Proc. Natl. Acad. Sci. USA
Vol. 96, pp. 6609–6614, June 1999
A small nucleolar RNA:ribozyme hybrid cleaves a nucleolar RNA
target in vivo with near-perfect efficiency
DMITRY A. SAMARSKY†‡, GERARDO FERBEYRE§¶, EDOUARD BERTRAND?‡‡, ROBERT H. SINGER?,
ROBERT CEDERGREN§††, AND MAURILLE J. FOURNIER†§§
†Department of Biochemistry and Molecular Biology, University of Massachusetts, Amherst, MA 01003;§Departement de Biochimie, Universite de Montreal,
Montreal, Quebec H3C 3J7, Canada; and?Departments of Anatomy and Structural Biology and Cell Biology, Albert Einstein College of Medicine,
Bronx, NY 10461
Communicated by John A. Carbon, University of California, Santa Barbara, CA, April 13, 1999 (received for review March 1, 1999)
to the yeast nucleolus by using the U3 small nucleolar RNA as
a carrier. The hybrid small nucleolar RNA:ribozyme, desig-
nated a ‘‘snorbozyme,’’ is metabolically stable and cleaves a
target U3 RNA with nearly 100% efficiency in vivo. This is the
most efficient in vivo cleavage reported for a trans-acting
ribozyme. A key advantage of the model substrate featured is
that a stable, trimmed cleavage product accumulates. This
property allows accurate kinetic measurements of authentic
cleavage in vivo. The system offers new avenues for developing
effective ribozymes for research and therapeutic applications.
A hammerhead ribozyme has been localized
Ribozymes have enormous potential in basic research and
biotechnology with special promise for applications in human
medicine (1, 2). In this context, a catalytic RNA is designed to
specifically recognize, interact with, and alter a target mole-
cule. The best-characterized ribozymes are hydrolytic nucle-
ases that perform site-specific cleavage of a target RNA (3, 4).
The hammerhead ribozyme has received most attention thus
far because of its early discovery, small size, and potential to
cleave almost any RNA molecule (5–7). Naturally occurring
hammerhead ribozymes cleave target sequences in the same
molecule, i.e., in cis (ref. 8; see also ref. 9). It is possible
however, to engineer artificial ribozymes, which cleave RNA
molecules in trans. In principle, this strategy can be used to
alter the level of any RNA in a living organism, and achieving
this goal has been a major focus of ribozyme research (see refs.
10 and 11). Despite much effort, progress has been slow in
attaining highly efficient cleavage of target molecules in living
cells (12–17). In no case known to us has complete or nearly
complete cleavage been observed in vivo for any trans-acting
hammerhead ribozyme. Overcoming this limitation is critical
if the full potential for modulating gene expression with
ribozymes is to be realized.
To achieve efficient cleavage in vivo, three key conditions
must be satisfied: (i) an intracellular concentration of ri-
bozyme that is not limiting, which requires a robust transcrip-
tion rate and/or good metabolic stability of the ribozyme; (ii)
a high degree of colocalization of the ribozyme to the same
cellular compartment(s) as the target molecule (13, 16); and
(iii) highly productive ribozyme–target interaction, i.e., good
substrate molecules (18). In this last regard, either or both
RNAs may be large and complex and bound to proteins; thus,
the potential for interference effects can be high. Lack of
detailed information about the in vivo structures and dynamics
of the interacting RNAs continues to be a major problem. In
the present study we sought to satisfy the requirements for
effective in vivo cleavage by using small nucleolar RNAs
(snoRNAs) as both catalytic and target molecules and to
demonstrate that complete or near-complete cleavage is in-
deed possible with a trans-acting ribozyme.
Yeast Strains and Manipulations. In vivo studies with
cis-acting snorbozymes were performed with a haploid yeast
strain transformed with appropriate plasmids. The host strain
in these experiments was YSD2 (MATa ura3 ade2 his3 leu2 lys1
trp1) prepared by sporulating the isogenic MH2 strain (kindly
provided by Molly Fitzgerald-Hayes, Univ. of Massachusetts,
Amherst, MA). In vivo studies with the trans-acting snor-
bozymes were conducted with diploid yeast strains containing
appropriate plasmid pairs. The latter strains were prepared as
follows. First, two isogenic haploid strains, YSD2 and YSD1,
obtained after sporulation of diploid MH2 and differing only
in mating type (a vs. ?), were transformed with plasmids
expressing either target- (YSD2) or ribozyme (YSD1)-
containing molecules. The transformants obtained were then
crossed to yield diploids expressing the desired pairs of RNA
molecules. Plasmids were introduced into yeast cells by using
a lithium acetate transformation procedure (19).
