Metal-free catalysts for the hydrolysis of RNA derived from guanidines, 2-aminopyridines, and 2-aminobenzimidazoles

Article abstract of DOI:10.1021/ja0443934

2-Aminopyridine and 2-aminobenzimidazole were chosen as structural analogues to substitute guanidinium groups in receptor molecules designed as phosphoryl transfer catalysts. Shifting the pK_a of the guanidinium analogues toward 7 was expected t

Full text of DOI:10.1021/ja0443934

Published on Web 01/27/2005
Metal-Free Catalysts for the Hydrolysis of RNA Derived from
Guanidines, 2-Aminopyridines, and 2-Aminobenzimidazoles
Ute Scheffer, Andreas Strick, Verena Ludwig, Sascha Peter, Elisabeth Kalden, and
Michael W. Go¨bel*
Contribution from the Institute for Organic Chemistry and Chemical Biology,
Goethe-UniVersity Frankfurt, Marie-Curie-Str. 11, 60439 Frankfurt am Main, Germany
Received September 15, 2004; E-mail: M.Goebel@chemie.uni-frankfurt.de
Abstract: 2-Aminopyridine and 2-aminobenzimidazole were chosen as structural analogues to substitute
guanidinium groups in receptor molecules designed as phosphoryl transfer catalysts. Shifting the pKa of
the guanidinium analogues toward 7 was expected to raise catalytic activities in aqueous buffer. Although
the pKa’s of both heterocycles are similar (6.2 and 7.0), only 2-aminobenzimidazole led to active RNA
cleavers. All cleavage assays were run with fluorescently labeled substrates and a DNA sequencer. RNase
contaminations would degrade RNA enantioselectively. In contrast, achiral catalysts such as 9b and 10b
necessarily induce identical cleavage patterns in RNA and its mirror image. This principle allowed us to
safely rule out contamination effects in this study. The most active catalysts, tris(2-aminobenzimidazoles)
9b and 10b, were shown by fluorescence correlation spectroscopy (FCS) to aggregate with oligonucleotides.
However, at very low concentrations the compounds are still active in the nonaggregated state. Conjugates
of 10b with antisense oligonucleotides or RNA binding peptides, therefore, will be promising candidates as
site specific artificial ribonucleases.
Scheme 1
Introduction
The chemistry of artificial phosphodiesterases and ribo-
nucleases is a challenging field of research due to the high
kinetic stability of phosphodiester ions against nucleophilic
attack.1 This functional group is known for reaction barriers up
to 100 kcal mol-1 in vacuo.1d,2 The remarkable value results
from charge repulsion, e.g., between dimethyl phosphate and
the methoxide ion.2b,c In aqueous solution, the dipole moment
of water efficiently stabilizes the transition state of phosphate
substitution. This phenomenon, electrophilic catalysis, is pre-
dicted by calculations and observed in many experiments.2b-e
Increased positive charge density further enhances the catalytic
power of electrophiles. Accordingly, lanthanide ions are among
the most successful artificial ribonucleases known today.1b In
aqueous solution around pH 7, the dominant nucleophilic species
are water and neutral hydroxy groups, respectively. General
bases to activate nucleophiles and general acids to protonate
leaving groups, therefore, may further contribute to phosphoryl
transfer catalysis.1c
and phenylethanol in DMF (Scheme 1).3a Two guanidinium
groups, when connected by appropriate spacers, showed sig-
nificant cooperativity in this process. Interestingly, the substitu-
tion of guanidinium ions in compound 1b by heterocyclic
analogues (2b) led to a 10-fold rate increase (Chart 1).
