Hydrolytic cleavage by a group I intron ribozyme is dependent on RNA structures not important for splicing Peik Haugen 1, *, Morten Andreassen 2 ,A ´ sa B. Birgisdottir 2 1 and Steinar Johansen 1,2 1 Department of Molecular Biotechnology, Institute of Medical Biology, University of Tromsø, Norway; 2 Faculty of Fisheries and Natural Sciences, Bodø Regional University, Norway DiGIR2 is the group I splicing-ribozyme of the mobile twin- ribozyme intron Dir.S956-1, present in Didymium nuclear ribosomal DNA. DiGIR2 is responsible for intron excision, exon ligation, 3¢-splice site hydrolysis, and full-length intron RNA circle formation. We recently reported that DiGIR2 splicing (intron excision and exon ligation) competes with hydrolysis and subsequent full-length intron circularization. Here we present experimental evidence that hydrolysis at the 3¢-splice site in DiGIR2 is dependent on structural elements within the P9 subdomain not involved in splicing. Whereas the GCGA tetra-loop in P9b was found to be important in hydrolytic cleavage, probably due to tertiary RNA–RNA interactions, the P9.2 hairpin structure was found to be essential for hydrolysis. The most important positions in P9.2 include three adenosines in the terminal loop (L9.2) and a consensus kink-turn motif in the proximal stem. We sug- gest that the L9.2 adenosines and the kink-motif represent key regulatory elements in the splicing and hydrolytic reac- tion pathways. Keywords: Didymium iridis; group I intron; ribozyme hydrolysis; RNA processing; RNA structures. A highly conserved catalytic core is responsible for the self-splicing reactions of group I intron ribozymes [1]. The secondary structures of paired segments (P1–P10 and the optional P11–P17) are organized into three-dimensional domains were P4–P6 and P3–P9 form the catalytic core [2,3]. The available crystal structure of the Tetrahymena intron ribozyme core reveals an active site preorganized to catalysis [3], which appears to contain at least three metal ions directly involved in the reaction [4]. The group I introns can be divided into five main subgroups named IA, IB, IC, ID and IE [2,5]. The great majority of the more than 1200 group I introns recognized within nuclear rDNA belong to the group IC1 and group IE [6]. While the Tetrahymena intron (Tth.L1925) is a prototype group IC1 intron, the Didymium twin-ribozyme intron Dir.S956-1 (and its DiGIR2 deri- vative) is the best studied of the group IE introns. Group I intron splicing is initiated by binding of an exogenous guanosine (exoG) into the guanosine binding site (GBS). Here, exoG is positioned for attack at the 5¢-splice site (SS) and splicing proceeds through two consecutive transesterification steps. In addition to the essential exon splicing reactions, Tth.L1925 also catalyze hydrolytic clea- vage at the 3¢-SS and the formation of truncated intron circles [1,7]. Hydrolytic cleavage at the 3¢-SS is initiated when the last intron nucleotide (TG) binds to the GBS prior to exoG. Splicing and hydrolysis are competing reactions leading to ligated exons and full-length intron circles, respectively [7]. We have identified and examined an unusual category of self-splicing group I introns with a complex structural organization and function [8–11]. These twin-ribozyme introns consist of two distinct ribozymes (GIR1 and GIR2) and a homing endonuclease gene (HEG). The DiGIR2 ribozyme, encoded by the Didymium iridis twin-ribozyme intron Dir.S956-1, catalyses intron splicing as well as a pronounced 3¢-SS hydrolysis and subsequent intron circu- larization in vitro as well as in vivo [7,8,12–14]. DiGIR2 represents the group IE introns, which has a different structural organization than the Tetrahymena group IC1 intron. Here we report structural requirements of the 3¢-SS hydrolysis reaction in DiGIR2, including the immediate 3¢-exon nucleotides, the P9.2 segment, and the GNRA tetra loops in L6 and L9b. Experimental procedures Plasmid constructions and in vitro mutagenesis 3 pDiGIR2 containing the DiGIR2 ribozyme with flanking exons is the basis for most constructs, and is previously described [8]. The P9.2 stem deletion, as well as the L6 and L9 GNRA to UUCG substitutions in DiGIR2, were introduced by using the PCR based Quick Change site- directed mutagenesis kit (Stratagene) in combination with the following PAGE-purified oligonucleotides: pDiGIR2 L6, OP294/OP295; pDiGIR2 L9, OP 296/OP297; and pDiGIR2DP9.2 4 , OP382/OP383. All other plasmid Correspondence to S. Johansen, Department of Molecular Biotechnology, Institute of Medical Biology, University of Tromsø, N-9037 Tromsø, Norway. Fax: + 47 77 645350, Tel.: + 47 77 645367, E-mail: Steinar.Johansen@fagmed.uit.no Abbreviations: GBS, guanosine binding site; HEG, homing endonuclease gene; IGS, internal guide sequence; rDNA, ribosomal DNA; SS, splice site. *Present address: Department of Biological Sciences and Center for Comparative Genomics, University of Iowa, Iowa City, IA 52242–1324, USA. (Received 30 October 2003, revised 24 December 2003, accepted 19 January 2004) Eur. J. Biochem. 