Flanking sequences with an essential role in hydrolysis of a self-cleaving group I-like ribozyme

Abstract

DiGIR1 is a group I-like ribozyme derived from the mobile twin ribozyme group I intron DiSSU1 in the nuclear ribosomal DNA of the myxomycete Didymium iridis. This ribozyme is responsible for intron RNA processing in vitro and in vivo at two internal sites close to the 5′-end of the intron endonuclease open reading frame and is a unique example of a group I ribozyme with an evolved biological function. DiGIR1 is the smallest functional group I ribozyme known from nature and has an unusual core organization including the 6 bp P15 pseudoknot. Here we report results of functional and structural analyses that identify RNA elements critical for hydrolysis outside the DiGIR1 ribozyme core moiety. Results from deletion analysis, disruption/compensation mutagenesis and RNA structure probing analysis all support the existence of two new segments, named P2 and P2.1, involved in the hydrolysis of DiGIR1. Significant decreases in the hydrolysis rate, k_obs, were observed in disruption mutants involving both segments. These effects were restored by compensatory base pairing mutants. The possible role of P2 is to tether the ribozyme core, whereas P2.1 appears to be more directly involved in catalysis.

INTRODUCTION

The biological role of group I ribozymes is to catalyze self-splicing through a series of transesterification reactions which result in the perfect ligation of flanking exon sequences (1). Most group I introns catalyze the splicing reaction as naked RNA in vitro, but are dependent on associated protein factors for efficient splicing in vivo. However, these factors may not be essential or nesessarily host specific, since the best studied of all group I introns, TtLSU1 from the ciliate Tetrahymena, works remarkably well in cognate positions in ribosomal DNA (rDNA) of distantly related organisms such as Escherichia coli and the yeast Saccharomyces cerevisiae (2–4) as well as some non-cognate positions in mammalian cells (5,6). Other group I introns require essential cellular protein factors for splicing and well-known examples are the Cbp2 protein and tyrosyl-tRNA synthetase, which are important or essential for some yeast and Neurospora mitochondrial group I introns, respectively (7–9).

The single example of a group I ribozyme with a biological function other than splicing is GIR1 from the nuclear twin ribozyme introns (10). Here, functional GIR1 ribozymes are found in the group I introns DiSSU1 of the myxomycete Didymium iridis (11) and NaSSU1 from different species of the amoeboeflagellate Naegleria (12), where they are positioned as insertions into the splicing ribozymes (GIR2) of the introns. GIR1 catalyzes a sequential hydrolytic cleavage at two internal processing sites (IPS1 and IPS2) (13), close to the 5′-end of the open reading frame (ORF) encoding the intron homing endonuclease (see Fig. 1A). Recently we showed that the Didymium GIR1 (DiGIR1) cleavage site IPS2 corresponds exactly to the 5′-end of the translated endonuclease mRNA in vivo (14). A similar result was obtained for NaSSU1 in Naegleria amoebae and further analysis in transfected yeast cells showed that the Naegleria GIR1 (NaGIR1) cleavage was essential for in vivo expression of the endonuclease (15). Comparisons to known introns suggest that both DiGIR1 and NaGIR1 have evolved from pre-existing, but distinct, eubacterial-like group I splicing ribozymes (13). Here, the transesterification reactions responsible for splicing have apparently been lost and replaced by hydrolytic cleavage at the IPS.

Figure 1.

Schematic presentation of the DiSSU1 intron and RNA transcripts. (A) Schematic organization of the intron and flanking rDNA sequences. DiSSU1 consists of two distinct group I ribozymes (DiGIR1 and DiGIR2) and an ORF encoding the I-DirI homing endonuclease. 5′ SS and 3′ SS, intron splice sites; IPS1/2, internal processing sites 1 and 2. (B) The paired regions P2 and P2.1 are indicated within DiGIR1. (C) Different RNA transcripts analyzed in the 3′ deletion study (I), P2 segment mutagenesis (II) and P2.1 segment mutagenesis (III). Crossed boxes indicate mutated sequences.

