The nrdB gene of bacteriophage T4 contains a group IA2 intron. We have investigated the kinetics of self-splicing by a shortened variant of nrdB pre-mRNA in the presence of the co-substrates guanosine and 2'-amino-2'-deoxyguanosine. The pH dependence of the first transesterification step displayed parallel linear correlations for the two different co-substrates up to pH 7, above which the reaction with guanosine levels off to become pH independent. The plot for the 30-fold slower reaction with 2'-aminoguanosine is linear up to pH 8-8.5 and then levels off. The linear correlations with slopes close to unity suggest that a deprotonation event accelerates the transesterification reaction and that a change in rate limiting step occurs at a first order rate constant of ~1 min-1 (i.e. for our system kcat/Km [approx] 105 M-1 min-1). The pH dependence of observed rate constants in different divalent metal ion mixtures, where the 2'-aminoguanosine-dependent reaction is enhanced 6- and 35-fold compared with that in magnesium, strongly supports this conclusion. This is, to our knowledge, the first report on an intact self-splicing group I intron where use of different co-substrates and divalent metal ions shows that a deprotonation enhances the rate and verifies that the transitions occurring during splicing of group I introns are all part of a common reaction sequence.
The self-splicing reaction by group I introns is catalysed by a guanosine co-substrate and consists of two consecutive transesterification reactions (1 ,2 ). The first transesterification involves a nucleophilic attack of the 3'-OH group of the guanosine co-substrate at the 5' splice site, leaving the guanosine residue attached to the intron and a free 3'-hydroxyl on the 5'-exon. The second transesterification involves nucleophilic attack of the free 3'-OH group of exon 1 at the 3' splice site and yields ligated exons and liberation of the co-substrate-intron species.
Divalent metal ions have been shown to promote the group I intron self-splicing reaction (1 ,2 ). Their role is suggested to be two-fold: structural and catalytic. For example, magnesium or manganese can promote both folding and catalysis of the Tetrahymena group I intron. We recently showed that a divalent metal ion interacts with the co-substrate and takes part in catalysis of the transesterification. These results support a two metal ion mechanism in self-splicing of group I introns (3 ).
Structural transitions facilitate formation of the different interactions that specify the 5' and 3' splice sites of the RNA precursor (1 ,4 -8 ). One major structural transition is the docking of helix P1 with the 5' splice site into the catalytic core of the intron. A conformational change is also proposed to occur when guanosine binds to the core of the Tetrahymena intron (9 ,10 ). Another structural transition comprises release of the bound co-substrate end of the intron and docking of the 3' splice site, a step assumed to engage the cleaved P1 loop structure in the so called P10 interaction with sequences at the 3' splice site.
Several self-splicing introns have been engineered into ribozymes capable of multiple turnover, thereby facilitating detailed kinetic analysis of the first transesterification step (11 -14 ). The characteristic pH rate profiles for such ribozymes show a biphasic behaviour, suggesting that the rate limiting chemical step is enhanced by deprotonation, but at higher pH is replaced by another pH- independent rate limiting step which may involve a conformational change (9 ,13 ). In this study we have utilized an intact group I intron from the bacteriophage T4 nrdB gene in order to study the pH dependence of a system where both transesterification steps are operating (15 ). The previously demonstrated functional co-substrate analogue 2'-amino-2'-deoxyguanosine (3 ,16 ) was used to monitor transitions between different rate limiting steps in the pH-dependent catalytic profile. Our results extend and further substantiate the earlier proposals of common transition states in the reaction mechanism of group I splicing, by showing that this also applies to intact group I introns, different co-substrates and different catalytic metal ions.
Deoxynucleotides, nucleotides, Nick columns (G-50), RNA Guard and T7 RNA polymerase were purchased from Pharmacia & Upjohn and guanosine, spermidine and diethylpyrocarbonate from Sigma. 2'-Amino-2'-deoxyguanosine was synthesized according to published procedures (17 ,18 ) and purified as described previously (16 ). Labelled [35S]UTP[alpha]S was purchased from Amersham. Restriction endonucleases DraIII and HpaI and DNase I were from Boehringer Mannheim, restriction endonuclease SpeI and T4 DNA ligase from Promega and shrimp alkaline phosphatase from US Biochemicals. All solutions were diethylpyrocarbonate treated.
The concentrations of the nucleoside co-substrates were determined spectrophotometrically at pH 1.0. Extinction coefficients used were 12 200 M-1 cm-1 at 256 nm for guanosine and 12 500 M-1 cm-1 at 255 nm for 2'-amino-2'-deoxyguanosine (17 ).
