|Home | About | Journals | Submit | Contact Us | Français|
The selection of sites for pseudouridylation in eukaryotic cytoplasmic rRNA occurs by the base pairing of the rRNA with specific guide sequences within the RNA components of box H/ACA small nucleolar ribonucleoproteins (snoRNPs). Forty-four of the 46 pseudouridines (Ψs) in the cytoplasmic rRNA of Saccharomyces cerevisiae have been assigned to guide snoRNAs. Here, we examine the mechanism of Ψ formation in 5S and 5.8S rRNA in which the unassigned Ψs occur. We show that while the formation of the Ψ in 5.8S rRNA is associated with snoRNP activity, the pseudouridylation of 5S rRNA is not. The position of the Ψ in 5.8S rRNA is guided by snoRNA snR43 by using conserved sequence elements that also function to guide pseudouridylation elsewhere in the large-subunit rRNA; an internal stem-loop that is not part of typical yeast snoRNAs also is conserved in snR43. The multisubstrate synthase Pus7 catalyzes the formation of the Ψ in 5S rRNA at a site that conforms to the 7-nucleotide consensus sequence present in other substrates of Pus7. The different mechanisms involved in 5S and 5.8S rRNA pseudouridylation, as well as the multiple specificities of the individual trans factors concerned, suggest possible roles in linking ribosome production to other processes, such as splicing and tRNA synthesis.
RNA harbors a large array of structurally diverse, modified nucleosides (33, 93). Pseudouridine (5-ribosyluracil; Ψ) was the first discovered and is the most abundant (for a review, see reference 16). The isomerization of U to Ψ involves the recognition of the specific U followed by the cleavage of the N-glycosidic bond, the rotation of the base, and the formation of a C-glycosidic bond (82). At the level of local RNA structure, Ψ contributes to greater stability relative to that of U by providing an additional hydrogen-bonding capability that allows a water-mediated bridge to form between the base and the sugar-phosphate backbone as well as improved base stacking (for a review, see reference 16).
In eukaryotes, Ψ is found at specific sites in the rRNAs, tRNAs, and snRNAs. The cytoplasmic rRNA precursor (pre-rRNA) of eukaryotes is abundantly modified. As a result, in addition to about 54 ribose-methylated and 10 base-methylated nucleosides (61, 89, 108), 46 Ψs are reported for the mature cytoplasmic rRNA of Saccharomyces cerevisiae (6, 7, 65, 83, 85). The pseudouridylation sites in eukaryotic, cytoplasmic rRNA are selected by the base pairing of the cytoplasmic rRNA with specific guide sequences present in the box H/ACA family of small nucleolar RNAs (snoRNAs) (for reviews, see references 10, 21, 54, and 114). These guide RNAs are associated with protein and function as ribonucleoprotein (RNP) particles, termed small nucleolar RNPs (snoRNPs).
Each box H/ACA snoRNP contains four core proteins in addition to the guide RNA (71). One of the core proteins, Cbf5 in yeast (dyskerin in humans), has significant similarity to TruB, a known Escherichia coli tRNA Ψ synthase (40, 55), and experimental and structural evidence supports the inference that the protein is indeed the catalytic component of box H/ACA snoRNPs (17, 60, 88, 115). Defective ribosome activity due to mutations in dyskerin has been implicated in the human disease dyskeratosis congenita (97, 112). The three other core box H/ACA snoRNP proteins also are essential for pseudouridylation, as demonstrated with yeast (12, 37), contributing to a stable, efficient complex capable of productive association with the substrate in a manner that establishes the characteristic spacing (14 to 16 nucleotides [nt]) between the active site and box H or ACA sequence element (5, 36, 38, 51, 60, 66, 81, 88, 90).
The goal of several initiatives is to identify and assign functions to the entire catalog of snoRNAs from various model organisms (14, 54, 59, 99, 100, 102). Presently, the population of snoRNAs in S. cerevisiae is the best characterized. Seventy-six snoRNAs have been identified in yeast (86). Remarkably, not only have the guide modification site pairings for most of the 2′-O-methylated (box C/D snoRNA targets) and Ψ sites in yeast been worked out, the guide assignments also have been experimentally verified by showing the specific loss of a particular modification(s) following the deletion of each snoRNA gene. Specifically, in the case of the box H/ACA type, 28 snoRNAs have been assigned to 44 of the 46 known Ψs in yeast rRNA (4, 102, 107).
