therapeutic applications of ribozymes

Small Efficient Hammerhead Ribozymes
Philip Hendry, Trevor J. Lockett, and Maxine J. McCall
1. Introduction
The hammerhead ribozyme was discovered as a self-cleavmg RNA
molecule in certain plant vtroids and satellite RNAs (1). Shortly after its
conserved features were defined (2,3), the hammerhead was shown to be able
to act as a true enzyme, cleaving multiple substrates m a bimolecular reaction
(4). The self-cleaving hammerhead can be divided in a number of ways into
two, or even three, separate strands (45). The most useful form has almost
all of the conserved nucleotides on the ribozyme strand, leaving minimal
sequence requirements in the substrate strand. To be cleavable the substrate
must possess the sequence 5’ UH (H is C, U, or A), where cleavage occurs to
the 3’ stde of H (6) (Fig. 1). This particular configuration has been the para-
digm for hammerhead ribozyme design since 1988. Here we describe varia-
tions on this basic design, with the constant theme being to minimize the size
of the ribozyme.
The advantages of minimizing the size of the ribozyme are several-fold. An
obvious advantage, for ribozymes which will be used as exogenously-supplied
therapeutics, is that the cost of synthesis is reduced if the number of rtbonucle-
otides is mimmized. A second advantage, for the exogenously-supplied
ribozyme, is that delivery to cells may be aided if the ribozyme is small. An
additional, and unexpected, advantage is that some minimized ribozymes
cleave their substrates faster in vitro than the analogous, standard ribozymes
do, although it is not yet known if this advantage is carried over to produce an
enhanced effect in vivo. Two strategies for minimizing the size of the hammer-
head ribozyme are described in this chapter. The first involves shortening the
hybridizing arms of the ribozyme (which results m shortening helix I and/or
From Methods m Molecular Me&we, Vol It Therapeutic Appkahons of Rlbozymes
Edlted by K J Scanlon 0 Humana Press Inc , Totowa, NJ
1
Hendry, Lockeft, and McCall

Helix III
1
Helix I
Substrate 5’ N N N N N N N U HI7 N N N N NM 3’
3’ r;r i il Ii Ii a,&, i.l I; il Ii i N 5’
A
%
A
G
Ribozyme
G12
AG uA6
C-G s
Helix II N-N
N.N
NON
Loop II N N
NN
Fig. 1 Schematic representation of the hammerhead rtbozyme m complex with its
complementary substrate. The ribozyme forms helix I with its 5’ arm and the substrate,
helix III with its 3’ arm and the substrate, and helix II and loop 2 with the nucleotldes
joming A, and G12 In the rtbozyme, all nucleotldes, except for the conserved C3 to A,
and GU to 4, I,
may be erther ribonucleotrdes or deoxyrlbonucleotldes The site of
cleavage m the substrate is shown by the downward arrow, 3’ to HI7 (H = C, U or A)
helix III in the ribozyme-substrate complex), whtle mamtaming maximal cleav-
age rates. The second involves shortening, or completely elimmatmg, helix II
in the ribozyme. The consequences of each of these modifications on the cleav-
age activrties will be discussed.
2. Guidelines for Design

The work we describe here is based largely on observations
made using
short substrates m vitro. To directly extrapolate these observations to the use of
ribozymes m vivo IS difficult because of the unknown factors that operate
wtthm a living organism. In particular, the ribozyme may be affected by the
activities of RNA binding proteins; these proteins can either enhance or retard
substrate binding and dissociation (7-9), may stabilize the ribozyme against
degradation (10,11), or may direct the rrbozyme to specific compartments
within the cell in a manner observed for some mRNAs (12). However, it 1s
important to understand the basic processes of ribozyme cleavage and, by mak-
ing some assumptions about the intracellular envtronment experienced by an
RNA molecule, tt should be possible to project some of this m vitro experience
mto therapeutic practice.
Hammerhead Ribozymes 3
2.7. Minimiring Arm Lengths
2.1.1. Introduction
The standard form of the hammerhead ribozyme, as used by most research-
ers, is shown in Fig. 1. It has the conserved nucleotides C3 to A, and Gi2 to
Ai5 i, 4 bp m helix II with a G C bp adjacent to A9 Glz, 4 nucleotides in loop II,
and a variable number of nucleotides in the 5’ and 3’ arms which, on bmdmg to
the substrate, form helix I and helix III, respectively. The goal here is to rede-
sign this hammerhead ribozyme so that it contains as few bases as possible in
the hybridizing arms without compromising the cleavage ability or specificity
of the ribozyme.
There are a number of steps m defining the hybridizing arms of a hammer-
head ribozyme. First, the target site within the RNA of interest must be chosen.
Good targets have the UH cleavage site located within an accessible region of
the RNA, so that the ribozyme 1s readily able to hybridize to the site (see Note 1).
Second, the number of base pairs to be formed between the ribozyme and sub-
strate, or the extent of complementarity, must be chosen. The aspects to con-

sider here are catalytic activity and specificity. Ribozymes that form a large
number of base pairs with the substrate are unlikely to turnover. Smaller
ribozymes with relatively short hybridizing arms are able to turnover rapidly
and therefore have the potential for high catalytic activity. Therefore, the extent
of complementarity should be such that the ribozyme-substrate complex
formed is (relatively) stable under the conditions of the experiment, and this
typically requires 1 l-l 7 bp. For specificity, the number of bp formed between
the ribozyme and substrate should be large enough to make the target sequence
unique, but not so large that imperfectly matched substrates form stable com-
plexes (13). Statistically, about 13 nucleotides are required to uniquely define a
particular site m a mRNA pool in a mammalian cell (see Note 2). Finally, once
the target site and the number of nucleotides to be bound has been decided, the
disposition of the nucleotides about the cleavage site must be determined. By
far, the most common arrangement has been to target an equal number of nucle-
otides on either side of Hi7. A ribozyme of this design we call a symmetric
ribozyme. We have recently shown that this design does not produce the most
rapid cleavage rates m vitro.
2.1.2. The Optimum Length for Helix I
We observed that the length of helix I in the ribozyme-substrate complex
has a very profound effect on the cleavage rate constant for that complex (14).
In a number of systems we have varied the length of helix I, both by varymg
the number of nucleotides to the 3’ side of the cleavage site of the substrates
and by changing the length of the 5’ hybridizing arms of the nbozymes. All
Hendry, Lockett, and McCall
Length of Helix I (bp)
Fig 2 Dependence of rate constants on length of helix II, with helix III constant at
10 bp substrates with varying numbers of nucleotrdes to the 3’ side of the cleavage site
and 10 nucleotides on the 5’ side are cleaved by their cognate rtbozymes; I Kr RA- 101
10; A Kr RB-IO/lo; 0 TAT RB-lO/lO. Reactions condlttons; 10 mA4MgCl,, 37°C
pH 7 13 From

ref. 14
these experiments were performed under ribozyme excess conditions with the
rrbozyme-substrate complex fully formed prior to initiation of the reaction.
The cleavage rate constants for substrates with varying numbers of nucleottdes
3’ of Hi7 by three different symmetrrc (10 + 10) ribozymes are shown m
Fig, 2.
Cleavage rate constants of 2 1 -mer substrates by ribozymes with varying lengths
of hybridizing arms are given m
Table 1.
Together these data demonstrate that
the optimum length for helix I m a hammerhead rrbozyme 1s about 5 or 6 nucle-
otides whether the length of the hehx 1s hmited by the length of the substrate or
ribozyme. Ribozymes wtth longer 5’ arms are potentially limited m their acttv-
tty by slow cleavage rates.
2.1.3. The Optimum Length for Helix III
To determine whether there was an optimum length for helix III, the cleav-
age rate constants were compared for ribozyme-substrate pairs with optimum,
or near optimum, helix I lengths and either 10 or 6 bp rn helix III. The variation
in rate constant observed is shown m
Table 2.
The effect is quite small, twofold
Hammerhead Ribozymes
5
Table 1
Cleavage Rate Constants for 21-mer Substrates
by Various Cognate Ribozymes
Substrate Rlbozyme III/P
TAT S21-lo/lob TAT RA- 1 O/l Ob 10110
TAT S21-IO/10 TAT RA- 1 O/5 5110
TAT S21-lO/lO TAT RA-5110 1015

