ABSTRACT
The Cbp2 protein facilitates the folding of a group I intron in the
COB
pre-mRNA of yeast mitochondria. Based on its ability to suppress mutations
affecting the autocatalytic reaction, the protein appears to play a role in the
selection of splice sites. Adding Cbp2 did not overcome the effects of
mutations in P1 whose primary effect was on the first step of splicing. In
contrast, most mutations affecting the ligation of exons were suppressed
in vitro
by Cbp2. These included mutations in P1, P9.0 and P10. In fact, a mutant
transcript lacking both P9.0 and P10 ligated efficiently in the presence of
Cbp2. P9.0 and P10 mutations also reduced the rate of cleavage at the 5'
splice junction, and this effect was only partially mitigated by adding Cbp2. A
competitive secondary structure near the 3'
splice junction blocked Cbp2-stimulated splicing, but this mutation could be suppressed by co-transcriptional splicing in the presence of Cbp2. Our data
underscore the importance of the interaction between the 5' and 3'
splice junctions in group I introns and suggest that nucleotide-nucleotide interactions that stabilize the structure of group I introns
can be superceded by protein-RNA interactions.
In group I introns, base pairing between the 5' splice junction and an internal guide sequence (IGS) selects the 5' splice site, which follows a conserved U[middot]G pair in this stem, called P1 (
1
-
5
). P1 is associated with the catalytic core of the intron through tertiary
interactions including that with the bulge J4/5 (
6
). Such tertiary contacts are reflected by the anti-cooperative binding of the guanosine substrate and oligonucleotide analogs
of the 5' splice junction (
7
,
8
) and are described as `docking' of P1 with the catalytic core (
9
,
10
). The spacing of P1 and P2 appears important in positioning the 5' splice site with respect to the catalytic core (
11
,
12
).
Several determinants of the 3' splice site have also been identified (Fig.
1
). The first is a pairing (P10) between the 3' exon and the internal guide sequence. That aligns the 3' and the 5' exons for the second step of splicing. The second,
designated P9.0, associates the 3'-end of the intron with its catalytic core (
13
-
15
). P9.0 was originally described as a 2 base pair (bp) helix between the
nucleotides immediately following P7 and the penultimate two nucleotides of the
intron. More recently, Jaeger
et al.
(
16
) showed that P9.0 is a composite between P9.0a, which forms between a conserved
guanosine that follows P7 and a cytosine near the 3' splice site, and P9.0b, which forms between the second and third
nucleotides downstream of P7 and the penultimate nucleotides of the intron.
While P9.0 is observed in all group I introns, the P10 pairing is not detected
in some introns (
17
). A G[middot]C base pair in P7, which binds the guanosine nucleophile for the first
step of splicing, has also been shown to bind the 3' terminal guanosine for the second step (
13
,
18
,
19
).
T7 RNA polymerase was isolated as described by Grodberg and Dunn (
41
). The purification of Cbp2 is described elsewhere (
39
). RNA ligase and restriction enzymes were purchased from Gibco-BRL or New England Biolabs. RNase A was obtained from Boehringer Mannheim.
Reagents for electrophoresis and radiolabeled nucleotides were obtained from
ICN. Placental RNase inhibitor was purchased from Promega. Activated nylon
membranes (Hybond-N Plus) were purchased from Amersham and nitrocellulose membranes (BA85)
were purchased from Schleicher and Schull. Other reagents were obtained either
from Sigma Chemical Company or from Fisher Scientific.
Several of the mutant transcripts used in this study have been described in
earlier publications (
42
,
43
). Those affecting P9.0 (e.g. A677U/A678U, U736A/U737A), the extension of P1
(A3U/U4A/A5C), the 3' exon (A+13G) and the P9.0/P10 double mutations are new and were prepared
by site directed mutagenesis using the double primer method of Zoller and Smith
(
44
). The template was a clone of
b
I5 and flanking exons from the
Hin
dIII site to the
Bgl
II site of
COB
inserted in the vector M13 mp19. All inserts were sequenced to verify that only
the desired alteration had been made. Several clones of each mutation were
tested to be sure that the behavior of the transcripts was consistent.
