Enhanced RNA binding of dimerized aminoglycosides

Article abstract of DOI:10.1016/S0968-0896(99)00071-1

Aminoglycoside antibiotics have recently emerged as an intriguing family of RNA binding molecules and they became leading structures for the design of novel RNA ligands. The demystification of the aminoglycoside-RNA recognition phenomenon is required for

Full text of DOI:10.1016/S0968-0896(99)00071-1

Bioorganic & Medicinal Chemistry 7 (1999) 1361±1371
Enhanced RNA Binding of Dimerized Aminoglycosides
Katja Michael, Hai Wangy and Yitzhak Tor*
Department of Chemistry and Biochemistry 0358, University of California, San Diego, La Jolla, CA 92093-0358, USA
Received 5 October 1998; accepted 11 December 1998
AbstractÐAminoglycoside antibiotics have recently emerged as an intriguing family of RNA binding molecules and they became
leading structures for the design of novel RNA ligands. The demysti®cation of the aminoglycoside±RNA recognition phenomenon
is required for the development of superior binders. To explore the existence of multiple binding sites in a large RNA molecule, we
have synthesized covalently linked symmetrical and nonsymmetrical dimeric aminoglycosides. These unnatural derivatives were
compared to their natural ``monomeric'' counterparts in their ability to inhibit the Tetrahymena ribozyme. The dimeric aminogly-
cosides inhibit ribozyme function 20 to 1.2Â103 fold more e€ectively than their natural parent compounds. The inhibition curves of
dimeric aminoglycosides have characteristic shapes suggesting the presence of at least two high anity-binding sites within the
ribozyme's three-dimensional fold. The interaction of a dimeric aminoglycoside with two complementary sites of the RNA molecule
is proposed. This binding motif may have implications on the development of new drugs targeting pivotal RNA molecules of bac-
terial and viral pathogens. # 1999 Elsevier Science Ltd. All rights reserved.
Introduction
ribozyme.10,16 The role played by the amino groups and
the in¯uence on their basicity by vicinal hydroxyl
groups has been elucidated. Electrostatic interactions
have been shown to be critically important in RNA
binding.10,17 A binding model based on structural elec-
trostatic complementarity proposes the three-dimen-
sionally de®ned projection of an aminoglycoside's
positively charged ammonium groups to centers of
negative charge density in the RNA host.16,18,19
Key RNA molecules of bacteria and retroviruses can be
targeted by aminoglycoside antibiotics. These 2-deoxy-
streptamine-containing amino sugars (Fig. 1) e€ec-
tively interfere with the bacterial protein biosynthesis and
viral protein±RNA interactions.1±4 The therapeutic use of
aminoglycosides is, however, limited due to their toxicity.5
These adverse e€ects may result from the lack of speci®c
binding to the RNA site of interest. The molecular details
of aminoglycoside±RNA binding are presently not fully
understood. The successful development of new anti-
bacterial and antiviral drugs based on this class of natural
products requires the elucidation of the elements involved
in RNA±ligand recognition. The interactions of ami-
noglycoside antibiotics with a number of seemingly unre-
lated RNA molecules4,6±12 suggest common recognition
patterns.13±15 We aim at elucidating RNA±small molecule
recognition by studying the interactions of chemically
modi®ed aminoglycosides with RNA.
In a preliminary account, we have reported the inhibi-
tion of a hammerhead ribozyme by natural aminogly-
cosides and their dimeric derivatives.20 Here we explore
the possibility of enhancing the binding anity between
aminoglycosides and RNA by targeting multiple bind-
ing sites that may coexist on a large RNA molecule.
Experimental support for the existence of multiple RNA
binding sites for aminoglycosides has been previously
obtained for the 67 nt RRE RNA of HIV-1,21 for the 59
nt TAR RNA of HIV-1,4 and for the 76 nt long
tRNAPhe 22
Dimerizing RNA binding ligands could, in
.
We have previously reported the structure±activity
relationships of deoxygenated aminoglycosides and
`amino-aminoglycosides' as inhibitors of a hammerhead
principle, lead to molecules capable of binding to two
proximal RNA sites resulting in enhanced binding a-
nity. To investigate this hypothesis, we have designed
covalently linked dimeric aminoglycosides shown in
Figure 2. A relatively long and conformationally ¯exible
linker was selected to allow the molecules to `scan' the
conformational space in their search for a second RNA
binding site. In selecting the tethering position, we have
taken into consideration our recent observation that the
600-OH in tobramycin is not essential for the inhibition
Key words: Aminoglycoside antibiotics; electrostatics; inhibition;
ribozyme.
* Corresponding author. Tel.: +1-858-534-6401; fax: +1-858-534-
y5383; e-mail: ytor@ucsd.edu
Current address: Clinical Micro Sensors, Inc., Pasadena, CA 91105,
USA.
0968-0896/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(99)00071-1

