Dissociation of long-chain duplex RNA can occur via strand displacement
in vitro
: biological implications
Dissociation of long-chain duplex RNA can occur via strand displacement in vitro : biological implications
Matthias
Homann
+
,
Wolfgang
Nedbal
and
Georg
Sczakiel*
Received August 27, 1996;
Accepted September 30, 1996
ABSTRACT
Hammerhead ribozymes with long antisense flanks (>50 bases) have been used
successfully to inhibit replication of human immunodeficiency virus type 1 (HIV-1) in living cells. To explain their increased efficacy versus antisense
controls or catalytically inactive derivatives, one can consider dissociation
of the ribozyme-product complex to allow a complete catalytic cycle. In this work we
investigated the dissociation of a double-stranded RNA with 56 bp
in vitro
. Dissociation was observed in the presence of single-stranded RNA with sequence complementarity to one of the duplex strands. A
displacement reaction between RNA single strands and the duplex, but not simple dissociation, was strongly
suggested by the concentration dependence of this process, the influence of additional non-complementary sequences on the single strand and by the unusually low
Arrhenius activation energy. The strand displacement reaction was slow
in vitro
at 37
o
C and physiological ionic strength, but was increased to
k
[approx]
10
3
-10
4
/M/s (
~
10
4
-fold) at higher temperatures by cetyltrimethylammonium bromide. This
compound is thought to enhance non-sequence-specific association of nucleic acids in a mechanistically similar
way to that in which cellular hnRNP proteins are thought to act, indicating
that strand displacement can be fast and, more importantly, could be tightly
regulated
in vivo
.
INTRODUCTION
Dynamic and regulatable interactions between cellular macromolecules are
biochemical prerequisites for living cells. Interactions of complementary RNAs
play a key role in the regulation of biological processes such as mRNA
translation, the splicing of pre-mRNA (
1
,
2
) and pathways involving antisense RNA (
3
). The association of complementary RNA has been described in thermodynamic and
kinetic terms in many cases (e.g.
3
-
7
). Kinetic models are appropriate in the case of naturally occuring antisense
RNAs, which belong to the best-studied long-chain complementary RNAs (
3
). The dissociation of RNA duplexes
in vitro
was found to be slow in most cases, depending on the length and base
composition of the double strand (
8
). The rate constants ranged between 10
-2
and 10
-3
/s for 10 bp duplexes (
9
) and even lower values were observed (
k
= 10
-4
/s) for dissociation of cleavage products in the case of short
trans
-cleaving hammerhead ribozymes forming 16 bp with their target RNA (
10
). However, little is known about the dissociation of longer RNA duplexes which
occur, for example, in the spliceosome (
11
) or when long hammerhead ribozymes with antisense flanks of 50-280 bases are applied intracellularly (
12
-
14
). In the latter case, the dissociation of long duplexes, i.e. the dissociation
of the ribozyme-product complex, is a prerequisite for the catalytic turnover of
ribozymes, which could explain the observed stronger inhibitory effects versus
antisense controls
in vivo
(
12
,
14
).
Here we describe the dissociation of a 56 bp duplex RNA in the presence of a
displacing single-stranded RNA at physiological ionic strength and temperature. The dissociation rate was dependent on the
presence and concentration of the RNA single strand with sequence
complementarity to one of the strands forming the duplex. These findings
suggest an associative mechanism rather than simple dissociation, i.e. the
displacing single strand interacts with the duplex and replaces the homologous
strand via formation of a ternary complex. This process could be strongly
enhanced in the presence of cetyltrimethlyammonium bromide (CTAB), a model
compound for cellular facilitator proteins (
15
).
MATERIALS AND METHODS
RNA synthesis
RNA was synthesized by run-off transcription
in vitro
from linearized plasmids by T7 RNA polymerase in a 200 [mu]l reaction mixture containing 10 [mu]g DNA template, 18 mM Na
2
HPO
4
, 2 mM NaH
2
PO
4
, 8 mM MgCl
2
, 20 mM dithiothreitol, 4 mM spermidine, 1 mM each NTP, 5 mM NaCl and 50 U T7
RNA polymerase. For the synthesis of [alpha]Y69 the plasmid p[alpha]Y69 was linearized with
Bgl
II, for the synthesis of [alpha]Y150 the plasmid p[alpha]Y150 was linearized with
Xho
I and for the synthesis of SR6 and AR6 the plasmids pRC-CMV-SR6 and pRC-CMV-AR6 were linearized with
Not
I as described before (
16
). The ribozyme [alpha]YRz195 was transcribed from
Eco
RI-linearized plasmid p[alpha]YRz195 (
17
). It contains 195 nt complementary to HIV-1 and 40 non-complementary nucleotides, including a hammerhead ribozyme domain at
its 5'-end (
17
).