Transcripts were generated from wild-type and mutant templates cloned in pT7T3-18 as described by Partono and Lewin (
33
). Transcripts were linearized with
Eco
RI or
Sma
I prior to transcription. The reaction conditions were those of Grodberg and
Dunn (
41
) and contained 20 mM sodium phosphate buffer, pH 7.7, 8 mM MgCl
2
, 4 mM spermidine, 4 mM dithiothreitol, 0.8 mM ATP, CTP and GTP, 0.2 mM
unlabeled UTP and 10 [mu]Ci [[alpha]-
32
P]UTP. Transcription reactions were conducted at 37oC for 2 h. Following transcription reactions, unincorporated nucleotides
were removed by chromatography on Sephadex G-50 (Pharmacia). For co-transcriptional splicing, unlabeled nucleotides were added at 0.1 mM
and reactions lasted 60 min. Polymerized RNA was recovered by precipitation
with ethanol and 0.5 M NH
4
CH
3
COO
-
.
For Cbp2-assisted splicing, reactions were conducted in low salt buffer (50 mM Tris-HCl, pH 7.5, 50 mM NH
4
Cl, 5 mM MgCl
2
). For autocatalytic reactions, a high salt buffer was used (50 mM Tris-HCl, pH 7.5, 1 M KCl, 50 mM MgCl
2
). GTP was present at 0.2 mM in all post transcriptional splicing reactions.
Cbp2-assisted reactions also contained 10 mM dithiothreitol. Reactions were
usually for 1 h at 37oC and were terminated by addition of EDTA to 50 mM and precipitation with
ethanol. Samples were dissolved in 90% formamide, 25 mM EDTA containing
bromophenol blue and xylene cyanol and resolved on gels made of 4 or 5%
polyacrylamide, 8 M urea and TBE buffer (89 mM Tris borate, pH 8.3, 2.5 mM
EDTA). Splicing products were visualized by autoradiography.
Rates of splicing were measured in duplicate time courses at 37oC in low salt buffer containing Cbp2 and various transcripts. Samples were
removed after 0, 1, 2, 4, 8 and 16 min and placed in tubes containing the formamide/EDTA loading buffer described above. Samples were then separated by electrophoresis and analyzed on a
PhosphorImager (Molecular Dynamics). Rates of 5' cleavage were estimated by the appearance of intron products (intron and
intron-3'-exon) expressed as a fraction of the intron present in the
precursor. The intron-associated precursor was calculated by the ratio of uridines in the intron
to the total uridines present in the precursor. Rates were measured when <20% of the precursor had reacted.
A double filter method (
45
) was used to measure affinity of Cbp2 and intron variants. This procedure was
modified by replacing the DEAE membranes with activated nylon membranes (Hybond-N Plus from Amersham). Each data point is the average of triplicate
determinations from the same dot blot. Binding conditions employed low salt
splicing buffer at 37oC for 15 min in the absence of GTP.
Detailed methods for modification of
b
I5 with hydroxyl radicals and 1-cyclohexyl-(2-morpholinoethyl)carbodiimide metho-
p
-toluenesulfonate (CMCT) and for analysis of modification sites based on
reverse transcription are described in an earlier paper (
39
).
When incubated in low salt conditions, a
32
P-labeled transcript containing
b
I5 (738 nt) and flanking exons was catalytically inert (Fig.
2
, lane 1). There was no splicing during the transcription reaction or the 60 min
incubation in low salt buffer. If Cbp2 was added to the transcription reaction
(lane 2), co-transcriptional splicing occurred. The ligated exons were detected as a
band of 454 nucleotides (nt), and intermediates included the intron-3'-exon and the free 5' exon. Similarly, if Cbp2 was incubated with a
completed transcript in low salt buffer, the same intermediates and products
were formed (lane 3). These products were identical to those of the
autocatalytic reaction (lane 4), with the exception that cyclization at a site
237 nt into the intron and guanosine addition within the 3' exon were diminished. Several bi-products evident in the autocatalytic reaction were reduced, these
include linear fragments of 237 and 501 nt resulting from internal cyclization
and hydrolysis (
33
).
