|Home | About | Journals | Submit | Contact Us | Français|
Expression of dsRNA complementary to small nucleolar RNAs (snoRNAs) in Trypanosoma brucei results in snoRNA silencing, termed snoRNAi. Here, we demonstrate that snoRNAi requires the nuclear TbDCL2 protein, but not TbDCL1, which is involved in RNA interference (RNAi) in the cytoplasm. snoRNAi depends on Argonaute1 (Slicer), and on TbDCL2, suggesting that snoRNA dicing and slicing takes place in the nucleus, and further suggesting that AGO1 is active in nuclear silencing. snoRNAi was next utilized to elucidate the function of an abundant snoRNA, TB11Cs2C2 (92nt), present in a cluster together with the spliced leader associated RNA (SLA1) and snR30, which are both H/ACA RNAs with special nuclear functions. Using AMT-UV cross-linking and RNaseH cleavage, we provide evidence for the interaction of TB11Cs2C2 with the small rRNAs, srRNA-2 and srRNA-6, which are part of the large subunit (LSU) rRNA. snoRNAi of TB11Cs2C2 resulted in defects in generating srRNA-2 and LSUβ rRNA. This is the first snoRNA described so far to engage in trypanosome-specific processing events.
RNA interference (RNAi) in Trypanosoma brucei was one of the first dsRNA-mediated silencing systems described in eukaryotes (1); since its discovery, RNAi has become one of the most useful tools for elucidating the function of protein coding genes in these organisms. In T. brucei, the natural RNAi pathway down-regulates transcripts from the retroposons, SLACS and INGI (2–4), and from a family of 147bp chromosomal internal repeats, termed CIR 147 (5), which are present in a subset of putative centromeric regions (6). The nature of the target transcripts suggests that one major function of RNAi in T. brucei is to promote genome stability, though RNAi may also have a role in chromosome segregation (2,7,8).
The ‘classical’ RNAi pathway, which is triggered by long dsRNA, is mediated by two key enzymes, Dicer, an enzyme of the RNase III family, and Slicer, a member of the Argonaute (AGO) protein family with an RNase H-type fold (9). Trypanosomes, like yeast, possess a single Argonaute (AGO1) homolog (2,10), but, unlike yeast, express two Dicer-like (DCL) proteins, TbDCL1 (3) and TbDCL2 (3,5). TbDCL1 is cytoplasmic and mediates the cytoplasmic RNAi response, which is harnessed in experimental RNAi to induce mRNA degradation. In contrast, TbDCL2, which is mostly nuclear (5), is central for dicing both retroposon and repeat-derived dsRNA and provides the first line of defense against these potentially dangerous molecules in the nucleus. On the other hand, both TbDCL1 and TbDCL2 contribute to the generation of retroposon-derived small interfering RNAs (siRNAs), suggesting that the two DCL proteins have both distinct and shared functions.
The existence of a nuclear phase of RNAi was also suggested by our previous studies demonstrating downregulation of small nucleolar RNAs (snoRNAs), triggered by expression of anti-sense RNA from an episomal vector in the monogenetic trypanosomatid, Leptomonas collosoma. This process was termed snoRNAi (11). We showed that snoRNAi in L. collosoma targets the mature snoRNA, but not its precursor, and we further provided evidence for the accumulation of siRNAs in snoRNAi-silenced cells. In T. brucei snoRNAi was achieved by expressing snoRNA information using a tetracycline inducible vector equipped with opposing T7 RNA polymerase (pol) promoters (11). Since the discovery of snoRNAi in trypanosomatids, the efficient silencing by siRNAs of telomerase RNA and 7SK RNA, two nuclear small RNAs, was reported, suggesting that nuclear RNAs can be down-regulated by RNAi in mammals, as well (12).
snoRNAs are one of the most abundant families of non-coding RNAs in eukaryotic genomes. Eukaryotic rRNA undergo both processing and extensive covalent modification that require the function of snoRNAs (13–16). The modifications include methylation of the 2′-hydroxyl residue (2′-O-methylation, Nm) (17) and isomerization of the uracil to pseudouridine (18,19). The snoRNAs guide these modifications following site-specific base-pairing with their targets. The snoRNAs that guide these modifications are designated by their specific sequence motifs: C/D boxed RNAs guide Nm, and H/ACA RNAs guide pseudouridylation (17–19). The C/D box RNAs, which are most relevant for this study, have 10–22nt sequences of perfect complementarity to the sequences within the mature rRNA (20).