Preparation of RNA double strands
The antisense RNA [alpha]Y69 (
16
) was radioactively labelled by incorporation of [[alpha]-
32
P]UTP (250 [mu]Ci) during
in vitro
transcription in 50 [mu]l transcription buffer (see above) lacking unlabelled UTP. Aliquots of 50 ng
labelled [alpha]Y69 were annealed with the complementary RNA SR6 (5 [mu]g) for 5 h in 200 [mu]l buffer containing 100 mM NaCl, 20 mM Tris-HCl, pH 7.4, and 10 mM MgCl
2
. RNase T1 was added (1000 U, 3 h) to remove single-stranded overhangs. The mixture was phenol extracted five times and the
resulting RNA duplex, ds56, was precipitated once in 0.3 M NaOAc, 2.5 vol
ethanol and subjected to gel filtration on Sephadex G50. The complete sequence
of the antisense strand of ds56 is 5'-AUCU
CCUUGAGGAGGUCUUCGUCGCUGUCUCCGCUUCUUC- CUGCCAUAGGAGAGCCUAAG
-3'. The 56 double-stranded bases in ds56 are underlined. Due to the fact that
RNase T1 cleaves 3' of G residues, ds56 also contains an AUCU overhang at the 5'-end of the antisense strand (Figs
1
and
2
B, `as') and an overhanging C at the 5'-end of the sense strand (Figs
1
and
2
B, `s').
Dissociation kinetics
The RNA double strand ds56 (0.4 nM) was incubated together with at least four
different concentrations of unlabelled single-stranded RNA (e.g. 0.4-5.2 [mu]M AR6) in 100 mM NaCl, 20 mM Tris-HCl, pH 7.4, 10 mM MgCl
2
for 6, 12 and 24 h at 37, 47 and 57oC in a volume of 6 [mu]l. The reaction was stopped by the addition of 15 [mu]l stop buffer (7 M urea, 50 mM Tris-HCl, 20 mM EDTA and an additional 0.5% SDS when 1 mM CTAB
was used in the dissociation experiment) and the reaction products were
analysed on 15% polyacrylamide gels containing 7 M urea. The gels were run at
room temperature at 10 V/cm for 12 h in a buffer containing 89 mM Tris, 89 mM
borate and 2.5 mM EDTA. The dissociation rates were determined by
quantification of the radioactivity in single-stranded and double-stranded RNA bands with a PhosphorImager (Molecular Dynamics). The
percentage of released single strand was plotted against time and the
individual half-lives (
t
1/2
) were determined graphically.
k
obs
can then be calculated from
k
obs
= ln2/
t
1/2
. Alternatively,
k
obs
can be derived by fitting plots with the computer program GRAFIT (Erithacus
Software, London, UK), giving identical results.
RESULTS
The dissociation of two associated complementary RNAs may follow either of two
kinetic schemes (Fig.
1
): first, a dissociative mechanism, i.e. dissociation of the RNA duplex occurs
by melting and separation of the complementary strands (Fig.
1
A); second, dissociation can be driven by a displacing third strand with
sequence complementarity to one of the two duplex-forming strands via the formation of a ternary complex (Fig.
1
B) which, in principle, is similar to some cases of enzyme product release (
18
). The two mechanisms can be distinguished by kinetic means. The associative
mechanism implies that the observed dissociation rate is not dependent on the
concentration of the `capture' strand provided that the association step (Fig.
1
B, represented by
k
3
) is rate limiting. Under these conditions the kinetics follow a second order
reaction and a pseudo first order reaction if either the displacer strand or
the duplex is in large excess. Conversely, the dissociative mechanism implies
that the observed dissociation rate is not dependent on the concentration of
the capture strand provided that the dissociation step is slower than re-association with the capture strand (Fig.
1
A). In this case, dissociation is also independent of biochemical properties of
the single strand, such as length and structure.
Dissociation rates are concentration dependent
To distinguish between these two possible mechanisms, we chose the HIV-1-derived antisense RNA [alpha]Y69 (69 nt) and its complementary target RNA SR6 (645 nt;
Fig.
2
A), which has been described in terms of structure and association kinetics (
16
). To synthesize the duplex RNA ds56, [alpha]Y69 was hybridized with SR6 and the resulting complex was treated with
RNase T1 to remove single-stranded overhangs, leaving the double-stranded portion of 56 bp (Fig.
2
B). The dissociation of ds56 was investigated in the presence of a large molar
excess of the single-stranded RNA AR6 at 57oC (Fig.
3
A). We observed a dependence of the rate of the dissociation reaction on the
presence and concentration of AR6 (Fig.
3
B). Thus, the dissociative mechanism (Fig.
