The Fission Yeast Protein Ker1p Is an Ortholog of RNA Polymerase I Subunit A14 in Saccharomyces cerevisiae and Is Required for Stable Association of Rrn3p and RPA21 in RNA Polymerase I*

A heterodimer formed by the A14 and A43 subunits of RNA polymerase (pol) I in Saccharomyces cerevisiae is proposed to correspond to the Rpb4/Rpb7 and C17/C25 heterodimers in pol II and pol III, respectively, and to play a role(s) in the recruitment of pol I to the promoter. However, the question of whether the A14/A43 heterodimer is conserved in eukaryotes other than S. cerevisiae remains unanswered, although both Rpb4/Rpb7 and C17/C25 are conserved from yeast to human. To address this question, we have isolated a Schizosaccharomyces pombe gene named ker1

Much attention has recently been focused on the A14 and A43 subunits in view of the structural and functional conservation of these two subunits in eukaryotes. A43 is conserved in a variety of eukaryotes (10) and shows amino acid sequence similarity to Rpb7 (a specific subunit of pol II), C25 (a specific subunit of pol III), and RpoE (a subunit of archaeal RNA polymerases) across multiple RNA polymerases (11). Furthermore, A43 forms a heterodimer with A14 that is similar to the Rpb4/Rpb7 (11,12), C17/C25 (13), and RpoF/RpoE (14) heterodimers in pol II, pol III, and archaeal RNA polymerases, respectively. It should be noted that Rpb4, C17, and RpoF have mutual sequence similarity and are grouped into a gene family, but no obvious homolog of A14 has been found in available data bases. A14 and Rpb4 are required for the stable assembly of A43 and Rpb7, respectively, in their respective RNA polymerases, suggesting a functional similarity of A14 to Rpb4 (5,11,15,16). The position of A14/A43 in the three-dimensional structure of pol I has been deduced to be similar to that of Rpb4/Rpb7, forming an upstream interface with the C-terminal domain of Rpb1 to interact with transcription factor IIB for pol II recruitment to the pol II promoter (4) and, furthermore, playing a role in the processing of the nascent RNA transcript (17). Consistent with the proposed position in pol I, A14/A43 also interacts with an rDNA-specific transcription factor (Rrn3p) for pol I recruitment to the rDNA promoter (10) and is able to bind to single-stranded RNA (18). Interestingly, C17/ C25 in pol III is also reported to interact with transcription factor IIIB, which recruits pol III to the pol III promoter (19).
The mechanism of the down-regulation of rDNA transcription (20 -24) is now believed to be as follows. Only a small fraction of pol I associated with Rrn3p is able to recognize the components of the preinitiation complex, resulting in pol I recruitment to the rDNA promoter (25)(26)(27)(28). A43 in pol I is responsible for associating with Rrn3p (10), and the association of A43 with Rrn3p is inhibited in post-log-phase cells (including nutrient-starved or growth-arrested cells) (24,26), resulting in a drastic decrease in pol I recruitment to the promoter (29). * This work was supported by the Core Research for Evolutional Science and Technology of the Japan Science and Technology Corporation, by a Human Frontier Science Program Organization grant, and by a grant for the promotion of the advancement of education and research in graduate schools from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the DDBJ/GenBank TM /EBI Data Bank with accession number(s) AB07137.
Thus, the molecular function of A43 and Rrn3p deserves further study to resolve long-standing questions regarding growthdependent transcription of rDNA (30).
It is firmly established that all 12 and all 17 subunits of S. cerevisiae pol II and pol III, respectively, are conserved in human pol II and pol III (31,32). However, it is not clear whether all 14 subunits identified in S. cerevisiae pol I are conserved in other eukaryotes (33,34). To gain further insight into the structure and function of pol I, we have been studying pol I of Schizosaccharomyces pombe, which is only distantly related to S. cerevisiae, but is amenable to genetic analysis (35). To date, it is known that S. pombe pol I consists of at least 12 subunits. The two largest, RPA190 and RPA140, are homologous to A190 and A135, respectively (36,37). The two smaller subunits, RPA42 and RPA17, correspond to AC40 and AC19, respectively (38,39). Five common subunits (Rpb5, Rpb6, Rpb8, Rpb10, and Rpb12) are shared by pol II and pol III (12), and SpRPA12 is a functional homolog of A12.2 (40). Thus, the 10-subunit core structure of pol I has been well conserved between the two yeasts through evolution. Moreover, the two specific subunits in S. pombe, RPA21 and RPA51, have been identified to be related to A43 and as a functional homolog of A49, respectively, suggesting that the pol I architecture in S. pombe is likely to be analogous to that in S. cerevisiae (41,42).