Recent genomic and functional studies revealed a very rich repertoire of snoRNAs in trypanosomes. Initial studies identified 21 clusters encoding 57C/D snoRNAs and 34H/ACA-like RNAs (21). Additional C/D snoRNAs were identified after partially mapping Nms on rRNA (22). An algorithm designed to identify the trypanosome single hairpin H/ACA-like RNA species carrying an AGA box at the 3′-end, identified additional H/ACA RNAs in T. brucei (23). At present, 46H/ACA have been identified in T. brucei that have homologs in the other trypanosomatid species (24).
rRNA processing in trypanosomes differs from maturation in most eukaryotes. The small subunit (SSU) rRNA in trypanosomes is the largest known so far, and the large subunit (LSU) rRNA is processed into six fragments, two large ones termed LSUα and LSUβ rRNAs, and four small fragments, termed the small rRNA (srRNA) (25–27). The only trypanosome snoRNAs involved in rRNA maturation identified to date are U3, snR30 and MRP RNAs (22,28,29). Several of the snoRNAs shown to function in rRNA processing in other eukaryotes, including U22, U8, U14, have not been identified to date in trypanosomes (20).
In this study, we examined the mechanism of snoRNAi in T. brucei, and found that silencing of the TB11Cs2C2 snoRNA requires TbDCL2, since co-silencing of TB11Cs2C2 and TbDCL2 nearly abolished snoRNAi. No effect on snoRNAi was observed in cells expressing a dominant negative TbDCL1 allele, previously shown to inhibit the cytoplasmic RNAi response (3). snoRNAi was also compromised in cells silenced for AGO1, suggesting that the snoRNAi requires the function of both TbAGO1 and TbDCL2. snoRNAi was then used to investigate the function of TB11Cs2C2 snoRNA, one of the most abundant snoRNA forms in trypanosomes, and previously implicated in methylation of LSU rRNA (30). Here, we demonstrate the interaction between TB11Cs2C2 and its targets using in vivo UV-induced AMT cross-linking. SnoRNAi of TB11Cs2C2 suggests that this RNA is involved in rRNA processing of LSUβ to release the small rRNA fragments, srRNA-2 and -6. This is the first snoRNA described to date that mediates trypanosome-specific rRNA processing events. The function of the neighboring TB11CsC1 snoRNA was also elucidated, and this snoRNA is implicated in processing of SSU rRNA. This study suggests that snoRNAi operates in the nucleus utilizing TbDCL2 and TbAGO1, and this process can be harnessed to elucidate the function of nuclear/nucleolar non-coding RNAs.
All the oligonucleotides used in this study are listed in Supplementary Figure S1.
Procyclic T. brucei strain 29–13 which carries integrated genes for T7 polymerase and the tetracycline repressor was grown in SDM-79 medium supplemented with 10% foetal calf serum in the presence of 50µg/ml hygromycin and 15µg/ml G418. The T7 RNA pol opposing silencing constructs were prepared as previously described (31) using oligonucleotides listed in Supplementary Data S1. Cells were transfected and clonal populations were obtained in microtiter plates, as previously described (32).
Total RNA was prepared with TRI-Reagent (Sigma), and 20μg/lane were fractionated on a 1.2% agarose, 2.2M formaldehyde gel. The RNA was visualized with ethidium bromide. The TbDCL2 and SmD3 mRNAs were detected using a randomly labeled probe (Random Primer DNA Labeling Mix, Biological Industries Ltd.). For analyzing snoRNAs and siRNAs, total RNA (10µg) was fractionated on a 10% polyacrylamide gel containing 7M urea. The RNA was transferred to a nylon membrane (Hybond; Amersham Biosciences), and probed with an anti-sense RNA probe specific to the gene.
RNA probes were synthesized on a PCR template generated using primers described in Supplementary Table S1, and containing the T7 polymerase promoter sequence.
RNA was prepared from T. brucei cells using the TRI-Reagent (Sigma). Primer extension analysis was performed as described (33,34) using 5′-end-labeled oligonucleotides specific to each target RNA. The extension products were analyzed on a 6% polyacrylamide/7M urea gel and visualized by autoradiography. Total RNA was prepared and analyzed by the RNase protection assay, as previously described (32). The RNase protection assay products were separated on 6% polyacrylamide/7M urea gels and visualized by autoradiography. To calculate differences in the level of extension products, the films were subjected to densitometric analysis using ImageJ.
