|Home | About | Journals | Submit | Contact Us | Français|
Small nucleolar ribonucleoprotein particles (snoRNPs) mainly catalyze the modification of rRNA. The two major classes of snoRNPs, box H/ACA and box C/D, function in the pseudouridylation and 2′-O-methylation, respectively, of specific nucleotides. The emerging view based on studies in yeast is that each class of snoRNPs is composed of a unique set of proteins. Here we present a characterization of mammalian snoRNPs. We show that the previously characterized NAP57 is specific for box H/ACA snoRNPs, whereas the newly identified NAP65, the rat homologue of yeast Nop5/58p, is a component of the box C/D class. Using coimmunoprecipitation experiments, we show that the nucleolar and coiled-body protein Nopp140 interacts with both classes of snoRNPs. This interaction is corroborated in vivo by the exclusive depletion of snoRNP proteins from nucleoli in cells transfected with a dominant negative Nopp140 construct. Interestingly, RNA polymerase I transcription is arrested in nucleoli depleted of snoRNPs, raising the possibility of a feedback mechanism between rRNA modification and transcription. Moreover, the Nopp140-snoRNP interaction appears to be conserved in yeast, because depletion of Srp40p, the yeast Nopp140 homologue, in a conditional lethal strain induces the loss of box H/ACA small nucleolar RNAs. We propose that Nopp140 functions as a chaperone of snoRNPs in yeast and vertebrate cells.
rRNA is heavily modified on specific nucleotides by 2′-O-methylation and pseudouridylation. These modifications occur cotranscriptionally in the nucleolus and number ~100 in vertebrates and 50 in yeast (Maden, 1990 ). The site specificity of these modifications is determined by a similar number of small nucleolar RNAs (snoRNAs). Two major classes of snoRNAs can be discerned based on short conserved sequence elements, box H/ACA and box C/D. The former guide the pseudouridylation of rRNA, whereas the latter determine the sites of 2′-O-methylation. The site specificity is achieved by base pairing of short stretches of the snoRNAs to complementary sequences of the rRNA flanking the nucleotides to be modified (for review, see Maxwell and Fournier, 1995 ; Smith and Steitz, 1997 ; Tollervey and Kiss, 1997 ). Although the function of these modifications is unknown, they cluster around the peptidyl transferase center of the ribosome, implying a role during peptide synthesis (Bakin et al., 1994 ).
Although the snoRNA-mediated mechanism of site selection has been worked out elegantly, little is known about the proteins associated with the snoRNAs or the enzymes catalyzing the modifications. However, recent studies in yeast indicate that each of the major small nucleolar ribonucleoprotein particles (snoRNPs) is endowed with its distinct set of proteins, box H/ACA snoRNPs with Cbf5p, Gar1p, Nhp2p, and Nop10p and box C/D snoRNPs with Nop1p, Nop5/58p, and Sik1/Nop56p (see Figure Figure6)6) (Lübben et al., 1995 ; Balakin et al., 1996 ; Ganot et al., 1997 ; Gautier et al., 1997 ; Henras et al., 1998 ; Lafontaine et al., 1998 ; Watkins et al., 1998a ; Lafontaine and Tollervey, 1999 ). The best-characterized protein is Cbf5p, a candidate pseudouridylase of rRNA (Jiang et al., 1993 ; Lafontaine et al., 1998 ). Its mammalian homologue is rat NAP57 (Meier and Blobel, 1994 ), which was identified in humans as the gene mutated in the X-linked bone marrow failure disorder dyskeratosis congenita (Heiss et al., 1998 ). Compared with yeast snoRNPs, even less is known about the composition of mammalian snoRNPs. In fact, aside from fibrillarin, the mammalian Nop1p orthologue, and a few bands on a gel, the composition of mammalian snoRNPs remains uncharacterized to date (Parker and Steitz, 1987 ; Tyc and Steitz, 1989 ; Caffarelli et al., 1998 ; Watkins et al., 1998b ).
We previously identified rat Nopp140, a highly phosphorylated nucleolar and coiled-body protein that shuttles between the nucleolus and the cytoplasm on intranuclear tracks (Meier and Blobel, 1990 , 1992 , 1994 ). Using Nopp140 antibodies, we subsequently identified the Nopp140-associated protein NAP57, the putative pseudouridylase of rRNA (Meier and Blobel, 1994 ; see also Koonin, 1996 ). Because mammalian rRNA pseudouridylase activity is dependent on snoRNAs, Nopp140 may in fact be associated with entire snoRNPs. This idea was supported by subsequent studies that suggested a role for Nopp140 in snoRNP transport between the nucleolus and the coiled bodies (Isaac et al., 1998 ).
To further investigate the role of Nopp140 in snoRNP function, we reexamined here in more detail the macromolecules that coprecipitate with Nopp140. We find that Nopp140 interacts with two discrete complexes corresponding to the mammalian box H/ACA and box C/D snoRNPs. Thus, GAR1 and box H/ACA snoRNAs specifically coprecipitate with NAP57. In addition to these box H/ACA snoRNP components, Nopp140 precipitates the newly identified NAP65 (the rat homologue of yeast Nop5/58), fibrillarin, and box C/D snoRNAs, all parts of box C/D snoRNPs. An in vivo interaction of Nopp140 with both major mammalian snoRNPs and its role in their intranuclear transport is indicated by the specific and exclusive depletion of snoRNP proteins from nucleoli of cells transfected with a dominant negative Nopp140 construct. The Nopp140-snoRNP interaction appears to be conserved in yeast, because genetic depletion of Srp40p, the yeast Nopp140 homologue (Meier, 1996 ), in a conditional lethal strain specifically reduces the level of box H/ACA snoRNAs. Based on these observations, we propose that Nopp140 functions as a chaperone of snoRNPs in yeast and vertebrate cells.
