Mouse RNA polymerase I 16-kDa subunit able to associate with 40-kDa subunit is a homolog of yeast AC19 subunit of RNA polymerases I and III.

We have previously isolated a mouse RPA40 (mRPA40) cDNA encoding the 40-kDa subunit of mouse RNA polymerase I and demonstrated that mRPA40 is a mouse homolog of the yeast subunit AC40, which is a subunit of RNA polymerases I and III, having a limited homology to bacterial RNA polymerase subunit α (Song, C. Z., Hanada, K., Yano, K., Maeda, Y., Yamamoto, K., and Muramatsu, M. (1994) J. Biol. Chem. 269, 26976-26981). In an extension of the study we have now cloned mouse RPA16 (mRPA16) cDNA encoding the 16-kDa subunit of mouse RNA polymerase I by a yeast two-hybrid system using mRPA40 as a bait. The deduced amino acid sequence shows 45% identity to the yeast subunit AC19 of RNA polymerases I and III, known to associate with AC40, and a local similarity to bacterial α subunit. We have shown that mRPA40 mutants failed to interact with mRPA16 and that neither mRPA16 nor mRPA40 can interact by itself in the yeast two-hybrid system. These results suggest that higher eukaryotic RNA polymerase I conserves two distinct α-related subunits that function to associate with each other in an early stage of RNA polymerase I assembly.

and pol III; two of them (AC40 and AC19) are shared by pol I and pol III; and the remaining seven subunits (A190, A135, A49, A43, A34.5, A14, and A12.2) are unique to pol I. The largest (A190) and the second largest (A135) are homologous to the ␤Ј and ␤ subunits of Escherichia coli RNA polymerase, respectively. AC40 and AC19 have a limited homology to E. coli ␣ subunit, indicating that they are functional homologs to ␣ (3). Thus analogous to E. coli core RNA polymerase ␣ 2 ␤␤Ј, these four subunits A190, A135, AC40, and AC19 may form a core in pol I. Indeed, AC40 and AC19 were shown to associate in vivo by a yeast two-hybrid system (4). In contrast to the well characterized yeast pol I, pol I from mammalian cells has not been sufficiently purified and analyzed biochemically. To study the structure and subunit composition of mammalian pol I, we have purified mouse pol I to apparent homogeneity and have shown that mouse pol I is composed of at least 11 subunits (180,114,44,40,27,20,18,16,15,14, and 12 kDa) and three associated proteins (PAF53, 51, and 49) and appears to have an organization similar to S. cerevisiae pol I (5,6). Indeed, molecular cloning of the mouse 40-kDa subunit of mouse pol I (hereafter referred to as mRPA40) has revealed that mRPA40 shares a high similarity with yeast AC40 (7) and a limited homology to E. coli ␣ subunit (5). Furthermore, pol I-associated factor PAF53 has been shown to have local similarities to yeast A49 subunit (6). However, the correspondence for the remaining subunits between yeast and mouse is not yet clear. Recently, conservation of five common subunits between yeast and mammalian pol II has been shown with identification of cDNAs encoding human homologs of yeast ABC27, ABC23, ABC14.5, ABC10␣, and ABC10␤ (8 -10). Remarkably, all but one, ABC27, can complement the corresponding null mutations in yeast (9,10). Considering the high conservation of components of RNA polymerases through evolution, we could expect that a smaller subunit related to yeast AC19 (and E. coli ␣) exists in mouse pol I and that the assumed homolog, if any, is able to interact with mRPA40. In this paper, using a yeast two-hybrid system, we have cloned mRPA16 cDNA encoding the 16-kDa subunit able to associate with mRPA40 and found that mRPA16 is indeed a homolog of yeast AC19. We have also shown that mRPA40 mutants failed to associate with mRPA16 and that neither mRPA40 nor mRPA16 could associate by itself. These results suggest that higher eukaryotic pol I conserves two distinct ␣-related subunits and give supportive evidence that they will form a heterodimer in the assembly pathway of pol I.