Multifunctional dinucleotide analogs for the generation of complex RNA conjugates

Article abstract of DOI:10.1016/S0040-4020(00)01114-5

Oligonucleotide conjugates are needed for in vitro selection schemes aiming at reactions between small, organic reactants. A general strategy is provided for the generation of the required RNA reactant conjugates based on multifunctional dinucleotide anal

Full text of DOI:10.1016/S0040-4020(00)01114-5

TETRAHEDRON
Pergamon
Tetrahedron 57 (2001) 1261±1268
Multifunctional dinucleotide analogs for the generation of
complex RNA conjugates
p
È
Felix Hausch and Andres Jaschke
Institute of Chemistry±Biochemistry, Free University Berlin, Thielallee 63, D-14195 Berlin, Germany
Received 23 October 2000; accepted 22 November 2000
AbstractÐOligonucleotide conjugates are needed for in vitro selection schemes aiming at reactions between small, organic reactants. A
general strategy is provided for the generation of the required RNA reactant conjugates based on multifunctional dinucleotide analogs. These
modi®ed dinucleotides allow for enzymatic ligation to native or enzymatically transcribed RNAs. They further contain a ¯exible poly-
ethylene glycol spacer for correct spatial positioning and a photolabile cleavage site for selective release. The dinucleotides can be
derivatized with the desired organic compounds by activated ester chemistry as was demonstrated by coupling to several nucleobases
and nucleotides. q 2001 Elsevier Science Ltd. All rights reserved.
1. Introduction
drawback, several classical selection experiments have also
been prone to side reactions at internal positions of the
nucleic acid.8,9
The past decade has seen an increasing interest in function-
alized oligonucleotides.1 The conjugation to chromophores
and ¯uorescent probes has facilitated the detection of
nucleic acids signi®cantly and nearly eliminated the use of
radioactively labeled oligonucleotides in standard tech-
niques such as Sanger sequencing or Northern/Southern
blotting.2 Metal chelators attached to nucleic acids serve
as arti®cial nucleases3 and conjugated markers such as
acridine or psoralene are used to study DNA±protein or
RNA±protein interactions. The advent of antisense or anti-
gene therapy has set new demands on the in vivo stability
and membrane permeability which were met by nuclease-
resistant or lipophilic modi®cations.4 The site-speci®c
immobilization of nucleic acids is essential for the fabrica-
tion of high-density DNA chips which has revolutionized
whole genome analysis.5
To expand the in vitro selection technique to previously
inaccessible reactions, conjugates of enzymatically gener-
ated nucleic acids are needed.10 In this approach, potential
(deoxy-) ribozymes can intramolecularily act on their
coupled reactant instead of on themselves. Through the
covalent linkage to their catalytically generated products
they become tagged, thus allowing for partitioning from
the unmodi®ed bulk of inactive sequences in the selection
step. The use of such RNA conjugates ultimately led to the
isolation of some remarkable new ribozymes catalyzing
fundamental organic reactions like amide or N-glycosidic
bond formation11,12 and Diels±Alder cycloaddition.13,14
The key step in such expanded selection protocols based on
oligonucleotide conjugates is the introduction of potential
reactants into enzymatically generated nucleic acids. While
several methods exist for the site-speci®c modi®cation of
oligonucleotides during their automated chemical synthesis
(e.g. by incorporating suitably functionalized phosphorami-
dites),15,16 labeling of native or enzymatically generated
DNA/RNA has been described only for a few selected
reporter groups with limited speci®city and ef®ciency.
Recently, complex RNA conjugates have been shown to be
a major improvement for in vitro selection experiments
aiming at the isolation of new catalytic oligonucleotides.6,7
In this technique, combinatorial nucleic acid libraries are
successively screened for catalytic activity followed by
enzymatic ampli®cation of the selected sequences. Since
an intramolecular catalysis event is exploited for the selec-
tion, the basic protocol using unmodi®ed DNA/RNA is
strictly restricted to oligonucleotide-modifying reactions.
Catalysts for reactions between two small, organic reactants
cannot be investigated using this approach. As an additional
We recently proposed photocleavable dinucleotide analogs
as versatile tools to tackle the inherent limitations of the in
vitro selection.17 They allow for ef®cient and site-speci®c
derivatization of randomized RNA transcripts and are
compatible with the overall selection scheme. Here we
present a general conjugation strategy based on these
dinucleotide analogs which vastly expands the repertoire
of potential reactants to be investigated and signi®cantly
Keywords: oligonucleotide derivatives; RNA conjugates; in vitro selection;
ribozymes; photocleavage.
p
Corresponding author. Tel.: 149-30-8385-2194; fax: 149-30-8385-5060;
e-mail: jaschke@chemie.fu-berlin.de
0040±4020/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020(00)01114-5

È
F. Hausch, A. Jaschke / Tetrahedron 57 (2001) 1261±1268
1262
reduces the preparative manipulations needed for their
conjugation.
whereas the ®ve hexaethylene glycol units of 2b should
be suf®cient to span the length of an extended 30 base
pair DNA duplex.19 Polyethylene glycol (PEG) was
chosen as a spacer because of its high ¯exibility,
chemical inertness and structural neutrality, thereby
minimizing a participation of the linker in the
catalysis.20
2. Results and discussion
2.1. Design and synthesis of the dinucleotide analogs
² An o-nitrobenzyl moiety21 which allows light-induced
cleavage of the linker under a variety of conditions.
