Report

Homologs of Small Nucleolar RNAs in Archaea

See allHide authors and affiliations

Science  21 Apr 2000:
Vol. 288, Issue 5465, pp. 517-522
DOI: 10.1126/science.288.5465.517

Abstract

In eukaryotes, dozens of posttranscriptional modifications are directed to specific nucleotides in ribosomal RNAs (rRNAs) by small nucleolar RNAs (snoRNAs). We identified homologs of snoRNA genes in both branches of the Archaea. Eighteen small sno-like RNAs (sRNAs) were cloned from the archaeon Sulfolobus acidocaldarius by coimmunoprecipitation with archaeal fibrillarin and NOP56, the homologs of eukaryotic snoRNA-associated proteins. We trained a probabilistic model on these sRNAs to search for more sRNAs in archaeal genomic sequences. Over 200 additional sRNAs were identified in seven archaeal genomes representing both the Crenarchaeota and the Euryarchaeota. snoRNA-based rRNA processing was therefore probably present in the last common ancestor of Archaea and Eukarya, predating the evolution of a morphologically distinct nucleolus.

Ribosome biogenesis in Eukarya occurs in the nucleolus. Several nucleolar proteins (NOPs), including fibrillarin, Nop56, and Nop58, and dozens of snoRNAs are involved in this process (1). The snoRNAs fall into two major classes: C/D box and H/ACA box RNAs. The C/D box snoRNAs are efficiently precipitated with antibodies against fibrillarin. Most C/D box snoRNAs target specific ribose methylations within rRNA, whereas most H/ACA box RNAs target specific conversions of uridine to pseudouridine within rRNA (2).

The general mechanism of C/D box snoRNA-targeted ribose methylation has been well established. Each snoRNA contains a 9- to 21-nucleotide (nt)–long sequence, located 5′ to the D or D′ box motif, that is complementary to an rRNA target sequence. Methylation is directed to the rRNA nucleotide that participates in the base pair 5 nt upstream from the start of the D or D′ box. It is likely that most, if not all, eukaryotic rRNA ribose methylations are guided by snoRNAs. In the yeastSaccharomyces cerevisiae, methylation guide snoRNAs have been assigned to all but four of the 55 rRNA ribose methylation sites (3).

SnoRNAs, which are apparently ubiquitous in Eukarya, have not been found in Bacteria or Archaea. However, the rRNA of the archaeonSulfolobus solfataricus (Sso) has been shown to contain 67 ribose methylation sites, a number similar to that found in eukaryotes (4). Even though Archaea are unicellular prokaryotic organisms that lack a nucleolus, their genomes encode homologs to the essential eukaryotic nucleolar proteins, fibrillarin and NOP56/58 (5, 6). On the basis of these observations, we decided to examine Archaea for the presence of sno-like RNAs (sRNAs).

To isolate sRNAs from the archaeon Sulfolobus acidocaldarius (Sac), we cloned the S. acidocaldarius homologs of the eukaryotic fibrillarin and NOP56/58 proteins, designated aFIB and aNOP56, using sequence information from a related species, S. solfataricus (7). The cloned genes were expressed in Escherichia coli, and the recombinant proteins were purified and used to raise polyclonal antibodies in rabbits. The two antibody preparations were each highly specific and recognize single polypeptides of the predicted size in total S. acidocaldarius cell extracts (Fig. 1A). The antibodies were used to monitor the size distribution of particles containing aFIB and aNOP56 in a glycerol gradient fractionation of partially purified cell lysate (Fig. 1A) (8). Both aFIB and aNOP56 sedimented as a large heterogeneous complex.