Precipitations were performed from rat liver nuclei (Meier and Blobel, 1994 ) and from whole cell extracts of buffalo rat liver cells (Li et al., 1997 ) for protein and RNA analysis, respectively. The antibodies used for the experiments were affinity-purified peptide antibodies directed against Nopp140 (Meier and Blobel, 1992 ) and NAP57 (Meier and Blobel, 1994 ). For the more stringent incubation conditions (see Figure Figure1A),1A), the 25 mM Tris, pH 8.1, extracts were adjusted to 150 mM sodium chloride, 1% Triton X-100, and 0.2% SDS; under less stringent conditions (see Figure Figure2,2, A and B), the addition of sodium chloride and detergent was omitted. However, both immunoprecipitates were repeatedly washed with the same stringency, i.e., 150 mM sodium chloride, 0.1% Triton X-100, and 0.02% SDS. The precipitates were analyzed by SDS-PAGE and silver staining (Merril et al., 1984 ) or transfer to nitrocellulose and subsequent amido black staining and immunodetection with ECL (Amersham Life Science, Arlington Heights, IL). The primary antibodies (and concentrations or dilutions) used for consecutive immunodetection were as follows: GAR1, Rab2B rabbit serum raised against human GAR1 (1:1000; Dragon, Pogaić, and Filipowicz, unpublished data); fibrillarin, D77 mouse monoclonal immunoglobulin G (IgG) (0.2 μg/ml; Aris and Blobel, 1988 ); CK2, polyclonal rabbit serum against the α subunit of casein kinase 2 (1:5000; Litchfield et al., 1994 ); NAP57, RL12 affinity-purified rabbit IgG (0.4 μg/ml; Meier and Blobel, 1994 ); and Nopp140, RF11 affinity-purified rabbit IgG (0.1 μg/ml; Meier and Blobel, 1992 ).
Amino-terminal peptide sequences of NAP65 and fibrillarin were obtained exactly as described for NAP57 (Meier and Blobel, 1994 ). The full-length cDNA of NAP65 was derived from the overlapping expressed sequence tags (ESTs) 108064, 111317, 105505, and 105838 and was constructed from EST105838 and EST108064, which were obtained from the American Type Culture Collection (Manassas, VA) on plasmids designated pTM131 and pTM135, respectively. The two ESTs were ligated with the use of PCR-based splicing by the overlap extension method (Vallejo et al., 1995 ) on an 870-nucleotide SphI/XhoI fragment, which was subsequently cloned into those sites of pTM131 to generate pTM136 (full-length NAP65 in pBluescript SK− [Stratagene, La Jolla, CA]). DNA sequencing was used to verify all sequences and discovered a base change in the EST105838 sequence at nucleotide 1356 of the full-length cDNA from A to G, resulting in the amino acid switch of lysine 396 to arginine. Indeed, arginine 396 is conserved in all metazoan homologues analyzed (see Figure Figure1C).1C). The sequence for full-length NAP65 was deposited in GenBank (accession number AF194371). NAP65 was transcribed/translated from pTM136 as described (Isaac et al., 1998 ). The NAP57 plasmid (pTM575) was generated previously (Meier and Blobel, 1994 ). Multiple protein sequence alignments were produced with the use of the CLUSTAL alignment method and were presented with the use of the BOXSHADE program through the Baylor College of Medicine Search Launcher (Smith et al., 1996 ).
Total RNAs were extracted from whole cell extracts and immunoprecipitates by phenol/chloroform and digested with RNase-free DNase I (Sigma Chemical, St. Louis, MO) to remove any contaminating traces of genomic DNA. Reverse transcription (RT)-PCR was performed with the use of the DNase I–treated RNA as template in the SuperScript One-Step RT-PCR system (BRL-Life Technologies, Grand Island, NY), as described by the manufacturer. The amplified DNAs were analyzed by 4% agarose (NuSieve, FMC BioProducts, Rockland, ME) gel electrophoresis, and ethidium bromide staining. Control amplifications with the use of Taq polymerase (Perkin Elmer-Cetus, Norwalk, CT) alone confirmed the absence of any genomic snoRNA sequences. The following snoRNA-specific primer pairs were used in the RT-PCR experiments: 5′-ACTCTCCCCGGGCTCTGT-3′ and 5′-TAGGAATATGCAGGCGCAGA-3′ for U17/E1; 5′-GAGAATTCTAAGCAGGATTTTACTACAATAT-3′ and 5′-CTCAGTGAGCTCATGTATGAGACCAAGCGT-3′ for E3; 5′-NNNNNNGAATTCCAAAACCATTCGTAG-3′ and 5′-NNNN-NNGAGCTCATCCAAGGAAGGAACTAGCCAAC-3′ for U14; and 5′-CCAGAGCCTGAAAAGGTGAA-3′ and 5′-CTCAGACAGTTCCTTCTGGA-3′ for U22.
