Chemical synthesis of an artificially branched hairpin ribozyme variant with RNA cleavage activity

Article abstract of DOI:10.1016/j.tet.2004.07.055

Due to the development in the field of RNA synthesis over the past decade of years, preparation of RNA oligonucleotides longer than 50 nucleotides is possible today. In this report, we describe the chemical preparation of a branched RNA molecule with RNA cleavage activity consisting of 81 nucleotides. It is derived from the hairpin ribozyme, a small catalytic RNA occurring in nature. The hairpin ribozyme consists of two separately folded domains (loop A and loop B domain), which can be joined in a number of different ways without loss of activity. In the construct presented here, 2′-deoxy-N4-(6- hydroxyhexyl)-5-methylcytidine was introduced to connect the loop B domain with the loop A domain via an artificial branch. The synthesized branched RNA is able to catalyze the cleavage of a number of suitable substrates. Compared with the corresponding non-branched reverse-joined ribozyme it cleaves its substrates only 5-fold slower. Surprisingly, no ligation activity could be detected.

Full text of DOI:10.1016/j.tet.2004.07.055

Tetrahedron 60 (2004) 9273–9281
Chemical synthesis of an artificially branched hairpin ribozyme
variant with RNA cleavage activity
a
Sergei A. Ivanov, Eugene M. Volkov, Tatiana S. Oretskaya and Sabine Muller
b
b
*
¨
a
¨
¨
¨
Ruhr-Universitat Bochum, Fakultat Chemie, Universitatsstrasse 150, 44780 Bochum, Germany
bDepartment of Chemistry and Belozersky Institute of Physico-Chemical Biology, Moscow State University, Vorob’evy gory,
Moscow 119899, Russian Federation
Received 22 April 2004; revised 12 July 2004; accepted 22 July 2004
Available online 25 August 2004
Abstract—Due to the development in the field of RNA synthesis over the past decade of years, preparation of RNA oligonucleotides longer
than 50 nucleotides is possible today. In this report, we describe the chemical preparation of a branched RNA molecule with RNA cleavage
activity consisting of 81 nucleotides. It is derived from the hairpin ribozyme, a small catalytic RNA occurring in nature. The hairpin
ribozyme consists of two separately folded domains0(loop A and loop B domain), which can be joined in a number of different ways without
loss of activity. In the construct presented here, 2 -deoxy-N4-(6-hydroxyhexyl)-5-methylcytidine was introduced to connect the loop B
domain with the loop A domain via an artificial branch. The synthesized branched RNA is able to catalyze the cleavage of a number of
suitable substrates. Compared with the corresponding non-branched reverse-joined ribozyme it cleaves its substrates only 5-fold slower.
Surprisingly, no ligation activity could be detected.
q 2004 Elsevier Ltd. All rights reserved.
1. Introduction
the desired sequence exchange reaction (lower part of
Fig. 1) turned out being less efficient (K. Bossmann,
S. Mu
¨ller, unpublished results). The low efficiency may be
The past few decades of research have revealed that RNA
can catalyze an amazing number of chemical reactions. The
structure of a number of ribozymes is known and
mechanistic details of RNA catalysis have been elucidated.
We can now begin to understand ribozymes well enough to
turn them into useful tools. The development of ribozymes
for site-directed alteration of an RNA sequence has been a
major goal in our laboratory. We have studied the ability of
the hairpin ribozyme1,2 to carry out RNA cleavage and
ligation and have combined two hairpin ribozymes into one
molecule henceforth dubbed twin ribozyme.3,4 A twin
ribozyme that was engineered by combination of two
conventional hairpin ribozymes mediates the removal of a
specific fragment from the parent RNA, followed by
ligation of a separately added repair oligonucleotide into
the resulting gap with 30% yield.4
attributed to the specific structure of the reverse-joined
hairpin ribozyme unit in the twin ribozyme. Due to a single
stranded linker connecting the loop A domain with the loop
B domain (Figs. 1 and 2) the length of the substrate
ribozyme duplex on the left of the cleavage site is limited to
six base pairs. Upon cleavage the resulting product can
easily dissociate making ligation at this site less efficient.
We, therefore, re-designed the structure of the so far used
reverse-joined hairpin ribozyme HP-RJWTC6 into
HP-RJTLB that consists of the loop B domain linked to
the loop A domain via a non-natural branch (Fig. 2). In
comparison to HP-RJWTC6 that due to its specific structure
can form only six base pairs with the 30-end of its substrate,
the branched ribozyme HP-RJTLB is able to bind substrates
of any length just by adaptation of the binding arm lengths.
Thus, fragments can be more strongly bound and conse-
quently ligated. In this paper, we describe the synthesis and
functional characterization of the artificially branched
ribozyme.
In addition to the conventional hairpin ribozyme, we also
used reverse-joined hairpin ribozymes,5,6 for twin ribozyme
engineering (Fig. 1).3
Even though these twin ribozymes catalyze the cleavage of
an RNA substrate at two specific sites (upper part of Fig. 1),
2. Results and discussion
2.1. Ribozyme design
Keywords: RNA synthesis; Branched RNA; Reverse-joined hairpin
ribozyme.
