Content uploaded by Jean-Pierre Perreault
Author content
All content in this area was uploaded by Jean-Pierre Perreault
Content may be subject to copyright.
Development and comparison of procedures for the
selection of delta ribozyme cleavage sites within the
hepatitis B virus
Lucien Junior Bergeron and Jean-Pierre Perreault*
RNA Group/Groupe ARN, De
Âpartement de Biochimie, Faculte
Âde Me
Âdecine, Universite
Âde Sherbrooke,
Sherbrooke, Que
Âbec J1H 5N4, Canada
Received July 10, 2002; Revised and Accepted September 9, 2002
ABSTRACT
Delta ribozyme possesses several unique features
related to the fact that it is the only catalytic RNA
known to be naturally active in human cells. This
makes it attractive as a therapeutic tool for the in-
activation of clinically relevant RNAs. However,
several hurdles must be overcome prior to the
development of useful gene-inactivation systems
based on delta ribozyme. We have developed three
procedures for the selection of potential delta
ribozyme target sites within the hepatitis B virus
(HBV) pregenome: (i) the use of bioinformatic tools
coupled to biochemical assays; (ii) RNase H
hydrolysis with a pool of oligonucleotides; and (iii)
cleavage assays with a pool of ribozymes. The
results obtained with delta ribozyme show that
these procedures are governed by several rules,
some of which are different from those both for
other catalytic RNAs and antisense oligonucle-
otides. Together, these procedures identi®ed 12
sites in the HBV pregenome that can be cleaved by
delta ribozymes, although with different ef®ciencies.
Clearly, both target site accessibility and the ability
to form an active ribozyme±substrate complex con-
stitute interdependent factors that can best be
addressed using a combinatorial library of either
oligonucleotides or ribozymes.
INTRODUCTION
The goal of gene therapy is to modulate the expression of a
speci®c gene. The ability of ribozymes (i.e. RNA enzymes) to
speci®cally recognize an RNA substrate, and subsequently
catalyze its cleavage, makes them attractive as therapeutic
tools for the inactivation of both viral RNAs and the mRNAs
associated with various diseases (1±3). Several successful
applicable ribozyme models have been tested both in vitro and
in a cellular environment (1±3).
Delta ribozyme possesses several unique features that are
all related to the fact that it is the only naturally occurring
catalytic RNA that has been discovered in humans (4,5). In
contrast to other catalytic RNAs, several hurdles remain to be
surpassed before we generate a useful gene-inactivation
system based on delta ribozyme. The initial step in the
development of a ribozyme capable of cleaving a natural RNA
molecule is the selection of the cleavage site with the greatest
potential for targeting. The speci®city of recognition is
derived from Watson±Crick base pairs formed between the
substrate and the ribozyme. For example, we have engineered
a57ntdelta ribozyme derived from the antigenomic strand of
the hepatitis delta virus (HDV) (Fig. 1A) (6,7). According to
its double-pseudoknot secondary structure, which is well
supported by experimental data, the substrate speci®city
depends on the formation of the P1 stem (i.e. the sequence
binding domain of the ribozyme), which includes one GU
wobble base pair followed by six non-speci®c Watson±Crick
base pairs (Fig. 1A) (4). Secondary and tertiary structures of a
target that are characterized by signi®cant intramolecular base
pairing are important features, which can in¯uence the
cleavage activity of ribozymes (8±10). Thus, even though
the appropriate target site is present, effective cleavage at that
site depends on its accessibility to the ribozyme. Target
sequences located in single-stranded regions of an mRNA
have a higher potential because they should be more
accessible to ribozyme binding than those in double-stranded
regions. Within the double-stranded regions the ribozyme
might compete unfavorably with intramolecular base pairing
when trying to bind its substrate (8±10). The development of
several procedures identifying potential target sites in RNA
molecules has been reported for various nucleic acid drugs,
including other ribozymes (11±22). Based on these reports, we
present three procedures developed in order to identify the
sites with the greatest potential for cleavage by delta ribozyme
in a model RNA. These procedures include the use of:
(i) bioinformatic tools coupled to biochemical assays
(13,15,16,22); (ii) ribonuclease H (RNase H) mapping in the
presence of a library of DNA oligonucleotides (11,20); and
(iii) cleavage assays using a library of ribozymes (12,14).
The pregenome RNA from the hepatitis B virus (HBV) was
chosen as model target because it offers several advantages.
We selected an RNA virus, instead of an mRNA, as our target
because the use of an mRNA would have limited our search
for potential cleavage sites to those near the 5¢-end in order to
ensure that cleavage resulted in the production of an inactive
protein. HBV has an extremely compact organization as all
*To whom correspondence should be addressed. Tel: +1 819 564 5310; Fax: +1 819 564 5340; Email: jperre01@courrier.usherb.ca
4682±4691 Nucleic Acids Research, 2002, Vol. 30 No. 21 ã2002 Oxford University Press
nucleotides have coding function (Fig. 1B) (23). The 3.2 kb
DNA genome replicates via reverse transcription involving an
RNA pregenome of ~3.4 kb whose secondary structure is
unknown. Moreover, HBV is a model target that has clinical
relevance as 350 million people worldwide are chronic
carriers (24,25).
MATERIALS AND METHODS
Bioinformatic analysis of HBV RNA
The sequence of the HBV variant used corresponded to the
insert of the plasmid pCHT-9/3091, which was kindly
provided by Dr M. Nassal (26). The most stable secondary
structures, in terms of energy, of this HBV variant were
predicted using the program RNA Structure 3.5 (http://rna.
chem.rochester.edu) (27). The resulting ®le was then analyzed
for the probability of binding to complementary RNA
oligonucleotides 7 nt long (which corresponds to the size of
the P1 stem formed by a substrate and a delta ribozyme) using
the software OligoWalk (13).
Eleven HBV sequences from various genotypes [DDBJ/
EMBL/GenBank accession numbers: A1 (X70185, X80924,
X97848, X97849, X97850, X97851 and Z35717), B1
(D00329 and D00330), C1 (X75665) and C2 (L08805)]
were retrieved from the database, aligned with the HBV
sequence present in plasmid pCHT-9/3091 (26) using the
ClustalW package (28), and minor adjustments introduced
manually (http://penelope.med.usherb.ca/labojp/pdf/hbv.pdf).
This sequence alignment was used to identify potential target
sites conserved in most of the genotypes. Other criteria,
including substrate speci®city for delta ribozyme cleavage,
were also used in order to identify the sites with the greatest
potential (see Results).
