Search

The Tetrahymena pre-rRNA self-splicing intron is shown to function in the unnatural context of an mRNA transcribed by RNA polymerase II in mammalian cells. Mutational analysis supports the conclusion that splicing in cells occurs by the same RNA-catalyzed mechanism established for splicing in vitro. Insertion of the intron at five positions spanning the luciferase open reading frame revealed 10-fold differences in accumulation of ligated exons and in luciferase activity; thus, the intron self-splices in many exon contexts, but the context can have a significant effect on activity. In addition, even the best self-splicing constructs, which produced half as much mRNA as did an uninterrupted luciferase gene, gave approximately 100-fold less luciferase enzyme activity, revealing an unexpected discontinuity between mRNA production and translation in cells. The finding that production of accurately spliced mRNA in cells does not guarantee a corresponding level of protein production is surprising, and may have implications for the development of trans-splicing ribozymes as therapeutics.

Group I introns possess a conserved guanosine residue at their 3' end, termed omegaG, that, in the case of the Tetrahymena pre-rRNA, is a major determinant of the second step of splicing. We examined the role of omegaG in self-splicing of the 249-residue group I intron of the Anabaena PCC7120 tRNAleu precursor. Contrary to observations with the Tetrahymena pre-rRNA intron, a mutation that places an adenosine residue at the omega position did not have a severe effect on the second step of splicing; neither 3' splice-site selection nor the rate of the second step was altered. The first step of splicing, however, was now readily reversed. This unexpected effect also resulted from a mutation that altered the nucleoside specificity of the intronic guanosine-binding site. The theme common to these mutations is that reversal of the first step of splicing results when there is not a strong interaction between the guanosine-binding site and the omega residue. This suggests that a major role of omegaG is to compete with the exogenous guanosine molecule added to the intron in the first step of splicing for the single guanosine-binding site of the intron. From these data, we are able to extend the mechanism for the self-splicing reaction of this intron by proposing two distinct conformational changes between the first and second steps of the splicing. The first of these is the exchange of the exogenous nucleoside for the omega nucleoside. This is the equilibrium that we can perturb by mutations at either the omega position or the guanosine-binding site. An additional conformational change then fully activates the intron for the second step of splicing.

A phylogenetically conserved guanine.uracil (G.U) pair defines the 5'-exon/intron boundary of precursor RNAs containing group I introns. In this wobble base pair, the G forms two hydrogen bonds with U in a base pairing geometry shifted from that of a canonical Watson-Crick pair. On the basis of thermodynamic measurements of synthetic base pair analogs (inosine, diaminopurine riboside, guanosine, or adenosine paired with U, C, or isocytidine) in place of the G.U pair, we have previously reported that the N2 exocyclic amine of the G is important for docking the 5'-exon into the active site of the Tetrahymena ribozyme [Strobel, S. A., & Cech, T. R. (1995) Science 267, 675-679]. Here we describe kinetic characterization of ribozyme-substrate combinations containing the same series of analogs. By measuring the rate constants of 5'-exon miscleavage (cleavage at incorrect phosphates), we demonstrate that the 5'-exon/intron boundary is primarily defined by the exocyclic amine of the G. The amine makes its contribution (2.5 kcal.mol-1) in the context of all three wobble pairs tested but fails to make a significant contribution (< 0.8 kcal.mol-1) when presented in a Watson-Crick base pairing geometry. We also demonstrate that the exocyclic amine makes a modest contribution to chemical transition state stabilization (1.0 kcal.mol-1 relative to an inosine-U pair). The majority of this transition state contribution (0.7 kcal.mol-1) is independent of that contributed by the 2'-hydroxyl of the neighboring U. This argues against the model in which substantial transition state stabilization is derived from a water molecule bridging between the exocyclic amine of G and the 2'-hydroxyl of U. Instead it suggests that the tertiary interaction between the exocyclic amine and its hydrogen bonding partner in the active site is slightly improved during the chemical transition. We conclude that the exocyclic amine of G is the primary contributor to many characteristics of reactivity that have been ascribed to the conserved G.U pair, including stabilization of the chemical transition state and definition of the 5'-exon/intron boundary.

The yeast mitochondrial group I intron b15 undergoes self-splicing at high Mg2+ concentrations, but requires the splicing factor CBP2 for reaction under physiological conditions. Chemical accessibility and UV cross-linking experiments now reveal that self-processing is slow because functional elements are not properly positioned in an active tertiary structure. Folding energy provided by CBP2 drives assembly of two RNA domains that comprise the catalytic core and meditates association of an approximately 100 nt 5' domain that contains the 5' splice site. Thus, the protein assembles RNA secondary structure elements into a specific three-dimensional array while the RNA provides the catalytic center. The division of labor between RNA and protein illustrated by this simple system reveals principles applicable to complex ribonucleoprotein assemblies such as the spliceosome and ribosome.

