New strategy for the synthesis of chemically modified RNA constructs exemplified by hairpin and hammerhead ribozymes

Abstract

The CuAAC reaction (click chemistry) has been used in conjunction with solid-phase synthesis to produce catalytically active hairpin ribozymes around 100 nucleotides in length. Cross-strand ligation through neighboring nucleobases was successful in covalently linking presynthesized RNA strands with high efficiency (trans-ligation). In an alternative strategy, intrastrand click ligation was employed to produce a functional hammerhead ribozyme containing a novel nucleic acid backbone mimic at the catalytic site (cis-ligation). The ability to synthesize long RNA strands by a combination of solid-phase synthesis and click ligation is an important addition to RNA chemistry. It is compatible with a plethora of site-specific modifications and is applicable to the synthesis of many biologically important RNA molecules.

Oligonucleotide chemistry is central to the advancement of core technologies such as DNA sequencing and genetic analysis and has impacted greatly on the discipline of molecular biology (1, 2). Oligonucleotides and their analogues are essential tools in these areas. They are produced by automated solid-phase phosphoramidite synthesis, a highly efficient process that can be used to assemble DNA strands over 100 bases in length (3). Synthesis of RNA is less efficient owing to problems caused by the presence of the 2′-hydroxyl group of ribose, which requires selective protection during oligonucleotide assembly. This reduces the coupling efficiency of RNA phosphoramidite monomers due to steric hindrance. In addition, side-reactions that occur during the removal (or premature loss) of the 2′-protecting groups cause phosphodiester backbone cleavage and 3′ to 2′ phosphate migration (4). Although several ingenious strategies have been developed to minimize these problems (5) and to improve the synthesis of long RNA (6), the chemical complexity of solid-phase RNA synthesis dictates that constructs longer than 50 nucleotides in length remain difficult to prepare. Most biologically important RNA molecules such as ribozymes (7), aptamers (8), and riboswitches (9, 10) are significantly longer than this, so new approaches to the synthesis of RNA analogues are urgently required. Although RNA synthesis by transcription might seem a viable alternative, it does not permit the site-specific incorporation of multiple modifications at sugars, bases, or phosphates. In contrast, automated solid-phase RNA synthesis is compatible with the introduction of fluorescent tags, isotopic labels (for NMR studies) and other groups to improve biological activity and resistance to enzymatic degradation. The scope and utility of important RNA constructs can be significantly extended by such chemical modifications, and the scale of chemical synthesis is potentially unlimited (11). These are important factors when planning structural studies on RNA or the development of therapeutic oligoribonucleotides analogues, and should be considered when devising new methods of RNA assembly. Enzymatic ligation might appear to be a method of achieving some of these objectives (12). T4 RNA ligase can be used to join nucleotides in single-stranded regions such as RNA loops, whereas T4 DNA ligase functions on double-stranded RNA. However, neither enzyme can be used to link together RNA strands at other positions, such as across the bases or sugars, or to create unnatural linkages. The use of ligases has other drawbacks; they are often contaminated with RNases that can partially degrade the ligation products, and the ligation protocols are quite complicated, requiring phenol extraction to remove the protein. Moreover, enzymatic ligation methods are not suitable for the large scale synthesis of RNA, and the yields of enzymatic RNA ligation are often low. This is partly due to the tendency of T4 RNA ligase to cyclize phosphorylated RNA strands and to join together incorrect strands.