Kr s21-lO/lO Kr RA-lO/lO lO/lO
Kr S21-lO/lO Kr RA-6110 1016
k&mm
0 63”
0.09 -I- 0.01
lo+ 1
0.10
6.7
Condltlons; pH 7 13,37”C, 10 mMMgC12
ONumber of bp m hehces III and I, respectively
bThe nomenclature for the substrates and rlbozymes 1s as follows The sequences of the
substrate molecules are taken from naturally-occurrmg mRNAs and are identified by their on-
gin, The TAT series are from the TAT gene of HIV-l, and the Kr senes are from the Krdppel
gene of
Drosophzla melanogaster
Rlbozymes are denoted by an R followmg the identifying
prefix, and substrates by the letter S and a number which indicates the number of nucleotldes m
the
substrate
There are three versions of hammerhead rlbozyme used m this chapter, and they
are denoted as nbozymes A, B and C Rlbozymes A (RA) are composed solely of RNA (with
the
exceptlon of the 3’ nucleotlde), rlbozymes B (RB) possess DNA m the arms that hybridize
to the substrate (with the exception of nucleotldes 15 1 and 15 2 which remam as RNA [Fig.
l]), and rlbozymes C (RC) are the same as rlbozymes B except that their helix II and loop II are
also composed of DNA. The number of nucleotldes m the hybridlzmg arms (for nbozymes) or
on each side of Cl7 (for substrates) are added to the name, with the first number referrmg the 5’
side and the second to the 3’ side For example, TAT I&4-5/10 IS an all-RNA rlbozyme with 5
nucleotldes m its 5’ arm and 10 nucleotldes m its 3’ arm, TAT S21-lO/lO 1s a 21-nucleotlde
substrate with 10 nucleotldes each side of C,,, and TAT RA-5/10 and TAT S21-lO/lO form a

complex with 10 bp m helix III and 5 bp m helix I (III/I = 10/5)
CAt pH 8 00
Table 2
Effect of Length of Helix Ill on Cleavage Rate Constants
S17-10/6 S13-616
Ribozyme III/I k,b,/min III/I k&mm
TAT RB-lO/lO 1 O/6 1.8 f 0.4 616 0.92 f 0 1
Kr RA-lO/lO 1016 4.8 f 0 6 616 3.4 f 1.0
Kr RB-lO/lO 1016 3 2 f 0.3 616 17+04
Condltlons 10 mMMgQ, 37”C, pH 7 13
at most, with the more efficient cleavage occurring in ribozyme-substrate pairs
possessing the longer helices III. Although we have not pursued this further,
there is no reason to suspect that cleavage efficiency will be impaired rf the
ribozymes are able to form even longer helices III with their substrates. How-
6 Hendry, Lockett, and McCall
ever, there is the danger that excessrvely long hybridizing arms are able to form
folded, stable structures that prevent substrate binding.
2.1.4. Multiple Turnover
The observations above relate to reaction condttrons m which the rrbozyme
is m excess of the substrate, and the rrbozyme and substrate were preannealed
before mrtration of the reaction by addition of Mg2+. Under these condmons,
substrate binding and product drssociatton have no effect on the observed cleav-
age rate constants, However, under multiple turnover conditions, a desirable
situation for therapeutic uses, the rates of substrate binding and product disso-
ciation must also be considered. The rate of substrate association 1s difftcult to
predict for large substrates and will be largely dependent on the structure of the
RNA m that region (15). On the other hand, for a given sequence, the rate of
dissociation of the cleavage product 1s expected to consistently decrease with
increasing length (15,16). Given that the optimum length for helix I is around 5
or 6 bp, and that duplexes of this length usually drssocrate quite rapidly, It 1s the

length of helix III that is most crucial m this respect. The rate constant for
dissociation of either cleavage product in vitro may be readily esttmated by a
number of techniques (see Note 3). As a rough guide, m condmons like that
encountered m biological systems, (pH 7.0, 37°C 100 mM NaCl), the rate
constants for drssociation of the helix III (after cleavage) are hkely to approach
that observed for the cleavage step when the length of helix III is m the range
5 -9 bp, depending on the sequence.
2.1.5. Summary
The most efficient hammerhead rtbozymes have 5 or 6 nucleotrdes in their 5’
hybridizing arms, so that they may form a helix I of 5 or 6 bp in complex with
their substrates. On the other hand, the ribozymes should have a minimum of 5
or 6 nucleotides m then 3’ hybridrzmg arms, so that they may form a helix 3 of
at least 5 or 6 bp m the complex. No diminution of cleavage rate constant under
rrbozyme excess condmons 1s expected for ribozymes with longer 3’ hybrrdiz-
mg arms, but the
turnover
rate under substrate excess conditions will be
adversely affected by excessrvely long helices III.
2.2. Minimizing or Eliminating Helix II
2 2.1. Introduction
Apart from the hybridizing arms, the other region of the hammerhead
rrbozyme which may be reduced m size, or even eliminated, IS helix II and loop
II. When hehx II and loop II are completely eliminated, so that the rrbozyme
consists of two strands with free ends at the conserved nucleotides Ag and Glz,
very slow cleavage of the substrate is observed (17,18). When Ag and Gr2 are
Hammerhead Ribozymes
7
linked, either by nucleotides or non-nucleotide chains, slow cleavage is
observed with linkers containing just 13 atoms, whereas more reasonable rates
are observed with linkers containing 25 atoms (I 7). These results demonstrate

that helix II is not essential for cleavage activity. However, the nature of the
linker which replaces helix II greatly affects the cleavage activity of the small
ribozyme (17,19).
Experiments to determine rate constants for the cleavage of short substrates
by ribozymes that have a truncated helix II have shown that the number of base
pairs m helix II can be reduced to two without loss of cleavage activity, relative
to the standard ribozyme with a four base-pan helix II (20). A further reduction
to one base pan m helix II results in a lo-fold loss of activity, and ehmmation
of helix II (where A9 and Gla are connected by a 4-nucleotide loop) results m
around a loo-fold loss of activity, relative to the standard ribozyme (20-22).
The short substrates with which these experiments have been carried out were
about 13 ribonucleottdes m length. When the substrate length, increases, for
example to 2 1 nt, with a concomitant increase m the lengths of hehces I and III,
the cleavage rates of ribozymes with one or no base pairs in helix II increases
relative to that of the standard ribozyme. Furthermore longer RNA transcripts
are cleaved in vitro by these small ribozymes faster than they are cleaved by
standard ribozymes (22). It is not known if the same relative rates of cleavage
by the small and standard ribozymes also occur in viva.
In order to distinguish these small ribozymes from the standard ribozymes,
we define a ribozyme with one base pair m helix II as a mmiribozyme, and a
ribozyme with no helix II as a minizyme.
2.2.2. Miniribozymes
A miniribozyme with nucleotides of sequence 5’rGUUUUC joining A9 and
Gt2 is shown m Fig. 3A. Maximal cleavage activity is conferred on the
mmiribozyme when the single base-pair that replaces helix II is a G.C in the
orientatron shown in the figure. The optimal sequence for the loop connecting
the G and C has not been determined, but a flexible loop of 4 ribouridmes gives
good activity, whereas a loop of three ribouridines has about 70% the effi-
crency of the 4-ribouridme loop in cleaving short substrates (unpubhshed data).
A loop of sequence 5’ UUUG had a cleavage rate constant about 10% that of