Because photocleavage is independent of pH, redox
potential, solvent polarity or ion strength, it offers a
highly stringent selection criterion. The regioselective
release of RNAs immobilized via their attached reac-
tants should be useful to suppress side reactions
which occur during the selection process and have
been shown to complicate the selection process.8,9
² A 30-terminal functional group to allow the attach-
ment of potential reactants or other small molecules
of interest. With biotin as a versatile af®nity tag, this
was realized during the automated synthesis of the
dinucleotide analogs 1a and 1b by using a biotiny-
lated solid support. As such appropriately pre-deriva-
tized solid supports are usually not commercially
available, we introduced a primary amine or a car-
boxyl group into the dinucleotide analogs 2 and 3.
These functional groups allow for the post-synthetic
Previous experiments have shown that photocleavable
linkers are compatible with the direct selection protocol.17
Therefore they can be used to expand the scope of the
method to reactions between two small molecules. To
access new classes of linker conjugates and thereby
apply the concept of photocleavable linkers to a broader
range of reactants, the dinucleotides 1±3 were
designed containing the following features in one molecule
(Fig. 1).
² Two cytidine residues and a 50-terminal phosphate group
for the T4 RNA ligase-mediated attachment to RNA
transcripts or other oligonucleotides.18
² Two to ®ve hexaethylene glycol (HEG) units to ensure
optimal positioning of RNA and coupled reactant. For
example, ligating linker 3a containing 18 ethylene glycol
units to the acceptor stem of a typical tRNA should allow
the attached reagent to interact with the anticodon loop,
Figure 1. Design of the multifunctional dinucleotide analogs 1±3 containing the 50-pCC residues for T4 RNA ligation, the photocleavable o-nitrobenzyl
group, up to ®ve hexaethylene glycol units and either a biotin, amino or carboxyl moiety at the 30-terminus.

È
F. Hausch, A. Jaschke / Tetrahedron 57 (2001) 1261±1268
1263
Figure 2. MALDI-TOF spectrum of the dinucleotide 2a in the positive detection mode using hydroxypicolinic acid as a matrix with added NH14 -ion exchanger
beads and excitation at 355 nm. The inset shows the IR-MALDI-TOF using succinic acid as matrix with excitation at 2940 nm.
attachment of potential reactants as carboxylic acids
or primary amines, respectively.
dinucleotide conjugates is the coupling of suitably activated
esters to the amino group of the dinucleotides 2a or 2b. This
was exempli®ed by the biotinylation of 2a with com-
mercially available sulfosuccinimidyl (biotinamido)
hexanoate. The reaction proceeded to near completion as
detected by HPLC analysis (data not shown).
The dinucleotide analogs were assembled by automated
synthesis using commercially available building blocks.
The photocleavable phosphoramidite 4 used to generate
the photocleavage site was synthesized in four steps
with 40% overall yield starting from o-nitrobenzalde-
hyde.21 The carboxyl-generating solid support 5 was
synthesized by immobilizing DMT-protected 4-hydroxy
butyric acid on an amino-functionalized CPG matrix.
After capping and detritylation, the 4-hydroxyl group
was esteri®ed with another 4-DMTO butyric acid via
the mixed anhydride procedure.22
While for some commonly used reagents preactivated
derivatives are commercially available (e.g. biotin, ¯uores-
cein, protected amino acids), for most target compounds the
appropriate NHS esters have to be prepared from the corre-
sponding carboxylic acids prior to conjugation with the
dinucleotide. In situ activation approaches24 with TSTU,
HBTU or EDC were not satisfying and led to complex reac-
tion mixtures without signi®cant amounts of the desired
products.
After the automated synthesis of the dinucleotide analogs,
the precursors of 3a and 3b were ®rst cleaved from the solid
support with aqueous TEA prior to the standard deprotec-
tion procedure to avoid amide formation. After desilylation,
the dinucleotides 1±3 were puri®ed by preparative HPLC
(shown for 3a in Fig. 4b) with 5±20% isolated yield and
characterized by MALDI-TOF mass spectroscopy. A typi-
cal spectrum of 2a is shown in Fig. 2 where the predicted
molecular mass peak at m/z 1789.18 g/mol can clearly be
detected. Since the UV excitation of the matrix during
MALDI-TOF measurements is nearly identical to the photo-
cleavage conditions of the o-nitrobenzyl moiety, two
additional peaks corresponding to the two expected photo-
fragments were observed, thereby con®rming the anti-
cipated cleavage of the photoactive o-nitrobenzyl unit.
Representatively, the integrity of dinucleotide 2a was
checked by IR-MALDI-TOF23 and the detection of the
correct molecular mass of [MH1] of 1787.3 g/mol clearly
validated the UV-MALDI-TOF measurements. As shown in
the inset of Fig. 2 no photofragmentation or other
decomposition was observed upon IR excitation.