Yeast cell growth (Ausubel et al., 1993 ; Meier, 1996 ; Lafontaine et al., 1998 ) and DNA manipulation (Maniatis et al., 1989 ; Meier, 1996 ) were performed according to standard procedures and as described previously. The plasmids used for the complementation studies of yeast Cbf5p by rat NAP57 were based on pACT2 (Clontech Laboratories, Palo Alto, CA), a yeast vector expressing the protein of interest as a carboxyl-terminal fusion of the Gal4 activation domain (GAD) under the ADH promoter. pTM113 contained full-length NAP57 amplified and cloned into the NcoI and EcoRI sites of pACT2. The haploid yeast strains YDL401 (CBF5) and YDL521-1 (GAL::cbf5; Lafontaine et al., 1998 ) and the diploid strains YCC130 (CBF5/cbf5::TRP1) and YWJ64-ts (cbf5-1/cbf5-1; Cadwell et al., 1997 ) were obtained from the indicated sources. From these strains, we generated the following transformants carrying the indicated plasmids (in parentheses): YYY64, YWJ64-ts (pTM113); YYY66, YWJ64-ts (pACT2); YYY68, YCC130 (pTM113); YYY69, YCC130 (pACT2); YYY138, YDL521-1 (pTM113); YYY139, YDL521-1 (pACT2); and YYY143, YDL401 (pACT2).
The synthetic lethal screen that generated YYY206 (Matα TRP1 lys2 ade2 ade3 ura3 can1 Δsrp40::HIS3 les2::LEU2), the srp40Δ les2 mutant carrying pGAL-SRP40 (pYY38), will be described elsewhere. pYY38 was constructed by cloning the SalI fragment containing SRP40 under the GAL10 promoter from pTM41 (Meier, 1996 ) into pRS317, which carries a LYS2 marker (Sikorski and Boeke, 1991 ). To allow growth of the wild-type strain (wt, YCH128) and the singly disrupted strains srp40Δ (YYY7) and les2 (YYY216) in lysine-free medium, they were transformed with pRS317 to generate YYY231, YYY232, and YYY236, respectively. Growth in lysine-free medium was required for the maintenance of pYY38 (pGAL-SRP40) in YYY206 (srp40Δ les2). The genetic backgrounds of the strains were as follows: wt, YCH128 (Matα TRP1 lys2 ade2 ade3 ura3 leu2 his3 can1; a kind gift from Susan Wente and Chris Hardy [Washington University School of Medicine, St. Louis, MO]); srp40Δ, YYY7 (Matα TRP1 lys2 ade2 ade3 ura3 leu2 can1 Δsrp40::HIS3); and les2, YYY216 (Matα TRP1 lys2 ade2 ade3 ura3 his3 can1 les2::LEU2).
For the Cbf5p and Srp40p depletion experiments, we followed essentially the protocol described for Cbf5p (Lafontaine et al., 1998 ). Total RNA was prepared (Schmitt et al., 1990 ), and 9 μg was loaded in each lane for Northern blot analysis. The snoRNAs U3, U14, snR3, snR11, and snR33 were detected by hybridization with the following 32P-labeled oligonucleotides: 5′-GGATTGCGGACCAAGCTAA-3′, 5′-CGAATGTTAAGGAACCAG-3′, 5′-TCGATCTTCGTACTGTCT-3′, 5′-GACGAATCGTGACTCTG-3′, and 5′-GATTGTCCAC-ACACTTCT-3′, respectively. To detect the CBF5 mRNA, CBF5 was amplified from genomic yeast DNA (Meier, 1996 ) and random prime labeled as described (Meier and Blobel, 1992 ). Strains YYY68 and YYY69 were used for tetrad analysis (Meier, 1996 ).
COS-1 cells were transiently transfected with the HA-tagged dominant negative Nopp140 carboxyl-terminal construct HA-NoppC (pWG13) and processed for indirect double immunofluorescence exactly as described (Isaac et al., 1998 ). The following primary antibodies were used at the dilutions given in parentheses: anti-recombinant Nopp140 serum (RH10 at 1:1000; Meier, 1996 ); anti-GAR1 serum (Rab2B at 1:50; Dragon, Pogaić, and Filipowicz, unpublished data); anti-RNA polymerase I serum (αPolIβ at 1:50 from Larry Rothblum [Geisinger Clinic/Weis Center for Research, Danville, PA]); and anti-HA ascites fluid (12CA5 at 1:200; Wilson et al., 1984 ). Secondary antibodies were rhodamine-labeled goat anti-rabbit IgG and fluorescein-labeled goat anti-mouse IgG antibodies, both from Boehringer Mannheim (Indianapolis, IN).
The RNA polymerase I (RNA pol I) run-on assays were performed as described previously (Moore and Ringertz, 1973 ) with the modifications described by Savino et al. (1999) . The cells were postfixed with 2% paraformaldehyde for 15 min, and 5-bromo-uridine-triphosphate (BrUTP) (Sigma Chemical) incorporation was detected with a mouse anti-5-bromo-2′-deoxyuridine (BrdU) mAb, F(ab′)2 fragments conjugated with FLUOS (Boehringer Mannheim) at a dilution of 1:5.
We previously used coimmunoprecipitation with Nopp140 to identify NAP57, a putative component of box H/ACA snoRNPs and pseudouridylase of rRNA (Meier and Blobel, 1994 ; Nurse et al., 1995 ; Lafontaine et al., 1998 ). To test if additional proteins coprecipitated with Nopp140, we reexamined the original Nopp140 precipitates with a more sensitive method, silver staining. Thus, after precipitation of Nopp140 with anti-peptide antibodies from rat liver nuclear extracts, the associated proteins were visualized by silver staining after separation by SDS-PAGE (Figure (Figure1A).1A). Careful comparison of the proteins precipitated in the absence (Figure (Figure1A,1A, lane 2) and presence (lane 3) of free competing peptide revealed the proteins specifically associated with Nopp140. In this way, eight minor bands of 110, 66, 65, 61, 50, 42, 38, and 24 kDa were detected in the Nopp140 precipitates in addition to the stoichiometric amounts of NAP57 (Figure (Figure1A,1A, lane 2, dots). Here we report on the identities of three of these proteins, one novel and two known.