Cross-linking was performed essentially as described in (35). Briefly, T. brucei cells were harvested at 1×107cells/ml and washed twice with PBS. Cells (~109) were concentrated and incubated on ice. 4′-Aminomethyl–trioxsalen hydrochloride (AMT) (Sigma) was added to the cells at a concentration of 0.2mg/ml. Cells treated with AMT were kept on ice and irradiated using a UV lamp at 365nm at a light intensity of 10µW/cm2 for 30min. Next, the cells were washed once with PBS and deproteinized by digestion with proteinase K (Roche) (200µg/ml in 1% SDS for 60min). RNA was prepared using TRI-Reagent (Sigma).
First, 250μg of RNA extracted from UV treated cells (+UV) and untreated cells (–UV), in hybridization buffer (20mM HEPES pH 8, 5mM MgCl2, 300mM KCl, 0.01% NP40, 1mM DTT) was heated for 2min at 80°C and then chilled on ice. Oligonucleotides linked to biotin (Figure S1) (8μg) were added, and the solution was incubated overnight at room temperature. Next, 50μl of blocked Neutravidin Sepharose beads (Pierce) were added, and the reaction was incubated for 2h at 4°C. Blocking of the beads was performed in 1ml blocking buffer [700μl DEPC treated water, 200μl WB100 (20mM HEPES pH 8, 10mM MgCl2, 100mM KCl, 0.01% NP40 (v/v), 1mM DTT), 50μl BSA (10mg/ml), 10μl Glycogen (20mg/ml), 40μl tRNA (11mg/ml)]. After binding of the RNA to the oligonucleotide-bound beads, the beads were washed five times in WB400 buffer [20mM HEPES pH 8, 10mM MgCl2, 400mM KCl, 0.01% NP40(v/v), 1mM DTT]. RNA was eluted from the beads using the TRI-Reagent (Sigma) and analyzed by primer extension (For TB11Cs2C2) or by RT–PCR (for TB11Cs2C1).
RNase H cleavage of rRNA was performed by annealing 200pmol of oligonucleotides as indicated in the figures (see Supplementary Figure S1 for specific primer sequences) with 20µg of total RNA at 65°C for 10min in 75mM KCl, 50mM Tris–HCl (pH 8.3), 3mM MgCl2, 10mM DTT. The samples were then placed on ice and two units of RNase H (NEB) were added. The samples were further incubated at 37°C for 60min, phenol/chloroform extracted, and ethanol precipitated. Samples were then subjected to splint labeling. For splint labeling, the RNA (10–20μg) was mixed with 150pmol of oligonucleotide (see Supplementary Data S1 for primer sequence) and heated for 2min at 85°C in 50mM Tris–HCl (pH 7.8), 10mM MgCl2 and 1mM DTT. The annealing reaction was quenched on ice for 30min; 50μCi of [α-32P]-dCTP (3000Ci/mmol) and five units of T7 DNA polymerase (Sequenase V. 2.0, USB) were then added, and the reaction was incubated for 1h at 37°C. RNA was separated on a 6% polyacrylamide–7M urea gel.
The RNA was treated with the ‘DNase-free’ reagent (Ambion) according to the manufacturer’s protocol for 30min to remove DNA contamination. Reverse transcription was performed by random priming (Reverse transcription system, Promega). The samples were heated for 5min at 70°C, followed by chilling on ice for 5min. Next, one unit of AMV-reverse transcriptase (Promega) was added, together with one unit RNase inhibitor (Promega) and the elongation reaction was performed according to the manufacturer’s instructions at 25°C for 10min and then at 42°C for 60min (Promega kit). The cDNA was used for PCR amplification using primers as specified in Supplementary Table S1. The cDNA was diluted to enable exponential amplification by PCR as determined by different dilutions of the cDNA. For rRNA analysis, the cDNA was prepared from ~1µg RNA, and was diluted ~1:500 for PCR amplification. To map the cross-linked adducts on rRNA, RNA obtained from AMT-UV cross-linking (250µg) after affinity selection was used to prepare cDNA, and the cDNA was diluted ~1:50. PCR was performed on 1µl of the diluted cDNA (described above), with the addition of 1µM primers and ReadyMix™Taq PCR Reaction Mix with MgCl2 (Sigma). The PCR conditions were as follows: 95° for 2min, followed by 25 cycles of 95° for 30s, 60° for 30s and 72° for 30s.