1362
K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
Figure 1. Natural aminoglycoside antibiotics.
Figure 2. Synthetic dimeric aminoglycosides.

K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
1363
of the hammerhead ribozyme HH16.10 A disul®de link-
age was chosen to maximize synthetic ¯exibility thus
facilitating the construction of symmetrical and non-
symmetrical dimers from similar building blocks.
been slightly altered compared to our preliminary
report.20 Scheme 1 illustrates the synthetic approach for
the heterodimeric aminoglycoside Tob±Neo (7) in
detail. Neomycin B (1) and tobramycin (2) were ®rst
protected as their Boc derivatives to give 9 and 12,
respectively. Treatment with triisopropylbenzenesulfo-
nyl chloride converted the primary hydroxyls to their
corresponding sulfonates 10 and 13. Treatment of the
fully Boc-protected sulfonates with bis(2-mercapto-
ethyl)ether in DMF in the presence of cesium carbonate
a€orded the thiol derivatives 11 and 14, respectively.
Reaction of tobramycin-derivative 14 with 2,20±dipyr-
idyldisul®de furnished the pyridyl±disul®de 15. This
derivative can be treated under mild conditions with
neomycin B derivative 11 to give the fully Boc-protected
dimer 16. Deprotection with TFA provided the desired
hetero-dimer 7. Fully Boc-protected aminoglycosides
with a mercapto substituent (e.g. 11 and 14), and fully
Boc-protected aminoglycosides with a pyridyldisul®de
substituent (e.g. 15) have been synthesized from neomy-
cin B (1), tobramycin (2) and kanamycin A (3). These
derivatives are the key components in the syntheses,
allowing for the formation of homo- and hetero-dimers
using the same strategy described above. After cleaving
o€ the Boc groups with tri¯uoroacetic acid, the dimeric
aminoglycosides were converted to their corresponding
free bases by ion exchange.
We have chosen the 388 nt Tetrahymena ribozyme L-21
Sca I as a functional model for a large RNA molecule
where the existence of multiple binding sites is plausible
due to sophisticated secondary and tertiary struc-
tures.23±25 This ribozyme catalyzes in trans the site spe-
ci®c phosphodiester hydrolysis of 50GGCCCUCUA
AAAA30 between U and A (Fig. 3).26 This strand scission
reaction can be inhibited by aminoglycoside antibiotics
and is therefore an excellent functional assay. The extent
of substrate cleavage in the presence of aminoglycosides
serves as an indirect measure for aminoglycoside binding
to the ribozyme. The comparison of individual inhibition
curves can reveal di€erences in the recognition patterns of
various aminoglycosidic ligands. Measuring inhibition
curves at di€erent ionic strengths can shed light on the
contribution of electrostatic interactions to binding. Two
key questions are addressed: (a) do dimeric aminoglyco-
sides bind to RNA with higher anities when compared
to their natural monomeric counterparts, and (b) what is
the role played by electrostatic interactions?
Here we report the synthesis of dimeric aminoglycosides
and demonstrate that dimerizing natural aminoglyco-
sides results in potent ribozyme inhibitors exceeding the
inhibitory activity of their natural monomeric counter-
parts by factors of 20-fold to 1.2Â103-fold. We have
focused on the binding behavior of neomycin B (1),
tobramycin (2) and their homodimeric derivatives Neo±
Neo (4) and Tob±Tob (5) in more detail. At high ionic
strength, the ribozyme binding anities of Neo±Neo (4)
remains practically unchanged, while the binding a-
nities of Tob±Tob (5) and the monomeric aminoglyco-
sides 1 and 2 are drastically weakened. We discuss the
exceptional behavior of Neo±Neo (4) based on the
degree of protonation at high ionic strength.
Investigation of aminoglycoside±RNA binding
The strength of the interaction between aminoglycosides
and RNA was indirectly estimated by investigating their
e€ect on ribozyme function. The inhibition of ribozyme
function was followed by measuring the ribozyme's
initial kinetics and by determining apparent IC50 values.
The ribozyme-catalyzed cleavage reaction was carried
out with 50-32P-labeled substrate in the presence or
absence of inhibitors at pH 7.0 under single-turnover
conditions. Reaction mixtures were resolved by poly-
acrylamide gel electrophoresis and the ratio of full
length substrate to its cleavage product was determined
by phosphorimage analysis. The inhibitory e€ect on the
reaction rate was investigated by taking out aliquots
and immediately stopping the reaction at di€erent time
intervals by adding EDTA-containing loading bu€er
(see Experimental). Initial kinetics was followed by
plotting the fraction of cleaved substrate versus time.
IC50 values were obtained from semilogarithmic plots of
the inhibition of the ribozyme versus aminoglycoside
concentration. The degree of inhibition was calculated
from the fraction of cleaved substrate after normal-
ization of the inhibition curves. Table 1 summarizes the
IC50 values obtained for all compounds investigated.
Results
Synthesis of dimeric aminoglycosides
A general synthetic strategy for the facile synthesis of
symmetrical as well as nonsymmetrical dimeric ami-
noglycosides has been developed. The synthesis has
Note that our inhibition assay responds to binding
events that interfere with ribozyme function directly.
However, a binding event at a site not involved in the
cleavage chemistry may in¯uence the inhibition curves
indirectly. High anity binding of aminoglycosides to
RNA sites irrelevant to ribozyme function decreases the
e€ective aminoglycoside concentration and can lead to
steep inhibition curves that are shifted towards higher
aminoglycoside concentrations.
Figure 3. Endonuclease reaction of the L-21 Sca I ribozyme derived
from Tetrahymena pre-rRNA. The substrate GGCCCUCUAAAAA,
base-paired with the 50 exon site of the ribozyme, is cleaved by exo-
genous GTP.