Analysis of Ribozyme Activity. The RNA molecules used in
the in vitro ribozyme studies were prepared by transcribing
DNA templates with T7 RNA polymerase. The templates were
prepared by PCR amplification of the plasmids used in the in
vivo ribozyme studies (see below). The cis reaction occurred
during the transcription incubation (1 h) with almost 100%
efficiency. In the trans ribozyme reactions, mutant and wild-
type ribozyme and substrate transcripts were quantified spec-
trophotometrically or estimated from [32P]UTP incorporation
measurements. A constant amount of substrate (2 nM) was
incubated with 10, 20, 40, 60, 80, and 100 nM ribozyme in 20
mM Tris?HCl (pH 7.9)/100 mM NaCl/10 mM MgCl2/2 mM
spermidine. Ribozyme and substrate were first combined in
the above conditions, but without Mg2?, heated at 80°C for 1
min, and cooled on ice. Reactions were started by adding Mg2?
and incubated for 1 h at 37°C. Under these conditions, ?40%
of the substrate was cleaved at the highest ribozyme concen-
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
PNAS is available online at www.pnas.org.
Abbreviation: snoRNA, small nucleolar RNA.
‡Present address: Howard Hughes Medical Institute and Program in
Molecular Medicine, University of Massachusetts Medical Center,
373 Plantation Street, Worcester, MA 01606.
¶Present address: Cold Spring Harbor Laboratory, Cold Spring Har-
bor, NY 11724.
††Deceased October 14, 1998.
‡‡Present address: Laboratoire de Jean-Marie Blanchard, Institute de
Genetique Moleculaire de Montpelier, Centre National de la Re-
cherche Scientifique, B.P. 5051, 1919 Route de Mende, 34033
Montpelier Cedex 01, France.
§§To whom reprint requests should be addressed. e-mail: 4nier@
tration (data not shown). The efficiency of ribozyme cleavage
in vivo was assessed by Northern blot analysis.
RNA isolation and Northern blot analyses were performed
as described (20). Total RNA for Northern blots was prepared
from cells grown overnight in selective liquid YNB medium
(0.67% yeast nitrogen base). Radiolabeled oligonucleotides
used to probe the Northern blots were C106, which recognizes
to all hybrid molecules used in the cis ribozyme studies as well
as substrate and product molecules featured in the trans
ribozyme studies (this probe specifically recognizes a unique
hairpin sequence in a modified U3*; see Fig. 1); and SD90,
which recognizes specifically ribozyme-carrying molecules
used in the trans ribozyme studies (this probe hybridizes to a
core sequence in the hammerhead ribozyme, on the 5? side of
the nucleotide used to prepare inactive ribozyme variants).
In situ hybridization and probe synthesis were performed as
described (21). The probes used were RNA molecules conju-
gated to either CY3 or Oregon green 488. These RNAs were
obtained by using in vitro transcription of plasmid templates
containing either the ribozyme or the U3* hybridization tag
DNA Constructs. The RNA molecules used in the in vitro
studies were prepared by transcribing DNA templates with T7
RNA polymerase according to a protocol provided by the
manufacturer (Promega). The DNA templates were prepared
by PCR amplification of the plasmids used in the in vivo studies
(see below) and are as follows. Active cis-snorbozyme, plasmid
pRS:T/C(?) with primers SD101 and SD102; inactive cis-
snorbozyme, plasmid pRS:T/C(?) with primers SD101 and
SD102; active trans-snorbozyme, plasmid pRS:C(?) with
primers SD99 and SD101; inactive trans-snorbozyme, plasmid
pRS:C(?) with primers SD99 and SD101; substrate molecule
(trans ribozyme analyses), plasmid pRS:T with primers SD100
and SD101. Before T7 transcription, the DNA templates were
treated with HpaI to generate RNA molecules with uniform 3?
ends. Plasmids for the in vivo snorbozyme studies were pre-
pared by substituting major portions of the 5? segment of the
U3-coding region with the appropriate experimental se-
quences, in plasmids pRU3*-313, pRU3(del)-316 or YEp-
U3(del) plasmids (see below). Replacements were performed
by using PCR mutagenesis strategies described (22). The DNA
templates and essential PCR primers were: pRU3*-313 and
SD75, for pRS:T/C(?); pRU3*-313 and SD76, for pRS:T/
C(?); pRU3(del)-313 and SD88, for pRS:C(?) and
YEp:C(?); pRU3(del)-313 and SD89, for pRS:C(?) and
YEp:C(?); pRU3*-313 and SD86, for pRS:T or pRS:T(9:9);
pRS:T and SD97, for pRS:T(6:6); pRS:T and SD95, for
pRS:T(6:9); pRS:T and SD96, for pRS:T(9:6); pRU3*-313 and
SD87, for pRS:T(12:12); and pRS:T and SD98, for
pRS:T(15:15). Plasmids pRU3*-313, pRU3(del)-316 and
YEp-U3(del) were prepared by inserting a 1.3-kb HindIII
fragment encoding U3* or U3(del) variants of the U3A gene
(22) into cloning vectors pRS313 (encoding U3*, into SalI
site), pRS316 [encoding U3(del), into SalI site], and YEp24 m
[encoding U3(del), into ClaI site]. Vectors pRS313 and
pRS316 are described by Sikorski & Hieter (23).