The snoRNA TB11Cs2C2 (92nt) is located in a cluster that includes SLA1, which guides pseudouridylation on the SL RNA (36,37), the snR30 homolog, which participates in rRNA processing (29), and another C/D snoRNA species, TB11Cs2C1 (76nt). The C/D snoRNAs TB11Cs2C1/C2 were implicated in guiding modifications on rRNA by non-conventional guiding rules (30). However, the methylation predicted to be modified by TB11Cs2C1 (30) is guided by another snoRNA (TB9Cs2C4) based on the conventional +5 rule that governs the methylation of all trypanosome snoRNAs (21,38). As opposed to the snoRNAs that guide modification, TB11Cs2C1/C2 are very abundant, like the U3 snoRNA (22). SnoRNAs are known to either guide modification, to function in rRNA processing, or to perform both functions.
In this study, we utilized snoRNAi of TB11Cs2C2 as a model both for elucidating the mechanism of snoRNAi and to investigate the possible role of this RNA in rRNA processing.
First, we examined which of the Dicers, TbDCL1 and/or TbDCL2, is essential for snoRNAi. The TB11Cs2C2 gene was cloned into the pZJM vector, expressing dsRNA from two T7 RNA pol opposing promoters (31). The construct was used to transfect parental cells, cells carrying a dominant negative TbDCL1 mutant (3), and a cell line silenced for TbDCL2. The dominant-negative TbDCL1 mutant was expressed as Tet-inducible ectopic copy carrying a dual BB2/FLAG tag and a base substitution of aspartate to glycine at amino acid 966 (3). This mutant is defective in cytoplasmic RNAi triggered by transfection of α-tubulin dsRNA, however its specific mode of action has not been investigated further (3). The results in Figure 1A–a demonstrate that, a cell line carrying the silencing construct for TB11Cs2C2 expressed the FLAG-tagged dominant-negative TbDCL1 upon tetracycline induction. To examine the phenotype of dominant-negative TbDCL1 and to verify that this mutant is defective in cytoplasmic RNAi, two cell lines were generated using the pZJM vector expressing dsRNA to silence SmD3 from the two T7 opposing promoters (31). One cell line contained the SmD3 silencing construct, and the second cell line contained both the SmD3 silencing construct as well as the dominant-negative TbDCL1. The results (Figure 1A-b) demonstrate that the presence of the dominant-negative TbDCL1 perturbed the silencing of SmD3 (compare lanes 2–4). Next, the TbDCL2 (Tb927.3.1230) was silenced by RNAi as described in ‘Materials and methods’ section, using the pZJM vector, described above (31). After establishing a clonal cell population, we introduced the silencing construct for TB11Cs2C2, carrying the blasticidin resistance gene. The silencing of TbDCL2 was confirmed by northern analysis (Figure 1A-c), as well as by demonstrating reduction in the level of CIR147 siRNAs (Figure 1A-d), whose production depends on TbDCL2 (5).
After confirming that silencing of TbDCL2 was efficient, the silencing of the TB11Cs2C2 snoRNA was examined by primer extension. The results, shown in Figure 1B and Supplementary Figure S2(A), quantitated by densitometric analysis, demonstrate that induction of TB11Cs2C2 silencing in parental cells reduced the level of the snoRNA by 80% (compare lanes 1 and 2). Similarly, the dominant-negative TbDCL1 mutant did not interfere with snoRNAi, since the level of snoRNA was also reduced by 85% upon tetracycline induction (compare lanes 3 and 4). However, when snoRNAi was examined under depletion of TbDCL2, only a minor reduction was observed in the snoRNA level, suggesting that TbDCL2 but not TbDCL1 is essential for snoRNAi.
To examine if, as in cytoplasmic RNAi, both Dicer and Argonaute are necessary to execute the silencing, the effect of TbAGO1 silencing on snoRNAi was examined. TbAGO1 was silenced using a T7 RNA pol opposing construct (31). Cell lines silenced for TbAGO1 were then transfected with the silencing construct for TB11Cs2C2, and a stable clonal cell line was established and further characterized. The western blot analysis presented in Figure 2A indicated that the TbAGO1 level was significantly reduced in this cell line, suggesting the efficient silencing of TbAGO1 in cells expressing the silencing constructs for both TbAGO1 and TB11Cs2C2. Next, the level of the snoRNA was examined by primer extension in the different cell lines. The results, shown in Figure 2B (lanes 1 and 2) and Supplementary Figure S2(B), demonstrate that silencing of snoRNA is compromised in cells silenced for TbAGO1. As a control, silencing of TB11Cs2C2 in the parental strain was examined (lanes 3 and 4). Thus, the results shown in Figures 1 and and22 suggest that snoRNAi requires both TbAGO1 and TbDCL2.