Here we describe click RNA ligation, a strategy that overcomes the above limitations and could lead to a step-change in the synthesis of biologically active RNA constructs. The CuAAC reaction (13, 14) was chosen for RNA ligation owing to its very high speed, efficiency, orthogonality with functional groups present in nucleic acids and compatibility with aqueous media. In this procedure, individual RNA oligonucleotides are assembled by automated solid-phase synthesis, purified by HPLC then chemically ligated by click chemistry (15) to produce much larger molecules. In this study we demonstrate the power of click ligation by synthesising a series of chemically modified hairpin and hammerhead ribozymes. The hairpin ribozyme belongs to a family of catalytic RNAs that cleave their RNA substrates in a reversible reaction to generate 2′,3′-cyclic phosphate and 5′-hydroxyl termini. It was discovered in tobacco ringspot virus satellite RNA where ribozyme-mediated self-cleavage and ligation reactions participate in processing RNA replication intermediates (16). It was chosen for the current study for three reasons: Its size is such that direct chemical synthesis is problematic and therefore click ligation becomes a valuable technique, its biophysical and catalytic properties have been extensively studied and are well understood (17), and its functionality can be extended by the incorporation of chemical modifications such as fluorescent tags (18). In this study three analogues of the hairpin ribozyme were synthesized from individual oligonucleotides by crosslinking the side-chains of modified uracil bases in opposite strands of the “unclicked” segments (trans-ligation). An RNA ribozyme (98-RNA) was synthesized and a DNA/RNA chimera (98-DNA/RNA) was made in which segments a and b contain tracts of DNA and segment c is entirely DNA (Fig. 1, Left). Ribozymes are known to tolerate deoxyribonucleosides outside the catalytic site and the incorporation of DNA offers the possibility of synthesising longer fragments for click ligation. Therefore we were anxious to confirm that the CuAAC reaction is compatible with DNA–RNA hybrids. A shorter 77-nucleotide version of the hairpin ribozyme was also prepared (77-RNA, Fig. 1, Right) to evaluate a simple splint-mediated modification of the above click ligation strategy. In a different approach, a hammerhead ribozyme (19, 20) was made by the splint-mediated cis-ligation of two RNA strands (Fig. 2). The hammerhead is a small catalytic RNA motif that is found in plant RNA viruses, satellite RNA, viroids, and repetitive satellite DNA (21) where it catalyzes cleavage and ligation of RNA during rolling circle replication (22, 23). It was chosen as a model to investigate the chemistry of a new triazole nucleic acid backbone mimic in click ligation, and to explore its biocompatibility.

Fig. 1.

Assembly of hairpin ribozymes by the CuAAC reaction. The substrate cleavage site (+1 - 1) is indicated by an arrow. 8% polyacrylamide gels: Right-hand lanes show CuAAC crude reaction mixtures and all other lanes show reactants (oligonucleotide segments). The solid arrows point to full-length ribozyme products in all three reactions, and the dashed arrow points to DNA splints in the 77-RNA reaction. Gels visualized by UV shadowing. (Left) The 98-mer RNA hairpin ribozyme 98-RNA and DNA–RNA hybrid 98-DNA/RNA. The sequences are the same but 98-DNA/RNA contains deoxyribose sugars at the positions marked with asterisks* and has a different pattern of dye-labeling (segment a = 5^′-cy5, segment b = 5^′-cy3, segment c = 5^′-FAM). The ES^- mass spectrum of the 98-DNA/RNA ribozyme is shown (found: 33169.4 Da, requires: 33170 Da). (Right) The truncated RNA 77-mer hairpin ribozyme 77-RNA. ES^- mass spectrum (found: 26983.6 Da, requires: 26983 Da).

Fig. 2.

Structure and sequence of the hammerhead ribozyme (green) and substrate (blue) with cleavage site in red. The triazole linkage lies in the catalytic pocket opposite to the cleavage site between C3 and U4. Nucleotides C3 and G8 form a base pair in the active conformation of the ribozyme–substrate complex (33).

Results and Discussion

Hairpin Ribozymes.