Oligonucleotides. Sequences of the oligonucleotides used in
this study are as follows: for SD13,* 5?-GCGGCTTAGGCT-
AAGCTAAGGCCAGC-3?; SD75, 5?-TGACTCTGTCGAC-
GATGAG-3?; SD76, 5?-TGACTCTGTCGACGTACCTG-
GCCACTGAATCCAACTTGGTTG-3?; SD87, 5?-GACTC-
CTGAATCCAACTTGGTT-3?; SD88, 5?-GACTCTGTCGA-
AATCCAACTTGGTTG-3?; SD90, 5?-ACGTCTACCTGTT-
TCGTCCTCACGGACTCAT-3?; SD95, 5?-ATCTGTGTCG-
ACCGAACCTGTCTAC-3?; SD96, 5?-ATCTGTGTC-
CCCACTGAATC-3? and for SD98, 5?-ATCTGTGTCGA-
bozyme) chimeras featured in the present study. (A) Map of S.
cerevisiae U3 (22). The yeast U3 molecule consists of (i) a 5? region
with little or no secondary folding; (ii) a single-stranded hinge region,
from exonucleolytic damage by a trimethyl guanosine (TMG) cap and
one or more proteins complexed with the box C?/D motif. Key
elements incorporated into the design of the snorbozyme system are
emphasized. The 5? domain contains highly conserved box elements
GAC, A?, and A, which are implicated in direct interaction with
pre-rRNA. The box C?/D motif is a nucleolar localization signal and
provides metabolic stability. The in vivo studies were carried out with
cells expressing wild-type U3, which is essential for growth, and one or
two U3 variants containing the experimental sequences. One variant
(U3*) contains a unique hybridization tag created by modifying the
sequence of one of the helical stems (upper left). The second is a
functional ‘‘mini-U3’’ molecule (U3del), in which all hairpin domains
are absent. The ribozyme and target sequences were introduced into
the 5? region of the U3 host molecules in place of conserved boxes A?
and A and a portion of the hinge segment. (B) Structures of the
hammerhead ribozyme elements featured, in cis and trans configura-
tions (8, 25). A G nucleotide essential for ribozyme cleavage is marked
with a filled triangle. Inactive snorbozymes containing a C at this
position were used in control experiments. The target and catalytic
sequences in the trans configuration are labeled T and C, respectively.
6610 Biochemistry: Samarsky et al. Proc. Natl. Acad. Sci. USA 96 (1999)
ACTGAATCCAACTTG-3?; SD99, 5?-CTCTACGTAATAC-
GACTCTACCTGTCTAC-3?; SD101, 5?-AAGTGGTTAAC-
TTGTCAGACTGCC-3?; SD102, 5?-CTCTACGTAATACG-
Recent advances in snoRNA research suggested to us that a
snoRNA molecule could be used as a ribozyme carrier to
target nucleolar RNAs. This approach was initially prompted
by identification of a common snoRNA structure motif, known
as the box C/D motif, which confers metabolic stability on the
snoRNA and also causes the RNA to be localized to the
nucleolus (ref. 21, and refs. therein). Other key advances were
of valuable information about the structure of the natural U3
snoRNP complex in this organism (refs. 22 and 24, and refs.
Snorbozyme Experimental System. The U3 snoRNA is
conserved among all eukaryotes and is required for processing
1. Major features include (i) a trimethyl guanosine cap, which
a 5? segment, which is involved in direct pre-rRNA binding and
is thought to be largely or completely single stranded; and (iii)
a highly folded 3? region, which contains the box C?/D motif
(an analog of the canonical box C/D motif in other snoRNAs;
see also above). Because the U3 segment involved in pre-
rRNA binding is known to be exposed in the mature snoRNP
particle (24), we reasoned that a ribozyme sequence placed in
this segment would be able to interact with a target molecule.