The splicing factor CBP2 is required to excise the yeast mitochondrial group I intron bI5 in vivo and at low magnesium ion concentrations in vitro. CBP2 binding is strengthened 20-fold by increasing Mg2+ concentrations from 5 to 40 mM, implying the protein binds, in part, to the same structure as that stabilized by the cation. The same transition is also observed as a cooperative increase in the rate of self-processing between 5 and 40 mM Mg2+, providing strong evidence for an RNA folding transition promoted by either Mg2+ or CBP2. The first step of splicing, guanosine addition at the 5' splice site, is rate limiting for exon ligation. At low (5 mM) magnesium ion, reaction (measured as kcat/Km or kcat) is accelerated 3 orders of magnitude by saturating CBP2. At near-saturating Mg2+ (40 mM), acceleration is 8- and 30-fold, for kcat and kcat/Km, respectively, so high magnesium ion concentrations fail to compensate completely for protein facilitation. Thus, self-splicing proceeds via two additional transitions as compared with reaction of the bI5-CBP2 complex, only the first of which is efficiently promoted by the cation. Guanosine 5'-monophosphate binds (Kd approximately 0.3 mM) with the same affinity to bI5 and the bI5-protein complex, supporting independent binding of the nucleophile and CBP2. Substitution of a phosphorothioate at the 5' splice site and pH profiles provide evidence that kcat is limited by chemistry at low pH and by a conformational step at high pH. Because binding by either Mg2+ or CBP2 increases the rate of chemistry more than the rate of the conformational step, in the physiological pH range (7-7.6) the protein-facilitated reaction is limited by a conformation step while self-splicing reaction is limited by chemistry. We conclude that CBP2 makes manifold contributions to bI5 splicing: binding compensates for at least two structural defects and accelerates the rate of the chemistry.

Ribozymes can be targeted to cleave specific RNAs, which has led to much interest in their potential as gene inhibitors. Such trans-cleaving ribozymes join a growing list of agents that stop the flow of genetic information. Here we describe a different application of ribozymes for which they may be uniquely suited. By targeted trans-splicing, a ribozyme can replace a defective portion of RNA with a functional sequence. The self-splicing intron from Tetrahymena thermophila was previously shown to mediate trans-splicing of oligonucleotides in vitro. As a model system for messenger RNA repair, this group I intron was re-engineered to regenerate the proper coding capacity of short, truncated lacZ transcripts. Trans-splicing was efficient in vitro and proceeded in Escherichia coli to generate translatable lacZ messages. Targeted trans-splicing represents a general means of altering the sequence of specified transcripts and may provide a new approach to the treatment of many genetic diseases.

The utilization of cryptic splice sites has been observed in a number of RNA splicing reactions. In the self-splicing group I intron of Tetrahymena thermophila, point mutations of either A57 or A95 promote cleavage at two sites other than the normal 5' splice site, suggesting that these nucleotides are involved in a common tertiary interaction. These results are unusual since A57 and A95 are neither at nor near the 5' splice site in the sequence or secondary structure. Cleavage at the alternative sites appears to occur by intron cyclization, a reaction with well-established structural and mechanistic similarities to the first step of RNA self-splicing. Alternative docking of P1 (the helix containing the 5' splice site paired to the internal guide sequence of the intron) into the catalytic core accounts for cleavage at the cryptic reaction sites. We propose that the A57/A95 interaction, along with an element implicated previously (J1/2), provide structural connectivity from the reaction site in P1 to the catalytic core of the Tetrahymena intron. It seems likely that RNA splicing in general will require such tertiary interactions to position RNA helices.

Group I introns, which are widespread in nature, carry out RNA self-splicing. The secondary structure common to these introns was for the most part established a decade ago. Information about their higher order structure has been derived from a range of experimental approaches, comparative sequence analysis, and molecular modelling. This information now provides the basis for a new two-dimensional structural diagram that more accurately represents the domain organization and orientation of helices within the intron, the coaxial stacking of certain helices, and the proximity of key nucleotides in three-dimensional space. It is hoped that this format will facilitate the detailed comparison of group I intron structures.

In the cyanobacterium Anabaena, the precursor to tRNA(Leu) has a 249-nucleotide group I intron inserted between the wobble and second bases of the anticodon; the intron self-splices during transcription in vitro [Xu, M. Q., Kathe, S. D., Goodrich-Blair, H., Nierzwicki-Bauer, S. A., & Shub, D. A. (1990) Science 250, 1566-1570]. By studying splicing of isolated pre-tRNA, we confirm that splicing occurs by the two-step transesterification mechanism characteristic of group I introns, resulting in excision of the intron and accurate ligation of the 5' and 3' exons. The first step, guanosine-dependent cleavage of the phosphodiester bond at the 5' splice site, occurs with kcat congruent to 14 min-1 and kcat/Km = 5 x 10(4) M-1 min-1 (32 degrees C, 15 mM MgCl2), unexpectedly efficient for a small group I intron. (kcat/Km is comparable to that of the Tetrahymena pre-rRNA intron, and kcat is an order of magnitude higher than any previously reported for a group I intron). The second step, ligation of the exons, is so slow (k = 0.3 min-1) that it is rate-limiting for splicing in vitro except at very low guanosine concentrations. Disruption of the base pairs that make up the anticodon stem of the tRNA dramatically reduces the rate of the first step of splicing, while compensatory mutations that restore base pairing generally restore activity. We suggest that the very short P1 helix of this pre-tRNA, with only three base pairs preceding the 5' splice site, is unstable without the additional base pairs in the anticodon stem.(ABSTRACT TRUNCATED AT 250 WORDS)