Plasmid pBS5[Delta]1-650 (Fig. 1 a) is a fusion derivative of plasmids pASS and pBS5 (19 ). Plasmid pASS contains 624 bp of pBS5 fused into the lacZ gene of pTZ18R such that the T7 RNA polymerase transcript contains the first 9 nt of the lacZ transcript followed by 13 intron-proximal nucleotides of exon 1 of the nrdB transcript, the entire 598 nt of the intron of the nrdB transcript, 13 intron-proximal nucleotides of exon 2 of the nrdB transcript and the downstream part of the lacZ transcript. Plasmid pBS5[Delta]1-650 was constructed by ligating a 2900 bp SpeI-DraIII fragment of pASS containing the T7 promoter, the chimeric lacZ::nrdB exon 1 and part of the intron to a 1826 bp SpeI-DraIII fragment of pBS5 containing the rest of the intron and the entire exon 2 of the nrdB gene. The resulting transcript contains the chimeric lacZ::nrdB exon 1, the native nrdB intron and the native nrdB exon 2.
Purified pBS5[Delta]1-650 DNA, linearized with HpaI, was used as template for in vitro transcription. The linearized DNA produces a transcript of 820 nt; 22 nt of chimeric lacZ::nrdB exon 1, 598 nt of nrdB intron and 200 nt of nrdB exon 2 (Fig. 1 b).
Trace amounts of [35S]UTP[alpha]S was used to label the pre-mRNA in in vitro transcription with T7 RNA polymerase as described previously (3 ). Purification of produced pre-mRNA was performed by DNase I treatment and desalting on G-50 columns (3 ).
The folding and splicing conditions of the pBS5[Delta]1-650 pre-mRNA transcript were essentially as previously described (3 ). The pre-mRNA was heat denatured and allowed to fold in the presence of divalent metal ions at 325C, which ensured that >85% of the RNA was highly reactive and homogeneous. The self-splicing reactions contained pre-mRNA in the nanomolar range and were initiated by addition of the co-substrate. The applied concentration ranges were for guanosine 0.85-30 [mu]M below and 0.85-20 [mu]M above pH 6.9 and for 2'-aminoguanosine 3-194 [mu]M below and 6-84 [mu]M above pH 7.5, which means that all reactions were performed with excess co-substrate. Splicing reactions were performed at 325C in the presence of 60 mM KCl, 4 mM MgCl2 and 40 mM buffer: MES-KOH, pH 5.8-6.2; PIPES-KOH, pH 6.2-7.8; Tris-HCl, pH 7.8-9.0. At pH 6.2 and 7.8 reaction rates were compared between two overlapping buffers and no difference due to buffer type was observed. The kobs measurements in divalent metal ion mixtures contained either 0.1 mM zinc and 4 mM magnesium or 0.3 mM manganese and 4 mM magnesium and as co-substrate 2'-aminoguanosine at saturating concentration (168 [mu]M).
Reactions were stopped at specific times by adding an aliquot of the reaction mixture to an equal volume of 10 M urea, 50 mM EDTA, 0.1% bromophenol blue and 0.25% xylene cyanol.
Samples were subjected to electrophoresis on 3.4% polyacrylamide (19:1 acrylamide/bisacrylamide) gels with 8 M urea in TBE buffer (0.089 M Tris base, 0.083 M boric acid, 2 mM EDTA, pH 8.0). Dried gels were exposed to a phosphorimager screen. Radioactivity was directly quantified by area integration on a Molecular Dynamics system. In order to achieve well-separated species of pre-mRNA (820 nt), co-substrate-IVS-exon 2 intermediate (799 nt) and co-substrate-IVS (599 nt), the gels were run such that ligated exons (222 nt) could not be accurately evaluated. Total RNA at each time point was calculated as the sum of remaining pre-mRNA, co-substrate-IVS-exon 2 * (221/212) and co-substrate-intron * (221/156), where the correction factors represent the number of uridine residues in the pre-mRNA divided by the number of uridine residues in co-substrate-IVS-exon 2 and co-substrate-IVS respectively.