By the same reasoning, we predicted that a target sequence
embedded in the same region in a second U3 molecule would
be accessible to the ribozyme. In addition to achieving good
accessibility, targeting cleavage to this particular segment of
U3 also was expected to yield a stable byproduct, which would
facilitate characterization of the system. In this regard, the
initial cleavage products are expected to be altered by endog-
enous exonucleases (see diagrams in Figs. 2 and 4). The short
5? segment would likely be degraded completely; however, the
larger 3? product would be protected from complete degra-
dation by proteins bound to the box C?/D motif. This final
byproduct would be both stable and smaller than the initial 3?
cleavage product. These properties would make measurements
of authentic in vivo cleavage both simple and reliable. We call
the hybrid snoRNA:ribozyme molecules snorbozymes.
Snorbozyme Cleavage in Cis. Our initial tests were carried
out with a cis-acting snorbozyme (Fig. 2). To distinguish this
molecule from natural U3, which is required for cell growth,
the snorbozyme was constructed from a U3 variant that
contains a unique hybridization tag (see Fig. 1). In our scheme,
the hammerhead ribozyme sequence (52 nt) was substituted
for a U3 segment of similar size (57 nt). Nearly complete
cleavage was observed with the cis-acting snorbozyme, both in
vitro (?95%) and in vivo (?90%). Control experiments with
mutant catalytic sequences demonstrated that cleavage is
ribozyme-dependent and the shorter, exonucleolytically
trimmed 3? byproduct is only formed in living cells and not in
vitro or after cell disruption. These results show the hammer-
head ribozyme is highly active in the context of the U3
snoRNA and that a stable byproduct accumulates as predicted.
Trans-Acting Snorbozyme. We next examined cleavage ef-
ficiency in the trans configuration. Two structural variants of
U3 were used, which allowed easy discrimination of experi-
mental and endogenous U3 molecules. The target RNA
contained the same hybridization tag used for the cis snor-
bozyme. The catalytic sequence was placed into a U3 mini-
variant, which is about half the size of wild-type U3. The
intracellular locations of the target- and ribozyme-containing
molecules were examined by using fluorescent in situ hybrid-
ization microscopy (Fig. 3). The patterns obtained show that
the ribozyme and target molecules colocalize precisely to the
nucleolus, as expected. Cleavage efficiency was assessed by
using Northern blotting, for transformants expressing catalytic
and target RNAs in various combinations. These included
expression of catalytic molecules from low- and high-copy
vectors and inactive catalytic RNA as a control (Fig. 4).
Coexpression of target and catalytic molecules from low-copy
plasmids, which produce approximately equimolar RNA
amounts, resulted in ?60% cleavage of the target molecule
(lane 7). The cleavage efficiency increased to ?90% when the
catalytic RNA was expressed from a high-copy vector, i.e., in
?10-fold molar excess (lane 8). These results show that the
trans-acting U3-hammerhead snorbozyme is a stable and
powerful catalyst in a cellular environment. To our knowledge,
this is the most potent trans-acting ribozyme activity yet
observed in vivo. Interestingly, in parallel analyses done in
vitro, only ?40% of the target U3 RNA was cleaved at a
saturating condition (data not shown). We presume that the
lower activity in vitro reflects different and less well controlled
folding properties of both the catalytic and substrate RNAs.
Optimization of in Vivo Cleavage. With a view to further
optimizing the trans cleavage efficiency in vivo, we examined
the effect of altering the complementarity between the cata-
lytic and substrate RNAs (Fig. 5). In the previous experiments,
the target–catalyst pair had the potential to form 18 bp,
corresponding to 9 bp each in helices I and III (see also Fig.
1). In the new experiments, the total complementarity was
varied from 12 to 30 bp by adjusting the sequences of the target
RNA. Coexpression of the test pairs—the target RNA from
the low-copy plasmid and catalytic RNA from the high-copy
plasmid—showed that the best cleavage efficiency occurred
snorbozyme was constructed in which a hammerhead sequence (52 nt)
was inserted into the full-size U3* molecule in place of nucleotides
10–67. The Northern blot identifies RNAs expressed in yeast (in vivo)
or with T7 RNA polymerase in vitro (in vitro 1). To verify that cleavage
and subsequent trimming of the in vivo-expressed snorbozyme did not
occur during RNA isolation, the in vitro-transcribed molecules were
added to intact yeast cells and passed through the same RNA isolation
procedure (in vitro 2). ? refers to the active ribozyme; ? refers to the
defective ribozyme (see Fig. 1B). Additional control samples corre-
spond to total RNA isolated from cells expressing the U3* or empty
host vector (V). The biochemical events leading to the final products
obtained in vivo and in vitro are shown schematically. The P2 product
is not identified in the Northern blot (in vitro lanes) because of its small
size (14 nt).