HeLa cell nuclear proteins that bind to single-stranded d(TTAGGG)n, the human telomeric DNA repeat, were identified and purified by a gel retardation assay. Immunological data and peptide sequencing experiments indicated that the purified proteins were identical or closely related to the heterogeneous nuclear ribonucleoproteins (hnRNPs) A1, A2-B1, D, and E and to nucleolin. These proteins bound to RNA oligonucleotides having r(UUAGGG) repeats more tightly than to DNA of the same sequence. The binding was sequence specific, as point mutation of any of the first 4 bases [r(UUAG)] abolished it. The fraction containing D and E hnRNPs was shown to bind specifically to a synthetic oligoribonucleotide having the 3' splice site sequence of the human beta-globin intervening sequence 1, which includes the sequence UUAGG. Proteins in this fraction were further identified by two-dimensional gel electrophoresis as D01, D02, D1*, and E0; intriguingly, these members of the hnRNP D and E groups are nuclear proteins that are not stably associated with hnRNP complexes. These studies establish the binding specificities of these D and E hnRNPs. Furthermore, they suggest the possibility that these hnRNPs could perhaps bind to chromosome telomeres, in addition to having a role in pre-mRNA metabolism.

The natural splice junction of the Tetrahymena large ribosomal RNA is flanked by hairpins that are phylogenetically conserved. The stem immediately preceding the splice junction involves nucleotides that also base pair with the internal guide sequence of the intervening sequence during splicing. Thus, precursors which contain wild-type exons can form two alternative helices. We have constructed a series of RNAs where the stem-loop in the 5' exon is more or less stable than in the wild-type precursor, and tested them in both forward and reverse self-splicing reactions. The presence of a stable hairpin in ligated exon substrates interferes with the ability of the intervening sequence to integrate at the splice junction. Similarly, the presence of the wild-type hairpin in the 5' exon reduces the rate of splicing 20-fold in short precursors. The data are consistent with a competition between unproductive formation of a hairpin in the 5' exon and productive pairing of the 5' exon with the internal guide sequence. The reduction of splicing by a hairpin that is a normal feature of rRNA structure is surprising; we propose that this attenuation is relieved in the natural splicing environment.

A ribozyme derived from the intervening sequence (IVS) of the Tetrahymena preribosomal RNA catalyzes a site-specific endonuclease reaction: G2CCCUCUA5 + G in equilibrium with G2CCCUCU + GA5 (G = guanosine). This reaction is analogous to the first step in self-splicing of the pre-rRNA, with the product G2CCCUCU analogous to the 5'-exon. The following mechanistic conclusions have been derived from pre-steady-state and steady-state kinetic measurements at 50 degrees C and neutral pH in the presence of 10 mM Mg2+. The value of kcat/Km = 9 x 10(7) M-1 min-1 for the oligonucleotide substrate with saturating G represents rate-limiting binding. This rate constant for binding is of the order expected for formation of a RNA.RNA duplex between oligonucleotides. (Phylogenetic and mutational analyses have shown that this substrate is recognized by base pairing to a complementary sequence within the IVS). The value of kcat = 0.1 min-1 represents rate-limiting dissociation of the 5'-exon analogue, G2CCCUCU. The product GA5 dissociates first from the ribozyme because of this slow off-rate for G2CCCUCU. The similar binding of the product, G2CCCUCU, and the substrate, G2CCCUCUA5, to the 5'-exon binding site of the ribozyme, with Kd = 1-2 nM, shows that the pA5 portion of the substrate makes no net contribution to binding. Both the substrate and product bind approximately 10(4)-fold (6 kcal/mol) stronger than expected from base pairing with the 5'-exon binding site. Thus, tertiary interactions are involved in binding. Binding of G2CCCUCU and binding of G are independent. These and other data suggest that binding of the oligonucleotide substrate, G2CCCUCUA5, and binding of G are essentially random and independent. The rate constant for reaction of the ternary complex is calculated to be kc approximately equal to 350 min-1, a rate constant that is not reflected in the steady-state rate parameters with saturating G. The simplest interpretation is adopted, in which kc represents the rate of the chemical step. A site-specific endonuclease reaction catalyzed by the Tetrahymena ribozyme in the absence of G was observed; the rate of the chemical step with solvent replacing guanosine, kc(-G) = 0.7 min-1, is approximately 500-fold slower than that with saturating guanosine. The value of kcat/Km = 6 x 10(7) M-1 min-1 for this hydrolysis reaction is only slightly smaller than that with saturating guanosine, because the binding of the oligonucleotide substrate is predominantly rate-limiting in both cases.(ABSTRACT TRUNCATED AT 400 WORDS)

Proteins are not the only catalysts of cellular reactions; there is a growing list of RNA molecules that catalyze RNA cleavage and joining reactions. The chemical mechanisms of RNA-catalyzed reactions are discussed with emphasis on the self-splicing ribosomal RNA precursor of Tetrahymena and the enzymatic activities of its intervening sequence RNA. Wherever appropriate, catalysis by RNA is compared to catalysis by protein enzymes.

The sequence of the cloned Tetrahymena ribosomal RNA intervening sequence (IVS) was altered at the site to which circularization normally occurs. The alterations caused circularization to shift to other sites, usually a nearby position which followed three pyrimidines. While a tripyrimidine sequence was the major determinant of a circularization site, both location of a sequence and local secondary structure may influence the use of that sequence. For some constructs circularization appeared to occur at the position following the 5' G, the nucleotide added to the IVS during its excision. Portions of the internal guide sequence (IGS), proposed to interact with the 3'exon were deleted without preventing exon ligation. Thus if the IGS-3'exon interaction exists, it is not essential for splicing in vitro.