The 42-kDa protein band was previously shown to correspond to the α subunit of casein kinase 2 by antibody reactivity (see Figure Figure2B)2B) (Li et al., 1997 ). The 38- and 65-kDa proteins were identified by amino-terminal peptide sequencing after precipitation of increased amounts of proteins and transfer to an Immobilon membrane, as described for NAP57 (Meier and Blobel, 1994 ). GenBank searches showed the 28 amino-terminal residues of the 38-kDa protein to be identical to those of rat fibrillarin, an integral component of box C/D snoRNPs (Figure (Figure1A)1A) (Tyc and Steitz, 1989 ; Balakin et al., 1996 ; Ganot et al., 1997 ). The positions of the unidentified residues in the fibrillarin peptide sequence (Figure (Figure1A,1A, X) matched exactly those of the arginines modified by NG,NG-dimethyl groups (Lischwe et al., 1985 ). In addition, the serine in position 6 was identified as a dehydroserine, indicating that it was phosphorylated (Figure (Figure1A,1A, circled P). Interestingly, fibrillarin purified from Novikoff hepatoma nucleoli was not modified at that position (Lischwe et al., 1985 ). Therefore, only a fraction of the total cellular fibrillarin appears to be phosphorylated at that residue, perhaps only the fraction associated with Nopp140.
Comparison of the amino-terminal peptide sequence of the 65-kDa protein band with the GenBank sequences revealed it to be an uncharacterized EST (EST105839; Lee et al., 1995 ). The full-length sequence was derived from overlapping ESTs and constructed by linking two ESTs together (EST105839 and EST108064). It encoded a protein of 534 amino acids with a calculated molecular mass of 60,043 Da and a theoretical isoelectric point of 8.7. Its high charge density (34% of all amino acids were charged) likely accounted for its slightly reduced mobility on SDS-PAGE. According to its mobility and association with Nopp140, we termed the protein NAP65. To ascertain that the cDNA constructed from the two ESTs was indeed full length, it was in vitro transcribed/translated in the presence of [35S]methionine and [35S]cysteine and analyzed by SDS-PAGE (Figure (Figure1B,1B, lane 2). Indeed, the resulting protein migrated as a single band of 65 kDa compared with NAP57 (lane 3), and no product was observed in the absence of exogenous DNA (lane 1).
GenBank searches further revealed NAP65 to be a highly conserved protein exhibiting sequence identities across its entire sequence of 98% (mouse), 95% (human), 58% (Drosophila melanogaster), 52% (Caenorhabditis elegans and Pisum sativum), 51% (Oryza sativa), and 45% (Saccharomyces cerevisiae). The alignment of these sequences is depicted in Figure Figure1C.1C. NAP65, like NAP57, even displayed homology to ORFs of archaea, e.g., 28% identity to the entire gene MJ0694 of Methanococcus jannaschii. In contrast to NAP57, however, no NAP65-related sequences were identified in eubacterial genomes. In addition, homologous ESTs were present in most other organisms whose genomes are being sequenced, but they did not amount to full-length sequences and therefore have been omitted from the alignment. Of all these putative NAP65 homologues, only the yeast Nop5/58p has been characterized to date (Gautier et al., 1997 ; Wu et al., 1998 ). Nop5/58p is an integral part of box C/D snoRNPs (Lafontaine and Tollervey, 1999 ). In summary, therefore, Nopp140 associates with NAP57, CK2, and two integral components of box C/D snoRNPs, fibrillarin and NAP65.
Because in our previous Nopp140 precipitations only minute amounts of proteins aside from NAP57 were detected (Figure (Figure1A)1A) (Meier and Blobel, 1994 ), we repeated the precipitations under slightly less stringent conditions (see MATERIALS AND METHODS). In addition, we performed the precipitations with peptide antibodies directed against NAP57 to test its association with snoRNPs. The precipitates were separated by SDS-PAGE, transferred to nitrocellulose, and stained with amido black (Figure (Figure2A).2A). The membrane was incubated consecutively with antibodies to Nopp140, NAP57, CK2, fibrillarin, and GAR1 (an integral component of box H/ACA snoRNPs) (Balakin et al., 1996 ; Dragon, Pogaić, and Filipowicz, unpublished data), and immunoreactivity was detected by secondary antibodies and ECL. The combined results of all of the antibody reactions generated by overlaying the individual films are depicted in Figure Figure2B.2B. Under the less stringent conditions, stoichiometric amounts of not only NAP57, the 57-kDa protein band, but also of fibrillarin and GAR1, the 38- and 27-kDa protein bands, precipitated with Nopp140 (Figure (Figure2,2, A and B, compare lanes 2). Amido black, however, stained only smaller amounts of NAP65, the 65-kDa band, in the Nopp140 precipitates (Figure (Figure2A,2A, dot). In contrast, precipitations with peptide antibodies against NAP57 yielded only stoichiometric amounts of GAR1 compared with NAP57 and smaller quantities of Nopp140 (Figure (Figure2,2, A and B, compare lanes 4). The presence of the α subunit of casein kinase 2 in the Nopp140 precipitates was detectable only by immunoreactivity (Figure (Figure2B,2B, lane 2) (Li et al., 1997 ). The specificity of the peptide antibody precipitations was established by competition with free peptide (Figure (Figure2,2, A and B, lanes 3 and 5). Together, these data indicate that Nopp140 coprecipitated protein components of both box C/D (fibrillarin and NAP65) and box H/ACA (NAP57 and GAR1) snoRNPs. NAP57, however, exclusively precipitated GAR1, a specific component of box H/ACA snoRNPs.