HBV and ribozyme DNA constructs
The plasmid pCHT-9/3091 contains a full-length copy of the
HBV pregenome (26). The HBV pregenome insert was
subcloned downstream of the T7 RNA promoter within the
vector pBlueScript SK (+/±) (Stratagene) using the SalI and
SacI restriction sites (Fig. 1B). The resulting plasmid was
named pHBVT7, and its identity was con®rmed by DNA
sequencing.
The construction of all ribozymes was performed as
described previously (10). Brie¯y, pairs of complementary
and overlapping DNA oligonucleotides, which corresponded
to the P1 stem region, were synthesized (Invitrogen), annealed
and ligated to PstI and SphI co-digested pdRzP1.1, yielding a
plasmid harboring a delta ribozyme referred to as pdRz-
HBVX, where X corresponds to the position of its potential
cleavage site within HBV. The sequences of all ribozyme
minigenes were con®rmed by DNA sequencing.
RNA synthesis
Ribozymes and HBV RNA were transcribed in vitro using
SmaI-linearized pdRz-HBVX and pHBVT7, respectively, as
templates. Run-off transcriptions were performed in the
presence of puri®ed T7 RNA polymerase (10 mg),
RNAguard (24 U, Amersham Biosciences), pyrophosphatase
(0.01 U, Roche Diagnostics) and linearized plasmid DNA
(5 mg) in a buffer containing 80 mM HEPES±KOH pH 7.5,
24 mM MgCl
2
, 2 mM spermidine, 40 mM DTT, 5 mM of each
NTP and with or without 50 mCi [a-32P]GTP (New England
Nuclear) in a ®nal volume of 100 mlat37°C for 4 h. Upon
completion, the reaction mixtures were treated with DNase
RQ1 (Amersham Biosciences) at 37°C for 20 min, puri®ed by
phenol:chloroform extraction and the nucleic acid precipitated
with ethanol. The HBV RNA products and ribozymes were
respectively fractionated by denaturing 5 and 7.5% poly-
acrylamide gel electrophoresis (PAGE; 19:1 ratio of acryl-
amide to bisacrylamide) in buffer containing 45 mM Tris±
borate pH 7.5, 7 M urea and 1 mM EDTA. The reaction
products were visualized either by UV shadowing or
autoradiography. The bands corresponding to the correct
sizes for the ribozymes and HBV RNAs were cut out and the
transcripts eluted overnight either at room temperature
(ribozymes) or at 4°C (HBV RNA) in a solution containing
0.5 M ammonium acetate and 0.1% SDS. The transcripts were
desalted on Sephadex G-25 (Amersham Biosciences) spun-
columns, and were then precipitated by the addition of 0.1 vol
Figure 1. Illustration of delta ribozyme and HBV RNA pregenome.
(A) Secondary structure of the engineered trans-acting delta ribozyme
bound to the HBV target. The pseudoknot P1.1 is illustrated by the dotted
lines. The homopurine base pair at the top of the P4 stem is represented by
two large dots, while the wobble base pair is represented by a single large
dot. Adjacent and in the P1 stem only the identity of the nucleotides essen-
tial for cleavage to occur are shown: H indicates A, C or U; D indicates A,
G or U; and N indicates A, C, G, U. The cleavage site is indicated by an
arrow. (B) Organization of the pHBVT7 vector. The four overlapping open
reading frames are shown (C, P, S and X).
Nucleic Acids Research, 2002, Vol. 30 No. 21 4683
3 M sodium acetate pH 5.2 and 2.5 vol ethanol, washed, dried,
resuspended in ultrapure water and the quantity determined by
either absorbance at 260 nm or 32P counting.
Synthesis of a combinatorial library of ribozymes
The library of delta ribozymes with a random P1 stem was
constructed using a PCR-based strategy (see inset in Fig. 4).
Three DNA oligonucleotides were used: (i) the antisense
oligonucleotides (Rz-pool; 5¢-GGGTCCCTTAGCCATC-
CGCGAACGGATGCCCANNNNNNACCGCGAGGAGGT-
GGACCC-3¢, where N = A, C, G or T) served as the primary
templates for the library of ribozymes; (ii) the sense primer
(T7-5¢Rz; 5¢-TTAATACGACTCACTATAGGGTCCACCT-
CCTCGCGGT-3¢) permitted the incorporation of the T7
RNA promoter upstream of the minigene; and (iii) the
antisense primer (3¢Rz; 5¢-GGGTCCCTTAGCCATCCGC-
GAACGG-3¢) ampli®ed the template. The PCR mixtures
contained 20 mM Tris±HCl pH 8.8, 6 mM MgCl
2
,10mM
KCl, 10 mM (NH
4
)
2
SO
4
, 0.1% Triton X-100, 0.2 mM of each
dNTP, 2 mM of each primer and 1 U Vent DNA polymerase
(New England Biolabs) in a ®nal volume of 100 ml. After an
annealing step of 1 min at 94°C, 20 cycles of 1 min at 94°C,
1 min at 50°C and 1 min at 76°C were performed. The PCR
products were puri®ed by phenol:chloroform extraction, the
nucleic acid precipitated with ethanol, resuspended in water,
and then in vitro transcriptions and puri®cations of the
ribozymes were performed as described above.
RNase H hydrolysis
RNase H reactions with 32P randomly labeled HBV RNA were
performed as a biochemical assay coupled with the data
obtained from the bioinformatic tools. HBV RNA (~0.1 mM,
~20 000 c.p.m.) and DNA oligonucleotides (7 nt, 5 mM) were
preincubated for 10 min at 25°C in a ®nal volume of 8 ml
containing 20 mM Tris±HCl pH 7.5, 20 mM KCl, 10 mM
MgCl
2
, 0.1 mM EDTA and 0.1 mM DTT. RNase H (0.5 U;
United States Biochemicals) was then added (2 ml) and the
samples incubated at 37°C for 30 min. The reactions were
quenched by adding 3 ml loading buffer (97% formamide,
0.025% xylene cyanol and 0.025% bromophenol blue) and
the mixtures fractionated through denaturing 5% PAGE
gels (19:1), which were analyzed with a PhosphorImager
(Molecular Dynamics).
