Genome-wide searching for pseudouridylation guide snoRNAs: analysis of the Saccharomyces cerevisiae genome

Abstract

One of the largest families of small RNAs in eukaryotes is the H/ACA small nucleolar RNAs (snoRNAs), most of which guide RNA pseudouridine formation. So far, an effective computational method specifically for identifying H/ACA snoRNA gene sequences has not been established. We have developed snoGPS, a program for computationally screening genomic sequences for H/ACA guide snoRNAs. The program implements a deterministic screening algorithm combined with a probabilistic model to score gene candidates. We report here the results of testing snoGPS on the budding yeast Saccharomyces cerevisiae. Six candidate snoRNAs were verified as novel RNA transcripts, and five of these were verified as guides for pseudouridine formation at specific sites in ribosomal RNA. We also predicted 14 new base-pairings between snoRNAs and known pseudouridine sites in S.cerevisiae rRNA, 12 of which were verified by gene disruption and loss of the cognate pseudouridine site. Our findings include the first prediction and verification of snoRNAs that guide pseudouridine modification at more than two sites. With this work, 41 of the 44 known pseudouridine modifications in S.cerevisiae rRNA have been linked with a verified snoRNA, providing the most complete accounting of the H/ACA snoRNAs that guide pseudouridylation in any species.

Figure 1

H/ACA model and snoGPS algorithm. (A) Schematic diagram of a consensus H/ACA snoRNA. A snoRNA in which both guide regions interact with a substrate RNA is shown. The classic H/ACA snoRNA sequence motifs are indicated, including left and right guide sequences, ‘H’ and ‘ACA’ boxes, 5′ and 3′ stems and the downstream U-rich region. The U-rich region is not part of the fully processed snoRNA. The model on which the ‘one-stem’ version of snoGPS is based consists of a single helix–bulge–helix stem with one pair of guide elements. (B) Schematic representation of the one-stem and two-stem snoGPS algorithms. Rectangles indicate length distribution tests. Rounded rectangles indicate nucleotide weight-matrix tests. For computational efficiency, tests are not necessarily computed in the order shown. The one-stem algorithm consists of tests 1–9. For the two-stem algorithm, snoGPS first searches for the hairpin structure containing the rRNA-guide sequences (tests 1–7). Then it searches both upstream (tests 10–13) and downstream (tests 14–17) for the second hairpin structure. To speed up the search, some tests have ranges on the allowable test results (shown to the right of the corresponding test). ‘minlen’ and ‘maxlen’ indicate the minimum and maximum allowed values for a length test. ‘Mismatch’ is the maximum allowed number of mismatches for a guide-region matching test and ‘min(match-mismatch)’ is the minimum excess of matches (Watson–Crick or G–U) over mismatches for a stem. If the score for a candidate is outside the allowed range, the sequence is rejected and the remaining tests are not executed.

Figure 2

Scores of randomized and background sequences. (A) Number of snoGPS hits per 12 Mb random genome sequence, when averaged over three 12 Mb random genomes and all 44 Ψ target sites. The graph indicates that a threshold score of 36 would be expected to include approximately two false positives/12 Mb. (B) mfold MFE-per-base scores for the known snoRNAs compared with those of the top 100 highest-scoring snoGPS hits found in one of the random sequence runs. Solid line is for known snoRNAs; dashed line for random snoGPS hits. Although there is considerable overlap between the two curves, the figure suggests that candidates with MFE-per-base scores greater than −0.15 kcal/mol/nt are probably false positives. (C) Percentage of sequences in the Not-Feature part of the S.cerevisiae genome with putative homologs in other Saccharomyces genomes. The dashed line shows that three to five homologs in other Saccharomyces genomes were found essentially for all known snoRNAs. The solid line is for an average of one-thousand 200 nt sequences randomly selected from the S.cerevisiae Not-Feature genome. The data suggest that Not-Feature sequences with fewer than three homologs outside S.cerevisiae are probably not H/ACA snoRNAs.

Figure 3

Phylogenetic comparison is consistent with the presumed secondary structure of newly discovered snoRNAs. The proposed secondary structures of the guide regions of the newly discovered S.cerevisiae (A) snR80 and (B) snR82 snoRNAs are shown along with the homologous sequences from several other Saccharomyces species. Nucleotides differing from S.cerevisiae are encircled. Those changes that presumably disrupt base pairing within the snoRNA or between the snoRNA and the rRNA are also shaded. The rRNA sequences are shown in light gray. The box ACA element at the 3′ end of each RNA is underlined. The Ψ at LSU-2350 is shown for the snR82 guide region. For clarity, only one of the two H/ACA hairpins for each of the snoRNA homologs is shown.

Figure 4

Experimental verification of predicted guide RNAs by northern hybridizations and primer extensions. (A) Northern-blot analysis of snoGPS snoRNA predictions. Total RNA was isolated from a wild-type (WT) and a strain in which snoGPS predicted snoRNA regions were chromosomally deleted. snR64 served as a loading control. (B) Primer extension 5′ end mapping of snR80, snR81 and snR85 from WT RNA relative to a 10 bp marker. (C) Primer extension 5′ end mapping of snR83 from WT RNA relative to a 10 bp marker. End-mapping of snR82, snR83 and snR84 has been reported previously (40). For snR82 and snR84, our results are very similar to those of (40) (data not shown). For snR83 our results show a strong stop signal corresponding to coordinates in reasonable agreement with those in the corrected version of (40) (see http://www.genetics.wustl.edu/eddy/publications/#McCutcheonEddy03).

Figure 5

The results of gene disruption experiments. Experimental verification of the loss of specific rRNA pseudouridines in (A) strains with candidate loci and an unconfirmed snoRNA gene disrupted and (B) strains with experimentally identified-snoRNA genes disrupted. Following treatment of RNA from several strains with or without CMC, primer extension was performed with various primers to check positions in the 18S (SSU) or 25S (LSU) subunit ribosomal RNAs. ‘WT’ represents RNA from yeast where the locus under question has not been disrupted.

Figure 5

The results of gene disruption experiments. Experimental verification of the loss of specific rRNA pseudouridines in (A) strains with candidate loci and an unconfirmed snoRNA gene disrupted and (B) strains with experimentally identified-snoRNA genes disrupted. Following treatment of RNA from several strains with or without CMC, primer extension was performed with various primers to check positions in the 18S (SSU) or 25S (LSU) subunit ribosomal RNAs. ‘WT’ represents RNA from yeast where the locus under question has not been disrupted.

Figure 6

Putative base pairings between Ψ target regions and known snoRNAs. Pairings are shown for (A) snR49 and snR161 and experimentally verified Ψ targets. (B) The putative snR3 and snR11 guide snoRNAs and LSU-2128; snR3 was experimentally shown to be the actual guide RNA. The association predicted for snR11, though plausible, was demonstrated experimentally to be incorrect. (C) The newly identified Ψ guide snoRNAs. (D) Potentially redundant guide snoRNAs.