Mrd1p is required for processing of pre-rRNA and for maintenance of steady-state levels of 40 S ribosomal subunits in yeast.

Ribosome biogenesis is a conserved process in eukaryotes that requires a large number of small nucleolar RNAs and trans-acting proteins. The Saccharomyces cerevisiae MRD1 (multiple RNA-binding domain) gene encodes a novel protein that contains five consensus RNA-binding domains. Mrd1p is essential for viability. Mrd1p partially co-localizes with the nucleolar protein Nop1p. Depletion of Mrd1p leads to a selective reduction of 18 S rRNA and 40 S ribosomal subunits. Mrd1p associates with the 35 S precursor rRNA (pre-rRNA) and U3 small nucleolar RNAs and is necessary for the initial processing at the A(0)-A(2) cleavage sites in pre-rRNA. The presence of five RNA-binding domains in Mrd1p suggests that Mrd1p may function to correctly fold pre-rRNA, a requisite for proper cleavage. Sequence comparisons suggest that Mrd1p homologues exist in all eukaryotes.

In eukaryotes, three of the four ribosomal RNAs are derived from a single RNA polymerase I transcript, the precursor rRNA (pre-rRNA). 1 Maturation of the pre-rRNA involves a series of sequential events that occur in the nucleolus (1,2). These include extensive post-transcriptional modifications of many of the nucleotides and several endo-and exonucleolytic cleavages (3,4) (see also Fig. 6A). The final maturation of the 18 S rRNA in the small 40 S subunit takes place in the cytoplasm (5), whereas the maturation of the rRNAs in the large 60 S subunit (5.8 and 25 S or 28 S rRNA from the pre-rRNA and 5 S rRNA separately synthesized by RNA polymerase III) occurs in the nucleus (3). The two ribosomal subunits are believed to be exported separately from the nucleus, and export involves the Ran cycle and specific nucleoporins (6 -8).
Processing of the pre-rRNA takes place in a multicomponent complex that is assembled co-transcriptionally (9,10) and requires the coordinated folding of the pre-rRNA as well as precise positioning of processing factors (3). This process is accompanied by the assembly of ribosomal proteins on the successive intermediates in an ordered manner to generate the two ribosomal subunits. In Saccharomyces cerevisiae, about 60 transacting proteins and 100 small nucleolar RNAs (snoRNAs) are involved in ribosome biogenesis. The snoRNAs are present in the form of snoRNA-protein (snoRNP) complexes in cells and can be divided into the box (C/D) snoRNAs, the box (H/A)CA snoRNAs, and the RNase MRP (11)(12)(13). The U3 snoRNP, one of the most abundant and best characterized snoRNPs, is essential for processing at the three early cleavage sites A 0 , A 1 , and A 2 that leads to the 18 S rRNA formation (9,14,15). In yeast, the U3 snoRNP shares several common protein components, Nop1p, Nop5p/58p, Nop56p, and Snu 13p (16 -20), with other box (C/D) snoRNPs. In addition, a set of proteins that are associated specifically with the U3 snoRNP have been identified, including Sof1p, Mpp10p, Imp3p, Imp4p, Lcp5p, Rcl1p, Rrp9p, and Dhr1p (21)(22)(23)(24)(25)(26)(27). Depletion of any of these proteins or of U3 snoRNA leads to inhibition of the early pre-rRNA cleavages at the A 0 -A 2 cleavage sites (in the case of Dhr1p, only the A 1 and A 2 sites) and reduction of the synthesis of 18 S rRNA.
Among the identified trans-acting nucleolar proteins, some are endo-and exonucleases and RNA helicases, whereas the role of others remains unknown (3). A shared feature of some nucleolar proteins that bind to pre-rRNA or snoRNA is that they contain a consensus RNA-binding domain (RBD), also called the RNA recognition motif. RBDs, which are also present in many proteins involved in different aspects of gene expression (28), have a typical ␤␣␤␤␣␤-fold (29) and mediate specific RNA binding. Two proteins essential for ribosome biogenesis, Nop4p (30) and nucleolin (31), each contain four RBDs. Nucleolin is involved in several steps in ribosome biogenesis (32). Recent NMR spectroscopy studies have shown that the first two N-terminal RBDs in nucleolin bind to an RNA stem loop present at several sites in the pre-rRNA, and it has been suggested that nucleolin may play a role as an RNA chaperone (33).