As mentioned above, several findings suggest that TB11Cs2C2 is likely to have a special function in rRNA processing: it is located in a cluster whose members are abundant RNAs with functions other than rRNA modification; it is not involved in methylation, as was initially proposed (30); and it is as abundant as U3 snRNP, which functions in rRNA processing (22).
To elucidate the function of this snoRNA, a bioinformatic search was conducted to identify potential targets. The entire rRNA cluster (GenBank AL929603.2) was analyzed by BLAST to identify potential base-pair interactions between the snoRNA and rRNA. Four potential base-pair interaction domains with rRNA were found; two sites with extensive complementarity between TB11Cs2C2 and the coding region of srRNA-2 and -6 (see Figure 4A for a schematic representation of the rRNA locus) and two weak interactions in ITS6 and ITS7 (not presented). The potential interactions between TB11Cs2C2 and srRNA-2 and -6 are depicted in Figure 3A and B. Based on this model, different domains of TB11Cs2C2 are involved in base-pair interaction with each of the targets. The secondary structure of TB11Cs2C2 is in agreement with the canonical structure of C/D snoRNA (20). In addition, a stem-loop structure is situated upstream to the interaction domain with srRNA-6. The validity of this structure is supported by phylogenic conservation among the trypanosomes (T. brucei, T. vivax and T. cruzi). Sequence differences that exist among the trypanosomatids include compensatory changes that maintain both the structure of the stems as well as the predicted base-pairing with the rRNA targets (Figure 3A).
To obtain biochemical evidence for these interactions, we carried out AMT treatment followed by UV cross-linking. AMT compounds are unique in their ability to freeze helical regions of nucleic acids. First, the planar AMT molecule intercalates within the double-helical region; upon irradiation at 365nm, covalent adducts with pyrimidine bases are formed (39). This method was used previously in trypanosomes to identify interactions between the spliced leader RNA and other small RNA species (35,40). AMT intercalates within stems formed either inter- or intra-molecularly. To identify intermolecular interactions, a protocol was devised to enrich the cross-linked adducts by affinity selection using an anti-sense biotinylated oligonucleotide complementary to TB11Cs2C2. After affinity selection, the adducts were mapped on TB11Cs2C2 by primer extension (note that reverse transcriptase usually stops one nucleotide before the cross-linked site). Only adducts that were enriched by the affinity selection were considered to have resulted from the interaction between TB11Cs2C2 and rRNA. To evaluate intramolecular cross-links, total RNA was extracted from AMT-treated cross-linked cells, separated on a denaturing gel, and the RNA band corresponding to the size of TB11Cs2C2 was excised and used to map the cross-linked adducts. This RNA should only contain intramolecular cross-linked adducts but not intermolecular cross-linked sites. Thus, primer extension stops common to lanes 4 and lane 6 in Figure 3C are the result of intramolecular cross-linking. These stops are marked with white arrowheads (Figure 3C) and are circled on the secondary structure diagram, and filled in gray (Figure 3A). The adducts were mapped to the proposed stems, supporting the proposed secondary structure of the molecule. In the unique stem located between positions 21 and 49nt of the RNA, four intermolecular cross-linked adducts were detected (U24, U34, U41, U47) and a single adduct (U9) was revealed in the terminal stem. An adduct was also mapped at position U16, in a single stranded domain, suggesting a possible tertiary intramolecular interaction. Interestingly, a stop corresponding to position 53 (marked with a black arrowhead) was also observed (Figure 3C). This stop was not found in RNA extracted from the gel, but was only enriched in RNA purified by affinity-selection, suggesting that this stop represents an intermolecular cross-link and not an intramolecular one. This cross-linked position lies exactly in the intermolecular duplex between TB11Cs2C2 and srRNA-6, depicted in Figure 3A. Note that an additional weak stop was observed one nt after the major stop, which results from reverse transcriptase stopping at the cross-linked site itself.
