ABSTRACT
We have previously proposed a hierarchical model for the folding mechanism of
the
Tetrahymena
ribozyme that may illustrate general features of the folding pathways of large
RNAs. While the role of elements in the conserved catalytic core of this
ribozyme during the folding process is beginning to emerge, the participation
of non-conserved peripheral extensions in the kinetic folding mechanism has not yet been addressed. We now show
that the 3
'
-terminal P9.1-P9.2 extension of the
Tetrahymena
ribozyme plays an important role during the folding process and appears to
guide formation of the catalytic core.
Group I introns catalyze their own excision from pre-RNA in a two step transesterification reaction which results in ligation
of the flanking exons and release of the free intron (
1
). They can be found in the genomes of a wide variety of organisms and are
characterized by a set of highly conserved base paired regions which together
form a core containing all the structural elements required for catalysis (
2
,
3
). Based on phylogenetic comparisons of the core sequences from a large number
of introns, a model for the three-dimensional architecture of the catalytic core has been proposed (
4
,
5
). The model shows the existence of two helical subdomains connected by a triple
helical scaffold, which probably orients the subdomains with respect to each
other (
6
-
8
). In addition to the catalytic core, group I introns contain a series of less
well-conserved peripheral extensions, whose variation defines subgroups of
related introns (
4
). While the extensions are not absolutely required for catalytic activity, they
are important for stabilizing the catalytic core to allow formation of the
active structure of the RNA at physiological Mg
2+
concentrations (
2
,
9
-
13
). For RNAs lacking some peripheral extensions their function may have been
taken over by proteins that specifically recognize the core structure. There is
evidence, for example, that loss of the P5abc extension in the
Tetrahymena
ribozyme can be complemented by the CYT-18 protein (
14
).
The group I intron from
Tetrahymena
pre-rRNA has been extensively characterized and serves as a model system for
studying RNA folding and catalysis (Fig.
1
) (
1
). In addition to the self-splicing reaction, a shortened version of the intron missing 21 nt from the 5'-end (the L-21
Sca
I
Tetrahymena
ribozyme) can catalyze cleavage of short oligoribonucleotides in
trans
with multiple turnover. The L-21
Sca
I ribozyme is moderately large (388 nt) and, like most ribozymes, requires Mg
2+
for formation of its active structure and catalysis. In the presence of Mg
2+
the catalytic core forms a single globular structure (
15
,
16
). While the two subunits of the core have previously been referred to as
domains (
8
,
10
,
17
), we suggest that the term subdomain more accurately describes their identity
as substructures of a globular domain. The P4-P6 subdomain includes the conserved P4 and P6 stems and is independently
stable outside of the context of the whole intron (
17
). The P3-P7 subdomain includes the helices P3 and P7, which are formed by base pairing between regions relatively far
apart in the linear sequence of the RNA.
We have previously proposed a model for the kinetic folding pathway of the
Tetrahymena
ribozyme, including both Mg
2+
-dependent and Mg
2+
-independent steps, in which the two main structural subdomains form hierarchically (
6
,
18
). The P4-P6 subdomain forms first and the overall rate limiting step is a Mg
2+
-independent rearrangement preceding stable formation of the P3-P7 subdomain. The P3 and P7 helices form in an interdependent
manner and the observed structural subdomains appear to correspond to kinetic
folding units. While the rate limiting step in our model may represent a number
of microscopic folding events, we have shown that one of these events is the
formation of the triple helical scaffold (
6
). This hierarchical model for folding of the group I ribozyme is consistent
with studies of Mg
2+
-induced folding at equilibrium (
10
,
19
).
Oligodeoxynucleotide probes were synthesized on a 1 [mu]mol scale on an Applied Biosystems DNA synthesizer, deprotected overnight at
65oC in 2 ml concentrated ammonium hydroxide and purified on 20% denaturing
polyacrylamide gels. Full-length bands were excised and eluted from the gel overnight into water at
4oC, followed by desalting on C
18
Sep-PaksR (Waters).