RNase H reactions were also performed with randomized
oligonucleotides (8 nt, 5 mM) and non-radioactive HBV RNA
(0.5 mM) under the conditions described above. After the
RNase H reaction, water (90 ml) was added and the mixture
phenol:chloroform extracted, the nucleic acids precipitated
with ethanol, washed, dried and conserved for further analysis
by primer extension.
Ribozyme cleavage assays
Cleavage reactions were carried out under single turnover
conditions ([Rz] > [S]). Internally 32P-labeled HBV RNA
(50 nM) was mixed with ribozyme (1 mM) in a 10 ml mixture
containing 50 mM Tris±HCl pH 7.5 and 10 mM MgCl
2
, and
then incubated at 37°C for 3 h. The reactions were stopped by
the addition of loading buffer, electrophoresed on denaturing
5% PAGE gels and analyzed with a radioanalytic scanner. The
speci®city of cleavage of several ribozymes was also veri®ed
by primer extension.
The cleavage of full-length HBV RNA (50 nM) was also
performed in the presence of the pool of delta ribozymes
(10 mM) in a ®nal volume of 10 ml, under the conditions
described above. After the reaction, the volume was made up
to 50 ml by adding water and the RNA precipitated with
ethanol, washed, dried and conserved for further analysis by
primer extension.
Primer extension analysis
A collection of DNA oligonucleotides complementary to the
HBV RNA was purchased from Invitrogen. The nomenclature
adopted referred to their respective complementary sequences
on the full-length HBV: HBV199±184 (5¢-TCATTAGTT-
CCCCCC-3¢), HBV343±329 (5¢-CTGTTTCTCTTCCAA-3¢),
HBV488±474 (5¢-GGCGAGGGAGTTCTT-3¢), HBV646 ±
632 (5¢-AGGAAAAGATGGTGT-3¢), HBV875±861 (5¢-ATA-
TACCCGCCTTCC-3¢), HBV1058±1044 (5¢-GTTGGGATT-
GAAGTC-3¢), HBV1245±1231 (5¢-GTGGAGACAGCG-
GGG-3¢), HBV1495±1481 (5¢-AAAAACCCCGCCTGT-3¢),
HBV1698±1684 (5¢-GCAGGATGAAGAGGA-3¢), HBV2040±
2026 (5¢-CCCAATACCACATCA-3¢), HBV2499±2485 (5¢-
ACCAAGCCCCAGCCA-3¢), HBV2784±2730 (5¢-AGA-
CGGAGAAGGGGA-3¢), HBV3028±3014 (5¢-CCCCCA-
ACTCCTCCC-3¢). The oligonucleotides (10 pmol) were 5¢-
end labeled in a mixture containing 10 mCi [g-32P]ATP
(3000 mCi/mmol; New England Nuclear), 50 mM Tris±HCl
pH 7.6, 10 mM MgCl
2
, 10 mM 2-mercaptoethanol and 12 U of
T4 polynucleotide kinase (United States Biochemicals) at
37°C for 30 min. The end-labeled oligonucleotides were
puri®ed on denaturing 20% PAGE gels, the relevant bands
excised from the gel and eluted overnight at room tempera-
ture, passed through G-25 spun-column, ethanol precipitated,
washed, dried and resuspended in water (60 ml). 5¢-32P-labeled
primer (6 ml) and 103reverse transcription buffer (0.6 mlof
500 mM Tris±HCl pH 8.3, 800 mM KCl and 100 mM MgCl
2
)
were used to dissolve the pellets resulting from either an
RNase H hydrolysis or a ribozyme cleavage assay performed
with a pool of either oligonucleotides or ribozymes, respect-
ively. The primer annealing step was performed by succes-
sively incubating the mixtures at 65°C for 2 min followed by
2 min on ice. The reactions were initiated by adding 0.8 mM of
each dNTP, 3.3 mM DTT and SuperscriptÔII Reverse
Transcriptase (100 U; Invitrogen) in a ®nal volume of 12 ml.
The samples were incubated at 45°C for 30 min, then ethanol
precipitated, washed and analyzed through 5% sequencing
PAGE gels. DNA sequencing reactions using the same primer
were migrated on the same gels in order to allow for
identi®cation of both the primer extension stops and the
cleavage sites of the ribozymes. The results were visualized
with a PhosphorImager.
RESULTS
Bioinformatic tools coupled to biochemical assays
Initially, an experimental procedure composed of three steps,
including the use of bioinformatic tools and biochemical
assays, was developed (Fig. 2A). The ®rst step consists of the
prediction of the RNA secondary structure and the subsequent
identi®cation of the sequences most likely to be in single-
stranded regions. Using the RNA folding algorithm in RNA
4684 Nucleic Acids Research, 2002, Vol. 30 No. 21
Structure 3.5 (27), eight structures were derived for the HBV
RNA pregenome (data not shown). With the exception of
minor local differences, the structures were similar and their
stabilities, based on the values of relative free Gibbs energy
[DG determined by free energy minimization (27)], varied by
<5%. Regardless of the HBV natural sequence variant in
question, the overall organizations of the structures obtained
were similar, suggesting that no important differences can be
attributed to a given genotype. The eight resulting structures
were then analyzed using the OligoWalk program (13). This
algorithm predicts the equilibrium af®nity of an RNA
oligonucleotide for a given sequence, taking into consider-
ation the predicted stability of the oligonucleotide-target helix
and the competition with the predicted secondary structures of
both the target and the oligonucleotide. In our case the
oligonucleotides correspond to the 7 nt composing the P1
binding domain of the delta ribozyme. The HBV sequence is
3358 nt in size and there are N ± L + 1 complementary
oligonucleotides of length L (where N and L are the lengths of
the RNA target and oligonucleotide, respectively, that is 3358
± 7 + 1). Thus, the output consisted of a ®le of 3352 possible
oligonucleotides, for each of which a DG value (corresponding
to an overall free Gibbs energy value estimated for the
interaction between an oligonucleotide and a given target) was
calculated. Using an arbitrary cut-off of DG < ±6.5 kcal/mol,
343 sequences were retained as potential target sites.
The second step consists of screening the sites at the
nucleotide sequence level based on two criteria. The ®rst of
these involved searching for speci®c sequence features known
to be critical for ef®cient cleavage by a trans-acting
antigenomic delta ribozyme. These features include: (i) the
®rst base pair of the P1 stem must be a GU wobble base pair
(29,30); (ii) there cannot be a guanosine at the cleavage site
(position ±1) as this results in an uncleavable substrate
(6,7,30); (iii) the presence of two consecutive pyrimidines at
position ±1 and ±2 dramatically reduces the cleavage
ef®ciency of a substrate (30); (iv) the presence of a cytosine
at position +4 (i.e. in the middle of the ribozyme strand of the
P1 stem) signi®cantly reduces the level of cleavage (31).