1
A) could be ruled out because this concentration dependence can only be
explained if the rate limiting step was association between the capture strand
AR6 and the released unlabelled single strand. The association rate constant
for this pair of RNAs is in the range 1 * 10
4
/M/s at 37oC and 1 * 10
5
/M/s at 57oC (data not shown). Thus, at the RNA concentrations used here, the half-lives of the association reaction should be in the range 1-10 s, which is at least 10
5
times faster than the observed half-lives for dissociation of ds56.
Dissociation rates are dependent on the length of the displacer strand
A number of single strands that are related to the displacer RNA AR6 and the
duplex ds56 respectively have significantly different displacing activities at
37 and 50oC (Table
1
). All RNA contain the same sequence stretch of 56 bases contained within the
duplex, however, they differ in the length of additional non-complementary sequences (Fig.
2
A). Though no simple length dependence is obvious in this experiment, it is
noteworthy that longer displacing strands, e.g. [alpha]YRz195 or AR6, led to 10- to 100-fold faster reaction rates as compared with the shorter [alpha]Y150 and [alpha]Y69. This observation is compatible with the
associative mechanism and may indicate an influence of the overall length
and/or secondary structure of the displacer strands on interaction with the
duplex. In contrast, this observation is not compatible with the dissociative
mechanism, since for all displacer strands association with the complementary
strand has been measured to be at least 10
5
-fold faster (Homann, unpublished results). Therefore, we conclude that a
strand exchange reaction takes place between duplex ds56 and single-stranded AR6 in which formation of a ternary complex is the rate limiting
step.
The activation energy of strand exchange
Further mechanistic insight comes from considering the Arrhenius activation
energy (
E
a
) of dissociation. The Arrhenius activation energy of dissociation of RNA double
strands with 6-18 bp was determined to be 21-25 kJ/mol/bp, irrespective of the length and base composition of
the double strand (
5
,
8
,
19
,
20
). For dissociation of ds56,
E
a
was calculated from the temperature dependence of
k
to be 153 kJ/mol (Fig.
6
) or 2.8 kJ/mol/bp. This value is nearly one order of magnitude smaller than
expected from an extrapolation of the data obtained for smaller duplexes (
5
,
8
,
19
,
20
) and, thus, is not compatible with a simple dissociation process.
Regulation of strand displacement
The rate limiting step in the proposed strand displacement reaction is
association of the duplex and the displacer strand with a second order rate
constant of
k
= 0.1/M/s at 37oC, which appears to be very small. However, strand displacement is strongly
enhanced in the presence of the low molecular weight compound CTAB (
21
), which is known to mechanistically mimic cellular proteins like hnRNP protein
A1 (
15
). At 1 mM CTAB, the reaction rates of strand displacement between ds56 and AR6
increased by a factor of 7000 at 37oC and 30 000 at 57oC (see table in Fig.
6
). This strong enhancement was accompanied by an increased instead of a lowered
Arrhenius activation energy (
E
a
= 215 kJ/mol versus 153 kJ/mol; Fig.
6
). A similar observation has been made in the case of the CTAB-mediated increase in the association rates of complementary RNA (Nedbal,
Homann and Sczakiel, unpublished data) and the results here are consistent with
the assumption that the first step of the associative pathway is enhanced,
which is association of the duplex and the single strand to form the ternary
complex.
.
Dependence of duplex dissociation on the displacing strand
Displacer
Total length
Temperature
Second order rate
RNA
(nt)
a
(oC)
constant
k
(per M/s)
AR6
645
37
0.1
50
2.0
[alpha]YRz195
235
37
0.1
50
2.0
[alpha]Y150
150
37
<0.01
50
<0.1
[alpha]Y69
69
37
<0.002
50
<0.01
a
All RNA contain the same 56 bases complementary to the sense strand of ds56 (Fig.
2
A and B) plus additional HIV-1-derived as well as vector-derived nucleotides.
DISCUSSION
In this work we describe a strand exchange reaction between a 56 bp duplex RNA
and RNA single strands with sequence complementarity to one of the duplex
strands. In accordance with an associative mechanism for dissociation (Fig.
1
B), the reaction is dependent on the concentration and biochemical properties of
the displacer strand, indicating that association of the duplex and the single
strand is the rate limiting step. Our kinetic data suggest the formation of a
ternary complex and the subsequent displacement of the homologous strand of the
duplex. We were unable to isolate the proposed ternary complex, probably due to
its fast dissociation. However, one can speculate on the nature of the ternary
complex and the molecular mechanism of strand displacement. For example,
association of the displacer RNA with the duplex could occur via melting of the
duplex ends, described as `breathing' (
6
,
8
), followed by invasion of the single strand and subsequent strand displacement.