We have isolated and characterized a novel yeast gene MRD1 (multiple RNA-binding domain). MRD1 encodes a protein with five consensus RNA-binding domains that is essential for viability. Depletion of Mrd1p leads to a decrease in the synthesis of 18 S rRNA and a decrease in the steady-state level of 40 S ribosomal subunits. Mrd1p associates with the 35 S pre-rRNA and the U3 snoRNA. We have found that Mrd1p is required for the initial A 0 -A 2 cleavages during processing of the 35 S pre-rRNA. Mrd1p is the founding member of a conserved family of proteins found in all eukaryotes, each with multiple RNA-binding domains. The high degree of conservation suggests that the multiple RNA-binding domains play an important structural role in organizing specific rRNA processing events.

EXPERIMENTAL PROCEDURES
Strains-S. cerevisiae strains are listed in Table I. Standard yeast methods and media are described within Guthrie and Fink (34). YPGal is similar to YPD medium except that 2% galactose replaces glucose. Yeast cells were transformed using the lithium acetate method (35). Transformants were selected on solid complete synthetic dextrose media lacking uracil, leucine, or histidine as required.
Strain HKDY8, an isogenic diploid derivative of haploid strain AA255 (36), was transformed using the NotI/SalI fragment of pS002 (see Table II; plasmids are described in the following section) containing the mrd1⌬1::hisG-URA3-hisG allele resulting in the Ura ϩ strain LWDY1. The URA3 marker in the mrd1⌬1 allele of LWDY1 was converted to LEU2 resulting in strain LWDY3. This was accomplished by transforming LWDY1 to Leu ϩ with a HindIII fragment from plasmid pUC4-ura3::LEU2 (obtained from Y. Kassir). LWDY1 was propagated on media containing 5-fluoroorotic acid (5-FOA); strain LWDY5, with the unmarked mrd1⌬2 allele, was isolated as a ura Ϫ papilant. Correct insertion of deletion constructs was confirmed by PCR and/or Southern blot analysis. Strains LWDY3 and LWDY5 were transformed to Ura ϩ with pS001 and sporulated, and tetrads were dissected. Strains LWY001 and LWY003 are segregants obtained from these dissections, respectively. LWY001 was transformed with plasmid pS004, and His ϩ transformants were grown on media containing 5-FOA resulting in strain LWY005. Similarly, LWY003 was transformed with pS005, and Leu ϩ transformants were propagated on media containing 5-FOA to obtain strain LWY007.
Plasmids-Plasmids and oligonucleotides are listed in Table II. Pfu polymerase (Stratagene) was used for all PCR amplifications. DNA constructs were propagated in XL-1 Blue Escherichia coli cells (Stratagene). The MRD1 gene (ORF YPR112c) was amplified by PCR using oligonucleotides MRD-p5 and MRD-p3 (Table II) as primers and genomic yeast DNA as template. The resulting fragment, containing 389 bases upstream of the initiating ATG codon and 474 bases downstream of the stop codon, was cloned into NotI and SalI restricted pRS316 (38), resulting in the plasmid pS001. Precise deletion alleles of MRD1 were created in two steps. First, using genomic DNA as template, a 409-bp NotI/BamHI flanked fragment containing sequences immediately 5Ј of the initiating ATG was amplified by PCR using primers mrdD-p5A and mrdD-p5 (Table II). In a parallel reaction, a 476-bp BamHI/SalI flanked fragment containing sequences immedi-ately 3Ј of the termination codon was amplified using primers mrdD-p3A and mrdD-p3B (Table II). After restriction with BamHI, these 5Јand 3Ј-PCR fragments were ligated and cloned into NotI-and SalIdigested pBS(KS)ϩ (Stratagene). In step two, either a 5.1-kb BamHI/ BglII fragment containing the hisG-URA3-kan r -hisG cassette obtained from pSE1076 (39) or a 1.8-kb BamHI fragment containing HIS3 was ligated into the BamHI site between the MRD1 5Ј-and 3Ј-flanking sequences, creating plasmids pS002 and pS003, respectively. An HA epitope-tagged allele of MRD1 in plasmid pS004 was created using a fusion PCR method (40) and primers HA-MRD-A, -B, -C, and -D (Table  II). The construct contains the MRD1 promoter region (up to position Ϫ389) followed by an N-terminal HA tag and the MRD1 ORF. This construct was inserted in NotI-SalI restricted pRS413 (41). Plasmid pS005 contains the MRD1 ORF, amplified by PCR using primer pairs MRD-GFP-5 and MRD-GFP-3 (Table II) using plasmid pS001 as template, inserted into NotI-and PstI-restricted pBFG-1/HA-GFP (42). A PstI-and NotI-flanked fragment encoding N-terminal HA-tagged Mrd1p, amplified by PCR using primer pair Gal-MRD-5 and Gal-MRD-3 (Table II) and pS001 as template, was inserted into PstI-NotIrestricted B2202, creating plasmid pS006.