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Media-SD synthetic medium is 2% glucose, 0.67% Bacto yeast nitrogen base (Difco) supplemented with the required bases and amino acids as described by Sherman et al. (11). SDϪLeu lacks leucine. SDϪLeuϪTrp lacks leucine and tryptophan. To select His ϩ transformants, 25 mM 3-amino-(1,2,4)-triazole (3AT), a * This work was supported in part by a research grant from the Asahi Glass Foundation and by a grant-in-aid for Scientific Research from the Ministry of Education, Science, and Culture 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 The yeast and E. coli strains and plasmids used in this study are listed in Table I. pY4 carrying GAL4 DB -mRPA40 fusion gene driven by PADC1 was constructed in the following way. A 1.4-kb SacI-SmaI fragment of PADC1-GAL4 DB cut out from pAS1 (12) was ligated between the SacI and SmaI sites of YEp351 (13). The resultant plasmid was digested with SmaI and BamHI and ligated with a 0.5-kb SmaI-BamHI fragment containing TADC1 derived from pPC97 (14), resulting in pNYN521. The cDNA encoding the entire coding region of mRPA40 (5) modified for subcloning in-frame to GAL4 DB was prepared and inserted between the NcoI and SmaI sites of pNYN521, creating pY4. pYN1068 was constructed as follows. A 0.7-kb SalI-NotI fragment was cut out from pPC67-B22 (Table I) by digestion with SalI and NotI and inserted between the SalI and NotI sites of pPC97 to express in-frame GAL4 DB -B22 (mRPA16) fusion protein. pI81 was identified in the pPC67 library as a clone expressing the GAL4 transcriptional activation domain (GAL4 TA )-mRPA40 (amino acid position 4 -355) fusion protein able to interact with GAL4 DB -B22 (mRPA16) fusion protein produced from pYN1068 in the yeast two-hybrid system. 2 Library Screening for mRPA40-interacting Clones-The yeast strain Y153 bearing UAS G -HIS3 and UAS G -LacZ as reporter genes was transformed with pY4 by the lithium acetate method (15), and Leu ϩ transformants were first selected on SDϪLeu plates. The Leu ϩ transformants (Y153 cells carrying pY4) were subsequently transformed with pPC67, a library of mouse embryonic cDNA fused to GAL4 TA , and the transformants were plated directly on SDϪLeuϪTrpϩ3AT medium. The plates were incubated for 7 days at 30°C, and His ϩ transformants were obtained. The His ϩ colonies were replica plated on SDϪLeuϪTrp medium and were tested for ␤-galactosidase activity by a filter lifting assay (16). LacZ ϩ clones were identified. Plasmid DNA was prepared from candidate clones and electroporated into E. coli MH1066 having trpC9830 that can be complemented by yeast TRP1 to recover the library-derived plasmids. The recovered plasmids were analyzed further as candidates.
DNA Sequencing-The cDNAs from library-derived plasmids were first subcloned between the SalI and NotI sites of pBluescript II KS(ϩ) plasmid because cDNAs were flanked by adaptors containing SalI at the 5Ј junction and NotI at the 3Ј junction site. Several deletions were then constructed, and their DNA sequences were determined by the dideoxy method with BcaBEST DNA polymerase (TaKaRa) on both strands with an ALF DNA sequencer (Pharmacia).
Immunoblot Analysis-The rabbit antibody against B22 protein encoded by B22 cDNA was prepared as follows. A 0.65-kb NruI-NotI fragment (from nucleotide 54 to the end of the cDNA containing the entire coding region of B22) was cut out from pPC67-B22. The NotI site was rendered blunt with the Klenow enzyme and inserted at the SmaI site of pGEX-1 (17) in-frame to the coding region of glutathione Stransferase (GST), resulting in pGEX1-B22. The plasmid pGEX1-B22 was transformed into E. coli TG1, and the synthesis of GST-B22 fusion protein was induced with the addition of 1 mM isopropyl-1-thio-␤-Dgalactoside. The fusion protein was affinity-purified as described (6) using glutathione-Sepharose 4B beads (Pharmacia) followed by a preparative sodium dodecyl sulfate (SDS)-polyacrylamide gel. The fusion proteins were excised from the gel, and polyclonal antibodies were prepared by MBL Co. (Nagano, Japan).
Immunoblot experiments were done as described (6). Pol I purified from mouse ascites cell MH134 as described (5) was subjected to SDSpolyacrylamide gel electrophoresis and transferred to an Immobilon membrane (Millipore). The membrane was treated with anti-B22 antibodies, and horseradish peroxidase-conjugated antibody to rabbit immunoglobulin was used as a second antibody. The bound antibodies were visualized by ECL reagents (Amersham Corp.).