Searching the 343 sequences for those that possess these four
features reduces the list to 15 potential target sites. The second
criteria we used was to analyze a sequence alignment of the
natural HBV variants in order to identify which of the
potential target sites are conserved. The identi®cation of the
highly conserved sequences in an RNA species should lead to
the development of nucleic acid drugs that have the ability to
speci®cally bind to all, or most, variants of this species.
Twelve HBV variants from representative genotypes were
aligned (Materials and Methods). This alignment revealed that
nine of the 15 potential target sites are relatively highly
conserved (i.e. conserved in at least 9 out of the 12 variants).
These sequences are listed in Table 1.
In order to give a biochemical dimension to the procedure,
the potential sites were then investigated using two assays in
solution. First, RNase H hydrolysis using 7 nt long DNA
oligonucleotides corresponding to the recognition domain of
the ribozyme (i.e. N
6
C) were performed. RNase H speci®cally
cleaves the RNA of an RNA±DNA duplex and can be used to
verify whether or not the binding of an oligonucleotide
occurred and is speci®c to a target sequence (10). Randomly
labeled HBV RNA was pre-incubated with unlabeled DNA
oligonucleotides, RNase H digested and the mixtures analyzed
on denaturing PAGE gels. A typical gel is shown in Figure 2B.
The presence of the oligonucleotides corresponding to
positions 1154±1160 and 1543±1549 allowed the RNase H
to cleave the HBV RNA, although at different levels (lanes 3
and 5). These results are summarized in Table 1 (upper part).
Three types of results were obtained: (i) the oligonucleotide
corresponding to position 1147 did not result in any detectable
cleavage under the conditions used; (ii) those for positions 409
and 2808 gave several non-speci®c products, suggesting that it
bound to the HBV RNA at more than one position, a result that
may be explained by the fact that RNase H requires only four
Figure 2. Selection of the sites with the greatest potential for delta ribo-
zyme cleavage based on the use of bioinformatic tools coupled to bio-
chemical assays. (A) Scheme of the steps that compose this procedure.
Adjacent to the OligoWalk is an illustration of an example RNA structure
that includes two oligonucleotides binding sites located in single-stranded
regions and one (indicated by the large X) that is unavailable because it is
located in a double-stranded region. (B) Typical results from biochemical
assays for two different sites. Lanes 1 and 2 are the negative controls for
the RNase H and ribozyme assays, respectively. In each case the HBV
RNA was incubated under the reaction conditions, but either the oligo-
nucleotide or the ribozyme were lacking. Lanes 3 and 5 are the RNase H
assays in the presence of the oligonucleotide directing the hydrolyses at
positions 1154 and 1543, respectively. Lanes 4 and 6 are the ribozyme
assays for cleavage at positions 1154 and 1543, respectively.
Nucleic Acids Research, 2002, Vol. 30 No. 21 4685
consecutive base pairs to be active (32); and (iii) the six others
that resulted in speci®c cleavage, albeit at different levels.
Secondly, the cleavage activity of the ribozymes was tested.
Delta ribozymes with appropriate recognition sequences were
synthesized by in vitro transcription from a minigene under
the control of the T7 RNA promoter cloned into pUC19 for
each of the six potential sites that gave speci®c RNase H
hydrolysis. The ability of each ribozyme to cleave the labeled
HBV RNA was tested under single-turnover conditions ([Rz]
> [S]). After incubation at 37°C, the reactions were quenched
and analyzed on PAGE gels. Typical results are shown in
Figure 2B. The Rz-1154 cleaved the HBV RNA poorly, while
the Rz-1543 appeared to be more ef®cient. Only two out of the
six ribozymes tested exhibited cleavage of the HBV RNA (i.e.
Rz-1154 and Rz-1543; see Table 1).
Thus, from the nine potential sites identi®ed in the
bioinformatic approach, only two were positive for in vitro
delta ribozyme cleavage. Clearly, the biochemical assays are
essential in order to validate the predictions derived from the
bioinformatic analyses. The robustness of the initial computer
predictions were further tested by analyzing seven additional
sites with DG > ±6.5 kcal/mol (data not shown). The sequences
of these sites respected the nucleotide requirements con-
sidered previously. Among these sites, three appear to be
accessible based on the RNase H assay (Table 1, lower part).
The ribozymes corresponding to these three sites were
synthesized and their cleavage activities accessed. Rz-486
did not detectably cleave the HBV RNA, but both Rz-212 and
Rz-513 did, although at different levels. Clearly, this approach
appears to have important limitations, at least in the case of the
delta ribozyme.
Biochemical approaches based on the use of libraries
Target site accessibility and the ability to form an active
ribozyme±substrate complex are two interdependent factors
whose relationship is complex. Such a relationship can be
better addressed by using a combinatorial method. Conse-
quently, we developed two procedures with a delta ribozyme
based on previous reports for both the hammerhead and
hairpin ribozymes (12).
A library of oligonucleotides. Figure 3A is a schematic
representation of the procedure using a library of oligonucle-
otides. In theory, all accessible sites within an RNA molecule
(i.e. those in single-stranded regions) would be speci®cally
bound by an oligonucleotide and the resulting RNA±DNA
heteroduplex subsequently hydrolyzed by RNase H (examples
in 11,20). The cleavage sites can then be identi®ed by
electrophoresis of primer extension reactions using 5¢-end-
labeled primers and the RNase H reaction products. From the
sequence of the substrate (HBV RNA), the sequence at
the cleavage site may be determined and, thus, that of the
complementary oligonucleotide (i.e. that acting as the
P1 domain in the ribozyme). Finally, the corresponding
ribozymes are synthesized and the cleavage activity tested.