Figure 1

Glycerol gradient sedimentation of aFIB- and aNOP56-containing particles present in S. acidocaldarius cell-free extracts. A sonicated cell extract was precipitated by addition of 35% ammonium sulfate, redissolved in buffer (50 mM tris, pH 8), layered onto a 35-ml 10 to 30% glycerol gradient in the same buffer, and sedimented in an SW27 rotor (10°C, 17 K, 16 hours). Fractions (1.5 ml) were collected. (A) Aliquots of every second fraction between 2 and 20 were simultaneously analyzed by Western blotting for the presence of aFIB and aNOP56 with the two antibodies prepared against the recombinant proteins expressed and purified from E. coli. The positions of 30S and 50S ribosomal subunits in the gradient are indicated. In the control, the aFIB and aNOP56 antibodies were shown to be highly specific for single polypeptides of the expected size (27 kD and 47 kD, respectively) in S. acidocaldarius crude cell extract (right). (B) Aliquots from every other gradient fraction between 4 and 14 were immunoprecipitated with antibody to aFIB (8), and RNA was recovered by phenol extraction from the precipitates (P) and the supernatants (S). Only about 0.1% of the RNA in each fraction was coprecipitated with the antibody; the bulk of the RNA was retained in the supernatant. As a control (4C), an aliquot of fraction 4 was immunoprecipitated with preimmune serum. To visualize the precipitated RNAs, we pCp end-labeled aliquots (0.005% and 2.5% of the total RNAs recovered from the supernatant and pellets, respectively) with RNA ligase and displayed them on an 8% denaturing polyacrylamide gel. The positions of tRNA and sRNA are indicated on the left. The precipitated RNA recovered from fraction 5 was separated on an 8% denaturing polyacrylamide gel, recovered by electroelution, and used as a template for RT-PCR cloning (9). An aliquot of the RNA recovered after electroelution was end-labeled and displayed on an 8% denaturing polyacrylamide gel (right). (C) Aliquots from every other gradient fraction between 4 and 14 were immunoprecipitated with antibody to aNOP56. Other details are as described above, except that recovered RNAs from fractions 6 to 8 and 10 to 13 were pooled and used for cDNA cloning. An aliquot of the pooled RNA was end-labeled and displayed on an 8% denaturing polyacrylamide gel (right).

To detect RNAs that associate with aFIB- and aNOP56-containing complexes, we immunoprecipitated aliquots from gradient fractions with either antibody to aFIB or antibody to aNOP56. Total RNA was extracted with phenol from the supernatants and the pellets, and a portion from each was 3′ end-labeled with32P-cytidine-5′,3′-bis-phosphate (pCp) and displayed by denaturing polyacrylamide gel electrophoresis (Fig. 1, B and C). The most abundant RNAs that were coimmunoprecipitated appear as a family of bands ranging in length from about 50 to 70 nt. This size class of RNAs, which is substantially shorter than eukaryotic C/D box snoRNAs, was invisible when total cellular RNA was labeled with pCp. To obtain cDNA clones, we gel-purified the RNAs precipitated from fraction 5 with antibody to aFIB and from fractions 6 to 8 and 10 to 13 with antibody to aNOP56, ligated them to the oligonucleotide AO30, used them as template for reverse transcription polymerase chain reaction (RT-PCR), and cloned them (9).

A total of 104 clones from the two immunoprecipitated RNA pools were sequenced. From these, one or more representatives of 18 different sequences that exhibited features characteristic of eukaryotic C/D box snoRNAs were recovered (Table 1) (2). Other clones contained small fragments of S. acidocaldarius 16S, 23S, and 5S rRNAs. The snoRNA-like clones contained well-defined C and D box motifs located near their 5′ and 3′ ends, respectively, and recognizable internal C′ and D′ box motifs, giving the RNAs a dyad repeat structure characteristic of eukaryotic methylation guide snoRNAs (10).

Table 1

The sequences of S. acidocaldarius(Sac) cDNA clones are aligned with the C, D′, C′, and D boxes as anchors. Dashes are gaps in the alignment. The GenBank accession numbers for the sRNA sequences are AF195095 throughAF195112.

View this table:

Primer extension analysis was used to confirm the presence of sRNAs within total RNA extracted from S. acidocaldarius. Each sRNA primer was designed to overlap the D box motif, the adjacent guide region, and a portion of the C′ box motif. Extension products were obtained for sR1 to sR17 (clone sR18 was identified later and not tested); a subset of these is illustrated in Fig. 2. The lengths of the products for all sRNAs were within 2 nt of the 5′ ends of the cDNA, except for sR3, 4, 6, and 8, which were between 3 and 5 nt longer than the respective cDNA clones (11).