RESULTS AND DISCUSSION
Isolation of Clones Encoding Proteins Able to Interact with mRPA40 Subunit-To identify mouse cDNAs encoding proteins able to associate with mRPA40, the reporter yeast Y153 carrying pY4 expressing GAL4 DB -mRPA40 fusion protein was transformed with mouse cDNA library pPC67. In total, about 1.5 ϫ 10 8 transformants were screened on SDϪLeuϪTrpϩ3AT plates. 250 His ϩ transformants were obtained in the initial His ϩ screening, and 30 of 250 His ϩ clones were found to be LacZ ϩ by the filter-lifting assay. We recovered library-derived plasmids from the 30 positive candidates individually, and the recovered plasmids were introduced into Y153 carrying either pY4 (the bait) or pYN901 (the control) to eliminate false positives. 21 clones showed His ϩ and LacZ ϩ in a GAL4 DB -mRPA40dependent manner. Subsequently the 21 clones were assigned into five groups by restriction enzyme mapping, and 5 representative clones were subjected to DNA sequence determination. Of the five clones sequenced, we found that cDNA designated B22 encodes a sequence that is highly homologous to the yeast pol I and pol III subunit, AC19 encoded by RPC19 gene (yRPC19) (18). Fig. 1 shows that Y153 cotransformed with pY4 (GAL4 DB -mRPA40) and B22 cDNA can activate HIS3 as well as LacZ, resulting in growth on plates containing 3AT (Fig. 1a) and blue color on a filter containing 5-bromo-4-chloro-3-indolyl-␤-D-galactoside (Fig. 1b). B22 encodes a protein of 133 amino acids with a calculated relative molecular mass of 15.1 kDa and a calculated isoelectric point of 6.5. Fig. 2 shows the nucleotide and deduced amino acid sequences of the protein encoded by  (14). pPC97 Derivative of pPC62 (14) to express GAL4 DB -fusion gene under P ADC1 , carrying the same multicloning site with pPC86 (14) LEU2, CEN6, ARSH4, and amp. pYN901 Derivative of YEp351 (13) carrying P ADC1 -GAL4 DB -T ADC1 derived from pPC97, LEU2, 2 and amp. pY4 Carries GAL4 DB derived from pAS1 (12) fused to mRPA40 in-frame, LEU2, 2 and amp. pPC67 Mouse cDNA library fused to GAL4 TA expressed by P ADC1 , TRP1, CEN6, ARSH4, and amp (14). pPC67-B22 The original plasmid recovered from the pPC67 library carrying about 0.7 kb of cDNA encoding B22 (mRPA16) fused to GAL4 AD . pGEX1-B22 Derivative of pGEX1 (17) carrying a glutathione S-transferase-mRPA16 fusion gene. pYN870 Derivative of YEp351 carrying full-length GAL4 at BamHI site. pI81 Derivative of pPC86 expressing GAL4 TA -mRPA40. pYN1068 Derivative of pPC97 expressing GAL4 DB -B22 (mRPA16). pNYN521 Derivative of YEp351 (13) carrying P ADC1 -GAL4 DB (from pAS1)-T ADC1 (from pPC97), LEU2, 2 and amp. pY4-A82D Derivative of pY4 containing an aspartic acid substitution at alanine 82 position of mRPA40 part. pY4-A82R Derivative of pY4 containing an arginine substitution at alanine 82 position of mRPA40 part.
B22 cDNA. The coding region starts at nucleotide 128 and ends at 526, and the polyadenylation signal (AATAAA) lies from 679 to 684. The putative first methionine shows a good context in translation initiation; 5 out 6 nucleotides are in accordance with Kozak's rule (18). The identity and similarity of the deduced amino acid sequence between the ORF of B22 and yeast AC19 are 44 and 83%, respectively (Fig. 3), suggesting that B22 cDNA encodes a mouse homolog of AC19. Almost the entire region of the ORF of B22 is well conserved between yeast and mouse, whereas the amino-terminal part of yeast AC19 (amino acid position 1-21) and the carboxyl-terminal part of the ORF of B22 (amino acid position 121-133) appear to be speciesspecific. Yeast AC19 is a phosphoprotein; the potential phosphorylation sites are observed at amino acid positions 54 -56, 91-97, and 136 -138 (19); however, the corresponding region of mRPA16 is not conserved well. Hence it remains to be determined whether mRPA16 is phosphorylated in vivo.