Comparable effects have been consistently observed in several
reactions. The enhanced catalytic power of 2-aminoimidazo-
linium has been attributed to the increased acidity of this ion
compared to guanidinium.3b Increased acidity reinforces hy-
drogen bonds with the substrate and improves transition state
stabilization. In addition, it enables the cation to participate in
general acid/base catalysis.4
Some time ago, we have studied the catalytic influence of
guanidinium ions on the reaction of catechol cyclic phosphate
Bis(guanidinium) compounds have been described by several
laboratories as useful phosphate receptors and transesterification
catalysts in nonaqueous solution.3,5 In water, however, high rate
increases are rarely observed.6 Furthermore, the rate advantage
(1) (a) Perreault, D. M.; Anslyn, E. V. Angew Chem., Int. Ed. Engl. 1997, 36,
432. (b) Trawick, B.; Daniher, A. T.; Bashkin, J. K. Chem. ReV. 1998, 98,
939. (c) Oivanen, M.; Kuusela, S.; Lo¨nnberg, H. Chem. ReV. 1998, 98,
961. (d) Zhou, D.-M.; Taira, K. Chem. ReV. 1998, 98, 991. (e) Kuimelis,
R. G.; McLaughlin, L. W. Chem. ReV. 1998, 98, 1027.
(2) (a) Lim, C.; Tole, P. J. Am. Chem. Soc. 1992, 114, 7245. (b) Dejaegere,
A.; Karplus, M. J. Am. Chem. Soc. 1993, 115, 5316. (c) Dejaegere, A.;
Liang, X.; Karplus, M. J. Chem. Soc., Faraday Trans. 1994, 90, 1763. (d)
Yliniemela, A.; Uchimaru, T.; Tanabe, K.; Uebayasi, M.; Taira, M. J. Am.
Chem. Soc. 1993, 115, 3032. (e) Taira, M.; Uchimaru, T.; Storer, J. W.;
Yliniemela, A.; Uebayasi, M.; Tanabe, K. J. Org. Chem. 1993, 58, 3009.
(f) Uchimaru, T.; Stec, W. J.; Taira, K. J. Org. Chem. 1997, 62, 5793.
(3) (a) Gross, R.; Du¨rner, G.; Go¨bel, M. W. Liebigs Ann. Chem. 1994, 49. (b)
Muche, M.-S.; Go¨bel, M. W. Angew. Chem., Int. Ed. Engl. 1996, 35, 2126.
(c) Kurz, K.; Go¨bel, M. W. HelV. Chim. Acta 1996, 79, 1967.
(4) By proton inventory studies, it has been shown recently that even
unmodified guanidinium ions are sufficiently acidic to protonate the
dianionic phosphorane transition state or intermediate of phosphoryl transfer
reactions: Piatek, A. M.; Gray, M.; Anslyn, E. V. J. Am. Chem. Soc. 2004,
126, 9878.
9
10.1021/ja0443934 CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 2211-2217
2211

A R T I C L E S
Scheffer et al.
Chart 1. Structures of Compounds 1-10
the solvent change from DMF to water. In addition, water itself
is a good electrophile due to its high polarity making it difficult
to surpass aqueous transition state solvation by guanidinium
groups. Finally, the protonation equilibrium of guanidines in
water is shifted toward the cation. Both compounds 1b and 2b,
as a result, are not sufficiently acidic to participate in proton
transfer steps. The remaining activity of 1b and 2b as pure
electrophilic catalysts is too weak to generate important rate
effects. We will show below that shifting the pKa of the cationic
groups toward 7 leads to promising synthetic RNA cleavers.
Increasing the number of charged groups gives a further boost
to catalytic power.
Results and Discussion
Synthesis of Catalysts. Derivatives of 2-aminopyridine were
obtained by heating the diamine 11 with 2-bromopyridine 12.
Subsequent reaction with di-tert-butyl dicarbonate led to a
mixture of bis-2-aminopyridine 4 (12%) and compound 13
(52%) that was transformed into product 3a by reagent 143b
(95%). 2-Aminobenzimidazoles are accessible by reaction of
the corresponding amines with thiocarbonyl diimidazole fol-
lowed by a second substitution with 1,2-phenylene diamine.7
A thiourea is formed in this step. It undergoes smooth cyclization
in the presence of HgO.7,8 Thus, benzylamine 15 could be
transformed into picrate 5a in 49% total yield. Analogous
procedures were used to prepare the catalyst picrates 6a and 8a
as depicted in Scheme 3. In the case of compound 7, the
benzimidazole moiety could be attached by heating the amine
precursor in the presence of 2-chlorobenzimidazole. However,
this procedure required high reaction temperatures and led to
products with insufficient yield and purity (20% after HPLC
separation).
Tris(2-aminoethyl)amine 21 (TREN) has been often used as
a building block in coordination chemistry. It also forms the
framework of several anion receptors based on guanidines and
ureas.9 For the synthesis of compound 9a (Scheme 4), TREN
Scheme 2
(6) For a Zn2+-bis(guanidinium) catalyst cleaving ApA, see: (a) Ait-Haddou,
H.; Sumaoka, J.; Wiskur, S. L.; Folmer-Andersen, F. J.; Anslyn, E. V.