the parent ribozyme (21).
The mmnibozyme may consist solely of ribonucleotides, or it may be syn-
thesized with a mixture of deoxyribonucleotides and ribonucleotides. A simple
configuration for the DNA/RNA hybrid has DNA in all positions except for
C3-A9 and Gi2-Ai5 2. The inclusion of DNA in the molecule reduces the costs
of synthesis, and may also give a degree of protection against degradation in
human serum (23) or in cells (24).
8 Hendry, Lockett, and McCall
3’ N N N N N PJ2A,u N N N N N N 5’
A cq~
A
G
& AGuA
C-G 9
u u
Miniribozyme
uu
3’ N N N N N f&Am N N N N N N 5’
A kG
A
Gu AG”A
T Gg
Minizyme
TTT
Fig. 3. Schematic representatton of a mmmbozyme and a mmizyme Formally, the
mmmbozyme has single G.C base pair replacing helix II, and the minizyme has no
helix II The sequences of nucleotides shown joining As and G12 in the miniribozyme
and mmizyme confer good cleavage activities on these molecules. The numbers of
nucleotides m the 5’ and 3’ arms of the mmnibozyme and mmizyme may be larger than
indicated here, without dtmnushing cleavage activity m vitro. All nucleotides, except

for the conserved CJ to A9 and G12 to A15 1, may be either ribonucleotides or deoxyri-
bonucleottdes.
DNA-contammg mimribozymes with d(GTTTTC) m place of helix II and
loop II, and with DNA hybridizing arms, cleave 13-mer substrates at approx
40% of the rate of analogous ribozymes with DNA in the hybridizing arms and
m helix II (Table 3). In these examples, both the mmn-ibozymes and ribozymes
form heltces I and III each of 6 base pairs m complex with the substrate. The
same mimribozymes cleave 21-mer substrates about twofold faster than the
analogous ribozymes (Table 3), and here they form a helix I and a helix III
each of 10 base pairs m complex with the longer substrate. Of more relevance
to biological applications, an all-RNA mmmbozyme cleaves an 809-nt RNA mol-
ecule in vitro much faster than does an analogous all-RNA ribozyme (Fig. 4),
Hammerhead Ribozymes
Table 3
Cleavage Rate Constants for 13.Mer and Pi-Mer Substrates
by DNA-Containing Miniribozymes (Mgttttc) and Ribozymes (FE)
Substrate
k&mm
Mgttttc RC-lo/10 Ratio Mgttttc/RC
TAT S 13 -6J6 0 175+0003 0.43 5 0 08 0.41
TAT S21-lo/10 0.9 Ik 0.2 0.45 * 0 09 20
Kr S13-6/6 0.59 f 0.09 1.6 2 0.3 0 37
Kr S21-lO/lO 3.0 zk 0.2 1.34 zk 0.08 22
Condltlons 10 mh4 MgCl,, 37”C, 50 mh4 Trls-HCl, pH 7 13
Sequences of nbozymes, with upper-case letters representing nbonucleotldes, and lower-
case letters representing deoxyribonucleotides, are as follows.
TAT Mgttttc: 5’ gtcctaggctCUGAUGAgtttttcGAAACttcctgga.
TAT RC 5’ gtcctaggctCUGAUGAgtccttttggacGAAACttcctgga
Kr Mgttttc 5’ ctccagtgtgCUGAUGAgttttcGAAACtcgcaaat
Kr RC 5’ ctccagtgtgCUGAUGAgtccttttggacGAAACtcgcaaat

and shows good activity over a much wider temperature range than does the
ribozyme (Fig. 5). Thus, at least m the examples studied to date, minirlbozymes
seem to be superior to full-size hammerheads m cleaving long transcripts in
vitro.
The relative rates at which mimribozymes cleave 13-mer and 21 -mer swb-
strates indicates that unlike full-stzed ribozymes, mimrlbozymes are not hm-
dered in their cleavage activity when forming extensive base pan-mg with the
substrate, particularly with respect to helix I. This gives an advantage to
mmiribozymes over ribozymes in that, for a target sequence of defined length,
the distribution of nucleotides on either side of the cleavage site is not restricted
in any way, whereas, for optimal cleavage by the ribozyme, the number of
nucleotides in the 5’ arm is restricted to 5 or 6 (see Subheading 2.1.2.). There-
fore the miniribozyme may be made with hybridizing arms of equal length,
which may assist turnover.
2.2.3.
Minizymes
A minizyme is a hammerhead ribozyme in which helix II and loop II have
been replaced by a short linker that contains no Watson-Crick base pairs (25’.
Minizymes have been made with linkers consisting of nucleotides (20,25,26),
or of short polymers of ethyleneglycol and phosphopropanedlol (17,19). The
rates of cleavage by mimzymes increase as the number of atoms m the chain
lmking Ag and G12 increases, with optimal rates being achieved with 25-31
atoms in the chain (27). For all-nucleotide linkers, this corresponds to 4 or 5
nucleotides conferring optimal activity.
IO
Hendry, Lockett, and McCall
60
time (hr)
Fig 4. Rates of cleavage of an 809-nucleotlde interleukm-2 transcript by
mterleukm-2 mimrlbozyme and ribozyme, at 37”C, 50 mA4 Tris-HCl, pH 8.0, 10 mA4

MgC12, with no
heat pretreatment. The minmbozyme has the sequence 5’ r(GUUUUC)
in place of helix II and loop II. Both mmiribozyme and ribozyme are made of RNA,
and have 8 nucleotides in the 5’ hybridizing arms and 6 nucleotides in the 3’ arms
Cleavage of the transcript occurs 82 nucleotldes from the 5’ end.
The first mmizymes synthesized had linkers of sequence d(TTTT) or
r(UUUU) (25). These minizymes were about loo-fold less-active than analo-
gous standard ribozymes m cleaving short substrates of about 13 nucleotides
(20-22). Like the mimrlbozymes however, symmetric (10 + 10) mmizymes
actually improve in cleavage activity when cleaving 21-mer substrates, such
that the mimzymes are typically only IO-fold less active than analogous
ribozymes (22). Thus it appears that mmizymes, like miniribozymes, and unlike
ribozymes, are not inhtbtted in their cleavage rates by the formation of long
(>6 bp) hehces I. Against a 428-nucleotide RNA target derived from the HIV-
1 TAT coding sequence, minizymes with these lmkers cleaved faster than did
the analogous ribozymes (22). Apparently, the small size and/or flexibility
enjoyed by the ribozyme variants that lack or possess substantially truncated
hehces II gives them some advantage over full sized ribozymes at the cleavage
of long RNA transcripts.
Recently, a mimzyme with the lmker of sequence d(GTTTT) has been
described (271, and it is shown in Fig. 3. The inclusion of the G at the 5’ posi-
Hammerhead Ribozymes 17
60
Temperature “C
Fig. 5 Time taken for 50% of an 809-nucleotlde mterleukm-2 transcript to be
cleaved by mterleukm-2 mmmbozyme and rlbozyme at various temperatures The
mmmbozyme and rlbozyme are described m
Fig. 4
tlon of the lmker was found to enhance the rate of cleavage relative to rates for
mmlzymes with all-pynmldme linkers. This minizyme, with d(GTTTT) linker

