Unexpected anisotropy in substrate cleavage rates by asymmetric hammerhead
ribozymes
Unexpected anisotropy in substrate cleavage rates by asymmetric hammerhead ribozymes
Philip
Hendry*
and
Maxine
McCall
CSIRO, Division of Biomolecular Engineering, PO Box 184,
North Ryde
, NSW 2113,
Australia
Received April 11, 1996
;
Revised and Accepted June 4, 1996
ABSTRACT
RNA substrates which form relatively short helices I and III with hammerhead
ribozymes are generally cleaved more rapidly than substrates which create
longer binding helices. We speculated that for optimum cleavage rates, one of
the helices needed to be relatively weak. To identify this helix, a series of
ribozymes and substrates of varying lengths were made such that in the complex,
helices I and III consisted of 5 and 10 bp respectively or
vice versa.
In two independent systems, substrates in the complexes with the shorter helix
I and longer helix III were cleaved one to two orders of magnitude more rapidly
than those in the complexes with the longer helix I and shorter helix III.
Similar results were obtained whether the numbers of base pairs in helices I
and III were limited either by the length of the hybridizing arms of the
ribozyme or the length of the substrate. The phenomenon was observed for both
all-RNA and DNA armed ribozymes. Thus, a relatively short helix I is required
for fast cleavage rates in pre-formed hammerhead ribozyme-substrate complexes. When helix III has 10 bp, the optimum length
for helix I is
~5 bp.
INTRODUCTION
Hammerhead ribozymes were discovered as the self-cleaving motifs in a number of small, circular pathogenic RNAs in plants (
1
-
3
). Uhlenbeck (
4
) showed that the ribozyme was able to act in a bimolecular fashion and as a
true enzyme, i.e. each enzyme is able to cleave multiple substrates. Haseloff
and Gerlach (
5
) divided the hammerhead into a form in which the majority of the conserved
nucleotides were located on the enzyme strand, with the only sequence
requirements for the substrate being UH just 5' of the cleavage site (H = A, U or C) (
6
-
8
). Since 1988 this configuration has been the paradigm for hammerhead ribozyme
design. The cleavage kinetics for the hammerhead ribozyme are usually analysed
according to the series of reversible reactions shown in Scheme 1, followed by
random dissociation of the two cleavage products P
1
and P
2
(
9
). The cleavage step is generally assumed to be essentially independent of the
length and sequence of the helices I and III which are formed when the 5' and 3' arms of the ribozyme bind to the substrate (Fig.
1
) (
10
).
MATERIALS AND METHODS
Nomenclature
This study investigates the cleavage kinetics of a number of substrates by
various hammerhead ribozymes. The sequences of the substrate molecules are
taken from naturally occurring mRNAs and are identified by their origin. The GH
series are from rat growth hormone, the TAT series are from the
TAT
gene of HIV-1 and the Kr series are from the
Krüppel
gene of
Drosophila melanogaster
. Ribozymes are denoted by an R following the identifying prefix and substrates
by the letter S, which are further identified by a number denoting their length
in nucleotides, e.g. Kr S13. There are two versions of hammerhead ribozyme used
in this paper and they are denoted as ribozymes A and B. Ribozymes A (RA) are
composed solely of RNA (with the exception of the 3'-nucleotide), whereas ribozymes B (RB) possess DNA in the arms that
hybridize to the substrate, with the exception of nucleotides 15.1 and 15.2,
which remain as RNA (Fig.
1
). The numbering system of Hertel
et al
. (
12
) is used throughout. The number of nucleotides in the hybridizing arms (for
ribozymes) or on each side of C
17
(for substrates) is given as a suffix to the name, with the first number
referring to the 5'-side and the second to the 3'-side. For example, Kr S17-10/6 is a 17mer substrate with 10 nt 5' of the C
17
and 6 nt on the 3'-side and is the substrate complementary to the ribozyme Kr RA-6/10. Figure
2
shows a selection of some of the substrates and ribozymes from the Krüppel system as a visual aid to the system of nomenclature.
