Efficient RNA ligation by reverse-joined hairpin ribozymes and engineering of twin ribozymes consisting of conventional and reverse-joined hairpin ribozyme units Sergei A. Ivanov, Ste ´ phanie Vaule ´ on and Sabine Mu ¨ ller Ruhr-Universita ¨ t Bochum, Bochum, Germany In recent years RNA has become the focus of develop- ment into new diagnostic and therapeutic schemes. Antisense-RNA, ribozyme, aptamer and siRNA tech- nologies have been developed and have found applica- tion in molecular medicine [1–7]. Signalling aptamers and aptazymes have been constructed that can sense a number of molecules in real time and thus are valuable diagnostic tools [8–10]. Furthermore, recently discov- ered riboswitches that regulate gene expression in vivo in response to specific metabolites [11–13] or tempera- ture [14] may lead to new RNA-based therapeutic strategies. Elucidation of the molecular principles of RNA functioning in a specific context has led to the engi- neering of RNA molecules with new functions. Two complementary strategies can be used in RNA engi- neering: rational design and directed evolution. Whereas directed molecular evolution relies on the cre- ation of a repertoire of modified RNAs from which beneficial variants are filtered, in a rational design experiment, defined changes in the nucleotide sequence and ⁄ or secondary structure of a specific RNA are planned on the basis of a preconceived idea. This requires detailed structural and mechanistic informa- tion on the parent RNA. In cases where this informa- tion is available, rational design has contributed to the development of new functional RNA, for example, sig- nalling aptamers and aptazymes [8–10]. Work in our laboratory has focused on the rational design of functional RNA, in particular on the development of hairpin-derived twin ribozymes for site-specific alteration of RNA sequences, and fluores- cent and affinity labelling of large RNA molecules [15–18]. The hairpin ribozyme catalyses the reversible site-specific cleavage of suitable RNA substrates, gen- erating fragments with a 2¢,3¢-cyclic phosphate and, respectively, a free 5¢-OH terminus [19,20]. In the reverse reaction, the oxygen atom of the free 5¢-OH group of one RNA fragment attacks the phosphorous of the cyclic 2¢,3¢-phosphate group of another, result- ing in ligation of the two fragments. In contrast to the hammerhead ribozyme, the conformation of the hair- pin ribozyme–substrate complex does not change signi- ficantly upon cleavage: the two cleavage fragments Keywords rational design; RNA catalysis; RNA ligation; sequence alteration; twin ribozyme Correspondence S. Mu ¨ ller, Ruhr-Universita ¨ t Bochum, Fakulta ¨ t Chemie, Universita ¨ tsstrasse 150, D-44780 Bochum, Germany Fax: +49 234 321 4783 Tel: +49 234 322 7034 E-mail: sabine.w.mueller@rub.de (Received 13 June 2005, accepted 15 July 2005) doi:10.1111/j.1742-4658.2005.04865.x In recent years major progress has been made in elucidating the mechanism and structure of catalytic RNA molecules, and we are now beginning to understand ribozymes well enough to turn them into useful tools. Work in our laboratory has focused on the development of twin ribozymes for site- specific RNA sequence alteration. To this end, we followed a strategy that relies on the combination of two ribozyme units into one molecule (hence dubbed twin ribozyme). Here, we present reverse-joined hairpin ribozymes that are structurally optimized and which, in addition to cleavage, catalyse efficient RNA ligation. The most efficient variant ligated its appropriate RNA substrate with a single turnover rate constant of 1.1 min )1 and a final yield of 70%. We combined a reverse-joined hairpin ribozyme with a conventional hairpin ribozyme to create a twin ribozyme that mediates the insertion of four additional nucleotides into a predetermined position of a substrate RNA, and thus mimics, at the RNA level, the repair of a short deletion mutation; 17% of the initial substrate was converted to the inser- tion product. 4464 FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS remain oriented close to each other in the ribozyme– product complex before dissociating to leave the free ribozyme behind [21–23]. Therefore, in the hairpin ribozyme reaction the entropic cost of ligation is rather low and can be compensated for by the favourable reaction enthalpy of formation of the 5¢,3¢-phospho- diester bond via ring opening of the 2¢,3¢-cyclic phos- phate [24]. This specific feature basically makes the hairpin ribozyme a better ligase than it is an endonuc- lease. However, dissociation of cleavage fragments from the ribozyme has to be considered and thus the ability to preferentially cleave or ligate a specific RNA substrate strongly depends on the stability of the pro- perly folded substrate–ribozyme complex. Strikingly, the hairpin ribozyme is an efficient ligase if the ribo- zyme substrate complex is folded into a stable secon- dary and tertiary structure. By contrast, if secondary and tertiary structure elements are less stable (yet sta- ble enough to form a catalytically competent complex) cleavage is favoured [25,26]. We exploited this specific feature of the hairpin ribozyme in a scheme of site- directed and patchwise exchange of RNA sequences [16,17]. Combination of two hairpin ribozymes into one molecule leads to twin ribozymes with two pro- cessing sites at a suitable substrate RNA. Because of the specific hairpin ribozyme cleavage–ligation charac- teristics described above a fragment of residing sequence is removed in the first part of the reaction followed by binding of another separately added RNA fragment in the gap left behind and its ligation to the final product [16]. In addition to the conventional hairpin ribozyme, we studied hairpin ribozymes with the two domains (loop A and loop