Table 1
To determine whether Cbp2 mediated splicing requires P1 and P10, we employed
transcripts containing point mutations that disrupt the pairing between the IGS
and the 5' or the 3' splice junction (Fig.
3
A). For transcripts with a defect in P10 no ligation of exons was observed in
autocatalytic reactions (Fig.
3
A, lane 3). This transcript, bearing a G+2C mutation, was subject to hydrolysis
at the 3' splice site, giving rise to the free linear intron, despite the lack of
ligation. Performing the reaction in the presence of Cbp2 (lane 2), however,
permitted ligation to proceed. This was also true if the alteration affecting
this base pair was in the guide sequence rather than the 3' exon (data not shown). The 836 nt intron-3'-exon species accumulated in the Cbp2-stimulated reaction, but not as much as one
would expect given the relative rates of the first and second steps of splicing
for this transcript (Table
1
). In the autocatalytic reaction (lane 3) the rate of 5' cleavage was less than the rate of hydrolysis at the 3' splice site. Consequently, this species did not accumulate.
Changing the second base of the intron from guanosine to cytosine reduces
cleavage at the 5' splice site and inhibits the second step of splicing (
43
). While this mutation affects a G[middot]C pair in P1, it also influences the alignment of P1 and P10 (Fig.
1
). As seen in Figure
3
B, Cbp2 was able to reduce the impact of this mutation on both steps of splicing
(lane 2). This results implies that base-pairing at this position is not essential for Cbp2 to recognize the
precursor RNA and that the riboprotein complex is active despite the weakened
P1.
P1 and P10 overlap with respect to the 5' bases of the IGS (Fig.
1
), and there appears to be a competition for IGS-pairing between the 5' splice site and the 3' splice site (
43
). To test the importance of this competition, we mutated positions 3-5 of the intron to the sequence U-A-C, permitting the formation of a 9 bp P1 stem that fully
overlaps P10. This mutation blocked ligation of exons in both autocatalytic and
Cbp2-stimulated splicing (Fig.
3
C). In the presence of Cbp2 (lane 2), the true intermediates of splicing (5' exon and intron-3'-exon) accumulated. In contrast, in high salt
autocatalytic conditions (lane 3), hydrolysis at the 3' splice site was observed, yielding free linear intron and the dead-end product consisting of the 5'-exon-intron. Hydrolysis at the 3' splice site was twice as rapid for this
mutant than for the wild type precursor (data not shown). There was also
substantial cleavage at the internal cyclization site giving a fragment of 501
nt.
Since Cbp2 did not suppress the P1 mutation G(229)C, it is possible that this
mutation altered the structure of the RNA so that the protein could not bind
with normal affinity. This premise was tested using nitrocellulose filter
binding to measure the equilibrium binding between Cbp2 and precursor RNA
molecules (Fig.
4
). Using the double filter method of Wong and Lohman (
45
), we could detect no significant differences in binding of Cbp2 to wild-type transcripts, to transcripts of P1 mutant G(229)C and to transcripts
of P10 mutant C(224)G. As a control we also tested the affinity of Cbp2 for the
antisense transcript and found at least 3-fold lower
K
d
for this transcript compared to the wild-type
b
I5 transcript (Fig.
4
). This experiment appears to exclude a major rearrangement in tertiary
structure leading to reduced affinity for Cbp2 in these RNA variants.
The results presented in Figure
3
suggest that the Cbp2 protein did not eliminate the need for the P1 helix to
determine the 5' splice site. In contrast, a stable P10 helix was not required for the
completion of splicing in the presence of Cbp2. P9.0b has also been implicated
in the selection of the 3' splice junction of group I introns (13,15,17,47). In
b
I5, this helix would be comprised of two adenosines at positions 677 and 678
hydrogen bonded with two uridines at positions 736 and 737, the penultimate
residues of the intron (Fig.