Immunoelectron Microscopy-Strain LWY005 was grown to late log phase and treated with 50 g/ml Zymolyase 100T (Seikagaku Corporation). Spheroplasts were fixed in 2% paraformaldehyde and 0.2% glutaraldehyde in PBS buffer for 2 h. The fixed cells were dehydrated in 70% ethanol for 1 h and embedded in LR White Resin (London Resin) at 50°C for 3 days. The thin sections were blocked with 3% bovine serum albumin in PBS for 45 min followed by incubation with primary antibodies (anti-HA (12CA5), diluted 1:1, and anti-Nop1p, diluted 1:100) with 1% bovine serum albumin in PBS for 1 h. The sections were washed in PBS and incubated for 30 min with secondary antibody (gold-conjugated mouse or rabbit IgG, 6 or 12 nm, Jackson Immuno-Research) diluted 1:20 in PBS containing 1% bovine serum albumin. Sections were fixed in 4% glutaraldehyde, stained with 5% uranyl acetate, and photographed in a Zeiss electron microscope at 80kV.
In Vivo Depletion of Mrd1p-The effect of diminished MRD1 expression was monitored in strain LWY008 expressing pGAL-HA-MRD1 (pS005) as described (43). Cell growth was monitored by measuring the optical density at 600 nm over a period of 36 h. Preparations of protein and RNA were obtained from cells harvested at the times indicated after the shift to glucose-based medium. For each time point, an equal amount of cells was used.
Analysis of Pre-rRNA Processing-Pulse-chase analysis of rRNA processing was done according to Ref. 44. Briefly, strain LWY008 expressing pGAL-HA-MRD1 (pS006) was grown in SGal to an A 600 of 1. Half of the cells were resuspended in SGal, and the other half were resuspended in SD medium, and the cultures were incubated for 24 h at 30°C. The cells were concentrated and pulse-labeled for 2 min with 250 Ci of [methyl-3 H]methionine, (PerkinElmer Life Sciences) and chased for 1, 3, and 10 min by addition of cold methionine to a final concentration of 5 mM. RNA was extracted using the hot phenol method (45), and aliquots containing 40,000 cpm were electrophoresed through 1.2% agarose-formaldehyde gels, transferred to nylon membranes, sprayed with EN 3 HANCE (PerkinElmer Life Sciences), and exposed to x-ray films.
Analysis of Polysomes and Ribosomal Subunits-Yeast strain LWY008 expressing pGAL-HA-MRD1 (pS006) was grown to early log phase (A 600 of 0.4) in YPGal and used to inoculate glucose-containing YPD media. At the times indicated, cells were harvested and lysed with glass beads, and polysomes were extracted as described (47). Extracts from 50-ml cultures were loaded onto 10 -50% sucrose gradients and centrifuged in an AH-627 rotor (Sorvall) at 4°C for 5 h at 26,500 rpm. Fractions were monitored at 260 nm. For quantification of total ribosomal subunits, cell extracts were prepared according to Ref. 8. Lysates from 25 ml of cultures were loaded onto 7-25% sucrose gradients and centrifuged in an AH-650 rotor at 4°C for 3 h at 49,500 rpm.