To test if Nopp140 and NAP57 associated only with the snoRNP proteins or also with their snoRNAs, the two proteins were precipitated from whole cell extracts and the coprecipitating RNAs were analyzed by RT-PCR with snoRNA-specific primers. The amplified cDNAs were visualized with ethidium bromide after agarose gel electrophoresis (Figure (Figure2C).2C). The presence of two representative box H/ACA snoRNAs, U17/E1 and E3, and two box C/D snoRNAs, U14 and U22, for which the rat or rodent sequences were available, was tested (Liu and Maxwell, 1990 ; Tycowski et al., 1996 ; Selvamurugan et al., 1997 ). All primers amplified DNA bands of the expected size from whole cell extracts (Figure (Figure2C,2C, lane 1). NAP57 exclusively coprecipitated the two box H/ACA snoRNAs, whereas Nopp140 coprecipitated the box H/ACA snoRNAs and lesser but distinct amounts of the box C/D snoRNAs (Figure (Figure2C,2C, lanes 2 and 4, respectively). No snoRNAs were detected if precipitation of NAP57 or Nopp140 was competed for with free peptide (Figure (Figure1C,1C, lanes 3 and 5, respectively). We conclude that Nopp140 and NAP57 associate with the protein and RNA components of snoRNPs, indicating that they associate with the intact particles.
The high sequence conservation between rat NAP57 and yeast Cbf5p (Meier and Blobel, 1994 ), together with their specific association with box H/ACA snoRNPs, prompted us to test if NAP57 could functionally complement the essential Cbf5p. For this purpose, we obtained a yeast strain whose genomic copy of CBF5 was placed under the control of the conditional GAL10 promoter (GAL::cbf5; Lafontaine et al., 1998 ). Genetic depletion of Cbf5p in this strain by growth on glucose-containing medium leads to significantly reduced levels of pseudouridines in rRNA, instability of box H/ACA snoRNAs, and growth arrest (Lafontaine et al., 1998 ). First, we tested for the ability of rat NAP57 to complement the growth-arrest phenotype of Cbf5p depletion in this strain (Figure (Figure3A).3A). Expression of a GAD-NAP57 fusion construct, in contrast to GAD alone, fully restored the growth of the Cbf5p-depleted strain (Figure (Figure3A,3A, compare lanes 2 and 3). Expression of two unrelated nuclear proteins fused to GAD, however, failed to restore growth (data not shown). The fusion of GAD to NAP57 was necessary to stabilize NAP57, because it was detected on immunoblots only when expressed as a fusion protein with GAD. In conclusion, NAP57 complemented the growth-arrest phenotype of a Cbf5p-depleted strain.
In addition to growth arrest, Cbf5p depletion caused a specific loss of box H/ACA but not box C/D snoRNAs (Figure (Figure3B,3B, lane 4) (Lafontaine et al., 1998 ). To test whether the rat NAP57 was able to stabilize the yeast box H/ACA snoRNAs, total RNAs were extracted from wild-type CBF5 and mutant GAL::cbf5 strains after growth in glucose-containing medium for 0 and 24 h and were analyzed for the presence of the snoRNAs by Northern blotting (Figure (Figure3B).3B). Indeed, GAD-NAP57 stabilized the box H/ACA snoRNAs (Figure (Figure3B,3B, lane 6), whereas GAD alone had no effect (lane 4). Surprisingly, the amount of box C/D snoRNAs appeared increased under the conditions in which the box H/ACA snoRNAs were lost (Figure (Figure3B,3B, lane 4). This was most likely caused by a relative overload of non-rRNAs when applying equal amounts of total RNAs, because Cbf5p depletion led to a decrease in rRNAs compared with non-rRNAs (Lafontaine et al., 1998 ). In agreement with this interpretation, expression of GAD-NAP57 restored the levels of box C/D snoRNAs (Figure (Figure3B,3B, compare lanes 4 and 6). This indicated that NAP57 functionally replaced Cbf5p in the box H/ACA snoRNPs and consequently restored normal rRNA synthesis in the Cbf5p-depleted strain. We did not test if the pseudouridylation of rRNA was restored. However, to ascertain that the complementation was indeed caused by GAD-NAP57 and not by Cbf5p through spurious release from glucose repression, we confirmed by Northern blotting that Cbf5p mRNA remained depleted in glucose-containing medium (data not shown). Therefore, rat NAP57 was able to functionally complement a yeast strain genetically depleted of Cbf5p.
Further experiments in a different background, however, revealed that the rat NAP57 only partially complemented Cbf5p function. Thus, we tested for the NAP57 complementation of a CBF5 temperature-sensitive strain (cbf5ts) and a CBF5 null strain (Cadwell et al., 1997 ). Surprisingly, GAD-NAP57, like GAD alone, failed to rescue growth of the cbf5ts strain at the nonpermissive temperature (Figure (Figure2C,2C, lanes 2 and 3). Because this finding was in contrast to our observations with the Cbf5p-depleted strain, we tested whether GAD-NAP57 was able to rescue a CBF5 null strain. For this purpose, a diploid strain heterozygous for CBF5 (CBF5/cbf5::TRP1; Cadwell et al., 1997 ) was transformed with pTM113 (GAD-NAP57) and sporulated, and the tetrads were dissected and allowed to germinate at room temperature and 30°C. All tetrads yielded only two viable spores, all harboring the wild-type CBF5 gene and many still carrying pTM113 (data not shown). This result indicated that NAP57 failed to complement a CBF5 null mutant. The difference in behavior of the GAL::cbf5 strain and the cbf5ts and CBF5 null strains toward NAP57 complementation is apparently caused by residual expression of CBF5 from the GAL10 promoter even when grown in glucose, as was observed previously for GAR1 in a GAL::gar1 strain (Girard et al., 1992 ). Therefore, NAP57 can complement the function of Cbf5p only in the presence of residual amounts of Cbf5p, indicating the existence of an additional function or of high-affinity binding sites for Cbf5p that NAP57 cannot rescue. Nevertheless, the mammalian NAP57 restored growth and stabilized the box H/ACA snoRNAs of a yeast strain genetically depleted of Cbf5p, supporting its function as a box H/ACA-specific snoRNP protein and putative pseudouridylase.