Angew. Chem., Int. Ed. 2002, 41, 4014. RNA cleavers based on amines:
(b) Yoshinari, K.; Yamazaki, K.; Komiyama, M. J. Am. Chem. Soc. 1991,
113, 5899. (c) Komiyama, M.; Yoshinari, K. J. Org. Chem. 1997, 62, 2155.
(d) Verheijen, J. C.; Deiman, B. A. L. M.; Yeheskiely, E.; van der Marel,
G. A.; van Boom, J. H. Angew. Chem., Int. Ed. 2000, 39, 369. (e) Petersen,
L.; de Koning, M. C.; van Kuik-Romeijn, P.; Weterings, J.; Pol, C. J.;
Platenburg, C.; Overhand, M.; van der Marel, G. A.; van Boom, J. H.
Bioconjugate Chem. 2004, 15, 576. (f) Michaelis, K.; Kalesse M. Angew.
Chem., Int. Ed. 1999, 38, 2243. (g) Michaelis, K.; Kalesse M. ChemBio-
Chem. 2001, 1, 79. (h) Scarso, A.; Scheffer, U.; Go¨bel, M.; Broxterman,
Q. B.; Kaptein, B.; Formaggio, F.; Toniolo, C.; Scrimin, P. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 5144. For imidazole based RNA cleavers see:
(i) Beloglazova, N. G.; Fabani, M. M.; Zenkova, M. A.; Bichenkova, E.
V.; Polushin, N. N.; Silnikov, V. V.; Douglas, K. T.; Vlassov, V. V. Nucleic
Acids Res. 2004, 32, 3887. (j) Fouace, S.; Gaudin, C.; Picard, S.; Corvaisier,
S.; Renault, J.; Carboni, B.; Felden, B. Nucleic Acids Res. 2004, 32, 151.
For RNA cleaving peptides, see: (k) Mironova, N. L.; Pyshnyi, D. V.;
Ivanova, E. M. In Artificial Nucleases; Zenkova, M. A., Ed.; Springer:
Berlin, 2004; p 151.
of compound 2b over 1b disappears when applied to RNA
hydrolysis (Scheme 2). For an explanation, several factors have
to be considered. First of all, ion pair stability is reduced by
(5) (a) Jubian, V.; Dixon, R. P.; Hamilton, A. D. J. Am. Chem. Soc. 1992,
114, 1120. (b) Jubian, V.; Veronese, A.; Dixon, R. P.; Hamilton, A. D.
Angew. Chem., Int. Ed. Engl. 1995, 34, 1237. (c) Smith, J.; Ariga, K.;
Anslyn, E. V. J. Am. Chem. Soc. 1993, 115, 362. (d) Perreault, D. M.;
Cabell, L. A.; Anslyn, E. V. Bioorg. Med. Chem. 1997, 5, 1209. (e) Kato,
T.; Takeuchi, T.; Karube, I. J. Chem. Soc., Chem. Commun. 1996, 953. (f)
Oost, T.; Filippazzi, A.; Kalesse, M. Liebigs Ann. Recl. 1997, 1005. (g)
Oost, T.; Kalesse M. Tetrahedron 1997, 53, 8421. (h) Zepik, H. H.; Benner
S. A. J. Org. Chem. 1999, 64, 8080. Selected recent papers on guanidinium
carboxylate recognition: (i) Schmuck, C.; Geiger, L. J. Am. Chem. Soc.
2004, 126, 8898. (j) Linton, B.; Hamilton, A. D. Tetrahedron 1999, 6027.
(k) Linton, B. R.; Goodman, M. S.; Fan, F.; van Arman, S. A.; Hamilton,
A. D. J. Org. Chem. 2001, 66, 7313. (l) Zafar, A.; Melendez, R.; Geib, S.
J.; Hamilton, A. D. Tetrahedron 2002, 58, 683. Crystal structure of the
sulfate salt of bis(guanidinium) compound 1: (m) Hutchings, M. G.;
Grossel, M. C.; Merckel, D. A. S.; Chippendale, A. M.; Kenworthy, M.;
McGeorge, G. Cryst. Growth Des. 2001, 1, 339. Crystal structure of
2-aminobenzimidazolium nitrate: (n) Bats, J. W.; Go¨rdes, D.; Schmalz,
H.-G. Acta Crystallogr. 1999, C55, 1325.