1
). To determine the importance of this pairing for the second step of splicing
we mutated the two A's to U's and the two U's to A's.
As seen in Figure
5
A, mutating the U736A, U737A mutant (P9.0-3') significantly reduced the ligation of exons in the autocatalytic
reaction (lane 3 and Table
1
) but ligation was increased in the presence of Cbp2. The A677U, A678U mutation
(P9.0-5') had a more profound effect on splicing. The first step of
autocatalytic splicing (cleavage at the 5' splice junction) was reduced and the second step (ligation of exons) was
blocked completely (lane 6). Adding Cbp2 (lane 5) increased the efficiency of 5' cleavage, but ligation was still inhibited. This inhibition may be
related to the ambiguity of the 5' cleavage site associated with this mutation. The A677U, A678U mutation
led to cleavage at a site two nucleotides into the intron (i.e. following U2)
in ~40% of the transcripts (lane 5). The effects of these lesions on the
kinetics of splicing of
b
I5 were quantitated by measuring the initial rate of 5' cleavage and the extent of exon ligation (Table
1
).
Because Cbp2 suppressed mutations in either P9.0 or P10, we monitored the effect
of Cbp2 on a double mutation containing a defects in both P9.0 (A677U, A678U)
and P10 (G+2C). The results are shown in Figure
5
B. Lanes 1-3 show the results for the P10 mutation alone. This defect inhibited the
second step of autocatalytic splicing (lane 3) but not in the presence of Cbp2
(lane 2). By contrast, the double mutant inhibited at the first step (lane 6),
but was spliced efficiently in the presence of Cbp2 (lane 5). The effects of
these mutations on the rate of 5' cleavage and the efficiency of ligation are summarized in Table
1
.
In both low salt buffer and in autocatalytic conditions, the 3' and 5' splice junctions are resistant to modification by hydroxyl
radicals, CMCT and RNase A (
39
). In contrast, nucleotides comprising the IGS are susceptible to modification
with these reagents in low salt buffer in the absence of Cbp2. Adding the
protein or high salt protects these residues, indicating that the protein
permits docking of the P1/P10 helices through tertiary interactions with the
intron core (
9
,
10
).
To examine the potential for base pairing and tertiary interactions within the 3'-end of
b
I5, we analyzed this region by modification with hydroxyl radicals and CMCT. The
former reagent modifies bases resulting in cleavage irrespective of secondary
structure but is sensitive to solvent accessibility. CMCT modifies uridines
and, to a lesser extent, guanidines and modifies unpaired residues
preferentially. Figure
6
shows a portion of the results for hydroxyl radical (lanes 1-4) and for CMCT (lanes 5-8) modification with a schematic indicating the protected bases.
The schematic is based on a series of experiments probing bases in a larger
region than that shown in these autoradiograms.
As an additional test of the importance of P10, we constructed a competitor of
this helix by instituting a hairpin within the downstream exon. The hairpin
arose through an A to G transition at position 13 of the exon (Fig.
7
) and sequestered 4 bases (GUAC) that normally pair with the IGS to form P10.
Existence of this structure was supported by partial RNase A digestion of 3'-labeled precursor RNA (data not shown). Transcripts bearing the
A(+13)G mutation were resistant to cleavage in the region predicted to form the
stem-loop structure, as expected for this single-strand specific nuclease. This mutation did not affect the affinity
of Cbp2 for
b
I5 (Fig.
4
).
The catalytic activity of group I ribozymes depends on nucleotide-nucleotide interactions such as hydrogen bonding and base stacking. These
contacts determine the secondary and tertiary structures of the ribozymes. This
work demonstrates that base pairs required for autocatalytic splicing are not
essential for protein-facilitated splicing. This result indicates that the binding of the
protein imparts structural stability on the RNA. We have begun our analysis on
the specificity of the exon binding domain of this group I intron. It is clear
that the analysis can be extended to the catalytic domain as well. Mohr
et al.