B) joined in reverse order [15,27,28] (Fig. 1). To further complement our work on the rational design of ribozymes for RNA sequence alter- ation we were interested in using reverse-joined hairpin ribozymes as building blocks for the construction of twin ribozymes. In order to evaluate the structural properties of reverse-joined hairpin ribozymes for func- tional design, we first studied the cleavage and ligation activity of reverse-joined hairpin ribozyme variants. Based on the results, the most suitable ribozyme was chosen for construction of a twin ribozyme. The ability of this twin ribozyme to mediate site-specific alteration of RNA sequence is demonstrated. Results RNA ligation by reverse-joined hairpin ribozymes Reverse-joined hairpin ribozymes, first introduced by Ohtsuka and colleagues [29,30], are derived from the conventional hairpin ribozyme by dissecting the two domains at the hinge between helix 2 and helix 3 and rejoining helix 4 to helix 1 via a linker of six unpaired residues [27] (Fig. 1). Linkers consisting of cytidine or adenosine have been studied and it has been found that a linker of six unpaired residues is suitable for connecting the two domains [29,30]. Further extending the linker length to 9, 12 or 18 residues increased the cleavage rate by only a factor of two [29]. Therefore, as well as to confine the conformational freedom of the ribozyme structure, we initially used an A 6 -linker for the design of reverse-joined hairpin ribozymes and stabilized helix 3 by a UUCG tetraloop cap [15,27]. Fig. 1. (A) Schematic presentation of how reverse-joined hairpin ribozymes are derived from the conventional hairpin ribozyme. Loop A and loop B domains of the conventional hairpin ribozyme are separated between helix 2 and helix 3. The loop B domain is turned through 180° and helix 4 is rejoined with helix 1 via a single-stranded linker to generate a reverse-joined hairpin ribozyme. (B) Secondary structure of reverse-joined hairpin ribozymes with substrates for ligation experiments. The circle indicates a 5¢-terminal fluorescein moiety used for detec- tion. cp, 2¢,3¢ cyclic phosphate. S. A. Ivanov et al. An engineered ribozyme for RNA sequence exchange FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS 4465 Under appropriate conditions, reverse-joined hairpin ribozymes with an A 6 -linker designed in our laborat- ory efficiently catalysed RNA cleavage [27]. However, ligation activity was rather poor; only about 1% of cleavage products ⁄ ligation substrates were ligated (S. A. Ivanov & S. Mu ¨ ller, unpublished results). Pre- vious linker length variation experiments were carried out to study the cleavage activity of reverse-joined hairpin ribozymes without looking at ligation [29,30]. Therefore, to search for reverse-joined hairpin ribo- zymes with improved ligation activity we reinvestigated the influence of linker length using the constructs shown in Fig. 1B. In order to favour ligation over cleavage, helix 2 was extended to 16 bp (Fig. 1B). This design should allow for a higher ligation yield due to more stable binding of the 5¢-ligation fragment [25,26]. Furthermore, we used two different short substrates, SG9 and SU9, for ligation. Whereas SU9 forms a 6 bp duplex with the ribozyme, binding of SG9 generates a duplex of only 5 bp due to the terminal G-A mis- match. Thus, the nonpaired adenosine residue at the hinge point can be integrated into the linker leading to further enhancement of the degrees of conformational freedom. We prepared reverse-joined hairpin ribo- zymes with single-stranded linkers of 6, 7, 8 and 12 adenosine residues and compared their activity for ligation of S17F5 cp with either SU9 or SG9 (Fig. 1B, Table 1). This experimental design allowed us to look at eight ribozyme–substrate complexes varying in the length of the single-stranded linker and ⁄ or the length of the duplex between the ribozyme and the 3¢-ligation substrate. Ligation reactions were carried out under conditions involving equimolar concentrations of ribo- zyme and ligation fragments, as well as under single turnover conditions (Table 1) in the presence of 10 mm MgCl 2 and 2 mm spermine; the polyamine was previ- ously found to be essential for efficient reverse-joined hairpin ribozyme catalysis [27]. Initially, we used a reaction temperature of 32 °C because this has been shown to be optimal [29,30]. However, we observed that ribozyme activity varied only slightly in measure- ments at 32 and 37 °C. The construct HP–RJWTA7 showed the fastest reaction kinetics and gave the high- est yields for ligation of the short fragment SG9 to S17F5 cp under equimolar concentrations, as well as under ribozyme saturation (Table 1). Thus, a linker length of eight adenosine residues in combination with a duplex of 5 bp between the 3¢-ligation substrate and the ribozyme seems to be most favourable for an efficient reaction among the studied species. HP–RJWTA7 is also an efficient endonuclease; it cleaves the substrate S40F3F5 with k obs ¼ 0.4 min )1 at equimolar concentrations of ribozyme and substrate (according to conditions used for sequence exchange reaction, see below). Design of a twin ribozyme and kinetic analysis As a result of the studies described above, the opti- mized structure of the reverse-joined ribozyme unit for twin ribozyme design consists of two domains joined by a single-stranded linker of eight adenosine residues and binds the 3¢-terminal part of its substrate via a duplex of 5 bp (Fig. 2A). The second part of the twin ribozyme consists of a three-way junction hairpin ribo- zyme as used in the twin ribozyme approach described recently [16]. In order to learn about the activity of both ribozyme units in the twin ribozyme, we determined the kinetic parameters for cleavage as well as ligation at the two sites in individual