(7) Perkins, J. J.; Zartman, A. E.; Meissner, R. S. Tetrahedron Lett. 1999, 40,
1103.
(8) Mohsen, A.; Omar, M. E.; Ragab, M. S.; Farghaly, A. M.; Barghash, A.
M. Pharmazie 1976, 31, 348.
(9) For the corresponding tris(guanidine), see: (a) Dietrich, B.; Fyles, D. L.;
Fyles, T. M.; Lehn, J.-M. HelV. Chim. Acta 1979, 62, 2763. (b) Pitsch, S.;
Scheffer, U.; Strick, A.; Go¨bel, M. W. HelV. Chim. Acta 2003, 86, 3740.
TREN derived urea derivatives as anion receptors: (c) Raposo, C.; Almaraz,
M.; Martin, M.; Weinrich, V.; Mussons, M. L.; Alcazar, V.; Cruz Caballero,
M.; Moran, J. R. Chem. Lett. 1995, 759. (d) Werner, F.; Schneider, H.-J.
HelV. Chim. Acta 2000, 83, 465. (e) Xie, H.; Yi, S.; Wu, S. J. Chem. Soc.
Perkin Trans. 2 1999, 12, 2751. TREN derived urea derivatives as organic
gelators: (f) de Loos, M.; Ligtenbarg, A. G. J.; van Esch, J.; Kooijman,
H.; Spek, A. L.; Hage, R.; Kellogg, R. M.; Feringa, B. L. Eur. J. Org.
Chem. 2000, 22, 3675. TREN derived guanidines as ligands for metal
9
2212 J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005

Metal-Free Catalysts for the Hydrolysis of RNA
A R T I C L E S
Scheme 4. Synthesis of Tris(benzimidazoles) 9 and 10
Scheme 3. Synthesis of Compounds 3-8
Table 1. Acidity Constants of Compounds 3b-8ba
was transformed into thiourea 22 by addition to excess o-
nitrophenylisothiocyanate (55%).7,10 After hydrogenation of the
nitro groups (23, 27%), HgO induced cyclization led to the final
product 9 (64%). The analogous compound 10 was prepared
from the Boc protected TREN derivative 24. Compound 10 is
equipped with an ester group to allow conjugation of RNA
ligands in future studies. Stepwise reaction of 24 with thiocar-
bonyl diimidazole and compound 25 formed thiourea 26 (88%)
which readily cyclized in the presence of HgO (97%). The
resulting benzimidazole 27 could be converted into product 10
via addition of o-nitrophenylisothiocyanate, reduction, and ring
closure as in the case of 9 (37% from 27). Hydrochloride salts
3b-10b were prepared from the corresponding picrates by
filtration over ion-exchange resin. Acidity constants in 100 mM
phosphate buffer, determined photometrically at 50 µM substrate
concentrations, are summarized in Table 1. While titrations of
determined
at [nm]
compound
pKa
λ
3b
4b
5b
6b
7b
8b
9b
6.2
6.5
6.9
6.6
7.1
7.0
n.d.
250
250
288
288
288
288
288
a Photometrically determined at 50 µM substrate concentration in 100
mM phosphate buffer. All data were averaged over a minimum of two
experiments.
compounds 9 and 10 behaved well in acidic solution, nonre-
producible effects occurred around pH 7 indicating some form
of aggregation. Furthermore, absorption versus concentration
curves did not fulfill Lambert-Beer’s law under these condi-
tions.
ions: (g) Raab, V.; Kipke, J.; Burghaus, O.; Sundermeyer J. Inorg. Chem.
2001, 40, 6964. (h) Wittmann, H.; Raab, V.; Schorm, A.; Plackmeyer, J.;
Sundermeyer, J. Eur. J. Inorg. Chem. 2001, 1937. (i) Walther, M.;
Wermann, K.; Go¨rls, H.; Anders, E. Synthesis 2001, 1327.
(10) Peyman, A.; Gourvest, J.-F.; Gadek, T. R.; Knolle, J. Angew. Chem., Int.
Ed. 2000, 39, 2874.
RNA Hydrolysis. DNA sequencers allow the online detection
of fluorescently labeled oligonucleotides. Based on this tech-
nique we and others have studied the degradation9b,11 or
extension12 of RNA chains. To achieve the best resolution of
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2213

A R T I C L E S
Scheffer et al.