Figure 2

Detection and 5′ end mapping of sRNAs fromS. acidocaldarius and S. solfataricus. Primers specific for the D box guide region of Sac sR1 to sR17 were 5′ end-labeled with γ32P-ATP and polynucleotide kinase and used in extension reactions with total RNA (10 μg) isolated fromS. acidocaldarius as template. (A) The extension products obtained with Sac sR1 and sR2 specific oligonucleotide primers were run alongside a 33P-DNA sequence ladder generated with the same primers and Sac sR1 or sR2 cDNAs as template. (B) The extension products obtained with Sac sR8, sR14, sR3, sR5, sR6, sR16, and sR10 specific primers and run with the DNA sequence ladder generated withSac sR8 cDNA clone. The main sR8 extension product is 3 nt longer than the 5′ end of the sR8 cDNA clone. For each extension reaction, the major extension product (>) and the approximate positions of the 5′ terminal nucleotide in the corresponding cDNA clone (•) are indicated beside the lane. (C) The primer extension reaction was as in (A), except that the primer was specific to Sso sR1 and total RNA from S. solfataricus was used as template. The DNA ladder was generated with Sac sR1 primer and the Sac sR1 cDNA clone as template. The Sac and Sso primers are complementary to the same region but differ at two internal positions.

To find additional homologs of our cloned S. acidocaldariussRNA genes, we ran BLASTN on each cDNA clone against the nonredundant nucleotide database (12) and recovered two weak hits against sequences in other Sulfolobus species:Sac-sR3 had a hit near the Sulfolobus shibataetop6B topoisomerase II gene (score = 40.1 bits, expectation value = 0.038), and Sac-sR1 had a hit that partially overlapped with the S. solfataricus aspartate aminotransferase gene (score = 38.2 bits, expectation value = 0.15). Although these candidates contained canonical C and D boxes, their authenticity as true sRNAs remained questionable because of their low scores. We tested for the presence of the Sac-sR1 homolog by primer extension analysis using S. solfataricusRNA as a template. A product with a length similar to that ofSac-sR1 was detected (Fig. 2C) and was designatedSso-sR1. Primer extension products for cloned S. acidocaldarius RNAs sR1 to sR17 and the apparent S. solfataricus sR1 homolog demonstrate the existence of archaeal snoRNA-like C/D box sRNAs.

To determine if these sRNAs might guide ribose methylation as in eukaryotes, we examined the sRNAs for potential guide sequences by comparison with S. acidocaldarius rRNA (13). Regions complementary to rRNA and adjacent to the D or D′ boxes were identified for 14 of the sRNAs (Table 2). Using the D/D′ box plus 5 nt rule, we predicted the locations of potential ribose methyl modifications in rRNA and experimentally tested for some of these sites using the deoxyribonucleotide triphosphate (dNTP) concentration-dependent primer extension assay (3,14). In this assay, ribose 2′-O-methyl sites cause characteristic pauses that are displayed in the reverse transcriptase reactions at low but not at high dNTP concentrations. We identified characteristic pauses at six predicted sites of methylation in S. acidocaldarius rRNA (Table 2). Several examples are shown in Fig. 3. Both Sac-sR1 and Sso-sR1 were predicted to target methylation to position U52 in the respective 16S rRNAs; pause sites were detected at this position in both rRNAs. Two of the sRNAs, sR10 and sR14, exhibit strong complementarities to S. solfataricus tRNAs (Table 2). The target nucleotide for sR14 is C34, the anticodon “wobble” base, which is commonly ribose methylated in eukaryotes (15). Not all eukaryotic C/D box snoRNAs containing complementary regions participate in ribose methylation (i.e., U3 and U8), so methylation guide function should not be assumed for all archaeal C/D box sRNAs. Gene disruption systems for S. acidocaldarius and most other Archaea are currently not available; consequently, we were not able to verify loss of predicted methylation sites upon disruption of sRNA genes. However, our evidence suggests that many of these sRNAs function as guides for ribose methylation, as in eukaryotes.

Figure 3

Detection of 2′-O-ribose methylation sites in rRNA. Positions of ribose methylation in rRNA were detected with the dNTP concentration-dependent primer extension pause assay (3, 14). Total RNA from S. acidocaldarius (A to C), S. solfataricus (A), or M. jannaschii (D) was used as template. The sequence ladders were generated from either DNA or RNA templates with the same primers used in the pause reactions. The position of pausing is indicated on the right along with the position of the methylated nucleotide; the pause characteristically occurs 1 nt upstream of the modification. When a sequence ladder is generated from DNA template, as in (A), the pause may occur 1 nt upstream of the modification or directly at the modification. The sequence of the sRNA guide and the complementary rRNA target are shown below each panel; the site of methylation in rRNA is in the base pair (boxed) positioned 5 nt upstream of the start of the D or D′ box.

Table 2

Annotations of S. acidocaldarius sRNAs.