271, 1015–1024 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04003.x constructs were made by cloning a PCR product into the SmaI site of a pUC18 vector using the Sure Clone Ligation Kit (Amersham Biosciences). PCR products were inserted randomly into a forward of reverse position, according to the M13 sequences flanking the multiple cloning site. The following primer combinations were used in order to generate PCR products from the pDiGIR2 tem- plate. Di347, OP129/OP347; Di348, OP129/OP348; Di349, OP129/OP349; Di350, OP129/OP350; Inv310–318, OP129/ OP486; Inv312–314, OP129/OP487; 5 D300–327, OP129/ OP488; D 6 291–299/328–333, OP129/OP489; Inv317–318, OP129/OP592; Inv315–316, OP129/OP593; Inv310–311, OP129/OP594; DiGIR2347, OP39/OP347; DiGIR2350, OP39/OP350. The OP129 and OP347 primer combination was used to generate PCR products from the pDiGIR2 L6 and pDiGIR2 L9 templates, resulting in the pDi347 L6 and pDi347 L9 constructs, respectively. OP129 and OP5 were used to amplify a product from pDiGIR2DP9.2 7 and generate the pDi5DP9.2 8 plasmid. Three different L9b mutants were generated from the following primer combi- nations. GUAA, OP559/OP560; GUGA, OP561/OP562; GAAA, OP547/OP548. Mutants in the P5 region were generated from the following primer combinations. CC-GG P5 receptor, OP549/OP550; inverted P5 receptor, OP858/ OP859; P5 hinge, OP860/OP861. Oligonucleotide sequences used in this work are available as supplement at the RNA Research Groups web site at http://www.fagmed.uit.no/ info/imb/amb. In vitro transcription, splicing and hydrolysis reactions Precursor RNAs were transcribed from T7 promoters on linearized plasmids in 25 lL 9 reactions at low magnesium concentration (2 m M MgCl 2 ). [ 35 S]CTP[a 10; 11 S] (10 lCiÆlL )1 ; Amersham Biosciences) was uniformly incorporated into the RNA transcripts. The following plasmids were linea- rized with EcoRI: pDi348, pDi349, pDi350, pDiGIR2, pDiGIR2L6, pDiGIR2L9, pDiGIR2DP9.2 12 , pDiGIR2.347, pDiGIR2.350, pDiGIR2, pDi347L6 and pDi347L9. pDi5DP9.2 13 with HindIII and pDi347 with SalI. RNAs used in quantitative experiments were cut from polyacryl- amide gels with surgical blades and incubated in 400 lL elution buffer (0.3 M NH 4 Ac, 0.1% sodium dodecyl sulfate, 10 m M Tris pH 8 and 2.5 m M EDTA pH 8) on a rotating wheel at 4 °C over night, purified through a 0.45 l M single use filter (Millipore) with a 1 mL syringe, and ethanol precipitated. RNA splicing was performed under self-splicing conditions essentially as described [8]. Hydrolytic cleavage at the 3¢-splice site was started by adding 15 lL of purified RNA (in DEPC 14 -treated water) to 30 lLofpreheated(50°C) buffer. Reactions were incubated under hydrolysis conditions (same as splicing conditions, but without GTP) at 50°Candsamples (5 lL) were collected, added to an equal volume of STOP solution, and immediately frozen at )70°C. Samples were separated on 8 M urea/5% polyacrylamide gels, followed by autoradiography. Computations To quantify RNA signals, phosphoimager screens were scanned after one to several days of exposure and the resulting images were analyzed by using the IMAGEQUANT 3.3 software. The 3¢ hydrolysis products were included as a theoretical value. Quantitative experiments involving impaired activity RNAs were performed once, while other experiments were reproduced between two and five times. The hydrolysis data were fit to the nonlinear first-order decay equation with end-point correction F t ¼ F 4 þ F 0 Â e Àk obs xt previously described [15,16]. Here, k obs is the calculated pseudo-first-order rate constant. Results and discussion In this work we have used a full-length splicing construct (DiGIR2) and a 5¢-truncated DiGIR2 construct (Di347) in mutational studies to analyze hydrolytic cleavage at the 3¢-SS (Fig. 1). Compared to DiGIR2, Di347 construct lacks the 5¢ exon, internal guide sequence (IGS), as well as the P1 and P2 elements. The 3¢ exon sequences are not essential for hydrolysis at the 3¢-splice site Sequences flanking both the Tetrahymena and the Physarum introns [17,18] have been shown to influence ontherateofin vitro splicing. To test for similar effects of the 3¢ exon on DiGIR2 hydrolytic cleavage at the 3¢-SS, mutations were introduced into the eight first positions of