The oligonucleotide building blocks required in the assembly of the hairpin ribozyme analogues (segments a, b, and c in Fig. 1) were prepared by automated solid-phase phosphoramidite synthesis using 2′-t-butyldimethylsilyl (TBS) protected monomers for the RNA tracts. The TBS method of RNA synthesis (24) was chosen as it is widely used, the requisite reagents are readily available, and it has been shown to be effective in the synthesis of a wide range of chemically modified RNA analogues (25). Oligonucleotide sequences are shown in Figs. 1 and 2 and in Table S1. For the trans-click ligation reactions, the individual oligonucleotide segments require an alkyne or azide group at the 5-position of a uracil base at the appropriate locus. The reaction scheme and the structures of the alkyne and azide-labeled nucleoside components 3 and 4 are shown in Fig. 3. Azides are unstable to P^III, and do not survive the conditions of phosphoramidite oligonucleotide synthesis. Therefore, the azide functionalities were introduced after oligonucleotide assembly by selective reaction of the N-hydroxysuccinimide ester of azidohexanoic acid 2 with the primary amino groups of the modified uracil bases (26). All oligonucleotides were purified by reversed-phase HPLC and analyzed by mass spectrometry prior to click ligation. For construction of the all-RNA and DNA/RNA 98-mer ribozymes the individual segments were annealed and covalently linked in a self-templated CuAAC reaction (27). The Cu^I catalyst was generated in situ from Cu^II sulfate and aqueous sodium ascorbate, and a water-soluble Cu^I-binding ligand 5 was added to prevent RNA degradation (28). During the reaction segment a was crosslinked to segment b near terminus B, and segment b to segment c near terminus C across the major grooves of the ribozyme stems. The crosslinking positions are indicated in Fig. 1, and the chemical structure of the resultant triazole linkage 6 is shown in Fig. 3. The two simultaneous click ligation reactions proceeded efficiently in the assembly of both ribozymes, and full-length constructs were purified by polyacrylamide gel-electrophoresis (Fig. 1, Left). Despite the self-templating, a minor side-reaction occurred due to cyclization of segment b, which contains both alkyne and azide. This impurity was easily removed during ribozyme purification. A shorter 77 nucleotide version of the hairpin ribozyme was also prepared (77-RNA, Fig. 1, Right), but in this case segment b was synthesized with two azide groups, one near the 5′- terminus and the other near the 3′-end, rather than alkyne and azide. Segments a and c were synthesized with an alkyne near their 3′- and 5′-ends respectively to ensure chemical complementarity with segment b. This facile change in strategy eliminated the possibility of unwanted cyclization of segment b and is an example of the flexibility offered by chemical ligation. Segments a, b, and c were labeled with cy5, cy3, and FAM respectively to facilitate fluorescence resonance energy transfer (FRET) studies (7, 29). There are many cases when RNA assembly cannot be self-templated, so an alternative strategy is necessary. For the smaller hairpin ribozyme 77-RNA we elected to explore the use of two complementary 24-mer DNA splints to hold the short arms of the ribozyme segments in place during click ligation. This method gave a clean reaction and the resultant short hairpin ribozyme was purified by polyacrylamide gel-electrophoresis.

Fig. 3.

The synthetic procedure used to covalently link the arms of the hairpin ribozyme across the major groove of the duplex.

The catalytic activity of the three hairpin ribozyme analogues was investigated by running cleavage reactions at 55 °C in the presence of five equivalents of the RNA substrate (Fig. 4). In all cases five cycles of denaturation (95 °C) followed by cleavage (55 °C) consumed the substrate. It was cleaved at the expected site (indicated in Fig. 1), as confirmed by mass spectrometry of the two fragments (ES^-, found: 5944 Da, 2154 Da, requires: 5943 Da, 2155 Da). The ribozymes also cleaved the substrate without temperature cycling (Fig. S3). The clicked ribozymes were robust and could be isolated from the cleavage gels and reused in subsequent reactions with no significant loss of activity. The fact that the triazole hairpin ribozyme cleaves its substrate rapidly and produces the same products as the natural ribozyme is not surprising as the click linkages are located remotely from the active site. This design feature can be conveniently incorporated into most large RNA constructs. This click linkage stretches across the major groove where it does not perturb the duplex or change its conformation (26). Cross-strand linking between nucleobases in RNA is unique to chemical ligation and cannot be achieved by enzymatic methods.

Fig. 4.