The formation of the numerous Ψs in the spliceosomal snRNAs of vertebrates is guided by a class of small RNAs that reside in the nucleoplasmic Cajal bodies and are closely related to the box H/ACA snoRNAs (44, 114). The box H/ACA small-Cajal-body RNAs also function as dyskerin-containing RNPs and contain the same small sequence motifs as the typical box H/ACA snoRNAs, as well as an additional sequence element implicated in the characteristic localization (92). Three of the six Ψs in and near the branch point recognition region of vertebrate U2 snRNA have counterparts in yeast U2 snRNA (63). Whereas the Ψs of human U2 snRNA are assigned box H/ACA small-Cajal-body RNA guides (59), initial work had shown that yeast U2 snRNA modification does not depend on the Cbf5 guide RNA system (63, 68). Instead, the classic protein-only enzymes Pus1 and Pus7 were shown to each catalyze the formation of individual Ψs in yeast U2 snRNA (63, 68). These proteins also pseudouridylate tRNAs in yeast (9, 68) and are homologs of eubacterial tRNA modification enzymes (49, 55, 76). However, recently a snoRNA was shown to guide the formation of the last of the three Ψs in yeast U2 snRNA (62). Since U2 snRNA is transcribed by RNA polymerase II (RNAP II), the finding revealed that box H/ACA snoRNP-mediated modification in yeast is not limited to RNAP I transcripts (18S, 5.8S, and 25S rRNAs).
In this work, we investigate the factors involved in the pseudouridylation of the two small rRNAs that are found in the large subunit of the S. cerevisiae cytoplasmic ribosome. To address the mechanisms involved in the pseudouridylation of 5S and 5.8S rRNAs, we examined the formation of Ψs in a strain impaired for Cbf5 activity. We show that 5.8S rRNA pseudouridylation requires functional Cbf5 and is guided by the snoRNA snR43. We demonstrate that the formation of Ψ in 5S rRNA is not dependent on a Cbf5-containing RNP and instead is catalyzed by Pus7, which also forms Ψ in U2 snRNA and tRNAs. Thus, the pseudouridylation of 5S and 5.8S rRNA is carried out by different mechanisms, with each of the trans factors involved displaying intriguing multispecificity.
The deletion of the snR43 gene was performed in the haploid YS602 (MATα ade2-101 his3-11,15 trpΔ901 ura3-52 leu2-3,112) S. cerevisiae strain by the replacement of the entire coding region with the TRP1 auxotrophic marker from Kluyveromyces lactis that was amplified from plasmid pBS1479 (87) with the primers WD440 (5′-CACGCGTGTTTTGTGCAGTGTATTACCAACTTGCGCATGCAAGGATATCATGATATCGAATTCCTGC-3′) and WD441 (5′-AGTATAAGAAAAATAAGCAAAATAAGTATACATAAAAATAAAT ACGAATAGTACGACTCACTATAGGG-3′), which add homologous sequences flanking snR43 to the ends of the marker gene. Following transformation with the PCR product (31), positive transformants (snR43Δ::TRP1) were selected on yeast nitrogen base plates lacking tryptophan, and the integration of the TRP1 marker at the correct location was verified in individual isolates by the isolation of genomic DNA followed by PCR with the primers WD328 (5′-CTCTAGAACTAGTGGATCC-3′) and WD442 (5′-GTCTAATGTGCTTGTTTGTGC-3′).
The haploid MATα pusΔ S. cerevisiae strains were purchased from Open Biosystems, Inc. (Huntsville, AL). The cbf5-D95A S. cerevisiae strain was described previously (115).