In the branched ribozyme HP-RJTLB the loop A domain
and the loop B domain are joined via a cytidine analogue
* Corresponding author. Tel.: C49-234-3227034; fax: C49-234-3214783;
e-mail: sabine.w.mueller@rub.de
0040–4020/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2004.07.055

9274
S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
2.2. Synthesis of the branched ribozyme
Our first attempts towards the synthesis of a branched RNA
employed the strategy shown in Figure 3, route A. The
phosphoramidite building block of thymidine was trans-
formed into the corresponding 4-triazolide 2 following a
strategy first introduced by Sung et al.9 The further strategy
involved standard RNA synthesis followed by treatment of
the protected RNA on the polymer with hexanolamine to
substitute the triazole moiety and to introduce the required
functionality for branching.
To evaluate this strategy, we first prepared a short model
oligoribonucleotide SI-SLA5 (for sequence details refer to
Section 4). After assembly of eight building blocks
including the modified monomer 2, the terminal 50-OH
group was acetylated and the polymer bound protected
oligomer was treated with aminohexanol/CH3CN to substi-
tute the triazole residue at the branching nucleotide. Then,
synthesis was continued with stepwise coupling of five more
adenosine units at the OH group of the amino hexanol linker
to yield the branched oligoribonucleotide. Analysis of SI-
SLA5 by enzymatic digestion, however, showed that the
oligomer had not the expected nucleoside composition (data
not shown). While less cytidines than expected could be
identified, the amount of detected adenosine residues was to
high. The most plausible explanation for this observation is
a side reaction during aminohexanol treatment. Position C-4
of cytidine residues is easy accessible for substitution
reactions.10 Thus, also the natural phenoxyacetyl protected
cytidine residue within the synthesized RNA chain may
have undergone partial transamination during hexanolamine
treatment, such that additional adenosine residues could be
attached also at this site. An alternative explanation might
be partial removal of the acetyl group at the 50-terminus of
the first synthesised chain during aminohexanol treatment.
This however, could be ruled out by further experiments
(see below).
Figure 1. Schematic presentation of site-directed alteration of RNA
sequence by an engineered twin ribozyme involving a reverse-joined
hairpin ribozyme unit. The twin ribozyme is designed to first cleave the
input substrate at two positions to remove the sequence fragment marked in
light grey (top). In the second step, a new RNA fragment (dark grey) is
bound in the gap left behind and ligated to the remaining fragments of the
substrate RNA. For validity of this concept see Ref. 4.
In order to circumvent this problem, we applied an
alternative strategy focussing on the synthesis of an
aminohexanol modified monomer and its incorporation
into RNA. For the synthesis of the modified monomer
building block 3 (Fig. 3, route B) we basically followed the
methodology introduced by Horn and co-workers for
synthesis of comb-type oligodeoxyribonucleotides.11
Briefly, the 50- and 30-hydroxyl groups of the sugar moiety
were protected with dimethoxytrityl- and tert-butyldi-
methylsilyl functionalities, respectively, following standard
procedures.12–14 The compound was transferred into the
4-triazolo derivative and the triazole group was substituted
with hexanolamine. The alcohol group at the linker was
converted into a levulinic acid ester followed by removal of
the 30-O-TBDMS group. Finally the phosphoramidite was
prepared following the standard protocol.15,16 As demon-
strated before, the levulinic acid residue is stable under the
conditions of oligonucleotide synthesis.13,17,18 Vice versa, it
can be quantitatively removed with hydrazine hydrate in
pyridine/acetic acid, leaving the 20-O-tert-butyldimethyl-
silyl protecting group and the succinyl linker attaching the
oligonucleotide to the solid support intact.19,20 However, the
routinely used and easily removable phenoxyacetyl (PAC)
groups for N-protection21 turned out to be too labile under
serving as branch point (Fig. 2). The cytidine analogue is
modified at C-4 with a linker providing an OH group to be
used for assembly of a third RNA chain. The exocyclic
amino group at C-4 of cytidine points into the major groove
of a double stranded RNA,7 such that sterical interfering of
RNA chains in the branch point is likely to be minimal.
Furthermore, the Watson-Crick face of the modified
nucleoside is free for base pairing with the opposite
guanosine in the substrate strand (Fig. 2). Since the overall
structure of an RNA duplex is not affected by a single
20-deoxy substitution,8 we decided to use a 20-deoxy
nucleotide instead of the natural ribonucleoside at the
branch point to keep preparation and handling of the
modified building block as facile as possible. The 20-deoxy
analogue is not located within a conserved sequence and
hence will not interfere with ribozyme activity.

S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
9275
Figure 2. Secondary structure of the branched ribozyme HP-RJTLB with three substrates S14F5, S24F5, S36F5, and of the reverse-joined hairpin ribozyme
HP-RJWTC6 with substrate S14F5. All substrates are 50-end labelled with fluorescein; arrows denote the cleavage site. The structure of the nucleoside at the
branch point (marked as Y in HP-RJTLB) is separately shown.
Figure 3. Synthesis of alternative phosphoramidite building blocks 2 and 3 for synthesis of the branched RNA. Route A: (i) DMTCl, DMAP, pyridine, 1.5 h, rt;
(ii) chloro-(2-cyanoethoxy)-diisopropylaminophosphine, EtN(iso-Pr)2, THF, 2 h, rt; (iii) 1,2,4-triazole, POCl3, Et3N, CH3CN, 1 h, 0 8C/rt. Route B:
(i) DMTCl, DMAP, pyridine, 1.5 h, rt; (ii) TBDMSCl, imidazole, DMF, 12 h, rt; (iii) 1,2,4-triazole, POCl3, Et3N, CH3CN, 1 h, 0 8C/rt; (iv) H2N–(CH2)6–
OH, CH3CN, 1 h, rt; (v) (a) levulinic acid, DCC, pyridine, 5 h, rt, levulinic anhydride, DMAP, pyridine, 1.5 h, rt; (vi) TBAF–3H2O, THF, 15 min, rt;
(vii) chloro-(2-cyanoethoxy)-diisopropylaminophosphine, EtN(iso-Pr)2, THF, 2 h, rt.