the 3¢-exon sequence (Fig. 2A) and analyzed in both the 5¢-truncated and full-length splicing DiGIR2 contexts. Precursor (Pre) RNAs were incubated under splicing conditions in time course experiments and the generated RNA species were separated on 8 M urea/5% polyacrylamide gels. Compared to the wt exon context (Di347; Fig. 2A), no reductions in hydrolytic cleavage of truncated transcripts were observed even when 2–8 exon positions were changed (Di348–50). Di347 and Di350 precursor RNAs were subjected to more extensive time course experiments including quantification of radioactive decay from the gels using phosphoimager screens. Fraction hydrolyzed RNA (Cut) of the precursor was plotted vs. time (Fig. 2B) and fitted into a nonlinear first- order decay equation. The observed rate constants (k obs ) are shown in Fig. 2B below the plot. Results indicate thattheimmediate3¢-exon sequence plays only a minor role in DiGIR2 hydrolysis, which corroborates the recent findings of the bacterial group IC3 ribozymes of Azoar- cus and Synechococcus pre-tRNA [19]. Same mutational changesasinDi350wereintroducedandtestedina DiGIR2 splicing context (DiGIR2.350). A time course experiment of DiGIR2.350 alongside the corresponding wild-type (DiGIR2.347) RNA is shown in Fig. 2C. The results indicate that the 3¢ exon sequence is not important for DiGIR2 splicing (see RNA 5), but some reductions in hydrolytic cleavage at the 3¢-SS (see RNA 3) and subsequent intron circle formation are observed (see RNA 1 and 6). This minor discrepancy between full- length splicing and 5¢-truncated transcripts may be due to RNA interaction of the proposed P10 (Fig. 1B), which is present in the full-length splicing transcript but not the truncated transcript. 1016 P. Haugen et al. (Eur. J. Biochem. 271) Ó FEBS 2004 Fig. 1. Schematic presentation and secondary structure model of the DiGIR2 ribozyme from Didymium iridis. (A) Schematic organization of the Dir.S956-1 intron (named according to [34]), encoding the two group I ribozyme motifs DiGIR1 and DiGIR2, and the I-DirI homing endo- nuclease. The 5¢-and3¢-splice sites (SS) are indicated, and flanking exon sequences are shown as open boxes. Below: the DiGIR2 (lacking DiGIR1 and I-DirIORF)andthe5¢-truncated DiGIR2 RNA (Di347) used in this study. (B) Secondary structure model of DiGIR2 (modified from [7,8]). Intron RNA nucleotides and exon nucleotides are presented as upper case and lower case letters, respectively. Paired segments (P) and intron nucleotides are numbered. Long-range base pairing interactions (P10 and P13) are shown, and the first transcribed nucleotide of the 5¢-truncated Di347RNA(position40)isindicatedbyanarrow 21 . Mutant regions in P5, P6, P9b, P9.2 and 3¢ exon are boxed. Ó FEBS 2004 RNA structures important for ribozyme hydrolysis 1 (Eur. J. Biochem. 271) 1017 GNRA to UUCG substitution of L9b, but not L6, results in reduced 3¢-SS hydrolysis GNRA tetra-loops have been shown to participate in long-range RNA-RNA tertiary interactions in highly structured RNAs, including the group I ribozymes, by interacting with specific receptor structures [2,3,20]. GNRA tetra-loops may also function as local stabilizers of stem-loop structures analogous to the UNCG family of tetra-loops [21]. UNCG tetra-loops have so far not been found to participate in tertiary RNA–RNA inter- actions. The DiGIR2 ribozyme contains two GNRA tetra-loops; a GUAA in L6 and a GCGA in L9b (Fig. 1B). To evaluate the role of L6 and L9b in 3¢-SS hydrolytic cleavage, the GNRA loops were replaced with UUCG, and the corresponding constructs were analyzed in a similar approach as described above. Results from time course experiments involving the 5¢-truncated RNAs are shown in Fig. 3A,B. A minor reduction in observed hydrolytic rate (k obs 0.085–0.077 min )1 ) was observed in the L6 UUCG substitution construct (Di347L6) compared to that of the wild type. However, the L9b UUCG replacement (Di347L9) resulted in a 10 fold reduction of 3¢-SS hydrolysis Fig. 2. Analysis of DiGIR2 3¢ exon sequences in hydrolytic cleavage and self-splicing. (A) Time course experiment (0–30 min) of 5¢-truncated DiGIR2 containing different sequence substitution within the eight first positions of the 3¢ exon. Mutant RNAs were subjected to splicing conditions [40 m M Tris pH 7.5, 10 m M MgCl, 200 m M KCl, 2 m M spermidine, 5 m M dithiothreitol, 0.2 m M GTP]. Exon nucleotides are presented as lower case letters and substituted positions are shaded. Pre, precursor RNA; Cut, 5¢ RNA product; 3¢SS, 3¢-splice site. (B) Di347 and Di350 RNAs subjected to hydrolysis conditions (identical to splicing conditions but without GTP) and plotted as fraction uncleaved precursor (pre/total) vs. time. Curves were fitted to the nonlinear first-order decay equation F t ¼ F 4 +F 0 · e )k obs xt and pseudo-first-order rate constants (k obs )were calculated. k obs variations represent differences between independent trials. RNA bands were quantitated by phosphoimager exposure with IMAGEQUANT version 3.3 software. (C) Self-splicing time course experiments (0–30 min) of DiGIR2.350 and wild-type DiGIR2.347. DiGIR2.347 was constructed in order to generate a DiGIR2 equivalent to DiGIR2.350 (short-3¢ exon sequences). Cir, intron RNA circle; Pre, precursor RNA; 5¢-E, 5¢ exon; Int, Intron; LE, ligated exons. 