1364
K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
Scheme 1. Reagents and conditions: (a) Boc2O, NEt3, DMSO/H2O, 60ꢀC, 5 h, 91%; (b) TIBSCl (2,4,6-triisopropyl bezenesulfonyl chloride), pyr.,
23ꢀC, 12 h; 66%; (c) bis(2-mercaptoethyl)ether, Cs2CO3, DMF, 23ꢀC, 12 h; 92%; (d) Boc2O, NEt3, DMSO/H2O, 60ꢀC, 4 h, 94%; (e) TIBSCl, pyr.,
23ꢀC, 12 h, 65%; (f) bis(2-mercaptoethyl)ether, Cs2CO3, DMF, 23ꢀC, 12 h; 94%; (g) 2,20-dipyridyldisul®de, MeOH, 23ꢀC, 5 h, 85%; (h) MeOH,
23ꢀC, 10 h, 60%; (i) (1.) 99% TFA, 23ꢀC, 3 min, (2.) Amberlite-400 (OH ), H2O, 23ꢀC, 2 h, quantitative.
Table 1. IC50 values for ribozyme inhibition by natural aminoglyco-
sides and their dimeric derivatives at low ionic strengtha
both aminoglycosides have been applied at the same
concentration range. Figure 6 illustrates representative
inhibition curves of all dimeric and monomeric
aminoglycosides studied at low ionic strength (50 mM
HEPES, pH 7.0, 10 mM MgCl2). The slopes of the inhi-
bition curves of all dimers are noticeably steeper than the
inhibition curves of the monomers. The extracted IC50
values are summarized in Table 1. All compounds inhibit
ribozyme function in the nM to mM concentration range.
Among them, the dimeric aminoglycosides are the most
ecient inhibitors with apparent IC50 values between
1.1Â102 and 7.7Â102 nM. The IC50 values of the mono-
mers neomycin B (1) and tobramycin (2) are 3.4 and
16 mM, respectively. Kanamycin A (3) is the least active
aminoglycoside antibiotic with an IC50 of 0.9 mM.
Aminoglycoside
IC50 (mM)
Neomycin B (1)
Tobramycin (2)
Kanamycin A (3)
Neo±Neo (4)
Tob±Tob (5)
Kan±Kan (6)
Tob±Neo (7)
Kan±Tob (8)
3.4‹1.45
16‹11
9.3Â102‹6Â102
0.19‹0.087
0.11‹0.072
0.77‹0.42
0.39‹0.17
0.26‹0.18
1
The ribozyme inhibition assay was carried out as described in the
Experimental. IC50 values are the average of at least three independent
experiments and are reported with the standard deviation.
Initial kinetics of the L-21 Sca I ribozyme
Aminoglycoside binding at high ionic strength
Qualitative initial kinetic experiments indicate that the
cleavage reaction rate is slowed down signi®cantly by all
aminoglycoside derivatives investigated. The dimers
Tob±Tob (5) and Kan±Tob (8) inhibit the cleavage
reaction more eciently than twice the amount of
tobramycin (2), kanamycin A (3) or a 1:1 tobramycin
(2):kanamycin A (3) mixture (Fig. 4).
The inhibitory activities of neomycin B (1), tobramycin
(2) and their homodimeric derivatives Neo±Neo (4) and
Tob±Tob (5) were determined at increasing NaCl con-
centrations. Under our conditions (50 mM HEPES, pH
7.0, 10 mM MgCl2, 0!300 mM NaCl), the ribozyme's
function is suciently well maintained. Control experi-
ments in the absence of aminoglycosides indicate that
the ribozyme loses less than 10% of its cleaving ability
at 300 mM NaCl. At 1 M NaCl, the ribozyme activity
drops to less than 50% of its original value.
Aminoglycoside binding at low ionic strength
A polyacrylamide gel of ribozyme inhibition obtained by
the titration of neomycin B (1) and its homodimeric deri-
vative Neo±Neo (4) is shown in Figure 5. For comparison,
Representative inhibition curves obtained at di€erent
levels of ionic strength are depicted in Figure 7. Strikingly,