In this study, we demonstrate that a newly isolated protein, Ker1p, is an ortholog of A14 and that the Ker1p/RPA21 heterodimer in S. pombe is the functional counterpart of A14/A43 in S. cerevisiae. We also show novel aspects of Ker1p that have not been previously observed in A14 and suggest that Ker1p is involved in growth-dependent transcription of rDNA.

EXPERIMENTAL PROCEDURES
Media, Strains, and Genetic Techniques-The yeast plasmids and strains used are listed in Table I. Minimal medium with or without thiamine and supplemented with appropriate amino acids and yeast extract/dextrose medium were prepared to grow S. pombe cells as described previously (35). Yeast extract/peptone/dextrose medium and synthetic dextrose medium were prepared as described previously (43). Synthetic dextrose medium lacking tryptophan and leucine and synthetic dextrose medium containing 25 mM 3-amino-1,2,4-triazole (3-AT) were used. Minimal medium containing 0.1-0.4 g/ml aureobasidin A was also used. Disruption of chromosomal ker1 ϩ was carried out as follows. Diploid cells (a cross between JY742 and JY745 cells) were transformed with the 4.5-kb XhoI-SacI linear fragment containing ker1⌬::ura4 ϩ from pYI186. To replace Ker1p with Ker1p-HA 3 , a 5.0-kb XhoI-SacI fragment from pYI176 (see below) was transformed into strain IZ2, resulting in YI28.
Plasmids-For two-hybrid screening, Gal4DB-rpa21 ϩ (pYI77) was constructed as follows. 0.52-kb rpa21 ϩ cDNA was amplified by PCR using pKI45 containing a full-length 0.52-kb cDNA of rpa21 ϩ (41) as a template and cloned between the SmaI and XhoI sites of pAS2-1 (44) to fuse Gal4DB to RPA21 in-frame. To construct deletion derivatives of rpa21 ϩ , pKI45 was also used as template DNA for PCR amplification. pYI106 expresses RPA21 with the N-terminal 56 amino acids truncated fused to Gal4DB, whereas pYI105 expresses RPA21 with the C-terminal 60 amino acids truncated fused to Gal4DB. To replace chromosomal ker1 ϩ with ker1 ϩ -HA 3 , a PCR-amplified 1317-bp XhoI-SmaI fragment of the 5Ј-untranslated region and open reading frame of ker1 ϩ and a 1017-bp NotI-SacI fragment of the 3Ј-flanking region of ker1 ϩ were cloned successively between the XhoI and SmaI sites and the NotI and SacI sites of pYN1237, generating pYI176. To express Ker1p under the control of the cytomegalovirus (CMV) promoter in S. pombe, full-length ker1 ϩ (441 bp) was amplified from JY742 DNA by PCR and cloned between the XhoI and BamHI sites of pAUR222 (TaKaRa), resulting in pYI195. To study the cellular localization of Ker1p, full-length ker1 ϩ was amplified from JY742 DNA, and the 441-bp fragment was cloned between the NotI and BamHI sites of pKS406 to express a GFP-Ker1p fusion protein, resulting in pYI193. A GFP-fibrillarin fusion construct was made using a 912-bp fragment of the fib gene encoding fibrillarin, which was amplified by PCR from JY742 DNA and cloned between the NotI and BamHI sites of pKS406, resulting in pYI200. To construct a disrupted ker1⌬::ura4 ϩ allele, we amplified the 880-bp 5Ј-untranslated sequence of ker1 ϩ flanked by the XhoI and BamHI sites and the 1.0-kb 3Ј-untranslated sequence of ker1 ϩ flanked by the NotI and SacI sites Derivative of KS(ϩ) cloned with full-length cDNA of rpa21 ϩ between XhoI and EcoRI sites pYI77 Derivative of pAS2-1 expressing Gal4DB-RPA21, TRP1, 2m pKS406 Derivative of pSGA (47) expressing GFP fusion proteins under control of nmt1 promoter, ars1, LEU2 pYI193 Derivative of pKS406 