In vitro
selection of hammerhead ribozymes containing a bulged nucleotide in stem II
In vitro selection of hammerhead ribozymes containing a bulged nucleotide in stem II
James B.
Thomson
,
Snorri
Th. Sigurdsson
,
Astrid
Zeuch
and
Fritz
Eckstein*
Max-Planck-Institut für Experimentelle Medizin, Hermann-Rein-Strasse 3, D-37075
Göttingen
,
Germany
Received August 27, 1996
;
Revised and Accepted October 7, 1996
ABSTRACT
Hammerhead ribozymes were transcribed from a dsDNA template containing four
random nucleotides between stems II and III, which replace the naturally
occurring GAA nucleotides.
In vitro
selection was used to select hammerhead ribozymes capable of
in cis
cleavage using denaturing polyacrylamide gels for the isolation of cleaving
sequences. Self-cleaving ribozymes were cloned after the first and second rounds of
selection, sequenced and characterised. Only sequences containing 5
'
-HGAA-3
'
, where H is A, C or U, between stems II and III were active; G was clearly not
tolerated at this position. Thus, only three sequences out of the starting pool
of 256 (4
4
) were active. The Michaelis-Menten parameters were determined for the
in trans
cleaving versions of these ribozymes and indicate that selected ribozymes are
less efficient than the native sequence. We propose that the selected ribozymes
accommodate the extra nucleotide as a bulge in stem II.
INTRODUCTION
The hammerhead ribozyme is an RNA motif which is capable of sustaining either
in trans
or
in cis
cleavage of a phosphodiester band (
1
-
3
) [for recent reviews see (
4
,
5
)]. The two-dimensional representation of the hammerhead ribozyme is depicted in Figure
1
. Cleavage specificity is controlled by the hybridising arms of the ribozyme,
which anneal with the substrate in a complementary fashion and direct cleavage
of the scissile phosphodiester bond. This activity is specifically directed to
occur after the third nucleotide of the cleavage triplet, at position H
17
, for which cleavage is limited to sequences of the form 5'-NUH-3' (where N is any nucleotide and H = A, U or C) (
6
-
8
). The ribozyme is composed of three [alpha]-helical regions, helices I, II and III, which flank the 11 single-stranded, conserved nucleotides of the catalytic core region.
This sequence of conserved nucleotides has a particular tertiary structure,
which has been elucidated by X-ray crystallography (
9
,
10
) and is further supported by a number of other biophysical techniques (
11
-
13
). As derived from the crystal structures, the catalytic core is built up of a
base mispairing and a uridine turn domain (Fig.
1
). Clearly, many of these tertiary interactions established from the crystal
structures form the ground-state of the hammerhead ribozyme, with additional information on hydrogen
bonding having been derived from chemical modification experiments (
14
-
17
). However, despite this wealth of data there is no clear indication of how
catalysis is achieved by the hammerhead ribozyme.
The technique of
in vitro
selection (
18
-
21
) is of great interest in application to the hammerhead ribozyme since it would
offer an opportunity to expand the number of nucleotide sequences after which
cleavage can occur and would also provide information regarding the tertiary
interactions within the conserved central core. Two
in vitro
selection strategies have already been applied to the hammerhead ribozyme and
have confirmed the importance of forming a stable stem II structure for
stabilisation of the adjacent A
9
[middot]G
12
, G
8
[middot]A
13
double mismatches of the central core region (
22
,
23
). This was consistent with previous work by Tuschl and Eckstein (
24
), which highlighted the importance for a stem II of at least two G[middot]C base pairs in order to attain optimal cleavage efficiency.
It was of interest to see if an extra nucleotide could be accommodated between
stems II and III, as it can be between stems I and II (
25
), and whether there were rigid sequence preferences for this region. Thus, this
could yield information regarding the conservation of nucleotides in this
central core region, which form the mispairing domain observed in the crystal
structures. A pool of ribozymes, containing four random nucleotides to replace
G
12
, A
13
and A
14
, was prepared by transcription from a DNA template. The active ribozymes were
subsequently selected, as outlined in Figure
2
, using a similar strategy as described by Nakamaye and Eckstein (
23
).