To generate an S. cerevisiae plasmid expressing the protein encoded by the PUS7 gene in a functional form, a region containing the entire PUS7 gene, as well as 261 bp upstream and 134 bp downstream, was amplified from S. cerevisiae genomic DNA using the primers WD446 (5′-ATTTCCGAGCTCGGTAGCACAGGAAGCGTCTAAAG-3′) and WD447 (5′-GGCTTTCTCGAGGTGTGCGATGCGCAGATACATATTTAC-3′) and Platinum Taq high-fidelity polymerase (Invitrogen, Carlsbad, CA). The PCR product was cloned into pCR2.1-TOPO (Invitrogen) and propagated in E. coli TOP10 (Invitrogen). The resulting plasmid was digested with XhoI and SacI restriction enzymes (New England Biolabs, Ipswich, MA), the sites for which were added in the initial PCR as part of the primers, and the PUS7 gene region was cloned into the polylinker of XhoI/SacI-cut pRS415 (103) to generate pPUS7(RS415). The plasmid was isolated from E. coli TOP10 and transformed into the haploid yeast pus7Δ strain.
Primer labeling and RNA sequencing were done as described previously (61, 70). The rRNA Ψ sites were detected by primer extension after the treatment of the RNA to create carbodiimide adducts at the sites of Ψ as described previously (102). The primers to sequence rRNA and detect the various Ψs are as follows: WD387 (102) for 25S rRNA, WD443 (5′-GCGTTCAAAGATTCGATGATTC-3′; purified by polyacrylamide gel electrophoresis) for 5.8S rRNA, and WD444 (5′-CCACTACACTACTCGGTCAGGC-3′) for 5S rRNA.
The snoGPS Web server was described previously (101); the query sequence used in the search for guides to the 5S and 5.8S rRNA targets consisted of the known H/ACA snoRNA collection (4, 86, 99, 102, 107) in FASTA format. snR43 homologs were identified using the fungal BLAST search provided by the Saccharomyces Genome Database (43); the alignment of snR43 homologs was done using the EMMA multiple-sequence alignment program that is part of EMBOSS (91) with initial formatting applied by PrettyPlot MulSeqs, provided via http://biotools.umassmed.edu/. Mfold (116) was used to consider local folding.
A single Ψ is known to occur in each of the two small rRNAs (5S and 5.8S) that are components of the large subunit of S. cerevisiae cytoplasmic ribosomes (Fig. (Fig.1)1) (75, 85, 94). To determine the factors involved in the pseudouridylation of these rRNAs, we first examined whether the snoRNP-associated Ψ synthase Cbf5 catalyzes the formation of either of these Ψs. We did this by using a mutant yeast strain (cbf5-D95A) that only expresses a form of Cbf5 that harbors a point mutation (D95A) in the identified catalytic domain. This mutant strain lacks the detectable pseudouridylation of rRNA in vivo and exhibits severely impaired cell growth (115). Total RNA was isolated from the mutant strain and a control strain with a wild-type CBF5 gene, and the Ψ sites in 5S and 5.8S rRNA were examined to see if there was a loss of pseudouridylation. RNA was treated with the chemical N-cyclohexyl-N′-(2-morpholinoethyl)carbodiimide methyl-p-toluenesulfonate (CMC) followed by mild alkali treatment, and Ψs were detected as strong, CMC-dependent primer extension stop bands at positions one nucleotide before the Ψ sites. Note that in our experiments there also were template-specific background ladders that served as internal controls to show that sufficient template RNA was present in each reaction. As illustrated in Fig. Fig.2,2, the formation of Ψ73 of 5.8S rRNA is abrogated in the cbf5-D95A strain but is present in the wild-type control (lanes 5 to 8), while the Ψ is observed at position 50 of 5S rRNA in both the cbf5-D95A mutant strain and a control strain (lanes 1 to 4). This result suggests that Ψ formation in 5.8S rRNA is dependent on the catalytic activity of the Ψ synthase Cbf5 in a box H/ACA snoRNP, whereas surprisingly Ψ formation in 5S rRNA is dependent on another Ψ synthase, presumably a protein enzyme.