Hairpin ribozyme substrate cleavage reactions. Lanes 1, 3, 5: cleavage reactions of 98-RNA, 98-DNA/RNA and 77-RNA ribozymes respectively with 5 equivalents of substrate. Lanes 2, 4, 6: control—uncleaved substrate. 20% polyacrylamide gels visualized by fluorescence.

The Hammerhead Ribozyme.

A minimal hammerhead ribozyme was synthesized by splint-mediated intrastrand click ligation (cis- ligation) of 3′-alkyne and 5′-azide oligoribonucleotides. This alternative method was chosen to investigate the suitability of click chemistry for linking RNA strands through their backbone rather than between the nucleobases. The success of this strategy in producing an active ribozyme requires the design of a triazole linkage that closely resembles a phosphodiester group. The structure of this linkage and the synthetic strategy to produce it are shown in Fig. 5. The key building block for the synthesis of the 3′-alkyne labeled RNA strand is 5′-DMT-3′-propargyl-5-methyldeoxycytidine 9. This was made in two steps, the first involving 3′-alkylation of 5′-DMT thymidine 7 with propargyl bromide in the presence of sodium hydride in THF to give intermediate 8 in 82% yield, a considerable improvement on the published method (30). In the second step, conversion of thymine to 5-methyl cytosine was carried out using phosphorus oxychloride, N-methylimidazole and aqueous ammonia. Compound 9 was coupled to succinylated aminoalkyl solid support to give derivatized support 10, which was used in the solid-phase synthesis of 3′-propargyl oligonucleotide 11 (Fig. 5). This solid support will be useful in the synthesis of oligonucleotides for various click ligation or labeling applications. The 5′-azide RNA strand 13 was made from the corresponding unmodified 5′-hydroxyl RNA congener 12 by treatment of the support-bound oligonucleotide with methyltriphenoxyphosphonium iodide followed by sodium azide. This 5′-azide labeling method has been used previously on DNA (31) and here we demonstrate its compatibility with RNA. The two components of the hammerhead ribozyme (11 and 13) were designed with extra thymidine nucleotides at the termini to aid gel-filtration, whereas the 5′-FAM dye on the DNA splint was included to differentiate it from the clicked ribozyme during reversed-phase HPLC purification. Splint-mediated ligation of the two RNA strands 11 and 13 proceeded smoothly to give ribozyme 14 with a triazole linkage in the backbone (Fig. 6A, lane 4). The modified hammerhead ribozyme was characterized by mass spectrometry (ES-, found: 7585 Da, requires: 7585 Da) and was shown to cleave its substrate, generating an identical product to that formed by the normal unmodified ribozyme (Fig. 6B, lanes 1 and 2). Cleavage occurred at the predicted scissile phosphodiester bond between cytidine and adenosine as confirmed by mass spectrometry of the two fragments (ES-, found: 6490 Da, 1495 Da; requires: 6490 Da, 1496 Da). Parallel experiments on a 5′-fluorescein labeled substrate demonstrated complete cleavage in 30 min at 37 °C, similar to the native ribozyme (Fig. 6C). This is a satisfying result, particularly as the unnatural triazole linkage is located at the active site between residues C3 and U4 of the ribozyme (Fig. 2). X-ray structures of the minimal (19, 20, 32) and full-length hammerhead ribozymes (33) confirm the critical positioning of C3. It is postulated that a base pair between C3 and G8 holds the ribozyme in its catalytically competent conformation (33). The observed activity of the triazole ribozyme confirms that in this structure the designed linkage is a suitable analogue of a phosphodiester group.

Fig. 5.

Synthesis of the hammerhead ribozyme. Synthesis of alkyne and azide RNA strands, construction hammerhead, and structure of triazole backbone.

Fig. 6.