9276
S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
Figure 4. Synthesis scheme of the branched RNA HP-RJTLB. Sequences of segments A, B and C as well as the structure of Y are shown in Figure 2. Details of
synthesis are given in the text.
the conditions of hydrazinolysis. Therefore, more stable
N-protecting groups were used: benzoyl for A and C,17 and
dimethylformamidine for G.21,22 The half life of N-benzoyl
protected adenosine and cytidine under the conditions of
hydrazinolysis was determined to be about 3 and 5 h,
respectively, while the dimethylformamidine group was
stable for the time of observation. Removal of the levulinyl
group by hydrazinolysis takes about 5 min, hence leaving
the major part of the N-protected nucleobases intact.
synthesis of segment C, the levulinic acid group at the
modified cytidine base was removed (Fig. 4, step V)
followed by assembly of segment C (Fig. 4, step VI).
Synthesis of all three segments of the branched oligo-
nucleotide was achieved with yields as usual for completely
unmodified non-branched oligoribonucleotides.
2.3. Deprotection, purification and characterization
Removal of base protecting groups and b-cyanoethyl groups
at the phosphates as well as cleavage from the polymer
support was accomplished using saturated methanolic
ammonia, followed by treatment with TEA-3HF/DMF
(3:1) to remove silyl groups at the 20-OH.23 In order to
avoid exposure of the branched RNA to long heating in
ammonia, the first of these two steps was carried out at
ambient temperature, however, for prolonged time. The
deprotected oligonucleotide was precipitated from n-buta-
nol and purified by electrophoresis through a 20%
denaturing polyacrylamide gel at 60 8C. We evaluated
these conditions as being essential to achieve sufficient
resolution of the desired product band (Fig. 5).
Chain assembly was carried out as shown in Figure 4. We
˚
used a 1000 A CPG support (instead of the routinely used
˚
500 A), to avoid sterical hindrance of reaction sites by
growing branched oligonucleotide chains.
Synthesis of segment A (Fig. 4, step I) was performed using
the standard procedure of automated RNA synthesis. The
branching nucleotide was incorporated with small changes
to the standard protocol. A higher excess of 3 and an
extended coupling time ensured 99.5% coupling efficiency
(Fig. 4, step II, for details refer to Section 4). Synthesis was
continued to assemble segment B, as before following the
standard protocol (Fig. 4, step III). Next, the 50-OH group at
the end of segment B was capped with an acetyl group to
prevent further reaction (Fig. 4, step IV). To initiate the
Due to the presence of branched and non-branched species it
was not possible to predict the migration properties of the

S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
9277
2.4. Cleavage experiments and kinetic characterization
Since HP-RJTLB is derived from a catalytic RNA structure
we next asked if activity has been preserved in the branched
ribozyme. To this end, we studied the ability of HP-RJTLB
to cleave three substrates of different length: S14F5, S24F5
and S36F5 (Fig. 2), and compared the results with the non-
branched reverse-joined hairpin ribozyme HP-RJ0WTC6.
All three substrates carry a fluorescein label at the 5 -end to
allow for reaction analysis with an automated DNA
sequencer as demonstrated previously.3 Reactions were
carried out in the presence of 2 mM spermine, since we have
found previously that activity of reverse-joined hairpin
ribozymes is considerably improved in the presence of the
polyamine.24 All substrates are cleaved by HP-RJTLB with
the same specificity as observed for HP-RJWTC6. The
results are summarized in Table 1.
Figure 5. Gel electrophoretic analysis (20% polyacrylamide) of HP-
RJTLB. Lane 1: crude product of HP-RJTLB synthesis. The arrow denotes
the product band. Lane 2: HP-RJTLB after purification and reloading onto
the gel.
product RNA. Therefore, the strongest band was cut out and
the RNA was eluted from the gel. In order to prove the
identity of the product, the obtained RNA was subjected to
nucleoside composition analysis. For this purpose samples
were digested with both nuclease P1 and alkaline phospha-
tase and the resulting nucleosides were separated and
identified by HPLC (Fig. 6). In the presence of nuclease P1
the phosphodiester bond between the aminolinker hydroxyl
group and the neighbouring nucleoside was totally hydro-
lyzed. We found snake venom phosphodiesterase being
less efficient in catalyzing this reaction (data not shown).
20-Deoxy-N4-(6-hydroxyhexyl)-5-methylcytidine could be
identified in the digestion mixture (marked as Y in Fig. 6),
which confirms successful incorporation of this
modified nucleoside into the analyzed oligonucleotide.
Integration of peak areas normalized to specific extinction
coefficients revealed a nucleoside composition which
agrees very well with theoretically calculated values,
demonstrating the superior quality of the synthesized
branched RNA.