MATERIALS AND METHODS
Nucleoside triphosphates and 2'-deoxynucleoside triphosphates were purchased from Boehringer
Mannheim. [[alpha]
32
P]ATP (3000 Ci/mmol), [[gamma]
32
P]ATP (5000 Ci/mmol), [[alpha]
35
S]dATP (3000 Ci/mmol), Sequenase quick-denaturing plasmid sequencing kit,
Taq
DNA polymerase and 10* reaction buffer, T4 polynucleotide kinase and 10* reaction buffer, MMLV-reverse transcriptase and 5* first strand buffer and Sequenase DNA polymerase
were purchased from Amersham. X-Ray film (X-OMAT XAR-5) was purchased from Kodak. Radioanalytical scanning was
performed on a Fuji BAS2000 Bio-imaging analyzer.
Eco
RI and
Bam
HI restriction endonucleases were purchased from NEB. Plasmid DNA purification
columns and PCR QiaQuick spin columns were purchased from Diagen (Düsseldorf, Germany).
Synthesis of oligonucleotides
The following DNA templates and primers were synthesised on an Applied
Biosystems 380A DNA synthesiser and were purified as previously described (
26
): template-A (71mer) 5'-d(GCCACACTGA CTATAGTTCC CTATAGTXXX XGCTTGCGCT CATCAGAGTG TGGC
TATAGT
GAGTCGTTAT
A)-3', which contains the T7 promoter region (underlined); PCR-A (37mer), 5'-d(GCGCTA
GAAT TC
T
ATAACGA
CTCACTATA
G CCACACT)-3', which restores the
Eco
RI restriction site (bold) and the T7 promoter region (underlined); RT-A (32mer), 5'-d(GGCGAT
GGAT CC
GCCACACT GACTATAGTT CC)-3', which restores the 3' region, helix I, which is removed during the cleavage
reaction and the
Bam
HI restriction site (bold).
Sequenase DNA polymerisation
Double-stranded DNA (pool 0) was prepared by mixing template A (40 [mu]M, 10 [mu]l) with PCR-A (40 [mu]M, 10 [mu]l). The strands were annealed by heating the
solution to 70oC for 5 min and cooling to room temperature. The following reagents were
then added to give the final concentrations indicated: Tris-HCl (pH 8.0, 65 mM), NaCl (50 mM), MgCl
2
(5 mM), dithiothreitol (5 mM), 2'-deoxynucleoside 5'-triphosphates (375 [mu]M each). Sequenase DNA polymerase (1.3 U) was then
added and the solution (50 [mu]l) incubated at 37oC for 1 h. The dsDNA was recovered by ethanol precipitation using NH
4
OAc (
27
).
T7 RNA transcription and RNA purification
dsDNA, produced from either Sequenase DNA polymerisation (pool 0) or from PCR
(pool 1), was used as the template for the T7 RNA polymerase catalysed
transcription. Each transcription was carried out in a volume of 250 [mu]l with the following final concentration of reagents: DNA (1 [mu]M), Tris-HCl (pH 8, 40 mM), MgCl
2
(20 mM), spermidine (1 mM), Triton X-100 (0.01%), dithiothreitol (10 mM), nucleoside 5'-triphosphates (4 mM each), T7 RNA polymerase (50 U/[mu]l) and [[alpha]-
32
P]ATP (in trace amount). Transcription mixtures were incubated at 37oC for 16 h (pool 0) or 8 h (pool 1), centrifuged and the supernatant
removed from the pyrophosphate precipitate. The RNA transcribed was recovered
from the supernatant by ethanol precipitation with NaOAc (
27
). The resultant RNA pellet was dried to remove any residual ethanol, dissolved
in water (25 [mu]l) and loaded onto a 12% denaturing polyacrylamide gel (0.4 mm thick, 50 W,
1.5 h). The desired length of RNA (46mer) was excised from the gel, using
xylene cyanol as a marker, and the RNA extracted by suspending the crushed gel
slice in NaOAc (1 M, pH 5.6, 200 [mu]l). After incubating for 2 h on ice and, with brief vortexing every 30 min,
the supernatant was removed. The extraction was repeated a second time, the
supernatants combined and the RNA precipitated by addition of 3 vol ethanol.