The GIR1s are less than 200 nt in size and represent the smallest known naturally occurring group I ribozymes. The evolved non-splicing function has resulted in an unusual structural organization including the discovery of an essential 6 bp pseudoknot (P15) within the catalytic core, as supported by comparative sequence analysis (10,12), site-directed mutagenesis (13), RNA structure probing (13,16) and in vitro selection (17). Another unusual structural feature of the GIR1 ribozymes compared to all other group I ribozymes is the lack of the P1 segment (10,16). In this report we present evidence for new structural elements in the flanking sequences that are essential for activity of the DiGIR1 ribozyme.

MATERIALS AND METHODS

Generation of templates for in vitro RNA transcription

The different DiGIR1 RNA constructs are named G1 (GIR1) followed by two numbers indicating the number of positions included upstream and downstream of the IPS1, respectively. Here, G1-162.22 contains 162 nt upstream and 22 nt downstream of IPS1. Constructs were made by standard PCR amplifications using Pfu DNA polymerase (Stratagene) on pDi162G1 as the template (13) and various primer combinations. The following 3′ deletion constructs were all made from OP39 (11) as the upstream primer in combination with a specific downstream primer (in parentheses): G1-162.5 (OP619, 5′-GGA TGC TTC CTT TCG GAA-3′), G1-162.8 (OP620, 5′-ACC GGA TGC TTC CTT TCG-3′), G1-162.11 (OP78, 5′-GAT ACC GGA TGC TTC CTT-3′), G1-162.16 (OP315, 5′-CTT GGG ATA CCG GAT GCT TCC TTT-3′), G1-162.22 (OP233, 5′-GAT TGT CTT GGG ATA CCG-3′), G1-162.30 (OP353, 5′-TTA GAT TTG ATT GTC TTG-3′), G1-162.37 (OP314, 5′-GGT ATC CTT AGA TTT GAT TGT CTT-3′), G1-162.46 (OP235, 5′-GCA CAG ATT GGT ATC CTT-3′) and G1-162.65 (OP12, 5′-TCA CCA TGG TTG TTG AAG TGC ACA GAT TG-3′). The following P2 and P2.1 mutants were also constructed (primer combinations in parentheses): G1-mutP2-5′ (OP388, 5′-AAT TTA ATA CGA CTC ACT ATA GGG AAC CCT AGT ATC ATG GCT AAT CA-3′, and OP233), G1-mutP2-3′ (OP39 and OP389, 5′-GAT TGT CAA CCC TTA CCG GAT GCT TCC TTT C-3′), G1-mutP2 (OP388 and OP389), G1-mutP2.1-5′ (OP312, 5′-AAT TTA ATA CGA CTC ACT ATA GGG TTG GGA AGT TAG TAC CCT AAT CAC CAT GAT GCA ATC-3′, and OP233), G1-mutP2.1-3′ (OP453, 5′-AAT TTA ATA CGA CTC ACT ATA GGG TTG GGA AGT ATC ATG GCT AAT CAG GTA CTA GCA A-3′, and OP233) and G1-mutP2.1 (OP311, 5′-AAT TTA ATA CGA CTC ACT ATA GGG TTG GGA AGT TAG TAC CCT AAT CAG GTA CTA GCA ATC GGG TTG AAC ACT TAA-3′, and OP233). PCR products were agarose gel purified, phenol/chloroform (1:1) extracted, ethanol precipitated and dissolved in DEPC-treated H₂O. All the upstream primers contain the T7 polymerase promoter sequence (underlined).