Preparation of oligonucleotides
Oligonucleotides were synthesized using an Applied Biosystems (Foster City, CA)
model 391 DNA synthesiser. Protected DNA phosphoramidite monomers were from
Millipore (Bedford, MA) or Auspep (Melbourne, Australia). RNA monomers,
protected at the 2'-hydroxyl with
t
-butyldimethylsilyl (
t
-BDMS) groups, were from the same sources. For convenience in the
syntheses, all the oligonucleotides have a deoxyribonucleotide at their 3'-end. Deprotection and purification of oligonucleotides were as
described previously (
13
), with the exception that removal of the
t
-BDMS group from the 2' position was achieved with the use of neat triethylamine
trihydroflouride for 24 h at room temperature, followed by precipitation of the
oligonucleotide with 10 vol 1-butanol prior to gel purification. The purity of each oligonucleotide was
checked by labelling its 5'-end with [
32
P]phosphate using T4 polynucleotide kinase (New England Biolabs, Beverly, MA)
and [[gamma]-
32
P]ATP (Du Pont, Wilmington, DE), electrophoresing the molecules on a 15%
polyacrylamide gel containing 7 M urea and visualizing the molecules by
autoradiography or using a Molecular Dynamics PhosphorImager system (Sunnyvale,
CA). The concentrations of the purified oligonucleotides were determined by UV
spectroscopy using the following molar extinction coefficients for the various
nucleotides at 260 nm: A, 15.4 * 10
3
; G, 11.7 * 10
3
; C, 7.3 * 10
3
; T/U, 8.8 * 10
3
l mol
-1
cm
-1
. All oligonucleotides were stored in distilled, deionized and autoclaved water at -20oC.
Oligonucleotide sequences
The oligonucleotides used in this study are as follows (capital letters refer to
ribonucleotides, lower case letters refer to deoxyribonucleotides).
Growth hormone system (GH).
GH RA-10/10, 5'-GACACUUCAU CUGAUGA GUCC UUUU GGAC GAAAC CCGCAGGt-3'; GH RB-10/10, 5'-gacacttcat CUGAUGA GUCC UUUU
GGAC GAAAC ccgcaggt-3'; GH S13-6/6, 5'-GCG GGU CAU GAA g-3'; GH S21-10/10, 5'-ACC UGC GGG UCA
UGA AGU GUc-3'.
Krüppel system (Kr).
Kr RA-10/10, 5'-CUCCAGUGUG CUGAUGA GUCC UUUU GGAC GAAAC UCGCAAAt-3'; Kr RB-10/10, 5'-ctccagtgtg CUGAUGA GUCC UUUU
GGAC GAAAC tcgcaaat-3'; Kr RA-6/10, 5'-AGUGUG CUGAUGA GUCC UUUU GGAC GAAAC UCGCAAAt-3'; Kr S13-6/6, 5'-GCG AGU CCA CAC
t-3'; Kr S21-10/10, 5'-AUU UGC GAG UCC ACA CUG GAg-3'; Kr S16-10/5, 5'-AUU UGC GAG UCC
ACA c-3'; Kr S16-5/10, 5'-C GAG UCC ACA CUG GAg-3'; Kr S14-10/3, 5'-AUU UGC GAG UCC
Ac-3'; Kr S15-10/4, 5'-AUU UGC GAG UCC ACa-3'; Kr S17-10/6; 5'-AUU UGC GAG UCC
ACA Ct-3'; Kr S18-10/7; 5'-AUU UGC GAG UCC ACA CUg-3'.
TAT system (TAT).
