Direct cloning of double-stranded RNAs from RNase protection analysis reveals processing patterns of C/D box snoRNAs and provides evidence for widespread antisense transcript expression

Manli Shen 2 Eduardo Eyras 0 1 Jie Wu 6 Amit Khanna 2 Serene Josiah 5 Mathieu Rederstorff 4 Michael Q. Zhang 3 Stefan Stamm 2 0 Catalan Institution for Research and Advanced Studies (ICREA) , Passeig Llus Companys 23, E08010, Barcelona, Spain 1 Universitat Pompeu Fabra , Dr Aiguader 88, E08003 2 Department of Molecular and Cellular Biochemistry, University of Kentucky , Lexington, KY , 40536, USA 3 MCB, UT Dallas, Richardson, TX 75080 , USA And TNLIST, Tsinghua University , Beijing, China 4 Nancy Universite /Biopo le , UMR 7214 AREMS CNRS-UHP, 9 avenue de la Fore t de Haye, 50500 Vandoeuvre-le` s-Nancy, France 5 Shire Human Genetic Therapies , Lexington, MA 02421, USA 6 Cold Spring Harbor Laboratory , 11724, Cold Spring Harbor, NY, USA We describe a new method that allows cloning of double-stranded RNAs (dsRNAs) that are generated in RNase protection experiments. We demonstrate that the mouse C/D box snoRNA MBII-85 (SNORD116) is processed into at least five shorter RNAs using processing sites near known functional elements of C/D box snoRNAs. Surprisingly, the majority of cloned RNAs from RNase protection experiments were derived from endogenous cellular RNA, indicating widespread antisense expression. The cloned dsRNAs could be mapped to genome areas that show RNA expression on both DNA strands and partially overlapped with experimentally determined argonaute-binding sites. The data suggest a conserved processing pattern for some C/D box snoRNAs and abundant expression of longer, noncoding RNAs in the cell that can potentially form dsRNAs. - RNA:RNA interactions play an important role in gene regulation, as shown by the recognition of pre-mRNA splice sites by snRNPs, and the regulation of mRNA function by miRNAs (1). All RNAs undergo extensive processing and are typically generated from longer precursor molecules. Recent high-throughput sequencing (HTS) data showed that RNAs previously viewed as metabolically stable, such as C/D and H/ACA snoRNAs as well as tRNAs undergo further processing resulting in shorter RNA forms (28). To fully understand how these RNAs are formed, it is necessary to clone them. One of the most precise ways to identify RNAs generated from a precursor RNA is to employ RNase protection analysis using a radioactively labelled antisense probe against the precursor. Hybridization of the probe to its target strand generates a dsRNA that is separated from other RNAs by removing all single-stranded RNAs using RNases. This method is well suited to study the processing of a defined larger RNA into smaller fragments, as these fragments can be detected by the shortening of the protected RNAs. RNase protection experiments are well established to give quantitative results. Although RNase protection experiments are highly sensitive and selective, their use is hampered by the inability to directly clone the protected RNA fragments, which is due to the lack of appropriate double-stranded RNAs (dsRNA) modifying enzymes. Previously, only dsRNAs from viruses that can be produced in large quantities could be cloned (9) and cloning has been demonstrated as a proof of principle using in vitro transcribed RNAs and model viral dsRNAs (10). To overcome this problem, we devised a technique to clone dsRNAs from standard RNase protection reactions. An overview of the method is given in Figure 1a. The method allows the identification of RNAs that are generated by processing of precursor RNAs. We were able to establish the processing pattern of RNAs derived from a C/D box snoRNA, MBII-85 (SNORD 116 in humans). Unexpectedly, we found evidence for abundant expression of endogenous RNAs that could form double strands in vivo. These endogenous RNAs overlap with genome regions that show evidence for expression of RNAs from both DNA strands. Some of these potentially dsRNA parts overlap with experimentally verified RNA:miRNA interaction sites. The high abundance of potential dsRNA sites indicates a biological role of RNA expression derived from opposite DNA strands that can be detected with this method. MATERIAL AND METHODS Cloning of dsRNAs Probe synthesis. We synthesize two RNA probes, a lowspecific activity probe for cloning purposes and a highspecific activity probe for detection of the RNAs. For cloning purposes, we synthesize an antisense RNA using all four cold NTPs at a concentration of 1 mM. To visualize the RNA, we spike this RNA with 1 ml of 32P a-UTP (800 mCi/mM) in a 20 ml reaction. This generates a lowspecific activity probe of 0.6 108 cmp/mg. To detect protected fragments, we synthesize one probe with high-specific activity using radioactive 32P a-UTP as the only source of UTP (specific activity 1.4 109 cmp/mg). We use the megascript kit (Ambion) for RNA synthesis. RNase protection. We incubate 100 mg total brain RNA with 500 ng spiked antisense probe. The RNA is prepared using trizol, to avoid loss of small RNAs. In a parallel experiment, we incubate 10 mg total brain RNA with 50 000 cpm of high-specific probe. Probes and cold RNAs were precipitated, dissolved in hybridization buffer and denatured at 95 C for 3 min. Hybridization is carried out overnight, in 10 ml of hybridization buffer at 42 C. Single-stranded RNA is digested by RNase T1 and A1 in 150 ml of RNase digestion buffer. Since both RNase A and RNase T1 leaves a 30 phosphate, we treat the reaction with shrimp nuclease for 300 in the same buffer to generate a free 30OH group. Prior to proteinase K digestion, 15 mg glycoblue (Ambion) is added. Hybridization buffer: 40 mM PIPES, 1 mM EDTA, 400 mM NaCl, 80% formamide, pH 6.4. RNase digestion buffer: 300 mM NaCl, 10 mM TrisCl, 5 mM EDTA, pH 7.4. RNases removal. RNases are removed by adding 15 ml of 10 mg/ml proteinase K and 15 ml of 10% SDS, followed by one hour incubation at 37 C and phenol/chloroform extraction. Removal of free nucleotides. Free nucleotides are removed by running the protected RNAs over a 1518% 8 M Urea, 1 TBE gel. To later visualize the protected bands, we combine the reactions made with the high- and lowspecific probes. After overnight autoradiography, the bands are cut out from the gel and the fragments are recovered by soakcrush in 3 M NH4Ac, 1% SDS solution overnight at 37 C. Fragments are recovered by adding 2.5 volume ethanol and 1 ml of glycoblue. Addition of the 30 linker A. The linker A sequence is: 50rAppCTGTAGGCACCATCAAT/3ddC. The rAppC moiety at its 50-end allows its ligation without ATP to the 30 OH of a nucleic acid. Its 30-end is blocked by inclusion of ddCTP. The first ligation is carried out in a 20 ml volume. The final concentration of linker A is 4 mM. The ligation is carried out for 2 h in a 20 ml reaction in 50 mM HEPES pH 8.3, 10 mM MgCl2, 3.3 mM DTT, 10 mg/ml BSA, (1 RNA ligation buffer, NEB), 8.3% (v/v) glycerol, 10% PEG 5000 and 20 U RNA ligase (NEB). 50 phosphorylation and removal of linker. We phosphorylate the 50-ends using polynucleotide kinase (NEB) and 1 mM ATP for 30 min, followed by precipi (...truncated)