expressing GFP-Ker1p pYI200 Derivative of pKS406 expressing GFP-fibrillarin pYN1235 Derivative of pGEM3ZpBR322ura4 ϩ pBR322 with BamHI site introduced into HindIII site (55) pYN1237 Derivative of KS(ϩ); HA 3 -TAG sequence (triple-HA epitope tagged with TAG) inserted between SmaI and SpeI sites; 2.5-kb BamHI-BamHI ura4 ϩ fragment excised from pYN1235 inserted at BglII site created downstream of HA 3 -TAG sequence pAUR224 Expression vector under control of CMV promoter, aur1 r , ars1 pYI195 Derivative of pAUR224 expressing full-length Ker1p under control of CMV promoter pYI176 Derivative of pYN1237 carrying 1.0-kb fragment of 5Ј-flanking and coding sequence of Ker1p-HA 3 , ura4 ϩ , and 1.0-kb fragment of 3Ј-flanking sequence of ker1 ϩ pGK100 Derivative of pDB248 carrying nuc1 ϩ /rpa190 ϩ (37) pYI185 Derivative of KS(ϩ) carrying 0.88-kb fragment of 5Ј-flanking region of ker1 ϩ between XhoI and BamHI sites and 1.0kb fragment of 3Ј-flanking region of ker1 ϩ between NotI and SacI sites pYI186 Derivative of pYI185 with ura4 ϩ inserted at BamHI site pREP41 Expression vector under control of modified nmt1 promoter, LEU2, ars1 (56) pREP81 Expression vector under control of modified nmt1 promoter, LEU2, ars1 (56) pYI40 Derivative of pREP81 with NdeI site converted to BglII site (41) pYI82 Derivative of pYI40 expressing rpa21 ϩ cDNA under control of weak nmt1 promoter (41)  pKI27 Derivative of pREP81 expressing full-length rrn3 ϩ under control of nmt1 promoter pYI210 Derivative of pYI40 expressing full-length ker1 ϩ under control of nmt1 promoter from the JY742 genome. Each PCR product was cloned successively between the XhoI and BamHI sites and between the NotI and SacI sites of pBluescript II KS(ϩ), resulting in pYI185. Then, the 2.5-kb BamHI-BamHI DNA fragment of ura4 ϩ obtained from pYN1235 was cloned into the BamHI site of the resulting plasmid, generating pYI186. To construct pKI27, full-length rrn3 ϩ was amplified by PCR from JY742 DNA and cloned between the SalI and SmaI sites of pREP81. To express ker1 ϩ under the control of the nmt1 promoter, full-length ker1 ϩ was amplified by PCR from JY742 DNA and cloned between the SalI and BamHI sites of pYI40, generating pYI210. Pfu DNA polymerase was used for PCR, and DNA sequencing analysis was used to confirm the PCR product. Two-hybrid Screening-pYI77 expressing a Gal4DB-RPA21 bait was transformed into the reporter strain Y190. Y190 carrying pYI77 was transformed with an S. pombe cDNA library fused to Gal4 activation domain in pGAD-GH (Clontech). The 3-AT-resistant and His ϩ transformants were screened on synthetic dextrose medium plates without Trp and Leu and containing 25 mM 3-AT. lacZ activation was examined by a filter lifting assay (38).
Fluorescence Microscopy of GFP Fusion Proteins-To visualize the nuclear chromatin region, cells were stained with 4Ј,6-diamidino-2phenylindole (DAPI) at 1 mg/ml. Fluorescent images were obtained with a Fujix HC-2500 CCD camera using a Zeiss Axioskop fluorescence microscope.
Immunoprecipitation-S. pombe cells were grown in yeast extract/ dextrose medium and harvested in mid-log phase. Preparation of cell extracts and immunoprecipitation with anti-HA epitope monoclonal antibody 12CA5 (Roche Applied Science) and anti-RPA190 antibody were carried out as described by Mitsuzawa et al. (45). Immunoblotting was performed essentially as described previously (39) using polyclonal antibodies against RPA190, RPA140, RPA21, and Rpb1 (pol II) (41,46).