The rRNA intervening sequence of Tetrahymena is a catalytic RNA molecule, or "ribozyme." A tertiary-structure model of the active site of this ribozyme has been constructed based on comparative sequence analysis of related group I intervening sequences, data on the accessibility of each nucleotide to chemical and enzymatic probes, and principles of RNA folding derived from a consideration of the structure of tRNA determined by x-ray crystallography. In the model, the catalytic center has a two-helix structural framework composed of the base-paired segments of the group I conserved sequence elements. The structural framework supports and orients the conserved nucleotides that are adjacent to the base-paired sequence elements; these conserved nucleotides are proposed to form the active site and to bind the 5' splice-site duplex and the guanine nucleotide substrate. Tests of the model are proposed.

Tetrahymena preribosomal RNA undergoes self-splicing in vitro. The structural components involved in recognition of the 5' splice site have been identified, but the mechanism by which the 3' splice site is recognized is not established. To identify some components of 3'splice site recognition, we have generated mutations near the 3' splice site and determined their effects on self-splicing. Alteration of the 3'-terminal guanosine of the intervening sequence (IVS), a conserved nucleotide in group I IVSs, almost eliminated 3' splice site activity; the IVS-3' exon splicing intermediate accumulated, and exon ligation was extremely slow. These mutations do not result in recruitment of cryptic 3' splice sites, in contrast to mutations that affect the 5' splice site. Alteration of the cytidine preceding the 3'-terminal guanosine or of the first two nucleotides of the 3' exon had similar but less severe effects on exon ligation. Most of the mutants showed some reduction (less than threefold) in GTP addition at the 5' splice site. A mutation that placed a new guanosine residue just upstream from the 3'-terminal guanosine misspliced to produce ligated exons with one extra nucleotide between the 5' and 3' exons. We conclude that multiple nucleotides, located both at the 3' end of the IVS and in the 3' exon, are required for 3' splice site recognition.

We have converted the intramolecular cyclization reaction of the self-splicing intervening sequence (IVS) ribonucleic acid (RNA) from Tetrahymena thermophila into an intermolecular guanosine addition reaction. This was accomplished by selectively removing the 3'-terminal nucleotide by oxidation and beta-elimination; the beta-eliminated IVS thereby is no longer capable of reacting with itself. However, under cyclization conditions, a free guanosine molecule can make a nucleophilic attack at the normal cyclization site. We have used this guanosine addition reaction as a model system for a Michaelis-Menten kinetic analysis of the guanosine binding site involved in cyclization. The results indicate that functional groups on the guanine that are used in a G-C Watson-Crick base pair are important for the cyclization reaction. This is the same result that was obtained for the guanosine binding site involved in splicing [Bass, B. L., & Cech, T. R. (1984) Nature (London) 308, 820-826]. Unlike splicing, however, certain additional nucleotides 5' to the guanosine moiety make significant binding contributions. We conclude that the guanosine binding site in cyclization is similar to, but not identical with, the guanosine binding site in splicing. The same binding interactions used in cyclization could help align the 3' splice site of the rRNA precursor for exon ligation. We also report that the phosphodiester bond at the cyclization site is susceptible to a pH-dependent hydrolysis reaction; the phosphodiester bond is somehow activated toward attack by the 3'hydroxyl of a guanosine molecule or by a hydroxyl ion.

Splicing of the ribosomal RNA precursor of Tetrahymena has previously been shown to require no protein in vitro; the cleavage-ligation activity is intrinsic to the RNA molecule. Analysis of the reaction kinetics with guanosine, which is a substrate in the reaction, and with several guanosine analogues suggests that guanosine binds to a specific site on the pre-rRNA. It appears that the RNA, like an enzyme, binds its substrate to promote the rate and specificity of a biological reaction.

The availability of Tetrahymena pre-rRNA of discrete size, produced by transcription of recombinant plasmids with bacteriophage SP6 RNA polymerase, has permitted a more detailed investigation of the self-splicing reaction. The predicted splicing intermediate, the product of cleavage by guanosine at the 5' splice site, was identified. This intermediate was tested in the intermolecular exon ligation reaction and found to be competent to undergo the second step of splicing. These results and others that evaluated the reactivity of the 5' and 3' splice sites independently show that splicing occurs in two separable steps. The 3' splice site was found to be susceptible to site-specific hydrolysis leaving a hydroxyl terminus. This is interpreted as an indication that the 3' splice site is activated for nucleophilic attack in general and for exon ligation in particular. Preliminary evidence for specific hydrolysis at the 5' splice site was also obtained. All of the newly characterized intervening sequence RNA-mediated reactions as well as those found previously are divided into three categories: transesterification by guanosine at sites following two or three pyrimidine nucleotides (and, as a minor reaction, at sites following other guanosine residues); transesterification by oligopyrimidines or by the 5' exon (which terminates with C-U-C-U-C-UOH) at the site following the 3'-terminal guanosine residue of the intervening sequence; and specific hydrolysis at the splice sites. One of the products of the reactions at the 3' splice site is a molecule that contains the 5' exon still attached to the intervening sequence. It has a 3'-terminal GOH and undergoes cyclization both at the normal cyclization site within the intervening sequence and at the 5' splice site. The finding that the splice site can act as a cyclization site, combined with the earlier observation that the normal cyclization site is subject to attack by guanosine mononucleotide, leads us to propose that all these reactions may be occurring in the same active site. Translocation (a conformational change) would then bring different oligopyrimidine sequences into the active site for attack by guanosine. On the basis of the experimental results, a model for the local structure at the active site is described. A key feature of the model is the interaction between the U at the end of the oligopyrimidine sequence, a G residue within the internal guide sequence in the intervening sequence, and another G residue that can be either the attacking group for transesterification or the 3'-terminal G of the intervening sequence.