To gain further support for the proposed interaction between TB11Cs2C2 and its two targets, the cross-linked species between the snoRNA and rRNA were analyzed by RNaseH mapping. The snoRNA-associated rRNA species were tagged by the splint labeling technique. In this method, a primer complementary to TB11Cs2C2 carrying a run of 8G nt at the 3′-end of the oligonucleotide was used as template to incorporate [α-32P]-dCTP at the 3′-end of TB11Cs2C2 using the enzyme T7 DNA polymerase (Supplementary Figure S1). In such an experiment, we anticipate identifying cross-links between TB11Cs2C2 and the different rRNA species linked to it. Indeed, upon UV cross-linking and splint labeling, a major cross-linked species appeared at the top of the gel [Figure 3D (a and b), lanes 2]. To map the cross-linked adducts, the rRNA was first specifically digested by RNaseH and specific oligonucleotides to rRNA and pre-rRNA (depicted schematically in Figure 3D). After digestion, the digested fragments carrying TB11Cs2C2, were detected by splint-labeling using the snoRNA-specific oligonucleotide. The results in Figure 3D–a demonstrate that upon RNaseH cleavage, the level of cross-linked species migrating at the top of gel was reduced and two new species of sizes ~130 and 170nt appeared. The ~130-nt fragment [marked as (1)] was most probably generated by cleavage of a species that was cross-linked to the srRNA-6 before processing, and was then processed by cleaving at the 3′-end of the srRNA-6. The ~170-nt fragment is most probably the result of cleavages at both the internal oligonucleotide as well as the oligonucleotide situated at ITS7 [marked as (2)]. The size on gel represents the size of the rRNA fragment and the size of cross-linked TB11Cs2C2. These data support the interaction of TB11Cs2C2 with srRNA-6. Validating the interaction of TB11Cs2C2 with srRNA-2 was even more important, since it is impossible to detect such interaction by primer extension because the interaction domain lies at the 3′-end of the snoRNA. The same methodology described in panel a, was used to demonstrate the association with srRNA-2. Upon cleavage, two species of 140nt were detected by using an oligonucleotide situated within the srRNA-2 (Figure 3D-b). These species are most probably generated from cleavage of the mature srRNA-2 carrying the cross-linked molecules. The heterogeneity in size of the fragments may stem from heterogeneity at the 5′-end of srRNA-2.
The data presented in panel D support the interaction of TB11Cs2C2 with both rRNA targets. It is not known if these interactions take place simultaneously or sequentially.
Based on the putative interaction domain between TB11Cs2C2 and rRNA (Figure 3A and B), we examined if this RNA is involved in rRNA processing. To this end, the cell line carrying the silencing construct for TB11Cs2C2 was used. To identify rRNA processing defects that might exist in these cells, RT–PCR and RNase protection assays were performed. Cells silenced for NOP58 were used as a control to calibrate the defects in rRNA processing. NOP58 is a structural protein that binds to all C/D snoRNAs and is essential for the stability of the particles. We previously reported the silencing of this factor (22). Total RNA was prepared from uninduced cells and NOP58 silenced cells after 3days of silencing. The levels of the precursors, internal transcribed spacer (ITS) or external transcribed spacer (ETS) of rRNA (schematically represented in Figure 4A), were measured by RT–PCR. The results (Figure 4B-a) demonstrate the accumulation of precursors for all the intergenic regions under conditions of NOP58 silencing, suggesting the C/D snoRNA family must participate in processing SSU, LSUα and LSUβ rRNA, but also in the cleavage of srRNAs. The same RT–PCR methodology was used to examine the defects in cells silenced for TB11Cs2C2. The results (Figure 4B-b) demonstrate increase in ITS5 and ITS7 levels in TB11Cs2C2-silenced cells. No other major changes were observed in the other ITS’s, suggesting the TB11Cs2C2 functions in processing events required to release srRNA-2 and -6. Next, an RNase protection assay was used to examine the level of srRNAs and their precursors. Anti-sense RNA probes were synthesized to the different domains as depicted in Figure 4C, and the protected fragments of mature RNA and the precursor were examined in uninduced cells and after 3 days of silencing (Figure 4C). The results demonstrate major changes in the level of ITS5 and ITS7 precursors (boxed) accompanied by clear reduction in the level of LSUβ and srRNA-6 (boxed) rRNAs. No change in precursors was observed for ITS6, suggesting that TB11Cs2C2 most probably mediates the processing upstream of srRNA-2 and downstream of srRNA-6. These processing events therefore release srRNA-2 and -6 from the precursor. These results are in agreement with the bioinformatic predictions and AMT-UV cross-linking (Figure 3).