1018 P. Haugen et al. (Eur. J. Biochem. 271) Ó FEBS 2004 rate (k obs 0.085–0.008 min )1 ). Time course experiments of the corresponding full-length splicing constructs (DiGIR2L6 and DiGIR2L9, respectively) and wild-type DiGIR2 are shown in Fig. 3C. No significant difference with respect to hydrolysis, circle formation, and exon splicing could be observed between processed DiGIR2 and DiGIR2L6 RNAs. We infer that the L6 GUAA tetra-loop is not involved in RNA–RNA tertiary interactions. How- ever, while the L9b substitution in DiGIR2 (DiGIR2L9) does not affect splicing (RNA5), the amounts of hydrolysis (RNA7) and intron circle formation (RNA1) are strongly reduced. These observations are consistent with an RNA– RNA tertiary interaction that involves the L9b GCGA tetra-loop. Search for an L9b tetra-loop receptor motif in P5 L9 GNRA tetra-loops, in combination with specific recep- tors in P5, are common tertiary interactions within group I introns [2,20,22,23]. Two different sequence contexts in P5 of DiGIR2 were tested for a possible receptor role with GCGA L9b. The first sequence context analyzed was based on the findings by Inoue and coworkers [23]. They reported that the J5/5a hinge in the Pneumocystis group IC1 intron may function as a receptor for the L9 GAAA tetra-loop. The correspondent region in DiGIR2 appears to be the P5 internal loop (Fig. 1B). Thus, the internal loop was deleted (D102–107, 121–124), 15 expressed as a DiGIR2 5¢-truncation construct, and analyzed for 3¢-SS hydrolysis. The deletion Fig. 3. Analysis of DiGIR2 L6 and L9 GNRA tetra-loops in hydrolytic cleavage and self-splicing. (A) Time course experiment (0–30 min) of 5¢-truncated variants containing tetra-loop substitutions, subjected to splicing conditions. Di347L6 and Di347L9 have GUAA to UUCG and GCGA to UUCG substitutions in L6 and L9, respectively. (B) Determination of hydrolytic cleavage rates, k obs , at hydrolysis conditions of Di347L6 and Di347L9. (C) Self-splicing time course experiments (0–30 min) of the L6 (DiGIR2L6) and L9 (DiGIR2L9) substitution constructs. See legends to Fig. 2 for abbreviations and experimental conditions. Ó FEBS 2004 RNA structures important for ribozyme hydrolysis 1 (Eur. J. Biochem. 271) 1019 had no effect on hydrolytic activity, consistent with that the P5 internal loop could not serve as a GCGA L9b receptor (data not shown). The second context analyzed includes the second and third base pairs in P5 stem, which harbors a potential CU:AG (positions 97, 98, 128 and 129) receptor motif (Fig. 1B). This motif, at the exact same position in P5, is frequently observed in combination with GNGA L9 tetra- loops in group IA introns [20]. Studies have shown that different GNRA tetra-loops appear to prefer a certain receptor motif, but with a significant cross reaction [24,25]. Here, GUGA, GUAA and GAAA tetra-loops preferen- tially interact with CU:AG, CC:GG and a 11 nt motif, respectively. To test for a possible P9b–P5 interaction, the Di347 construct containing four different P9b GNRA tetra-loop motifs (GCGA, GUGA, GUAA and GAAA), in combination with two different putative P5 receptor motifs at the second and third base pair positions (CU:AG and CC:GG) were analyzed for hydrolytic cleavage. Further- more, the CU:AG motif was inverted to a nonreceptor sequence (GA:UC) and analyzed together with both the wild-type GCGA and the GUGA P9b tetra-loops. Whereas all the GNRA L9b tetra loops tested supported hydrolysis well compared to UUCG, the wild-type tetra-loop (GCGA) was always the most efficient one followed by GUGA. However, no significant reduction in hydrolytic cleavage rate with respect to wild type and mutant P5 constructs could be found (data not shown). In summary, comparative data support a P5 stem receptor [2,20,22], but we were not able to gain further experimental evidence probably due to a significant cross-reaction between the receptor motifs used in our approach. Deletion of P9.2 dramatically reduces 3¢-SS hydrolysis The P9.2 paired segment is present in many nuclear group IC1 