Analysis of nuclear extracts was performed essentially as described (10). The extracts were centrifuged through 10 -40% sucrose gradients in an AH-650 rotor for 2 h at 50,000 rpm. The fractions were analyzed by Northern and Western blotting. 35 S pre-rRNA was detected with a probe representing part of the 5Ј-ETS, and HA-Mrd1p was detected with anti-HA antibodies.
Immunoprecipitation and Immunoblot Analysis-Yeast strain LWY007 expressing HA-MRD1-GFP (pS005) was grown to mid-log phase (A 600 of 0.5) in 50 ml of YPD. Cells were washed and resuspended in 0.7 ml of lysis buffer (10 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 150 mM NaCl, 1 mM dithiothreitol, 0.2% Nonidet P-40) containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride and 2.5 g/ml each of aprotinin, leupeptin, pepstatin A) and 5 mM of RNase inhibitor vanadyl ribonucleoside complex (Invitrogen). Cells were lysed by vortexing with an equal volume of glass beads (0.4 -0.5 mm, Sigma) for 5 min. Aliquots of cleared lysates (0.5 ml) were incubated with 100 l of protein A-Sepharose and either 12CA5 or GFP antibody for 2 h at 4°C. Mock incubations to which only protein A-Sepharose was added were carried out in parallel. Eluted proteins were analyzed by Western blotting using anti-Nop1 antibody and ECL reagents (Amersham Biosciences).
To detect pre-rRNA, RNA was resolved on 1.2% agarose-formaldehyde gels. To analyze snoRNAs, RNA was electrophoresed through a 6% denaturing polyacrylamide gel, and individual snoRNAs were detected by hybridization with oligonucleotide probes as indicated.

RESULTS
The MRD1 Gene Is Essential for Cell Viability-We have recently identified a gene in the dipteran Chironomus tentans (CTE314912) called Ct-RBD-1 2 that contains six consensus RNA-binding domains. The presence of six RBDs is unusual, and Ct-RBD-1 appears to be involved in ribosomal synthesis and/or function. Based on sequence similarity with Ct-RBD-1, we have identified a gene in S. cerevisiae, ORF YPR112c on chromosome XVI, which encodes a protein with five RBDs (Fig.  1A). We have named this previously uncharacterized gene MRD1. Mrd1p is comprised of 887 amino acid residues, and in addition to the five RBDs, Mrd1p contains several putative nuclear localization sequences. Computer searches identified Mrd1p homologues, each with a similar domain architecture, in Saccharomyces pombe (O13620), Caenorhabditis elegans (Q9XU67), Drosophila melanogaster (Q9VT19), and human (Q9UFN5). The S. pombe homologue contains five RBDs, whereas the C. elegans, D. melanogaster, C. tentans, and human proteins all have six RBDs (Fig. 1B).
To characterize the function of Mrd1p, we constructed a mrd1 null allele by replacing the ORF in one of the chromosomal copies of MRD1 with URA3 in the diploid strain HKDY8. Subsequent tetrad analysis showed a 2:2 segregation for viability, and no Ura ϩ spore-derived colonies were recovered, indicating that the MRD1 gene is essential. Microscopic analysis revealed that the mrd1⌬ spores germinated, but cell division stopped after three to four generations. To confirm the essential nature of MRD1, the heterozygous LWDY3 strain containing one wild-type MRD1 gene and one disrupted allele marked with LEU2 was transformed with a URA3 centromeric plasmid (pS001) carrying the MRD1 gene under its cognate promoter. This strain was sporulated, and tetrad analysis showed that plasmid-borne MRD1 was able to support wild-type growth in the cells carrying the mrd1 null allele; Leu ϩ Ura ϩ spores were recovered at the expected frequency.