In vitro transcription, ribozyme self-cleavage and RNA structure probing

All precursor RNAs were in vitro transcribed using a T7 RNA polymerase kit (Stratagene). The transcription reactions were performed according to the procedure recommended by the manufacturer. RNAs were uniformly labeled by the incorporation of 0.5 µl [³⁵S]CTP (10 mCi/ml) (Amersham) in a 50 µl reaction. The RNA was then phenol/chloroform extracted, ethanol precipitated and dissolved in an appropriate volume of DEPC-treated H₂O. Prior to the self-cleavage reaction, RNA was preincubated for 5 min at 45°C in acetate buffer, pH 5.5, containing 25 mM MgCl₂ and 1 M KCl to allow correct folding of the RNAs (see 16). Time course experiments were initiated by raising the pH to 7.5 by adding 4 vol of a start buffer containing 47.5 mM HEPES–KOH pH 7.5, 25 mM MgCl₂ and 1 M KCl. Reactions were terminated at selected time points by the removal of aliquots and the addition of 2 vol of stop solution (95% formamide, 50 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol FF). RNAs were stored at –70°C. Reacted RNAs were resolved on 5% polyacrylamide–7 M urea gels and visualized by autoradiography or exposed to a PhosphorImager screen (Molecular Dynamics). The transcripts used for probing experiments were made by in vitro transcription of a copy of a PCR product corresponding to G1-162.65 cloned into a plasmid vector and subsequently linearized at the NcoI site introduced in OP12 (18). These transcripts include 64 nt downstream of IPS1. Chemical and enzymatic structure probings were performed as described previously (13).

Data analysis

RNA bands were quantitated by phosphoimager exposure with ImageQuant v.3.3 software. All kinetic data were analyzed as described in Jabri et al. (16), except that we include a theoretical value for the 3′ hydrolysis product due to the small and variable size and co-migration with salt in the gel. In short, the data were fitted using Enzfitter v.1.05 (Elsevier-Biosoft) to the following non-linear first order decay equation with an end point correction: (fraction precursor)_t = (fraction precursor)_{t = ∞} + (fraction precursor)_{t = 0} × exp(–k_obs × t) to obtain values for k_obs and end points of reactions. All graphs were drawn using Matlab 4 (Mathworks Inc.). k_obs describes the slope of these plots and is a measure of the reaction rate for the hydrolysis reaction (16). All transcripts were analyzed two to eight times in parallel and the results were highly reproducible (0–3% cleavage variation). Thus, error bars in the presentations are omitted for clarity of the figures.

RESULTS

3′ deletion analysis of DiGIR1

We have previously reported a 5′ deletion analysis of DiGIR1, concluding that 162 nt positions upstream of the IPS1 cleavage site (construct G1-162.57) are required in vitro for optimal ribozyme activity (11). To evaluate the functional importance of downstream flanking sequences, a similar 3′ deletion analysis was performed. Nine different constructs were produced by PCR amplification of pDi162G1 (13) using the upstream primer OP39 containing the T7 promoter and various downstream primers (Fig. 1C). The PCR products were in vitro transcribed to generate DiGIR1 RNA with 5, 8, 11, 16, 22, 30, 37, 46 or 65 nt downstream of IPS1. The pre-RNAs, all containing 162 nt upstream of IPS1, were preincubated in an acetate buffer (pH 5.5) to allow correct folding of the ribozyme. The hydrolysis reaction was initiated by raising the pH to 7.5. Each RNA construct was incubated and then terminated at seven different time points (0–240 min) and the cleavage products separated on a denaturing polyacrylamide gel and exposed to a phosphoimager screen to be quantitated.

A plot of the fraction of uncleaved precursors versus time is presented in Figure 2. The shortest DiGIR1 RNA (G1-162.5) with only 5 nt downstream of the IPS was found to be competely inactive (Table 1) and is not included in the plot. At least 22 nt are required to cleave 70–80% of the precursor, which appears to be the end point of the reaction (13). The rate constant (hydrolysis rate, k_obs) and end point were obtained from a best fit plot of each reaction (see Materials and Methods) and values are presented in Table 1. The hydrolysis rate of 0.018 min^–1 for G1-162.22 is similar to that previously observed for NaGIR1 (0.027 min^–1) containing 19 nt downstream of the IPS (13). Interestingly, a significant increase in k_obs was observed when the 3′-region was increased from 22 to 65 nt (0.018 to 0.072 min^–1) without an improvement in end point of the reaction (Fig. 2, lower panel, and Table 1).

Figure 2.

DiGIR1 3′ deletion analysis presented as a plot of fraction uncleaved precursor versus time. All graphs are best fit curves to the non-linear first order decay function: F_t = F_∞ + F₀ × e^{–kobs × t}. ³⁵S-labeled DiGIR1 RNAs were subjected to hydrolysis conditions for different lengths of time (0, 5, 15, 30, 60, 90 and 240 min), separated on gels and quantitated after exposure to a phosphoimager screen. Result from G1-162.5 RNA is not included in the plot because no cleavage activity could be detected. The graphs are split into an upper and a lower panel for clarity of the figure.