The short substrate S14F5 as well as S24F5 that extends
over the branch point are cleaved by HP-RJTLB with only
about 5-fold lower rates than observed with the non-
branched ribozyme HP-RJWTC6. Interestingly, cleavage of
the substrate S36F5 that is fully paired with the ribozyme
proceeds with rather slow rate, kreact for HP-RJTLB is
another 5-fold down compared with the other two substrates
S14F5 and S24F5. A similar result has been obtained with
HP-RJWTC6: S36F5 is cleaved 3-fold slower than the two
shorter substrates. Also, the end-point of the reaction is
rather low. Only 7% of the total S36F5 substrate is cleaved
by HP-RJTLB, 45% by HP-RJWTC6. This result might be
attributed to an increased ligation activity of both ribozymes
with the long substrate. The 50-terminus of S36F5 forms a
contiguous 16 base pair helix with the 30-terminus of the
ribozyme. As mentioned above the natural hairpin ribozyme
is also an efficient ligase and it has been shown earlier that
structural stabilisation shifts the cleavage/ligation equi-
librium towards ligation. Fedor and co-workers have
Figure 6. Nucleoside composition analysis of HP-RJTLB by reverse phase HPLC. Lane 1: digestion mixture of synthesized HP-RJTLB. The peak marked with
an asterisk results from a component of the digestion buffer or the enzyme (see also lane 2). Lane 2: enzymatic digestion mixture without RNA as control. Lane
3: marker nucleoside Y: 20-deoxy-N4-(6-hydroxyhexyl)-5-methylcytidine. The calculated and experimentally found (in parenthesis, data results from three
independent digestion experiments) base composition for HP-RJTLB is shown.

9278
S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
Table 1. Kinetic constants for cleavage reactions of HP-RJTLB and HP-RJWTC6
Substrate
HP-RJTLB
kreact!102 minK1
HP-RJWTC6
kreact!102 minK1
K1, nM
End-point of
reaction (%)
K1, nM
End-point of
reaction (%)
S14F5
S24F5
S36F5
71.0G10.3
84.3G11.5
152.7G31.2
2.4G0.1
2.3G0.1
0.044G0.004
80
70
7
12.8G2.4
93.0G15.6
46.5G11.9
13.0G0.4
12.4G0.7
3.8G0.2
95
90
45
demonstrated that cleavage takes place rapidly only when
cleavage products dissociate rapidly and that cleavage rates
fall when cleavage products remain bound in stable base
paired helices. This was taken as evidence that bound
products are re-ligated.25–28 What holds for the convention-
al hairpin ribozyme also applies to reverse-joined hairpin
ribozymes, ligation activity increases with the length of
duplexes (S. Ivanov, S. Mu¨ller, unpublished observations).
Therefore, the observed slow cleavage of S36F5 by HP-
RJTLB and HP-RJWTC6 might be interpreted as enhanced
ligation due to stable bound cleavage products and in
consequence re-ligation. This interpretation gains further
support by the observation, that both ribozymes cleave only
a fraction of the long substrate, while cleavage of short
substrates went on further (Table 1). To additionally support
this result we carried out ligation experiments with both
ribozymes providing the two cleavage fragments as
substrates (Fig. 7).
HP-RJTLB. The ligation reaction has been carried out
several times with varied concentrations of the two
fragments and of the ribozyme as well as under differing
reaction conditions. However, in no case ligation activity
could be detected. Obviously, the low activity of the
branched ribozyme HP-RJTLB for cleavage of the long
substrate S36F5 is not the result of re-ligation. Comparison
of HP-RJTLB with HP-RJWTC6 shows that the branched
ribozyme HP-RJTLB cleaves all three substrates to a
smaller extent than HP-RJWTC6 and the cleaved fraction
decreases with the length of substrates (Table 1). This might
imply that a fraction of HP-RJTLB is trapped in inactive
conformations. With increasing length of the substrate the
ability to fold into the required active structure may be
decreased, such that only a small fraction of the substrate
can be cleaved. Accordingly, also ligation would be
hampered with long substrates. Strikingly, cleavage might
occur only from temporarily folded species that are
insufficiently stable to allow for ligation. Extending the
linker that connects the two domains in the branch point
might help to improve ribozyme folding in a way that the
two domains have more conformational freedom to be
Whereas as expected, the reverse-joined ribozyme HP-
RJWTC6 readily ligated the two fragments (yield: 11%),
very surprisingly no ligation activity could be detected for
Figure 7. Schematic presentation of the ligation assay and data analysis. The substrate S17F5p carries a 20,30-cyclic phosphate and is ligated with S19 to yield
the product S36F5.

S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
9279
positioned in an optimal orientation to each other and to
interact via a specific net of hydrogen bonds that is known to
stabilize the three-dimensional complex.29
Pharmacia Biotech). Buffers for ribozyme digestion and
substrate cleavage were prepared using autoclaved de-
ionized water, filtered through a sterile 0.2 mm pore filter,
and stored at K20 8C prior to use.
3. Conclusion
4.2.1. Preparation of modified phosphoramidite building
blocks 50-O-(4,40-dimethoxytrityl)-4-(1,2,4-triazolyl)-
thymidine-30-O-(2-cyanoethyl-N,N-diisopropylphosphor-
amidite) 2. 1,2,4-Triazole (780 mg, 11.29 mmol) was
suspended in dry CH3CN (15 ml), with intensive stirring,
in an ice bath. Dry triethylamine (1.8 ml, 12.95 mmol) was
added followed by dropwise addition of POCl3 (0.24 ml,
2.70 mmol). The mixture was left to react for 45 min an0d
then filtered through cellulose0 into a solution of 50-O-(4,4 -
dimethoxytrityl)-thymidine-3 -O-(2-cyanoethyl-N,N-diiso-
propylphosphoramidite) (447 mg, 0.60 mmol) in dry
CH3CN (1 ml). After 1 h at room temperature quantitative
conversion of the educt into product (TLC analysis) was
observed. Ethylacetate (100 ml) was added to the reaction
mixture and the organic layer was washed with an equal
volume of 10% aqueous NaHCO3 followed by brine. The
aqueous solution was back-extracted with ethylacetate. The
organic layers were combined, dried over Na2SO4, filtered
and concentrated to dryness in vacuo resulting in yellow oil.