TB11Cs2C1 is located between snR30 and TB11Cs2C2 in the same locus, as described above. This snoRNA is also very abundant (22). As stated above, this snoRNA is not involved in modification. To examine if TB11Cs2C1 may function in rRNA processing, a bioinformatic search described above was performed, to examine the potential interaction of this snoRNA with rRNA. The results, shown in Figure 5A and B, demonstrate two potential base-pair interactions; one within rRNA, as previously suggested (30), and one in the ETS close to the A0 cleavage site. To examine if these interactions take place in vivo, these potential associations were examined by AMT UV cross-linking, as described above. The RNA was subjected to affinity selection using an anti-sense biotinylated oligonucleotide complementary to TB11Cs2C1. Primer extension performed on the SSU rRNA domain, using RNA extracted from UV irradiated cells and affinity selected with the snoRNA anti-sense oligonucleotide, indicated a cross-linked adduct on SSU rRNA at position U1900, further supporting the interaction between TB11Cs2C1 and SSU rRNA. The specificity of the cross-linking is evident, since the stop was eliminated following irradiation at 254nm, which photoreversed the cross-linking (Figure 5C).
To evaluate the interaction with the ETS and TB11Cs2C1, we used the ‘RNA walk’ methodology that was recently developed in our group (41). The method is based on UV induced AMT cross-linking in vivo followed by affinity-selection of the hybrid molecules and mapping the intermolecular adducts by RT–PCR or real-time PCR. Domains carrying cross-linked adducts fail to efficiently amplify by PCR as compared with non-cross linked domains. To this end, RNA was prepared from the cross-linked cells, and affinity purified with the anti-sense oligonucleotide to TB11Cs2C1. Following cDNA preparation using random primers, PCR was performed on the cDNA using three pairs of primers, as indicated in Figure 5D-a. The results in Figure 5D-b demonstrate that all domains were amplified from cDNA prepared from control cells. However, the fragment carrying the potential cross-linked domain could not be amplified when cDNA from the AMT UV cross-linking preparation was used, suggesting that this domain is cross-linked to rRNA. Interestingly, the other domains located on the pre-rRNA were more efficiently amplified after AMT-UV cross-linking and affinity selection (Figure 5D-b, compare lanes 2 and 4 with 1 and 3), since pre-rRNA is enriched by the affinity-selection procedure.
To verify that TB11Cs2C1 is indeed cross-linked at the relevant positions involved in interactions with the two targets, the affinity-selected RNA was used in primer extension with primer specific to TB11Cs2C1. The results suggest that primer extension stops were only observed in the affinity selected RNA at positions located within the TB11Cs2C1-rRNA interaction domains (Figure 5A and B). The adduct due to interaction with U22 results from cross-linking to SSU rRNA, whereas the adducts on U32 and U35 most probably result from interaction with the 5′-ETS. These data suggest that TB11Cs2C1 interacts with both SSU rRNA and pre-rRNA (5′-ETS), and therefore is most probably involved in SSU rRNA processing. The ultimate proof of the involvement of this snoRNA in rRNA processing could have been obtained by snoRNAi. However, we failed to obtain >45% reduction in the level of snoRNA in the silenced cells, and as a result of the inefficient knock-down, rRNA processing defects could not be detected. Only two kinds of snoRNAs exist, those that guide methylation and/or those that are involved in rRNA processing. The methylation that exists in the interaction domain of TB11Cs2C1 and SSU, is not guided by TB11Cs2C1 as previously shown (30), but is guided by TB9Cs2C4 (21) according to the +5 rule. This strongly suggests that TB11Cs2C1 is involved in rRNA processing of SSU.
In this study, the mechanism of snoRNAi was elucidated and was found to require both TbAGO1 (Slicer) and TbDCL2 (Dicer). The dependence on TbDCL2 for snoRNAi stems from the localization of the target in the nucleolus.
The data obtained in this study are in agreement with a very recent study showing that the accumulation of transcripts from the CIR147 repeats in T. brucei is only slightly affected by TbDCL1 knock-out, but requires TbDCL2 activity (5). These repeats generate bona fide siRNAs harboring a 5′ monophosphate and a modified 3′-end (5). Like snoRNAs, CIR147 RNAs never leave the nucleus, and therefore their silencing depends on the nuclear Dicer isoform. It is possible that dsRNA produced by T7 RNA pol opposing constructs in the transgenic cells expressing the silencing construct for TB11Cs2C2 are transported to the cytoplasm and are digested to siRNAs by TbDCL1. However, due to the absence of the target in the cytoplasm, these siRNAs would not be effective silencers. In order to mediate snoRNAi, siRNAs must be in the nucleus and interact with both AGO1 and the target snoRNA. Interestingly, in ago1−/– cells, CIR147 transcripts accumulate in the nucleolus, raising the intriguing possibility that silencing may take place in this compartment (8). The presence of the RNAi machinery in the nucleolus may explain why snoRNAi is so efficient.