Data were analysed using the KaleidaGraph graphing and curve fitting package (Synergy Software). Observed rate constants (kobs) were obtained by fitting linear equations to plots of the natural logarithm of remaining precursor fraction (lnF[pre-mRNA]) versus time for time points up to 80% consumption of pre-mRNA. Non-linear curve fits with equations for consecutive reactions (cf. Fig. 1 c and legend to Fig. 3 ) were used to compare k2 (the forward rate constant for the first transesterification reaction) and k3 (the forward rate constant for the second transesterification). Km and kcat values for single turnover were determined from Hanes-Woolf plots, [s]/kobs versus [s]. The reaction is not truly catalytic but the kinetics obey the mathematical expression v = k1k2 [G][pre-mRNA]/(k-1 + k2). [G] is constant since it is in large excess, as opposed to enzyme catalysed reactions where the enzyme concentration is constant. The expression is identical to the Michaelis-Menten equation where 1/Km equals k1/(k-1 + k2) and k2 equals kcat. k2 is considerably smaller than k-1, which gives Km [approx] Kd (Ks), the dissociation constant. To facilitate comparison with other systems (mainly ribozymes derived from group I introns) we have chosen to use Km and kcat as they describe the same chemical events in truly catalytic systems. Determinations of kobs were typically based on four time points and those of kcat and Km were typically based on seven different co-substrate concentrations, as specified above. The kinetic constants at pH 7.2 were from Sjögren et al. (3 ). These determinations consist of at least seven different co-substrate concentrations and each single time curve was based on more than five time points.
The influence of the starting 5'-triphosphate close to the upstream splice junction of the shortened nrdB pre-mRNA was investigated by treating ~0.5 pmol pre-mRNA transcript with 5 U shrimp alkaline phosphatase at 375C in 20 mM Tris-HCl, pH 7.5, containing 5 mM MgCl2. The observed rate constants of the dephosphorylated pre-mRNA transcript at pH 6.85 (5 [mu]M guanosine), 7.2 (1.3 and 15 [mu]M guanosine), 7.6 (5 [mu]M guanosine) and 9.0 (30 [mu]M guanosine) were not significantly different from those of untreated pre-mRNA transcripts.
The cloned group IA2 intron-containing nrdB gene of bacteriophage T4 was engineered to a shortened variant where the wild-type exon 1 sequence of 664 bp was substituted with 9 bp of the lacZ gene followed by 13 bp of intron-proximal exon 1 nrdB sequence (Fig. 1 a and b). This facilitates detailed evaluation of the reaction kinetics of the resulting shortened pre-mRNA, which self-splices according to the reaction pathway of group I introns (Fig. 1 c). Splicing of group I introns consists of two `chemical' steps (i.e. breaking and making of covalent bonds) with intermediate formation of co-substrate-intron-exon 2 (1 ). Such an intermediate accumulates at the early stages of the self-splicing reaction of our shortened nrdB pre-mRNA and later disappears, with concomitant production of ligated exons and G-IVS (Fig. 2 ). Control experiments showed that no liberated exon 2 was formed during self-splicing and that no IVS-exon 2 intermediate or other splicing products were formed after extended incubation of the pre- mRNA in the absence of co-substrate (data not shown). Thus, the co-substrate-IVS-exon 2 intermediate is specific for co-substrate catalysed self-splicing. Both magnesium and manganese divalent ions can promote folding of the shortened pre-mRNA into a catalytically active structure (3 ).
The rate of guanosine-dependent splicing of the shortened T4 nrdB pre-mRNA increases linearly with increasing pH in the range 5.8-7.0. Above pH 7.0 the total rate of reaction becomes independent of changes in pH. The apparent Km for guanosine is constant over the entire pH range tested, which means that kcat/Km reflects the behaviour of kcat. The maximal kcat/Km is 1.3 * 105 M-1 min-1. The lower half of the pH profile of log(kcat/Km) has a slope of 0.94 (Fig. 4 ), implying that the splicing reaction is dependent upon deprotonation. As suggested earlier for engineered Tetrahymena and Anabaena ribozymes (9 ,13 ), this kind of pH behaviour can mean either titration of a catalytically important group with a pKa of ~7 or a switch from a pH-dependent catalytic step to another pH-independent rate limiting step at a kcat of 1.3 min-1 (kcat/Km [approx] 105 M-1 min-1).
For further characterization of the rate limiting step of the self- splicing reaction we have utilized the fact that 2'-aminoguanosine is a co-substrate for self-splicing of shortened T4 nrdB pre-mRNA (3 ,16 ). As the kcat value with 2'-aminoguanosine as co-substrate is ~30-fold lower than that with guanosine, it is possible to differentiate between the two explanations for the pH dependence described above. In the case of a titratable group with a pKa of ~7 the 2'-aminoguanosine-dependent reaction would also level off just above pH 7, whereas in the case of interference from another rate limiting step the 2'-aminoguanosine-dependent reaction would be expected to level off at a similar kcat (or kcat/Km) value as the guanosine-dependent reaction. If a conformational step becomes rate limiting it seems most appropriate to compare first order rate constants (i.e. kobs at saturating concentration of co-substrates or kcat) since this kind of transition is unimolecular.
