In this report, we describe a family of RNAs that are derived from the snR13
locus of S. cerevisiae
. The most abundant of these, snR13F, is a TMG-capped, C/D box-containing snoRNA that most likely functions in rRNA maturation in the nucleolus (23
). Three overlapping RNAs, snR13R, snR13F, and snR13T, were detected in wild-type cells by Northern blotting. The steady-state levels of these RNAs changed when strains carrying the temperature-sensitive mutation sen1-1
were shifted to the restrictive temperature. snR13R and snR13T increased in abundance with similar kinetics. The level of snR13F declined in both wild-type and sen1-1
cells at the restrictive temperature, obscuring any potential change due to the sen1-1
mutation itself. The reason for this decline at an elevated temperature is unknown.
When in vivo-labeled, nascent snR13 RNA was analyzed in both wild-type and temperature-shifted sen1-1 cells, similar changes were observed except that the amount of snR13F RNA synthesized was severely reduced in sen1-1 cells shifted to the restrictive temperature. This result indicates that the rate of synthesis is significantly reduced. We identified in wild-type and temperature-shifted sen1-1 cells several additional nascent RNAs that were not detected on Northern blots. Two long forms of snR13, called RL and RS, were detected. RL is probably equivalent to the snR13R RNAs detected on Northern blots, because it responds to inactivation of Sen1p in the same way as snR13R. Similar amounts of RS were detected at the permissive and restrictive temperature. The relationship of RS to RL is not yet known. Our results indicate that a function provided by Sen1p, which is most likely an RNA unwinding activity, is required for the synthesis of snR13F. This is consistent with an interpretation in which snR13F is derived from RL by a posttranscriptional pathway leading to 3′ end formation.
We used RNase H mapping and primer extension to characterize the three stable RNAs that accumulate on Northern blots. These RNAs are distinguished by a 5′ truncation, a 3′ extension, or both. snR13R is a mixed population of RNAs that share in common a ~1,300-nucleotide extension of the 3′ end of snR13F. RNase H mapping indicates that R form RNAs contain two discrete 5′ ends and two discrete 3′ ends. Depending on the distribution of 5′ and 3′ ends, there are two to four species of snR13R. Using primer extension, we showed that one of the 5′ ends of snR13R is identical to the 5′ end of snR13F (residue A1 in Fig. and ). Since Midwestern blotting and TLC analysis indicate that snR13F contains a TMG cap, it seems likely that a subpopulation of snR13R may also contain a TMG cap. We could not detect a TMG-capped R form on Midwestern blots, but this could be because of low abundance, poor transfer to nylon membranes, or comigration with other TMG-capped RNAs.
The changes in synthetic rates and levels of accumulation of the snR13 RNAs result from a requirement of Sen1p to both stabilize the 5′ terminus and to promote formation of the 3′ terminus from a 3′-extended RNA. The truncated 5′ end found among snR13R RNAs begins at nucleotide U17 and is therefore missing the first 16 nucleotides found in snR13F RNA. snR13T also begins at U17. snR13T RNA was detected on Northern blots but not on Midwestern blots and is therefore not likely to contain a TMG cap. Based on these findings, we propose that snR13F, snR13T, and the 5′-truncated version of snR13R are all derived from a common monomethyl-G-capped primary transcript that becomes hypermethylated to form a single TMG-capped RNA.
The 5′ truncation of snR13R and snR13T may result from exonucleolytic degradation or endonucleolytic cleavage of a single primary transcript that begins at nucleotide A1, producing a capless, 5′-truncated RNA. If the two 5′ ends were derived from two primary transcripts initiating at two different transcriptional start sites, all of the RNAs might be expected to have a TMG cap which is synthesized in conjunction with transcription by RNA polymerase II. The lack of a TMG cap in snR13T is inconsistent with this model. Our interpretation of the data implies that Sen1p is required for maintenance of the 5′ end. In absence of Sen1p function, the 5′ end is removed up to a position approaching the location of the C/D base-paired stem.
Primer extension stops were detected at positions U15 and U104. Both stops map within the sequence C2U4 that is repeated three times in snR13F and snR13T and once more in snR13R. The truncated 5′ end beginning at U17 in a subpopulation of snR13R and in snR13T maps within C2U4 repeat 1. The 3′ ends of snR13F and snR13T are located within C2U4 repeat 3. Because of the apparent significance of C2U4 repeats, we analyzed a substitution of CA2U3 for C2U4 repeat 1, which contains the U15 and U17 primer extension stops, and substitutions of CA2U3 and A6 for C2U4 repeat 3, which contains the 3′ terminus of the snR13F and snR13T RNAs (Fig. shows the locations of these repeats).
If C2U4 sequence repeats 1 and 3 were to function as signals for transcription initiation and termination, respectively, we would expect that mutations that impair the function of these sequences might cause increased accumulation of RNAs containing extended 5′ and/or 3′ ends. Our results show that this is not the case. The mutation in repeat 1 caused decreased accumulation of snR13F RNA without a concomitant increase in 5′-truncated snR13T and 3′-extended snR13RR. Both of the mutations in repeat 3 failed to cause detectable increases in the accumulation of snR13T and snR13R RNAs. Instead, these mutations resulted in the accumulation of a pair of novel RNAs that differ markedly from R forms. The novel doublet RNAs failed to migrate according to size and may not contain a significant 3′ extension, indicating a potential level of structural complexity not expected for simple 3′-extended transcripts. These results make it unlikely that C2U4 repeats function in transcription initiation or termination. Instead, they are likely to play a role in maturation.