experiments (Fig. 3). Substrate RNAs S40F5dA15 and S40F5dA31 were synthesized to be cleaved at either of the specific sites. Cleavage at the second site was abolished by replacing the attack- ing 2¢-OH group with a hydrogen atom. Specifically, either A15 or A31 was substituted by deoxyadenosine. Cleavage reactions were carried out under single turn- over conditions delivering the kinetic constants shown Table 1. Ligation parameters of reverse-joined hairpin ribozymes with single stranded linkers of varying lengths. Ribozyme HP-RJWTA6 HP-RJWTA7 HP-RJWTA8 HP-RJWTA12 3¢-substrate SU9 SG9 SU9 SG9 SU9 SG9 SU9 SG9 3¢-terminal base of substrate U G U G U G U G 5¢-terminal base of ribozyme A A A A A A A A Linker length, n 6778891213 Duplex length, bp 6 5 6 5 6 5 6 5 k lig ,min )1 0.564 0.938 0.894 1.140 0.535 0.865 0.263 0.447 K D ,nM 140.8 141.1 109.3 75.3 73.9 97.3 76.6 85.4 Ligation yield (%) under equimolar conditions 35.2 43.7 35.0 43.3 31.5 43.0 21.7 33.1 under single turnover conditions 55.8 61.0 63.3 70.0 54.1 57.6 35.3 43.0 An engineered ribozyme for RNA sequence exchange S. A. Ivanov et al. 4466 FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS A B Fig. 2. (A) Reaction scheme for HP–TWRJ-mediated fragment exchange. Substrate RNA S40F3F5 is annealed to HP–TWRJ (left) and cleaved at two defined sites. The fragment extending between the two cleavage sites (16-mer, shown in red) is replaced on the ribozyme by the oligonucleotide S20 cp (20-mer, shown in green) which subsequently becomes ligated to the flanking substrate fragments to form the HP–TWRJ–product complex (right). Green circles indicate 5¢-and3¢-terminal fluorescein moieties used for detection. Reaction was carried out in the absence and presence, respectively, of the oligonucleotide S6-anti, which is complementary to the six 5¢-terminal nucleotides of HP–TWRJ (for details refer to main text). (B) Model duplexes used for determination of melting points. The sequence of the educt and prod- uct duplex corresponds to the sequence of HP–TWRJ with initial substrate S40F3F5 and with product P44F3F5, respectively. Fig. 3. Secondary structures of ribozyme substrate complexes used for measuring cleavage (A) or ligation (B) rates at either site. Cleavage and ligation at the second site was abolished by replacing the essential adenosine in the substrate strand (A15 or A31) by deoxyadenosine. Grey circles indicate 5¢- and 3¢-terminal fluorescein moieties used for detection. S. A. Ivanov et al. An engineered ribozyme for RNA sequence exchange FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS 4467 in Table 2. A similar setup was used to determine liga- tion parameters: fragments with dA instead of A (S35F5dA15cp and S29dA31) were used for ligation with the corresponding substrate SG9 or S15F5 cp, respectively. The results show that the twin ribozyme- mediated cleavage, as well as ligation, proceeds with similar activity at both individual sites; cleavage rates are virtually identical, ligation rate constants vary by a factor of only about 2.5. Alteration of RNA sequence by the twin ribozyme HP–TWRJ The twin ribozyme HP–TWRJ was designed to pro- mote the insertion of four specific nucleotides into a predetermined site of an arbitrarily defined substrate RNA as illustrated in Fig. 2A. In the ribozyme– substrate complex, a stretch of four nucleotides in the central part of the ribozyme strand (GAUU) bulges. Cleavage at both predefined sites releases a 16-mer that can easily dissociate from the ribozyme because of the destabilizing bulge. The added RNA fragment S20 cp contains the four additional nucleotides complement- ary to the GAUU loop in the ribozyme strand. Hence, binding of this oligonucleotide to the gap left by disso- ciation of the 16-mer, converts the previously inter- rupted duplex into a continuous one extended by 4 bp. For thermodynamic characterization of the system, we determined the melting points of model duplexes cor- responding to the central part of the twin ribozyme– substrate and ribozyme–product complexes (educt and product duplex, Fig. 2B). There was a significant dif- ference in the melting temperature of the two duplexes (Table 3), supporting our strategy to drive the reaction by a change in duplex stability. On the basis of the melting temperatures obtained using varying duplex concentrations, thermodynamic parameters were deter- mined. The binding enthalpy for both duplexes varied by 71.4 kcalÆmol )1 and fragment exchange was associ- ated with a favourable DG of )15.2 kcalÆmol )1 (see Experimental procedures and Table 3). This should drive the reaction in the desired direction and favour product formation. The validity of the experimental design was checked by analysing the time course of the reaction using a fluorescence assay as described previously [15]. The substrate RNA S40F3F5 was incubated with an equi- molar amount of ribozyme at 37 °C. We used a reac- tion temperature of 37 °C in order to compare the results with our previous twin ribozyme studies [16]. The reaction was allowed to proceed for 30 min, after which S20 cp was added in equimolar quantities to ribozyme and initial substrate, and the reaction was left to proceed at 37 °C for another 120 min. We ini- tially chose this experimental design in order to be able to observe the individual reaction steps. However, as found later, the reaction occurs in a similar manner when initial substrate, ribozyme and S20 cp are present in the reaction mixture from the beginning (S. Vaule ´ on &S.Mu ¨ ller, unpublished observations). Data for the original setup are given in Fig. 4. After 30 min, char- acteristic cleavage products were detected (Fig. 4B, lane 2). Addition of the fragment S20 cp led to the for- mation of new products detected as three additional bands (lanes 3 and 4). These signals correspond