The library used was composed of oligonucleotides 8 nt in
length corresponding to one residue before the cleavage site,
which has to be single-stranded (i.e. position ±1), and the
seven residues of the recognition domain of the ribozyme (i.e.
the P1 stem). The libraries were designed taking into account
the sequence speci®cities described earlier for delta ribozyme
cleavage. More speci®cally, the nucleotide at position ±1
relative to the cleavage site cannot be a guanosine (30),
therefore the 3¢-end nucleotide of the oligonucleotide cannot
be a C. The ®rst base pair of the P1 stem has to be a GU
wobble pair (29,30), therefore the seventh nucleotide of all
oligonucleotides was a cytosine (i.e. allowing the formation of
a GC base pair). The larger library of oligonucleotides
corresponds to the sequence 5¢-N
6
CD-3¢(i.e. 12 288 different
sequences), while the smaller one corresponds to the sequence
5¢-SSNDMSCD-3¢(i.e. 576 different sequences) (where N
indicates A, C, G or U; D indicates A, G or U; S indicates C or
G; and, M indicates A or C). The latter library was designed to
include more of the sequence preferences of trans-acting delta
ribozymes, and to favor stronger binding of the substrate to the
ribozyme due to the presence of a higher percentage of GC
residues.
A typical example of the identi®cation of a potential site is
illustrated in Figure 3B. In this case the primer extension
was performed using an oligonucleotide complementary to
positions 632±646 of the HBV RNA. In the absence of any
Table 1. Compilation of the potential sites identi®ed by the computer method
Cleavage
position
HBV P1 stem ORF targetedaDG
(kcal/mol)
RNase H
assayc
Rz
assayc
409 GCCCCUA C and P genes ± 6.6 n.s. n.d.
523 AGAUCUC P gene ±8.0 ++ ±
524 GAUCUCA P gene ±8.0 + ±
946 GCAUGGG PreS1 gene ±8.5 + ±
1147 GCCCUCA PreS1 gene ±9.7 ± n.d.
1154 GGCUCAG PreS1 gene ±8.5 +++ +
1543 GGACUUC S genes ±7.1 +++ ++
2250 GGCCUAU P gene ±8.3 + ±
2808 GCACCUC X gene ±8.1 n.s. n.d.
212bGGGUGGG C genes ± 0.3 +++ +
486bGCCUCGC P gene ± 0.9 +++ ±
513bGCGUCGC P gene ±3.0 +++ +++
n.d., not determined; n.s., not speci®c.
aSee Figure 1B.
bThese sites were suggested not accessible.
cRelative level of cleavage.
4686 Nucleic Acids Research, 2002, Vol. 30 No. 21
oligonucleotide no band was observed in the speci®c region as
compared with lanes 1 and 2 (lane 3). In the presence of the
oligonucleotides from the 5¢-N
6
CD-3¢library, several bands
were detected (lane 2). This indicates the presence of a single-
stranded region accessible to several oligonucleotides binding
on different copies of the RNA target, and results in cleavage
at various contiguous positions. In the presence of oligo-
nucleotides from the 5¢-SSNDMSCD-3¢library (lane 1), fewer
bands were observed. This re¯ects the smaller expansion of
the sequences composing the library. However, since each
sequence is present in a higher concentration (i.e. for the same
absolute amount of oligonucleotides) the concentration of
cleaved RNA, and consequently cDNA, is larger. This assay
suggested that a ribozyme may be engineered for cleavage at
position 513 (i.e. Rz-513). Rz-513 speci®cally and ef®ciently
cleaved the HBV RNA as shown by the primer extension
assay in Figure 3C.
Since each primer produced a readable sequence of
200±300 bases, several primers were required in order to
map the full HBV RNA. An average of two potential sites
were detected for each primer extension reaction (i.e. per
~200±300 bases) based on the detection of either one, or
stretches of, band(s). Most of the time, either one or two
consecutive bands of weak intensity were detected.
Preliminary experiments have shown that such bands do not
indicate productive sites. When we considered the detection of
three or more consecutive bands in primer extension, we
retrieved only six potential sites (Table 2). In each case, the
region was analyzed to determine the most appropriate
sequence in terms of cleavage by a delta ribozyme (i.e.
position ±1 to the cleavage site must be A, C or U). With the
exception of Rz-887, ®ve of the six ribozymes exhibited a
relatively ef®cient cleavage of the HBV RNA under the
conditions used (Table 2). Of these ®ve active ribozymes, four
were conserved (only Rz-574 was not) with respect to the
sequence alignment of the HBV variants (i.e. present in at least
9 out of the 12 variants).
A combinatorial library of ribozymes. Finally, we tried an
approach based on a library of delta ribozymes (Fig. 4A). The
most important advantage of this approach should be that the
detection of a cDNA band is directly related to the cleavage of
a ribozyme, whereas the bands produced by the previous
RNase H-based assay were an indirect indication. Conse-
quently, this method is theoretically better because the
selection is based not only on a mimic of the binding site,
but rather on both binding and cleavage by the ribozyme.
A pool of ribozymes was synthesized using a minigene
under the control of a T7 RNA polymerase promoter (see
inset, Fig. 4A). The minigene was made from a pool of
antisense oligonucleotides in which the positions correspond-
ing to the P1 stem were randomized, and the second strand
synthesized using a sense primer that formed the T7 RNA
promoter. The library was based on the randomization of the
six positions of the P1 stem (i.e. 5¢-N
6
U-3¢, giving 4096
ribozymes). This library took into account the requirement for
a U for the formation of the GU wobble base pair adjacent to
the cleavage site (29,30). The ribozymes (i.e. 2.4 nM of each
individual ribozyme) were incubated with HBV RNA (50 nM),
and the cleavage sites identi®ed by electrophoresis of primer
extension reactions as was done for the RNase H approach
Figure 3. Selection of the sites with the greatest potential for delta ribo-
zyme cleavage based on the use of an RNase H assay in the presence of a
library of oligonucleotides. (A) Scheme illustrating the procedure. HBV
RNA and cDNA are represented by the full and dashed lines, respectively.
RT is reverse transcriptase. (B) Autoradiogram of a sequencing gel reveal-
ing the single-stranded region that includes position 513. Lanes 1 and 2 cor-
respond to the reactions performed in the presence of the 5¢-SSNDMSCD-3¢
and 5¢-N
6
CD-3¢libraries, respectively (where N = A, C, G or U; D = A, G
or U; M = A or C; and S = C or G). Lane 3 is the reaction performed in the
absence of any oligonucleotide. Adjacent to the gel is a DNA sequencing
reaction, and the corresponding sequence is shown on the right side.
(C) Autoradiogram of a sequencing gel of the cleavage assay for Rz-513.
Lanes 1 and 2 are reactions performed in the absence of either ribozyme or
MgCl
2
, respectively. Lane 3 is the complete reaction. After the cleavage
reaction, primer extension was performed and migrated beside a sequencing
reaction. The sequence is shown adjacent to the gel. The arrow indicates the
cleavage site.