Biochemical Fractionation of pol I-pol I was partially purified as described previously (46). Whole cell extract from strain YI28 was loaded onto a nickel-nitrilotriacetic acid-agarose column. The proteins eluted with 200 mM imidazole were loaded onto a DEAE-Sephadex A25 column and eluted with a 50 -620 mM ammonium sulfate gradient. Fractions were examined by SDS-PAGE, followed by Western blotting using antibodies against RPA190, RPA21, and HA. (Ker1p was tagged with HA 3 in strain YI28.) Phosphatase Treatment-Whole cell extract was prepared form strain YI28, and 1.6 mg of protein was immunoprecipitated with anti-RPA190 antibody (10 l of antiserum) as described above. The precipitates were washed three times with 20 mM HEPES-KOH (pH 7.6), 150 mM potassium acetate, 20% glycerol, 0.1% Nonidet P-40, and 1 mM dithiothreitol and once with HM buffer (50 mM HEPES-KOH (pH 7.6) and 1 mM MgCl 2 ). The pellet was resuspended in 1 ml of HM buffer, divided into four aliquots, centrifuged again, resuspended with 100 l of HM buffer, and incubated for 10 min at 30°C. Calf intestine alkaline phosphatase (30 units, 1.5 l; Roche Applied Science) was added to one tube and incubated for 20 min at 30°C. The reaction was stopped by addition of SDS sample buffer and heating at 95°C for 5 min. In controls, sodium pyrophosphate (final concentration of 5.4 mM) was added with or without alkaline phosphatase, and the sample was then treated as described above. No treatment was performed for the fourth sample. All samples were subjected to 8% SDS-PAGE, followed by immunoblot analysis with anti-HA antibody.

Identification of a Novel Protein (Ker1p) That
Interacts with the RPA21 Subunit-To identify protein(s) that interact with RPA21, we generated a Gal4DB-RPA21 fusion construct in pAS2-1 (pYI77) and introduced it into the S. cerevisiae twohybrid reporter strain Y190. Subsequently, we introduced an S. pombe cDNA library fused to the Gal4 activation domain into the Y190 strain carrying pYI77. We selected ϳ10 7 Leu ϩ transformants and screened colonies showing 3-AT resistance and a lacZ-positive phenotype. In total, 27 transformants showing 3-AT resistance and the lacZ-positive phenotype were obtained, and the responsible plasmids carrying cDNA fused to the Gal4 activation domain were retrieved (data not shown). Nucleotide sequencing of the retrieved plasmids indicated that all of the cDNAs encoding the protein shown to interact with RPA21 were derived from the same gene; one group lacked the C-terminal 30 amino acids, and another retained the full-length gene, indicating that the C-terminal 30 amino acids are not required for interaction with RPA21 in the yeast two-hybrid method. The gene isolated by the two-hybrid system encodes a protein of 147 amino acids with a calculated molecular mass of 16,976 Da and a calculated pI of 6.25. The predicted protein is very hydrophilic and contains many charged amino acids: 21 lysine residues, 9 arginine residues, 24 glutamic acid residues, and 7 aspartic acid residues (see Fig. 7). Therefore, we have named this protein Ker1p (for lysine (K) and glutamic acid (E)-rich protein 1) and the gene encoding it ker1 ϩ . No proteins homologous to Ker1p were observed in an initial data base search.
Apparent Molecular Mass of Ker1p-HA 3 -To determine the apparent molecular mass of Ker1p, a YI28 strain expressing Ker1p-HA 3 was constructed. Whole cell extracts prepared from YI28 and the parental strain IZ2, in which Ker1p had not been tagged, were subjected to SDS-PAGE, followed by immunoblotting with anti-HA monoclonal antibody 12CA5. Fig. 1A shows that Ker1p-HA 3 was detected as a doublet of bands at 30 and 32 kDa, including a triple-HA sequence (4.3 kDa). Since the calculated molecular mass of Ker1p is ϳ17 kDa, it appears that Ker1p-HA 3 migrates abnormally on SDS-polyacrylamide gel, for unknown reasons.
Ker1p Is Phosphorylated-The predicted amino acid sequence of Ker1p suggested that it contains many consensus phosphorylation sites for protein kinase A (Ser 14 ), protein kinase C (Ser 14 Fig. 7). We considered the possibility that Ker1p is phosphorylated and that both phosphorylated and non-phosphorylated forms were detected as doublet bands by immunoblotting in Fig. 1A. Therefore, Ker1p-HA 3 was first immunoprecipitated with anti-HA antibody, and the immunoprecipitates were then treated with alkaline phosphatase in the absence or presence of a phosphatase inhibitor. As shown in Fig. 1B Fig. 1A (lane 2). Therefore, we conclude that the 30-kDa band represents non-phosphorylated Ker1p-HA 3 and that the 32-kDa band represents phosphorylated Ker1p-HA 3 .