A general procedure for the derivatization of the dinucleo-
tide therefore consists in a separate activation of carboxylic
acids followed by coupling to the amino-functionalized
dinucleotide analogs. This was demonstrated by the conju-
gation of the nucleobases adenine, nicotinic acid and orotic
acid to 2b (Fig. 3). With these conjugates in hand, RNA
catalyzed nucleobase/nucleotide metabolism reactions
could be investigated which are fundamental to a hypo-
thetical RNA world.25 For the coupling of adenine, the car-
boxyl moiety ®rst had to be introduced by reacting 6-chloro
purine 6 with caproic acid 7 to yield 8a26. The three nucleo-
bases 8a±c were then activated with DCC in DMF (HMPTA
for 8a due to superior solubility) followed by a conversion
to the corresponding NHS esters after the addition of
N-hydroxy succinimide. The dicyclohexyl urea byproduct
was precipitated and the activated esters 9a±c were puri®ed
by crystallization from petroleum ether:isopropanol (6:1).
All products were characterized by mass spectroscopy and
NMR which revealed signi®cant traces of unreacted orotic
acid (35%) in 9c, possibly due to hydrolysis during puri®ca-
tion. Since this did not inhibit the following derivatization
experiments, further puri®cation was not necessary. The
reactivity of the activated esters towards hydrolysis
2.2. Coupling of the dinucleotide analogs
The most straightforward way to generate derivatized

È
F. Hausch, A. Jaschke / Tetrahedron 57 (2001) 1261±1268
1264
Figure 3. (a) Synthesis and N-hydroxysuccinimide (NHS) activation of the nucleobases N-(6-purinyl)-caproic acid (8a), nicotinic acid (8b) and orotic acid
(8c); (b) HPLC puri®cation of the coupling reactions of the nucleobase±NHS esters 9a±c to the amino-functionalized dinucleotide 2b (gradient B).
increased in the order 9a,9b,9c. Whereas 9a was moder-
ately stable in aqueous neutral solutions (e.g. HPLC
buffers), rapid precipitation was observed during the
coupling of 9c indicating very fast hydrolysis. When stored
at 2208C under argon, no decomposition was detected for
any of the NHS esters over several months.
functional groups interfering with the coupling chemistry or
they might lack the carboxyl group necessary for conjuga-
tion. Inherent instability might also preclude a coupling
procedure as in Fig. 3. To address these classes of
compounds, the dinucleotide analogs 3a and 3b were
designed and synthesized carrying a 30-terminal carboxyl
group. Preliminary experiments27 have shown that this func-
tionality can be used for conjugation to primary amines.
Since in this approach the amine component is now in
large excess, the limiting carboxyl group is amenable to in
situ activation with EDC at pH 6. This was demonstrated by
coupling the dinucleotide 3a to N6-(6-aminohexyl)-AMP
11b and to 5-(biotinamido)-pentylamine 11c. Fig. 4b
depicts the HPLC chromatograms of the starting material
3a and of the products after pre-puri®cation by PAGE and
desalting on a Sephadex G10 column. To our surprise, no
side reactions occurred at the 50-phoshate group of 3a as
suggested by the literature.28 This was corroborated by
identifying the correct photofragments of the products 12b
and 12c by MALDI-TOF mass spectroscopy. For reactants
susceptible to decomposition the reaction times can be
shortened as was demonstrated by the coupling of the
ATP analog 11a to 3a on an analytical scale (Fig. 4c).
During the coupling reactions an approximately 40-fold
excess of activated ester ensured over 80% conversion of
the dinucleotide 2b. As shown in Fig. 3b, conjugation with
adenine or nicotinic acid resulted in signi®cantly higher
retention times during HPLC puri®cation. Besides the
desired products and traces of unreacted dinucleotides, the
main byproducts were identi®ed as hydrolyzed or quenched
esters of 9a±c and NHS. Because the orotic acid conjugate
and the underivatized dinucleotide analog 2b have similar
retention times, traces of the unreacted starting material
could not be resolved. Additional HPLC analysis of isolated
10c and co-injection with unmodi®ed dinucleotide 2b,
however, veri®ed a conjugation to orotic acid. Preparative
HPLC yielded ca. 50% of isolated product under various
reaction conditions and scales. The correct identities of
the isolated products were con®rmed by MALDI-TOF
mass spectroscopy. Due to the speci®c photofragmentation
pattern side reactions at the cytidine residues could be
excluded.