(
30
) concluded that the CYT-18 protein of
Neurospora
can suppress mutations in the T4
Td
intron by stabilizing the intron core. Gampel
et al.
(
48
) have shown that Cbp2 is not able to suppress a mutation in the guanosine
nucleotide binding site of
b
I5. A systematic study of the
b
I5-Cbp2 interaction is worthwhile, since Cbp2 is the physiologic partner of
this ribozyme in mitochondria.
We have previously shown that Cbp2 stabilizes tertiary interactions within
b
I5 so that the IGS surface of P1 becomes inaccessible to solvent (
39
). Our present data (Fig.
6
) suggest that splicing conditions also render the guanosine binding site in P7
and P9.0a less accessible to modification. We infer that Cbp2, by stabilizing
P1 and leading to its docking with the intron core, also fulfils the
requirements for selection of the 3' splice site. Our results suggest that the formation of P9.0a correlates
with the tertiary contact required for 3' splice site selection. Our results are consistent with those of Jaeger
et al
. (
16
) who suggested that P9.0a contributes to the stability of the
sunY
ribozyme. The potential for forming a G[middot]C pair between analogous positions is conserved among group I introns (
17
).
There is evidence from the
Tetrahymena
ribozyme that the guanosine binding site in P7 binds the 3' terminal guanosine of the intron following the first step of splicing (
13
,
18
,
49
). The rate of the ligation reaction is significantly reduced by mutating the
conserved guanosine at the 3'-end of the intron (
18
,
19
,
49
).
Tetrahymena
introns disrupted in both P9.0 and P10 could catalyze exon ligation at rates
between 1 and 30% that of the wild-type intron (
13
). We found that Cbp2 suppressed mutations in both P9.0 and P10 in
b
I5 (Fig.
5
B) that block ligation under autocatalytic conditions. The effect of the double
mutation was surprising for two reasons. First, Cbp2 permitted the second step
of splicing in a transcript lacking two determinants of the 3' splice site. Second, Cbp2 stimulated ligation of the double mutant was
more efficient than that of either of the single mutants. This result suggests
that the interference imposed by the P9.0-5' mutation (A677U, A678U) was reduced by the P10 mutation. The P9.0-5' mutation leads to a misalignment of P1 with the
intron core [as suggested by its effect on the 5' splice site (Fig.
5
A, lane 5)]. The P10 mutation G+2C affects a position at which P1 and P10 are in
competition for binding the IGS (Fig.
1
) (
43
). Reducing the affinity of the 3' exon for the IGS may, consequently, permit a more stable P1. In
addition, Cbp2, by constraining the flexibility of the catalytic core of
b
I5, might increase the productive binding of the 3' terminal guanosine in the absence of P9.0 or P10.
Our analysis of intron mutants demonstrates that sequences near the 3' splice site influence the first step of splicing. The fact that altering
the 5' component of P9.0b had greater impact than mutating its 3' pairing partner suggests that the 5' sequence has importance in addition to its participation
in P9.0. The proximity of the A677U, A678U mutant to the guanosine binding site
in P7 may explain the effect of this mutation on the first step of splicing.
Most group IA introns contain two purines in the 5' component of P9.0b (
17
). In contrast, if the P9.0 structure were more important than the sequence of
the dinucleotides that comprise it, then restoring the pairing by construction
of a double mutation should restore activity. In fact, the double mutant
spliced inefficiently under autocatalytic conditions (Fig.
5
A, lane 9) but was able to splice in the presence of Cbp2 (lane 8). It would be
difficult to characterize these base changes as compensatory, because the
double mutant did not approach wild type activity (Table
1
). These data indicate that the P9.0b pairing does not play the same role in
b
I5 as it does in the
Tetrahymena
ribozyme, where similar mutations blocked ligation but compensated for each
other. Rather, the P9.0b pairing makes a limited contribution to selecting the
3' splice site, but does contribute to the 5' cleavage reaction (Fig.