to the 29- and 35-mer resulting from ligation of the 20-mer to either the 9- or 15-mer cleavage product, and to the desired product RNA (P44F3F5) resulting from liga- tion of the 20-mer to both fragments as shown in Fig. 4A. Continuing the reaction for 15 min after the addition of S20 cp (total reaction time: 45 min) resul- ted in conversion of 5% of starting material into the 44-mer product P44F3F5, whereas 13% of the sub- strate S40F3F5 remained unprocessed (Fig. 4B, lane 3, Fig. 4C). After another 105 min of incubation, further enrichment of products involving ligation to the 20-mer fragment S20 cp was observed. The ratio of product to initial substrate increased; 9% of substrate RNA was converted to the 44-mer and 5% was left unchanged (Fig. 4B, lane 4, Fig. 4C). These results demonstrate favourable cleavage of the substrate fol- lowed by dissociation of the 16-mer versus favourable ligation of the 20-mer. The yield of product RNA P44F3F5 could be fur- ther increased by stabilization of the ribozyme active Table 3. Thermodynamic parameters of model duplexes. T m (°C) a DH (kcalÆmol )1 ) DS (calÆK )1 Æmol )1 ) DG 37 °C (kcalÆmol )1 ) Educt duplex 38.1 )78.4 )223.0 )9.2 Product duplex 72.7 )149.8 )404.2 )24.4 a At 500 nM oligonucleotide concentration. Table 2. Kinetic parameters of HP–TWRJ catalyzed cleavage and ligation reactions. Substrates k react , (min )1 )K(nM) k react ⁄ K, (min )1 lM )1 ) Cleavage S40F5dA15 0.57 ± 0.01 16.1 ± 2.4 35.4 S40F5dA31 0.32 ± 0.02 12.0 ± 4.6 26.6 Ligation SG9 + S35F5dA15cp 0.55 ± 0.03 22.9 ± 8.5 24.0 S29dA31 + S15F5 cp 1.38 ± 0.09 34.8 ± 8.1 39.6 An engineered ribozyme for RNA sequence exchange S. A. Ivanov et al. 4468 FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS structure. Theoretical analysis of HP–TWRJ folding (using software rna structure 4.0) showed that HP–TWRJ, in addition to the desired minimal energy structure (DG ¼ –70.7 kcalÆmol )1 ), can fold into an alternative nonfunctional structure with a virtually identical Gibbs’ free energy (DG ¼ –70.5 kcalÆmol )1 ). A short antisense oligonucleotide 3¢-CCCTCT-5¢, complementary to the six 5¢-terminal nucleotides of HP–TWRJ (Fig. 2A), assists proper folding; folding analysis of this system revealed an energy difference between both competing structures of 10 kcalÆmol )1 in favour of the desired functional structure. Carrying out the sequence-exchange reaction described in the presence of this antisense oligonucleotide increased the yield of the final product to 17% (Fig. 4C). To validate our data we repeated the reaction, using for sequence exchange a 20-mer fragment internally labelled with a Cy5 moiety [S20 cp(Cy5)] (Fig. 5). This experimental setup allowed us to detect ligation pro- ducts not only by fragment lengths analysis with an ALF DNA sequencer (detection of fluorescein emis- sion upon excitation at 488 nm), but also by virtue of their unique fluorescence at 700 nm using a LI-COR DNA sequencer (detection of Cy5 emission upon exci- tation at 680 nm). As shown in Fig. 5A, there is a clear conversion of the fast-running Cy5-labelled 20-mer into three slower running species corresponding to ligation products of the Cy5-labelled 20-mer with fluorescein-labelled 9-mer [29-mer (F3, Cy5)] and 15-mer [35-mer (F5, Cy5)], respectively, and to final 44-mer RNA product labelled with Cy5 and two fluo- rescein moieties [44-mer (F3, F5, Cy5)]. In both experiments (compare Figs 4B and 5A) there is a strong 35-mer signal resulting from ligation of the 15-mer produced by cleavage in the first step of the reaction to the added 20-mer fragment S20 cp. This illustrates that ligation at the site of the conven- tional hairpin ribozyme proceeds somewhat faster (2.5-fold, Table 2) than ligation at the reverse-joined ribozyme site. Discussion The twin ribozyme HP–TWRJ was designed to mediate the specific exchange of two RNA fragments (Fig. 2). A B C Fig. 4. (A) Fragmentation and ligation scheme (compare with Fig. 2). cp, 2¢,3¢-cyclic phosphate. (B) Monitoring the reaction. Lane 1, start of cleavage reaction. Lane 2, mixture after 30 min at 37 °C (immediately before adding S20 cp). Lane 3, mixture 15 min after addition of S20 cp (incubation at 37 °C continued, total reaction time: 45 min). Lane 4, mixture after additional 105 min at 37 °C (areas of peaks corresponding to S40F3F5 and P44F3F5 indicate 9% conversion to full length product and 5% remaining substrate). For additional details see main text. Peak heights are standardized by the data processing software, such that total peak integrals of different lanes are not constant. (C) Time course of fragment exchange reaction in the absence (solid lines) and presence (dashed lines) of the antisense oligonucleotide S6-anti. S. A. Ivanov et al. An engineered ribozyme for RNA sequence exchange FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS 4469 During the process, 16 nucleotides of residing substrate sequence are exchanged for 20 nucleotides, which are added to the reaction mixture as a separate synthetic RNA fragment. Recently, we communicated the devel- opment of a twin ribozyme consisting of two hairpin ribozymes connected in tandem that can catalyse the same fragment-exchange reaction [16]. The twin ribo- zyme described here was more challenging, because it involves a reverse-joined hairpin ribozyme unit and required more extensive design and evaluation. Reverse-joined hairpin ribozymes were introduced nearly 10 years ago [29,30]. Since then they have attrac- ted little attention: there has been no follow-up demon- strating the catalytic potential of these interesting ribozyme structures beyond the work of the initial developers and our laboratory. We studied the ligation activity of reverse-joined hairpin ribozyme variants. Interestingly, reverse-joined hairpin ribozymes act as highly efficient