TAT RA-10/10, 5'-GUCCUAGGCU CUGAUGA GUCC UUUU GGAC GAAAC UUCCUGGa-3'; TAT RA-6/6, 5'-UAGGCU CUGAUGA GUCC UUUU GGAC
GAAAC UUCc-3; TAT RB-10/10, 5'-gtcctaggct CUGAUGA GUCC UUUU GGAC GAAAC ttcctgga-3'; TAT RA-10/5, 5'-GUCCUAGGCU CUGAUGA GUCC
UUUU GGAC GAAAC UUc-3; TAT RA-5/10, 5'-AGGCU CUGAUGA GUCC UUUU GGAC GAAAC UUCCUGGa-3'; TAT S13-6/6, 5'-GGA AGU CAG CCU a-3'; TAT S21-10/10, 5'-UCC AGG AAG UCA GCC UAG GAc-3'; TAT S16-10/5, 5'-UCC AGG AAG UCA GCC t-3'; TAT S16-5/10, 5'-GAA GUC AGC CUA GGA c-3'; TAT S14-10/3, 5'-UCC AGG AAG UCA GC-3'; TAT S15-10/4, 5'-UCC AGG AAG UCA GCc-3'; TAT S17-10/6, 5'-UCC AGG AAG UCA GCC Ua-3'; TAT S18-10/7, 5'-UCC AGG AAG UCA GCC UAg-3'; TAT S19-10/8, 5'-UCC AGG AAG UCA GCC UAG g-3'.
Ribozyme excess kinetic experiments
The experiments were performed at 37oC with ribozyme and substrate (labelled at the 5'-end with [
32
P]phosphate) in 10 mM MgCl
2
and 50 mM buffer (Tris, pH 8.0 or 7.13, or Mes, pH 6.42), using the following
procedure. The substrate concentrations were in the range 2-4 [mu]M (typically 2 [mu]M) and the ribozyme concentration was at least 1.5 times that of
the substrate (typically 3 [mu]M). In several experiments with Kr RB-10/10 and Kr S14-10/3, the substrate concentration was reduced to 100 nM and the
ribozyme concentration was kept at 3 or increased to 6 [mu]M. These control experiments were performed to check that at these
concentrations the substrate was fully bound by the ribozyme. The ribozyme and
substrate together in buffer were pre-treated by heating to 85oC for 2 min, centrifuging briefly and then placing at the reaction
temperature for a few minutes. The reaction was normally initiated by the
addition of MgCl
2
. In a control experiment, the ribozyme Kr RB-10/10 and substrates Kr S16-10/5 and Kr S21-10/10 were pre-incubated separately in buffer containing 10 mM MgCl
2
at 85oC and, after cooling to 37oC, the reaction was initiated by mixing the two solutions. Samples
were removed at various times and quenched by addition to 2 vol gel loading
buffer containing 90% formamide and 20 mM EDTA. The fraction of substrate
cleaved in each sample was determined by separation of the substrate from the 5' product in a 15% polyacrylamide gel containing 7 M urea and quantifying
the amounts of each using a Molecular Dynamics PhosphorImager system and
ImageQuant software (Molecular Dynamics, Sunnyvale, CA). The kinetic parameters
were obtained by fitting the data for percentage of product formed (
P
t
) at any given time (
t
) to the equation
P
t
=
P
8
- [exp(-
k
obs
t
)
P
D
)
where
P
t
is the amount of product at time
t
,
P
8
is the amount of product at
t
= [infinity],
k
obs
is the first order rate constant for the reaction and
P
D
is the difference between the percentage of product at
t
= [infinity] and
t
= 0. This is a first order kinetic equation from which
k
obs
,
P
8
and
P
D
are determined by least squares fitting of the data.
P
8
was typically in the region 0.6-0.8, i.e. ~60-80% of the substrate was cleaved at the end of the reaction.
The observed rate constants,
k
obs
, presented in the Tables are the means +- SD of at least two independent experiments.
RESULTS
Long armed ribozymes cleave short substrates faster than long substrates
Ribozymes with 10 nt in each hybridizing arm form complexes with symmetric 13 nt
substrates that contain 6 bp each in helices I and III, while the same
ribozymes form complexes with symmetric 21 nt substrates that have 10 bp in
each of the helices. As shown in Table
1
, such ribozymes cleave 13mer substrates much faster than the corresponding
21mer substrates, with the exception of TAT RA-10/10. The reactivity of TAT RA-10/10 appears to be anomalous and is apparently modulated by some
spurious base pairing (see Discussion). Since short double-helices are generally less stable than longer double-helices, the relative cleavage rates of 13 versus 21 nt substrates
by 10/10 ribozymes is consistent with a cleavage mechanism that requires one or
both of helices I and III to be relatively unstable.
