Results from higher eukaryotic systems indicate that the 5′ ends of many snoRNA species are generated by 5′→3′ exonuclease activities (7
). Two 5′→3′ exonucleases, Xrn1p and Rat1p, have been identified in S. cerevisiae
, and we therefore tested whether mutants defective in these activities are also defective in the 5′ processing of pre-snoRNA species. The data presented here establish that Xrn1p and Rat1p play roles in the formation of the 5′ ends of the yeast snoRNAs, snR190, U14, U18, and U24. In each case, the rat1-1
mutation led to some accumulation of 5′-extended species at 37°C, but this was stronger in the xrn1-Δ rat1-1
double-mutant strain, whereas the xrn1
-Δ single mutation alone had little effect. After 2 h at the nonpermissive temperature, depletion of the mature snoRNAs was observed in the rat1-1 xrn1
-Δ strains, indicating that the major biosynthetic pathway was inhibited. As rat1-1
strains rapidly cease growth at 37°C (2
), stronger depletion of the mature snoRNAs was not observed at later time points. The pre-rRNA spacer fragments and the 5′-extended snoRNA and 5.8S rRNA species were, however, present at relatively low levels compared to the mature rRNA and snoRNAs. The xrn1
-Δ allele is a gene disruption construct (14
), whereas the rat1-1
allele is a temperature-sensitive point mutation (2
). It is not clear whether Rat1p retains some residual processing activity in the rat1-1
strain or whether the accumulated RNAs were degraded by another pathway. In the rat1-1
mutant strain, the nuclear poly(A) signal is lost after prolonged incubation at the nonpermissive temperature (2
), suggesting that there is some residual activity in the mutant. A complex of 3′→5′ exonucleases processes the 3′ end of the 5.8S rRNA (30
) and degrades the excised pre-rRNA spacer 5′ to site A0
). This complex may also contribute to turnover of the RNAs that accumulate in the 5′→3′ exonuclease mutants.
Northern and primer extension data indicate that snR190 and U14 are synthesized from a common, dicistronic transcript. In the rat1-1
and xrn1-Δ rat1-1
strains, ladders of 5′-extended snoRNAs that extend to positions −42 for snR190 and −55 for U14 were detected. These sites do not have any obvious homology to each other or to consensus snoRNA promoter sequences. Moreover, U14 position −55 lies only 12 nt 3′ to the mature snR190 region, making it unlikely to be a transcription start site. We propose that snR190 position −42 and U14 position −55 represent intermediate sites in the processing of larger pre-snoRNA species. These could be the products of either endonuclease or 5′→3′ exonuclease digestion. However, exonuclease digestion would presumably have to involve an exonuclease(s) other than Xrn1p and Rat1p, and other 5′→3′ exonuclease have not been identified in yeast extracts (19
). Moreover, the 5′ cap structure would be expected to confer protection against exonucleases, and an endonuclease activity is more probable. The furthest-upstream primer extension stop that we detected lies 302 nt 5′ to snR190. It has not been established whether this represents the transcription start site or is a further upstream processing site; however, the 5′ end of the next open reading frame lies only 190 nt upstream of this site, making this likely to be the start site.
U18 and U24 are synthesized from the introns of host genes that also encode mRNAs. U24 can be synthesized only from the debranched intron lariat; the snoRNA was found almost entirely in circular form in a mutant which lacks intron-debranching activity. This strongly indicates that both 5′ and 3′ processing of the pre-snoRNA are exclusively exonucleolytic. Moreover, little if any processing of U24 can occur on the unspliced pre-mRNA. In the case of U18, synthesis of the mature snoRNA was reduced to ~30% of the wild-type level in the debranching mutant, indicating that the major processing pathway is also via exonuclease digestion. Residual processing might be due to endonuclease cleavage of the intron lariat or exonuclease digestion of the unspliced pre-mRNA. For both U18 and U24, pre-snoRNAs that were 5′ extended to the intron 5′ splice site in the rat1-1 and xrn1-Δ rat1-1 strains accumulated, indicating that these are the 5′→3′ exonucleases responsible for processing the pre-snoRNAs. Primer extension specifically on pre-U24, using a 3′-flanking oligonucleotide, failed to detect the mature 5′ end of the snoRNA in the rat1-1 strains at 37°C, demonstrating the inhibition of 5′ processing.
In each case the accumulation of 5′-extended snoRNA species was much stronger in the rat1-1
strain than in the xrn1
-Δ strain, indicating that Rat1p is the major pre-snoRNA-processing activity in wild-type cells. Since Rat1p functions in the nucleus, the presumed site of pre-snoRNA processing, while Xrn1p functions in the cytoplasm (17
), it may be that the processing activity normally resides only in Rat1p, with Xrn1p functioning to process the accumulated pre-snoRNAs in the rat1-1
Together, the data suggest the models shown in Fig. . We envisage that snR190 and U14 are synthesized from a dicistronic pre-snoRNA species which extends from a position 302 nt 5′ to snR190 to beyond the 3′ end of U14 (Fig. A). This is processed, probably endonucleolytically, at positions 42 nt 5′ to snR190 and 55 nt 5′ to U14, within the intergenic spacer region. These processing reactions are followed by 5′ and 3′ trimming reactions which generate the mature snoRNAs. In contrast, U24 (Fig. B) is processed from the excised pre-mRNA intron. In wild-type cells processing is probably exonucleolytic from the debranched intron lariat.
FIG. 7 (A) Model for the processing of pre-snR190 and pre-U14. The coding sequences of snR190 and U14 lie in the same orientation in the genome and are separated by only 67 nt (46). We propose that they are synthesized from a common precursor which extends from (more ...)
In general, the host genes for vertebrate snoRNAs encode protein products that have some relationship to ribosome synthesis or function. This coexpression may facilitate the coordinated synthesis of the protein and snoRNA products. The data reported here extend the interaction between the synthesis of the snoRNAs and the function of the nucleolus by demonstrating that the snoRNAs and pre-rRNAs are processed by common components. It is possible that as the snoRNAs developed, they simply made use of whatever processing machinery was available. Alternatively, the use of common components might have been selected because of the obvious possibilities that it offered for coregulation of the synthesis of the rRNAs and snoRNAs. Such coregulation might indeed be the reason that so many snoRNAs, but no other known small RNA species, are synthesized by such excision mechanisms.
All studies on the in vitro processing of vertebrate snoRNAs have implicated 5′→3′ exonuclease activities in formation of the 5′ ends of the snoRNAs (7
; reviewed in references 20
). In no case have the nucleases yet been identified, but we strongly predict that, at least in some cases, these activities will involve the vertebrate homologs of Rat1p and Xrn1p (5