View this table:

We next asked whether sRNAs are found in other Archaea. We retrained a previously developed eukaryotic snoRNA search program with the verified S. acidocaldarius sRNA genes (3,16) and used it to screen the available archaeal genome sequences. We first searched the genome sequence of the closely related archaeon S. solfataricus (17). The program identified dozens of sRNA candidates, each of which had the potential to target a modification to a particular position in rRNA. We designed primers complementary to the 20 top-scoring candidate sRNAs and performed primer extensions on S. solfataricus total RNA to detect stable RNAs. Ten candidates (Sso sR1 to sR10), all ranking within the top 13 candidates by score, generated products of the anticipated size, 2 to 6 nt upstream of the predicted C box. An alignment of the 10 verified S. solfataricus sRNAs, plus three high-scoring, untested sRNA predictions (sR11 to sR13), is available (18). Six predicted target ribose methylation sites were assayed with the dNTP concentration-dependent primer extension assay (Fig. 3), and four showed reverse transcription pauses characteristic of ribose methylation (18). Three additional target site predictions are known to be modified at the homologous position in S. acidocaldarius 16S rRNA (13, 18).

Sulfolobus is a member of the Crenarchaeota, one of the two main phyla of Archaea; the other phylum, the Euryarchaeota, is evolutionarily distant. Complete genome sequences are available from archaeal species covering a wide range of genera, including both the Crenarchaea (Aeropyrum pernix) and the Euryarchaea (Methanococcus jannaschii, Archaeoglobus fulgidus, Methanobacterium thermoautotrophicum,Pyrococcus horikoshii, Pyrococcus abyssi, andPyrococcus furiosus) (5, 17). In searching these genomes for guide sRNAs, we found strong candidates in six of the seven species (18).

The searches of the M. jannaschii (Mja) andA. fulgidus (Afu) genomes gave eight and four strong sRNA hits, respectively; guide regions in most of these candidates exhibit complementarity to rRNA (18). The presence of all eight Mja sRNAs was confirmed by primer extension analysis on M. jannaschii total RNA (18). We attempted to verify seven of the ribose methylation sites predicted by the Mja sRNAs. Five sites showed concentration-dependent pauses indicative of ribose methylation, and the two other sites showed concentration-independent pauses, inconclusive for ribose methylation (18). An example pause site predicted by Mja-sR6 at position C2034 in 23S rRNA is shown (Fig. 3D). The D box guide region ofMja-sR8 predicts methylation of the anticodon wobble base for the intron-containing precursor of tRNA-Met. We did not test any tRNAs for ribose modifications, although the wobble base in tRNA-Met is known to be ribose-methylated within another hyperthermophilic crenarchaeon (19). The search of the Aeropyrum pernix (Ape) genome produced 23 candidate sRNAs (18). There were no strong sRNA hits in the genome ofM. thermoautotrophicum (20).

The genomes of three Pyrococcus species have been sequenced:P. horikoshii (Pho), P. furiosus(Pfu), and P. abyssi (Pab) (17). These related sequences enabled us to infer support for sRNA predictions using comparative sequence analysis. From separate genome searches followed by comparative analysis, we identified 57 groups of homologous Pyrococcus sRNA genes (21). Forty-seven groups were found in all three species, eight were found in only two species, and two were unique to single species. Examples of two of these groups, sR3 and sR4, are illustrated (Fig. 4), and the complete set is available online, as are the alignment, annotation, and genomic distribution of the candidate sRNAs found in P. horikoshii (18).

Figure 4

Guide region sequence similarity in archaeal sRNA. Three sets of sRNAs that are predicted to direct 2′-O-ribose methylation to homologous sites in 16S or 23S rRNA are aligned either over their entire length or over their D′ or D box guide regions. The guide complementarities to regions of 16S or 23S rRNA are shown above and below the guide regions; predicted sites for methylation within the rRNAs that are homologous (H) or nonhomologous (NH) are boxed. An archaeal consensus is included with each set of aligned sequences in bold; X in the consensus indicates the presence of compensatory nucleotide substitutions in the sRNA guide and rRNA target region. For other details, see text.