5
A). Similar results have been reported by others (
50
,
51
). Cbp2 was able to reduce the impact of the incorrect nucleotides or the lack
of the P9.0 structure. We have also tested a guanosine to uridine mutation,
G(676)U, at P9.0a and found that it blocks the second step of splicing and can
be suppressed by Cbp2 (data not shown). This mutation had no effect on the
fidelity of 5' cleavage but greatly reduced that step.
Given the alignment of P1, P10 and P9.0 and their proximity to the guanosine
binding site in P7 (Fig.
1
), it is not surprising that mutations directed at the 3' splice junction should influence the 5' splice site and, conversely, that a mutation in P1, such as G2C,
should reduce the second step of splicing (
43
). Some, but not all, mutations affecting P1 were suppressible by Cbp2. In
particular, the base pair between C(-4) in the 5' exon and G(228) in the IGS was essential. While Cbp2 appears to
make contacts with the IGS and the 5' exon (
38
,
39
), these positions do not form UV-crosslinks with the protein.
The capacity of intron 5 to splice during the transcription reaction (Fig.
2
) and the ability of the exon 6 mutation A(+13)G to ligate co-transcriptionally (Fig.
8
B) implicate the protein in the sequential folding of the intron. We have
demonstrated that co-transcriptional addition of Cbp2 alters the folding of L1 during
transcription (
37
) and that Cbp2 stabilizes the secondary structure of L1 and P1 (
39
). The A(+13)G mutation inhibited the ligation of exons in the Cbp2-stimulated reaction, but did not affect the first step of splicing (Fig.
8
A), and had no impact on either step of autocatalytic splicing (lane 3). A(+13)G
does not change a known binding site of the protein on this RNA (
38
,
39
). The difference in the protein-stimulated and the autocatalytic reactions probably reflects the increased
flexibility of the RNA in the high-salt conditions relative to that in the protein complex. Emerick and
Woodson (
52
) noted that rRNA precursors containing the
Tetrahymena
IVS sometimes misfolded during transcription and must be re-folded for splicing to occur. It is apparent that the stem-loop in exon 6 causes a similar phenomenon for
b
I5. Adding Cbp2 during transcription permits the correct folding of the
determinants of the 3' splice-site. In low salt conditions, the alternative secondary structure
otherwise predominates. Burke and colleagues, using a model system based on a
group I intron from
Azoarcus
, demonstrated that the substrate for the ligation reaction is a P1-P10 complex (
53
,
54
). Because of the overlap between P1 and P10 in
b
I5, we view this complex as a dynamic rather than a static structure. The role
of Cbp2 in determining the 3' splice site seems related to its ability to secure P1 and to promote
tertiary interactions between P1 and the catalytic center of the intron (
38
,
39
).
This work was supported by a grant from the National Institute of General
Medical Sciences (RO1 GM12228). Support facilities were also provided by the
Interdisciplinary Center for Biotechnology Research and by the Center for Mammalian Genetics at the University of Florida.
Transcript
Cbp2-dependent splicing
Autocatalytic splicing
%5' cleavage
% ligation
%5' cleavage
% ligation
Wild type
4.7
a
64 +- 4
b
2.6
62 +- 2
P9.0-3'
3.6
43 +- 3
2.0
30 +- 1
P9.0-5'
2.2
1.4 +- 1
0.03
nd
c
P9.0-5'/P9.0-3'
0.1
25 +- 10
nd
nd
P10
3.7
4.9 +- 2
0.42
nd
P9.0/P10
1.6
62.5 +- 3
nd
nd
Cbp2 did not overcome the effects of a point mutation in the IGS that disrupts
P1 (lanes 4-6). An identical result was obtained if the point mutation affecting this
base pair was in the 5' exon (data not shown). A double mutation that restored base-pairing (a C to G replacement in exon 5 and G to C in the IGS) also
restored the ligation of exons in the autocatalytic reaction (lanes 7-9). Additional fragments were produced as a consequence of guanosine
addition at a site in the 3' exon that resembles the 5' splice site (
46
). This competing reaction was suppressed by the addition of Cbp2 (lanes 8 and
9) indicating that the specificity of splicing was increased.
REFERENCES
Return