To check the applicability of the multifunctional linkers 1±3
for the generation of novel RNA reactant conjugate
libraries, the derivatized dinucleotides 10a±c were ligated
overnight to
a randomized 157nt RNA pool with
2.3. Conjugation with complex molecules
approximately 80% ef®ciency as determined by a gel shift
during PAGE puri®cation. No in¯uence of the length of the
PEG spacer or the conjugated 30-terminal reactant on
the ligation ef®ciency was observed. A deoxy version of
Many reagents of interest are neither amenable to a separate
activation as NHS-esters nor are they suited for in situ acti-
vation. Especially natural products may contain additional

È
F. Hausch, A. Jaschke / Tetrahedron 57 (2001) 1261±1268
1265
Figure 4. (a) Primary amines 11a±c used for in situ coupling to the carboxy-terminated dinucleotide 3a; (b) HPLC puri®cation of the crude dinucleotide 3a
after automated synthesis (top) and its EDC-mediated conjugation products with the primary amines 11b (middle) and 11c (bottom) after PAGE and size
exclusion chromatography; (c) Autoradiography of a denaturing 20% polyacrylamide gel analyzing the EDC conjugation products of 3a. Lane 1 with 12a for
2 h, lane 2 with 12c for 25 min and lane 3 with 12c for 2 h.
the dinucleotide analogs containing two deoxycytidines
(dC) instead of ribocytidines (rC) at the 50-end could also
be ligated to RNA pools, albeit with only half the ef®ciency
compared to the RNA linker (data not shown).
20-OH of the riboses or exocyclic amines of the nucleo-
bases.29
Suitably derivatized dinucleotides (e.g. with puromycin)
may also improve stringency in peptide selection experi-
ments based on RNA±peptide fusions.30 In combination
with recently described multifunctional primers, the
presented coupling chemistry can be applied to the genera-
tion of various DNA conjugates using appropriate PCR
protocols.31
With the RNA libraries conjugated to the dinucleotides
10a±c in hand, RNA-catalyzed nucleobase/nucleotide
metabolism reactions can be investigated which are
fundamental to a hypothetical RNA world. These nucleo-
base±RNA conjugates are currently being used in selection
experiments for ribozymes synthesizing their own building
blocks by catalyzing N-glycosidic bond formations. Along
similar lines, AMP/ATP±RNA conjugates like 12a and 12b
might allow selection experiments for ATP-generating
ribozymes which are crucial for activating these building
blocks, e.g. for subsequent RNA-catalyzed polymeriza-
tion.25
3. Conclusion
The dinucleotide analogs described here provide a straight-
forward and generally applicable way for site-speci®c
derivatization of larger RNA transcripts. This allows for
the expansion of the in vitro selection protocol to new,
previously inaccessible reactions. The desired reactants
can be coupled to the dinucleotides either via NHS-
activated carboxyl groups or with lower yields as primary
amines by in situ EDC-mediated condensation. The
dinucleotide analogs presented in this work should serve
as valuable tools in a variety of in vitro selection experi-
ments as well as in RNA interaction studies.
Recently, the strategy of RNA±reactant conjugates coupled
via photocleavable linkers was also applied to the selection
of acyl-transfer catalysts. The isolated ribozymes
accelerated the aminoacylation of small, external oligo-
nucleotides. These ®ndings demonstrated the regioselective
selection pressure exerted by the photocleavable linker
since the aminoacylation of the single 30-terminal hydroxyl
group was favored in the presence of 160 competing internal

È
F. Hausch, A. Jaschke / Tetrahedron 57 (2001) 1261±1268
1266
4. Experimental
Na2CO3 and 1.00 g (7.0 mmol) 6-chloro purine 6 in 10 ml
water was adjusted to pH 9.5±10 by addition of 1 ml HCl
(18%). The mixture dissolved upon re¯uxing and after 1 h
the pH was readjusted with additional Na2CO3. After 3 h of
re¯uxing the reaction mixture was cooled to 08C, the
product was precipitated by addition of 10 ml acetic acid
and recovered by ®ltration. The crude product was puri®ed
twice by dissolving in 0.1 M NaOH and precipitating with
acetic acid to yield 0.85 g (3.4 mmol, 50%) of HPLC-
analyzed pure product 8a after drying in vacuo. Rf (silica
gel)ˆ0.51 (CH2Cl2:MeOH:CF3COOHˆ8:1.5:0.5). HPLC
4.1. General
All organic reagents were purchased from Aldrich unless
otherwise noted. Reversed phase HPLC was performed on
a C-18 column (Beckman) with a 0.1 M triethylammonium
acetate pH 7 (TEAAc) buffer and increasing percentage of
acetonitrile (gradient A: 0.8%±16% in 15 min and 16%±
24% in 20 min or gradient B: 0.8%±18% in 5 min and
18%±21.2% in 20 min, both followed by an 80% aceto-
nitrile wash). Elution pro®les were detected by measuring
the absorbance at 270 nm.
1
(gradient A)ˆ16, 74 min. H NMR (250 MHz, 4% NaOD/
D2O): dˆ1.39±1.5 (quintet, 2H, Jˆ7.5 Hz); 1.59±1.78 (m,
4H); 2.24±2.3 (t, 2H, Jˆ7.5 Hz); 3.53±3.58 (t, 2H,
Jˆ7.5 Hz); 8.01 (s, 1H); 8.21 (s, 1H). 13C NMR
(62.9 MHz, DEPTˆ1/0/2): dˆ28.40 (2); 28.88 (2);
31.54 (2); 40.32 (2); 43.44 (2); 122.69 (0); 152.24 (1);
154.20 (1); 156.43 (0); 160.70 (0); 186.34 (0). MS (EI):
m/zˆ249 (24.5%, M1z); 190 (36%); 162 (35%); 148
(100%); 135 (45%).