and DNA hybrldlzmg arms, cleaved a 15-nucleotlde synthetic substrate only
fivefold slower m vitro at 37°C than did a standard ribozyme with DNA m the
hybridizing arms. In human cells, it was as effective as the DNA-armed
ribozyme m inhibiting the production of the protein coded for by the targeted
mRNA (see
Note
4). Thus, the removal of helix II from the ribozyme did not
affect its biological actlvlty.
2.2.4. Summary
Helix II and loop II of the standard rlbozyme may be replaced by the
sequence r(GUUUUC) or d(GTTTTC) for a mmirlbozyme with good cleavage
activity, and by r(GUUUU) or d(GTTTT) for a minizyme. In v&o, mmlzymes
and minirlbozymes cleave very short substrates more slowly than do standard
rlbozymes, but m at least two cases they cleave long RNA transcripts faster.
Long hehces I and III do not impede cleavage rates, with the cleavage rate
constants lmprovmg marginally m going from 6 bp in each to 10 bp m each.
Naturally, the rates of turnover m condltlons of substrate excess would be expected
12
Hendry, Lockett, and McCall
to diminish as the length of the hybridizing hehces increased in this range. A
minizyme with a d(GTTTT) linker was as effective as an analogous rlbozyme
m human cells, even though its in vitro cleavage of a short substrate was slower.
Further testing m vlvo is needed to determine the effectiveness of
mmlrrbozymes relative to mmlzymes.
2.3.
Conclusion
This chapter has described a number of ways in which the basic structure of
the hammerhead rlbozyme can be modified. With these observations, the design
rules for the hammerhead rlbozyme have been extended and clarified. The 5’
arm of the standard ribozyme (that which forms part of helix I) should have

around five nucleotides for optimal cleavage rates The 3’ arm is not limited m
the number of nucleotldes, but should not be so long that it impedes turnover or
generates problematic mtramolecular folding. If these design rules are fol-
lowed, the standard hammerhead ribozyme is the most effective of the agents
against short substrates m vitro. If a particular application requires a partlcu-
larly long helix I, or the rlbozyme is exogenously synthesized and cost of syn-
thesis and ease of delivery is a consideration, then the best design may well be
a mmlzyme or mml-nbozyme. The cleavage of long transcripts at least m vitro
would appear to be best achieved by minizymes or miniribozymes.
3. Notes
1 Target accessiblllty 1s one of the major considerations in any nucleic-acid based
therapy. Although not considered in this chapter, a number of approaches to this
problem have been attempted mcludmg exammatlon of computer-predicted sec-
ondary structures (28), probmg the RNA of interest with ollgonucleotides that
possess mtrmslc or inducible cleavage capabilities or provide a substrate for
RNase H (29,301
or
analyzing the cleavage products generated by a population of
random-armed rlbozymes (31)
2 In a typical mammalian cell it has been estimated that there are approx 20,000
different mRNA molecules (32), and so, if their average length 1s 2 kb, there 1s a
sequence complexity of 4 x lo7 nucleotldes. Statistically then, the mmlmum num-
ber of nucleotldes reqmred to uniquely define a target sequence in a typical mam-
mallan cell 1s about 13 nucleotldes (413 - 7 x 107).
3. The measurement of the rate constant for dissoclatlon of cleavage product(s) can
be achieved by a number of methods One method would be to measure, for a
series of helix III lengths, the cleavage rate constants (k2) under rlbozyme excess
conditions and the turnover number (k,,,) under substrate excess conditions, at
the point where the two constants diverge, it IS likely that the rate constant for
substrate dlssoclatlon has become rate limiting. Another technique involves a

pulse-chase experiment where the amount of labeled product of Interest bound to
the rlbozyme 1s determined by native gel electrophoresis (33).
Hammerhead Ribozymes 13
4. The suppression of mterleukm 2 (IL2) expression in human peripheral blood
mononuclear (PBMN) cells m tissue culture experiments by a mimzyme and a
ribozyme was examined. The agents were designed to bind to 15 nucleotides at
the 5’ end of the IL2 mRNA with cleavage occurring 20 nucleotides from the
ATG start codon. The mmizyme had deoxyribonucleotides m the hybridizing
arms and in the d(GTTTT) linker joining Ag and Gt2. The nbozyme had analo-
gous DNA hybrtdizmg arms, and an RNA helix II of 4 base pairs The mimzyme,
ribozyme, and vartous control oligonucletides were transfected at concentrations
of 5, 10, and 20 l.&! into PBMN cells, for a period of 6-8 h, at which time the
cells were stimulated to express IL2 by addition of PHA The levels of IL2
secreted into the supfmatant after 16 h were measured using both bio- and ELISA
assays The mimzyme inhibited the production of interleukin-2 protein to an
extent comparable to that obtained by the DNA-armed nbozyme, and both molecules
were more effective than an inactivated mmizyme (by A,, to Gi4 substitution) with
d(GTTTT) linker, a 15-nucleotide antisense DNA, and a 15nucleotide DNA control
of nonsense sequence. None of the molecules were toxic to the cells (27)
References
1 Symons, R. H (1992) Small catalytic RNAs. Annu Rev Bzochem 61,64 l-671
2. Forster, A C. and Symons, R. H (1987) Self-cleavage of plus and mmus RNAs of
a vmrusoid and a structural model for the active sites Cell 49,2 1 l-220
3. Forster, A C and Symons, R H. (1987) Self-cleavage of virusoid RNA is per-
formed by the proposed 55-nucleotide active site. Cell 50,9-16
4 Uhlenbeck, 0. C. (1987)A small catalytic oligoribonucleotide. Nature 328,596-600.
5 Koizumi, M., Iwai, S., and Ohtsuka, E. (1988) Cleavage ofspecific sites ofRNA by
designed ribozymes. FEBS Lett. 239,285-288
6. Haseloff, J. and Gerlach, W. L (1988) Simple RNA enzymes with new and highly
specific endoribonuclease activities Nature 334, 585-591