(A). CuAAC reaction to synthesize the hammerhead ribozyme, 20% polyacrylamide gel: Lanes 1, 2: starting oligonucleotides (alkyne and azide), lane 3: DNA splint, lane 4: crude reaction mixture. (B). Hammerhead ribozyme cleavage reaction with nonfluorescent substrate. Lane 1: S + TR, lane 2: S + NR, lane 3: TR, lane: 4 NR, lane 5: S. (S = uncleaved substrate, NR = normal ribozyme control, TR = triazole ribozyme). Gels A and B visualized by UV shadowing. TR (24-mer) is longer than NR (16-mer) due to the additional thymidines at the termini. (C). Hammerhead ribozyme cleavage reaction with 5′-fluorescein labeled substrate. Lane 1: FS, lane 2: FS + NR 30 min, lane 3: FS + TR 30 min, lane 4: FS, lane 5: FS + NR 1 h, lane 6: FS + TR 1 h. (FS = fluorescent substrate, gel imaged by fluorescence).

To further demonstrate that the triazole linkage is a viable substitute for a phosphodiester group we carried out UV-melting studies on the natural and triazole ribozymes with their RNA substrate in the absence of magnesium ions (to prevent substrate cleavage). We obtained melting curves that were very similar (Fig. S4), with a slight lowering in melting temperature in the triazole case (ΔT_m = 1.1 °C). We then hybridized the native and triazole hammerhead ribozymes to their DNA complement to produce DNA–RNA hybrid duplexes. In this case the triazole analogue was almost as stable as its natural counterpart (ΔT_m = 3.8 °C) (Fig. S4). The CD spectra of the above constructs indicate that the triazole modification does not change the duplex conformation (Fig. S5). Although it would not generally be necessary to place the triazole modification at a critical catalytic position when designing RNA constructs we have shown here that such a substitution can lead to stable biologically active ribozymes, thereby demonstrating the biocompatibility of the triazole linkage.

Conclusion

In this study click chemistry has been used to ligate presynthesized oligonucleotides to construct RNA molecules and DNA/RNA chimeras up to 100 nucleotides in length. The method is compatible with site-specific modifications, mixed backbones and various reporter groups. The reaction is quick, simple, amenable to large scale synthesis, and highly efficient. It is a very flexible procedure that can be used with a diverse range of alkyne and azide-labeled oligonucleotides, which are accessible from commercial sources. It creates novel chemical linkages that cannot be produced by enzymatic methods. Click ligation is therefore an important addition to RNA chemistry and biochemistry.

Two different strategies were employed in this study, although others could also have been used (27, 34, 35). One variant of the click reaction (trans-ligation) was carried out between the nucleobases with self-templating, and also by splint-mediated ligation. The resultant synthetic hairpin ribozymes contain multiple fluorescent labels which can be used to monitor RNA folding and melting. The hammerhead ribozyme was constructed by an alternative strategy, templated cis-ligation, to produce a novel triazole nucleic acid backbone mimic that is compatible with catalytic activity. The chemistry described here could be applied to the synthesis of many important RNA molecules such as riboswitches (9, 10, 36), siRNA delivery systems (37), multivalent aptamers (38), and components of ribosomes (39). Using click ligation, several variants of segments of such RNA constructs could be prepared on a large scale, purified and ligated in different combinations to provide a number of full-length RNA analogues for structural and biochemical studies. It is also interesting to contemplate the possible advantages of using chemical and enzymatic RNA ligation in tandem for the synthesis of RNA constructs beyond the size limit of either individual method. This could be achieved using oligonucleotide building blocks functionalized with 5′-phosphates and alkynes/azides. The enzymatic ligations could be carried out first, then subsequent introduction of Cu(I) and additional alkyne/azide oligonucleotides to the mixture would trigger a set of orthogonal click ligation reactions. This strategy could be used to place native and modified (nonhydrolyzable) linkages at specific positions in the final construct.

Materials and Methods

Mass spectra of ribozymes were recorded on a Bruker micrOTOF™ II focus ESI-TOF MS instrument in ES^- mode and data were processed using MaxEnt. Sterile plastic/glassware and buffers were used for all procedures involving RNA analogues.

Assembly of hairpin ribozymes 98-RNA, 98-DNA/RNA and 77-RNA.