.
Cleavage rate constants for symmetric 13 and 21 nt substrates by their cognate
ribozymes with 10/10 hybridizing arms
Ribozyme
S13-6/6
S21-10/10
III/I
a
k
obs
(min
-1
)
III/I
a
k
obs
(min
-1
)
GH RA-10/10
b
6/6
1.6 +- 0.6
10/10
0.5 +- 0.2
GH RB-10/10
b
6/6
5 +- 1
10/10
1.9 +- 0.2
TAT RA-10/10
c
6/6
0.24 +- 0.05
10/10
0.6 +- 0.1
TAT RB-10/10
c
6/6
4.5 +- 0.3
10/10
0.44 +- 0.07
Kr RA-10/10
c
6/6
9 +- 3
10/10
0.62 +- 0.03
Kr RB-10/10
c
6/6
7.6 +- 0.6
10/10
2.7 +- 1.6
Conditions: pH 8.0, 10 mM MgCl
2
.
a
Number of base pairs in helices III and I.
b
At 30oC.
c
At 37oC.
Cleavage is rapid when helix I is short
In order to elucidate which of the helices was required to be less stable, a
series of substrates were made which would form helices of different lengths
with the symmetric (10/10) ribozymes. The cleavage rates of these substrates by
their cognate ribozymes are shown in Table
2
. With the exception of TAT RA-10/10, the ribozymes cleaved the substrates which formed 5 bp in helix I
(S16-10/5) much more efficiently than the substrates which formed 5 bp in helix
III (S16-5/10), where in each case the other helix consisted of 10 bp. In a control
experiment, TAT S16-10/5 and TAT S16-5/10 were cleaved with similar efficiency by the short armed
ribozyme TAT RA-6/6 (
k
obs
= 3.1 and 3.8 min
-1
respectively at pH 7.13), demonstrating that there was no intrinsic difference
in the cleavability of the two TAT substrates.
This phenomenon also applies when the length of the helix is limited by the
length of the hybridizing arms of the ribozyme rather than the length of the
substrate. Two asymmetric ribozymes, TAT RA-10/5 and TAT RA-5/10, were synthesized and their ability to cleave the symmetric
21mer substrate TAT S21-10/10 was examined. The results are given in Table
3
. Clearly, the same conclusions can be drawn: a symmetric 21mer substrate in a
complex with a helix I of 5 bp is cleaved much more efficiently than when in a
complex with a helix I of 10 bp. The generality of the effect was confirmed by
the observation that the ribozyme Kr RA-6/10 cleaved the substrate Kr S21-10/10 with a rate constant of 6.7 min
-1
at pH 7.13 (Table
3
), nearly 70-fold greater than the symmetric ribozyme Kr RA-10/10 was able to cleave the same substrate (
k
obs
= 0.1 min
-1
).
.
Cleavage rate constants for symmetric 21 nt substrates by their cognate
asymmetric ribozymes
Ribozyme
Substrate
III/I
k
obs
(min
-1
)
TAT RA-10/5
TAT S21-10/10
5/10
0.09 +- 0.01
TAT RA-5/10
TAT S21-10/10
10/5
10 +- 1
Kr RA-6/10
Kr S21-10/10
10/6
6.7 +- 0.1
Conditions: 10 mM MgCl
2
, 37oC, pH 7.13.