and group IE introns, including the Tetrahym- ena intron. However, no clear functional role has been assigned to this peripheral structural element. To test a possible functional importance in splicing and hydrolytic cleavage, a deletion study of the P9.2 element was performed. The first DiGIR2 deletion mutant to be analyzed lacks the P9.2 structure (positions 293–331; DiGIR2DP9.2) 16 . The corresponding 5¢-truncated and full- length splicing constructs were transcribed and analyzed by the same approach as described above. Time course experiments are presented in Fig. 4, and revealed that the P9.2 deletion dramatically reduces 3¢-SS hydrolysis (Fig. 4A). In fact, no hydrolytic cleavage was detected in the 5¢-truncated construct after % 24 h of incubation (data not shown). Surprisingly, the P9.2 deletion appears to increase the splicing efficiency of DiGIR2. Figure 4B shows that DiGIR2)P9.2 RNA was processed to essentially excised intron and ligated exons. Faint signals corresponding to products of 3¢-SS hydrolytic cleavage (RNAs 3 and 7) and intron circle formation (RNAs 1 and 6) are observed, indicating that some hydrolytic activity are still present in the full-length splicing construct compare to 5¢-truncated construct. In order to evaluate the rate of hydrolytic cleavage catalyzed by DiGIR2 and DiGIR2DP9.2 17 , we included the Tetrahymena ribozyme (Tth.L1925) in a comparative analysis. RNA obtained from full-length splicing constructs of the ribozymes was incubated under hydrolysis conditions (without GTP). The results are presented as an autoradio- gram (Fig. 4C) and as a plot of fraction cleaved precursor vs. time (Fig. 4D). DiGIR2 and Tth.L1925 were found to have very similar hydrolytic cleavage rates at their 3¢-SS at this reaction condition with a calculated k obs of 0.080 and 0.073 min )1 , respectively. The P9.2 deletion in DiGIR2 reduces the hydrolysis reaction approximately 10 fold (k obs % 0.007 min )1 ), which is similar to that observed in the P9b tetra-loop substitution mutant (Fig. 3B). Nucleotide positions within the L9.2 are essential for hydrolysis The observation that the P9.2 deletion dramatically affects hydrolytic cleavage, but not splicing, suggests a more direct role in ribozyme hydrolytic function. Two additional P9.2 deletion constructs were thus generated, and include a proximal- (positions 291–299, 328–333) and a distal (posi- tions 300–327) stem deletion (Fig. 5A). The corresponding 5¢-truncated constructs were transcribed and analyzed by the same approach as described above, and found to completely abolish the hydrolytic reaction (data not shown). These results further support an important role of P9.2 in 3¢SS hydrolytic cleavage. We infer that the distal sequences of P9.2 are essential as deletion of positions 300–327 did not support hydrolysis. Furthermore, shortening of P9.2 by the proximal deletion suggests a positional effect of the distal sequences. To test the importance of the P9.2 loop sequence (L9.2), five different substitution mutants were generated in the 5¢-truncation constructs, in vitro transcribed, and subjected to cleavage conditions. Indeed, L9.2 was found to be essential for hydrolytic cleavage as substitution by inverting all the L9.2 positions (positions 310–318; CGCUACAAA to GCGATGTTT) became inactive (k obs less than 0.001 min )1 ;Fig.5B,C). 18 Some group I introns contain terminal loops within the P9 domain that are engaged in long-range base–pairing interactions, e.g. P13 between L9.1 and L2.1, and P17 between L9 and L5 [26]. Interestingly, a putative base–pairing interaction exists in DiGIR2 between L9.2 (pos. 312–314) and L5 (pos. 114–116). However, experiments including the L9.2 mutant (Inv312– 314) excluded this possibility (k obs ¼ 0.077 min )1 vs. 0.085 min )1 of wild type; Figs 5B,C). The remaining L9.2 positions were changed in pairs (310/311, 315/316 and 317/318; Fig. 5B), and the corresponding results are presented in Fig. 5C. All substitution mutants were found to affect hydrolytic cleavage, with most dramatic effect at positions 315–318 (AAAU; k obs ¼ 0.002–0.003 min )1 ). The impaired hydrolytic cleavage of the L9.2 substitu- tions is rescued by high Mg 2+ concentrations in the absence of K + ions. Further biochemical characterizations of the L9.2 mutants were performed, including hydrolysis at different mono- and divalent cation concentrations. The corres- ponding 5¢-truncation constructs were analyzed at three different Mg 2+ concentrations (5, 10 and 50 m M )and 0m M KCl. A surprising observation was that the presence of 200 m M K + (standard conditions) during the reaction has a negative effect on 3¢-SS hydrolysis. Cleavage rates 1020 P. Haugen et al. (Eur. J. Biochem. 