Chart 2. Sequences of Oligonucleotides 30-35
all possible fragment bands, a short DNA spacer was placed
between the fluorescent dye and the RNA part. The presence
of very short fragments complicating the analysis is thus
avoided. Two sequences were chosen as substrates: a 31mer
hairpin structure derived from HIV-1 TAR (30) and a linear
29mer (31,13 Chart 2). Four additional deoxynucleotides have
been attached to the 3′ end of RNAs 32 and 33 to improve the
separation of substrates and the longest degradation products.
The RNA part of 33 represents the TAR sequence of 30 but
was synthesized from enantiomeric nucleotides.9b DNA 34,
diluted with the unmodified DNA 35, served as a dye-labeled,
nondegradable oligomer for aggregation studies.
Special care was taken to avoid contaminations by natural
ribonucleases. Since catalysts 3b-9b could not be prepared
under sterile conditions, their solutions were purified by
ultrafiltration.14 In a first set of experiments, all catalysts were
tested at high concentrations (10 mM, 37 °C, see Figure 1 and
Table 2). The limited solubility required pH 6.0 for compounds
3b and 4b. For other catalysts pH 7.0 was chosen. The degree
of cleavage after incubation with 4b for 20 h (<1% for 30 and
2% for 31) was not much above the background level (<1% in
most runs). More hydrolysis was seen with compound 3b (4%/
6%). Remarkably, the single benzimidazole unit of 5b was
sufficient to exceed these results: 13% of RNA 30 and 15% of
31 were hydrolyzed. While a second imidazolinium ion (6b)
did not change the reactivity, an aminopyridine unit increased
the catalytic power (7b). By far the highest rates of hydrolysis
were obtained with bis(benzimidazole) 8b. However, the latter
experiment worked at the solubility limits of this compound.
Figure 1. RNA cleavage induced by compounds 5b, 6b, 7b, and 8b at 10
mM catalyst concentration (120-140 nM RNA, 50 mM Tris-HCl pH 7.0,
0.01% SDS, 37 °C, 20 h).
Table 2. Catalyst Efficiency at 10 mM Concentrationa
% degradation
linear
hairpin
compound
pKa
RNA 31
RNA 30
3b
4b
5b
6b
7b
6.2
6.5
6.9
6.6
7.1
7.0
61
41
21
<11
13
15
15
31
n.d.
2.4
1.3
4.6
5.2
<1
13
31
86
1.1
3.4
1.3
2.3
<1
8b
36b
37b
38b
39b
a 120-140 nM RNA, 50 mM Tris-HCl, pH 7.0 (1pH 6.0), 0.01% SDS,
37 °C, 20 h.
Excessive RNA precipitation was observed in several runs.
The resulting poor signal-to-noise ratios made quantitative
analysis in those cases impossible (see Figure 1, lane k). For
the same reason, no data could be obtained for tris(benzimida-
zole) 9b. To minimize the loss of labeled oligonucleotides by
nonspecific hydrophobic interactions, we added 0.01% of
sodium n-dodecyl sulfate (SDS) to the cleavage buffer.15 A small
set of amines (hydrochlorides of 36-39, Chart 3) was tested as
a control since RNA hydrolysis by NH2 groups is a well
documented phenomenon.6b-h Amine promoted degradation was
(11) (a) Gla¨sner, W.; Merkl, R.; Schmidt, S.; Cech, D.; Fritz, H.-J. Biol. Chem.
Hoppe-Seyler 1992, 373, 1223. (b) Schmidt, C.; Welz, R.; Mu¨ller, S. Nucleic
Acids Res. 2000, 28, 886.
(12) Hey, M.; Hartel, C.; Go¨bel, M. W. HelV. Chim. Acta 2003, 86, 844.
(13) Hall, J.; Hu¨sken, D.; Pieles, U.; Moser, H. E.; Ha¨ner, R. Chem. Biol. 1994,
1, 185.
(14) Podyminogin, M. A.; Vlassov, V. V.; Giege´, R. Nucleic Acids Res. 1993,
21, 5950.
(15) Santoro, S. W.; Joyce, G. F. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 4262.
9
2214 J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005

Metal-Free Catalysts for the Hydrolysis of RNA
A R T I C L E S
Table 3. Catalyst Efficiency at 1 mM Concentrationa
Chart 3. Amines Tested in the RNA Cleavage Assay
% degradation
linear
hairpin
compound
pKa
RNA 32
RNA 30
5b
6b
7b
8b
9b
6.9
6.6
7.1
7.0
n.d.