Cleavage activity in the cis configuration. A cis-acting
Biochemistry: Samarsky et al.Proc. Natl. Acad. Sci. USA 96 (1999)6611
with an intermediate and asymmetric complementarity of 15
bp [9 bp in helix III and 6 bp in helix I (lane 7)]. These results
suggest that 9:6 bp complementarity provides optimal binding
and cleavage in vivo, which is in good agreement with in vitro
data. Lengthening helix I to more than 6 bp has been observed
to slow the rate of cleavage in vitro, in some cases by as much
as 100 fold. Shortening the total complementarity to less than
10–12 bp, on the other hand, also reduced the efficiency (27;
see also ref. 28). Strikingly, the cleavage efficiency was nearly
complete in the optimal case shown here.
Kinetics of the in Vivo Snorbozyme Cleavage. An important
advantage of the new in vivo system is that the ribozyme,
substrate and cleavage molecules can all be quantified, as
predicted. Product accumulation is unusual in ribozyme stud-
ies, and this property makes accurate in vivo kinetic studies
possible for the present system. Although detailed measure-
ments are beyond the scope of the present report, several
preliminary observations can be made from the results ob-
tained, for both the cis- and trans-acting ribozymes. The
efficiency of the cis-acting snorbozyme in vivo is ?90%. If the
half-life of the U3 derivatives is similar to that of natural U3,
estimated at 90 min (29), the rate of the ribozyme reaction is
deduced to be 0.08 min?1in vivo (see Appendix). In trans, the
best cleavage efficiency observed was better yet, for a ri-
bozyme–substrate complementarity of 9:6 bp and expression
of the ribozyme from a high-copy vector (?10-fold excess over
the target). Assuming an efficiency of ?95%, the rate of
reaction, kobs, is estimated at ?0.15 min?1(see Appendix).
These values are consistent with the lower range of activities
observed in vitro with trans-acting hammerhead ribozymes
(30), indicating that the maximal reaction rate, i.e., the rate of
the cleavage step, may be roughly comparable in vivo and in
vitro. When the ribozyme and the substrate have 9:9 bp of
complementarity, the reaction rate, kobs, increased from 0.01
min?1(60% cleavage), when the ribozyme was expressed from
a low-copy vector, to 0.08 min?1(90% cleavage), when the
ribozyme was expressed from a high-copy vector. This result
suggests that for the ribozyme levels evaluated, the rate of
ribozyme–substrate association is limiting in vivo. We estimate
the ribozyme concentration to be in the range of 3 ?M
(low-copy vector) to 30 ?M (high-copy vector) in the nucle-
olus, significantly more than the concentration required to
saturate the reaction in vitro (see Appendix). Although pre-
liminary in nature, this is direct evidence that ribozyme–
substrate binding is much slower in vivo than in vitro. Finally,
it is known that the rate of product release affects the overall
activityofatrans-actingribozyme invitro (31).However,inthe
intracellular location of the trans-acting ribozyme components was
examined by fluorescent in situ hybridization microscopy. Catalytic
and target RNA molecules were expressed in yeast cells from separate
low copy plasmids (see text and Fig. 4). Yeast nucleoli were identified
by using a probe specific for the box C/D U14 snoRNA. (A) Hybrid-
ization with a probe specific for the catalytic molecule. (B) Hybrid-
ization with a probe specific for an internal segment of the substrate
RNA. This probe recognizes both the full-length target molecule and
the cleaved and trimmed byproduct (see also Fig. 4). (C) Double
target and cleaved byproduct molecules. The labels in each image
correspond to the RNA molecules identified.
The U3-based snorbozymes localize to the nucleolus. The
was assessed with snorbozyme genes that encode the catalytic and
target elements in separate transcripts. The catalytic portion of the
ribozyme (46 nt) was embedded in the mini-U3 variant (U3del) in
place of nucleotides 6–60; C(?) refers to the active ribozyme, and
C(?) refers to the defective ribozyme. The target sequence (19 nt) was
placed into U3*, replacing the same wild-type segment; the target
RNA is designated T. U3 snoRNA derivatives containing substrate
and catalytic sequences were coexpressed in yeast cells from two
different low-copy plasmid vectors (V1 and V2). To determine
whether the ribozyme level was limiting, the catalytic RNA was also
expressed from a multicopy vector (V*). The diagram depicts the
principal events predicted to occur, and the Northern blot identifies
the RNA species observed. Bands marked on the Northern blot
correspond to: target RNA (T), cleaved and trimmed byproduct (P1?),
catalytic RNA (C), U14 snoRNA (?130 nt) as a control for RNA
loading, and the positions of 5.8S (?160 nt) and 5S (?120 nt) rRNAs.