Full-length L-21
Sca
I and truncated L-21
Nhe
I ribozymes were prepared by transcription from plasmid pT7L-21 (
20
) linearized with
Sca
I or
Nhe
I (New England Biolabs) respectively. Transcription reactions (100 [mu]l) were performed in 40 mM Tris-HCl, pH 7.5, 2 mM spermidine, 10 mM dithiothreitol, 25 mM MgCl
2
, 1 mM each GTP, CTP and UTP, 0.1 mM ATP and 250 [mu]Ci [[alpha]-
32
P]ATP (3000 Ci/mmol; New England Nuclear) for 3.5 h at 37oC using 500 U T7 RNA polymerase (New England Biolabs) and 15 [mu]g linearized plasmid template. Full-length transcription products were purified on 6% denaturing
polyacrylamide gels and eluted from the gel overnight at 4oC into buffer containing 10 mM Tris-HCl, pH 7.5, 1 mM EDTA and 0.3 M sodium acetate. The RNA was ethanol
precipitated and resuspended in 10 mM Tris-HCl, pH 7.5, 0.1 mM EDTA. Ribozyme concentrations were determined by
Cerenkov counting.
To determine folding rates the oligonucleotide hybridization assay was performed
as described (
18
). RNA (final concentration 1 nM) in 60 [mu]l buffer containing 1 mM Tris-HCl, pH 7.5, 0.01 mM EDTA was annealed by heating to 95oC for 45 s followed by equilibration at 37oC for 3 min. Folding was initiated by addition of an equal
volume of 2* folding buffer (1* = 50 mM Tris-HCl, pH 7.5, 10 mM MgCl
2
, 10 mM NaCl, 1 mM dithiothreitol), aliquots (10 [mu]l) were taken at the times indicated and added to 10 [mu]l 1* folding buffer containing oligonucleotide probe and RNase H (final concentration 0.1 U/[mu]l; United States Biochemical). Oligonucleotide binding and RNase H cleavage were allowed to
proceed for 30 s before the reaction was quenched with 14 [mu]l stop solution (90 mM EDTA and marker dyes in 82% formamide). The zero time points were obtained by adding oligonucleotide probe and RNase
H in 2* folding buffer immediately after annealing in a separate reaction. Final
probe concentrations were 20 [mu]M for the P3 probe and 60 [mu]M for the P6 probe. Products were separated on 6% denaturing
polyacrylamide gels and quantitated using a Molecular Dynamics PhosphorImager.
The data were fitted to a single exponential,
f
=
f
equ.
+ (
f
0
-
f
equ.
) exp(-
k
obs.
t
), where
f
is the fraction cleaved at time
t
,
f
equ.
is the fraction cleaved at equilibrium,
f
0
is the fraction cleaved at time
t
= 0 and
k
obs.
is the observed rate constant. This fitting procedure allows the end points to vary and independent values for the rate constant and equilibrium end point
are obtained.
To obtain the Mg
2+
concentration dependence of folding at equilibrium, the Mg
2+
concentration in the folding buffer was varied and kinetic experiments as
described above were performed at each Mg
2+
concentration. During the probe binding/RNase H cleavage step the Mg
2+
concentration was always adjusted to 10 mM. The equilibrium end point from the
fit of the data to single exponentials yielded the fraction of RNA folded at
equilibrium for each Mg
2+
concentration. The data were fitted to an expression for two state binding of
n
Mg
2+
ions,
f
= 1/{([Mg
2+
]/[Mg
2+
]
1/2
)
n
+ 1}, where
f
is the fraction cleaved at equilibrium,
n
is the number of Mg
2+
ions bound per RNA molecule and [Mg
2+
]
1/2
is the mid point of the transition.
To observe RNA folding kinetically as well as at equilibrium, we have developed a kinetic assay based on hybridization of complementary
oligodeoxynucleotide probes and RNase H cleavage (
18
). Folding is induced by the addition of Mg
2+
, and the fraction of RNA still accessible to oligonucleotide binding and/or
RNase H cleavage is determined at increasing folding times. The rate of the
transition from the accessible, unfolded to the inaccessible, folded state can
thus be measured. The assay requires that there is a substantial difference in
the accessibility of the RNA to the probes at different Mg
2+
concentrations. This condition is only met for certain regions of the RNA,
including the P3 and P7 helices in the P3-P7 subdomain and the P4 and P6 helices in the P4-P6 subdomain (
18
). Using different oligonucleotide probes folding of both the P4-P6 and the P3-P7 subdomains can therefore be followed specifically and
independently.