Table 1. Hydrolysis rate (k_obs) and reaction end points.

RNA	kobs	End point
G1-162.65	0.072	0.25
G1-162.46	0.056	0.38
G1-162.37	0.043	0.31
G1-162.30	0.02	0.19
G1-162.22	0.018	0.2
G1-162.16	0.025	0.54
G1-162.11	0.023	0.56
G1-162.8	0.012	0.81
G1-162.5	~0.000	~1.00
G1-mutP2-5′	0.011	0.32
G1-mutP2-3′	0.011	0.6
G1-mutP2	0.037	0.23
G1-mutP2.1–5′	0.003	0.76
G1-mutP2.1–3′	0.002	0.57
G1-mutP2.1	0.016	0.18

The 6 bp terminal P2 segment tethers the DiGIR1 ribozyme core

Naegleria GIR1 ribozymes appear to contain a P2 segment consisting of 4–7 bp, connected to P10 via an internal loop (10,16). Chemical modification analysis of one of the ribozymes, NanGIR1, supports the existence of P2 (16). An inspection of the corresponding flanking sequences in DiGIR1 suggested a 6 bp P2 segment (Fig. 3), consistent with the requirement 22 nt downstream sequence to cleave 80% of the precursor (G1-162.22 RNA; Table 1). In order to evaluate the significance of the DiGIR1 P2 base pairing in catalysis, mutations were made to disrupt base pairing in P2 (Fig 3, left insert). The resulting mutant RNAs, G1-mutP2-5′ and G1-mutP2-3′, were incubated and assayed in time course experiments under hydrolysis conditions as described for the 3′ deletion RNAs. Both the disruption base pair mutants showed a decrease in the reaction end point and k_obs (Fig. 4 and Table 1). The compensatory mutant, G1-mutP2 (Fig. 3), which restored base pairing of the P2 stem, restored hydrolysis activity above the level of G1-162.22 RNA (Fig. 4). Here, the k_obs increased from 0.018 to 0.037 min^–1 (Table 1). These results support the existence of P2 and imply an important, but not essential, role in hydrolysis of DiGIR1.

Figure 3.

Secondary structure presentation of DiGIR1 including the flanking P2 and P2.1 segments. The mutation strategy, based on base pair disruption and compensation of P2 and P2.1, are boxed at the top.

Figure 4.

Site-specific mutagenesis of P2. (A). Time course experiment of DiGIR1 with unchanged P2 (G1-162.22), P2 5′ disrupted (G1-mutP2-5′), P2 3′ disrupted (G1-mut P2-3′), and P2 compensated (G1-mutP2) RNAs. ³⁵S-labeled RNAs were submitted to hydrolysis, analyzed at different time points (0, 5, 15, 30, 60, 90 and 240 min) and separated on a 5% denaturing polyacrylamide gel. Only the precursor RNA (pre-RNA) and the 5′ ribozyme core product (5′ product) are shown. (B). Plot of fraction uncleaved precursor versus time, based on the experiment presented above.

The P2.1 segment in the 5′-flanking region is essential for DiGIR1 catalytic activity

In a previous report we noted a stable hairpin structure (renamed P2.1) located 5′ of the DiGIR1 ribozyme core (11). An RNA lacking the putative structure (G1-127.57 RNA) was found to be inactive when incubated under hydrolysis conditions (11), suggesting an important function. To test this assumption, a mutation strategy similar to that of P2 presented above was performed. Two mutant constructs that disrupted the base pairing in P2.1 were made (Fig. 3, right insert), in vitro transcribed and incubated and assayed under hydrolysis conditions. Both the disruption base pair mutants, G1-mutP2.1-5′ and G1-mutP2.1-3′, showed a pronounced decrease in the reaction end point and k_obs (Fig. 5 and Table 1). At 240 min incubation, cleavage was barely detectable in any of the mutants (Fig. 5A). The compensatory mutant RNA, G1-mutP2.1 (Fig. 3), which restored base pairing of the P2.1 stem, restored hydrolysis activity to the level of G1-162.22 RNA (Fig. 5) with similar k_obs values (0.018 and 0.016 min^–1) and end point values (0.20 and 0.18). We also tested if the individual bases in the P2.1 loop (L2.1) were responsible for this effect by generating a mutant version of the G1-162.22 RNA where the L2.1 CUAAUC was changed to GGUUGG. No decrease in activity could be observed when incubated under hydrolysis conditions (data not shown). These results support the P2.1 hairpin structure, but not its individual bases, to be essential for hydrolysis of DiGIR1.