To test if Nopp140 interacted with snoRNPs in vivo, we took advantage of a dominant negative Nopp140 construct identified previously (Isaac et al., 1998 ). We had found that expression of the conserved carboxyl terminus of Nopp140, NoppC, chased endogenous Nopp140, NAP57, and fibrillarin out of the nucleolus but left the nucleolar localization of all other antigens tested (nucleolin, B23/NO38, UBF, and RNA pol I) unaffected. Figure Figure4A,4A, a, shows that GAR1 is also depleted from the nucleoli (open arrows) in NoppC-transfected cells (solid arrow). Therefore, the localization of all snoRNP proteins to which antibodies are currently available (NAP57, fibrillarin, and GAR1) is affected by the expression of NoppC but not the localization of other nucleolar proteins. This strongly indicates an interaction of Nopp140 with box H/ACA and box C/D snoRNPs in the cell.
Surprisingly, despite the depletion of the snoRNPs, the nucleolus appeared unaltered by phase-contrast imaging (Figure (Figure4,4, a′′, b′′, and c′′). Therefore, we tested if RNA pol I transcription, one of the main nucleolar functions, was affected in NoppC-transfected cells (Figure (Figure4B).4B). For this purpose, we studied run-on transcription in permeabilized cells by BrUTP incorporation and subsequent indirect fluorescent detection with anti-BrdU antibodies that cross-react with BrU (Wansink et al., 1993 ). In untransfected cells, BrUTP is incorporated into rRNA, as indicated by the bright fluorescence in the nucleoli (Figure (Figure4B,4B, compare b and b′′). Transfected cells in this case were identified by the mislocalization of endogenous Nopp140 in a punctate pattern in the nucleoplasm, as described previously (Figure (Figure4B,4B, solid arrow in b′) (Isaac et al., 1998 ). In all of these transfected cells, BrUTP incorporation was completely absent (Figure (Figure4B,4B, solid arrow in b), indicating an inhibition of RNA pol I transcription. Interestingly, the nucleolar localization of RNA pol I itself remained unaffected in NoppC-transfected cells (Figure (Figure4C,4C, compare c and c′). These results support the idea that the presence of snoRNPs in the nucleolus is required for transcription to occur. For example, the snoRNAs could coat the rRNA during transcription by hybridization to its complementary sequences. Finally, neither snoRNPs nor transcription seems to be required for the phase-dense appearance of the nucleolus in light micrographs.
Because Nopp140 interacted in mammalian cells with snoRNPs and NAP57 could partially complement the function of its snoRNP protein counterpart in yeast, we tested if Srp40p, the yeast Nopp140 homologue (Meier, 1996 ), also interacted with snoRNPs. For this purpose, we took advantage of a strain in which the nonessential SRP40 gene was rendered essential by the mutation of a gene, LES2, which caused synthetic lethality with the SRP40 deletion (srp40Δ). LES2 was identified in a synthetic lethal screen with SRP40 with the use of random lacZ LEU2 insertions throughout the genome by transformation with a mutagenized yeast library (Burns et al., 1994 ; our unpublished observations). The LEU2 marker inserted in LES2 allowed us to determine that only a single additional gene, aside from SRP40, was mutated in the srp40Δ les2 strain and to segregate the single les2 mutation from the srp40Δ deletion. Growth of the srp40Δ les2 double mutant was dependent on the presence of SRP40 supplied on a plasmid under its own promoter (data not shown) or under the conditional GAL10 promoter (pGAL-SRP40; Figure Figure5A).5A). Thus, the srp40Δ les2 (pGAL-SRP40) strain grew as well as wild-type yeast when Srp40p was expressed (Figure (Figure5A,5A, top, compare lanes 1 and 4) but not when Srp40p expression was repressed by glucose (Figure (Figure5A,5A, bottom, lane 4). Therefore, we generated a conditional lethal strain whose growth was dependent on the expression of Srp40p. The singly disrupted srp40Δ and les2 strains showed no or little growth defect, respectively, on glucose compared with wild-type yeast (Figure (Figure5A5A bottom, compare lanes 1, 2, and 3). However, les2 growth was severely diminished in the absence of glucose in medium containing raffinose, galactose, and sucrose (Figure (Figure5A,5A, top, lane 3). The latter phenotype was rescued by additional SRP40 expression from its own promoter on a low-copy-number CEN plasmid, indicating a tight relationship between the SRP40 and LES2 genes (data not shown, but see Figure Figure5A,5A, top, lane 4).