Given that Ψ formation in 5.8S rRNA is dependent on Cbf5 functioning in a putative box H/ACA snoRNP, we sought to identify the particular guide RNA in the RNP particle. Recent experimental and computational approaches have most likely revealed the complete collection of S. cerevisiae snoRNAs (4, 20, 25, 61, 69, 102, 107). Particularly underscoring the comprehensive nature of the current collection, a recent experimental approach to purify and identify the full array of S. cerevisiae box H/ACA snoRNAs based on associations with H/ACA snoRNPs discovered only one novel snoRNA, yet the set lacked only a single box H/ACA snoRNA that was previously recognized (107). Therefore, we began our search by examining the pseudouridylation pockets of the known H/ACA snoRNAs for complementarity to the sequences immediately flanking the 5.8S Ψ site. As shown in Fig. Fig.3A,3A, snR43 displays a significant potential for base pairing with the region modified in 5.8S rRNA, positioning U73 at the pocket for isomerization. snR43 previously had been predicted, and subsequently shown, to guide Ψ966 of yeast 25S rRNA (84, 107); in fact, the same guide elements are involved (Fig. (Fig.3B3B).
As an independent technique to identify putative guides for 5.8S rRNA that is not limited to considering the previously identified pseudouridylation pockets of the box H/ACA snoRNAs, we turned to the computational method described by Schattner and colleagues (101, 102). This approach, on the snoGPS server, identified only snR43 as a candidate and suggested the same guide RNA-target pairing revealed by a manual inspection of the known pockets.
With two analytical approaches indicating snR43 as the guide, we investigated whether snR43 is involved in forming Ψ in 5.8S rRNA in vivo. To do this, we generated a strain in which the gene for snR43 is deleted by gene replacement with a selectable marker. RNA from the deletion strain and the control parental strain in which snR43 is intact was prepared and used for CMC treatment, and this was followed by primer extension to examine several Ψ sites in rRNA to see if pseudouridylation is altered. In the parent strains, the Ψ is seen in 5.8S rRNA (Fig. (Fig.3C,3C, lanes 1 and 2). The formation of Ψ73 in 5.8S rRNA is indeed abolished in the snR43 deletion strain (Fig. (Fig.3C,3C, lanes 3 and 4), demonstrating that snR43 is involved in the formation of the Ψ of 5.8S rRNA. Primer extension was performed with the same RNA samples using primers specific for a region in 25S rRNA, and, consistent with the reported role for snR43 as the guide for 25S-Ψ966 (107), the formation of Ψ966 is abrogated in the snR43 deletion strain (Fig. (Fig.3D,3D, lanes 7 and 8). Importantly, CMC-dependent signals corresponding to several nearby Ψs persist, showing that isomerization to Ψ still occurs in the snR43 deletion strain for surrounding sites in the region of 25S rRNA examined in Fig. Fig.3D3D (lanes 6 and 8). Taken together, the results show that snR43 serves as a guide for specific Ψ sites in both 5.8S and 25S rRNA using the same guide elements.
The secondary structure of a typical eukaryotic box H/ACA snoRNA consists of two primary extended stem-loops, each separated by a single-stranded hinge region containing the box H motif, all followed by the single-stranded ACA motif three nucleotides from the 3′ end of the RNA (10, 21, 54, 114). If an extended 5′ or 3′ stem-loop functions as a guide domain, then an internal bulged region is present forming the pseudouridylation pocket. snR43 contains an additional 53-nt stem-loop after the box H sequence that we refer to here as the insert stem-loop (Fig. (Fig.4).4). While not unique, such an arrangement is not observed in most eukaryotic snoRNAs. Other S. cerevisiae snoRNAs (snR42, snR46, and snR191) contain a similarly positioned, additional stem-loop, yet in two of the three cases the insert stem-loop is much smaller, i.e., 36 nt for snR42 and 28 nt for snR46. With regard to the extended nature of the insert stem-loop, snR43 resembles the S. cerevisiae snR191 and Euglena gracilis snoRNA h1, both of which have significant stem-loops (98 and 48 nt, respectively) between flanking box H and ACA domains (4, 98).