MALDI-TOF measurements were performed on a Bruker
Re¯ex spectrometer using hydroxypicolinic acid as a matrix
containing additional NH14 -ion exchanger beads. After
excitation with a Nd±YAG laser at 355 nm, cations were
detected by time-of-¯ight measurement. Alternatively, 2,6-
dihydroxybenzoic acid was used as a matrix in the negative
detection mode with excitation at 377 nm. o-Nitrobenzyl
containing samples were prone to photodegradation result-
ing in a characteristic loss of 16 mass units thereby obscur-
ing the exact determination of these peaks.
4.1.4. N-(6-Purinyl) caproyl-N-oxysuccinimide (9a).
350 mg (1.4 mmol) N-(6-purinyl) caproic acid (8a) and
176 mg (1.5 mmol) N-hydroxy succinimide were dissolved
in 1 ml of dry hexamethylphosphortriamide (HMPTA)
under argon. The reaction was started by addition of
0.31 g (1.5 mmol) DCC at 08C. After leaving the reaction
overnight at room temperature the dicyclohexyl urea by-
product was ®ltered off and rinsed with 0.1 ml HMPTA.
The crude product was precipitated with 10 ml petroleum
ether and 1.2 ml isopropanol for 2 h at room temperature
and recovered by ®ltration. This residue was redissolved in
4 ml acetonitrile:acetone (1:1) and recrystallized with 7 ml
petroleum ether:isopropanol (6:1) to yield 90 mg 9a
(0.26 mmol, 18%) as a white solid after drying in vacuo.
1H NMR (250 MHz, DMSO-d6): dˆ1.34±1.46 (2H); 1.54±
1.7 (m, 4H); 2.57 (s, 1H); 2.63±2.69 (t, 2H, Jˆ7.5 Hz); 2.78
(s, 4H); 3.44 (br, 2H); 7.48 (s, 1H); 8.06 (s, 1H); 8.14 (s,
1H). 13C NMR (62.9 MHz): dˆ23.99; 25.43; 28.66; 30.14;
33.64; 36.45; 118.66; 138.54; 149.45; 152.36; 154.46;
168.97; 170.25. MS (FAB): m/zˆ347 (M1H1).
4.1.1. Orotidyl-N-oxy-succinimide (9c). 1.20 g (10.4 mmol)
N-hydroxy-succinimide (NHS) and 1.53 g (10.0 mmol)
orotic acid 8c (Fluka) were dissolved in 25 ml of dry DMF
in an argon atmosphere. After addition of 2.26 g (11.0 mmol)
dicyclohexyl carbodiimide (DCC) the solution was stirred
overnight at room temperature. The precipitated solid was
®ltered off, the ®lter was rinsed with 5 ml DMF, and the
®ltrate was diluted with 180 ml of a mixture of petroleum
ether (Merck, 40±608C) and isopropanol (6:1). After 2 h at
room temperature the crude product was recovered by ®ltra-
tion, redissolved in 5 ml DMF and recrystallized from 31 ml
petroleum ether: isopropanol (6:1) to yield 0.71 g of 9c
1
(2.8 mmol, 28%) as a white solid after drying in vacuo. H
NMR (250 MHz, DMSO-d6): dˆ2.84 (s, 4H, NHS); 6.3 (s,
1H, CH); 11.58 (s, 2H, NH); traces of orotic acid and NHS at
dˆ2.58; 6.0; 10.82; 11.3. 13C NMR (62.9 MHz, DMSO-d6):
dˆ25.64; 106.54; 137.14; 150.74; 157.09; 163.29; 169.73,
traces of orotic acid: 25.31; 103.27; 142.67; 150.95; 161.85;
164.18; 172.90.
4.2. Synthesis of the dinucleotide analogs
The dinucleotide analogs 1a and 1b were synthesized on a
Pharmacia Gene Assembler on a 1.3 mmol DMT-on scale
using phosphoramidites from Chemgenes, a biotinylated
solid support from Glen Research and the photocleavable
building block 4. Coupling times were 90 s except for the
cytidine residues (720 s) and the chemical phosphorylation
(3£90 s).
4.1.2. Nicotinyl-N-oxy-succinimide (9b). 1.45 g (12.6 mmol)
N-hydroxy succinimide and 1.48 g (12.0 mmol) nicotinic acid
8b (Sigma) were dissolved in 25 ml dry DMF under argon and
treated with 2.7 g (13.0 mmol) DCC overnight at room
temperature. The precipitate was ®ltered off, the ®ltrate was
concentrated by rotary evaporation and crystallized from
petroleum ether:isopropanol (6:1) at room temperature. The
white solid was recovered by ®ltration and dried under high
The dinucleotide analogs were deprotected in 1 ml 33%
NH4OH/ethanol (3:1) overnight at 558C. After lyophyliza-
tion, the samples were treated with 300 ml 1 M nBu4NF in
THF for 48 h, quenched with 0.3 ml 2 M TEAAc (pH 7) and
diluted to 10 ml with water. The crude product was passed
over a 1.5 ml Sephadex A25 column (pre-equilibrated with
10 ml 0.05 M TEAAc), washed with 15 ml 0.1 M TEAAc
and eluted with 2 M TEAAc. The UV active fractions were
pooled, lyophylized and puri®ed by HPLC. The lyophilized
products were quanti®ed by UV spectroscopy at 270 nm and
characterized by UV-MALDI-TOF mass spectroscopy.