271) Ó FEBS 2004 Fig. 4. Analysis of DiGIR2 P9.2 deletion in hydrolytic cleavage and self-splicing. (A)Timecourseexperiment(0–30min)ofDi347andDi5)P9.2 subjected to splicing conditions. (B) Self-splicing time course experiment (0–30 min) of DiGIR2 and DiGIR2DP9.2. The 3¢-SS hydrolysis and intron circle 22;2322;23 formation are strongly reduced in DiGIR2DP9.2 compared to DiGIR2, while splicing appears unaffected. (C) Time course experiment of DiGIR2, Tetrahymena ribozyme (Tth.L1925), and DiGIR2DP9.2, subjected to hydrolysis conditions (without GTP). DiGIR2DP9.2 was incubated up to 21 h (1260 min) in order to complete the hydrolysis reaction for a more accurate calculation of the rate constant. M, RNA size marker. (D) Plot of fraction uncleaved precursor (pre/total) vs. time of the hydrolysis reactions presented in (C), and subsequent determination of corresponding rate constants (k obs ). See legends to Fig. 2 for abbreviations and experimental conditions. 24 Ó FEBS 2004 RNA structures important for ribozyme hydrolysis 1 (Eur. J. Biochem. 271) 1021 were found to be significantly higher at 10 m M Mg 2+ with 0m M K + than 10 m M Mg 2+ with 200 m M K + (compare Figs 5C and 6). Furthermore, the impaired hydrolytic cleavage in the mutant constructs is rescued by increasing the Mg 2+ concentrations, and at 50 m M all the mutants are at, or close to, the wild type level rate (Fig. 6). Inhibition of Mg 2+ -dependent ribozymes by monovalent cations has previously been noted [27–29], and suggested to be due to monovalent cations displacement of Mg 2+ from essential sites within the ribozymes [27]. Experiments using the hammerhead ribozyme showed that instead of having a coordinated stimulating effect on ribozyme activity, Na + ions inhibit divalent ion mediated ribozyme reactions at lower concentrations, while rescuing the negative effect at higher (>3 M ) concentrations [29]. Our observation that the impaired hydrolytic cleavage of the L9.2 substitutions is rescued by high Mg 2+ concentrations only in the absence of monovalent K + ions suggests that magnesium plays an important role in hydrolysis, but is being displaced (maybe from L9.2) in the presence of monovalent ions. Fig. 5. Analysis of DiGIR2 P9.2 substitutions and deletions in hydro- lytic cleavage. (A) Secondary structure presentations of P9.2 and the P9.2 deletions (dashed lines) introduced into 5¢-truncated DiGIR2 ribozymes. (B) Nucleotide substitutions introduced into the L9.2 loop region of the 5¢-truncated DiGIR2. (C) Plot of fraction uncleaved precursor (pre/total) vs. time of the hydrolysis reactions presented in B (hydrolysis conditions), and subsequent determination of corres- ponding rate constants (k obs ). See legends to Fig. 2 for abbreviations and experimental conditions. Fig. 6. The effect of DiGIR2 L9.2 substitutions on hydrolysis by increased Mg 2+ concentration. Time course experiment (0–150 min) of the wild type DiGIR2 and L9.2 substitution constructs schematically presented in Fig. 5B at hydrolysis conditions with different magnesium concentrations, but at 0 m M KCl. See legends to Fig. 2 for abbrevia- tions and experimental conditions. 1022 P. Haugen et al. (Eur. J. Biochem. 271) Ó FEBS 2004 Functional implications of P9.2 in hydrolysis Our finding that P9.2 is an essential structural element in hydrolytic cleavage, but does not affect splicing, contrasts experimental data from the Tetrahymena group IC1 intron. P9.2 has been shown to be nonessential for splicing and 3¢-SS hydrolysis [30], but a deletion construct lacking both P9.1 and P9.2 had a significant decrease in the folding rate of the catalytic core [31]. The detailed global structure of Tetrahymena P9.2 is not known as peripheral extensions outside the catalytic core were not included in the crystal structure determination [3], but Fe(II)ÆEDTA cleavage data and modeling indicate that P9.2 is pointing outwards from thecore[26]. What is the functional role of the essential L9.2 nucleo- tides in DiGIR2 hydrolysis? One possibility is that L9.2 nucleotides participate in tertiary contacts with other parts of the molecule. Whereas all attempts to obtain supporting indication or evidence of regular base-pairing interactions have failed, the important L9.2 adenosines could still be involved in, e.g. a minor helix packing interaction in an unidentified, distally located receptor within DiGIR2 19 [32]. An alternative possibility is that P9.2 may serve a more direct role in hydrolytic cleavage catalysis, perhaps by presenting hydrolysis-dependent metal-ion (e.g. magnes- ium) to the active site, as indicated by results presented in Fig. 6. P9.2 could