The pellet was then washed twice with ethanol/water (7:3, 2 * 150 [mu]l) and the pellet thoroughly dried before being dissolved in water (25 [mu]l).
RT-PCR
Reverse transcription was carried out in a 30 [mu]l reaction volume with the following final concentrations of reagents: RT-A (833 nM), Tris-HCl (pH 8.0, 20 mM), KCl (100 mM), MgCl
2
(3 mM), gelatine (200 [mu]g/ml), dNTPs (666 nM each), RNase inhibitor (1 U/[mu]l) and MMLV-RT (0.67 U/[mu]l). The protocol observed was essentially that as described by
Nakamaye and Eckstein (
23
) with the exception that after incubation for 1 h at 42oC the RNA in the mixture was destroyed by addition of NaOH (2 M, 20 [mu]l) and water (150 [mu]l). This was then incubated for a further 1 h at 37oC and the cDNA precipitated from this mixture by addition of
NH
4
OAc (10 M, 50 [mu]l) and ethanol (750 [mu]l). The cDNA pellet was then washed twice with ethanol/water (7:3) dried
and dissolved in water (110 [mu]l). Using the following PCR cycle, 94oC (30 s), 55oC (10 s), 72oC (2 min), as described by Long and Uhlenbeck (
22
), the number of PCR cycles required to amplify the cDNA was established by
removing 10 [mu]l aliquots every fifth cycle and determining the extent of product formation
by agarose (2.5%) gel electrophoresis. The remainder of the cDNA was then
amplified in ten 100 [mu]l reactions. The DNA produced by this procedure was isolated by two ethanol
precipitations; first of all using NH
4
OAc, to remove the triphosphates, and then from NaOAc. This DNA was suitable for
use in a further T7 transcription reaction or for cloning and sequencing. Cloning and sequencing was carried out as previously described by Nakamaye and Eckstein (
23
).
Kinetics of intermolecular ribozyme cleavage
The Michaelis-Menten parameters for
in trans
ribozyme cleavage were carried out under single turnover (
28
) and multiple turnover (
24
) conditions as previously described.
RESULTS
The selection experiments described here were carried out using a previously
described procedure (
23
; Fig.
2
), and the topic of
in vitro
selection has been extremely well reviewed (
18
-
21
). In brief, the protocol consisted of transcription of a random pool of RNA
from a random DNA template and isolation of the shorter cleavage product
(46mer) on a 12% denaturing polyacrylamide gel. The RNA was reverse-transcribed using primer RT-A, which restores the 3'-nucleotides lost through the cleavage and includes the
Bam
HI restriction site. The resultant cDNA was amplified by PCR using, in addition
to RT-A, the primer PCR-A, which restores the T7 RNA polymerase promoter and the
Eco
RI restriction site. The dsDNA was subsequently utilised for cloning or as a
template in subsequent selection cycles.
During the selection the enrichment of each of the pools with
in cis
cleaving hammerhead ribozymes was established by transcribing each pool and
measuring the extent of cleavage after 3 h incubation. RNA transcribed from
pools 0, 1 and 2 DNA cleaved to approximately 4, 24 and 65% respectively and
DNA pools 1 and 2 were cloned and sequenced to yield the following sequence
data.
