, 43Marchler G. Schuller C. Adam G. Ruis H. EMBO J. 1993; 12: 1997-2003Crossref PubMed Scopus (413) Google Scholar, 44Cherry J.R. Johnson T.R. Dollard C. Schuster J.R. Denis C.L. Cell. 1989; 56: 409-419Abstract Full Text PDF PubMed Scopus (144) Google Scholar, 45Matsuura A. Treinin M. Mitsuzawa H. Kassir Y. Uno I. Simchen G. EMBO J. 1990; 9: 3225-3232Crossref PubMed Scopus (54) Google Scholar), 3) mpk1 (46Watanabe Y. Irie K. Matsumoto K. Mol. Cell. Biol. 1995; 15: 5740-5749Crossref PubMed Scopus (171) Google Scholar), and 4) rpb1, rpc31, andrpc53 (7Stettler S. Chiannilkulchai N. Denmat S.H. Lalo D. Lacroute F. Sentenac A. Thuriaux P. Mol. & Gen. Genet. 1993; 239: 169-176Crossref PubMed Scopus (80) Google Scholar). These genetic characteristics displayed by thessd1 mutation, in addition to the fact that Ssd1p is a cytoplasmic RNA-binding protein, suggest that SSD1 may be involved in the expression of various genes at the post-transcriptional level by controlling RNA metabolism.