Mrd1p Co-localizes with the Nucleolar Protein Nop1p and Is Also Present in the Nucleoplasm-To determine the intracellular location of Mrd1p, we constructed an allele of MRD1 (pS005) that is under the control of the phosphoglycerate kinase promoter and that encodes a Mrd1p fusion protein with an N-terminal HA and a C-terminal GFP. The HA-Mrd1p-GFP protein complemented mrd1 null alleles and thus is functional. The HA-Mrd1p-GFP fusion protein is predominantly in the nucleus and excluded from both the vacuoles and the cytoplasm ( Fig. 2A). In contrast, in cells with the plasmid pBFG-1/HA-GFP alone, the green fluorescence was spread throughout the cytoplasm (Fig. 2C), consistent with previous reports (42). As compared with DAPI staining (Fig. 2B), it was evident that HA-MRD1p-GFP was present throughout the nucleus but that a more strongly stained crescent pattern was present within the fluorescent nuclei. This pattern suggests that Mrd1p localizes to the nucleolus.
To more precisely determine the subcellular localization of Mrd1p, a functional N-terminal HA-tagged Mrd1p was expressed under its cognate promoter in the CEN vector pRS413 (pS004). The HA-Mrd1p was localized by immunoelectron mi- Mrd1p Is Essential for Pre-rRNA Processing croscopy. The immunolabeling was found throughout the nucleus including the nucleolus (Fig. 2D). Little if any labeling was detected in the cytoplasm. Double labeling confirmed that Mrd1p is present in the nucleolus with an overlapping distribution of the nucleolar marker Nop1p (Fig. 2E). Combined, our data show that Mrd1p is a nuclear protein that is enriched in the nucleolus.
In Vivo Depletion of Mrd1p Impairs Growth and Decreases the Level of Mature 18 S rRNA-To determine the function of the MRD1 gene, we analyzed the effect of depletion of Mrd1p by using a conditional system for phenotypic analysis. The N-terminal HA-tagged MRD1 ORF was placed under the control of a galactose promoter (pS006). On YPGal, the pGal-HA-MRD1 allele complements the mrd1 null mutation in strain LWY008, indicating that the HA-Mrd1p is functional.
In YPGal liquid medium, strain LWY008 showed a doubling time of ϳ2.2 h. Following transfer to YPD medium, in which HA-MRD1 expression is repressed, the growth rate remained similar to that of control cells in YPGal for the first 12 h. Thereafter the doubling time began to increase significantly. After 30 h in YPD, the generation time increased to 6 h (Fig.  3A). Concomitant with the decrease in the growth rate, the amount of HA-Mrd1p was clearly diminished after 6 h of incubation in the YPD medium, became very low after 9 h, and was undetectable after 12 h (Fig. 3B). In contrast, the expression level of Nop1p remained constant during growth in YPD medium (Fig. 3C).
The cessation of growth caused by reduced levels of Mrd1p was reversible. The viability of cells grown in YPD media for 20, 30, or as long as 40 h was examined. Around 95, 90, and 80% of the cells, respectively, began dividing again and formed colonies when plated on YPGal media. No colonies formed when cells were plated on YPD media.
We determined the steady-state levels of rRNAs following Mrd1p depletion and found that the amount of 18 S rRNA was reduced dramatically, whereas the abundance of the 25 S rRNAs was not affected (Fig. 3D). It is unlikely that the decreased steady-state level of 18 S rRNA was due to the effect of cell death because (a) at least more than 90% of cells grown in YPD medium were still viable at the relevant time points based on the cell viability test and (b) the 5.8 S rRNA levels were not affected (data not shown). Further, the levels of Nop1p did not change during the experiment (Fig. 3C). These observations indicate that Mrd1p could have a direct role in 18 S rRNA metabolism.
Mrd1p Depletion Leads to a Deficiency of 40 S Ribosomal Subunits-Since Mrd1p is located in the nucleolus and its depletion leads to a decrease in cell growth and a reduction in the level of 18 S rRNA, we explored the possible function of Mrd1p in ribosome biogenesis. In Fig. 4, it is demonstrated that depletion of Mrd1p leads to drastic changes in the polysome profile. As compared with the profile in cells expressing Mrd1p (Fig. 4A), there is a gradual decrease in polyribosomes and 80 S monosomes with time (Fig. 4, B and C). Simultaneously, there is a decrease in the amount of 40 S ribosomal subunits and a relative increase in 60 S ribosomal subunits. The 40 S ribosomal subunit deficit was further examined by analysis of the total amount of ribosomal subunits (Fig. 4D). The A 260 60 S:40 S ratio for the LWY008 strain increased drastically after 24 h in YPD medium (Fig. 4E). Taken together, these data show that depletion of Mrd1p leads to a decrease of 40 S ribosomal subunits.