ligases. The most active variant HP–RJWTA7 ligated two substrates with 43% yield when all reactants (ligation substrates and ribozyme) were incubated at equimolar concentrations; under single turnover conditions the yield increased to 70% (Table 1). A conventional hairpin ribozyme variant corresponding to the 3¢-terminal region of the twin ribozyme HP–TWRJ delivered only 29% of ligated product (compared with 43% for HP–RJWTA7, data not shown). No high-resolution structure is available for reverse-joined hairpin ribozymes. However, the observed functionality implies that the active confor- mation of reverse-joined hairpin ribozymes involves a similar relative orientation of the two ribozyme domains to that seen in the crystal structure of the con- ventional hairpin ribozyme [21,31]. Variation in linker length and ⁄ or the length of the duplexes flanking the single-stranded linker will, therefore, influence the posi- tions of the two loops in the folded structure. Our results indicate that a single-stranded linker of eight adenosine residues, as in HP–RJWTA7, and a 5 bp duplex between the 3¢-ligation substrate and the ribo- zyme allows proper folding of the complex into the act- ive conformation with the required contacts between loops A and B. Thus, to the best of our knowledge, this is the first example of a reverse-joined hairpin ribozyme that, under appropriate conditions, can ligate two suit- able substrates with up to 70% yield. AB Fig. 5. (A) Monitoring the HP–TWRJ mediated fragment exchange involving a Cy5-labelled oligonucleotide S20 cp(Cy5). Lane 1, S20 cp(Cy5) control. Lanes 2 and 3, incubation of S20 cp(Cy5) with HP–TWRJ in the absence of initial substrate under conditions of fragment exchange reaction after 5 and 80 min. Lane 4, incubation of S20 cp(Cy5) with initial substrate S40F3F5 in the absence of HP–TWRJ under conditions of fragment-exchange reaction. Lane 5, mixture after 30 min cleavage reaction at 37 °C (immediately before adding S20 cp(Cy5)). Lanes 6, 7 and 8, mixture 15, 120 and 180 min, respectively, after addition of S20 cp(Cy5) (incubation at 37 °C continued). Lane 9, Cy5 labelled 29-mer as length control. Double bands result from an isomeric mixture (cis- ⁄ trans-isomers) of the Cy5 moiety used for labelling. (B) Schematic presentation of HP–TWRJ mediated fragment exchange (compare with Fig. 2). Green circles indicate 5¢-and3¢-terminal fluorescein moieties; the red circle indicates a Cy5 moiety used for detection. An engineered ribozyme for RNA sequence exchange S. A. Ivanov et al. 4470 FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS The reverse-joined hairpin ribozyme HP–RJWTA7 was combined with a three-way junction hairpin ribo- zyme to generate the twin ribozyme HP–TWRJ. The newly designed structure mediates the site-specific exchange of two patches of RNA sequence with up to 17% yield, which is a remarkable improvement com- pared with earlier versions of this twin ribozyme [15,32]. The yield can possibly be increased further by the additional destabilization of ribozyme–substrate complexes in the region containing the sequence to be exchanged (Fig. 2B, red). Although a bulge of four nucleotides has been introduced to weaken binding of this sequence, there is still a duplex of eight contiguous base pairs hampering dissociation of the fragment to be exchanged after cleavage. Further reduction of the length of this duplex would facilitate dissociation. However, this implies that the two ribozyme units of the twin ribozyme are located closer together and this may interfere with proper folding of the twin ribozyme due to a sterical clash between both catalytic units. In our previous studies, we observed that the sequence and length of the helix between the two loops contain- ing the cleavage ⁄ ligation site, as well the size of the bulge, influence the exchange efficiency [16,32,33]. Therefore, the specific design of a custom-designed twin ribozyme will strongly depend on the substrate to be processed, such that variation of the distance between the two sites seems reasonable only at this stage. In summary, the demonstration of functionality of HP–TWRJ provides proof of the principle that, in addition to the conventional hairpin ribozyme, reverse- joined hairpin ribozymes can also be used as building blocks for the construction of twin ribozymes. Even though HPTWRJ is not as efficient as its tandem- configured relative [16], its successful construction and the demonstration of its functionality supports the idea of creating novel RNA catalysts by rational design. Furthermore, reverse-joined hairpin ribozymes can act on RNA substrates that are not readily accepted by conventional hairpin ribozymes. For example, a hair- pin ribozyme variant with the terminal base pair of the ribozyme–substrate duplex at the hinge changed from 3 0 -CU-5 0 5 0 -GA-3 0 to 3 0 -GA-5 0 5 0 -CU-3 0 displayed 10-fold lower cleavage activity compared with the wild-type ribozyme [15]. By contrast, corresponding variants of the reverse-joined hairpin ribozyme showed no significant difference in cleavage behaviour. (C. Schmidt & S. Mu ¨ ller, unpub- lished results). We attributed this result to enhanced coaxial stacking of the two domains in the relevant variant of the hinged conventional hairpin ribozyme leading to enrichment of an extended inactive confor- mation [34,35]. Because of the single-stranded linker, joining the two domains in reverse-joined hairpin ribo- zymes, coaxial stacking is less probable and therefore not sensitive to the sequence at the hinge point. A useful application for twin ribozymes is site-speci- fic labelling or functionalization of transcripts in vitro and possibly in vivo as we have recently demonstrated with tandem configured twin ribozymes [18]. The twin ribozyme accepts RNA fragments that are