potentially access the catalytic core-region during hydrolysis analogous to P1 during splicing. Experi- mental data from the Tetrahymena intron have provided strong evidence that the active site contains three magnes- ium ions directly involved in catalysis [4]. The model for transitionstateinteractionswithintheactivesitesuggest that two of the metal ions are bound to the guanosine cofactor and that the third metal ion interacts with a 3¢ atom of the nucleotide preceding the intron. However, to our knowledge none of the metal ions have been specifically assigned to hydrolytic cleavage. Regardless of the exact functional role in hydrolysis of the L9.2 nucleotides, hydrolysis is also dependent on proximal sequences in the P9.2 stem. These sequences harbor a consensus kink (K)-turn motif (Fig. 1B), fre- quently observed in ribosomal RNA and other structural RNAs as a protein binding motif [33]. The K-turn motif in P9.2 appears essential to DiGIR2 hydrolysis as deletion (D291–299/328–333) 20 completely abolish the 3¢-SS hydro- lysis (Fig. 5A). The K-motif in DiGIR2 probably both positions the essential L9.2 nucleotides for their functional role in 3¢-SS hydrolysis, and binds to a specific nuclear protein factor. Recently we described that DiGIR2 processing proceeds in two independent reaction pathways (Fig. 7), one leading to intron splicing and the other to full-length intron circularization via 3¢-SS hydrolysis [7]. Here we suggest that the K-motif in P9.2 represents a key regulatory element between the reaction pathways, and that a protein factor may exists in the Didymium nucleus that participates in the regulation. The observation that splicing appears much more efficient than hydrolysis in vivo compared to in vitro [8,12–14] is consistent with this proposal. Acknowledgements We thank Henrik Nielsen for discussions. This work was founded by grants from Norwegian Research Council, Norwegian Cancer Society, Aakre Foundation for Cancer Research, and Simon Fougner Hart- manns Foundation. References 1. Cech, T.R. (1990) Self-splicing of group I introns. Annu. Rev. Biochem. 59, 543–568. 2. Michel, F. & Westhof, E. (1990) Modelling of the three-dimen- sional architecture of group I catalytic introns based on com- parative sequence analysis. J. Mol. Biol. 216, 585–610. 3. Golden, B.L., Gooding, A.R., Podell, E.R. & Cech, T.R. (1998) A preorganized active site in the crystal structure of the Tetrahymena ribozyme. Science 282, 259–264. 4. Shan, S., Yoshida, A., Sun, S., Piccirilli, J.A. & Herschlag, D. (1999)ThreemetalionsattheactivesiteoftheTetrahymena group Iribozyme.Proc. Natl Acad. Sci. USA 96, 12299–12304. 5. Shu, S.O., Jones, K.G. & Blackwell, M. (1999) A group I intron in the small subunit rRNA gene of Cryptendoxyla hypophloia, an ascomycetous fungus: evidence for a new major class of group I introns. J. Mol. Evol. 48, 493–500. 6. Cannone, J.J., Subramanian, S., Schnare, M.N., Collett, J.R., Du D’Souza, L.M.Y., Feng, B., Lin, N., Madabusi, L.V., Muller, K.M., Pande, N. & Shang, Z., Yu, N. & Gutell, R.R. (2002) The Comparative RNA Web (CRW) Site: an online database of comparative sequence and structure information for ribosomal, intron, and other RNAs. BMC Bioinform. 3, 1–31. 7. Nielsen, H., Fiskaa, T., Birgisdottir, A ˚ .B., Haugen, P., Einvik, C. & Johansen, S. (2003) The ability to form full-length intron RNA circles is a general property of nuclear group I introns. RNA 9, 1464–1475. 8. Decatur, W.A., Einvik, C., Johansen, S. & Vogt, V.M. (1995) Two group I ribozymes with different functions in a nuclear rDNA intron. EMBO J. 14, 4558–4568. 9. Einvik, C., Decatur, W.A., Embley, T.M., Vogt, V.M. & Johansen, S. (1997) Naegleria nucleolar introns contain two group Fig. 7. 25 Schematic representation of the functional implications for hydrolysis in group I intron. The splicing pathway (left) is initiated by exogenous guanosine (exoG) and results in ligated exons and spliced out free intron. This pathway benefits the host organism. The circu- larization pathway (right) is initiated by hydrolysis at the 3¢-splice site (SS), at the [ohgr]G residue (TG), and results in nonspliced exons and full-length (FL) intron circle. This pathway is a selfish feature for the intron [7]. The hydrolysis step is dependent on the P9-domain RNA structures P9.2 and L9b, intron structures not important in the splicing pathway. Ó FEBS 2004 RNA structures important for ribozyme hydrolysis 1 (Eur. J. Biochem. 271) 1023 I ribozymes with different functions in RNA splicing and pro- cessing. RNA 3, 710–720. 10. Einvik, C., Elde, M. & Johansen, S. (1998) Group I twintrons: genetic elements in myxomycete and schizopyrenid amoebo- flagellate ribosomal DNAs. J. Biotechnol. 