To test the effect of the P9.1-P9.2 extension on folding of the P3-P7 subdomain we examined the folding kinetics of the L-21
Nhe
I ribozyme, in which the extension is deleted, using an oligonucleotide probe
targeting the P3 helix (Fig.
1
). Since the P3-P7 subdomain forms in an interdependent manner, the probe targeting P3
reports on formation of the entire subdomain (
6
). The folding rate in L-21
Nhe
I RNA was decreased compared with full-length L-21
Sca
I ribozyme (Fig.
2
) (
k
obs.
= 0.32 +- 0.04/min for L-21
Nhe
I versus 0.72 +- 0.14/min for L-21
Sca
I). Because the rearrangement monitored with this probe is a slow step required for stable formation of the P3-P7 subdomain (formation of intermediate I
3
from I
2
; see Fig.
3
), the decreased rate suggests that stabilization of the P3-P7 subdomain by the P9.1-P9.2 extension may be involved in this slow step. Alternatively,
deleting the 3'-terminal extension may have resulted in a change in the rate
limiting step.
To compare our data with the published equilibrium Mg
2+
-dependence of P3-P7 formation in the context of the L-21
Nhe
I ribozyme (
10
) we determined the fraction of RNA in which P3 is accessible to our
oligonucleotide probe at equilibrium at a series of Mg
2+
concentrations (Fig.
2
B). The fraction of ribozyme folded at each Mg
2+
concentration was obtained from the equilibrium end point of a kinetic experiment as described above. Each point therefore represents an
independent experiment. The mid point of the transition ([Mg
2+
]
1/2
= 3.3 mM) was shifted to a higher Mg
2+
concentration compared with L-21
Sca
I ([Mg
2+
]
1/2
= 0.97 mM) (
18
).
We have continued our investigation of the kinetic folding pathway of the
Tetrahymena
group I intron, which serves as a model system for RNA folding, by examining
the role of a peripheral extension outside the catalytic core of the ribozyme
during folding. Deletion of the P9.1-P9.2 extension at the 3'-end of the ribozyme decreases the rate of the slow step in
our proposed minimal kinetic scheme (Fig.
3
). This slow rate represents the conversion of intermediate I
2
, in which the P4-P6 subdomain but not the P3-P7 subdomain is formed, to the transient intermediate I
3
. I
3
is competent to rapidly bind Mg
2+
, allowing stable formation of the P3-P7 subdomain and thus I
F
, an intermediate in which both the P4-P6 and the P3-P7 subdomains are present (
18
). Folding of the fast forming P4-P6 subdomain, and therefore formation of I
2
, is not perturbed in L- 21
Nhe
I RNA, nor has Mg
2+
binding become rate limiting. The decreased rate is therefore due to a direct
effect on the slow step in the minimal folding mechanism and not to a change in
the rate limiting step within our kinetic scheme.
The observed decrease in the folding rate upon deletion of the P9.1-P9.2 extension is only 2-fold and therefore not very dramatic. The folding rate of L-21
Sca
I RNA was measured numerous times with separate preparations of each reagent
over a period of almost 1 year and in none of these experiments was a rate as
low as that observed here for L-21
Nhe
I RNA ever measured. Conversely, the folding rate of the L-21
Nhe
I ribozyme was independently determined several times using separate
preparations of reagents and the rate was never in the range observed for wild-type RNA. The effect is therefore highly reproducible, which leads us to
believe that the 2-fold difference, although small, is real.
Chemical modification experiments have provided evidence for a tertiary
interaction between the 3'-terminal extension and bases in loops L2 or L2.1, which may help
lock the P3-P7 subdomain in place (
11
) and be responsible for stabilization of P3-P7 (
10
). In the
sunY
group I intron a similar stabilization of the core by elements of a 3'-terminal domain was observed (
9
). Our results suggest that the proposed tertiary interaction between the P9.1-P9.2 extension and L2 or L2.1 in the
Tetrahymena
intron is formed during the slow step of the minimal folding mechanism and is
therefore present in the transient intermediate I
3
. The identity of the nucleotides in P9.1-P9.2 and L2/L2.1 participating in the tertiary interaction has not yet
been established and it was therefore not possible to test the involvement of
this interaction more specifically. We propose that the P9.1-P9.2 extension is important not only for stabilization of the P3-P7 subdomain, but also helps guide folding of the RNA by limiting
the mobility of P3 and P7 and, along with the triple helical scaffold,
promoting proper association of the P4-P6 and P3-P7 subdomains.