Pool 1 DNA
Forty-eight colonies were picked and from the subsequent run-off transcripts 10 of the clones appeared to give a cleavage
product. These ten and a further eight clones, which demonstrated no cleavage,
were sequenced and fell into the following four categories.
(i) Self cleaving ribozymes
. Seven clones fell into this category and only three sequences were represented, where the random region was AGAA (3
times), CGAA (2 times) and UGAA (2 times). These three motifs cleaved to 82, 89 and 91% (Fig.
3
) during a 30 min transcription reaction, which is similar to the 95% of the native
hammerhead sequence.
Pool 2 DNA
From the 15 clones picked and transcribed
in vitro
, nine supported
in cis
cleavage. Of these nine the sequences CGAA (
7
) and AGAA (
2
) were the only representatives.
a
Multiple turnover conditions using 5-25 nM ribozyme and 50-1500 nM substrate concentrations.
b
Taken from ref. 24.
In trans
cleavage of selected ribozymes
Ribozyme sequences were synthesised for
in trans
cleavage and targeted against a 19mer substrate (Fig.
4
, left; Table
1
), which contains the sequences in helices I and III used in the selection
experiment, and also against a 12mer substrate (Fig.
4
, right; Table
2
), for which the native ribozyme has been well characterised in this laboratory (
24
). Using the 19mer substrate (Fig.
4
, left; Table
1
) no cleavage was observed under multiple turnover conditions and Michaelis-Menten parameters were established using single turnover conditions. The
three selected ribozymes had catalytic efficiencies only 3-fold lower than the native sequence.
Multiple turnover cleavage could be observed when the ribozymes were targeted
against a 12mer substrate (Fig.
4
, right). The results (Table
2
) indicate that they cleave the substrate with between 10- and 37-fold lower catalytic efficiency, compared to the native GAA
sequence. In order to investigate the possibility that the additional
nucleotide is incorporated into stem II, forcing C
11.1
into bulge, Rz A-bulge was synthesised (Fig.
5
), which contains an adenosine to replace C
11.1
. This ribozyme was targeted against the 12mer substrate (Table
2
) and demonstrated a similar catalytic efficiency to the selected ribozymes.
DISCUSSION
In vitro
selection has already been utilised in hammerhead ribozymes in order to examine
the importance of the stem-loop II region in the cleavage reaction (
22
,
23
). This confirmed the importance of a G
10.1
[middot]C
11.1
base pair for the closing of stem II and the base mismatch region, formed by
the A
9
[middot]G
12
, G
8
[middot]A
13
and U
7
[middot]A
14
mispairs, which connects stems II and III in a near continual [alpha]-helix (Fig.
1
). The activity of hammerhead ribozymes containing shortened stem II sequences (
24
) have demonstrated that G
10.1
[middot]C
11.1
and C
10.2
[middot]G
11.2
base pairs are the minimum requirement for optimal catalytic efficiency.
Ribozymes with a four nucleotide linker between G
10.1
and C
11.1
are also functional but cleave with less than one tenth the activity of
ribozymes containing a stem II (
24
,
29
).
In vitro
selection also identified a ribozyme similar to the hammerhead as one of the
self-cleaving motifs, amongst a variety of others, isolated from a pool of tRNA
molecules containing a 100 nucleotide random insert in one of the loops (
30
).
In this paper a part of the central core region has been randomised to extend
this selection technique to investigate the nucleotide requirements of the
single-stranded region between nucleotides 11.1 and 15.1. The extra nucleotide
was added, since it has been reported that an additional nucleotide can be
incorporated into the single-strand region between stems I and II (
31
) without significant loss of ribozyme activity.
Selected sequences
Enrichment of the random pools with cleaving sequences was achieved with each
round of selection and the DNA from pools 1 and 2 were cloned and sequenced to
show how the selection progressed as the incubation time was lowered from 16 to
8 h. Pool 1 had three sequences with self-cleaving activity;
viz
. AGAA, represented three times and CGAA and UGAA, each represented twice. The
DNA from pool 2 had only two cleaving sequences;
viz
. CGAA seven times and AGAA twice, UGAA was not present. It is not clear why
this selection favours the CGAA sequence so strongly, since from the run-off transcripts of the cleaving clones (Fig.