18 S rRNA Synthesis Is Impaired upon Mrdp1 Depletion-To determine whether the reduction in the levels of 18 S rRNA and 40 S ribosomal subunits following the depletion of Mrd1p was due to defects in pre-rRNA processing, the pre-rRNA processing pathway was directly followed by using pulse-chase labeling analysis (Fig. 5; see also Fig. 6A for a schematic view of the pre-rRNA processing). In the cells expressing the MRD1 gene, the 35 S pre-rRNA was rapidly processed to yield 27 and 20 S pre-rRNA intermediates as detected after 1 min of chase (Fig.  5, SGal). After 3 min of chase, the 35 S precursor disappeared, and most of the 27 and 20 S intermediates were processed to 25 S rRNA and 18 S rRNA. The processing appeared to be completed within 10 min since all of the labeling was chased into mature 25 and 18 S rRNAs. However, in the Mrd1p-depleted cells, processing is slowed down with an increased amount of the 35 S precursor (Fig. 5, SD). After up to 3 min of chase, a significant amount of 35 S precursor was still present. Processing into 27 S pre-rRNA intermediates and 25 S rRNA was not affected. In contrast, processing into the 20 S pre-rRNA intermediate and 18 S rRNA was severely impaired. Thus, accumulation of 35 S pre-rRNA followed by reduced formation of the 20 S pre-rRNA intermediate leads to a net decrease and delay in 18 S rRNA production in the Mrd1p-depleted cells. This is consistent with our observations from the ribosomal gradient analysis that the amount of 40 S subunits decreased.
Mrd1p Is Required for the Initial Processing of Pre-rRNA at the A 0 -A 2 Sites-To assess the role of Mrd1p in pre-rRNA processing in more detail, we analyzed the steady-state levels of the pre-rRNA processing intermediates upon depletion of Mrd1p. The yeast pre-rRNA processing pathway and the locations of the oligonucleotide probes used to detect 5Ј-ETS, -ITS1, and -ITS2 are shown in Fig. 6A. Equal amounts of total RNA prepared from the LWY008 strain (harvested after growth for 0, 16, 24, and 30 h in YPD medium) were subjected to Northern blot analysis (Fig. 6B). When probe 1, which hybridizes to a sequence upstream of the A 0 site, was used, little 35 S was detected in the control RNA extract (Fig. 6B, probe 1, lane 1; see also probes 2-4, lane 1), consistent with the known rapid processing of this precursor in wild-type cells. In the Mrd1pdepleted cells, the amount of the 35 S pre-rRNA increased. This accumulation was also detected with all probes that could detect the 35 S pre-rRNA precursor (Fig. 6B, probes 2-4, lanes  2-4), showing the inhibition of cleavage at site A 0 . Meanwhile, we also detected an increase of the 23 S aberrant pre-rRNA intermediate (Fig. 6B, probe 1, lanes 2-4), which is the product of cleavage of the 35 S pre-rRNA at site A 3 in the absence of prior cleavage at sites A 0 -A 2 (48). The 23 S aberrant pre-rRNA intermediate could also be seen when hybridizing with probes 2 and 3 at the 5Ј-side of A 2 and A 3 cleavage sites, respectively (Fig. 6B, probes 2 and 3, lanes 2-4). We also detected a decrease in the amount of 20 S (Fig. 6B, probe 2), which is consistent with the inhibition of cleavage at A 2 . Hybridization with probe 4, specific for the precursor containing 5.8 and 25 S rRNAs, showed that the cleavages at the A 3 and downstream sites were not affected upon Mrd1p depletion (Fig. 6B, probe 4).