The optimum lengths of helices I and III
To determine the optimum length of helix I, two series of substrates (TAT and
Kr) were prepared with varying numbers of nucleotides on the 3'-side of the cleaved nucleotide H
17
and a constant 10 nt on the 5'-side. Rate constants for cleavage of the substrates by TAT RB-10/10, Kr RA-10/10 and Kr RB-10/10 were measured at pH 7.13 and 8.0, and also
at pH 6.42 for Kr RA-10/10 only. The results of experiments at pH 7.13 are shown in Figure
3
. These data show that with a helix III of 10 bp, the optimum length of helix I
is 5 or 6 bp. Similar trends were obtained for all the ribozymes at all pHs
studied (data not shown).
Control experiments
DISCUSSION
Throughout this series of experiments we have used high concentrations of
ribozyme in excess over the substrate so that essentially all the substrate is
in complex with ribozyme. Therefore, association of the ribozyme and substrate,
dissociation of the cleavage products from the ribozyme and rate of binding of
Mg
2+
play no role in determining the observed cleavage rate constants. These results
demonstrate for the first time that the cleavage step
k
2
(Scheme 1) is strongly dependent on the length of helix I. The effect appears
to be quite general. There are clearly some fundamental processes occurring in
the cleavage of substrates by the hammerhead ribozyme which have hitherto been
overlooked.
A rapid pre-equilibrium?
Our observations are consistent with the scheme for the hammerhead cleavage
reaction proposed earlier (
11
) (Scheme 2), in which a conformational change in the initial ribozyme-substrate complex is required to allow cleavage to occur. They are
consistent with a mechanism which involves a rapid equilibrium between an
active and an inactive conformation. The position of the equilibrium is
apparently dependent on the length and/or stability of helix I, with the amount
of active conformer being maximal around 5 bp. The requirement for a rapid pre-equilibrium comes from the observation that at all lengths of helix I,
there is a similar pH dependence of
k
obs
(Fig.
4
). If the conformational change were slow and rate limiting (for example with a
helix I of 10 bp), it may be expected that the reaction would be independent of
pH. An alternate scheme in which the active and inactive conformations exchange
via the free ribozyme and substrate can be eliminated because exchange via that
pathway should be much too slow, especially for complexes with extensive base
pairing.
There is ample precedent for the existence of an inactive conformation of the
ribozyme-substrate complex. The crystal structures (
14
,
15
) of the hammerhead ribozyme complexed with a substrate analogue show an
inactive conformation in which the latent nucleophile is not correctly
positioned for in-line attack at the phosphate. The structures also show that certain
phosphate and sugar groups in the 5' arm of the ribozyme lie close to phosphate groups in the distal part of
helix II, near loop II. Specifically, in the crystal structure by Pley
et al
. (
14
), the interactions involve the fourth and fifth nucleotides in the 5' arm of the ribozyme (nt 2.4 and 2.5) and 2 nt in helix II (nt 11.3 and
11.4); in the structure by Scott
et al
. (
15
), the interactions are between nt 2.5 and nt 11.4. Since we have observed that
cleavage rates decline significantly when helix I is >6 bp in length (i.e. when
helix I around nt 2.5 is stable and the interaction with helix II is
strongest), it is tempting to speculate that these interhelix interactions are
characteristic of the inactive conformation.
Minizymes (ribozymes lacking helix II) are poor cleavers of short substrates
compared with ribozymes (
16
), but they are not further inhibited by long helices I. For example, minizymes
with 10 nt in each hybridizing arm invariably cleave symmetric 21mer substrates
more efficiently than symmetric 13mer substrates. In addition, no improvement
in cleavage rate constants is observed for these minizymes cleaving 10/10
substrates compared with 10/5 substrates (where there are 10 and 5 bp
respectively in helices I of the complexes; data not shown). This observation
supports the hypothesis that some form of interaction between helices I and II
is responsible for stabilizing the inactive conformation.
In retrospect, inhibition of the ribozyme by the long helix I, rather than
substrate dimerization (
17
,
18
), accounts for our earlier observations (
19
) that minizymes with 10/10 hybridizing arms had activity similar to full-sized all-RNA ribozymes with 10/10 hybridizing arms when cleaving 21mer
substrates.