3
) all three selected sequences cleave to ~85-90%. It is very probable that UGAA is in fact represented in pool 2,
since only 15 colonies were picked for analysis and it may have been detected
if a larger number of colonies had been screened.
All the selected ribozymes were synthesised chemically and their
in trans
cleaving parameters established (Tables
1
and
2
). Using the 19mer substrate (Fig.
4
a, Table
1
), which contains the hybridising arms used in the selection cycle, the selected
ribozymes displayed catalytic efficiencies similar to that of the native. In
contrast, when the 12mer substrate was used (Fig.
4
b, Table
2
), the selected ribozymes had catalytic efficiencies which were up to 37-fold lower than the native.
The selected ribozymes, whether targeted against the 19- or 12mer substrate, all have similar catalytic efficiencies. With the
19mer substrate, efficiencies were only slightly lower than the native and this
was mainly due to a slight decrease in
k
cat'
. That
k
cat'
for the native is so much lower than
k
cat
obtained with the short substrate under multiple turnover could be due to the
existence of a preequilibrium for the long ribozyme-substrate complex (
32
-
34
). With the 12mer substrate, the catalytic efficiencies of the selected
ribozymes were lower than the native by over a factor of ten. The catalytic
efficiency of this ribozyme has been shown to be very dependent upon the
structure of stem-loop II, with variations in the loop sequence lowering the catalytic
efficiency by up to 3-fold (
24
). Presumably the helical destabilisation caused by the additional nucleotide
lowers the catalytic efficiency of the selected ribozymes.
The much lower catalytic efficiencies could also be a consequence of the
selection having been carried out using different hybridising arms, although it
has never been demonstrated that the sequence of stems I and III affect the
efficiency of ribozyme cleavage. The trend in catalytic efficiency is reflected
mainly by a lowering of the
k
cat'
or
k
cat
values, whilst
K
m
' or
K
m
remains relatively constant. Thus, the selected ribozymes appear to bind the
substrate equally as well as the native hammerhead. Only with the 12mer
substrate did CGAA, AGAA and Rz A-bulge demonstrate an ~2- to 3-fold increase in
K
m
, which is indicative of the presence of alternative conformations inhibiting
substrate binding (
35
).
Structure of the selected ribozymes
The selected ribozymes indicate that 5'-HGAA-3' is the only sequence tolerated in an active
hammerhead ribozyme. The additional nucleotide was only observed at the 5'-end of the sequence, suggesting that it is incorporated into stem
II and cannot be tolerated in the central core or stem III. This is reasonable
since nucleotides 15.1 and 15.2 base pair with conserved nucleotides 16.1 and
16.2, which are the first two nucleotides of the cleavage triplet. As a result,
there is no real possibility for accommodation of an additional base at the
bottom of stem III. It is sensible therefore that the additional nucleotide is
accommodated into stem II, where there is greater sequence tolerance for
forming the closing base pair, thus forcing C
11.1
into a bulged position (Fig.
5
). Although in most naturally occurring hammerhead ribozymes a G
10.1
[middot]C
11.1
base pair is found, a C
10.1
[middot]A
11.1
mismatch is present in the small barley yellow dwarf virus (sBYDV) (
36
) and an additional uridine is found between nucleotides A
9
and G
10.1
in the (+) strand of the lucerne transient streak virus (
25
). However,
in vitro
selection on the sBYDV sequence demonstrated that although a mismatch in stem
II was tolerated, it was not optimal for the cleavage reaction and
randomisation of positions 7, 10.1 and 11.1 yielded the more active sequence
containing the standard G
10.1
[middot]C
11.1
base pair (
23
).
Rz A-bulge (Table
2
, Fig.