17, 63–74. 11. Johansen, S., Einvik, C. & Nielsen, H. (2002) DiGIR1 and NaGIR1: naturally occurring group I-like ribozymes with unique core organization and evolved biological role. Biochimica 84, 905–912. 12. Johansen, S. & Vogt, V.M. (1994) An intron in the nuclear ribosomal DNA of Didymium iridis codes for a group I ribozyme and a novel ribozyme that cooperate in self-splicing. Cell 25, 725– 734. 13. Vader, A., Nielsen, H. & Johansen, S. (1999) In vivo expression of the nucleolar group I intron-encoded I-DirI homing endonuclease involves the removal of a spliceosomal intron. EMBO J. 18, 1003– 1013. 14. Vader, A., Johansen, S. & Nielsen, H. (2002) The group I-like ribozyme DiGIR1 mediates alternative processing of pre-rRNA transcripts in Didymium iridis. Eur. J. Biochem. 269, 5804–5812. 15. Jabri, E., Aigner, S. & Cech, T.R. (1997) Kinetics and secondary structure analysis of Naegleria andersoni GIR1, a group I ribo- zyme whose putative biological function is site-specific hydrolysis. Biochemistry 36, 16345–16354. 16. Einvik, C., Nielsen, H., Nour, R. & Johansen, S. (2000) Flanking sequences with an essential role in hydrolysis of a self-cleaving group I-like ribozyme. Nucleic Acids Res. 28, 2194–2200. 17. Rocheleau, G.A. & Woodson, S.A. (1995) Enhanced self-splicing of Physarum polycephalum intron 3 by a second group I intron. RNA 1, 183–193. 18. Cao, Y. & Woodson, S.A. (2000) Refolding of rRNA exons enhances dissociation of the Tetrahymena intron. RNA 6, 1248– 1256. 19. Ikawa, Y., Naito, D., Shiraishi, H. & Inoue, T. (2000) Structure- function relationships of two closely related group IC3 intron ribozymes from Azoarcus and Synechococcus pre-tRNA. Nucleic Acids Res. 28, 3269–3277. 20. Costa, M. & Michel, F. (1995) Frequent use of the same tertiary motif by self-folding RNAs. EMBO J. 14, 1276–1285. 21. Varani, G. (1995) Exceptionally stable nucleic acids hairpins. Annu. Rev. Biophys. Biomol. Struct. 24, 379–404. 22. Jaeger, L., Michel, F. & Westhof, E. (1994) Involvement of a GNRA tetraloop in long–range RNA tertiary interactions. J. Mol. Biol. 236, 1271–1276. 23. Ikawa, Y., Shiraishi, H. & Inoue, T. (2000) A small structural element, Pc-J5/5a, plays dual roles in a group IC1 intron RNA. Biochem. Biophys. Res. Comm. 274, 259–265. 24. Costa, M. & Michel, F. (1997) Rules for RNA recognition of GNRA tetraloops deduced by in vitro selection: comparison with in vivo evolution. EMBO J. 16, 3289–3302. 25. Ikawa, Y., Naito, D., Aono, N., Shiraishi, H. & Inoue, T. (1999) Trans-activation of the Tetrahymena group I intron ribozyme via a non-native RNA–RNA interaction. Nucleic Acids Res. 27, 1859–1865. 26. Lehnert, V., Jaeger, L., Michel, F. & Westhof, E. (1996) New loop–loop tertiary interactions in self-splicing introns of subgroup IC and ID: a complete 3D model of the Tetrahymena thermophila ribozyme. Chem. Biol. 3, 993–1009. 27. Chowrira, B.M., Berzal-Herranz, A. & Bruke, J.M. (1993) Ionic requirements for RNA binding, cleavage, and ligation by the hairpin ribozyme. Biochemistry 32, 1088–1095. 28. Kuo, T.C. & Herrin, D.L. (2000) A kinetically efficient form of the Chlamydomonas self-splicing ribosomal RNA precursor. Biochem. Biophys. Res. Commun. 273, 967–971. 29. Zhou, J.M., Zhou, D.M., Takagi, Y., Kasai, Y., Inoue, A., Baba, T. & Taira, K. (2002) Existence of efficient divalent metal ion- catalyzed and inefficient divalent metal ion-independent channels in reactions catalyzed by hammerhead ribozyme. Nucleic Acids Res. 30, 2374–2382. 30. Ikawa, Y., Ohta, H., Shiraishi, H. & Inoue, T. (1997) Long–range interaction between the P2.1 and P9.1 peripheral domains of the Tetrahymena ribozyme. Nucleic Acids Res. 25, 1761–1765. 31. Zarrinkar, P.P. & Williamson, J.R. (1996) The P9.1 – P9.2 per- ipheral extension helps guide folding of the Tetrahymena ribo- zyme. Nucleic Acids Res. 24, 854–858. 32. Nissen, P., Ippolito, J.A., Ban, N., Moore, P.B. & Steitz, T.A. (2001) RNA tertiary interactions in the large ribosomal subunit: the A-minor motif. Proc.NatlAcad.Sci.USA98, 4899–4903. 33. Klein, D.J., Schmeing, T.M., Moore, P.B. & Steitz, T.A. (2001) The kink-turn: a new RNA secondary structure motif. EMBO J. 20, 4214–4221. 34. Johansen, S. & Haugen, P. (2001) A new nomenclature of group I introns in ribosomal DNA. RNA 7, 935–936. 1024 P. Haugen et al. (Eur. J. Biochem. 271) Ó FEBS 2004 . N., Shiraishi, H. & Inoue, T. (1999) Trans-activation of the Tetrahymena group I intron ribozyme via a non-native RNA RNA interaction. Nucleic Acids Res hydrolysis (RNA7 ) and intron circle formation (RNA1 ) are strongly reduced. These observations are consistent with an RNA RNA tertiary interaction that involves