To test if depletion of Srp40p in the conditional strain affected the stability of snoRNAs, as in the case of Cbf5p (Figure (Figure3B),3B), the various strains were grown for 0 and 24 h in glucose-containing medium. The stability of the snoRNAs was tested on Northern blots with antisense oligonucleotide probes to two box C/D (U3 and U14) and three box H/ACA (snR3, snR11, and snR33) snoRNAs (Figure (Figure5B).5B). As expected from their undiminished growth under these conditions, the levels of all tested snoRNAs remained unaffected in the wild-type and single-mutant strains (Figure (Figure5B,5B, lanes 1–6). However, depletion of Srp40p in the double mutant led to the specific reduction of the tested box H/ACA but not the box C/D snoRNAs (Figure (Figure5B,5B, compare lanes 7 and 8). Therefore, Srp40p depletion exhibited the same phenotype as that of the integral box H/ACA proteins Cbf5p, Nhp2p, and Nop10p (Henras et al., 1998 ; Lafontaine et al., 1998 ; Watkins et al., 1998a ). These findings demonstrated that Srp40p genetically interacts with box H/ACA snoRNAs and suggested that Srp40p in yeast associates with snoRNPs, such as Nopp140, in mammalian cells. Therefore, these data add to the previously presented evidence that Srp40p is the bona fide Nopp140 homologue (Meier, 1996 ) and suggest that Srp40p and consequently Nopp140 play a role in snoRNP function.
The previously determined and the novel associations of Nopp140 described here can be distilled into a model of Nopp140 interactions (Figure (Figure6).6). Nopp140 is the first protein shown to interact with both major classes of snoRNPs, box H/ACA and box C/D. This is in contrast to the integral components of snoRNPs, which are specific for each class. In addition, a Nopp140-snoRNP interaction appears to be conserved in evolution, because Srp40p, the yeast Nopp140 homologue, genetically interacts with at least one class of snoRNPs.
The unique protein composition of box H/ACA and box C/D snoRNPs depicted in Figure Figure66 was derived from recent experiments in yeast (Lübben et al., 1995 ; Balakin et al., 1996 ; Ganot et al., 1997 ; Gautier et al., 1997 ; Henras et al., 1998 ; Lafontaine et al., 1998 ; Watkins et al., 1998a ; Lafontaine and Tollervey, 1999 ) and from the specific precipitation of human box H/ACA snoRNAs with GAR1 antibodies (Dragon, Pogaić, and Filipowicz, unpublished data). All of these proteins are highly conserved and exhibit sequence identities that range from 45 to 70% between yeast and mammals, including the comparison of the novel rat NAP65 and yeast Nop5/58p (Figure (Figure6,6, mammalian proteins in bold, yeast homologues in italic). Our studies demonstrate that not only are the protein components conserved from yeast to mammals but that the protein compositions of both major classes of snoRNPs are conserved as well. Thus, based on their coprecipitation, NAP57, GAR1, and box H/ACA snoRNAs form a specific complex analogous to yeast box H/ACA snoRNPs. Nopp140, however, in addition to these box H/ACA snoRNP components, precipitates fibrillarin, the novel NAP65, and box C/D snoRNAs. Because the box H/ACA snoRNP proteins and RNAs form a distinct complex, the other components coprecipitating with Nopp140 likely constitute a separate complex, as they do in yeast, i.e., box C/D snoRNPs. In fact, the snoRNP composition is so highly conserved between mammals and yeast that rat NAP57 functionally replaced its yeast counterpart in at least one strain genetically depleted of its homologue Cbf5p. This finding is similar to the functional complementation of the yeast box C/D snoRNP protein Nop1p by its human orthologue fibrillarin (Jansen et al., 1991 ). The establishment of the composition of mammalian snoRNPs has gained clinical importance as a result of the identification of dyskerin, the human NAP57 homologue, as the protein mutated in the X-linked recessive bone marrow failure syndrome dyskeratosis congenita (Heiss et al., 1998 ).
At present, there are antibodies available against only three mammalian snoRNP proteins: fibrillarin, NAP57, and GAR1. All of these proteins associate with Nopp140 and behave according to the emerging model for snoRNP composition (Figure (Figure6).6). In addition, Nopp140 coprecipitated the novel NAP65 and several unidentified bands (Figure (Figure1A,1A, open dots). It is likely that some of these bands correspond to additional snoRNP proteins. For example, the 24-kDa band is a good candidate for the mammalian NHP2 homologue, whereas the 66- or 61-kDa band could correspond to the vertebrate Sik1p/Nop56p homologue. GAR1 is not visible because it comigrates on that gel with the IgG light chains (Figure (Figure1A,1A, asterisks), and NOP10 is expected to migrate with the gel front. Analysis of the residual bands will possibly identify additional snoRNP proteins, particularly those of the box C/D class, because NAP57, GAR1, NHP2, and NOP10 are predicted to make up the complete complement of box H/ACA snoRNP proteins (Henras et al., 1998 ; Watkins et al., 1998a ). Interestingly, a 65/68-kDa protein band was previously identified to specifically cross-link to box C/D snoRNAs (Caffarelli et al., 1998 ; Watkins et al., 1998b ). This protein likely corresponds to the NAP65 characterized here. Therefore, those findings support our evidence that NAP65 is a specific part of box C/D snoRNPs, as demonstrated for its yeast counterpart Nop5/58p (Lafontaine and Tollervey, 1999 ).