We asked whether predicted rRNA methylation sites occurred at homologous rRNA positions in different archaeal genera. We view our site predictions with caution, as the sRNA complementarities are short and few have been experimentally tested. Nonetheless, on the basis of an rRNA multiple alignment (22), a total of 19 predicted methylation sites were conserved between two or more genera. Figure 4A shows 16S Um52, a confirmed modification inSulfolobus, which we predict is guided by sR1 inSulfolobus and by sR4 in Pyrococcus. However,Sulfolobus sR1 and Pyrococcus sR4 also have dissimilar D′ associated guide sequences that are predicted to target methylation to nonhomologous positions (16S Um33 in S. solfataricus and 16S Am361 in Pyrococcus).Figure 4B shows that the predicted guide sequences for a site in 23S rRNA (Sac U2692, Ape U2714, andPho U2673) contain four separate nucleotide substitutions that are matched by compensatory substitutions in 23S rRNA, strong evidence that this sRNA/rRNA interaction is evolutionarily conserved. In nearly all cases, the intergenera sequence similarity between sRNAs that predict methylation at a homologous site is limited to the interacting guide region. In only one instance, we detected some end-to-end sequence similarity between two sRNAs from different archaeal genera: Pho-sR39 and Mja-sR6 (Fig. 4C). Moreover, the guide sequences can be either both in the same position (i.e., both D box associated) or in different positions (i.e., one D′ and the other D box associated; see Fig. 4B). Therefore, simple relationships of homologous sRNAs with homologous methylation sites are not obvious, and it remains uncertain whether sRNA guide sequences directing methylation to a homologous site are related to each other by common ancestry or by sequence convergence.

In general, all the archaeal sRNAs we identified are small, usually 50 to 60 nt in length, whereas human and yeast methylation guide snoRNAs average roughly 75 and 100 nt, respectively (3, 10). A much larger proportion of archaeal sRNAs appear to have the ability to guide methylation from both D′ and D boxes as “double guides.” On the basis of program predictions and comparative sequence analysis among Pyrococcus groups, we estimate that the majority of verified and putative archaeal sRNAs have two guide regions, whereas only 20% of human and yeast snoRNAs have been reported to be double guides (3, 10). Often, the predicted target sites of double-guide sRNAs are within the same RNA molecule, and often, they are closely linked. For example,Sso-sR1 appears to direct methylation with D′ and D box guides to positions U33 and U52 in 163S rRNA (Fig. 4A). This is in contrast to yeast snoRNA double guides, in which there is no apparent correlation between molecules targeted by the same snoRNA.

The number of sRNAs revealed by the search program seems to correlate with the optimum growth temperature of the organism:Pyrococcus species (95°C) have more than 50 putative sRNAs, whereas M. thermoautotrophicum (65°C) has no easily recognizable sRNAs. This may imply that a larger number of methylation modifications in rRNA might be required to fold or stabilize rRNA at high temperature (4) or that sRNAs are easier to recognize in hyperthermophiles because their gene features are more canonical.

In eukaryotes, snoRNAs do not act solely on rRNA. A number of cellular and viral RNAs transit through the nucleolus during maturation and at least one of these, the spliceosomal snRNA U6, is a substrate for snoRNA guide-directed methylation (23). Three cloned, verified Sac sRNAs (Table 2) do not appear to target any known stable RNAs (18), and several archaeal sRNAs exhibit complementarity to various tRNAs. Four of the sRNAs we identified (thePyrococcus sR40 genes and Afu sR3) reside within the intron of the genes encoding tRNA-Trp. Our program detected these putative intronic sRNAs because they appeared to be capable of targeting methylation to sites within rRNA (18). However, Daniels and co-workers (24) have independently identified these sRNAs and suggest that the D′ and D box guides are targeting methylations to positions C34 and C39 within the intron-containing precursor tRNA. These observations suggest that both ribosomal and nonribosomal RNAs may be substrates for sRNA guide-directed methylation in Archaea.

Thus, it appears that an RNA-based guide mechanism for directing specific RNA 2′-O-ribose methylations was an established feature in the common ancestor of Archaea and Eukarya (5). In Bacteria, there is a low abundance of 2′-O-methylation and pseudouridylation in rRNA, and neither a fibrillarin homolog nor C/D box sRNAs have been described. Nonetheless, the existence of sRNA-directed modifications in bacterial stable RNAs remains a possibility.

  • * Present address: Department of Genetics, M322, Stanford University School of Medicine, 300 Pasteur Drive, Stanford, CA 94305–5120, USA.

  • To whom correspondence should be addressed. E-mail: patrick.p.dennis{at}ubc.ca

REFERENCES AND NOTES

View Abstract

Navigate This Article