Based on these results, we conclude that depletion of Mrd1p leads to increased levels of 35 S pre-rRNA and the 23 S aberrant pre-rRNA intermediate. These effects are concordant with impaired pre-rRNA processing at sites A 0 , A 1 , and A 2 . Consistently, we observed decreased levels of the 20 S pre-rRNA intermediate and 18 S rRNA. Although our results are indicative of severely impaired processing, some 20 S pre-rRNA intermediate and 18 S rRNA were still produced, and 35 S and the aberrant 23 S pre-rRNA intermediate did not accumulate to very high levels.
Mrd1p Is Associated with the 35 S Pre-rRNA and U3 snoRNA-Our results show that Mrd1p is involved in the ini-tial cleavages of the 35 S pre-rRNA. If so, Mrd1p is most likely physically associated with the pre-rRNA precursor. In Fig. 7A, it is shown that 35 S pre-rRNA and Mrd1p are both present in 80 -90 S particles in the nucleus. As a further test, we immunoprecipitated cell extracts from cells expressing the HA-Mrd1p-GFP fusion protein with anti-GFP or anti-HA antibodies. In Fig. 7B, it is shown that 35 S pre-rRNA could be detected in the immunoprecipitated Mrd1p complex. These results suggest that Mrd1p interacts with the 35 S pre-rRNA and are consistent with the fact that Mrd1p is necessary for efficient cleavage at the A 0 -A 2 cleavage sites.
Impaired pre-rRNA processing at the A 0 -A 2 processing sites is also observed after depletion of the snoRNAs U3, U14, and snR10 and several nucleolar proteins. We therefore used coimmunoprecipitation to assay whether Mrd1p is associated with some of these factors. Extracts from cells expressing the HA-Mrd1-GFP fusion protein were immunoprecipitated with anti-GFP antibodies. The immunoprecipitated RNA was separated on 6% polyacrylamide gels and probed with oligonucleotides specific to several box (C/D) (U3, U14, U18, snR190) snoRNAs and the box (H/A)CA (snR11) snoRNAs. Only U3 snoRNA specifically co-immunoprecipitated with the Mrd1 fusion protein (Fig. 7C). The same result was obtained when anti-HA antibody was used for immunoprecipitation (data not shown). We also showed that the U3 snoRNP protein Nop1 could be co-immunoprecipitated with Mrd1p (Fig. 7D).
Finally, we tested the stability of the U3, U14, U18, snR11, and snR190 snoRNAs following Mrd1p depletion (Fig. 8). The hybridization signals from Mrd1p-depleted cells were not different from those seen from cells grown in galactose medium. This suggests that the effect of Mrd1p depletion is not due to a decrease in the amount of snoRNAs.

Mrd1p Together with Several Other Components Contribute
to the Initial Processing of Pre-rRNA-The MRD1 gene is essential for cell growth. We demonstrated that Mrd1p is required for maintaining steady-state levels of the 40 S ribosomal subunit. Analysis of the effect of Mrd1p depletion showed that the 35 S pre-rRNA and the aberrant 23 S pre-rRNA intermediate increased and that the levels of the 20 S pre-rRNA intermediate and 18 S rRNAs were reduced. This is characteristic of a specific inhibition of processing at the A 0 -A 2 sites. The aberrant 23 S pre-rRNA intermediate cannot be further processed to 18 S rRNA and is known to be rapidly degraded by the  Total RNA was extracted from LWY008 cells grown in YPD for the indicated times and separated on 6% polyacrylamide gels. After electroblotting to a nylon filter, the RNA was hybridized with oligonucleotide probes specific for the indicated snoRNAs. exosome (49). The inhibition of cleavage at the A 0 -A 2 sites was significant, but some 18 S rRNA could still be made. The fact that we did not detect any changes in the steady-state levels of the U3, U14, U18, snR11, and snR190 snoRNAs showed that the effect of Mrd1p depletion is not due to a defect in the synthesis or stability of these snoRNAs. Rather, our results showing that Mrd1p is associated with the 35 S pre-rRNA in the 80 -90 S pre-rRNP particle suggest a more direct role for Mrd1p in pre-rRNA processing.