conjugated with a dye (here Cy5). A number of other dyes and modifications are incorporated equally well even in rather long and structured RNA molecules [18]. Thus, the twin-ribozyme strategy paves the way for site-speci- fic labelling ⁄ modification of RNA molecules that are too long for chemical synthesis. Furthermore, apart from fragment-exchange reactions, simple clipping of a fragment of desired length and sequence from a nat- ural RNA may be a useful application. For example, RNA fragments that involve modified nucleobases are easily obtainable from naturally occurring RNA by the use of twin ribozymes. Subsequently, these fragments can be investigated using various analytical methods. Another potential application is genotyping of single nucleotide polymorphisms in human genomes [36]. Subsequent analysis of appropriate RNA fragments by MS can reveal if a certain nucleotide in the target gene has been altered [37,38]. Thus, the development and application of twin ribozymes may lead to a number of interesting strategies in molecular biology, genome analysis and possibly molecular medicine. It is cer- tainly advantageous having a number of twin-ribozyme variants that can be adapted to and optimized for a specific target. It has been shown previously that the sequence of the substrate-binding domain of the hairpin ribozyme can be adapted to cleave a desired RNA substrate, just a few conserved nucleobases are required [39]. The sequence requirements for conven- tional and reverse-joined hairpin ribozyme substrates are virtually the same [29,30]. However, the distinct mode of joining the two ribozyme domains in both variants leads to distinct acceptance of specific sequences particularly in the region of the domain hinge. The new designed twin ribozyme HP–TWRJ is thus a valuable addition to the existing variants of a tandem configured hairpin ribozyme. Experimental procedures Synthesis of ribozymes and substrates Reverse-joined hairpin ribozymes HP–RJWTAn (n ¼ 6, 7, 8, 12) and the twin ribozyme HP–TWRJ were transcribed in vitro from oligonucleotide templates using T7 RNA S. A. Ivanov et al. An engineered ribozyme for RNA sequence exchange FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS 4471 polymerase essentially as described previously [15]. Briefly, double-stranded DNA templates were generated from two synthetic primers (BioTez, Berlin, Germany) overlapping by 15 complementary bases (for generation of templates for transcription of HP–RJWTAn) or 32 complementary bases (for generation of the template for transcription of HP–TWRJ). After primer annealing, DNA templates were completed with DNA polymerase I, Klenow fragment exo – (Fermentas, St. Leon-Rot, Germany). Transcription was carried out with T7 RNA polymerase at 37 °C in standard transcription buffer (40 mm Tris ⁄ HCl pH 7.9, 6 mm MgCl 2 , 10 mm dithiothritol, 10 mm NaCl, 2 mm spermidine) over 3 h (HP–RJWTAn), or in Hepes buffer (20 mm Hepes pH 8.0, 10 mm Mg(OAc) 2 ,10mm NaOAc, 1 mm dithio- threitol, 25 lgÆmL )1 bovine serum albumin) for 4.5 h (HP– TWRJ). Proteins were removed by phenol ⁄ chloroform extraction and RNA was precipitated from ethanol. Ribo- zymes were purified by denaturing gel electrophoresis (7 m urea, acrylamide ⁄ bis-acrylamide 19 : 1) on a 10% poly- acrylamide gel (HP–TWRJ) or a 20% polyacrylamide gel (HP–RJWTAn). Product-containing bands were eluted with 2 m LiClO 4 overnight at room temperature and precipitated from acetone. Substrate oligoribonucleotides SU9, SG9 and S40F3F5 were synthesized using the phosphoramidite method on a 1 lmole scale using an automated DNA ⁄ RNA synthesizer (Gene Assembler Special, Pharmacia Biotech, Freiburg, Germany). 5¢-Fluorescein labelling was achieved by solid- phase coupling of ‘fluoreprime’ phosphoramidite (Amer- sham Biosciences, Freiburg, Germany). For labelling at the 3¢-end, controlled pore glass was used to which fluorescein was attached via a thiourea functionality as a succinate linkage (ChemGenes Corp., Wilmington, USA). It con- tained a dimethoxytrityl-protected hydroxyl group which after deprotection was used for chain elongation. All RNA substrates were purified by electrophoresis on 15–20% denaturing polyacrylamide gels, eluted with 2 m LiClO 4 and precipitated from acetone. For internal Cy5 labelling of the 20-mer fragment S20 cp(Cy5), a deoxythymidine carrying an amino function (amino modifier C6dT phosphoramidite, ChemGenes Corp.) was built into a 29-mer oligonucleotide during chemical synthesis (5¢-GUCCAGAAA-NH 2 C6dT-CUCC- CUCACAGUCCUCUUU-3¢). After standard deprotection and gel purification the amino function was coupled with Cy5-NHS ester (Amersham Biosciences). To this end, 30 nmol of the oligonucleotide were solved in 500 lL car- bonate buffer (0.1 m pH 8.5) and mixed with 1 mg dye. The reaction was allowed to proceed for 1 h at room tem- perature in the dark with occasional shaking. The reaction was stopped by ethanol precipitation. The side product N-hydroxysuccinimid was removed by washing the resolved precipitate over a NAP column; the labelled oligonucleotide was separated from nonlabelled species by gel electrophor- esis and then cleaved with a conventional hairpin ribozyme HP–WTTL [27] to yield the 20-mer fragment S20 cp(Cy5) with 2¢,3¢-cyclic phosphate. In the same way RNA frag- ments S17F5 cp, and S20 cp containing a 2¢,3¢-cyclic phos- phate group (cp) were obtained from cleavage of appropriate chemically synthesized RNA molecules with HP–WTTL [27]. Extinction coefficients of ribozymes and substrates were calculated with oligoanalyzer 3.0 (http:// www.idtdna.com) taking into account the absorption of fluorophores. Ligation experiments Individual ribozymes were mixed with the respective sub- strate, SU9 or SG9 in Tris ⁄ HCl (pH 7.5) buffer, heated at 90 °C for 1 min followed by