The viscous material was purified on a silica gel column
under isocratic conditions using Et3N/CHCl3 (0.5:99.5, v/v)
as eluent. The desired fractions were pooled and evaporated
to obtain 2 as white foam (392 mg, 82.0%). TLC analysis:
Rf (2)Z0.73 (both isomers) (AcOEt/Et3N/CH2Cl2, 34:8:58,
A ribozyme 81 nucleotides in length was prep0ared by
chemical synthesis and the modified nucleoside 2 -deoxy-
N4-(6-hydroxyhexyl)-5-methylcytidine was introduced in
order to create an artificial branch at a pre-defined position
of the RNA chain. The branched ribozyme is functional:
cleavage rates were only 5-fold lower compared with the
corresponding non-branched ribozyme. To the best of our
knowledge, HP-RJTLB is the first example of a chemically
synthesised artificially branched catalytic RNA. Even
though, HP-RJTLB in its present form does not catalyze
ligation of RNA fragments and, therefore, is not suitable to
be used as building block of a twin ribozyme for RNA
repair,4 the herein presented results demonstrate, that the
existing strategies for chemical preparation of RNA allow
for the development of new catalytic structures that cannot
be prepared enzymatically. This considerably increases the
range of structural modification for analysis and functional
design of RNA molecules. We expect our results to greatly
enhance the prospects of synthetic RNA molecules in
molecular biology and biochemistry studies. Further
investigations into optimization of the structure of reverse-
joined hairpin ribozymes for the twin ribozyme approach4
are in progress.
1
v/v). H NMR (CDCl3): dZ9.21 (s, 1H, H-triazolyl), 8.29
(d, 1H, H–C6), 8.01 (d, 1H, H-triazolyl), 7.35–7.20 (m, 9H,
H-phenyl), 6.76 (d, 04H, H-phenyl), 6.27 (q, 1H, H–C10),
4.61 (m, 1H, H–C3 ), 4.05 (m, 1H, H–C40), 3.72 (s, 6H,
CH3O), 3.62–3.45 (m, 1H, Ha-C50; 2H CH2OP; 2H H–C
i-Pr), 3.29 (m, 1H, Hb-C50), 2.77 (m, 1H, Ha-C20), 2.56 (t,
1H, H–CHCN), 2.35 (m, 1H, Hb-C20; 1H, H–CHCN), 1.87,
1.86 (ss, 3H, CH3–C5), 1.18–0.97 (m, 12H, CH3 i-Pr). 13C
NMR (CDCl3): dZ158.7 (C phenyl), 158.1 (C4), 153.9
(C2), 153.3, 146.7 (C triazolyl), 145.0 (C phenyl), 144.2
(C6), 135.2 (C phenyl), 130.2–127.2 (CH phenyl), 117.4
(CH2CN), 113.4–113.3 (CH phenyl), 105.8 (C5), 87.5, 87.4,
86.9 (Cq, C10,C40), 72.1 (C30), 62.3 (C50), 58.0 (CH2OP),
55.2 (OCH3), 41.2 (C20), 43.2 (CH2CN), 24.6 (CH-iPr),
20.2 ((CH3)2iPr), 16.4 (CH3–C5). 31P NMR (CDCl3): dZ
150.6, 150.0. FAB-MS, m/z: Calcd (MCNaC) 818.87,
found 818.40 (MZC42H50O7N7P).
4. Experimental
4.1. General methods
NMR studies were carried out on a Bruker AM 300 NMR
spectrometer at 300 K using approx. 30 mM sample
1
solutions in CDCl3. H, 13C and 31P NMR spectra were
recorded at 300, 75 and 121 MHz, respectively. Chemical
shifts were measured in relation to tetramethylsilane (dZ
0 ppm) for 1H NMR and 13C NMR spectra and 85%
phosphoric acid (dZ0 ppm) for 31P NMR spectra. Mass
spectra (FAB or ESI) of compounds were taken with a
Quadrupole Mass Spectrometer HP 5995 A. Analytical
HPLC was performed using an Eurospher 100 C18 column
(4.6!250 mm, 5 mm) (Knauer GmbH, Germany). Gravity
chromatography was performed on a 2.5!15 cm column
packed with silica gel 60 (0.063–0.200 mm, Merck,
Germany) with an appropriate solvent as eluent. Thin-
layer chromatography (TLC) of reaction mixtures and
purified compounds was carried out using pre-coated silica
gel 60 F254 plates (Merck).
4.2.2.
20-Deoxy-50-O-(4,40-dimethoxytrityl)-N4-(6-
hydroxyhexyl-(O-levulinyl))-5-methyl-cytidine 30-O-(2-
cyanoethyl-N,N-diisopropylphosphoramidite) 3. The
modified nucleoside was prepared mainly as described by
Horn et al.11 and converted into the phosphoramidite
building block. Briefly, the procedure involved protection
of 50- and 30-hydroxyl groups of 20-deoxyuridine followed
by conversion of the base into the 4-triazolo derivative.
Next, the triazolo group was substituted with 6-amino-
hexanol and the alcohol group at the linker was converted
into a levulinic acid ester. Finally the0 30-O-TBDMS group
was removed for preparation of the 3 -O-phosphoramidite.