Mrd1p is co-immunoprecipitated with U3 snoRNA (and the U3 snoRNP-associated Nop1p). It is established that U3 snoRNA base-pairs to the 35 S pre-rRNA at the 5Ј-ETS and also at the 5Ј-stem loop within the 18 S rRNA. These interactions are needed for the early A 0 -A 2 cleavages and A 1 and A 2 cleavages, respectively (11). Our immunoprecipitation results therefore are in agreement with the interpretation that Mrd1p and the U3 snoRNA are both bound to the 35 S pre-rRNA precursor. Although it is likely that these interactions occur in a region of the 35 S rRNA precursor containing the 5Ј-ETS and 18 S rRNA, our data do not rule out other alternatives.
Several components have been shown previously to be necessary for processing at the A 0 -A 2 sites. These include the U3, U14, snR30, and snR10 snoRNAs and the snoRNP proteins, Nop1p, Sof1p, and Gar1p (16, 21, 50 -54), the putative helicases Fal1p (44), Rrp3p (55), and Rok1p (56), the dimethylase Dim1p (57), and the nucleolin-like Nsr1p (58). Mutations in different components have shown that processing at the three sites is interconnected but not necessarily coupled (57,59). Many different components are involved in all three processing events, indicating that the A 0 , A 1 , and A 2 cleavages take place in a multi-snoRNP complex (60). Processing at the A 0 -A 2 sites is also coupled to cleavage at the A 3 site, suggesting that contacts between processing factors can bridge longer distances in a large pre-rRNA-protein complex (61,62). The structure of Mrd1p and its association with the 35 S pre-rRNA and U3 snoRNA, combined with its involvement in the processing of three separate cleavage sites, are consistent with a model in which Mrd1p binds to the pre-rRNA and/or snoRNAs in a complex that is required for the processing reactions.
Mrd1p May Play a Role in Organizing the Pre-rRNA in the Multicomponent RNA-protein Processing Complex-A notable feature of Mrd1p is the presence of five RNA-binding domains. One or up to four RBDs are present in a large number of different proteins (29). The structure of several individual RBDs (63, 64) as well as two consecutive RBDs, complexed with RNA, have been determined (33,65,66). Each RBD has a ␤␣␤␤␣␤-fold. Individual RBDs interact with specific sequences of up to 12 nucleotides, present in stem loops or in linear RNA. Two consecutive RBDs in nucleolin (33) bind on opposite sides of an RNA loop, thus stabilizing the stem loop, whereas in sex-lethal (66) and poly(A)-binding protein (65), two consecutive RBDs bind and stretch the RNA.
It is most likely that the five RBDs in Mrd1p are all folded into the compact RBD ␤␣␤␤␣␤-structure and that they contribute to binding to specific sequences in the 35 S pre-rRNA or in snoRNAs. In Mrd1p, the distances between the five RBDs are unusually long, 285, 127, 76, and 42 amino acid residues (Fig.  1A). These regions could facilitate specific interactions with other proteins within the pre-rRNA-protein complex.
Why is Mrd1p necessary for the initial pre-rRNA processing steps? Cleavages at the A 0 -A 2 sites require many different components in the pre-rRNP processing complex. The conformation of the pre-rRNA and the snoRNAs may be important to accomplish a cleavage-competent substrate. It has recently been proposed that nucleolin via its RBDs binds a GC-rich sequence present several times in pre-rRNA, to induce or sta-bilize stem-loop structures, and serves as an RNA chaperone (33). It has also been proposed that a cleavage-competent pre-rRNP structure requires that several RNA-binding proteins interact with the different spacer sequences in the 35 S pre-rRNA (67). Proteins have been identified that assist in the formation of specific rRNA tertiary structures; thus, these proteins inhibit other possible alternative base pairings that can kinetically trap rRNA molecules in inactive conformations (68). Based on the current understanding of Mrd1p, we suggest that Mrd1p binding to pre-rRNA facilitates the proper folding of rRNA for cleavage at the A 0 -A 2 processing sites. Mrd1p has a similar sequence organization to proteins in several eukaryotic organisms (see "Results"), and it has a similar nuclear localization to proteins in Diptera and human. 2 This suggests that Mrd1p is evolutionarily conserved in eukaryotes and that the functional properties of Mrd1p revealed here are of general importance.