Samples are given two-part names, which refer to the vector pairs
carrying the target and catalytic RNA genes, respectively. Combina-
tions of empty vectors and experimental genes are also included as
controls. The samples are arranged as follows: lanes 1 and 2, target
molecule (T) only, produced in the presence of low- and high-copy
empty vectors (V2 and V2*); lanes 3 and 4, active catalytic RNAs
produced from low- and high-copy vectors, C(?) and C(?)*, respec-
tively, in the presence of empty vector V1; lanes 5 and 6, inactive
catalytic RNA produced from low- and high-copy vectors, C(?) and
C(?)*, respectively, in the presence of empty vector V1; lanes 7 and
8, target molecules coexpressed with active catalytic RNA encoded by
low- and high-copy vectors, C(?) and C(?)*, respectively; lanes 9 and
10, target molecules coexpressed with inactive catalytic RNA encoded
by low- and high-copy vectors, C(?) and C(?)*, respectively. The
minor bands observed just above the catalytic RNAs (C) correspond
to molecules with extensions at the 3? end (22). The cleavage efficiency
for active ribozyme produced from low- and high-copy vectors is
estimated at ?60% and 90%, respectively (lanes 7 and 8; compare with
lanes 9 and 10).
Cleavage in trans. The in vivo cleavage efficiency in trans
6612 Biochemistry: Samarsky et al. Proc. Natl. Acad. Sci. USA 96 (1999)
present study, product release did not seem to be an issue in
vivo. The fact that cleaved and nontrimmed product does not
accumulate in the cells indicates that the initial cleavage
product is rapidly degraded and is not associated with the
Taken together, the results of this study allow a number of
important conclusions and predictions to be made. First, and
most significant, the findings validate the long-standing pre-
diction that gene expression in a living cell can, indeed, be
effectively blocked with a powerful artificial ribozyme. It is
clear now that a trans-acting hammerhead ribozyme can cleave
a target RNA with near-perfect efficiency in vivo. It follows
that all key reaction requirements have been satisfied, includ-
ing ribozyme stability, colocalization of the catalytic and target
RNAs, and accessibility of the relevant RNA regions.
In addition to showing that a trans-acting ribozyme can
cleave RNA nearly completely, we have established an effec-
tive strategy for delivering a ribozyme to a specific subcellular
site for targeting a specific type of RNA, in the present case the
nucleolus and an artificial nucleolar RNA. This particular
approach could be valuable for modulating levels of natural
snoRNAs and ribosomal RNAs as well as other RNAs pro-
posed to be associated with the nucleolus. The latter class
includes various mRNAs, tRNAs and splicing snRNAs, and
the RNA components of telomerase, the signal recognition
particle, MRP RNase and RNase P (ref. 32, and refs. therein).
Because artificial box C/D snoRNAs are also known to localize
to coiled bodies in mammalian cells (ref. 21, and refs. therein),
it may also be possible to target snorbozymes to RNAs in
coiled bodies, such as splicing snRNAs and the U7 snRNA
involved in histone mRNA maturation (33–38). By extension
of the concept used here, it seems reasonable to predict that
in different cellular compartments can be designed by creative
use of other stabilizing and localizing determinants.
Beyond establishing an effective approach to localization
and cleavage, the U3 snorbozyme system described has the key
advantage of yielding a stable cleavage byproduct that can be
quantified. Formation of this stable and, importantly, endog-
enously trimmed product provides the twin benefits of (i)
distinguishing authentic in vivo cleavage from cleavage that
may occur during or after RNA purification and (ii) the ability
to measure product formation quantitatively. These properties
allow detailed and accurate kinetics studies to be performed in
vivo. Our initial results indicate that the in vivo performance
of the trans-acting U3-based snorbozymes compares favorably
with that of small model ribozyme and substrate molecules
analyzed in vitro. The rate of cleavage is quite comparable,
whereas substrate binding requires high snorbozyme concen-
The present study also demonstrates the feasibility of using
yeast for in vivo ribozyme studies. Earlier results with this
organism were much less striking, with only 20–50% reduction
of target mRNA levels observed for a trans-acting ribozyme
(39, 40). The success experienced here indicates that the
powerful genetic advantages of yeast can now be exploited to
characterize and optimize ribozyme performance in vivo.