Recently, based on comparisons of sequenced yeast genomes, it has been possible to ascertain and add support for the existence of small structured RNAs (67, 69, 102). In an effort to provide support for the predicted secondary structure and to gain further insight into the function of snR43 in related yeasts, we searched for snR43 homologs from the available genomes of several hemiascomycetes (30). These putative snR43 homologs were manually inspected for the potential to form box H/ACA snoRNA-like secondary structures. Our alignment, based upon primary sequences and secondary structures, is presented in Fig. Fig.4A.4A. The 5′ portion of the snR43 species, specifically the guide elements and portions of the insert stem-loop, is more conserved than the 3′ end. The high conservation at the sequence level of a substantial portion of the insert stem-loop, in particular the 3′ portion of the loop, is unexpected; however, the Yarrowia lipolytica sequence shows considerable variation in this region. Sequences of the hemiascomycete snR43 homologs also can be readily modeled using Mfold (116) to form secondary structures reminiscent of the entire S. cerevisiae snR43, with the H and ACA boxes similarly configured relative to the three primary stem-loops (data not shown).
Given the high conservation of the guide elements and overall structural similarity to the functional S. cerevisiae snoRNA, the aligned sequences likely correspond to orthologs of snR43 in the other yeasts analyzed. In contrast to the variation seen in the orthologs of snR43, the rRNA is highly conserved (30). The region of 25S rRNA proposed to interact with snR43 is 100% conserved among these hemiascomycetes, while the corresponding regions in the 5.8S rRNAs display only a few differences. There is some variation in the snoRNA 3′ guide element distal to the site of pseudouridylation and outside of the region proposed to interact with S. cerevisiae 25S rRNA; these sequence changes are not always accompanied by compensating changes in rRNA sequence and, thus, tend to weaken the potential interaction with 5.8S rRNA and, in some cases, strengthen the proposed interaction with 25S rRNA (Fig. (Fig.4B).4B). Overall, the high degree of conservation in the rRNA and the base-pairing potential displayed by the homologs to the known sites of snR43 action suggest that these snoRNAs also function as guides for the sites corresponding to 5.8S-Ψ73 and 25S-Ψ966 in these other organisms. Thus, this dual action of snR43 does not appears to be unique to S. cerevisiae or even simply to the members of the closely related Saccharomyces sensu stricto group.
As shown in Fig. Fig.22 (lanes 1 to 4), Ψ formation in 5S rRNA depends on a Ψ synthase other than Cbf5, presumably a protein enzyme acting without a guide RNA. Unlike the case for the 5.8S rRNA Ψ, using the known collection of yeast H/ACA snoRNAs, no well-ranking, convincing candidate is proposed by the snoGPS server for the 5S rRNA Ψ. Nine Ψ synthases, including Cbf5, were identified in the yeast genome based on sequence and motif analyses (2, 3, 19, 55); later, a synthase with a highly divergent sequence was added to bring the total to 10 (63). In order to identify the synthase catalyzing 5S rRNA pseudouridylation, we began investigating the status of 5S rRNA pseudouridylation in strains in which the other nonessential Ψ synthase genes are deleted. The disruption of the gene corresponding to the 5S rRNA Ψ synthase would result in the loss of the 5S rRNA Ψ, presuming that the activity is nonredundant. As illustrated in Fig. Fig.5A,5A, primer extension analysis following CMC treatment shows the loss of the CMC-dependent stop in the pus7Δ strain (lanes 9 and 10) and not in the pus1Δ and pus4Δ strains (lanes 5 to 8), suggesting that Pus7 catalyzes the formation of Ψ50 in 5S rRNA of S. cerevisiae.