20-deoxy-50-O-(4,40-dimethoxytrityl)-N4-(6-hydroxyhexyl-
(O-levulinyl))-5-methyl-cytidine 30-O-(2-cyanoethyl-N,N-
diisopropylphosphoramidite) 3 was obtained as light yellow
foam (392 mg, 83.2%). TLC analysis: Rf (3)Z0.66, 0.75
4.2. Reagents and solvents
Reagents and solvents for preparation of modified phos-
phoramidites were purchased from Sigma-Aldrich Chemie
GmbH (Germany) and Merck (Germany), fully protected
ribonucleoside phosphoramidites and long-chain alkyl-
amine CPG-columns for RNA-synthesis were purchased
from Glen Research (USA). Substrate RNAs were 50-end
labelled using ‘fluoreprime’ fluorescein amidite (Amersham

9280
S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
(both isomers) (n-C6H14/AcOEt/Et3N, 5:25:2.4, v/v). 1H
NMR (CDCl3): dZ7.58 (s, 1H, H–C6), 7.33–7.15 (m, 9H,
H-phenyl), 6.75 (d, 4H, H-phenyl), 6.39 (m, 1H, H–C10),
4.52 (m, 1H, H–C30), 4.02 (m, 1H, H–C40), 3.99 (t, 2H,
CH2-Lev), 3.73, 3.72 (ss, 6H, CH3O), 3.51–3.42 (m, 4H,
CH2; 2H, CH2OP, 2H H–C i-Pr), 3.22 (m, 2H, H–C50),
2.70–2.66 (m, 2H, CH2-Lev), 2.56–2.47 (m, 1H, Ha-C20;
1H, H–CHCN), 2.31 (t, 1H, H–CHCN), 2.18 (m, 1H,
Hb-C20), 2.12 (s, 3H, CH3-Lev), 1.55 (m, 4H, CH2), 1.37 (s,
3H, CH3-C5), 1.30 (m, 4H, CH2), 1.22–0.93 (m, 12H, CH3
i-Pr). 13C NMR (CDCl3): dZ206.8 (Me–CO), 172.8
(O–CO), 163.1 (C4), 158.6 (C phenyl), 156.2 (C2), 144.4
(C phenyl), 137.1 (C6), 135.5 (C phenyl), 130.2–127.0 (CH
phenyl), 117.6 (CH2CN), 113.3–113.2 (CH phenyl), 101.7
(C5), 86.6, 85.5 (Cq, C10,C40), 72.1 (C30), 64.5 (C50), 58.1
(CH2OP), 55.3 (OCH3), 46.9 (C20), 45.7 (CH2-linker), 43.2
(CH2CN), 40.9 (CH2-linker), 37.9 (CH2-Lev), 29.9, 29.2
(CH2-linker), 28.5 (CH3-Lev), 28.0 (CH2-Lev), 26.4, 25.6
(2CH2-linker), 24.7–24.5 (CH-iPr), 20.1–19.4 ((CH3)2iPr),
12.4 (CH3–C5). 31P NMR (CDCl3): dZ150.1, 149.5.
FAB-MS, m/z: Calcd (MCNaC) 965.10, found 965.03
(MZC51H68O10N5P).
cytosine, dimethylformamidine with guanine. Unmodified
phosphoramidites were coupled using the standard proto-
col.30 For coupling of the branching nucleotide the
following changes to the protocol were made: a 40-fold
excess of phosphoramidite over the solid phase was used
and the coupling time was extended from 5 to 24 min. The
coupling reaction was carried out in two steps: first, 100 ml
of the 0.2 M phosphoramidite solution were cycled over the
column and coupling was allowed to proceed for 12 min.
Immediately after this, the same amount of fresh phosphor-
amidite solution was transferred to the column and the
coupling reaction was extended for another 12 min. The
following nucleotides (segment B) were coupled without
changes to the standard protocol. Prior to synthesis of
segment C (see Figs. 2 and 4), the terminal 50-OH group of
segment B was acetylated by treating the polymer support
manually five times with 100 ml Capping A solution (Ac2O
in pyridine) and 100 ml Capping B solution (1-methylimi-
dazole in pyridine), each turn taking 3 s. The column was
washed thoroughly with acetonitrile (5 ml), then removed
from the synthesizer and treated with 6 ml of a freshly
prepared solution of 0.5 M hydrazine hydrate in pyridine/
acetic acid (4:1, v/v) for 5 min (solution was slowly passed
through the column using a syringe). The column again was
washed thoroughly with anhydrous acetonitrile (7 ml) and
transferred back to the synthesizer. Assembly of segment C
was carried out again following the standard protocol.
4.3. RNA synthesis
Oligoribonucleotides were synthesized by the phosphor-
amidite method essentially as described previously.3,30 An
automated DNA/RNA synthesizer (Gene Assembler Special,
Pharmacia) was used for chain assembly at a 1 mmol scale.
Fluorescein labelled RNA substrates (S14F5, S24F5 and
30
˚
S36F5) were synthesized on CPG 500 A, as described.