Nucleic Acids Research, 2002, Vol. 30 No. 21 4687
described above. A typical example of the identi®cation of a
site is illustrated in Figure 4B. In this case the primer
extension was performed using an oligonucleotide comple-
mentary to positions 329±343 of the HBV RNA. The presence
of the pool of ribozymes resulted in the detection of an
additional band corresponding to cleavage at position 303
(lane 1) as compared with the results obtained in the absence
of either MgCl
2
(lane 2) or the ribozyme (lane 3). Scanning the
entire length of the HBV RNA revealed six potential sites
(Table 3). Subsequently, speci®c delta ribozymes were made
from the minigenes and their cleavage activities assessed. Five
out of the six cleaved the HBV RNA, with the Rz-303, Rz-
1543, Rz-2833 and Rz-2861 recognizing conserved sequences
of the HBV variants (i.e. present in at least in 9 out of 12
variants).
DISCUSSION
Procedures for the selection of sites
This work presents the development of three procedures for
the identi®cation of delta ribozyme cleavage sites within an
RNA target, a step essential to the development of a gene
therapy. The ®rst approach consists of an experimental
scheme composed of three steps, including the use of
bioinformatic tools coupled to biochemical assays (Fig. 2).
At its heart, this procedure is based on computer prediction of
the RNA secondary structure, a procedure that is the most
popular method of determining the structures of all kinds of
RNA (13,16) and elucidating the regions that are most likely
to be single-stranded. Of the nine potential sites obtained by
the computer-assisted analysis, six were both speci®c and
Table 2. Compilation of the potential sites identi®ed by the RNase H method
Cleavage position HBV P1 stem ORF targetedaRNase H assay (N)bRz assayc
167 GGCAAGC C genes + (3) +++
279 GGGCCUA C genes + (4) +++
461 GGUCCCC P gene + (4) ++
513 GCGUCGC P gene + (9) +++
574 GGGGAAC P gene + (3) +
887 GAAACAA P gene + (7) ±
aSee Figure 1B.
bN indicates number of cleavage bands.
cRelative level of cleavage.
Figure 4. Selection of the sites with the greatest potential for delta ribozyme cleavage sites based on a ribozyme cleavage assay performed in the presence of
a ribozyme library. (A) Scheme illustrating the procedure. HBV RNA and cDNA are represented by the full and dashed lines, respectively. RT is reverse tran-
scriptase. The PCR strategy used to synthesize the library is shown in the inset in which oligonucleotides A and C served as sense and antisense primers,
respectively, to amplify the randomized sequences of the templates (oligonucleotide B). (B) Autoradiogram of a sequencing gel revealing the cleavage site at
position 303. Lane 1 is the reaction performed in the presence of ribozymes and MgCl
2
, while MgCl
2
and the ribozyme were omitted in the reactions in lanes
2 and 3, respectively. Adjacent to the gel is a DNA sequencing reaction, and the corresponding sequence is shown on the right side. The arrow indicates the
cleavage site.
4688 Nucleic Acids Research, 2002, Vol. 30 No. 21
effective for the RNase H hydrolysis (Table 1). However, only
two members of this subset were cleaved by the appropriate
ribozymes. Although the two ®rst steps were performed
rapidly, as the coupling of the biochemical approach is
essential to providing relevant results, this procedure did not
result in any signi®cant gains in terms of time. We concluded
that this bioinformatic approach has severe limitations, at least
in the case of delta ribozyme. Many reasons can be put forth to
explain the discrepancy between the computer predictions and
sites effectively cleaved by the ribozymes, including the fact
that the tertiary structures of both the ribozyme and the RNA
target were not taken into account by the software. However,
the most important reason is the fact that, even today, the
prediction of RNA structure remains a complicated task for
long RNA molecules (8±10).
Based on our results, a combinatorial method is the best
means of overcoming the complexed interdependent nature of
target site accessibility and the ability to form an active
ribozyme±substrate complex. In the RNase H procedure, the
presence of combinatorial libraries of oligonucleotides mimics
the binding of the ribozyme. Taking into account some of the
substrate speci®city features of the delta ribozyme cleavage
allowed us to both limit the expansion of the library and to
`personalize' it to this catalytic RNA. Preliminary results
showed that only the presence of a stretch of bands,
corresponding to consecutive stops of the primer extension,
was indicative of a productive site. The binding domain of the
substrate is restricted to seven bases, but a larger single-
stranded domain is required in order to both satisfy the tertiary
structure of the ribozyme and permit its molecular mechanism
(which involves a conformational transition of the ribozyme±
substrate complex) (7). The procedure based on a combin-
atorial library of delta ribozymes randomized for their
recognition sequences offers the advantage of selecting sites
that are bound by a ribozyme and subsequently cleaved. In that
case only single bands were seen. Both combinatorial methods
allowed us to identify six potential sites, ®ve of which were
cleaved in vitro by the appropriate ribozymes in each case.
However, none of the selected sites were identi®ed by both the
oligonucleotide and ribozyme libraries. This observation
indicates that none of the methods permits the identi®cation
of all potential sites, which is clearly a limiting factor.
Substrate speci®city for delta ribozyme cleavage
Using small model substrates, the speci®city for delta
ribozyme cleavage has been determined in vitro (29±31,33).
Brie¯y, the P1 stem is composed of the 7 nt that base pair with
the ribozyme; hence, these nucleotides can be de®ned as
internal determinants of substrate speci®city. Moreover, it has
been demonstrated that the nucleotide composition at pos-
itions ±1 to ±4 from the scissile bond in¯uences the ef®ciency
of cleavage of small substrates (31,33). For example, the
presence of two, single-stranded, consecutive pyrimidines at
positions ±1 and ±2 decreases the levels of cleavage to varying
degrees depending on the identity of the residues in question.
The nucleotides in positions ±1 to ± 4 do not directly contribute
to substrate recognition (e.g. substrate binding), but the
identity of these residues is essential for ef®cient cleavage
(10,30). Therefore, they contribute to substrate speci®city,
likely as external determinants. In this study, the nucleotides at
positions ±1 to ±4 had various compositions. Most of the
cleavage sites retained do not possess consecutive pyrimidines
at positions ±1 and ±2, or if they did they were inef®ciently
cleaved, supporting the idea that the rules established with
small substrates are consistently respected in a long RNA
target. However, we also observed exceptions to these rules:
for example, the Rz-513 exhibited an ef®cient cleavage even
though two cytosines are located at positions ±1 and ±2. This
result shows that the situation of a small substrate is slightly
different to that of a long molecule such as HBV RNA. In fact,
there seems to be a complex situation in which several factors,
including the secondary structure motifs in close proximity to
a potential site, as well as long interactions, modulate the
ef®ciency of the cleavage.