4.8
4.6
4.5
23.1
92.9
2.4
3.0
3.2
15.3
87.6
a 120-140 nM RNA, 50 mM Tris-HCl, pH 6.0, 0.01% SDS, 37 °C, 20
h. All data were averaged over two experiments.
somewhat above the background level (<1%) but rarely
surpassed 3% (see Table 2).
The most promising candidates (5b-9b) were then compared
under conditions ensuring sufficient solubility (pH 6.0, 1 mM
catalyst concentration). Among the biscationic compounds, again
8b induced the highest cleavage rates. Combining three ami-
nobenzimidazoles in structure 9b, therefore, should lead to
highly active catalysts. Indeed, almost complete hydrolysis of
all RNA substrates was observed in the presence of 9b (Figure
2, Table 3). The reaction is specific for RNA, since no
degradation is seen in the DNA part of the substrates.
In these initial assays, compound 9b clearly showed up as
the most promising catalyst. The more detailed studies, therefore,
Figure 3. Time-dependent RNA cleavage in the presence of 1 mM catalyst
9b (120-140 nM RNA, 50 mM Tris-HCl pH 6.0, 0.01% SDS, 37 °C).
Figure 4. pH dependent RNA cleavage at 1 mM concentration of 8b (120-
140 nM RNA, 50 mM Tris-HCl, 0.01% SDS, 37 °C, 20 h).
concentrated on 9b and its analogue 10b. The time course of
RNA hydrolysis was studied next. The decrease of substrate
concentration versus time at 1 mM 9b and pH 6.0 could be
fitted to a first-order rate law (Figure 3). The resulting half-
lives are 120 min for the linear RNA 32 and 200 min for the
TAR analogue 30, respectively. This corresponds to a formal
first-order rate constant of 3.3 × 10-6 s-1 for a mean
phosphodiester linkage in substrate 32 (29 susceptible bonds).
The pH is expected to influence the rates of benzimidazole
induced RNA hydrolysis. Systematic studies with compound
9b were hampered by the limited solubility above pH 7.
However, at 1 mM concentration they could be carried out with
bis(benzimidazole) 8b. As shown in Figure 4, cleavage rates
increase steadily with increasing pH. At any pH, they are far
above background levels. pH versus rate curves may show the
protonation state of 8b or 9b relevant to catalysis. The
Figure 2. RNA cleavage induced by compounds 5b, 6b, 7b, 8b, and 9b
at 1 mM catalyst concentration (120-140 nM RNA, 50 mM Tris-HCl pH
6.0, 0.01% SDS, 37 °C, 20 h).
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2215

A R T I C L E S
Scheffer et al.
Table 4. RNA Cleavage as a Function of [9b]a
% degradation
concentration
9b [ M]
linear
hairpin
µ
RNA 32
RNA 30
250
200
150
120
100
80
87.4
85.6
63.3
60.3
16.6
6.2
48.7
44.8
27.1
26.7
8.5
6.4
60
3.9
6.5
40
5.2
5.2
20
6.3
7.7
10
4.1
1.1
a RNA cleavage in the presence of 0.01% SDS. 120-140 nM RNA, 50
mM Tris-HCl, pH 6.0, 37 °C, 20 h.
requirement for such conclusions, however, would be the
absence of complicating effects such as pH dependent aggrega-
tion. With the nonideal behavior of 9b in mind, mechanistic
interpretations of Figure 4 are highly dangerous (see below).
Table 4 shows the dependence of the total amount of RNA
cleavage on catalyst concentration. The reaction is still fast at
[9b] ) 250 µM. At concentrations around 100 µM, a sudden
decrease of cleavage occurs. Significant hydrolysis above
background levels can be observed at catalyst concentrations
as low as 20 µM. While the specific numbers in Table 4 are
taken from a single dilution experiment and reflect the statistical
error of the method, several repetitions of the whole dilution
series could reproduce the sudden rate decrease below 100-
150 µM. Essentially the same was seen with catalyst 10b.