To verify that the absence of Pus7 is responsible for the loss of the formation of the 5S rRNA Ψ in the deletion strain, we restored Pus7 expression to the pus7Δ strain. The PUS7 gene, along with 261 bp upstream and 134 bp downstream, was amplified from the yeast genome and cloned into a LEU2/CEN-based, low-copy-number yeast shuttle plasmid, which subsequently was transformed into the pus7Δ strain; the transformation of an empty plasmid served as a control. RNA from both strains was analyzed by CMC treatment followed by primer extension to determine the pseudouridylation state of position 50 in the 5S rRNA. Whereas the pus7Δ strain harboring the empty vector fails to generate the 5S rRNA Ψ (Fig. (Fig.5B,5B, lanes 5 to 8), the addition of the PUS7 gene to the plasmid rescues the formation of the 5S rRNA Ψ (Fig. (Fig.5B,5B, lanes 9 to 12). These results confirm that PUS7 gene deficiency is responsible for the loss of the 5S rRNA Ψ in the deletion strain.
Pus7 is a multisite, multisubstrate Ψ synthase in vivo, and the activity of recombinant Pus7 in the absence of other factors has been demonstrated with U2 snRNA and tRNA substrates in vitro (9, 63). In those substrate RNAs, the seven nucleotides encompassing the modification site conform to a consensus sequence (9). Presumably, these sequences form, at least in part, the synthase recognition site (9); however, for U2 snRNA it was shown that activity also depends on an additional element located downstream of the 7-nt consensus sequence critical for recognition and/or modification by Pus7 (63). Strikingly, the alignment of the seven nucleotides surrounding the sites of S. cerevisiae Pus7 activity (9, 104) demonstrates that the modification site in 5S rRNA matches the consensus sequence seen in other Pus7 substrates (Fig. (Fig.6).6). Hence, the observed Pus7 dependency of the formation of the 5S rRNA Ψ is reasonable in light of the previously characterized targets of Pus7 activity. Taken together, the results concerning 5S rRNA and the prior characterization of Pus7 indicate that Pus7 directly catalyzes the pseudouridylation of 5S rRNA, and catalysis most likely is not dependent on a guide RNA or additional protein factors.
By using a strain in which Cbf5, the Ψ synthase enzyme associated with the box H/ACA snoRNP complexes, is compromised for activity, we have shown that while the formation of the Ψ in S. cerevisiae 5.8S rRNA is associated with snoRNP activity, the formation of the Ψ in S. cerevisiae 5S rRNA is independent of Cbf5. Our analysis suggests that 5.8S rRNA modification at this position is not unique to S. cerevisiae and most likely occurs in a wide range of hemiascomycetes, a group that covers an evolutionary breadth comparable to that of the entire chordate phylum (24). Likewise, the formation of the Ψ in 5S rRNA has been reported for other hemiascomycetes (75, 105).
The synthesis of 5.8S rRNA in yeast is complex (27-29), and we have now shown that an snoRNA is involved; snR43 uses the same guide sequences that function elsewhere (position 966 in 25S rRNA) to direct the pseudouridylation of 5.8S rRNA. snR43 adds to the list of yeast snoRNAs that guide pseudouridylation at more than one site in rRNA, bringing the total to 13, nearly half of the 28 guide H/ACA snoRNAs. snR43 and snR49 are the only S. cerevisiae snoRNAs known to use the same guide domain to target the pseudouridylation of two different species of mature rRNA.
The nucleotide equivalent to yeast 25S-Ψ966 in human 28S rRNA also is a Ψ (83), and the site is one of two assigned to the 133-nt snoRNA ACA9 (54, 59). Only the guide elements in ACA9 show any resemblance to snR43, constrained by the need to base pair with a homologous site in 28S rRNA; ACA9 does not target human 5.8S rRNA. A similar phenomenon has been noted for the yeast and human snoRNAs guiding the highly conserved Ψs in helix 69 of large-subunit rRNA (4). However, in that case the overall secondary structure and length both are similar, unlike the case of yeast snR43 and its human counterpart, for which even those characteristics differ markedly due to the insert stem-loop and the greater than 70-nt difference in size. Human and plant 5.8S rRNAs do not contain a Ψ at a position equivalent to that of yeast 5.8S-Ψ73, yet neighboring nucleotides are modified (Fig. (Fig.1B)1B) and assigned to particular snoRNAs (14, 59).