The RNA molecules transcribed from many eukaryotic genes are interrupted by intervening sequences, which are removed by a process called RNA splicing. One structurally related group of intervening sequences, the group I intervening sequences, are found in a variety of microorganisms. Some of these, including the group I intervening sequence from the ribosomal RNA precursor of Tetrahymena thermophila, have been shown to mediate their own splicing in an RNA-catalyzed reaction. Following its excision from the ribosomal RNA precursor, the Tetrahymena intervening sequence acts as an enzyme, cutting and rejoining RNA substrates.

The self-splicing rRNA intron of Tetrahymena thermophila contains two stem-loop structures (P9.1 and P9.2) near its 3' end that are not conserved among group I introns. As a step toward deriving the smallest active self-splicing RNA, 78 nucleotides encompassing P9.1 and P9.2 have been deleted. This deletion has no effect on the core catalytic activity of the intron, as judged by its ability to catalyze poly(C) polymerization and other related reactions. In contrast, reactions at the 3' splice site of the rRNA precursor--exon ligation and intermolecular exon ligation--take place with reduced efficiency, and exon ligation becomes rate-limiting for self-splicing. Moreover, intermolecular exon ligation with pentaribocytidylic acid is inaccurate, occurring primarily at a cryptic site in the 3' exon. A deletion of 79 nucleotides that disrupts P9, as well as removing P9.1 and P9.2, has more severe effects on both the first and second steps of splicing. P9, a conserved helix at the 5' edge of the deletion point, can form stable alternative structures in the deletion mutants. This aberrant folding may be responsible for the reduced activity and accuracy of reactions with mutant precursors. Analysis of the cryptic site suggests that choice of the 3' splice site may not only depend on sequence but also on proximity to P9. In the course of these studies, evidence has been obtained for an alternative 5' exon-binding site distinct from the normal site in the internal guide sequence.

Oligonucleotide-directed mutagenesis has been used to alter highly conserved sequences within the intervening sequence (IVS) of the Tetrahymena large ribosomal RNA precursor. Mutations within either sequence element 9L or element 2 eliminate splicing activity under standard in vitro splicing conditions. A double mutant with compensatory base changes in elements 9L and 2 has accurate splicing activity restored. Thus, the targeted nucleotides of elements 9L and 2 base-pair with one another in the IVS RNA, and pairing is important for self-splicing. Mutant splicing activities are restored by increased magnesium ion concentrations, supporting the conclusion that the role of the targeted bases in splicing is primarily structural. Based on the temperature dependence, we propose that a conformational switch involving pairing and unpairing of elements 9L and 2 is required for splicing.

Using short oligoribonucleotides as ligated exon substrates, we show that splicing of the Tetrahymena rRNA group I intron is fully reversible in vitro. Incubation of ligated exon RNA with linear intron produces a molecule in which the splice site sequences of the precursor are reformed. Reversal of self-splicing is favored by high RNA concentration, high magnesium and temperature, and the absence of guanosine. 5' exon sequences that can pair with the internal guide sequence of the intron are required, whereas 3' exon sequences are not essential. Integration of the intron into ligated exon substrates that have the ability to form stem-loop structures is reduced at least one order of magnitude over short, unstructured substrates. We propose that the formation of these structures helps drive splicing in the forward direction. We also show that the Tetrahymena intron can integrate into a beta-globin transcript. This has implications for transposition of group I introns.

The specificity of reactions catalyzed by the Tetrahymena pre-rRNA intervening sequence (IVS) was studied using site-specific mutagenesis. Two sequences required for 5' splice-site selection during self-splicing were defined. Single-base changes in either a 5' exon sequence or a 5' exon-binding site within the IVS disrupt their ability to pair and result in inefficient or inaccurate splicing. Combinations that restore complementarity suppress the effect of the single-base changes. Sequence alterations in the 5' exon-binding site also change the specificity of two other reactions: intermolecular exon ligation (trans-splicing) and the enzymatic nucleotidyltransferase activity of the IVS RNA. Thus the substrate specificity of an RNA enzyme can be changed in a manner predictable by the rules of Watson-Crick base-pairing.