We thank Dr. Molly Fitzgerald-Hayes for providing yeast strain
MH2 and Drs. John J. Rossi, John M. Burke, and Thomas L. Mason
for critical reading of the manuscript. The research was supported by
National Institutes of Health Grant GM 19351 and National Science
Foundation Grant MCB-9419007 (D.A.S. and M.J.F), a grant from the
Medical Research Council of Canada (G.F. and R.C.), and National
Institutes of Health Grants GM 54887 and GM 57071 (E.B. and
R.H.S.). This paper is dedicated to the memory of one of the
coauthors, Dr. Robert Cedergren, a wonderful scientist, mentor, and
Parameters Used in the in Vivo Kinetics Calculations. The
efficiency of the cis-acting snorbozyme in vivo is ?90%. If the
half-life of the U3 derivatives is similar to that of natural U3,
estimated at 90 min (29), the rate of the ribozyme reaction is
deduced to be 0.08 min?1in vivo (for the formulas used in
calculations see ref. 41). In trans, the best cleavage efficiency
observed was better yet, for a ribozyme–substrate comple-
mentarity of 9:6 bp and expression of the ribozyme from a
high-copy vector (?10-fold excess over the target). Assuming
an efficiency of ?95%, the rate of reaction, kobs, is estimated
at ?0.15 min?1(see below). When the ribozyme and the
substrate have 9:9 bp of complementarity, the reaction rate,
kobs, increased from 0.01 min?1(60% cleavage) when the
bozyme cleavage activity. (A) A set of substrates able to form different
The relevant target region is on top, paired with the corresponding
ribozyme moiety on the bottom (the catalytic core of the ribozyme Rz
is omitted for simplicity). The trans-acting ribozyme complex exam-
ined earlier is the 9:9 pair. (B) Efficiency of cleavage was assessed by
using Northern blot analysis. Each substrate RNA was coproduced
with an active (?) or inactive (?) ribozyme molecule or in the
presence of an empty host vector (V). Variable amounts of truncated
substrates are evident in the negative control samples. The yield of
these molecules reflects differences in the 5? sequences of the sub-
strates and the growth stage of the cells used in the analysis (more is
present at later stages of growth). These products are presumed to
arise from random degradation of the substrate 5? end. The origin of
the truncated substrate variant in lane 17 is not clear. Accumulation
of this product might reflect a nonspecific cleavage process that
depends on the inactive ribozyme. Alternatively, cleavage might be
mediated by the ‘‘inactive’’ ribozyme, which may have low residual
activity that is magnified in this particular, high-affinity context. The
RNA species marked with triangles (lanes 7–9 and 16–18) most likely
originate from alternative folding of target molecules. Similar effects
have been observed previously with internally modified U2 and U3
RNAs (22, 26). Interestingly, these target molecules are completely
resistant to ribozyme cleavage.
Effect of flanking complementarity on trans-acting ri-
Biochemistry: Samarsky et al. Proc. Natl. Acad. Sci. USA 96 (1999)6613
ribozyme was expressed from a low-copy vector, to 0.08 min?1
(90% cleavage) when the ribozyme was expressed from a
high-copy vector. This effect suggests that for the ribozyme
levels evaluated, the rate of ribozyme-substrate association is
limiting in vivo. The ribozyme concentration is estimated to be
in the range of 3 ?M (low-copy vector) to 30 ?M (high-copy
vector) in the nucleolus, significantly higher than the concen-
tration required to saturate the reaction in vitro (30). The
volume of the yeast nucleolus is roughly 0.5 ? 10?15liters (1
?m ? 1 ?m ? 0.5 ?m; E.B., unpublished data). Assuming
ribozyme expressed from a low-copy vector (1–3 copies per
cell) occurs at a concentration approximating that of natural
U3 (normally from two gene copies), i.e., about 1,000 mole-
cules per cell, as reported by Hughes et al. (42), the ribozyme
concentration is estimated to be 3 ?M.
In Vivo Trans-Cleavage Model. The ribozyme cleavage of an
RNA target in trans and in vivo can be described by the
following kinetic model:
where E is the ribozyme; S is the substrate; P1 and P2 are the
two cleavage products; P1? is exo-trimmed P1; and & signifies
a complex. kdegis the degradation rate of natural U3 and is
assumed to be the same for all U3 derivatives involved here.
Based on estimation of the U3 half-life as 90 min (29), kdeg
should be ?0.008 min?1. k2is the rate of the cleavage step and
is assumed to be much faster than kdeg. In agreement, k2is
estimated to be 0.08 min?1in the cis-acting ribozyme exper-
iment. k3is the rate of conversion of P1 into P1? and is also
assumed to be much faster than kdeg. Indeed, k3is much faster
attained in vivo implies that the concentration of each molec-
ular species is constant over time. If we define the overall
reaction rate, kobs, as the fraction of substrate which is cleaved
per unit time, then kobs? k2? [E&S]?[S0], where [S0] is the
total concentration of substrate. From the equilibrium equa-
tions, it follows that in vivo, kobs? kdeg? [P1?]?[S0]. The value
for kobs can be calculated experimentally, because kdeg is
known, and [P1?]?[S0] ratio can be estimated from the North-
Rossi, J. J. (1995) Trends Biotechnol. 13, 301–306.