Given that guide H/ACA snoRNAs are involved in the formation of Ψs in transcripts produced by polymerases other than RNAP I, the fact that no snoRNA is involved in the modification of 5S rRNA, a product of RNAP III, was not obvious a priori. By testing known RNA guide-independent Ψ synthases of yeast, we have identified Pus7 as the synthase catalyzing the formation of the Ψ in 5S rRNA. This is the first report of a Ψ in cytoplasmic eukaryotic rRNA that is not snoRNA guided. Importantly, now all 46 U's known to be converted to Ψs in S. cerevisiae rRNA are associated with a modifying trans-acting factor.
On the basis of transcriptional coregulation, Pus7 was included in a set of genes that function in ribosome and rRNA biosynthesis (109). Here, we have directly linked Pus7 activity to the modification of 5S rRNA. The targets of Pus7 all share a 7-nt consensus at the site of modification (9). How Pus7 recognizes its natural targets still is not known. Presumably, there is more determining a Pus7 substrate than the short 7-nt consensus encompassing the modification site, because matches are moderately frequent in RNA (data not shown). Moreover, even within a recognized substrate, only particular sites are modified, while other matches are not. Particularly illuminating is the fact that roughly 30 nt downstream of the site pseudouridylated by Pus7 in S. cerevisiae 5S rRNA (nt 79 to 85) is a perfect match to the particular version of the 7-nt sequence found at the site of Pus7 action in U2 snRNA, yet position 50 is the only Ψ in S. cerevisiae 5S rRNA (75, 85). Similarly, U2 snRNA has a match downstream (nt 52 to 58) as well, within a region previously shown to be important for modification by Pus7, yet this site is not modified either (62, 63). The inability of Pus7 to modify substrate RNAs at additional sites that match the consensus may be due to the masking of these sites by a protein(s). Alternatively, only certain portions of substrates may be suitable for modification by Pus7, and perhaps only at certain stages of RNA folding, as seen with other Ψ synthases and enzymes modifying uridine (41). Consistent with this possibility, only an altered conformation of tRNA, called the lambda form (42) and featuring an unfolded D-stem/loop, could be docked computationally into the presumed active site of E. coli TruD (26). Until now, natural substrates for studying Pus7 have been in short supply; our discovery that the very abundant 5S rRNA is a Pus7 substrate means that the pus7Δ strain should provide large amounts of 5S rRNA/RNP substrate, thereby leading to a better understanding of recognition and catalysis by Pus7.
Until recently, the prevailing view was that the pseudouridylation and 2′-O methylation of eukaryotic cytoplasmic rRNA were guided by snoRNAs in snoRNP complexes. Now, for each of the two classes of nucleotide modification that are introduced primarily by snoRNPs, an exception has been identified in which rRNA modification depends on classic protein-only enzymes. Our elucidation of Pus7 as the synthase responsible for the pseudouridylation of 5S rRNA, combined with the revelation that the protein Spb1 acts in the 2′-O methylation of a site in yeast 25S rRNA (58), means that snoRNP-independent mechanisms need to be considered for both types of rRNA modification. The reason for the different mechanisms and particular targets is not obvious. Possible reasons for different mechanisms are that (i) separate machinery enables coupling to separate regulatory networks, (ii) they assure that certain modifications are formed irrespective of the bulk of rRNA modifications, and (iii) the bacterium-like site-specific protein mechanism may have been retained from a common ancestor (Spb1 is related to RrmJ, the enzyme that modifies the neighboring site in bacteria) (58). While there presently is a paucity of knowledge on their exact role, no single snoRNP-catalyzed modification has been identified as being overly critical in yeast (8, 52). Perhaps the rRNA modifications introduced by guide RNA-independent enzymes are more important and necessitate separate machinery. Whereas the Spb1-guided modification has been shown to be essential for normal ribosome production and translation (58), the significance of the Pus7-catalyzed Ψ in 5S rRNA is not obvious from its position, and a critical role is not supported by phylogenetic conservation outside of ascomycetes.