AAAAAK
CGUGK
The model oligoribonucleotide SI-SLA5 (
XAAðdTÞ
˚
Deblocking was performed using saturated methanolic
ammonia for 22 h at 25 8C with the branched RNAs SI-
SLA5 and HP-RJTLB or over 12 h at ambient temperature
with substrate RNAs S14F5, S24F5 and S36F5. Removal of
the 20-O-silyl-protecting groups was achieved by treating
the samples with triethylamine trihydrofluoride in DMF
(3:1, v/v, 0.8 ml) for 1.5 h at 55 8C. The reaction was
quenched by addition of water (0.2 ml) and the RNA chains
were precipitated from n-butanol.
with XZ4-triazolo-dT), was synthesized on a 500 A CPG
support. The modified phosphoramidite 2 was dissolved in
dry acetonitrile to give a 0.12 M concentration and filtered
through 0.45 mm Teflon filters prior to use. Synthesis was
performed ‘trityl off’. Phenoxyacetyl groups were used for
protection of the amino functions of adenine, cytosine and
guanine. Unmodified phosphoramidites were coupled using
the standard protocol.30 The coupling time for 2 was
extended from 5 min (standard protocol) to 10 min. The
following nucleotides CGUG were coupled without changes
to the standard protocol. Prior to synthesis of the side-chain
(A)5, the terminal 50-OH group of the main-chain was
acetylated by manually treating the polymer support five
times with 100 ml Capping A solution (Ac2O in pyridine)
and 100 ml Capping B solution (1-methylimidazole in
pyridine), each turn taking 10 s. The column was washed
thoroughly with acetonitrile (5 ml), then removed from the
synthesizer and treated with 10 ml of a freshly prepared
solution of 0.33 M aminohexanol in acetonitrile for 45 min
(solution was slowly passed through the column using a
syringe). The column again was washed thoroughly with
anhydrous acetonitrile (10 ml), transferred back to the
synthesizer and washed again with anhydrous acetonitrile
(5 ml). Assembly of the side-chain was carried out
following the standard protocol.
The branched RNAs were purified by 20% denaturing (7 M
urea) PAGE (acrylamide/bis-acrylamide 19:1) on 8!10 cm
glass plates at 60 8C in TBE buffer at 100 V. The gel was
soaked in ethidium bromide solution; visualization of RNA
was achieved by irradiation with UV-light (254 nm). The
band corresponding to the desired product was excised from
the gel and eluted with 2 M LiClO4 overnight at room
temperature. The oligonucleotide then was precipitated
from acetone. Fluorescein labelled RNA substrates were
purified using 20% denaturing polyacrylamide gels. Bands
corresponding to the desired products were visualized both
at 254 and 366 nm, excised from the gel and treated as
described above.
4.4. Nucleoside composition analysis
For synthesis of the branched ribozyme HP-RJTLB, CPG
˚
with a pore size of 1000 A was used. The branching
nucleotide building block 3 was dissolved in dry acetonitrile
at a concentration of 0.2 M and filtered through 0.45 mm
Teflon filters prior to use. Synthesis was performed ‘trityl
off’. Benzoyl was used for N-protection with adenine and
Oligoribonucleotides (0.08 O.D.260) were dissolved in 10 ml
of nuclease P1 buffer (40 mM AcONa, 2 mM (AcO)2Zn pH
5.3) and a freshly prepared solution of nuclease P1 from
Penicillium citrinum in the same buffer (10 ml, 10 mg/ml)
was added. After incubation at 37 8C for 12 h, 10!
dephosphorylation buffer (4 ml), alkaline phosphatase from

S. A. Ivanov et al. / Tetrahedron 60 (2004) 9273–9281
9281
calf intestine (3 ml, 3 U) and water (13 ml) were added to
References and notes
obtain a final volume of 40 ml. Reaction was allowed to
proceed for 3 h at 37 8C. Individual reaction mixtures were
analysed by HPLC using a RP-18 column with buffer A
(0.1 M ammonium acetate, pH 6.5), and a gradient of buffer
B (0.1 M ammonium acetate, pH 6.5, 60% CH3CN): 0–15%
B over 20 min, 15–40% B over 30 min and 40–100% over
2 min at 37 8C and a flow rate of 1 ml/min.
1. Hampel, A.; Tritz, R. Biochemistry 1989, 2, 4929–4933.
2. Haseloff, J.; Gerlach, W. L. Gene 1989, 82, 43–52.
3. Schmidt, C.; Welz, R.; Mu¨ller, S. Nucleic Acids Res. 2000, 28,
886–894.
4. Welz, R.; Bossmann, K.; Klug, C.; Schmidt, C.; Fritz, H.-J.;
Mu¨ller, S. Angew. Chem., Int. Ed. 2003, 42, 2424–2427.
5. Komatsu, Y.; Kanzaki, I.; Koizumi, M.; Ohtsuka, E. J. Mol.
Biol. 1995, 252, 296–304.
Peak areas in conjunction with extinction coefficients at
260 nm(C:7.4,U:9.9,G:11.5,A:15.4, dT: 8.7, inl/mmol cm)
were used to calculate relative concentrations of monomers.
6. Komatsu, Y.; Kanzak, I.; Ohtsuka, E. Biochemistry 1996, 35,
9815–9820.
7. Wada, T.; Kobori, A.; Kawahara, S.; Sekine, M. Eur. J. Org.
Chem. 2001, 4583–4593.
4.5. Cleavage reactions
8. Wyatt, J. R.; Walker, G. T. Nucleic Acids Res. 1989, 17,
7833–7842.
Cleavage reactions were performed under single-turnover
conditions and each experiment was repeated at least once.
A mixture of ribozyme (20–220 nM) and substrate (10 nM)
in Tris–HCl buffer (15 mM, pH 7.5) was heated at 90 8C for
1 min followed by incubation at 32 8C for 15 min.