1
vacuum to yield 1.67 g (63%) of the desired product 9b. H
NMR (250 MHz, DMSO-d6): dˆ2.89 (s, 4H, NHS); 7.64±
7.72 (m, 1H); 8.42±8.49 (m, 1H); 8.92±8.98 (m, 1H); 9.28±
9.3 (m, 1H). 13C NMR (62.9 MHz, DEPTˆ1/0/2): dˆ25.60
(2); 121.08 (0); 124.49 (1); 137.77 (1); 150.54 (1); 155.65
(1); 161.01 (0); 170.21 (0). MS (EI): m/zˆ220 (2.3%, M1z);
106 (100%, M±NHS1z); 78 (50%, M±NHS±CO1z).
4.1.3. N-(6-Purinyl)-caproic acid (8a). A suspension of
1.67 g (12.7 mmol) caproic acid 7, 0.77 g (7.0 mmol)

È
F. Hausch, A. Jaschke / Tetrahedron 57 (2001) 1261±1268
1267
1a: 2.4 A270 (125 nmol, 10%); Rtˆ31.6 min (gradient A);
mass (m/z)ˆ2488.8 (calculated [M±H]2ˆ2487.6 g/mol);
30-terminal photofragment (m/z)ˆ931.8 (calculatedˆ
930.6 g/mol).
at 378C. The product was puri®ed with 20% denaturing
PAGE, visualized by autoradiography, eluted in water,
concentrated by lyophylization and desalted on a 0.8 ml
Sephadex G10-column.
4.2.2. Derivatization of the dinucleotide 2b. The conju-
gates 10a±c were generally synthesized in 10 ml reactions
containing 2.4 mM dinucleotide analog 2b, 0.1 M of the
respective N-hydroxysuccinimide-activated base 9a±c,
0.4 M K2HPO4 (pH 8.0) and 20% DMF at room tempera-
ture. After 2±3 h the reactions were quenched with 2 ml
ammonium acetate (1 M) for 30 min, diluted to 100 ml
with water and puri®ed by HPLC (gradient B).
1b: 3.2 A270 (165 nmol, 13%); Rtˆ18.1 min (gradient B);
mass (m/z)ˆ3179.3, 3164.5 (calculated [M1H]1ˆ
3178.6 g/mol); 50-photofragment (m/z)ˆ1904.2, 1889.2
(calculatedˆ1903.4 g/mol); 30-biotinylated photofragment
(m/z)ˆ1277.0 (calculatedˆ1276.2 g/mol).
Dinucleotide 2a was synthesized on an Applied Biosystems
391 synthesizer on a 1 mmol DMT-on scale using a 30-
aminomodi®er-C6-CPG support from Chemgenes.
Deprotection and puri®cation were analogous to the
dinucleotides 1 yielding 5.4 A270 (286 nmol, 22%) eluting
at 21.5 min (gradient A). UV-MALDI-TOF: Mass found
(m/z)ˆ1789.18 (mass calculated [M1H]1ˆ1788.28 g/
mol); cytidine-containing photofragment (m/z)ˆ1216.33
(calculatedˆ1214.82 g/mol); amine-containing photo-
fragment (m/z)ˆ575.1 (calculatedˆ574.46 g/mol).
10a: 0.27 OD270 (14 nmol, 60%); Rtˆ17.5 min; mass found
(m/z)ˆ3052.1
(calculated
[M1H]1ˆ3051.1 g/mol);
cytidine-containing photofragment (m/z)ˆ1904.3 (calcu-
lated: 1903.4 g/mol); adenine-containing photofragment
(m/z)ˆ1149.6 (calculated: 1149.6 g/mol).
10b: 0.25 OD267 (13.2 nmol, 55%); Rtˆ16.4 min; mass
found (m/z)ˆ2925.7 (calculated [M1H]1ˆ2924.9 g/mol);
cytidine-containing photofragment (m/z)ˆ1904.1 (calcu-
lated: 1903.4 g/mol); nicotinyl-containing photofragment
(m/z)ˆ1023.9 (calculated: 1023.4 g/mol).
Dinucleotide 2b was synthesized on a Milligen 8800 DNA
Synthesizer on a 160 mmol DMT-on scale and deprotected
with 12 ml 33% NH4OH and 4 ml EtOH overnight at 558C.