As anticipated, there are no translation stop codons in-frame preceding the ATG at nucleotide 128 in the original B22 clone. Therefore, to confirm the predicted ORF (133 amino acids), several overlapping cDNA clones were isolated independently by the screening of mouse MH134 ascites cell cDNA libraries with B22 as a probe. We found the longest cDNA contained the same ORF but extended 17 nucleotides (1ϳ17 in Fig. 2) at the 5Ј end of B22. The results indicate that the cloned B22 contains an entire ORF, and ATG at nucleotide 128 is an authentic translation-initiating ATG. The region between GAL4 TA and the first ATG at nucleotide 128 did not appear to be involved in the interaction with mRPA40 since a derivative of B22 consisting of the ORF of 133 amino acids fused directly to GAL4 TA on pPC86 showed the interaction with mRPA40 in the yeast twohybrid system. 3 Northern blot analysis was performed using B22 as a probe and detected an endogenous B22 mRNA in mouse ascites MH134 cells as a single transcript of 0.8ϳ0.9-kb size, which is sufficient for encoding B22. 3 Mouse Pol I 16-kDa Subunit Is a Homolog of Yeast AC19 -Mouse pol I consists of at least 11 subunits, ranging from 180 to 10 kDa (5). To examine which subunit of mouse pol I is encoded by B22 cDNA, we prepared specific polyclonal antibodies against the polypeptide encoded by B22. Purified mouse pol I was subjected to SDS-polyacrylamide gel electrophoresis followed by Western blot analysis using anti-B22 antibodies. We found that anti-B22 antibodies detect a subunit with an apparent molecular mass of 16 kDa (Fig. 4), which is very close to the calculated molecular mass of B22 (15.1 kDa). We have designated the protein encoded by B22 cDNA mouse RPA16 (mRPA16) according to the apparent molecular mass of the subunit on SDS-polyacrylamide gel. The results show that the 16-kDa subunit, mRPA16, in mouse pol I is a homolog of yeast AC19 shared by yeast pol I and pol III. The demonstration that mammalian pol I contains a homolog to yeast AC19 and the homolog, mRPA16, can interact with mRPA40, which is a homolog to the AC40 subunit, raises the question of whether both mRPA40 and mRPA16 are also contained in mammalian pol III since AC19 and AC40 are shared by yeast pol I and pol III (7,19). Western blot analysis using the antibodies against mRPA40 shows that human pol III contains a subunit corresponding to mRPA40, suggesting that mRPA40 is shared by mouse pol I and pol III. 4 It remains to be determined whether the mRPA16 subunit is present in mammalian pol III.
Two Distinct ␣-Related Subunits in Both Pol I and Pol II-In addition to mRPA40 identified previously, our identification of the mRPA16 underscores the existence of two ␣-related subunits in eukaryotic nuclear RNA polymerases. Two distinct ␣-related subunits have also been identified in both yeast and human pol II: B44.5 and B12.5 in yeast (20,21) and hRPB33 and hRPB14 in human (22,23). Furthermore, we have recently identified mouse RPB14, a homolog to yeast B12.5, 4 and mouse RPB31, a homolog to yeast B44.5, in mouse pol II. 2 Taken together, it is now clear that two distinct ␣-related subunits are conserved in eukaryotic pol I and pol II in contrast to prokaryotic RNA polymerases containing a homodimer of ␣ subunit. The function of the ␣ subunit in prokaryotes may be assigned to the two different subunits in eukaryotes. It should be noted that the RNA polymerase of archaebacteria Sulfolobus acidocaldarius composed of 13 subunits also contains a subunit D corresponding to AC40 (or B44.5) and a subunit L related to AC19 (or B12.5); D and L form a subcomplex in the RNA polymerase (24,25).