In this respect it is noteworthy that the Arrhenius activation energy of 153
kJ/mol measured for the displacement reaction between ds56 and AR6 corresponds
to the melting of 7-8 bp in short RNA duplexes (
8
). A related reaction is known from recA-catalysed strand exchange in homologous recombination (for reviews see
22
,
23
). In this reaction, recA-covered double and single strands are aligned by non-Watson-Crick pairing before homologous pairing allows formation of a
stable joint molecule from which branch migration proceeds in the 5' -> 3' direction relative to the displaced strand (
24
). For the initial stable pairing, however, only 8 nt are necessary (
25
).
Alternatively, the duplex ends may adopt rather stable alternative secondary
structures, e.g. intramolecular fold-back structures, that are trapped by the single strands. In both cases,
the observed differences in the displacing activities of RNA single strands as
listed in Table
1
may be influenced by differences in their ability to interact with the ends or
certain terminal structures of the duplex. Such differences might be due to
different secondary structures of the displacer strands at sites complementary
to the duplex ends. However, when using ds56-derived RNA duplexes of 48 and 58 bp with duplex ends differing from ds56,
strand displacement was observed in a similar way (Homann, unpublished
results). This indicates that the described reaction is not limited to the
double strand ds56 or due to unique structural properties of the RNA molecules
used here.
One could speculate that an alternative way of formation of a ternary complex
involves formation of a partial triple helix between the duplex and the
displacer strand. There is a polypurine stretch in the sense strand of the
duplex ds56 and, correspondingly, polypyrimidine stretches in both antisense
strands (positions 28-42 in ds56). However, if a triple helix was formed, it is likely that it
is formed in an orientation in which the incoming displacer strand is
antiparallel with the homologous strand of the duplex. Consequently, it would
be parallel with its complementary strand of the duplex (C. Hélène, personal communication), which cannot explain strand exchange.
Alternatively, invasion of the duplex RNA could occur via a loop element of the
single strand, as is indicated by the interaction between a GNRA loop and the
minor groove of an RNA helix which was observed in a group I intron of the
bacteriophage T4 (
26
,
27
). However, since a GNRA loop cannot be found in the predicted secondary
structures of either of the displacer strands, there is no evidence supporting
this model.
The rate constant
k
= 0.1/M/s at 37oC for the displacement reaction seems to be too small to be relevant for
cellular reactions. However, many viral (
28
,
29
) and cellular proteins (
30
-
32
) are known to increase the rate constants for association of nucleic acids by
up to three orders of magnitude. In the presence of the facilitator CTAB, the
rate constants for strand exchange increased dramatically to
k
= 10
3
-10
4
/M/s (Fig.
6
). This observation supports the model of formation of the ternary complex as
rate limiting. Furthermore, the strong CTAB-mediated increase in strand exchange indicates that the reaction can be
regulated and could play an important role
in vivo
.
The results described here may have several biological implications. Firstly,
catalytic turnover of ribozymes could be imagined assuming displacement of
cleaved substrates of a ribozyme-product complex by an incoming uncleaved substrate in living cells which
is compatible with the observed stronger inhibitory effects of long-chain hammerhead ribozymes versus antisense controls (see for example
12
-
14
,
17
). We tested this possibility with an `asymetric' ribozyme (
17
) forming 63 bp with its substrate. In this case, the uncleaved substrate was
able to displace the cleaved product with a similar rate as was measured for
the strand exchange between ds56 and AR6 (M. Homann, unpublished results). In
view of the potential biological and clinical role of
trans
ribozymes, it may further be of interest to analyse target sequences with
respect to the ability to displace cleaved products from the ribozyme-product complex, which could be done by means of
in vitro
selection techniques. Here, the aim is to identify pairs of ribozyme-product RNAs where the RNA substrate serves as a good displacer and the
products are good leaving strands.
Further, at a more basic level, the energy barrier for RNA strand exchange may
be much lower than expected for simple dissociation, indicating that
interactions between RNA could be more dynamic than expected and would not
necessarily need helicases or single strand stabilizing proteins. There is, for
example, an extensive base pairing interaction between the small nuclear RNAs
U4 and U6 in the splicesosome (
2
). However, the switch from the inactive to the active spliceosome is correlated
with displacement of U4 from U6, which then forms base pairs with U2 to a
similar extent (
11
). According to the mechanism described in this work, U2 might play a more
active role in displacing U4 from U6. It should be noted, however, that neither
the intracellular concentrations of nucleic acids nor the influence of
subcellular localization on RNA-RNA interactions are known.
ACKNOWLEDGEMENTS
We thank H.zur Hausen for continuous support and interest. We also wish to
cordially thank C.Hélène, J.Reinstein and R.S.Goody for stimulating suggestions and
discussions, as well as P.Romby for helpful comments. This work was supported
in part by the Deutsche Forschungsgemeinschaft, the European Union (HCM) and
the Bundesministerium für Forschung und Technologie.
REFERENCES
1 Gesteland,R. and Atkins,J. (1993) The RNA World. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