7 Tsuchihashi, Z., Khosla, M., and Herschlag, D. (1993) Protein enhancement of
hammerhead ribozyme catalysis. Science 262,99-102
8 Bertrand, E L. and Rossi, J. J. (1994) Facilitation of hammerhead rtbozyme cataly-
sis by the nucleocapsid protein of HIV-l and the heterogeneous nuclear ribonucleo-
protein Al. EMBO J 13,2904 2912
9. Herschlag, D. M , Khosla, M., Tsuchihashi, Z., and Karpel, R. L. (1994) An RNA
chaperone activity of non-specific RNA binding proteins m hammerhead ribozyme
catalysis. EMBO J 13,2913-2924.
10. Sioud, M. (1994) Interaction between tumour necrosis factor alpha ribozyme and cellular
proteinsInvolvement in ribozyme stability and activity. J MoZ Biol. 242,619-629.
11. Sioud, M., Opstad, A., Zhao, J. Q., Levitz, R., Benham, C , and Drhca, K. (1994) In
vivo decay kinetic parameters of hammerhead nbozymes Nucleic Acids Res 22,
5571-5575
12. H111, M. A., Schedlich, L., and Gunning, P. (1994) Serum-induced signal transduc-
tion determines the peripheral location of beta-actm mRNA within the cell. J. Cell
Biol 126,1221-1229.
14 Hendry, Lockett, and McCall
13. Herschlag, D. (199 1) Imphcatrons of rlbozyme kmetlcs for targeting the cleavage
of specific RNA molecules m VIVO* more isn’t always better Proc. Nat1 Acad Scz
USA 88,692 l-6925
14 Hendry, P and McCall, M J (1996) Unexpected amsotropy in substrate cleavage
rates by asymmetric hammerhead ribozymes. Nucleic Aczds Res 24,2679-2684.
15 Young, S. and Wagner, R. W. (1991) Hybridlsatton and dissociation rates of
phosphodiester or modified ohgodeoxynucleotides with RNA at near-physiologi-
cal conditions Nucleic Aczds Res 19,2463-2470.
16 Porschke, D , Uhlenbeck, 0 C., and Martm, F. H (1973) Thermodynamtcs and
kmetics of the hehx-co11 transition of oligomers contammg GC pairs. Bzopolymers
12,1313-1335
17 Hendry, P, Moghaddam, M J., McCall, M. J , Jennings, P. A., Ebel, S , and Brown, T.
(1994) Using lmkers to investigate the spatial separation of the conserved nucleotides

Ag and G12 in the hammerhead ribozyme. Bzochzm Blophys Acta 1219,405-412.
18. Lustig, B , Lm, N H , Smith, S. M , Jermgan, R L , and Jeang, K -T (1995) A
small modified hammerhead rtbozyme and Its conformatlonal charactertsttcs
determined by mutagenesis and lattice calculation Nuclezc Aczds Res 23,353 l-3538.
19 Benseler, F , Fu, D -J , Ludwig, J , and McLaughlin, L W (1993) Hammerhead-
like molecules containing non-nucleoslde linkers are active RNA catalysts J Am
Chem. Sot. 115,8483,8484.
20 Tuschl, T. and Eckstem, F. (1993) Hammerhead nbozymes. importance of stem-
loop II for activity Proc Natl Acad Scz USA 90, 6991-6994.
21. Long, D. M and Uhlenbeck, 0. C (1994) Kinetic characterization of mtramolecu-
lar and intermolecular hammerhead RNAs with stem II deletions Proc Natl Acad
Scl USA 91,6977-698 1
22 Hendry, P., McCall, M. J , Santiago, F. S , and Jennings, P. A (1995) In vitro acttv-
ity of mmimised hammerhead ribozymes. Nucleic Acids Res. 23,3922-3927.
23 Shimayama, T , Nishikawa, F., Nishikawa, S , and Tana, K (1993) Nuclease resis-
tant chimeric rlbozymes containing deoxyribonucleotides and phosphorothioate
linkages. Nucleic Acids Res 21,2605-26 11
24 Taylor, N. R , Kaplan, B E , Swiderskt, P , Ll, H , and Rosst, J J (1992) Chtmenc
DNA-RNA hammerhead rlbozymes have enhanced m vitro catalytic efficiency and
increased stability m viva Nucleic Acids Res 20,4559-4565
25. McCall, M J , Hendry, P , and Jennings, P A (1992) Minimal sequence requue-
ments for ribozyme activity Proc Nat1 Acad Scz USA 89,5710-5414.
26 Goodchtld, J. and Kohh, V (1991) Rtbozymes that cleave an RNA sequence from
human unmunodeflciency virus. the effect of flanking sequences on rate Arch
Blochem Blophys 284,386391.
27. Sloud, M., Opstad,A , Hendry, P, Lockett, T J , Jennings, P. A , and McCall, M J.
(1997) A munmised hammerhead ribozyme with activity against mterleukm-2 m
human cells. Biochem. Blophys Res Commun 231,397-402
28. L’Hmllier, P J , Davis, S. R , and Bellamy, A R (1992) Cytoplasmic delivery of
ribozymes leads to efficient reduction in alpha-lactalbumm mRNA levels m C 1271

mouse cells EMBO J 11.4411-4418
Hammerhead Ribozymes 15
29 Godard, G , Francois, J C , Duroux, I., Asselme, U., Chassignol, M., Thuong, N ,
Helene, C., and Satsonbehmoaras, T. (1994) Photochemically and chemically
activatable antisense ohgonucleottdes. comparison ofthen reactivities towards DNA
and RNA targets. Nucleic Acids Res 22,4789-4795
30. De Young, M B., Kmcadedenker, J., Boehm, C. A., Rick, R. P., Mamone, J. A.,
McSwiggen, J A , and Graham, R. M. (1994) Functional characterization of
ribozymes expressed using U 1 and T7 vectors for the intracellular cleavage of ANF
mRNA. Bzochemzstry 33, 12,127-12,138.
31 Lieber, A. and Strauss, M (1995) Selection of efticient cleavage sites m target
RNAs by using a ribozyme expression library Mol Cell Bzol 15, 540-55 1
32 Alberts, B , Bray, D., Lewis, J , Raff, M., Roberts, K., and Watson, J D (1994)
Molecular Bzology of the Cell, 3rd ed., Garland, New York, p 369
33 Hertel, K J , Herschlag, D , and Uhlenbeck, 0. C. (1994) A kinetic and thermo-
dynamic framework for the hammerhead ribozyme reaction. Blochemzstry 33,
3374-3385.
2
The Hairpin Ribozyme
Discovery and Development for Gene Therapy
Arnold Hampel
1. Introduction
1.1. Discovery of the Hairpin Ribozyme
The mmimum catalytic center of (-)sTRSV was identified, biochemically
characterized and named the hairpin ribozyme (1,2). Following Initial identifi-
cation of the mmtmum catalytic sequence, we identified a trans-catalytic reac-
tion and biochemically characterized this reaction. The mmimum sequence for
catalytic activity was found to be a 50 nucleotide (nt) ribozyme and a 14 nt
substrate. This was determined by modeling and cut-down experiments. A 50 nt

catalytic RNA could cleave a 14 nt substrate RNA and the reaction occurred in
trans. The reaction proceeded without depletion of the 50 nt RNA component,
and therefore was catalytic. It had true Michaelis-Menten kmetics allowmg
determmation of KM, bat, energy of activation, Mg2+ dependence, and the pH
optima.
1.2. The Two-Dimensional Structure of the Hairpin Ribozyme
We determined the two-dimensional structure of the hairpin by making an
extensive collection of site-specific mutants in both the rtbozyme and the sub-
strate. By comparison of catalytic activity for the native sequence with mutants
containing both mismatch and alternate base pairs for regions of predicted base
pairing, we determined base pairs to identify four hehces and five loops for the
ribozyme-substrate complex. The overall structure was hanpin-like, so I named
it the hairpin ribozyme (2-4) (Fig. 1).
From
Methods rn Molecular Medmne,
Vol
11 Jherepeuttc Applmtlons of Rtbozymes
Edlted by K J Scanlon 0 Humana Press Inc , Totowa, NJ
17
78
Site of cleavage
Hampel
Catalytic RNA
Loop 4
3’
50
A
Loop 5
RMA
3’