The slow formation of I
3
from I
2
may consist of several microscopic steps, including formation of the triple
helical scaffold connecting the two subdomains and possibly base pair formation
in P3 and P7. We have shown that the triple helical scaffold is formed during the slow step (
6
), but can at present not distinguish whether formation of the triple helix
scaffold and stabilization of the P3-P7 subdomain by the P9.1-P9.2 extension are discrete steps or whether they occur in a
concerted rearrangement. A double mutant ribozyme in which both the 3'-terminal extension was deleted and the triple helix scaffold was
disrupted was too unstable to allow careful analysis of folding kinetics (data
not shown).
Determination of the Mg
2+
dependence of formation of the P3 helix in L-21
Nhe
I RNA showed a shift of the transition mid point to a higher Mg
2+
concentration, and confirmed published observations (
10
,
11
) that the P3-P7 subdomain is destabilized in this truncated ribozyme. While the exact
value of the transition mid point ([Mg
2+
]
1/2
= 3.30 mM) differs somewhat from that observed using Fe(II)-EDTA as a footprinting probe ([Mg
2+
]
1/2
= 1.83 mM;
10
), the difference in the apparent mid points is not surprising given the large
difference in size between the two probes used [small hydroxyl radicals
generated in the presence of Fe(II)-EDTA versus large oligonucleotides]. The different temperatures at which
the experiments were performed (42oC for the hydroxyl radical footprinting, 37oC for the kinetic oligonucleotide hybridizations) may also contribute
to this discrepancy. Although formation of I
3
from I
2
itself does not involve binding of Mg
2+
, the observed increase in the equilibrium Mg
2+
requirement for P3-P7 subdomain formation can at least in part be accounted for by the
decreased rate of formation of I
3
. Since rapid Mg
2+
binding to I
3
drives the otherwise unfavorable conversion of I
2
to I
3
(Fig.
3
), a decrease in the rate of this conversion will result in an increase in the
Mg
2+
concentration required for formation of I
F
, even if the dissociation constant of Mg
2+
binding to I
3
is unaltered. The present kinetic results thus complement previous observations
made at equilibrium (
10
,
11
).
In the cellular environment it may be that production of mature RNA for
transcripts containing group I introns is limited by proper folding of the
intron, rather than by catalysis. Partially or completely unfolded RNA may be
bound rapidly and non-specifically by proteins, which can inhibit formation of the active
structure. Even a relatively minor increase in folding rate afforded by a
peripheral extension may therefore confer a selective advantage by preventing
the core of the RNA from becoming kinetically trapped in unproductive
conformations or complexes.
These results may have implications for the folding mechanisms of other large
RNAs, many of which also consist of phylogenetically highly conserved core
regions surrounded by less conserved peripheral extensions. Such an
organization can be found in group I (
4
,
5
) and group II introns (
21
), in the RNA component of RNase P (
22
) and in rRNAs (
23
,
24
). In several cases it has been demonstrated that peripheral extensions can
stabilize the core region of a large RNA (
2
,
9
,
10
-
12
), suggesting that some general functions of the extensions may be conserved
between different RNAs. The importance of the P9.1-P9.2 extension during folding of the
Tetrahymena
ribozyme suggests that, in addition to stabilizing the final conformation, non-conserved peripheral extensions may also guide the folding process in large, highly structured RNAs.
We thank Dan Treiber and Martha Rook for critical reading of the manuscript.
This work was supported by grants from the Searle Scholar Program of the
Chicago Community Trust, the Rita Allen Foundation and the Camille and Henry
Dreyfus Foundation. P.P.Z. is a pre-doctoral fellow of the Howard Hughes Medical Institute.
REFERENCES
Return