Any contribution to the substrate speci®city for delta
ribozyme cleavage is important for further development
in vivo. As any given 7mer that corresponds to the length of
the P1 stem (the binding domain) might be found at a
frequency of one in 47nt (i.e. 16 384), duplicate targets might
be present in the population of cellular mRNA, and cleavage
of these RNAs could prove toxic to the cell. Therefore, the
modulation of cleavage ef®ciency caused by the sequences in
positions ±1 to ±4 of the cleavage site might be of primary
importance in minimizing this problem. The net effect of this
is to extend from 7 to 11 contiguous nucleotides the domain of
the substrate that contributes to determining the ability of an
RNA molecule to be cleaved by delta ribozyme. Furthermore,
it is important to remember that the complexity of an RNA
structure drastically reduces the number of potential sequen-
ces that can be considered as potential sites to target. In this
study, the use of a combinatorial library including thousands
of different ribozymes did not allow us to retrieve too many
active ribozymes even though the sequence complementary to
the recognition sequence was present within the HBV RNA.
Clearly, the secondary and tertiary structures of the RNA
molecules, as well as their complexation with other cellular
Table 3. Compilation of the potential sites identi®ed by the library of delta ribozymes
Cleavage position HBV P1 stem ORF targetedaRz pool assaybRz assayb
303 GUGGUUU C genes ++ +++
1365 GUAAACC Pre S2 + ±
1543 GGACUUC S genes ++ ++
2624 GUUGUCC P gene + ++
2833 GUCUGUG X gene + +++
2861 GUGCACU X gene + +++
aSee Figure 1B.
bRelative level of cleavage.
Nucleic Acids Research, 2002, Vol. 30 No. 21 4689
components (e.g. with proteins forming RNPs) also make
important contributions to minimizing the problem of lack of
speci®city.
Targeting HBV
Together, the three procedures used in this work yield a
collection of 12 delta ribozymes with the potential to target
throughout the HBV RNA pregenome. Comparing the sites
identi®ed here, within the HBV RNA, with those of other
reports, based on the use of other catalytic RNAs like the
hammerhead and the hairpin ribozyme (14,24,25,34±37),
shows no similarities. Moreover, when we consider the sites
targeted in other HBV works it appears impossible to engineer
delta ribozyme cleaving at these positions, primarily because
the delta ribozyme has different characteristics than other
catalytic RNAs. On the other hand, the extension of this work
would be to show whether or not some, or all, of the delta
ribozymes of the collection would be effective in controlling
the propagation of HBV in vivo. However, previous reports
have demonstrated the existence of relatively good correl-
ations between in vitro and in vivo for both oligonucleotide
and ribozyme effects (19,20,22,38).
ACKNOWLEDGEMENTS
We thank Dr Michael Nassal for providing the HBV vector.
This work was supported by grants from both the Canadian
Institutes of Health Research (CIHR) and Health Canada to
J.-P.P. The RNA group is supported by grants from both the
CIHR and Fonds FCAR (Que
Âbec). L.J.B. was a recipient
of a pre-doctoral fellowship from FCAR/FRSQ and the
Georges Phe
Ânix Fondation. J.-P.P. is an investigator from
the CIHR.
REFERENCES
1. Lewin,A.S. and Hauswirth,W.W. (2001) Ribozyme gene therapy:
application for molecular medicine. Trends Mol. Med.,7, 221±228.
2. Michienzi,A. and Rossi,J.J. (2001) Intracellular applications of
ribozymes. Methods Enzymol.,341, 581±596.
3. Sun,L.Q., Cairns,M.J., Saravolac,E.G., Baker,A. and Gerlach,W.L.
(2000) Catalytic nucleic acids: from lab to application. Pharmacol. Rev.,
52, 325±347.
4. Shih,I.H. and Been,M.D. (2002) Catalytic strategies of the hepatitis delta
virus ribozymes. Annu. Rev. Biochem.,71, 887±917.
5. Le
Âvesque,D., Choufani,S. and Perreault,J.P. (2002) Delta ribozyme
bene®ts from a good stability in vitro that becomes outstanding in vivo.
RNA,8, 464±477.
6. Ananvoranich,S. and Perreault,J.P. (1998) Substrate speci®city of d
ribozyme cleavage. J. Biol. Chem.,273, 13182±13188.
7. Mercure,S., Lafontaine,D., Ananvoranich,S. and Perreault,J.P. (1998)
Kinetic analysis of dribozyme cleavage. Biochemistry,37,
16975±16982.
8. Campbell,T.B. and Cech,T.R. (1995) Identi®cation of ribozymes within a
ribozyme library that ef®ciently cleave a long substrate RNA. RNA,1,
598±609.
9. Birikh,K.R., Berlin,Y.A., Soreq,H. and Eckstein,F. (1997) Probing
accessible sites for ribozymes on human acetylcholinesterase RNA. RNA,
3, 429±437.
10. Roy,G., Ananvoranich,S. and Perreault,J.P. (1999) Delta ribozyme has
the ability to cleave in trans an mRNA. Nucleic Acids Res.,27, 942±948.
11. Lima,W.F., Brown-Driver,V., Fox,M., Hanecak,R. and Bruice,T.W.
(1997) Combinatorial screening and rational optimization for
hybridization to folded hepatitis C virus RNA of oligonucleotides with
biological antisense activity. J. Biol. Chem.,272, 626±638.
12. Yu,Q., Pecchia,D.B., Kingsley,S.L., Heckman,J.E. and Burke,J.M.
(1998) Cleavage of highly structured viral RNA molecules by
combinatorial libraries of hairpin ribozymes. J. Biol. Chem.,273,
23524±23533.
13. Mathews,D.H., Burkard,M.E., Freier,S.M., Wyatt,J.R. and Turner,D.H.
(1999) Predicting oligonucleotide af®nity to nucleic targets. RNA,5,
1458±1469.
14. zu Putlitz,J., Yu,Q., Burke,J.M. and Wands,J.R. (1999) Combinatorial
screening and intracellular antiviral activity of hairpin ribozymes
directed against hepatitis B virus. J. Virol.,73, 5381±5387.