The resulting lyophylized residue was desilylated in 1 ml
triethylammonium trihydro¯uoride at room temperature for
48 h, desalted on a Sephadex G10-column and puri®ed by
preparative HPLC on a C18-Eurosphere column (Knauer)
yielding 150 A270 (8 mmol, 5%) eluting at 13.3 min
(gradient B). Mass found (m/z)ˆ2821.0 (mass calculated
10c: 0.26 OD274 (13.4 nmol, 56%); Rtˆ12.8 min; mass
found (m/z)ˆ2959.0 (calculated [M1H]1ˆ2957.9 g/mol);
cytidine-containing photofragment (m/z)ˆ1904.2 (calcu-
lated: 1903.4 g/mol); orotidyl-containing photofragment
(m/z)ˆ1056.9 (calculated: 1056.4 g/mol).
[M1H]1ˆ2821.2 g/mol);
cytidine-containing
photo-
4.2.3. Derivatization of the dinucleotide 3a. 20 nmol of
partially 50-32P labeled dinucleotide 3a were treated with
2 mmol of 5-(biotinylamido)-pentylamine 11c or N6-(6-
aminohexyl)-AMP 11b and 1 mmol EDC in 4 ml 2 M
imidazole/HCl for 30 h and with 2 ml of fresh 0.25 M
EDC for another 4 h. The reaction mixtures were then
electrophoresed on a 20% denaturing polyacrylamide gel.
The crude products were identi®ed by autoradiography,
eluted with water, desalted on a 4 ml Sephadex G10 column
and puri®ed by HPLC (gradient A).
fragment (m/z)ˆ1904.2 (calculatedˆ1903.4 g/mol); amine-
containing photofragment (m/z)ˆ918.4 (calculatedˆ
918.8 g/mol).
The dinucleotides 3a and 3b were synthesized as described
for 1 starting with solid support 5. Prior to deprotection the
dinucleotide analogs 3 were ®rst cleaved from the solid
matrix in 0.75 ml H2O/TEA (2:1) for 12 h at 558C and
then deprotected with an additional 0.5 ml 33% NH4OH/
EtOH (3:1) for 6 h at 558C. Desilylation and puri®cation
were performed as for the analogs 1.
12b: 68 mOD265 (2 nmol, 10%); Rtˆ26.8 min; mass found
(AMP-containing photofragment): m/zˆ955.7 (calculated
[M±H]2ˆ955.5 g/mol).
3a: 4 A270 (200 nmol, 15%); Rtˆ23.6 min (gradient A);
mass (m/z)ˆ2084.8 (calculated [M±H]2ˆ2084.1 g/mol);
30-terminal photofragment (m/z)ˆ527.4 (calculatedˆ
527.1 g/mol).
12c: 0.12 OD265 (6.25 nmol, 30%); Rtˆ28.9 min; mass
found (biotinylated 30-photofragment): m/zˆ837.5 (calcu-
lated [M±H]2ˆ837.6 g/mol).
3b: 2.4 A273 (127 nmol, 10%); Rtˆ11.7 min (gradient B);
mass found (30-carboxyl-containing photofragment):
m/zˆ874.1; (calculated [M1H]1ˆ873.7 g/mol).
4.3. Transcription of the RNA pool and ligation to the
dinucleotide derivatives
4.2.1. Radioactive labeling of the dinucleotide 3a.
300 pmol of the dinucleotide 3a were dephosphorylated
by 2 U shrimp alkaline phosphatase (USB) in 10 ml buffer
(20 mM Tris±HCl, pH 8.0, 20 mM MgCl2) for 1 h at 378C,
then mixed with 1.5 ml 0.1 M EDTA and heated to 808C for
10 min. After cooling, the reaction mixture was diluted with
2.5 ml T4 kinase buffer (10£, MBI Fermentas), 1.5 ml
MgCl2 0.1 M, 2.5 ml dithiothreitol (DTT) 35 mM, 5 ml
g-32P ATP (Amersham, 10 mCi/ml) and 2 ml T4 polynucleo-
tide kinase (10 U/ml, MBI Fermentas) and incubated for 2 h
The 157nt-RNA was synthesized by transcription of
10 pmol double-stranded DNA template in a 200 ml reac-
tion containing 80 mM Hepes pH 7.5, 22 mM MgCl2, 1 mM
spermidine, 10 mM DDT, 0.2 mg/ml BSA, 4 mM of each
nucleoside triphosphate (Roche), 1 U pyrophosphatase
(Sigma), 100 mCi a-32P CTP (Amersham) and 200 U T7
RNA polymerase. After 4 h at 378C the DNA template
was digested with 40 U DNase I (Roche) for 30 min at
378C. The RNA was puri®ed by denaturing PAGE (5%).