U GUAU&WAC U
GUG CUGG
. . . .
U
l ee l eee l eme l ee**ee
CAC A A GACC A -A
m m Sbase 1
G CAZIAG
20
AGA
Loop 3 Helix 4 Loop 2 Helix 3
Helix 2 Loop 1 Helix 1
Fig. 1. The (-)sTRSV hairpin ribozyme. The ribozyme-substrate complex consisted
of four helices and five loops as shown. Base pairing occurred between the ribozyme
and substrate in helices 1 and 2. The length of helix 1 can be optimized for each sub-
strate cleaved and in general varied between 6 and 12 bp. The substrate had a BN*GUC
requirement where B is G, C, or U and cannot be A. The corresponding V nucleotide in
the ribozyme is C,G, or A.
1.3. Development of the Hairpin Ribozyme for Gene Therapy
Following its discovery, biochemical characterization and determination of
two-dimensional structure, we engineered the hairpin ribozyme to cleave het-
erologous substrate RNAs (2,3). This led to our development of the hairpin
ribozyme system for human gene therapy and other applications for down-
regulation of gene expression (5-8). Targeting rules for cleavage of heterolo-
gous substrates were determined (3). The substrate had a sequence requirement
of BN*GUC where the * is the site of cleavage (Fig. 1). The nucleotide B is G,
U, or C but not A.
Sequence searches were done for a number of systems, including HIV- 1, to
identify sequences containing BN*GUC for use as possible target sites (2,5).
Using HIV-I as an example, ribozymes were made to a number of potential

targets and cleavage efficiency of the ribozymes to these targets determined.
Optimization was done by varying the length of helix 1 to identify its optimal
length for maximum catalytic efficiency (kJKM). In general the optimal length
of helix 1 varied between 6-12 bp with 8 bp being a useful first approximation.
The ribozyme was improved by making number of sequence change in
regions of the ribozyme containing nonessential nucleotides. These greatly
improved catalytic activity for certain targets. By replacing loop three with a
specific tetraloop sequence, catalytic efficiency improved by as much as 30
times (6). Those ribozymes that had the best catalytic efficiency were used for
gene therapy in tissue culture cells. Hairpin ribozymes that we developed by
these methods have been approved by the RAC (Recombinant DNA Advisory
Committee) for human use and will soon be tested as potential AIDS therapeu-
Hairpin Ribozyme
19
tics in humans by Dr. Flossie Wong-Staal at the University of California-San
Diego (8).
2. Materials
1. Synthetic DNA oligonucleotides coding for ribozymes and substrates which in-
clude the T7 promoter sequence followed by the initiating nucleotides CCC (for
ribozyme templates) or CGC (for substrate templates). When transcribed, the fol-
lowing DNA templates will give the RNA sequences of the native (-)sTRSV
hairpin ribozyme found in Fig. 1.
a. Ribozyme DNA Template Sequence, Start of transcription: 3’ATTATGCTG
AGTGATAT”CCCTTTGTCTCTTCAGTTGGTCTCTTTGTGTGCAACA
CCATATAATGGACCATS
b. Substrate DNA Template Sequence. Start of transcription: 3’ATTATGCTGA
GTGATAT’YZGCACTGTCAGGACAAA
2. An oliogodeoxynucleotide complementary to the T7 promoter sequence. This
sequence can be used generally for transcribing all ribozyme or all substrate
sequences. The sequence of the T? complements (see Note 1) were:

a. For Transcribing Ribozymes: S’TAATACGACTCACTATAGGG3’
b. For Transcribing Substrates: S’TAATACGACTCACTATAGCG3’
3. Transcription buffer 2X, consisting of 8% polyethylene glycol3000,0.2% Triton
X- 100, 2 mM spermidine, 10 mM DTT, 80 mMTris, pH 8.0, and 12 mM MgC12.
Store frozen.
4. 10 m&Y NTP solution, composed of ATP, CTP, GTP, and UTP each at 10 mA4
stored frozen at -20°C in aliquots. Repeated freeze/thaw should be avoided.
5. Labeled nucleotide IZX~~P-CTP (10 pCi/pL, 3000 Ci/mmol) from ICN (Duarte,
CA) for transcript labeling. This reagent was stored frozen and used within 1 wk
of the reference date. When in use, keep the reagent on ice. Repeated freeze/thaw
and holding at room temperature will inactivate the sample.
6. T7 RNA polymerase 20 U/pL from Ambion (Austin, Texas [cat. no. 20841) (see
Note 2).
7. Gel extraction buffer for extracting RNA transcripts from gels, consisting of OSM
ammonium acetate, 2 mMNa* EDTA, and 0.5 mg/mL SDS.
8. Glycogen 20 mg/mL Boehringer-Mannheim (Germany, cat. no. 901393).
9. Cleavage buffer 4X consisting of 8 mM spermidine, 48 mA4 MgC&, and 160 mM
Tris, pH 7.5.
10. The PC-compatible program Tablecurve 2D v.3 for Windows 3.2 from Jandel
Scientific Software Co., San Rafael, CA.
3. Methods
3.1. Preparation of Ribozyme and Substrate RNA
1. DNA oligonucleotides were chemically synthesized by standard methods, puri-
fied by reverse phase HPLC (see Note 3), dried, dissolved in distilled water,
annealed with the complementary T7 promoter oligonucleotide at a 1: 1 molar
ratio in 20 mMTris, pH 7.5, heated to 9O”C, and cooled. Store frozen.
20 Hampel
2. T7 transcriptlon was at 37°C in 50 pL of 400 ng DNA, 40 mMTris, pH 8 0,6 mA4
MgCl*, 5 mMDTT, 1 mM spermldme, 4% PEG-3000,O 1% Trlton X- 100, 1 mM
NTP, 20 mC1 [a3*P]CTP, 40 U RNasm (Promega), and 20 U T7 RNA polymerase

(Amblon) for 3 h Addition of 2 5 pL 10 mA4 GTP at 30 mm improved transcrip-
tlon Two units of DNase (Amblon) were added after 3 h (see Note 4) and the
incubation continued at 37’C for one additional hour (see Note 5) RNA was
ethanol precipitated, resuspended m 8 pL H20, and denatured by adding 6 &
98% formamide dye and heated to 90°C for 2 min The sample was snap-cooled
on ice and separated on 10% PAGE-8M urea gel for rlbozymes and 15% PAGE-
8M urea gel for substrates (see Note 6)
3 The desired bands (see Note 7) were exclzed, macerated m 400 ,uL gel extraction
buffer, shaken for 1 h and centrifuged at 14,OOOg for 10 mm The supernatant was
ethanol precipitated with 1 pg glycogen as carrier, the pellet washed 2x with 70%
ethanol 2 mMNa2EDTA (the SDS must be removed) and quantitated for radloac-
tivlty by Cherenkov counting Plcomoles (pmoles) of C nt m the sample was
calculated*
pmoles of C m sample = $1 in isolated transcript
x [pmoles CTP added/(@ CTP added x decay factor)]
(1)
Pmoles of C m the sample was converted to pmoles RNA using the number of
C residues in a given RNA transcript:
pmoles of RNA transcript
= pmoles C in sample/(moles C/mole of RNA)
(2)
Yields varied with different templates, but this method produced up to 400 pmoles
or5pgofRNA
3.2. Cleavage and Determination of Catalytic Parameters
1. In order to carry out a RNA cleavage reaction, a first approximation of ribozyme
and substrate concentrations were used Initial concentrations of substrate (400 n&f)
and rrbozyme (80 nM) give a 5:l substrate:nbozyme ratio. These solutions were
each heated to 90°C for 2 mm and cooled on ice Just prior to use A typical cleav-
age reaction consisted of
2 pL rlbozyme