Other work
Other studies are consistent with the notion that helix I needs to be relatively
short or weak and helix III can be much longer and/or more stable for rapid
cleavage of a substrate by hammerhead ribozymes. In the context of a very long
helix III, helix I was able to be reduced to as few as 3 bp without loss of
activity (
20
), however, in that case, despite the fact that the ribozyme and substrate were
pre-annealed, the observed rate constants were only ~0.01 min
-1
at pH 8, 20 mM MgCl
2
, which is ~1000-fold slower than would be expected for efficient cleavage of short
substrates. Mismatches in either of helix I or III close to the conserved
domain were shown to cause a dramatic decrease in activity, but mismatches more
distal in helix I had only marginal effects (
21
). In the context of relatively short substrate binding helices (
22
), mismatches close to the core in helix I were tolerated much more than in
helix III, where mismatches in any of the four innermost base pairs caused a
significant decrease in
k
2
. In the context of a hammerhead with very long helices I and II, it was
observed that elongation of helix III from 3 to 9 bp caused a >100-fold increase in observed rate constant (
23
). In another study (
24
), cleavage of relatively long transcripts with excess ribozyme showed very
significant decreases in activity when helix III was reduced from eight to four
effective base pairs, albeit under conditions where the cleavage event did not
appear to be rate determining.
Denman
et al
. (
25
) have studied a number of ribozymes targeted to the [beta]-amyloid peptide precursor, where their stated aim was to investigate
the effect of destabilizing helix I. They concluded that the most active
ribozyme was one with helices I and III of 7 and 8 bp respectively, but this
study was complicated by multiple substrate conformations and the observed
cleavage rates were very low. Amiri and Hagerman (
23
) claim that a pre-annealed ribozyme with very long helices I and II and a helix III of 9 bp
is able to cleave its annealed substrate with a rate constant >5 min
-1
at pH 8, 25oC, upon addition of 10 mM MgCl
2
. However the accompanying figure (figure 2 in Amiri and Hagerman;
23
) appears to show cleavage under those conditions with a rate constant of ~0.5 min
-1
. These examples, if sustained, might argue against the generality of the effect
we have described. However, an effect similar to our observations has been
noted in a further unrelated system (P. Keese and M. Stapper, personal
communication).
The reactivity of TAT RA-10/10
The ribozyme TAT RA-10/10 displays reduced activity; it cleaves its 13mer substrate TAT S13-6/6 >40-fold slower than does the shorter analogue TAT RA-6/6 at pH 8.0 (
11
). The source of the anomalous behaviour must lie in the terminal nucleotides of
the 5' arm of the ribozyme, since TAT RA-10/5 is similarly repressed, cleaving TAT S13-6/6 with a rate constant of only 0.09 +- 0.01 min
-1
at pH 7.13, 50-fold slower than the cleavage of that substrate by TAT RA-5/10 and TAT RA-6/6 under the same conditions. In this context we note that
the terminal nucleotides in the 5' arm of the ribozyme have the base sequence 5'-GUCC, which is complementary to four bases in helix II, and
in addition there is a high degree of complementarity between the sequences A
2.5
-C
3
and G
8
-U
L2.1
. Thus, the anomalous cleavage rates of TAT RA-10/10 appear to be caused by spurious base pairing interactions.
Conclusion
Hammerhead ribozyme cleavage proceeds most rapidly when the length of helix I is
~5 bp. This observation is consistent with a pre-equilibrium between active and inactive conformations of the ribozyme-substrate complex. There are very clear implications for
ribozyme design in this study. For optimum
in vitro
cleavage rates, the 5' hybridizing arm should contain of the order of 5 +- 1 nt. The number of nucleotides in the 3' hybridizing arm is less critical, however, there should
be greater than ~5 nt. If multiple turnovers are required, then the rate of dissociation of
the 5' product of cleavage from the 3' arm of the ribozyme (i.e. the rate of dissociation of helix III
after cleavage) must be considered.
REFERENCES
1 Buzayan,J.M., Gerlach,W.L. and Bruening,G. (1986) Nature, 323, 349-353.