Figure 5.

Site-specific mutagenesis of P2.1. (A). Time course experiment of DiGIR1 with unchanged P2.1 (G1-162.22), P2.1 5′ disrupted (G1-mutP2.1–5′), P2.1 3′ disrupted (G1-mut P2.1–3′) and P2.1 compensated (G1-mut P2) RNAs. ³⁵S-labeled RNAs were analyzed as described in the legend to Figure 4. (B). Plot of fraction uncleaved precursor versus time, based on the above experiment.

RNA structure probing analysis supports the existence of P2 and P2.1 in DiGIR1

In order to obtain independent support for the existence of P2 and P2.1, both enzymatic (RNases V1, T1 and T2) and chemical (DMS and DEPC) probing experiments were performed on DiGIR1 RNA using the primers C80 and C81 as previously described in Einvik et al. (13). The results from the structure probing experiments are summarized in Figure 6A, superimposed on the secondary structure model of flanking DiGIR1 sequences. The probing pattern is in general agreement with the secondary structure model supported by mutational analysis. However, some inconsistencies are observed, in particular in the chemical probing. This is probably due to the existence of alternative conformations in the population of molecules, as also indicated by the presence of a significant fraction of uncleavable RNA. Interestingly, whereas the 3′ branch of the P2.1 stem shows several strong RNase V1 signals, consistent with an exposed double-stranded RNA configuration, no signals were seen at the 5′ branch of the same segment (Fig. 6B). A possible explanation for this observation is that the 5′ branch of P2.1 is protected from the large RNase probe by other parts of the GIR1 structure. Several DMS and RNase T2 signals are seen in L2.1, indicating a single-stranded loop region not involved in secondary or tertiary interactions in the synthetic RNA in this assay.

Figure 6.

Structure interpretations of flanking sequence regions. (A) Summary of the structural probing data, based on primer extension as well as 5′- and 3′-end-labeled RNAs, are superimposed on the DIGIR1 sequence. Chemical probes: dimethyl sulfate (DMS) and diethylpyrocarbonate (DEPC). Enzymatic probes: RNase T1 (T1), RNase T2 (T2) and RNase V1 (CV). Large and small symbols in the figure represent strong and weak signals, respectively. A comparison of the secondary structures of flanking regions in GIR1 from Didymium (DiGIR1) and Naegleria (NaGIR1). The NaGIR1 sequence is a consensus based on five distinct Naegleria introns (10). R = G or A; K = G or U; Y = U or C; W = A or U. Nucleotide positions in parentheses are not present in some of the NaGIR1 variants. (B) Primer extension analysis of DiGIR1 RNA probed with RNase T2 and RNase V1 as well as an untreated control (No). The primer extension reactions were fractionated on an 8% denaturing acrylamide gel in parallel with a set of sequencing reactions made by dideoxy sequencing of DiGIR1 using AMV reverse transcriptase (G, A, T and C refer to the sequence of the strand complementary to DiGIR1). A cluster of three consectutive cleavage sites corresponding to G98–G101 on the 3′-side of the P2.1 stem is seen in the RNase V1-treated RNA. Also note the lack of RNase V1 cleavages in the 5′-side of the stem. (C) Proposed secondary structure of the 5′-UTR hairpin (from 14). Sequences involved in the P10 and P2 pairings are indicated.