5
) was synthesised to test the idea that the selected ribozymes accommodated the
additional nucleotide by placing C
11.1
into a bulged position in stem II and creating a base pair between G
10.1
and the first random position. If this was the case, then a ribozyme containing
an A-bulge would be as active as the CGAA selected ribozyme, which contains a C-bulge. The Michaelis-Menten parameters of this sequence (Table
2
), targeted against the 12mer substrate, are only slightly different from that
of the CGAA sequence implying that an A-bulge is tolerated fairly well in stem II. Bulges are very common in RNA
structures and helical destabilisation and/or disruption is dependent on the
flanking base pairs. However, generally they create a kinking of the helix by
around 10o (
37
,
38
) and are not found to be particularly destabilising. Therefore, it is not
unreasonable that the additional nucleotide could be accommodated as a bulge in
stem II, and this would presumably be less destabilising to the overall
structure of the hammerhead ribozyme than if it needed to be incorporated
between stem II and the G/A mismatches. The double G/A mismatch extends the [alpha]-helix of stem II into the central core and it is an essential
structural feature of the hammerhead ribozyme. Double G/A mismatches are common
in RNA structures and, depending on the closing base pairs, they do not
significantly destabilise helical DNA (
39
,
40
) or RNA (
41
-
43
). The drop in catalytic efficiency from CGAA to UGAA and down to AGAA, in
particular with the 19mer substrate, is consistent with the formation of a
progressively less stable G
10.1
[middot]H base pair. The results imply that the non-Watson-Crick G
10.1
[middot]A and a G
10.1
[middot]U base pairs, formed by ribozymes AGAA and UGAA respectively (Fig.
5
), do not impair formation of the double G[middot]A mismatches.
A bulge can be accommodated into stem II where it is part of an extended [alpha]-helix with a distal double G[middot]A mismatch. Thus, the lower catalytic efficiency of the
selected ribozymes (Table
2
) can be explained by the destabilisation of stem II through the introduction of a bulged nucleotide. This also offers an explanation for the absence of GGAA as a cleaving sequence since a G cannot form a stable
mispair with G
10.1
.
Conclusions
Hammerhead ribozymes were selected from a pool containing four random
nucleotides incorporated between helices II and III. These experiments
demonstrate that the GAA sequence between stems II and III is a very strongly
conserved motif. The additional nucleotide is most likely incorporated into
stem II causing nucleotide C
11.1
to bulge out. All the selected sequences cleave
in cis
and
in trans
cleavage efficiencies are only slightly lower than the native.
ACKNOWLEDGEMENTS
We are extremely grateful to P. A. Heaton, Y. Berlin and K. Birikh for the
critical reading of the manuscript and U. Kutzke for expert technical support.
We thank one of the referees for pointing out the possible existence of a
preequilibrium in one of the complexes. This work was supported by the Deutsche
Forschungsgemeinschaft and an EMBO fellowship to S.Th.S.
REFERENCES
1 Haseloff,J. and Gerlach,W.L. (1988) Nature, 334, 585-591.MEDLINE Abstract
5 Sigurdsson,S.T., Thomson,J.B. and Eckstein,F. (1996) In Simons,R.W. and Grunberg-Manago (ed.), RNA Structure and Function. Cold Spring Harbour Laboratory Press, Cold Spring Harbor, NY, in press.
6 Ruffner,D.E., Stormo,G.D. and Uhlenbeck,O.C. (1990) Biochemistry, 29, 10695-10702.MEDLINE Abstract
7 Shimayama,T., Nishikawa,S. and Taira,K. (1995) Biochemistry, 34, 3649-3654.MEDLINE Abstract
27 Sambrook,J., Fritsch,E.F. and Maniatis,T.C. (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbour Laboratory Press, Cold Spring Harbor, NY.
28 Heidenreich,O. and Eckstein,F. (1992) J. Biol. Chem., 267, 1904-1909.MEDLINE Abstract