The pseudouridylations of the RNAP II-transcribed U2 snRNA and the RNAP I-transcribed pre-rRNA are known to be linked by an snoRNP-dependent pathway; snR81 uses different guide domains to target a Ψ site in rRNA and one in U2 snRNA (62, 102). We have now demonstrated a link between the pseudouridylation of U2 snRNA and the RNAP III-transcribed 5S rRNA by the snoRNA-independent synthase Pus7, which also participates in the modification of tRNAs (RNAP III transcripts). The 5S rRNA genes are interspersed between the rRNA repeats in S. cerevisiae and are, by definition, nucleolar. The observed spatial juxtaposition of pre-tRNAs and the nucleolus indicates that a major part of tRNA biogenesis also is compartmentalized in the nucleolus with rRNA synthesis and ribosome assembly (11). In fact, additional experiments have shown that tRNA genes are proximal to the nucleolus in yeast (34, 35, 106, 110). Finally, snRNAs have been detected in the Xenopus laevis nucleolus, in which some of the modifications occur (57, 113). These observed associations suggest the possibility that Pus7 activity is connected to the nucleolus and, more importantly, point to the possibility that the modification status provides a signal allowing the coregulation of the production and turnover of rRNAs, spliceosomes, and tRNAs. These observations also suggest a link to the production of ribosomal proteins, since the Pus7-catalyzed Ψ35 in U2 snRNA is important for the high-efficiency splicing of S. cerevisiae introns (79, 80, 111), which are preferentially located in ribosomal protein genes (32). Indeed, the interdependencies are consistent with the concept that the production of RNAP II and RNAP III transcripts is regulated in response to pre-rRNA synthesis via an extensive network of interactions (18, 56, 74, 95, 96).
Recent work has shown that the modification of tRNA is important for its stability and that specific degradation pathways are triggered by undermodification (1, 45, 47); thus, the hypomodification-triggered degradation of RNA could provide a means of coupling modification activity to a regulatory network. Interestingly, Pus7 is one of several factors directly linked to a rapid tRNA degradation pathway that acts on specific hypomodified tRNAs (1). Another RNA degradation pathway triggered by tRNA hypomodification also shows links to 5S rRNA synthesis in yeast (48). Furthermore, the interaction of 5S rRNA and ribosomal protein L5 is critical for 5S rRNA stability and incorporation into the pre-60S ribosomal particle (22, 23, 73, 77), and it has been proposed that the unstable nature of unbound 5S rRNA probably is connected to a lack of modification and that this is part of a network keeping pre-rRNA processing linked to RNA polymerase III activity (13). Interestingly, altered 5S rRNA stability also is observed when RNAP I activity is uncoupled genetically from attenuating its activity in response to stress (18, 56). Our elucidation of Pus7 as the factor involved in the pseudouridylation of yeast 5S rRNA will facilitate the further examination of the possible link between 5S rRNA stability and modification.
The specificity for multiple sites demonstrated by the factors studied here has important implications for assigning functions to enzymes or guide RNAs in other organisms. Pus7 shows multisite, multisubstrate specificity that is unparalleled. Furthermore, snR43 shows dual specificity; the same guide elements of snR43 are used to modify pre-rRNA at different sites that correspond to mature 5.8S and 25S rRNA sequences. The exact base-pairing potential seems to differ slightly, with the pairings for the 5.8S rRNA target being lengthier (Fig. 3A and B). The high degree of flexibility in the guide pairings (Fig. (Fig.4B)4B) makes the assigning of function difficult for organisms in which the exact modifications have not been mapped or in which the stringent verification of the guide RNA by effecting the disruption of its activity is not convenient. Likewise, the increasing number of targets for bacterium-like Ψ synthases makes assigning the exact extent of their functions more tentative. Moreover, for enzymes or guide RNAs for which the disruption of activities causes observed phenotypes, one has to be cautious in assigning roles to particular modifications when unidentified ones may yet exist.
We thank Maurille J. Fournier and Michael W. Gray, in whose laboratories this study was completed. We thank Jeremy Ho for assistance in the preparation of the snR43-deletion strain.
The manuscript is dedicated to the memory of James Ofengand, a pioneer in the field of Ψ and the synthases.
This work was supported by U.S. Public Health Service grant number GM19351 to Maurille J. Fournier.
Published ahead of print on 10 March 2008.