The pH dependence of 2'-aminoguanosine catalysed splicing was studied in the pH range 6.7-9.0. With this co-substrate the rate of splicing was found to increase over the entire pH range studied (cf. Fig. 5 ). The Km of 2'-aminoguanosine is constant up to pH 7.8 but then increases 3- to 4-fold at higher pH. This suggests that a titrating group may influence the binding of this co-substrate. It is unlikely, however, that protonation/deprotonation of the 2'-amino group of the co-substrate is causing this effect, since the reported pKa of 2'-amino-2'-deoxyuridine is 6.2 (20 ). For the 2'-aminoguanosine catalysed reaction the plot of log(kcat/Km) versus pH is linear with a slope of 0.94 up to pH 8.0, after which it gradually levels off and reaches a kcat/Km of 0.3 * 105 M-1 min-1 at pH 8.8 (Fig. 4 ). The pH rate profile obtained with 2'-aminoguanosine thus suggests a change from a pH-dependent to a pH-independent step above pH 8.5, which is at least one pH unit higher than the corresponding reaction with the guanosine co-substrate. This strongly suggests that self-splicing of the shortened T4 nrdB intron studied here switches from the first transesterification step to another rate limiting step at a rate constant of ~1 min-1 (cf. Fig. 5 ).
The pH dependence of the self-splicing reaction of the shortened T4 nrdB pre-mRNA was also investigated in the presence of divalent metal ion mixtures. It was earlier observed that addition of low concentrations of manganese or zinc ions in the magnesium- containing reaction mixtures at pH 7.2 increased the observed rate constants ~6-fold and 35-fold respectively with 2'-aminoguanosine, whereas those with guanosine as co-substrate were unaffected (3 ). Figure 5 , in which kobs for reactions in metal ion mixtures at saturating co-substrate concentration is plotted together with kcat for the magnesium-dependent reactions, shows that the general pH-dependent behaviour of the observed reaction rate is true also for splicing in metal ion mixtures. In manganese/magnesium mixtures the 2'-aminoguanosine-dependent reaction increases with increasing pH over almost the entire pH range studied, with a slope of 1.3 between pH 5.8 and 7.4. In zinc/magnesium mixtures the 2'-aminoguanosine-dependent rate is similar to the rate of guanosine catalysed splicing in magnesium buffer and a levelling off of the pH-dependent increase seems to coincide with that of the guanosine reaction. These results further substantiate the hypothesis that a deprotonation precedes or is part of the chemical step that is rate limiting up to a maximal rate of ~1 min-1, after which a common pH-independent step becomes rate limiting.
We have analysed the pH dependence of the self-splicing reaction of an intact group IA2 intron. Our system consists of a shortened pre-mRNA of the nrdB gene from bacteriophage T4 in which most of the native exon 1 sequences have been deleted and the chimeric 22 nt exon 1 contains only 13 nrdB-derived intron-proximal nt. The P1 stem of the nrdB group IA2 intron is proposed to consist of only 5 bp, which is short in comparison with most other group I introns (21 ). In addition, the shortened nrdB pre-mRNA probably lacks additional structural elements 5' of the P1 stem due to the short exon 1 sequence. Yet, under physiological in vitro conditions (4 mM Mg2+, pH 7.2, 32oC) the shortened pre-mRNA self-splices with a rate (kcat/Km ~105 M-1 min-1) comparable with other group I introns.
The self-splicing reaction of the shortened pre-mRNA containing the nrdB intron accumulates a co-substrate-IVS-exon 2 intermediate. In contrast, a splicing intermediate could not be distinguished in the self-splicing reaction of wild-type nrdB pre-mRNA (16 ), in which exon 1 is considerably longer. It is interesting to note that a co-substrate-IVS-exon 2 intermediate has also been observed in other self-splicing systems, e.g. for the group I intron in pre-tRNALeu of Anabaena, the group I intron in pre-tRNAIle of Azoarcus and a shortened pre-mRNA of the mitochondrial cytochrome b group I intron of Saccharomyces cerevisiae (13 ,22 -24 ). These group I introns share the feature of having short exon 1 sequences, as is also the case in our engineered T4 nrdB system.
This study was supported by grants to B.-M.S. and R.S. from the Swedish Natural Science Research Council.
*To whom correspondence should be addressed. Tel: +46 8 164150; Fax: +46 8 152350; Email: bitte@molbio.su.se
REFERENCES
+Present address: Pharmacia & Upjohn, Lund, Sweden