Use of Cis- and Trans-Ribozymes to Remove 5′ and 3′ Heterogeneities From Milligrams of In Vitro Transcribed RNA

Adrian R. Ferr-D'Amar 0 Jennifer A. Doudna 0 0 Department of Molecular Biophysics and Biochemistry, Yale University , PO Box 208114, New Haven, CT 06520-8114, USA - In vitro transcription with phage T7 RNA polymerase is the method of choice for obtaining multi-milligram quantities of RNA for structural studies. However, run-off transcription with this enzyme results in molecules that are heterogeneous at their 3-, and depending on template sequence, 5-termini. For transcripts longer than ~ 50 nucleotides (nt), these impurities cannot be removed by preparative purification techniques. Use of cis-delta, or trans-VS ribozymes allows preparation of homogeneous RNA with any 3-terminal sequence. If present, 5 heterogeneity can be overcome with a cis-hammerhead ribozyme. During the course of a structural investigation of Group II self-splicing introns, we sought to prepare a 70 nt RNA molecule comprising domains V and VI (d56) of the ai5g intron (Fig. 1). Run-off transcription from a plasmid linearized with the restriction enzyme BsaI to generate a DNA terminus ...TAGCC-3 on the template strand resulted in six to eight different molecules, the shortest of which (~ 30% of the full-length transcript) is the desired product (Fig. 2, lane A). Although resolved on an analytical gel, these different molecules could not be separated on a preparative scale either by gel electrophoresis or chromatography, making the resulting RNA inadequate for biophysical studies. Addition of random nucleotides to the 3-terminus of run-off transcripts by T7 RNA polymerase is well documented (1); we chose to use a ribozyme to cleave the 3-end of our transcript to make it homogeneous. The hammerhead (HH) and hairpin ribozymes have been employed previously to cleave transcripts (2, and references therein). However, the well-characterized hammerhead has sequence requirements 5 to the cleavage site that are incompatible with our desired product. This catalytic RNA needs the sequence UX (X G) to precede the cleavage site. The hairpin ribozyme, which needs (G/C/U)N instead, is prone to aberrant cleavage (2). Furthermore, when used to remove 3-termini, both require sequence complementarity with nucleotides internal to the product, necessitating ribozymes of different sequence for cleaving different constructs. In contrast, the hepatitis delta virus ribozyme (d ) and the Neurospora Varkud satellite RNA ribozyme (VS) have minimal sequence requirements: d will cut after any base other than G (3, and references therein), while VS will cleave efficiently after any base other than C (4). Thus, these two ribozymes used in concert should allow cleavage after any desired sequence. * To whom correspondence should be addressed Run-off transcription from a template encoding d56 followed by the 24 nt substrate stemloop required by VS to cut in trans (VSsl) resulted in the expected 94 nt product. In the presence of the trans-acting ribozyme RNA, the desired 70 nt d56 was obtained, contaminated, however with ~ 10% of RNAs that appear to be 1 and 2 nt longer (Fig. 2, lanes C and D). To investigate if these were the result of aberrant cleavage by VS, we transcribed a d56 which is followed by a single 3-terminal U, in turn followed by the d ribozyme. This resulted in a similar pattern of contaminants (Fig. 2, lane B). These longer products are not artifacts of partial dephosphorylation, since acid treatment of the full-length RNA to open the 23 cyclic phosphate followed by phosphatase treatment preserves the pattern, with slightly slower mobilities overall (data not shown). In order to determine the nature of this residual heterogeneity, the dephosphorylated RNAs were 5-end-labeled, separated on a denaturing gel, excised individually, and subjected to partial RNase T1 digestion. This revealed that they differ exclusively at their 5-terminus (data not shown). 5-terminal heterogeneity in T7 RNA polymerase transcripts is thought to result from incorporation of abortive dinucleotides and trinucleotides during initiation of further rounds of transcription (5). We examined the effect of varying the nucleotide triphosphate concentrations during transcription on the ratio of product to contaminants, without succeeding in obtaining a more pure product. The problem was solved by incorporating a hammerhead ribozyme on the 5-end of d56 together with the VSsl at the 3-terminus (Fig. 1). The hammerhead ribozyme cleaved itself off the product almost completely during the course of a 2 h transcription (Fig. 2, lane E). Addition of VS ribozyme in trans resulted in the generation of a homogeneous d56 RNA of the expected mobility (Fig. 2, lane F). We note that the hammerhead ribozyme itself is uniform in length, implying it has a homogeneous 5-terminus. The analytical transcription of the homogeneous d56 presented in Figure 2 was carried out in the presence of an approximately equimolar amount of separately transcribed and purified VS ribozyme. Titration experiments showed that after incubation for 4 h under transcription conditions, one-tenth of this amount of ribozyme achieved complete cleavage of substrate (data not shown). For large scale (20 ml) transcription reactions, we found that simultaneous transcription, with a 1:10 ratio of plasmids encoding the trans-VS ribozyme and HH-d56-VSsl, resulted in almost quantitative conversion of precursor into processed, homogeneous d56, which was easily separated from the ribozymes and cleaved VSsl by preparative denaturing polyacrylamide gel electrophoresis. The yield of purified d56 was ~ 0.5 mg RNA/ml transcription reaction. The use of the trans-acting VS ribozyme for the preparation of large quantities of homogeneous RNA transcripts that we introduce here has several advantages over the previously well-documented use of cis-hammerhead ribozymes. First, the VS ribozyme has minimal sequence requirements 5 to the cleavage site. Secondly, introduction of the VS substrate stem loop into the template is accomplished readily, using PCR, with a 3 primer of modest length. Thirdly, one trans-acting VS ribozyme can be used to cleave multiple RNA constructs, and even recycled for further reactions. Fourthly, in many cases, it is sufficient simply to transcribe in the same vessel the RNA of interest and the VS ribozyme. Finally, when valuable nucleotides (for instance, uniformly isotopically labeled nucleotides for NMR spectroscopy) are employed in transcription, use of atrans-acting ribozyme prepared separately with conventional nucleotides would allow considerable savings of precursors, which would otherwise be incorporated into a cis-ribozyme. We thank K. Zhou for excellent technical assistance, R. Collins for the VS ribozyme-encoding plasmid, and O. Uhlenbeck for a discussion on 5 heterogeneity. A.R.F. is a fellow of the Jane Coffin Childs Memorial Fund for Medical Research. J.A.D. is a Lucille P. Markey Scholar in Biomedical Science. This work was supported by grants from the Jane Coffin Childs (...truncated)