Interestingly, the cleavage pattern induced by compounds 9b
and 10b does not reflect the secondary structure of the TAR
model 30 (see Figure 2, lane k). When repeated with the
enantiomeric RNA 33,9b the total amount of degradation at each
catalyst concentration and the cleavage pattern were indistin-
guishable from the data produced with natural RNA 30. At
catalyst concentrations below 100 µM, however, some experi-
ments with natural RNA 30 exhibited preferential cleavage after
pyrimidines. This was never observed with enantiomeric
substrate 33 and can only be explained as a minor contamination
effect.9b
Relevance of Aggregation for the Activity of Catalysts 9b
and 10b. The idea that aggregation phenomena involving tris-
(benzimidazole) catalysts and RNA may be crucial for the
interpretation of cleavage data was triggered by two observa-
tions: the deviation of Lambert-Beer’s law and the denaturation
of RNA secondary structure. Fluorescence correlation spectros-
copy (FCS) is a highly sensitive method for the detection of
aggregates.16 Thus, we have determined the diffusion time of
the dye-labeled DNA 34 by FCS as a function of catalyst
concentration and the presence of further additives. A non-
degradable oligonucleotide was chosen to avoid time dependent
effects caused by substrate hydrolysis. In Tris buffer pH 6.0
containing 0.01% sodium n-dodecyl sulfate (SDS) and low
concentrations of 10b, the diffusion time of 34 was 150-180
µs (24 °C), consistent with the molecular mass of 7.3 kDa. The
data could be fitted to a one-component model. In contrast, with
catalyst concentrations above 100 µM, large aggregates contain-
ing several copies of 34 appeared. As shown in Figure 5, a good
Figure 5. As detected by FCS, the presence of detergents has a critical
influence on the aggregation of catalysts 9b and 10b. (A) 50 mM Tris-HCl
pH 6.0, 0.01% SDS. (B) 50 mM Tris-HCl pH 6.0, 0.01% Triton X-100.
Aggregation correlates with RNA cleavage yields. Left ordinate: diffusion
time of oligonucleotide 34 (25 nM 34 + 175 nM 35). Right ordinate:
percentage of cleaved RNA 33 (150 nM).
correlation exists between cleavage of enantio-RNA 33 and
aggregation of DNA 34. However, when SDS was replaced by
the nonionic detergent Triton X-100 (0.01%, 160 µM), ag-
gregates of 34 could be seen at concentrations of 10b as low as
4 µM. With Triton, no sudden catalyst deactivation below 100
µM occurs (Figure 5). When surfactants were completely
omitted, the critical value of [10b] leading to aggregation of
34 again is 4 µM. These strange looking results are easily
explained by noting that 0.01% SDS corresponds to 350 µM, a
value far below the critical micelle concentration of 8 mM but
equaling the concentration of benzimidazole units when [9b]
or [10b] is 117 µM. In the final experiment, therefore,
aggregation of 34 as a function of [10b] was studied in the
presence of variable SDS concentrations. In all cases, aggregates
of 34 disappeared when the point of equivalence between the
anionic surfactant and the benzimidazole units was reached. It
is our current view that both catalysts 9b and 10b self-assemble
at pH 6.0 and concentrations down to 4 µM independently from
the presence of nucleic acids. This is supported by finding
identical aggregation thresholds with 25 nM DNA 34 alone or
a mixture of 25 nM 34 plus 175 nM unlabeled DNA 35.
Nonionic surfactants, added to minimize unspecific hydrophobic
interactions, interfere neither with this process nor with RNA
binding to such aggregates. SDS, in contrast, may not influence
the assembly of 9b or 10b, but it binds to the aggregates due to
the combination of negative charge and hydrophobicity thus
(16) Mu¨ller, J.; Chen, Y.; Gratton, E. Methods Enzymol. 2003, 361, 69.
9
2216 J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005

Metal-Free Catalysts for the Hydrolysis of RNA
A R T I C L E S
Figure 7. Conformers of protonated 2-aminopyridines.
known that N-substituted guanidinium ions form stable ion pairs
only when they can offer two parallel NsH bonds to the guest
(corresponding to “E” in Figure 7).5j Thus the ability to form
undisturbed ion pairs between phosphate and heterocycle may
be decisive for efficient acid/base catalysis. Unfavorable con-
formational equilibria could also account for the moderate and
unpredictable transphosphorylation activities of some N-sub-
stituted guanidinium compounds.5h
Figure 6. Degradation of hairpin RNA 33 induced by catalyst 10b in the
nonaggregated state at low concentrations. The cleavage pattern is consistent
with the secondary structure shown in Chart 2.
preventing the interaction with oligonucleotides. The nonideal
nature of the RNA-catalyst solutions has to be kept in mind to
avoid misinterpretation of data (e.g., the pH dependency shown
in Figure 4).