The Tetrahymena rRNA intervening sequence (IVS) excises itself from the pre-rRNA and then mediates its own cyclization. We now find that certain di- and trinucleotides with free 3' hydroxyl groups reopen the circular IVS at the cyclization junction, producing a linear molecule with the oligonucleotide covalently attached to its 5' end. This linear molecule recyclizes with release of the added oligonucleotide. Thus the IVS RNA, like an enzyme, lowers the activation energy for both forward and reverse cleavage-ligation reactions. Certain combinations of pyrimidines are required for circle reopening. The most reactive oligonucleotide is UCU. This sequence resembles those preceding the major and minor cyclization sites in the linear IVS RNA (UUU and CCU) and the 5' splice site in the pre-rRNA (UCU). We propose that an oligopyrimidine binding site within the IVS binds the sequences upstream of each of these target sites for cleavage-ligation.

Group I self-splicing introns have a 5' splice site duplex (P1) that contains a single conserved base pair (U.G). The U is the last nucleotide of the 5' exon, and the G is part of the internal guide sequence within the intron. Using site-specific mutagenesis and analysis of the rate and accuracy of splicing of the Tetrahymena thermophila group I intron, we found that both the U and the G of the U.G pair are important for the first step of self-splicing (attack of GTP at the 5' splice site). Mutation of the U to a purine activated cryptic 5' splice sites in which a U.G pair was restored; this result emphasizes the preference for a U.G at the splice site. Nevertheless, some splicing persisted at the normal site after introduction of a purine, suggesting that position within the P1 helix is another determinant of 5' splice site choice. When the U was changed to a C, the accuracy of splicing was not affected, but the Km for GTP was increased by a factor of 15 and the catalytic rate constant was decreased by a factor of 7. Substitution of U.A, U.U, G.G, or A.G for the conserved U.G decreased the rate of splicing by an even greater amount. In contrast, mutation of the conserved G enhanced the second step of splicing, as evidenced by a trans-splicing assay. Furthermore, a free 5' exon ending in A or C instead of the conserved U underwent efficient ligation. Thus, unlike the remainder of the P1 helix, which functions in both the first and second steps of self-splicing, the conserved U.G appears to be important only for the first step.

Certain RNA molecules can mediate their own cleavage or splicing or act as enzymes to promote reactions on substrate RNA molecules. Thus, RNA is not restricted to being a passive carrier of genetic information but can have an active role in directing cellular biochemistry. These findings suggest the possibility that other cellular RNAs, including the RNA components of small nuclear ribonucleoproteins, of the ribosome, and of various ribonucleoprotein enzymes, are catalysts. RNA enzymes (ribozymes) can be used as sequence-specific RNA cleavage agents in vitro, providing useful tools for biochemical studies of RNA. On a more speculative note, ribozymes directed against viral RNAs have the potential of serving as therapeutic agents. Finally, some infectious agents, including hepatitis delta virus and perhaps poliovirus and rhinoviruses, are themselves ribozymes, providing potential targets for pharmaceuticals.

Group I introns include many mitochondrial ribosomal RNA and messenger RNA introns and the nuclear rRNA introns of Tetrahymena and Physarum. The splicing of precursor RNAs containing these introns is a two-step reaction. Cleavage at the 5' splice site precedes cleavage at the 3' splice site, the latter cleavage being coupled with exon ligation. Following the first cleavage, the 5' exon must somehow be held in place for ligation. We have now tested the reactivity of two self-splicing group I RNAs, the Tetrahymena pre-rRNA and the intron 1 portion of the Neurospora mitochondrial cytochrome b (cob) pre-mRNA, in the intermolecular exon ligation reaction (splicing in trans) described by Inoue et al. The different sequence specificity of the reactions supports the idea that the nucleotides immediately upstream from the 5' splice site are base-paired to an internal, 5' exon-binding site, in agreement with RNA structure models proposed by Davies and co-workers and others. The internal binding site is proposed to be involved in the formation of a structure that specifies the 5' splice site and, following the first step of splicing, to hold the 5' exon in place for exon ligation.

The intervening sequence (IVS) of the Tetrahymena thermophila ribosomal RNA precursor undergoes accurate self-splicing in vitro. The work presented here examines the requirement for Tetrahymena rRNA sequences in the 5' exon for the accuracy and efficiency of splicing. Three plasmids were constructed with nine, four and two nucleotides of the natural 5' exon sequence, followed by the IVS and 26 nucleotides of the Tetrahymena 3' exon. RNA was transcribed from these plasmids in vitro and tested for self-splicing activity. The efficiency of splicing, as measured by the production of ligated exons, is reduced as the natural 5' exon sequence is replaced with plasmid sequences. Accurate splicing persists even when only four nucleotides of the natural 5' exon sequence remain. When only two nucleotides of the natural exon remain, no ligated exons are observed. As the efficiency of the normal reaction diminishes, novel RNA species are produced in increasing amounts. The novel RNA species were examined and found to be products of aberrant reactions of the precursor RNA. Two of these aberrant reactions involve auto-addition of GTP to sites six nucleotides and 52 nucleotides downstream from the 3' splice site. The former site occurs just after the sequence GGU, and may indicate the existence of a GGU-binding site within the IVS RNA. The latter site follows the sequence CUCU, which is identical with the four nucleotides preceding the 5' splice site. This observation led to a model where where the CUCU sequence in the 3' exon acts as a cryptic 5' splice site. The model predicted the existence of a circular RNA containing the first 52 nucleotides of the 3' exon. A small circular RNA was isolated and partially sequenced and found to support the model. So, a cryptic 5' splice site can function even if it is located downstream from the 3' splice site. Precursor RNA labeled at its 5' end, presumably by a GTP exchange reaction mediated by the IVS, is also described.