incubation at 32 °C for 15 min. The second ligation substrate S17F5 cp was added and the mixture was incubated at 32 °C for another 10 min. Reactions were started by the addition of MgCl 2 and spermine. The final volume of the reaction mixture was 10 lL, final concentrations were: 200 nm ribozyme, 20 or 200 nm SU9 and SG9, 200 nm S17F5 cp, 10 mm MgCl 2 , 2mm spermine, 15 mm Tris ⁄ HCl, pH 7.5. Ligation was allowed to proceed at 32 °C for 30 min. Aliquots (2.5 lL) were removed at suitable intervals and reactions were quenched by addition to 3 lL stop-mix (10 mm EDTA in 90% formamide) followed by intensive vortexing, heating at 90 °C for 1 min, and immediate cooling on ice. Each experiment was repeated at least once. Samples were ana- lysed by PAGE on 8% denaturing gels (7 m urea) using an ALF DNA sequencer (Pharmacia Biotech) as described pre- viously [15]. Data were processed and analysed using dna fragment analyzer 1.2 Software (Pharmacia Biotech). Kinetic constants of ligation reactions were determined under single turnover conditions. Reactions were carried out essentially as described above. However, owing to very fast ligation the required volumes of MgCl 2 stock solution as well as of spermine stock solution were placed on the inner wall of the Eppendorf tube and reaction was started by intensive vortexing. The final volume of the reaction mixture was 15 lL; final concentrations were: 80–500 nm ribozyme, 20 nm SU9 or SG9, respectively, 88–550 n m S17F5 cp, 10 mm MgCl 2 ,2mm spermine, 15 mm Tris ⁄ HCl, pH 7.5. Reaction analysis was as described above. Appar- ent first-order rate constants (k obs ) were obtained from a plot of product formation against time in the linear phase of reaction. Rate and dissociation constants (k lig and K D ) were calculated from linear curve fitting using plots of k obs against k obs ⁄ [E], where [E] ¼ ribozyme concentration (Eaddie–Hofstee plot). The margin of error was < 10% in all measurements. Determination of kinetic parameters Kinetic parameters of twin ribozyme cleavage reactions using individual substrates S40F5dA15 and S40F5dA31 An engineered ribozyme for RNA sequence exchange S. A. Ivanov et al. 4472 FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS were determined under single turnover conditions. Substrate was mixed with ribozyme in Tris ⁄ HCl (pH 7.5) buffer, hea- ted at 90 °C for 1 min and incubated at 37 °C for 15 min. Reactions were started by addition of MgCl 2 and spermine, portions of which were placed before on the Eppendorf tube wall as described in ligation experiments. The final volume of the reaction mixture was 15 lL; final concentra- tions were: 40–200 nm ribozyme, 20 nm substrate, 10 mm MgCl 2 ,2mm spermine, 15 mm Tris ⁄ HCl, pH 7.5. Kinetic parameters of twin ribozyme ligation reactions were determined under single turnover conditions, using the following protocol. Ribozyme was mixed with the respective substrate, SU9 or SG9 in Tris ⁄ HCl (pH 7.5) buffer, heated at 90 °C for 1 min followed by incubation at 37 °C for 15 min. The second ligation substrate S17F5 cp was added and the mixture was incubated at 37 °C for another 10 min. Reactions were started by addition of MgCl 2 and spermine. The final volume of the reaction mixtures was 20 lL; final concentrations were: 40–200 nm ribozyme, 20 nm short substrate, 44–220 nm long substrate, 10 mm MgCl 2 ,2mm spermine, 15 mm Tris ⁄ HCl. Samples were processed as des- cribed above; parameters of cleavage as well as of ligation were determined from Eaddie–Hofstee plots. Sequence exchange reaction A mixture of ribozyme and substrate in Tris ⁄ HCl (pH 7.5) buffer was heated at 90 °C for 1 min followed by incuba- tion at 37 °C for 15 min. The cleavage reaction was started by addition of MgCl 2 and spermine. The final volume of the reaction mixtures was 20 lL; final concentrations were: 220 nm ribozyme, 220 nm substrate, 11 mm MgCl 2 , 2.2 mm spermine, 16.5 mm Tris ⁄ HCl, pH 7.5. After 30 min, an aliquot (2 lL) was removed from the reaction mixture and substituted by an equal volume con- taining the 20-mer fragment S20 cp (2 lL), such that ori- ginal RNA fragments (ribozyme, initial substrate and added 20-mer) had a final concentration of 200 nm. Aliqu- ots (2.5 lL) were taken at suitable time intervals and added to 3 lL of stop-mix on ice. Samples were analysed using an ALF DNA sequencer, and data were processed using alf fragment manager as described previously [15]. When the reaction was carried out with the Cy5-labelled 20-mer S20 cp(Cy5), samples were also analysed using a DNA Sequencer 4200 (LI-COR Biosciences, Bad Homburg, Ger- many) and the data were processed with gene imagir 4.05 (LI-COR). Melting temperatures and thermodynamic parameters To generate duplexes individual amounts of oligonucleotides (for sequence see Fig. 2B) were mixed in reaction buffer con- taining 10 mm Tris ⁄ HCl, pH 7.5 and 10 mm MgCl 2 , heated at 90 °C for 5 min followed by slow cooling to 20 °C. Individual samples with duplex concentrations of 500 nm, 750 nm,1lm, 2.5 lm and 5 lm have been prepared. Melting profiles were measured at 260 nm in 1 mL quartz cells using a UV visible spectrometer CARY 1E (VARIAN INC., Palo Alto, USA) equipped with a temperature control device. Thermostat temperature was varied from 20 to 90 °C with a heating rate of 0.2 °CÆmin )1 . To avoid solvent evaporation 100 lL mineral oil was placed on the sample surface. Melting points T m (°C) were defined as the maxima of the first derivation. Individual values for DH and DS were obtained by plotting R ln[C] against 1 ⁄ T m (K )1 ). The free enthalpy of the exchange of the short fragment (Fig. 2, red) for the longer fragment (Fig. 2, green) on the ribozyme at 298 K (DG) was defined as DG(298 K) ¼ DG product duplex (298 K) – DG educt duplex (298 K). Acknowledgements Financial support by DFG and Ju ¨ rgen Manchot Foundation as well as a PhD studentship to SV by the Ju ¨ rgen Manchot Foundation is gratefully acknow- ledged. References 1 Kurreck J (2003) Antisense technologies. Improvement through novel chemical modifications. Eur J Biochem 270, 1628–1644. 