15. Scherr,M., Rossi,J.J., Sczakiel,G. and Patzel,V. (2000) RNA accessibility
prediction: a theoritical approach is consistent with experimental studies
in cell extracts. Nucleic Acids Res.,28, 2455±2461.
16. Amarzguioui,M., Brede,G., Babaie,E., Grotli,M., Sproat,B. and Prydz,H.
(2000) Secondary structure prediction and in vitro accessibility of mRNA
as tools in the selection of target sites for ribozymes. Nucleic Acids Res.,
28, 4113±4124.
17. Scherr,M., LeBon,J., Castanotto,D., Cunliffe,H.E., Meltzer,P.S.,
Ganser,A., Riggs,A.D. and Rossi,J.J. (2001) Detection of antisense and
ribozyme accessible sites on native mRNAs: application to NCOA3
mRNA. Mol. Ther.,4, 454±460.
18. Barroso-delJesus,A. and Berzal-Herranz,A. (2001) Selection of targets
and the most ef®cient hairpin ribozymes for inactivation of mRNAs using
a self-cleaving RNA library. EMBO Rep.,21, 1112±1118.
19. Kru
Èger,M., Beger,C., Welch,P.J., Berber,J.R. and Wong-Staal,F. (2001)
C-SPACE (cleavage-speci®c ampli®cation of cDNA ends): a novel
method of ribozyme-mediated gene identi®cation. Nucleic Acids Res.,
29, e94.
20. Sohail,M., Hochegger,H., Klotzbu
Ècher,A., Le Guellec,R., Hunt,T. and
Southern,E.M. (2001) Antisense oligonucleotides selected by
hybridization to scanning arrays are effective reagents in vivo.
Nucleic Acids Res.,29, 2041±2051.
21. Pan,W.H., Devlin,H.F., Kelley,C., Isom,H.C. and Clawson,G.A. (2001)
A selection system for identifying accessible sites in target RNAs. RNA,
7, 610±621.
22. Ding,Y. and Lawrence,C.E. (2001) Statistical prediction of single-
stranded regions in RNA secondary structure and application to
predicting effective antisense target sites and beyond. Nucleic Acids Res.,
29, 1034±1046.
23. Tong,C.Y.W. (2000) Genetic variations of hepatitis B virus. Curr. Opin.
Infect. Dis.,13, 481±487.
24. Welch,P.J., Tritz,R., Yei,S., Barber,J. and Yu,M. (1997) Intracellular
application of hairpin ribozyme genes against hepatitis B virus. Gene
Ther.,4, 736±743.
25. Weinberg,M., Passman,M., Kew,M. and Arbuthnot,P. (2000)
Hammerhead ribozyme-mediated inhibition of hepatitis B virus X gene
expression in cultured cells. J. Hepatol.,33, 142±151.
26. Nassal,M. (1992) The arginine-rich domain of the hepatitis B virus core
protein is required for pregenome encapsidation and production viral
positive-strand DNA synthesis but not for virus assembly. J. Virol.,66,
4107±4116.
27. Mathews,D.H., Sabina,J., Zuker,M. and Turner,D.H. (1999) Expanded
sequence dependence of thermodynamic parameters provides improved
prediction of RNA secondary structures. J. Mol. Biol.,288, 911±940.
28. Thompson,J.D., Higgins,D.G. and Gibson,T.J. (1994) Clustal WÐ
improving the sensitivity of progressive multiple sequence alignment
through sequence weighting, position-speci®c gap penalties and weight
matrix choice. Nucleic Acids Res.,22, 4673±4680.
29. Nishikawa,F., Fauzi,H. and Nishikawa,S. (1997) Detailed analysis of
base preferences at the cleavage site of a trans-acting HDV ribozyme: a
mutation that changes cleavage site speci®city. Nucleic Acids Res.,27,
403±410.
30. Desche
Ãnes,P., Lafontaine,D.A., Charland,S. and Perreault,J.P. (2000)
Nucleotides ±1 to ±4 of hepatitis delta ribozyme substrate increase the
speci®city of ribozyme cleavage. Antisense Nucleic Acid Drug Dev.,10,
53±61.
31. Ananvoranich,S., Lafontaine,D.A. and Perreault,J.P. (1999) Mutational
analysis of the antigenomic trans-acting delta ribozyme: the alterations
of the middle nucleotides located on the P1 stem. Nucleic Acids Res.,17,
1473±1479.
32. Hogrefe,H.H., Hogrefe,R.I., Walder,R.Y. and Walder,J.A. (1990) Kinetic
analysis of Escherichia coli RNase H using DNA-RNA-DNA/DNA
substrates. J. Biol. Chem.,265, 5561±5566.
4690 Nucleic Acids Research, 2002, Vol. 30 No. 21
33. Shih,I.H. and Been,M.D. (2001) Energetic contribution of non-essential
5¢sequence to catalysis in a hepatitis delta virus ribozyme. EMBO J.,20,
4884±4891.
34. Feng,Y., Kong,Y.Y., Wang,Y. and Qi,G. (2001) Inhibition of hepatitis B
virus by hammerhead ribozyme targeted to the poly(A) signal sequence
in cultured cells. Biol. Chem.,382, 655±660.
35. Beck,J. and Nassal,M. (1995) Ef®cient hammerhead ribozyme-mediated
cleavage of the structured hepatitis B virus encapsidation signal in vitro
and in cell extracts, but not in intact cells. Nucleic Acids Res.,23,
4954±4962.
36. Kim,Y.K., Junn,E., Park,I., Lee,Y., Kang,C. and Ahn,J.K. (1999)
Repression of hepatitis B virus X gene expression by hammerhead
ribozymes. Biochem. Biophys. Res. Commun.,257, 759±765.
37. Passman,M., Weinberg,M., Kew,M. and Arbuthnot,P. (2000) In situ
demonstration of inhibitory effects of hammerhead ribozymes that are
targeted to the hepatitis Bx sequence in cultured cells. Biochem. Biophys.
Res. Commun.,268, 728±733.
38. Kato,Y., Kuwabara,T., Warashina,M., Toda,H. and Taira,K. (2001)
Relationships between the activities in vitro and in vivo of various kinds
of ribozyme and their intracellular localization in mammalian cells.
J. Biol. Chem.,276, 15378±15385.
Nucleic Acids Research, 2002, Vol. 30 No. 21 4691