We have substituted all nucleotides at intron nucleotide 260 (N260) in transcripts related to the self-splicing Tetrahymena rRNA. Substitution slightly affects the binding and reaction of GTP with this group I catalytic center; kcat/Km varies over a three-fold range. The base of N260 therefore communicates with the rG site, but is unlikely to bond directly to GTP. Different nucleotides at this position also alter the binding of L-arginine to the intron, measured by inhibition of the reaction with GTP. Effects of similar small magnitude on interaction of RNA with both GTP and L-arginine support the previous argument from kinetic and structural comparison (Yarus, M. (1988) Science 240, 1751) that placed the two ligands of the RNA in the same binding site. G260 RNA shows the greatest affinity for both D- and L-arginine, but uniquely lacks stereoselectivity for the amino acid. Therefore G260 alters spatial relations within the G site, otherwise conserved in C260, U260, and A260 RNA's. Guanyl urea was used as a probe for the G/guanidino H-bonding part of the rG/arginine site. G260 RNA's dissociation constant for guanyl urea is similar to that of the other RNA's, suggesting that G260 RNA is unaltered at the G/guanidino end of the rG/arginine binding site. To account for all observations, we suggest that the G260 substitution alters the relative location of the RNA backbone near the 5' exon-intron junction, making this location more flexible and closer to the alpha-NH3+'s of L- and D-arginine.

Splicing of the Tetrahymena ribosomal RNA precursor is mediated by the folded structure of the RNA molecule and therefore occurs in the absence of any protein in vitro. The Tetrahymena intervening sequence (IVS) has been inserted into the gene for the alpha-donor fragment of beta-galactosidase in a recombinant plasmid. Production of functional beta-galactosidase is dependent on RNA splicing in vivo in Escherichia coli. Thus RNA self-splicing can occur at a rate sufficient to support gene expression in a prokaryote, despite the likely presence of ribosomes on the nascent RNA. The beta-galactosidase messenger RNA splicing system provides a useful method for screening for splicing-defective mutations, several of which have been characterized.

The dinucleotide CpUOH, when incubated with self-splicing Tetrahymena pre-rRNA in the absence of GTP, functions as a 5' exon. It cleaves the precursor exactly at the 3' splice site and becomes covalently ligated to the 3' exon. Other oligonucleotides with sequences that resemble CUCUCU, the sequence at the 3' end of the 5' exon, can add to the 3' exon in this reaction. Such splicing in trans is most readily explained by a site within the intervening sequence that binds the last few nucleotides of the 5' exon. This binding site functions in splice site recognition and is also part of the active site of the ribozyme. The mechanism by which 5' splice sites are selected in Tetrahymena rRNA and group I mitochondrial RNA splicing is like that used in nuclear mRNA splicing, in that it involves specific pairing of bases adjacent to the splice site with a complementary RNA sequence.

Group I introns fold to form an active site to mediate their own RNA splicing. Sequence elements conserved among the available set of 66 group I introns are compiled. Comparative sequence analysis leads to the prediction of some conserved structural features that have not been widely appreciated. The possible significance of conserved nucleotides within base-paired duplexes is discussed; they might be involved in base triplets or alternate pairing interactions.

The intervening sequence (IVS) of the Tetrahymena ribosomal RNA precursor is excised as a linear RNA molecule which subsequently cyclizes itself in a protein-independent reaction. Cyclization involves cleavage of the linear IVS RNA 15 nucleotides from its 5' end and formation of a phosphodiester bond between the new 5' phosphate and the original 3'-hydroxyl terminus of the IVS. This recombination mechanism is analogous to that by which splicing of the precursor RNA is achieved. The circular molecules appear to have no direct function in RNA splicing, and we propose the cyclization serves to prevent unwanted RNA from driving the splicing reactions backwards.

The sequence requirements for splicing of the Tetrahymena pre-rRNA have been examined by altering the rRNA gene to produce versions that contain insertions and deletions within the intervening sequence (IVS). The altered genes were transcribed and the RNA tested for self-splicing in vitro. A number of insertions (8-54 nucleotides) at three locations had no effect on self-splicing activity. Two of these insertions, located at a site 5 nucleotides preceding the 3'-end of the IVS, did not alter the choice of the 3' splice site. Thus the 3' splice site is not chosen by its distance from a fixed point within the IVS. Analysis of deletions constructed at two sites revealed two structures, a hairpin loop and a stem-loop, that are entirely dispensable for IVS excision in vitro. Three other regions were found to be necessary. The regions that are important for self-splicing are not restricted to the conserved sequence elements that define this class of intervening sequences. The requirement for structures within the IVS for pre-rRNA splicing is in sharp contrast to the very limited role of IVS structure in nuclear pre-mRNA splicing.