Ker1p Is Localized Predominantly in the Nucleolus-Because
Ker1p interacts with RPA21 of pol I, which localizes specifically in the nucleolus, we examined whether Ker1p also localizes in the nucleolus using a GFP-Ker1p fusion protein. Fig. 2 (A-C) shows that GFP-Ker1p formed a dense, crescent-shaped structure that occupied one side of the nucleus and that the crescentshaped region was not stained well by DAPI. The observed crescent-shaped structure with much reduced DNA staining is the most obvious characteristic of the yeast nucleolus (37,47,48). However, GFP-Ker1p was also observed in the DAPIstained region (Fig. 2B), and it might be possible that GFP-Ker1p also localizes outside the nucleolus due to overproduction under the control of the strong nmt1 promoter. For controls, we examined localization of GFP itself and observed a clear cytoplasmic distribution (Fig. 2, D-F). We also examined the localization of the nucleolar protein fibrillarin (GFP-fibrillarin) expressed in the same GFP fusion vector and found that it localized specifically in the crescent-shaped nucleolus (Fig. 2, G-I). As expected, the crescent-shaped region in the nucleus shows much lower DAPI staining in Fig. 2G. Taken together, we conclude from these results that Ker1p is predominantly localized in the nucleolus, although it is also possible that a certain fraction of Ker1p localizes in the nucleoplasm.
Co-immunoprecipitation of Ker1p and pol I-The apparent nucleolar localization of Ker1p prompted us to study the physical interaction of RPA21 (pol I) and Ker1p in vivo. Therefore, extracts prepared from cells expressing Ker1p-HA 3 were immunoprecipitated with anti-HA antibody (12CA5) beads, and coprecipitated proteins were detected by immunoblotting. Fig. 3A shows that pol I subunits RPA190, RPA140, and RPA21 co-immunoprecipitated with Ker1p, suggesting that Ker1p associates with pol I in vivo. The association is specific for pol I because the pol II subunit Rpb1 was not co-immunoprecipitated (Fig. 3A, fourth panel). Conversely, Ker1p co-immunoprecipitated with RPA140 and RPA21 when anti-RPA190 antibody was used for immunoprecipitation (Fig. 3B). These results confirm that Ker1p associates with pol I in vivo. No subunit was precipitated without the specific antibodies (lane 3). It should be noted that the bands of Ker1p again appeared to be doublets (lane 2), suggesting that the upper and lower bands correspond to the 32-and 30-kDa forms of the protein, respectively, as shown in Fig. 1B. The results suggest that pol I associates with both phosphorylated and non-phosphorylated Ker1p.
Ker1p Is Co-fractionated Biochemically with pol I-The above results suggest that Ker1p is a pol I subunit. To confirm biochemically that Ker1p is a novel pol I subunit, we purified pol I from an S. pombe strain (YI28) expressing both Ker1p-HA 3 and RPA140 tagged with a His 6 -FLAG epitope. The whole cell extract was first affinity-purified using a nickel-agarose column and then fractionated by DEAE-Sephadex A25 column chromatography. We observed that Ker1p co-eluted with the peak fractions (fractions 13 and 14) of pol I detected through the RPA190 and RPA21 subunits (Fig. 4). Although more rigorous biochemical purification is needed, the elution pattern through the DEAE column confirms that Ker1p is a subunit of pol I. We noted that fractions 11 and 12 might also contain pol I without Ker1p and RPA21, but we did not examined these  (lanes 2 and 3) were derived from the light chain of immunoglobulin G. fractions further in this study.
Genetic Interaction between Ker1p and RPA21-We have previously shown that overproduced Rrn3p is able to suppress the temperature-sensitive growth defect of rpa21 mutants, indicating a genetic interaction between RPA21 and Rrn3p (41). To examine the genetic interaction between Ker1p and RPA21, we introduced a multicopy vector expressing Ker1p under the control of the CMV promoter into the three temperature-sensitive rpa21 mutants (ts152, ts296, and ts2817) (41) and examined whether the growth defects of the mutants were suppressed at the restrictive temperature. As shown in Fig. 5, overproduction of Ker1p clearly suppressed the growth deficiency of the three mutants, indicating a genetic interaction between Ker1p and RPA21. Since A14 was shown previously to be required for stable association of A43 with pol I in S. cerevisiae (11), it appears that Ker1p is also required for stable association of RPA21 with pol I.