The predicted secondary structure of snR13F (Fig. ) suggests that C2
repeats 1 and 3 may reside near each other in three-dimensional space. The two repeats flank a base-paired region that overlaps with the C and D boxes. In our studies we detected 5′ and 3′ ends within repeats 1 and 3, respectively, but failed to detect RNAs with an endpoint in repeat 2 or in a new repeat created by the snR13-RNEW
mutation. These results indicate that the presence of a C2
repeat is not sufficient to specify a site for end formation. According to this interpretation, C2
repeats may play a role in maturation by providing a preferred context to the neighboring base-paired C/D box. Additional evidence supporting this view comes from a comparison of sequences surrounding C and D boxes in other snoRNAs. When the sequences of other snoRNAs were examined, we found that several contained either an exact C2
match or a close match differing by a single nucleotide (Table ). However, some snoRNAs such as snR10 and snR31 do not possess any sequences resembling C2
). These observations indicate that C2
is most likely a preferred sequence but not one that is globally required in all snoRNAs.
TABLE 3 Yeast RNAs with snR13-like 3′sequences
The phenotypes conferred by sen1-1
and by mutations in C2
repeats 1 and 3 resemble the phenotypes of C/D box mutations described by others (6
). For example, deletion of a short stem that immediately flanks the C and D boxes in human U14 snoRNA results in maturation defects. Also, Xenopus
U16 and U18 RNAs require intact C and D boxes for end maturation. The 3′ end of Xenopus
U16 is formed within a sequence context identical to that found for snR13 and proceeds by a reaction requiring protein factors that have been partially purified. We speculate that one of the factors could be Sen1p in yeast and its homologue in higher eukaryotes.
We tested these ideas further by analyzing double mutants carrying sen1-1 and each of the mutations in C2U4 repeats 1 and 3. When mutations in repeat 1 were combined in the same haploid strain with the sen1-1 mutation, the abundance of snR13F was significantly reduced at the restrictive temperature. We have not yet determined whether this is due to a reduced rate of formation of snR13F or to an increased rate of decay of the mature RNA. When mutations in repeat 3 were combined with the sen1-1 mutation, R forms accumulated at the expense of the novel doublet RNAs upon a shift from permissive to restrictive temperature. This finding indicates that an epistatic relationship exists between Sen1p and C2U4 repeat 3. Based on this finding, we propose that Sen1p is required prior to repeat 3. Furthermore, the epistatic relationship suggests a connection between Sen1p, C2U4 repeat 3, and the C/D box-fibrillarin snoRNP complex. Repeat 3 may provide a good sequence context that influences the binding of C/D box proteins that both protect snR13F from degradation and promote 3′ end maturation. Sen1p might be recruited into the snoRNP complex to facilitate 3′ end formation. Alternatively, Sen1p might promote the assembly of a snoRNP complex.
There are two reasons for suspecting that Sen1p may play a direct role in the formation of RNA termini and may be recruited into or assist in the assembly of snoRNP complexes. The changes in synthetic rates and levels of accumulation caused by loss of Sen1p function commence immediately upon temperature shift, which would be expected if Sen1p plays a direct role in snoRNA maturation. By contrast, the accumulation of tRNA precursors observed in strains carrying sen1-1
does not correlate with growth temperature, indicating that an indirect effect of Sen1p on tRNA splicing is likely (42
). Most importantly, snR13F coimmunoprecipitates with wild-type Sen1p, suggesting that an RNA-protein physical interaction is likely (39
). It remains to be determined whether snR13R and snR13T RNAs are also present in precipitates because their abundance in wild-type strains is very low.
Direct evidence that the novel doublet RNAs represent trapped intermediates in a maturation pathway is still lacking. We were unable to detect newly synthesized RNAs corresponding to the novel doublet RNAs in wild-type cells, but they may still exist as extremely labile RNAs. If they correspond to intermediates in 3′ end formation, their existence suggests that 3′ end formation involves more than one step. We are currently investigating the structures of these RNAs to determine why they exhibit anomalous migration on gels and whether they differ in length at the 5′ end in a manner similar to snR13F, snR13T, and the subpopulations of the snR13R RNAs.
Inactivation of Sen1p caused time-dependent changes in the accumulation of the snoRNAs snR10, snR11, snR13, snR31, snR40, and snR42, which include both C/D box and ACA snoRNAs, and the snRNA U5 (28
). In addition, other RNAs have recently been shown to exhibit altered accumulation upon inactivation of Sen1p, including RNAs of unexpected size containing U5, snR40, and snR45 sequences (39
). Since loss of Sen1p function affects multiple RNAs and since it is recruited into RNP complexes containing as many as 20 snoRNAs (39
), we suspect that Sen1p plays a role in the maturation of numerous snoRNAs.
mutation causes pleiotropic defects in RNA metabolism, including effects on tRNA and rRNA processing (39
). Also, temperature-shifted sen1-1
cells exhibit altered localization of nucleolar proteins such as fibrillarin (38
). Since changes in rRNA processing and nucleolar organization do not appear until several hours after temperature shift of sen1-1
cells but perturbations of snR13 synthesis commence immediately, it seems likely that global perturbations in snoRNA synthesis are primary defects in sen1-1
cells. The nucleolar and rRNA defects probably develop as a secondary consequence because rRNA maturation requires snoRNAs. The primary defect responsible for the accumulation of intron-containing pre-tRNAs in sen1-1
cells is still unknown. Our studies raise the possibility that a small RNA whose synthesis depends on Sen1p could be required for tRNA synthesis.
snR13R RNAs extend into and include all of the downstream ORF YDR472w. At present, we are unable to interpret the significance of this finding. Recent evidence indicates that YDR472w produces a polyadenylated mRNA that is likely to be translated into a protein product (21a
). Studies are in progress to determine whether the snR13R RNAs encode a protein in addition to producing a snoRNA.