It is currently unknown if snoRNAi has a physiological role, or if this mechanism simply takes advantage of the machinery that exists in the nucleus/nucleolus for silencing retroposons and the CIR147 transcripts (5). However, since we know so little about how snoRNAs are regulated and processed in the cells, it is still possible that snoRNAi has a physiological role in these processes. Interestingly, silencing of two small nuclear RNAs, namely telomerase RNA and 7SK RNA, was reported in mammalian cells using siRNAs. This suggests that DICER and AGO2 must also operate in the nucleus of mammalian cells to execute similar silencing (12).
This study also sheds light on the complex processing of rRNA in trypanosomes (Figure 4A). The cleavages that excise the SSU rRNA are different in trypanosomes as compared to other eukaryotes such as yeast and mammals (42). In trypanosomes, the first cleavage is between the SSU rRNA and 5.8S RNA at position B1, and only later do cleavages occur at positions A′, A0, A1 and A2 (43). In yeast and mammals, the cleavages to release the SSU rRNA first take place at A1 and A0, and only later do cleavages separate the SSU rRNA from 5.8S RNA (42). So far, only the U3 snoRNA was implicated in processing rRNA in the 5′-ETS domain of trypanosome rRNA (44,45). Another RNA that is implicated in processing of SSU rRNA is snR30, and this RNA was identified in the locus that also contains TB11Cs2C1 and TB11Cs2C2, described in this study (29). There are two additional snoRNAs that were shown to release the SSU rRNA in other eukaryotes, U14 (46) and U22 (47). Our bioinformatic searches failed to identify U14 and U22 homologs among the snoRNAs identified so far. TB11Cs2C1 has the potential to function in SSU rRNA processing, since it interacts with SSU rRNA at the 3′-end of the RNA as well as at a position upstream to A0. Although there is no structural resemblance between TB11Cs2C1 and U14, TB11Cs2C1 may nevertheless function as the functional trypanosome homolog of U14.
Interestingly, recent deep-sequencing analysis of small RNPs in T. brucei (Ullu and Michaeli unpublished results) identified highly abundant C/D molecules; these candidates most probably include previously unidentified snoRNA and/or snoRNA that carry out trypanosome-specific processing events. Work is in progress, using snoRNAi, to examine the distinct roles of these snoRNAs in rRNA processing.
The first snoRNA found to be involved in srRNA processing, TB11Cs2C2, was analyzed in this study. This RNA has the potential to interact with both srRNA-2 and -6, and AMT-UV cross-linking performed in this study confirmed these interactions. The defects observed in rRNA processing suggest that this snoRNA may direct the cleavage upstream from srRNA-2 and downstream from srRNA-6. Interestingly, we did not observe accumulation of ITS6, located between srRNA-2 and srRNA-6, suggesting that there must be another snoRNA that mediates the cleavage of srRNA-2 from the precursor. Indeed, when the rRNA defects were examined in cells depleted for H/ACA RNAs via CBF5 silencing (29), a significant accumulation of ITS6 was observed, suggesting that there must exist an H/ACA RNA that is involved in trypanosome-specific rRNA processing at this site.
In summary, snoRNAi reflects the activity of the nuclear silencing machinery of trypanosomes, which most probably evolved to silence repeats and retroposons, but may have additional novel functions that have not been identified to date. However, snoRNAi can be utilized experimentally for functional analysis of nuclear/nucleolar ncRNAs. This study highlights the function of two abundant trypanosome snoRNPs, TB11Cs2C2, which functions in srRNA processing, and TB11Cs2C1, which is implicated in SSU rRNA processing, since it is no longer believed to function in rRNA methylation. TB11Cs2C1 may represent a functional homolog of snoRNA(s) found in other eukaryotes, which process rRNA at similar sites. The use of snoRNAi, together with deep-sequencing data, is expected to help identify additional trypanosome-specific snoRNAs in the near future.
Supplementary Data are available at NAR Online.
Funding for open access charge: Israel-US Binational Science Foundation (BSF), by an International Research Scholar’s Grant from the Howard Hughes Foundation to SM; Public Health Service Grant (grant number AI28798) to EU. S.M. holds the David and Inez Myers Chair in RNA silencing of diseases.
Conflict of interest statement. None declared.