Ker1p Has a Limited Homology to S. cerevisiae A14 -No apparent homologous protein was initially found when the Ker1p sequence was used in a data base search. However, the results obtained above clearly indicate that Ker1p is a pol I subunit and that it interacts with RPA21. Because RPA21 is an ortholog of A43 of S. cerevisiae, we re-examined the homology of Ker1p to S. cerevisiae A14, which heterodimerizes with A43. Previously, A14 was suggested to have homology to a putative open reading frame of IPF1568 from Candida albicans (11), suggesting that the A14 gene family is conserved in C. albicans, although no genetic or biochemical evidence was presented. This suggestion prompted us to directly investigate the homologies among Ker1p, A14, and IPF1568 (hereafter referred to as C. albicans A14), and Fig. 6 shows an alignment of these proteins constructed using ClustalW. We found that Ker1p shows 21% identity and 27% similarity to the 126-amino acid sequence of S. cerevisiae A14 and that the N-terminal 60 amino acids of Ker1p show especially high identity (37%) and similarity (43%) to S. cerevisiae A14. The N-terminal region also shows significant identity between the A14 subunits of S. cerevisiae and C. albicans (11), and Ker1p shows 26% identity and 36% similarity to the 132-amino acid sequence of C. albicans A14. The local identity of Ker1p to the S. cerevisiae and C. albicans A14 subunits is especially high (38 and 42%, respectively) between amino acids 42 and 65 of Ker1p, and this region contains a motif that may be conserved between the S. cerevisiae and C. albicans A14 subunits (SQLKRIQR), as already suggested by Peyroche et al. (11). We conclude from these results that Ker1p is an ortholog of A14 from both S. cerevisiae and C. albicans. ker1 ϩ Is Required for Growth Only at High Temperatures-To examine whether ker1 ϩ is essential for cell growth, we replaced one of the chromosomal copies of ker1 ϩ with a disrupted ker1⌬::ura4 ϩ allele in the S. pombe diploid. A Ura ϩ transformant (YI29) was chosen and subjected to tetrad analysis upon sporulation (Fig. 7A). Of the 20 asci dissected, one yielded four viable spores; seven yielded three viable spores; and the remaining 12 yielded two viable spores on yeast extract/peptone/dextrose medium plates at 30°C. Large colonies were invariably Ura Ϫ , whereas all of the small colonies, including the extremely small ones, were Ura ϩ . Correct disruption of the ker1 ϩ locus in the Ura ϩ segregant YI30 was verified by PCR (data not shown). Growth of the colonies was tested at 25, 30, and 36°C; none of the Ura ϩ colonies grew at 36°C, but all of the Ura Ϫ colonies did, indicating that ker1 ϩ is not essential for cell growth at 30 or 25°C, but is required for cell growth at 36°C (Fig. 7B).
ker1⌬ Is Suppressed by Overproduction of RPA21 or Rrn3p-We subsequently examined whether complete deletion of ker1 ϩ (ker1⌬) is also suppressed by overproduction of RPA21. Fig. 8 shows that overexpression of RPA21 suppressed the growth defect of ker1⌬, again suggesting (see Fig. 5) that Ker1p is required for stable association of RPA21 with pol I. This result also suggests that RPA21 can associate with pol I independently of Ker1p. Because the rDNA-specific transcription factor Rrn3p interacts with RPA21 (41), we also examined Each of the temperaturesensitive mutants (ts152, ts296, and ts2817) transformed with pYI195 (ker1 ϩ ) or pAUR222 (vector) was re-streaked on minimal medium with thiamine (ϩthi) containing aureobasidin A and incubated for 5 days at 36°C. ker1 ϩ was expressed under the control of the CMV promoter. FIG. 6. Amino acid sequence alignment of S. pombe Ker1p (Sp) with S. cerevisiae A14 (Sc) and C. albicans IPF1568 (Ca). Identical and similar residues are highlighted in black and gray, respectively. The alignment was generated with the ClustalW program using the ID matrix. (The default BLOSUM matrix yielded a different alignment, in which Ker1p appeared to be less homologous to S. cerevisiae A14.) whether overexpression of Rrn3p is able to suppress the temperature-sensitive growth of ker1⌬. As shown in Fig. 8, overproduction of Rrn3p suppressed the temperature-sensitive phe-notype of ker1⌬, suggesting that Ker1p interacts with Rrn3p and directly stabilizes the association of Rrn3p with pol I. However, the alternative possibility remains that overproduced Rrn3p can interact with RPA21 and perhaps stabilize the association of RPA21 with pol I without the participation of Ker1p. Fig. 8 also shows that multiple copies of rpa190 ϩ were unable to suppress the ker1⌬ phenotype.