2 @, substrate
2 pIa dH20
2 pL 4X cleavage buffer
Control reactlons mcluded mcubatlon of substrate m cleavage buffer wlthout
ribozyme and mcubatlon of rlbozyme m cleavage buffer without substrate
2 The above solutions were incubated for 1 h at 37’C, formamide dye mix (6 pL)
added to stop the reactions, and they were placed on ice. The samples were heated
to 9O“C, snap-cooled and electrophoresed on 15% PAGE/8 M urea gels
3 The gel was covered with Saran Wrap, aligned with carefully marked autoradiog-
raphy film, stored m an autoradlographic cassette, and exposed at -80°C over-
Harrpm Ribozyme
21
mght Film development revealed the location and intensity of the radioactive
substrate and product bands
4 After realignment of the film and gel, the bands were exclzed and counted (see
Notes 8 and 9)
5 The fraction of substrate converted to product was calculated as follows
fraction cleaved = [cpm (P)/[cpm (S) + cpm (P)]
(3)
where P was product and S was substrate remaining Since the ratio of rlbozyme to
substrate was I:5 m this example, a cleavage fraction of more than 0 20 was nec-
essary for rlbozyme turnover to occur If no turnover was observed the reaction
was single event It was then necessary to alter the ribozyme sequence to improve
its cleavage efficiency, i.e , the ribozyme arm hybrldlzmg to the substrate to form
helix 1 may be excessively long such that product departure was very slow
6 Based on the results from the initial cleavage reaction, a time-course experiment
was designed to verify multiple turnover and give a first approxlmatlon for the
range of kinetic constants.
7 The results of the time-course experiment could be used to calculate an approxi-
mate KM and k,,, by integrating the Mlchaelis-Menten equation These approxl-

matlons could be used to design condltlons for formal kinetic analysis
8 Determination of k,,, and KM values for the ribozyme and substrate were done
with multiple turnover reactions to give true Michaelis-Menten parameters (9)
(see Note 10) Initial velocity was determined using a range of substrate concen-
trations around the projected KM,
with ribozyme concentration fixed and limit-
mg. Multiple time points were done for each rlbozyme/substrate combmatlon to
show linearity of the reaction and determine initial velocity For each kinetic
assay the percent cleavable substrate was corrected for uncleavable substrate
For example, by these T7 transcnptlon methods, typically 80-90% of the sub-
strate was cleavable
9 The values for substrate concentration (x) and imtlal velocity b) were fit to the
Michaelis-Menten equation using a curve fitting program such as the Jandel SC]-
entific Tablecurve:
y = A . xl(B + x)
(4),
where A is V,,, = [Rz]
l
k,,, and B is KM The best fit curve gives a statistical
estimate (r2) of how well the data are defined by the equation, shows a 90%
confidence interval and gives values for A and B (see Note 11)
10 Catalytic efficiency 1s calculated by the ratio of k,,JKM This 1s a useful catalytic
parameter
3.3. Mutagenesis
3.3.1. Determination of the Two-Dimensional Structure
of the Hairpin Rlbozyme
1. The methodology has two basic components. postulation and testmg of the model,
2 Postulation of the model was initially done by computer modeling using methods
such as described by Tabler and Sczakiel (10). RNA foldmg programs gave the
22 Hampel

preferred mmimal energy structures that provided a starting point for the expen-
mental determmatton of the RNA secondary structure.
3 To test the model based on minimum energy predtctrons, it was necessary to
directly mutagemze bases, and determine the effect of these changes on catalyttc
acttvtty. This was done by making changes m the DNA template for the substrate
and rlbozyme to provide extensive mformatlon without the necessity of selec-
tion. Mutagenesis data can both ehmmate and suggest a model from a group of
minimum energy structures predicted by computer.
4 Mutagenesis was designed to locate helmal and loop structures by identifying
Watson-Crick base pans The approach was to mutate predicted base pans to
both mismatch and alternattve predicted base pans Each mutation was tested
for catalytic acttvtty in compartson to the native nonmutated rtbozyme to com-
pare the effects of the mutation on catalytic activity Four helical regions and
5 loops m the hatrpm ribozyme-substrate complex were identified by this
method (Fig. 1)
5 The two-dtmenslonal structure of the hairpin rtbozyme, as determined by these
methods, consists of a basic catalyttc unit composed of two hehces (hehces 3 and
4) and three loops. Two of the loops are oppostte each other, loops 2 and 4, while
loop 3 IS at the end of a hatrpm stem Helix 3 has 4 bp and helix 4 has 3 bp When
the substrate binds to the ribozyme, two addmonal helices and two additional
loops form. The two hehces are hehces 1 and 2 Helix 2 has a maximum of 4 bp,
with 4 bp bemg optimal. Helix 1 is of variable length The two loops formed
between the rtbozyme and substrate are loops 1 and 5 Loop 1 1s m the rtbozyme
sequence and 1s charactertzed by 4 nt. The substrate loop is also 4 nt and has the
sequence A*GUC Cleavage of substrate 1s at the *.
3.4. Applications to Gene Therapy
3.4.1. Determhation of Substrate Targeting Rules
1 By using mutagenests methods described above, the targeting rules for the sub-
strate were determined The substrate, for the (-)sTRSV based halt-pin rtbozyme,
must have a BN*GUC sequence where * IS the site of cleavage (3) The GUC was

the preferred sequence with the G base absolutely required. The UC bases were
preferred as there was very low cleavage acttvtty (k,,,&) when they are changed
For example changing the C results m a 12 x reduction m cleavage efficiency for A
or G and 30 x reduction for U m thts posttion. The B nucleottde can be a C, G, or U
but cannot be an A for trans cleavage (see Note 12)
3.4.2. identification of Target Sites
1, Using appropriate search methods the DNA sequence databases were searched to
identify BN*GUC sequences m the selected RNA transcripts
2 If the target gene had potential heterogeneity it was necessary to identify regions
of high homology for the target site. By using sequence comparison methods
such as BLAST, these regions of homology were identified (11)
Hairpin Ribozyme
23
3. In addition to having the B*NGUC requirement and being m conserved sequence
regions, the target sttes should also be in regions of mimmal interference from
RNA structure or bound proteins. These preferred regions are often near the 5’
cap, near the 3’ termmus and near sphce acceptor sites Sequences near splice
donor sites m general have a high degree of secondary structure and should be
avoided.
3.4.3. Design of the Ribozyme
1. With identification of an appropriate target sequence, a hairpin rtbozyme was
designed to cleave this target sequence The conventional hairpin rtbozyme will
have the features shown m Fig. 1
2 The ribozyme component of helix 2 was designed to base pair with standard G.C
and A:U base pairs to the four nt upstream of the N*GUC of the target sequence.
We have not seen any advantage m usmg wobble base pairs
3 The ribozyme component of helix 1 was designed to base pair with the 10 nt
downstream of the N*GUC sequence. Again, standard Watson-Crick base pairs
were used We have not seen any obvtous constramts m the sequence of helix 1
This was notably true m the first base pair followmg the *GUC All four nt were