DISCUSSION

We have presented evidence that the self-cleaving group I-like ribozyme DiGIR1 contains structural elements critical for hydrolysis outside the ribozyme core moiety. One of these structures, the 9 bp P2.1 hairpin, was found to be essential for catalytic activity and is unique among the group I ribozymes. Structure probing data, summarized in Figure 6, indicate that the 5′ branch of the stem is protected from RNase V1 cleavage, suggesting a closer association of P2.1 with other parts of the ribozyme structure. RNA structure probing using an azido-PNA photoaffinity probe complementary to the 5′ branch of DiGIR1 P10 imply a P2.1 location proximal to the catalytic core region (manuscript in preparation). What might be the functional role of P2.1 in hydrolysis? One possibility is that P2.1 is directly involved in tertiary interactions with other parts of the molecule in order to stabilize the ribozyme core. However, several observations do not support such RNA–RNA interactions as the critical factor for hydrolysis. These include the lack of convincing base pairing to the L2.1 nucleotides and the lack of a recognizable tetraloop receptor motif in the P2.1 stem which could potentially interact with the GAAA loop in P9 (see Fig. 3). Furthermore, the L2.1 mutant and the compensatory restored mutant G1-mutP2.1 did not affect the catalytic activity, showing that the hairpin structure rather than its individual bases is important for hydrolysis. An alternative explanation is that P2.1 may serve a more direct role in catalysis. DiGIR1, similar to all other ribozymes known, is dependent on magnesium (or other metal ions) for catalytic activity (13). Whereas three catalytic metal ions have been identified within the active site of the Tetrahymena group I ribozyme (19), none has to our knowledge been specifically assigned to 3′ splice site hydrolysis. We therefore speculate that P2.1 may present a catalytic metal ion to the active site by folding back onto the core facilitating hydrolytic cleavage of IPS1, a cleavage site resembling the 3′ splice site in the Tetrahymena ribozyme (13).

The flanking P2/P10 motif is conserved among the Didymium and Naegleria GIR1 structures (Fig. 6A). The 3′ deletion analysis (Fig. 2) and the disruption compensation mutations (Fig. 3) support an important role in DiGIR1 hydrolysis for the P10 and P2 segments, respectively. A similar result was obtained for NaGIR1 (16). The individual base pairs in P10 are highly conserved in DiGIR1 and NaGIR1 and probably serve in organizing IPS2. The P2 segment is more variable at the sequence level, but still important for ribozyme activity. We suggest that P2 stabilizes the GIR1 structure by tethering the 5′ and 3′ flanking sequences together. Flanking exon sequences have been reported to both inhibit and enhance splicing of the Tetrahymena and Physarum group I introns (5,6,20–23). Interestingly, the exons in the anticodon stem from the Anabaena pre-tRNA group I intron were found to be essential for splicing (24), a result similar to that of the flanking sequences in DiGIR1. The observed structural and functional resemblance between GIR1 (Fig. 6A) and eubacterial pre-tRNA group I ribozymes is intriguing in the light of a possible common evolutionary origin (13).

The 5′-end of the polysome-associated I-DirI mRNA corresponds exactly to IPS2, indicating DiGIR1 to be involved in intron endonuclease expression in vivo (14). The intron endonuclease, I-DirI, initiates intron homing in vivo when haploid intron-containing and intron-less Didymium amoebae are mated (25). I-DirI appears to be a rare example of a protein expressed from an RNA polymerase I transcript (14). Here, a hairpin structure was proposed in the 5′-untranslated region (UTR) in order to stabilize the uncapped mRNA. Interestingly, the 3′ branches of both P2 and P10 overlap with the proposed 5′-UTR hairpin (Fig. 6C). We have recently reported that DiGIR1 performs sequential hydrolytic cleavages at IPS1 and IPS2 and results from molecular modeling imply that IPS1 cleavage induces a conformational change in the ribozyme in order to cleave at IPS2 (13). We propose that the flanking RNA sequences of DiGIR1 play a dual role in I-DirI mRNA processing. First, it folds into the cleavage-competent P2–P2.1–P10 configuration. After IPS cleavage, P2 and P10 dissociate and, consequently, the nucleotides involved refold into the 5′-UTR hairpin of the I-DirI mRNA. This structure may target trans-acting factors either involved in nuclear mRNA stabilization, nuclear–cytoplasmic translocation or even translation. These possibilities are currently under investigation.