2 Scanlon KJ (2004) Anti-genes: siRNA, ribozymes and antisense. Curr Pharm Biotechnol 5, 415–420. 3 Long MB, Jones JP 3rd, Sullenger BA & Byun J (2003) Ribozyme-mediated revision of RNA and DNA. J Clin Invest 112, 312–318. 4 Puerta-Fernandez E, Romero-Lopez C, Barosso-del- Jesus A & Berzal-Herranz A (2003) Ribozymes: recent advances in the development of RNA tools. FEMS Microbiol Rev 27, 75–97. 5 Rimmele M (2003) Nucleic acid aptamers as tools and drugs: recent developments. Chembiochem 4, 963–971. 6 Alisky JM & Davidson BL (2004) Towards therapy using RNA interference. Am J Pharmacogenom 4, 45– 51. 7 Berkhout A (2004) RNA interference as an antiviral approach: targeting HIV-1. Curr Opin Mol Ther 6, 141–145. 8 Burgstaller P, Jenne A & Blind M (2002) Aptamers and aptazymes: accelerating small molecule drug discovery. Curr Opin Drug Discov Devel 5, 690–700. 9 Famulok M & Verma S (2002) In vivo-applied func- tional RNAs as tools in proteomics and genomics research. Trends Biotechnol 20, 462–466. 10 Silverman SK (2003) Rube Goldberg goes (ribo) nuc- lear? Molecular switches and sensors made from RNA. RNA 9, 377–383. S. A. Ivanov et al. An engineered ribozyme for RNA sequence exchange FEBS Journal 272 (2005) 4464–4474 ª 2005 FEBS 4473 [...]... Site-directed alteration of RNA ¨ sequence mediated by a small engineered twin ribozyme Angew Chem Int Ed 42, 2424–2427 17 Muller S (2003) Engineered ribozymes as molecular ¨ tools for site-specific alteration of RNA sequence Chembiochem 4, 991–997 ´ 18 Vauleon S, Ivanov SA, Gwiazda S & Muller S (2005) ¨ Site-specific fluorescent and affinity labeling of RNA using a small engineered ribozyme Chembiochem in... Correlating structural dynamics and function in single ribozyme molecules Science 296, 1473–1476 23 Nahas MK, Wilson TJ, Hohng S, Jarvie K, Lilley DMJ & Ha T (2004) Observation of internal cleavage and ligation reactions of a ribozyme Nat Struct Mol Biol 11, 1107–1113 24 Nesbitt SM, Erlacher HA & Fedor MJ (1999) The internal equilibrium of the hairpin ribozyme: temperature, ion and pH effects J Mol Biol... S ¨ (2004) Chemical synthesis of an artificially branched hairpin ribozyme variant with RNA cleavage activity Tetrahedron 60, 9273–9281 29 Komatsu Y, Kanzaki I, Koizumi M & Ohtsuka E (1995) Modification of primary structures of hairpin ribozymes for probing active conformations J Mol Biol 252, 296–304 30 Komatsu Y, Kanzaki I & Ohtsuka E (1996) Enhanced folding of hairpin ribozymes with replaced domains... Biol 286, 1009–1024 25 Fedor MJ (1999) Tertiary structure stabilization promotes hairpin ribozyme ligation Biochemistry 38, 11040–11050 4474 S A Ivanov et al 26 Fedor MJ (2000) Structure and function of the hairpin ribozyme J Mol Biol 297, 269–291 27 Welz R, Schmidt C & Muller S (2001) Spermine sup¨ ports catalysis of hairpin ribozyme variants to differing extents Biochem Biophys Res Commun 283, 648–654... regulation: from transcription attenuation to riboswitches and ribozymes Trends Microbiol 12, 473–475 14 Chowdhury S, Ragaz C, Kreuger E & Narberhaus F (2003) Temperature-controlled structural alterations of an RNA thermometer J Biol Chem 278, 47915–47921 15 Schmidt C, Welz R & Muller S (2000) RNA double ¨ cleavage by a hairpin- derived twin ribozyme Nucleic Acids Res 28, 886–894 16 Welz R, Bossmann... Ferre-D’Amare AR (2001) Crystal structure of a hairpin ribozyme inhibitor complex with implications for catalysis Nature 410, 780–786 32 Muller S, Welz R, Ivanov S & Bossmann K (2004) ¨ Twin ribozymes In Highlights in Bioorganic Chemistry: Methods and Application (Schmuck C & Wennemers H, eds), pp 404–419 Wiley-VCH, Weinheim 33 Drude I (2005) Design und Synthese von Twinribozymen zur Reparatur von Insertions... Large-scale genotyping by mass spectrometry: experience, advances and obstacles Curr Opin Chem Biol 6, 31–38 38 Sauer S, Gelfand DH, Boussicault F, Bauer K, Reichert F & Gut IG (2002) Facile method for automated genotyping of single nucleotide polymorphisms by mass spectrometry Nucleic Acids Res 30, 22 39 Perez-Ruiz M, Barroso-delJesus A & Berzal-Herranz A (1999) Specificity of the hairpin ribozyme Sequence... NG, Kotzorek G, Heckman JE & Burke JM (1998) Structural basis for heterogenic kinetics: reengineering the hairpin ribozyme Proc Natl Acad Sci USA 95, 6091–6096 35 Murchie AIH, Thomson JB, Walter F & Lilley DMJ (1998) Folding of the hairpin ribozyme in its natural conformation achieves close physical proximity of the loops Mol Cell 1, 873–881 36 Kirk BW, Feinsod M, Favis R, Kliman RM & Barany F (2002)... sequences participating in the autolytic processing of satellite tobacco ringspot virus complementary RNA Gene 82, 53–61 20 Hampel A & Tritz R (1989) RNA catalytic properties of the minimum (-) sTRSV sequence Biochemistry 28, 4929–4933 ´ 21 Rupert PB, Massey AP, Sigurdsson ST & Ferre´ D’Amare AR (2002) Transition state stabilization by a catalytic RNA Science 298, 1421–1424 22 Zhuang X, Kim H, Pereira...An engineered ribozyme for RNA sequence exchange 11 Mandal M & Breaker RR (2004) Gene regulation by riboswitches Nat Rev Mol Cell Biol 5, 451–463 12 Mandal M, Lee M, Barrick JE, Weinberg Z, Emilsson GM, Ruzo WL & Breaker RR (2004) A glycine-dependent riboswitch that uses cooperative . Efficient RNA ligation by reverse-joined hairpin ribozymes and engineering of twin ribozymes consisting of conventional and reverse-joined hairpin ribozyme. ability of this twin ribozyme to mediate site-specific alteration of RNA sequence is demonstrated. Results RNA ligation by reverse-joined hairpin ribozymes Reverse-joined