The tris(2-aminobenzimidazoles) 9b and 10b proved to be
among the most efficient metal free catalysts for RNA cleavage.
At 1 mM catalyst concentration, the half-life of RNA compares
well with the data of many metal complexes. As shown by
fluorescence correlation spectroscopy, however, the active
species under those conditions are aggregates of substrate and
catalyst leading to denaturation of RNA secondary structure.
Aggregation critically depends on the nature and concentration
of further additives such as SDS. Compound 10b, when tested
at very low concentration, was shown to be active in the
nonaggregated state as well. Conjugates of 10b with sequence
specific RNA ligands, therefore, have good prospects of success.
The benzimidazole catalyzed hydrolysis is specific for RNA.
Hence the overall mechanism must involve nucleophilic attack
by 2′ hydroxy groups. Unfortunately, catalyst aggregation
prevents further mechanistic insight from pH versus rate
correlations. Such experiments will be accessible with oligo-
nucleotide conjugates of 10b in hand. Apart from the synthetic
tasks, future work will try to dissect the different mechanistic
aspects of aminobenzimidazole catalysis.
A central question remains: if RNA cleavage mainly cor-
relates with aggregated forms of compound 9b or 10b, what
can be expected for the catalytic activity of conjugates with
antisense oligonucleotides or RNA binding peptides acting as
single molecules? Figure 6 compares the degradation of the
enantio-TAR model 33 at 125 µM and 1.9 µM concentrations
of 10b without any surfactant. In contrast to Figure 2 (lane k),
the cleavage pattern starts to reflect the TAR secondary structure.
At [10b] ) 1.9 µM, a strong differentiation between nucleotides
around the bulge, the loop, and the double helical parts is seen.
Analogous assays with natural RNA 30 led to identical cleavage
patterns. 20% of RNA 33 are degraded within 20 h (10%
cleavage at [10b] ) 0.9 µM) thus demonstrating the catalytic
activity of tris(benzimidazoles) in the nonaggregated state.
Conclusions
The online detection of fluorescently labeled oligoribonucle-
otides allowed the precise and convenient characterization of
RNA cleaving catalysts. By comparing reactions of naturally
configurated and enantio-RNA, compelling evidence was found
to rule out contamination effects that may corrupt the cleavage
assays. Of all compounds tested, derivatives of 2-aminobenz-
imidazoles turned out to be powerful RNA cleavers. When
compared with guanidines, the increased reactivity can be
attributed to the pKa change from 14 to 7. It is more difficult to
explain the failure of the 2-aminopyridines 3 and 4. With pKa’s
around 6.2, these heterocycles are good candidates for general
acid/base catalysis. In contrast to all other guanidinium ana-
logues present in compounds 1-10, the exocyclic CsN bond
does not represent a local C2 axis. Arguing by analogy with
peptide bonds, the Z conformation should be preferred (Figure
7, amino NsH antiperiplanar to the CdN bond).17 It is also
Acknowledgment. This paper is dedicated to the memory of
Professor Jacques H. van Boom (1937-2004). Financial support
by the European Community is gratefully acknowledged
(European Network on the Development of Artificial Nucleases,
HPRN-CT-1999-00008). We also would like to thank Prof.
Stefan Pitsch, EPFL Lausanne, for providing us with a sample
of enantiomeric RNA 33.
Supporting Information Available: Experimental details,
spectroscopic and analytical data for all new compounds. Figure
S1: titration curve to determine the pKa of compound 5b. Figure
S2: RNA cleavage induced by amines 36-39. Figure S3:
Time-dependent RNA cleavage in the presence of 1 mM catalyst
9b. Table S1: diffusion times (FCS) of oligonucleotide 34 as
a function of different additives. This material is available free
of charge via the Internet at http://pubs.acs.org.
(17) Neutral 2-alkylaminopyridines normally adopt the E conformation, espe-
cially when bound to guest molecules such as carboxylic acids. Confor-
mational polymorphism due to E/Z isomerization, however, is known from
2-arylaminopyridines: Bensemann, I.; Gdaniec, M.; Polonski, T. New J.
Chem. 2002, 26, 448.
JA0443934
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2217