Neither mRPA16 nor mRPA40 Can Associate by Itself in the Yeast Two-hybrid System-In E. coli it has been established that the assembly pathway of core polymerase is: 27). Hence, dimerization of ␣ is an important step for the initiation of enzyme assembly in E. coli. In yeast, earlier work suggested that ␣-related subunits like B44.5 or AC40 play roles in the assembly pathway of cognate enzymes. For example, using temperature-sensitive mutants of B220 (␤Ј homolog), B150 (␤ homolog), and B44.5 (␣-related subunits), Kolodziej and Young (28) observed similar steps for the assembly of core pol II, i.e. B44.5 and B150 form a subcomplex first, and the subcomplex subsequently associates with B220. In case of pol I (and pol III), AC40 is proposed to act at the early step of pol I (and pol III) assembly since heat inactivation of temperature-sensitive AC40 caused a complete absence of stable subcomplexes of both pol I and pol III (7). Furthermore, Lalo et al. (4) suggested that AC19 and AC40 would associate with each other, but neither would by itself. From these results we can infer that ␣-related subunits in

TABLE II
Protein interaction tests for mRPA40 and mRPA16 proteins in the yeast two-hybrid system: quantitative assay of ␤-galactosidase activity Fusion proteins were expressed from plasmids pYN870 (GAL4 fulllength), pPC97 (GAL4 DB ), pPC86 (GAL4 TA ), pY4 (GAL4 DB -mRPA40), pI81 (GAL4 TA -mRPA40), pYN1068 (GAL4 DB -mRPA16), and pPC67-B22 (GAL4 TA -mRPA16). Double transformants were grown on SD-Leu-Trp medium, and ␤-galactosidase activity was assayed as described by Guarente (32). ␤-Galactosidase activity was assayed in at least three transformants and expressed as a percentage of ␤-galactosidase activity synthesized by full-length GAL4. higher eukaryotes are also involved in the early step of enzyme assembly by forming either heterodimers or homodimers. Therefore, we examined whether mRPA16 or mRPA40 can self-associate in the yeast two-hybrid system. All pairwise combinations of plasmids expressing GAL4 DB -mRPA40, GAL4 TA -mRPA40, GAL4 DB -mRPA16, or GAL4 TA -mRPA16 were cotransformed into reporter yeast Y153, and the possible interactions were assessed by the quantitative assay of ␤-galactosidase activities. Table II clearly shows that neither selfcombination of mRPA40 nor mRPA16 could induce ␤-galactosidase activity (6th and 10th lines), but combinations between mRPA40 and mRPA16 could induce the ␤-galactosidase (7th and 9th lines), indicating that these two ␣-related subunits of mouse pol I do not self-dimerize, but preferentially form heterodimers.
It is also noted that two ␣-related subunits, mRPB31 and mRPB14, in mouse pol II, also associate but not self-associate. 2 Therefore these results strongly indicate that in eukaryotic nuclear RNA polymerases a heterodimer of the two distinct ␣-related subunits plays an function analogous to that of bacterial ␣-related homodimer. Of course rigorous proof of this evidence must await in vitro reconstitution of the enzyme.
Involvement of the ␣-Motif of mRPA40 in the Interaction between mRPA40 and mRPA16 -Several lines of evidence suggest that the conserved region, the ␣-motif, in ␣-related subunits is involved in homo-or heterodimer formation (4,7,28,29). Therefore, we tested first whether mRPA40 mutants bearing an amino acid substitution in their ␣-motif interact with mRPA16. Alanine 82, one of the conserved amino acids in the ␣-motif of mRPA40 (Fig. 5), was changed to arginine or aspartic acid using a site-directed mutagenesis 4 since the corresponding amino acid in yeast RPC40 (alanine 64) plays an essential role in yeast (4). The growth of the reporter strain Y153 cotransformed with the plasmids expressing GAL4 TA -mRPA16 and either GAL4 DB -mRPA40(A82R) or GAL4 DB -mRPA40(A82D) was tested on SDϪLeuϪTrpϩ3AT plates at 25 or 34°C. Fig. 6 shows that both mutations cause, interestingly, a temperaturesensitive interaction between the mRPA40 mutant and mRPA16 in yeast; the reporter yeasts expressing mutants mRPA40(A82D) or A82R along with mRPA16 did not grow on the plate containing 3AT at 34°C, but they did grow at 25°C, indicating that the alanine 82 is crucial, directly or indirectly, for the interaction. The result suggests that the ␣-motif of mRPA40 functions for the interaction with mRPA16. Encouraged by these results, we are currently mutagenizing mRPA16 randomly to investigate which amino acid residues are indispensable for the interaction.
We have now identified two core subunits in mouse pol I, mRPA16 and mRPA40, both of which are yeast homologs. Thus, it will be interesting to know if remaining small specific subunits of pol I are also homologs of yeast. It also remains to be examined whether AC19 can be replaced with mRPA16 in yeast.