È
F. Hausch, A. Jaschke / Tetrahedron 57 (2001) 1261±1268
1268
8. Wilson, C.; Szostak, J. W. Nature 1995, 374, 777±782.
9. Jenne, A.; Famulok, M. Chem. Biol. 1998, 5, 23±34.
Full-length RNA (157 nt) was identi®ed by auto-
radiography, eluted with 0.3 M NaOAc pH 5.5 and precipi-
tated with ethanol, yielding 3.2 nmol. To suppress folding of
the randomized RNA pool into undesired secondary struc-
tures interfering with ligation, the 30-terminus was routinely
annealed to 4 nmol of a complementary oligonucleotide at
958C. For RNA substrates with accessible 30-ends, this
hybridization was not essential. Ligation of the hybridized
RNA was performed in a 200 ml reaction with 50 mM Tris±
HCl pH 7.8, 20 mM MgCl2, 10 mM DTT, 20 mg/ml bovine
serum albumin, 4 mM adenosine triphosphate (MBI), 10%
DMSO, 100 mM of the respective dinucleotide derivative
10a±c and 300 U T4 RNA ligase (MBI) with 70% conver-
sion. Electrophoresis on a 5% denaturing polyacrylamide
gel yielded 0.9 nmol (30%) of the isolated ligation product.
È
10. Jaschke, A. Catalysis of organic reactions by RNAÐ
strategies for the selection of catalytic RNAs. In The Many
Faces of RNA, Eggleston, D. S., Prescott, C. D., Pearson, N. D.,
Eds.; Academic Press: San Diego, 1998; pp 179±190.
11. Zhang, B.; Cech, T. R. Nature 1997, 390, 96±100.
12. Unrau, P. J.; Bartel, D. P. Nature 1998, 395, 260±263.
13. Tarasow, T. M.; Tarasow, S. L.; Eaton, B. E. Nature 1997,
389, 54±57.
È
14. Seelig, B.; Jaschke, A. Chem. Biol. 1999, 6, 167±176.
15. Beaucage, S. L.; Iyer, R. P. Tetrahedron 1993, 49, 1925±
1963.
16. Beaucage, S. L.; Iyer, R. P. Tetrahedron 1993, 49, 6123±
6194.
17. Hausch, F.; Jaschke, A. Bioconjugate Chem. 1997, 8, 885±
890.
È
18. Igloi, G. Anal. Biochem. 1996, 233, 124±129.
19. Wong, J. C.; Kuhl, T. C.; Israelachvili, J. N.; Mullah, N.;
Zalipsky, S. Science 1997, 275, 820±822.
Acknowledgements
This work was supported by the Deutsche Forschungs-
gemeinschaft (Ja 794/2 and SFB 344-B10 to A. J.). We
È
are indebted to C. Frauendorf and Dr. W. Schroder for tech-
nical assistance and to Drs E. Nordhoff, P. Franke, S.
È
È
20. Jaschke, A.; Furste, J. P.; Cech, D.; Erdmann, V. A.
Tetrahedron Lett. 1993, 34, 301±304.
21. Ordoukhanian, P.; Taylor, J. S. J. Am. Chem. Soc. 1995, 117,
9570±9571.
È
Hahner, A. Schafer and O. Holzmann for spectroscopic
È
22. Hovinen, J.; Guzaev, A.; Azhayev, A.; Lonnberg, H.
measurements. The authors wish to thank Professor V. A.
Erdmann for additional support.
Tetrahedron Lett. 1993, 34, 8169±8172.
23. Nordhoff, E.; Cramer, R.; Karas, M.; Hillenkamp, F.;
Kirpekar, F.; Kristiansen, K.; Roepstorff, P. Nucleic Acids
Res. 1993, 21, 3347±3357.
References
24. Bannwarth, W.; Knorr, R. Tetrahedron Lett. 1991, 32, 1157±
1160.
1. Goodchild, J. Bioconjugate Chem. 1990, 1, 165±187.
2. Wojczewski, C.; Stolze, K.; Engels, J. W. Synlett 1999, 10,
1667±1678.
25. Bartel, D. P.; Unrau, P. J. Trends Biol. Sci. 1999, 24, M9±
M13.
È
È
26. Schafer, K.; Bullesbach, E. E.; Jolles, P.; Zahn, H. Biol. Chem.
Hoppe-Seyler 1986, 367, 757±768.
3. Trawick, B. N.; Daniher, A. T.; Bashkin, J. K. Chem. Rev.
1998, 98, 939±960.
È
27. Hausch, F.; Jaschke, A. Tetrahedron Lett. 1998, 39, 6157±
6158.
4. Akhtar, S.; Agrawal, S. Trends in Pharmacol. Sci. 1997, 18,
12±18.
28. Chu, B. C. F.; Wahl, G. M.; Orgel, L. E. Nucleic Acids Res.
1983, 11, 6513±6529.
5. Niemeyer, C. M.; Blohm, D. Angew. Chem., Int. Ed. Engl.
1999, 38, 2865±2869.
È
29. Ruppert, T.; Jaschke, A. Manuscript in preparation.
È
6. Frauendorf, C.; Jaschke, A. Angew. Chem., Int. Ed. Engl.
1998, 37, 1378±1381.
30. Roberts, R. W.; Szostak, J. W. Proc. Natl. Acad. Sci. USA
1997, 94, 12297±12302.
È
31. Hausch, F.; Jaschke, A. Nucleic Acids Res. 2000, 28, e35.
È
7. Jaschke, A.; Seelig, B. Curr. Opin. Chem. Biol. 2000, 4, 257±
262.