Dissociation of Ker1p from pol I in Post-log-phase Cells-The data in Fig. 8 suggest that Ker1p is involved in stabilizing the association of Rrn3p with pol I, either directly or indirectly. Because Rrn3p is released from pol I in post-log-phase or growth-arrested cells (23,24,26), we examined whether Ker1p is also released from pol I, resulting in destabilization of Rrn3p in pol I in post-log-phase cells. As shown in Fig. 9, pol I from cells expressing Ker1p-HA 3 in mid-and post-log phase was immunoprecipitated with anti-RPA190 antibody, and the relative amounts of RPA190, RPA140, and Ker1p were compared. We observed a drastic decrease in the ratio of Ker1p (both phosphorylated and non-phosphorylated) to RPA190 in pol I prepared from post-log-phase cells (Fig. 9A, right panels, compare lanes 2 and 3 with lane 1), although the ratio of RPA140 to RPA190 did not change significantly, suggesting that Ker1p is dissociated from pol I in the post-log phase. The dissociation of Ker1p from pol I may cause instability of Rrn3p in pol I, either directly or indirectly, resulting in dissociation of Rrn3p from pol I, which inactivates rDNA transcription. DISCUSSION In this study, we have shown that Ker1p isolated by a yeast two-hybrid system using RPA21 as bait is the counterpart of the S. cerevisiae pol I subunit A14. This is the first demonstration that a pol I-specific A14 ortholog is conserved in eukaryotes other than S. cerevisiae, despite no apparent homolog of A14 being identified in the S. pombe genome. We have successfully aligned the amino acid sequence of almost the entire length of Ker1p with those of A14 and IPF1568 (Fig. 6), indicating that these subunits are indeed grouped into a gene family. Our investigation of Ker1p has, however, also revealed features of Ker1p that are distinct from those of A14. First, Ker1p is phosphorylated (Fig. 1), whereas phosphorylation of FIG. 7. Gene disruption of ker1 ؉ . A: left panel, the 4.5-kb XhoI-SacI DNA fragment used for ker1 ϩ disruption; right panel, tetrads of diploids (ker1⌬::ura4 ϩ /ker1 ϩ ) grown at 30°C. B, growth of four haploid segregants derived from the ascospore produced by tetrad dissection of the ker1⌬::ura4 ϩ /ker1 ϩ diploid.
FIG. 8. Temperature-sensitive growth of a ker1⌬ strain is suppressed by overexpression of rpa21 ؉ or rrn3 ؉ , but not rpa190 ؉ . vector, ker1 ϩ , rpa190 ϩ , rrn3 ϩ , and rpa21 ϩ indicate strain YI28 (ker1⌬) carrying pREP41 (an empty vector), pYI210 (ker1 ϩ ), pGK100 (rpa190 ϩ ), pKI27 (rrn3 ϩ ), and pYI82 (rpa21 ϩ ). Each transformant was re-streaked on minimal medium plates without leucine and thiamine and incubated at 36 or 30°C for 5 days.  2 and 3) at the times indicated in B. Left panels, for detection of each subunit in the crude extract, 50 g of each extract was loaded onto an SDS-polyacrylamide gel. Ker1p-HA 3 , RPA190, RPA140, and Rpb1 (pol II for a control) were detected by immunoblot analysis with antibodies against HA, RPA190, RPA140, and Rpb1, respectively. Right panels, pol I in the crude extract was immunoprecipitated with anti-RPA190 antibody. The immunoprecipitates (IP) were subjected to SDS-PAGE, followed by immunoblot analysis with antibodies against HA, RPA190, RPA140, and Rpb1. We noted that bands detected by anti-Rpb1 antibody were nonspecific. KER1-P, phosphorylated Ker1p; KER1, Ker1p. B, strain YI28 was grown in yeast extract/dextrose medium at 30°C. The culture was harvested at the indicated times (time points 1, 2,  and 3). The A 600 values of the culture at time points 1, 2, and 3 were ϳ1.0 (mid-log phase), ϳ5.3, and ϳ6.8 (post-log phase), respectively. The doubling times at each point were ϳ2 h (time point 1), ϳ6 h (time point 2), and ϳ20 h (time point 3). In this study, we defined a temporary slow growth phase, such as those at time points 2 and 3, as a post-log phase. A14 has never been observed. Second, Ker1p is suggested to interact, either directly or indirectly, with Rrn3p and to stabilize the association of Rrn3p with pol I in vivo (Fig. 8), whereas it is unclear whether A14 affects the stability of Rrn3p with pol I in vivo. Third, Ker1p is released from pol I in post-log-phase cells (Fig. 9), whereas such instability of A14 in post-log-phase or growth-arrested cells is unknown in S. cerevisiae. The significance of these differences must await future studies to determine whether A14 can be phosphorylated or does dissociate from pol I in post-log-phase cells.