active m this position. Note that a 10 bp helix 1 was designed, but a range of helix
1 nbozyme-substrate lengths were assayed m order to optimize the length of
helix 1. By making a ribozyme capable of forming a range of lengths of helix 1
up to 10 bp, the range of hehx 1 lengths could be tested with a single ribozyme
4. The ribozyme DNA template sequence:
a Ribozyme DNA Template Sequence Start of transcription. 3’ATTATGC
TGAGTGATAT”CCCNNNNNNNNNNTCTTNNNNTGG
TCTC TTTGTGTGCAACACCATATAATGGACCATS
Transcription was done as above to make the desired hairpin rtbozyme We have
tested activity of a number of ribozymes both with and without the S’GGG
sequence prior to the sequence of the ribozyme, and have found the addition of
GGG to the 5’ ribozyme termmus has no deleterious effect on activity
3.4.4. Design of the Substrate
1 A prmciple m design of the substrate for in vitro testing was to keep it short, Long
m vitro substrates were not useful due to extensive interfering secondary struc-
ture whrch occurs m long RNA m the absence of cellular factors
Secondary struc-
ture m the substrate introduces variables which mterfere wtth the obJecttve of
determination of m vitro catalytic activity
2. In order for the hairpin ribozyme to be effective, the target sequence must be
exposed. The use of long helix 1 lengths to compete out secondary structure is
not suggested because the result is essentially an antisense effect with very low
turnover rates. It is difficult to predict if a structure will be exposed m vlvo or not
This needs to be determined experimentally by testing for in vivo activity.
3 The substrate DNA template sequence designed to produce a full length substrate
with 10 bp helix 1 was as follows.
24 Hampel
a Substrate DNA Template Sequence Start of transcription 3’ATTATGC
TGAGTGATAT”GCGNNNNNCTGNNNNNNNNNN5’
4 Transcription was carrted out as described above

5. Substrate RNA bands representing a range of helix 1 lengths were isolated from the
gel In addition to the full-length transcript, a range of transcripts representing sub-
strate with sequentially one less nucleottde on the 3’ termmus was obtained from
the transcription ladder The bands m the ladder were due to premature termmatton
of transcription which was common from single stranded templates. Each
sequentially smaller transcript represented one less nucleottde on the 3’ termmus
confirmed by direct RNA sequencing (I2,13), and thus serially shortened helix 1
lengths were produced m the formed ribozyme-substrate complex These sub-
strate bands were isolated and correspond to helix 1 lengths of 5-l 0 bp
3.4.4. Optimization of Helix 1 Length
1. Using the range of substrates generated from transcription, cleavage assays were
carried out to determine optimal helix 1 length. The correct way to do the analy-
SIS was to determme catalyttc efficiency (kcat/I&) with multiple turnover reac-
tions for each helix 1 length This entailed much mltial effort A quicker
approximatton was to carry out the cleavage assay with multiple turnover using
high substrate concentrattons, 1 e , higher than the initial estimate of I& From
the mtttal velocity of this reaction a pseudo first order rate constant kobs was
calculated. The high substrate concentration used caused the {Khl + [5’j} term m
the denommator of the Michaelis-Menten equation to approach [5J and conse-
quently the velocity of the reactton approached Vmax Thus at high substrate
concentrations, kobs approached kcat This method IS satrsfactory for an uuttal
screen of helix 1 lengths
3.4 5. Tetraloop Modification of the Hairpin Rlbozyme
1 The tetraloop addition to the hairpin rtbozyme was a modtfication which has
greatly improved catalyttc effctency for certain hairpin nbozymes (6) Loop 3 of
the native hairpin ribozyme was replaced with the GGAC(UUCG)GUCC
tetraloop sequence (Fig. 2)
2 The tetraloop forms a very stable stem loop structure (24) and thus likely stabi-
lizes the rtbozyme itself against thermal denaturation Depending on the specific
target sequence to be cleaved by the ribozyme, the tetraloop addition to the

ribozyme either had no effect on activity, decreased activity slightly or, activity
increased The change m the catalytic parameters of the HIV- 1 pol specific hair-
pin rtbozyme was most signrficant When the tetraloop additton was made to the
basic hairpin rtbozyme, the KM decreased from 42 nMto 6.7 nA4and k,,, increased
from 0 2/min to 0 5/mm to give an overall mcrease m catalytic efficiency (k,,J
Kh?) of 15 times (6)
3 The halt-pin tetraloop ribozyme was designed to cleave specific target sequences
followmg the same targeting rules for helix 1 and 2 as for the conventional hair-
pm ribozyme described above.
Hairpin Rlbozyme
site
of cleavage
25
Catalytic RNA
Loop 4
i
G
i
3’
c
GUCCGUG
GuAuAuuAc u
CUGG
. . . . . . .
. . . .
. . . .
l e.**ee
U
CAGGCAC A
AGACCAmA

~ &j&JSbasel
U
CAAAG
AGA
Loop 3 Helix 4 Loop 2 Helix 3 Helix 2 Loop 1 Helix 1
Fig. 2. Hairpin tetraloop ribozyme. Loop 3 of the native hairpin nbozyme structure
was replaced with a GGAC(UUCG)GUCC tetraloop sequence
4. The DNA template correspondmg to the tetraloop hairpin ribozyme was as
follows
a DNA template for the tetraloop haupm ribozymes Start of transcription:
3’ATTATGCTGAGTGATAT”CCCNNNNNNNNNNTCTTNNNNTGGTCT
CTTTGTGTGCCTGAAGCCAGGCACCATATAATGGACCAT 5’
5. Synthesis of DNA, transcrtption, and determination of catalytic activity were
done as above for the conventional hairpin ribozyme
4. Notes
1. The DNA single stranded templates began transcription with a GGG or GCG
sequence at the 5’ end of the desired transcript, which improved T7 RNA poly-
merase mrtiation (151 By extendmg the complementary strand of the T7 pro-
moter 3 nt past the mitiation site, transcription levels were also improved
2 A variety of highly purified, high concentration T7 RNA polymerases are cur-
rently commercially available. We have not obtained equal success from all of
the preparations, however, Ambion T7 RNA polymerase produced fewer extrane-
ous transcripts (both shorter and longer) which made band identification caster
3. The method of oligodeoxynucleotide purification used affects RNA transcription
yield Our laboratory used HPLC purification by a Aquapore RP-300 7 micron
Brownlee reverse phase column eluted with a gradient of acetonitrile/triethyl-
ammonium acetate This method was fast, and the volatile solvents used for elution
were easily removed m the spin vat to give a highly purified salt-free sample.
4 A vartety of reaction times were tested for their effect on yield Although presum-
ably longer reaction times should result m higher yields, we did not observed this

to be the case
5 DNase I was used to terminate the transcription reaction because we found small
quantities of ohgodeoxynucleotides copuritied with the RNA transcripts during
gel electrophoresis. Since the DNA was complementary to the RNA sequences,
this could interfere with ribozyme activity.
6. Transcription reactions can either be loaded directly into large capacity lanes on
polyacrylamide gels or ethanol precipitated The former procedure was more