Site-specific mutagenesis of the self-splicing Tetrahymena intron has been used to investigate the function of C109-G212, a conserved base pair in the P4 stem of group I introns. Mutation of C109 to G affects splicing only slightly, whereas mutation of G212 to A or C reduces the rate of splicing substantially (500-fold reduction in kcat/Km under standard in vitro splicing conditions for the G212C mutant). Splicing activity of the compensatory double mutant (C109G:G212C) is intermediate between those of the two single mutants. Thus, the stability of the P4 stem as well as the identity of the base at position 212 are important for self-splicing. Single and double mutants containing the G212C substitution have a decreased temperature optimum for self-splicing and are partially Mg2+ suppressible, both indicative of structural destabilization. Chemical structure mapping indicates that the mutations do not redirect the global folding of the RNA, but affect the structure locally and at one other site (A183) that is distant in the secondary structure. We propose that, in addition to its pairing in P4, G212 is involved in a base triplet or an alternate base pair that contributes to the catalytically active tertiary structure of the ribozyme.

In previous studies of transcription and splicing of the ribosomal RNA precursor in isolated Tetrahymena nuclei, we found that the intervening sequence (IVS) was excised as a unique linear RNA molecule and was subsequently cyclized. In the present work, transcription at low monovalent cation concentration is found to inhibit splicing and to lead to the accumulation of a splicing intermediate. This intermediate contains splicing activity that either is tightly bound to the RNA or is part of the RNA molecule itself. The intermediate is able to complete the excision of the IVS when it is incubated with a monovalent cation (75 mM (NH4)2SO4), a divalent cation (5-10 mM MgCl2) and a guanosine compound (1 microM GTP, GDP, GMP or guanosine). ATP, UTP, CTP and guanosine compounds without 2' and 3' hydroxyl groups are inactive in causing excision of the IVS. Accurate excision of the IVS, cyclization of the IVS and (apparently) ligation of the 26S rRNA sequences bordering the IVS all take place under these conditions, suggesting that a single activity is responsible for all three reactions. During excision of the IVS, the 3' hydroxyl of the guanosine moiety becomes linked to the 5' end of the IVS RNA via a normal phosphodiester bond. When GTP is used to drive the reaction, it is added intact without hydrolysis. Based on these results, we propose that Tetrahymena pre-rRNA splicing occurs by a phosphoester transferase mechanism. According to this model, the guanosine cofactor provides the free 3' hydroxyl necessary to initiate a series of three transfers that results in splicing of the pre-rRNA and cyclization of the excised IVS.

Ribozymes are RNA molecules that catalyze biochemical reactions. Fe(II)-EDTA, a solvent-based reagent which cleaves both double- and single-stranded RNA, was used to investigate the structure of the Tetrahymena ribozyme. Regions of cleavage alternate with regions of substantial protection along the entire RNA molecule. In particular, most of the catalytic core shows greatly reduced cleavage. These data constitute experimental evidence that an RNA enzyme, like a protein enzyme, has an interior and an exterior. Determination of positions where the phosphodiester backbone of the RNA is on the inside or on the outside of the molecule provides major constraints for modeling the three-dimensional structure of the Tetrahymena ribozyme. This approach should be generally informative for structured RNA molecules.

A shortened form of the intervening sequence of the self-splicing RNA from Tetrahymena thermophila catalyzes sequence-specific cleavage of RNA. Cleavage site selection involves a base-pairing interaction between the substrate RNA and a binding site within the intervening sequence. Single-base changes in this binding site were previously shown to alter substrate specificity in a predictable manner. To examine the generality with which substrate specificity can be altered, six variant catalytic RNAs (ribozymes) have been produced with two- or three-base changes in the active site. Each ribozyme cleaves its predicted substrate. The conditions required for good reactivity and for discrimination against cleavage at mismatched sites vary and were independently determined for each ribozyme.

The intervening sequence (IVS) of the Tetrahymena rRNA precursor catalyzes its own splicing. During splicing the 3'-hydroxyl of guanosine is ligated to the 5' terminus of the IVS. One catalytic strategy of the IVS RNA is to specifically bind its guanosine substrate. Deoxyguanosine (dG) and dideoxyguanosine (ddG) are found to be competitive inhibitors of self-splicing. Comparison of the kinetic parameters (Ki = 1.1 mM for dG; Ki = 5.4 mM for ddG; Km = 0.032 mM for guanosine) indicates that the ribose hydroxyls are necessary for optimal binding of guanosine to the RNA. dG is not a substrate for the reaction even at very high concentrations. Thus, in addition to aiding in binding, the 2'-hydroxyl is necessary for reaction of the 3'-hydroxyl. A second catalytic strategy of the IVS RNA is to enhance the reactivity of specific bonds. For example, the phosphodiester bond at the 3' splice site is extremely labile to hydrolysis. We find that dG and ddG, as well as 2'-O-methylguanosine and 3'-O-methylguanosine, reduce hydrolysis at the 3' splice site. These data are consistent with an RNA structure that brings the 5' and 3' splice sites proximal to the guanosine binding site.

A shortened form of the self-splicing ribosomal RNA (rRNA) intervening sequence of Tetrahymena thermophila acts as an enzyme in vitro. The enzyme catalyzes the cleavage and rejoining of oligonucleotide substrates in a sequence-dependent manner with Km = 42 microM and kcat = 2 min-1. The reaction mechanism resembles that of rRNA precursor self-splicing. With pentacytidylic acid as the substrate, successive cleavage and rejoining reactions lead to the synthesis of polycytidylic acid. Thus, the RNA molecule can act as an RNA polymerase, differing from the protein enzyme in that it uses an internal rather than an external template. At pH 9, the same RNA enzyme has activity as a sequence-specific ribonuclease.