Couture, L. A. & Stinchcomb, D. T. (1996) Trends Genet. 12,
Symons, R. H. (1992) Annu. Rev. Biochem. 61, 641–671.
Pyle, A. M. (1993) Science 261, 709–714.
Bratty, J., Chartrand, P., Ferbeyre, G. & Cedergren, R. (1993)
Biochim. Biophys. Acta 1216, 345–359.
McKay, D. B. (1996) RNA 2, 395–403.
Forster, A. C. & Symons, R. H. (1987) Cell 49, 211–220.
Symons, R. H. (1997) Nucleic Acids Res. 25, 2683–2689.
Haseloff, J. & Gerlach, W. L. (1988) Nature (London) 334,
Usman, N. & Stinchcomb, D. T. (1996) in Catalytic RNA, eds.
Eckstein, F. & Lilley, D. M. J. (Springer, Berlin), pp. 243–264.
Sarver, N., Cantin, E. M., Chang, P. S., Zaia, J. A., Ladne, P. A.,
Stephens, D. A. & Rossi, J. J. (1990) Science 247, 1222–1225.
Sullenger, B. A. & Cech, T. R. (1993) Science 262, 1566–1569.
Scanlon, K. J., Ishida, H. & Kashani-Sabet, M. (1994) Proc. Natl.
Acad. Sci. USA 91, 11123–11127.
Beck, J. & Nassal, M. (1995) Nucleic Acids Res. 23, 4954–4962.
Burke, J. M. (1997) Nat. Biotechnol. 15, 414–415.
Montgomery, R. A. & Dietz, H. C. (1997) Hum. Mol. Genet. 6,
Lieber, A. & Strauss, M. (1995) Mol. Cell. Biol. 15, 540–551.
Rose, M. D., Winston, F. & Heiter, P. (1990) Methods in Yeast
Genetics: A Laboratory Course Manual (Cold Spring Harbor Lab.
Press, Plainview, NY).
Samarsky, D. A., Balakin, A. G. & Fournier, M. J. (1995) Nucleic
Acids Res. 23, 2548–2554.
Samarsky, D. A., Fournier, M. J., Singer, R. H. & Bertrand, E.
(1998) EMBO J. 17, 3747–3757.
Samarsky, D. A. & Fournier, M. J. (1998) Mol. Cell. Biol. 18,
Sikorski, R. S. & Hieter, P. (1989) Genetics 122, 19–27.
Mereau, A., Fournier, R., Gregoire, A., Mougin, A., Fabrizio, P.,
Luhrmann, R. & Branlant, C. (1997) J. Mol. Biol. 273, 552–571.
Clouet-d’Orval, B. & Uhlenbeck, O. C. (1997) Biochemistry 36,
Shuster, E. O. & Guthrie, C. (1988) Cell 55, 41–48.
Hendry, P. & McCall, M. (1996) Nucleic Acids Res. 24, 2679–
Stage-Zimmermann, T. K. & Uhlenbeck, O. C. (1998) RNA 4,
Hughes, J. M. X. & Ares, M., Jr. (1991) EMBO J. 10, 4231–4239.
Thomson, J. B., Tucsl, T. & Eckstein, F. (1996) in Catalytic RNA,
eds. Eckstein, F. & Lilley, D. M. J. (Springer, Berlin), pp.
Hertel, K. J., Herschlag, D. & Uhlenbeck, O. C. (1994) Biochem-
istry 33, 3374–3385.
Pederson, T. (1998) Nucleic Acids Res. 26, 3871–3876.
Brasch, K. & Ochs, R. L. (1992) Exp. Cell Res. 202, 211–223.
Carmo-Fonseca, M., Pepperkok, R., Carvalho, M. T. & Lamond,
A. I. (1992) J. Cell Biol. 117, 1–14.
Matera, A. G. & Ward, D. C. (1993) J. Cell Biol. 121, 715–727.
Frey, M. R. & Matera, A. G. (1995) Proc. Natl. Acad. Sci. USA
Gall, J. G., Tsvetkov, A., Wu, Z. & Murphy, C. (1995) Dev. Genet.
Wu, C. H., Murphy, C. & Gall, J. G. (1996) RNA 2, 811–823.
Ferbeyre, G., Bratty, J., Chen, H. & Cedergren, R. (1995) Gene
Ferbeyre, G., Bratty, J., Chen, H. & Cedergren, R. (1996) J. Biol.
Chem. 271, 19318–19323.
Donahue, C. P. & Fedor, M. J. (1997) RNA 3, 961–973.
Hughes, J. M. X., Konings, D. A. M. & Cesareni, G. (1987)
EMBO J. 6, 2145–2155.
6614 Biochemistry: Samarsky et al.Proc. Natl. Acad. Sci. USA 96 (1999)