Our data demonstrate that Nopp140 interacts with both major classes of snoRNPs. Thus, we show that proteins and RNAs of box H/ACA and box C/D snoRNPs coprecipitate with Nopp140 under low-stringency conditions (Figure (Figure2).2). Furthermore, proteins of both classes are specifically chased out of the nucleolus by the dominant negative Nopp140 carboxyl terminus (Figure (Figure4A).4A). Interestingly, the association of Nopp140 with box H/ACA snoRNPs appears tighter than that with the box C/D class (depicted by the thick arrows in Figure Figure6).6). This is supported by the coimmunoprecipitation of stoichiometric amounts of NAP57, a box H/ACA snoRNP protein, under conditions that precipitate only small amounts of box C/D proteins (Figure (Figure1A).1A). A preferred interaction with box H/ACA snoRNPs is also suggested by the genetic depletion studies of Srp40p, the yeast Nopp140 homologue (Figure (Figure55B).
What is the nature of the Nopp140-snoRNP interaction? Is it transient or is Nopp140 an integral component of snoRNPs? We suggest that Nopp140 only transiently associates with snoRNPs. Thus, Nopp140 is easily isolated as a single species (away from snoRNPs) under low-ionic-strength or high-salt (500 mM sodium chloride) conditions (Meier and Blobel, 1990 ). Moreover, most box C/D snoRNP proteins already dissociate from Nopp140 at physiological salt concentrations (Figure (Figure1),1), and purified yeast box H/ACA snoRNPs appear to lack the yeast Nopp140 homologue Srp40p (Lübben et al., 1995 ; Watkins et al., 1998a ). It is likely that snoRNPs remain intact under conditions in which Nopp140 dissociates, because intra-snoRNP interactions survive even the harsh conditions of cesium chloride gradients (Lübben et al., 1995 ). Such a transient and reversible association of Nopp140 with snoRNPs is consistent with it functioning as a chaperone of snoRNPs, either for their intranuclear transport or during snoRNP-rRNA association and dissociation (see below).
It is unclear why Nopp140 exhibits a preferred affinity for NAP57 and if all of this NAP57 is snoRNP associated. Surprisingly, extensive yeast two-hybrid analysis with both proteins, full-length and individual domains, failed to reveal any interaction (our unpublished results). Although this could be a peculiarity of the yeast two-hybrid system, it indicates that Nopp140 may not bind to box H/ACA snoRNPs through NAP57. Alternatively, the highly and mostly positively charged carboxyl terminus of NAP57 would appear to be a good interacting partner with the highly phosphorylated repeat domain of Nopp140. This possibility is supported by the fact that such a charged carboxyl terminus is also present in NAP65, which could provide the Nopp140 handle for box C/D snoRNPs. Furthermore, these charged carboxyl termini are conserved in Cbf5p and Nop5/58p, the yeast counterparts of NAP57 and NAP65, respectively. In summary, however, the components of snoRNPs that bind directly to Nopp140 remain to be identified and/or confirmed. It will be interesting to determine what governs the Nopp140-snoRNP association and what fraction of Nopp140 at any given time is associated with these particles.
rRNA modification in vertebrate cells occurs cotranscriptionally (Maden, 1990 ). This is supported by our finding that the lack of snoRNPs in nucleoli leads to transcription arrest. Although it cannot be ruled out that other factors contribute to our observation, it is interesting to speculate that newly transcribed rRNA emerging from RNA pol I is immediately covered by snoRNPs. Consequently, the absence of these snoRNPs could induce a negative feedback mechanism of rRNA transcription. It is possible that transcription is physically restrained by the accumulation of misfolded rRNA or that there is a direct interaction between the snoRNPs and RNA pol I. For example, Nopp140 itself could provide such a link because it has been reported to function as a transcription factor in another system (Miau et al., 1997 ). Although the interaction between snoRNPs and the RNA pol I transcription machinery is an intriguing possibility, it needs to be further investigated, e.g., in in vitro transcription systems.
In conclusion, we report here on the interaction of Nopp140 with both major classes of mammalian snoRNPs. What is the function of this interaction? Three possibilities come to mind. First, Nopp140 shuttles between the nucleolus, the coiled bodies, and the cytoplasm and may as such facilitate the transport of snoRNPs through the nucleoplasm, e.g., between the nucleolus and the coiled bodies. A snoRNP transport role for Nopp140 is supported by the specific depletion of snoRNPs from the nucleolus upon expression of the dominant negative NoppC construct. Second, Nopp140 may aid in the biogenesis of snoRNPs, similar to the function of the SMN and SIP proteins in the assembly of spliceosomal snRNPs (Fischer et al., 1997 ). Third, Nopp140 may be directly involved in the function of snoRNPs, the modification and processing of rRNA. As such, it may aid the binding or release of the snoRNPs to or from the nascent rRNA. Therefore, its function could be analogous to that of SR proteins, which are required for the assembly and function of the spliceosome (for review, see Valcarcel and Green, 1996 ). These possibilities are not mutually exclusive and may be partially overlapping. Nopp140 appears to exhibit a conserved function as a chaperone of snoRNPs.
We thank John Aris, Jonathan Backer, John Carbon, Witold Filipowicz, Denis Lafontaine, Larry Rothblum, David Tollervey, and Susan Wente for reagents and Denis Lafontaine, Maria Savino, and Susan Smith for helpful suggestions. We are grateful for the technical assistance by Judah Goldberg and to the Analytical Imaging Facility of Albert Einstein College of Medicine for the use of the equipment. We appreciate the critical reading of the manuscript by Witold Filipowicz, Susan Smith, and Jonathan Warner. This work was supported by grants from the National Institutes of Health and the Howard Hughes Medical Institute–Research Resources Program for Medical Schools
In agreement with our data, hNop5/Nop58, the human homologue of rat NAP65 and yeast Nop5/58p, was now also shown to be a specific component of box C/D snoRNPs (Lyman et al. . RNA 5, 1597–1604).