Magnesium chloride was added to a concentration of
10 mM and the mixture was left at 32 8C for another
10 min. The cleavage reaction was started by addition of
spermine (pH 7.5, final concentration 2 mM). The final
volume of the reaction mixture was 20 ml. Aliquotes (3 ml)
were removed at six suitable time intervals and the reaction
was quenched by addition to 3 ml of stop-mix (10 mM
EDTA in 90% formamide). Samples were heated to 90 8C
for 1 min, then immediately cooled on ice and subsequently
subjected onto a 15% denaturing gel (7 M urea) using an
ALF DNA sequencer as described previously.3 Initial rates
of reactions were obtained from a plot of product formation
against time within the linear phase of reaction. Kinetic
parameters were determined from linear curve fitting using
Eaddie–Hofstee plots.
9. Sung, W. J. Nucleic Acids Res. 1981, 9, 6139–6151.
10. Duncan, B. K.; Miller, J. H. Nature 1980, 287, 560–561.
11. Horn, T.; Chang, C. A.; Urdea, M. S. Nucleic Acids Res. 1997,
25, 4835–4841.
12. Jones, R. A. In Oligonucleotide Synthesis: A Practical
Approach; Gait, M. J., Ed.; Oxford University Press: Oxford,
1990; pp 23–34.
13. Sproat, B. S.; Beijer, B.; Grøtli, M.; Ryder, U.; Morand,
K. L.; Lamond, A. I. J. Chem. Soc., Perkin Trans. 1 1994,
419–431.
14. Cameron, M. A.; Cush, S. B.; Hammer, R. P. J. Org. Chem.
1997, 62, 9065–9069.
15. Scaringe, S. A.; Francklyn, C.; Usman, N. Nucleic Acids Res.
1990, 18, 5433–5441.
16. Beigelman, L.; Karpeisky, A.; Matulic-Adamic, J.; Haeberli,
P.; Sweedler, D.; Usman, N. Nucleic Acids Res. 1995, 23,
4434–4442.
17. Iwai, S.; Ohtsuka, E. Nucleic Acids Res. 1988, 16, 9443–9456.
18. Sørensen, M. D.; Meldgaard, M.; Rajwanshi, V. K.; Wengel, J.
Bioorg. Med. Chem. Lett. 2000, 10, 1853–1856.
19. Oligvie, K. K.; Nemer, M. J.; Hakimelahi, G. H.; Proba,
Z. A.; Lucas, M. Tetrahedron Lett. 1982, 23, 2615–2618.
20. Ganeshan, K.; Tadey, T.; Nam, K.; Braich, R.; Purdy,
W. C.; Boeke, J. D.; Damha, M. J. Nucleosides Nucleotides
1995, 14, 1009–1013.
4.6. Ligation reactions
For ligation the two following substrates were0 used: GUC
CUC UUG CAC GGU CCG C (S19) and 5 -fluorescein
labelled F-CAG 0AAA UCU GGC UCA CAcp (S17F5p),
which carries a 2 , 30-cyclic-phosphate group. Ligation was
performed with equimolar concentrations of substrates and
ribozyme (200 nM). A mixture of ribozyme and 19-mer in
Tris–HCl buffer (15 mM, pH 7.5) was heated at 90 8C for
1 min followed by incubation at 32 8C for 15 min. Then the
17-mer and MgCl2 (final concentration of MgCl2 10 mM)
were added and incubated at 32 8C for another 10 min. The
reaction was started by addition of spermine (pH 7.5) to a
concentration of 2 mM. The final volume of the reaction
mixture was 20 ml. Aliquotes (3 ml) were removed after 30,
60 and 120 min. The reaction was quenched by pipetting the
aliquots into 3 ml of stop-mix (10 mM EDTA in 90%
formamide). Samples were heated to 90 8C for 1 min, then
immediately cooled on ice and subsequently subjected onto
a 15% denaturing gel (7 M urea) using an ALF DNA
sequencer as described.3
21. Wu, T.; Ogilvie, K. K.; Pon, R. T. Nucleic Acids Res. 1989, 17,
3501–3517.
22. Damha, M. J.; Ganeshan, K.; Hudson, R. H. E.; Zabarylo,
S. V. Nucleic Acids Res. 1992, 20, 6565–6573.
23. Wincott, F.; DiRenzo, A.; Shaffer, C.; Grimm, S.; Tracz, D.;
Workman, C.; Sweedler, D.; Gonzalez, C.; Scaringe, S.;
Usman, N. Nucleic Acids Res. 1995, 23, 2677–2684.
24. Welz, R.; Schmidt, C.; Mu¨ller, S. Biochem. Biophys. Res.
Commun. 2001, 283, 648–654.
25. Hegg, L. A.; Fedor, M. J. Biochemistry 1995, 34,
15813–15828.
26. Fedor, M. J. Biochemistry 1999, 38, 11040–11050.
27. Nesbitt, S. M.; Erlacher, H. A.; Fedor, M. J. J. Mol. Biol. 1999,
286, 1009–1024.
28. Yadava, R. S.; Choi, A. J.; Lebruska, L. L.; Fedor, M. J. J. Mol.
Biol. 2001, 309, 893–902.
´
29. Rupert, P. B.; Ferre-D’Amare, A. R. Nature 2001, 410,
´
Acknowledgements
780–786.
30. Welz, R.; Mu¨ller, S. Tetrahedron Lett. 2002, 43, 795–797.
We gratefully acknowledge financial support by the DFG
and the Pinguinstiftung.