A comparison of the pol I subunits in S. pombe and S. cerevisiae is shown in Table II. 10 subunits constituting the core structure (RPA190, RPA140, RPA42, RPA17, Rpb5, Rpb6, Rpb8, Rpb10, Rpb12, and SpRPA12) are conserved from S. cerevisiae to S. pombe pol I. RPA190 and RPA140 were not examined, but seven of the remaining eight subunits (all except Rpb8) could substitute for the corresponding subunits in S. cerevisiae, suggesting functional conservation of most of the subunits (38 -40, 49). In three specific subunits, including Ker1p, RPA51 was tested and found to rescue an rpa49 mutation in S. cerevisiae (42). Although RPA21 encodes only 174 amino acids and appears to be much diversified from S. cerevisiae A43 (S. cerevisiae A43 contains 326 amino acids), its role in pol I recruitment to the rDNA promoter is conserved (41), and it is plausible that the N-terminal region conserved in the A43 gene family plays a role in the interaction with Rrn3p. The question of whether S. pombe pol I conserves a counterpart of S. cerevisiae A34.5 is unresolved. We believe that the primary sequence of the A34.5 homolog, if any such protein exists, may be poorly conserved in S. pombe. Mouse pol I has been found to contain distinct subunits from lower eukaryotes, such as PAF67 and PAF49 (33,34,50). These results tempt us to speculate that lower eukaryotes such as yeast might conserve the 14 subunits found in S. cerevisiae and that higher eukaryotes might have more specific subunits such as PAF67 and PAF49, in addition to the 14 conserved subunits.
It is known that A14 is not phosphorylated in S. cerevisiae pol I (1, 51, 52), whereas here Ker1p was found to be phosphorylated, suggesting that specific subunit phosphorylation has also evolved independently among pol I subunits. Indeed, S. cerevisiae A43 is multiphosphorylated (53), whereas mammalian A43 is barely phosphorylated (24). It has been argued that A43 must be phosphorylated to associate with Rrn3p in S. cerevisiae, whereas Rrn3p phosphorylation is a prerequisite for the association of A43 with Rrn3p in mammalian cells (24,53,54). In this context, the functional dissection of Ker1p phosphorylation/dephosphorylation may provide a novel insight into the pol I recruitment mechanism.
It appears that Ker1p is required for the stability of RPA21 based on multicopy suppression experiments. (i) ker1⌬ exhibits a temperature-sensitive growth deficit, and the temperaturesensitive deficit is suppressed by overproduction of RPA21 (Fig.  8). (ii) The temperature-sensitive growth deficit of three rpa21 mutants (ts152, ts296, and ts2817) is suppressed by overexpression of ker1 ϩ (Fig. 5). Since it is known that A14 is also required for the stability of A43 in S. cerevisiae (9,11), the role of Ker1p may be, as expected, similar to that of A14. To verify the Ker1p function, purification of pol I from extracts of ker1⌬ mutants deserves future study. Unexpectedly, as shown in Fig. 8, overproduction of Rrn3p also suppresses the temperature-sensitive phenotype of the ker1⌬ mutant, suggesting that Ker1p is also required for the stable association of Rrn3p with pol I. Alternatively, these suppression data could suggest that overproduction of Rrn3p suppresses the instability of RPA21 in pol I without participation of Ker1p, leading to indirect suppression of the temperature-sensitive phenotype of ker1⌬. Clearly, future biochemical experiments are required to reveal whether Rrn3p directly interacts with Ker1p.
Accumulating evidence has shown that dissociation of Rrn3p from pol I in post-log-phase or growth-arrested cells causes a decrease in pol I recruitment to the promoter, resulting in a decrease in or halting of rDNA transcription. As described above, it has been argued that post-translational modification (phosphorylation/dephosphorylation) of A43 and Rrn3p regulates the stability of the pol I-Rrn3p complex, causing dissociation of Rrn3p from pol I in post-log-phase or growth-arrested cells. The immunoprecipitation experiments in Fig. 9, performed using anti-RPA190 antibody, showed that the amounts of both forms of Ker1p (phosphorylated and non-phosphorylated forms) relative to RPA140 and RPA190 are reduced drastically after cells enter the post-log phase. The results suggest that dissociation of Ker1p from pol I in post-log-phase cells is one of the regulatory mechanisms of Rrn3p dissociation from pol I since the dissociation of Ker1p may lead to instability of RPA21 and Rrn3p in pol I. It is also possible that release of Ker1p induces certain modification(s) of RPA21 and Rrn3p that result in release of Rrn3p from pol I. Therefore, it is tempting to speculate that the association/dissociation of Ker1p might primarily regulate growth-dependent transcription of rDNA in S. pombe.