The three segments of each ribozyme (Fig. 1) (10 nmol of each) in 0.2 M NaCl (950 μL) were annealed by heating at 80 °C for 5 min, cooling slowly to room temperature (1 h) then maintaining the temperature at 0 °C for 15 min. A solution of Cu^I click catalyst was prepared from trishydroxypropyltriazole ligand 5 (Fig. 3) (28) (3.5 μmol in 0.2 M NaCl, 50.0 μL), sodium ascorbate (50.0 μmol in 0.2 M NaCl, 50.0 μL), and CuSO₄·5H₂O (0.5 μmol in 0.2 M NaCl, 5.0 μL). This solution was added to the annealed ribozyme segments and the reaction mixture was kept at 0 °C for 15 min, then at room temperature for 1 h. A NAP-10 gel-filtration column was used to remove reagents. The clicked hairpin ribozymes were analyzed and purified by denaturing 8% polyacrylamide gel electrophoresis. To synthesize the short hairpin ribozyme (77-RNA) two complementary 24-mer DNA splints (10 nmol of each) (Table S1), were added and the reaction was carried out under the above conditions.

Templated assembly of triazole hammerhead ribozyme (Fig. 6A).

Alkyne hammerhead segment, azide hammerhead segment and hammerhead splint (Fig. 2 and Table S1), (10 nmol of each) in 0.2 M NaCl (200 μL) were annealed by heating at 80 °C for 5 min and cooling slowly to room temperature (1 h). To a solution of trishydroxypropyltriazole ligand 5 (28) (0.7 μmol) in 0.2 M NaCl (50.0 μL) was added sodium ascorbate (1.0 μmol in 0.2 M NaCl, 2.0 μL) followed by CuSO₄·5H₂O (0.1 μmol in 0.2 M NaCl, 1.0 μL) under argon. This mixture was added to the annealed oligonucleotides and the reaction was left under argon at room temperature for 1 h, made up to 1 mL with water and gel-filtered to remove reagents (NAP-10). The ligated oligonucleotides were analyzed using denaturing 20% polyacrylamide gel electrophoresis and purified by reversed-phase HPLC.

Cleavage of substrate with native and clicked hammerhead ribozymes (Fig. 6 B and C).

The native and clicked hammerhead ribozymes (0.2 nmole of each) were dissolved in 30 μL Tris-HCl buffer (50 mM Tris-HCl, 10 mM MgCl₂, pH 7.6) and 0.1 nmole of the fluorescent hammerhead RNA substrate in 30 μL of the same buffer was added to each ribozyme. The reaction mixtures were incubated at 37 °C for 30 min after which time half the sample (30 μL) was removed, mixed with formamide (30 μL) and frozen in liquid nitrogen. After a further 30 min the remaining 30 μL of the reaction mixture was treated in the same way. The samples were then analyzed by 20% denaturing polyacrylamide gel electrophoresis with fluorescent detection. Cleavage of the nonfluorescent hammerhead substrate was carried out in the same way using 1.25 nmole of the ribozymes and 1.0 nmole of the substrate. The reaction was analyzed after 1 h and the gel was visualized by UV shadowing.

Cleavage of substrate with 98-RNA, 98-DNA/RNA and 77-RNA hairpin ribozymes.

The clicked hairpin ribozymes (0.15 nmol of each) and substrate (0.75 nmol) were dissolved in 40 μL Tris-HCl buffer (50 mM Tris-HCl, 10 mM MgCl₂, pH 7.6), vortexed, and heated in a thermal cycler (PCR machine) at 55 °C for 30 min then 5 cycles of 95 °C for 2 min and 55 °C for 30 min. Formamide (40 μL) was added and the reaction mixture was heated at 80 °C for 5 min to denature the complex, cooled on ice and analyzed by 20% denaturing polyacrylamide gel electrophoresis. To cleave the substrate for analysis by mass spectrometry the clicked hairpin ribozyme and substrate (1.0 nmol of each) were dissolved in 80 μL of Tris-HCl buffer and heated at 55 °C for 1 h, after which the solution was made up to 1 mL with water and desalted by NAP-10 gel-filtration.