K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
1365
increasing the NaCl concentration from 0 to 300 mM
has little in¯uence on the binding characteristics of the
dimeric derivative Neo±Neo (4). Its IC50 value increases
only fourfold and the steepness of its inhibition curves is
retained. In contrast, the IC50 values of Tob±Tob (5),
tobramycin (2) (data not shown) and neomycin B (1)
dramatically shift by a factor of 100 when the ionic
strength is increased from 0 to 300 mM NaCl. Despite
the loss of binding anity at increased NaCl con-
centrations, the inhibition curves of Tob±Tob (5)
remain steep regardless of the ionic strength.
cations.24 In analogy to the hammerhead ribozyme,
where four critical magnesium ions in the catalytic core
were suggested to be displaced by positively charged
aminoglycosides,17,18 aminoglycosidic inhibitors of the
Tetrahymena ribozyme may follow a similar mechanism.
Displacing critical metal ions by the ammonium groups
of aminoglycosides could lead to a disruption of the
correct ribozyme folding, thus inhibiting its function.
Our studies con®rm the correlation between the number
of amino groups on the ligand and its RNA binding
anity. Neomycin B (1) is the most active inhibitor
among the natural products investigated, followed by
tobramycin (2) and kanamycin A (3). All dimeric ami-
noglycosides inhibit the ribozyme much more e€ectively
than natural aminoglycosides (Table 1). Although the
number of positive charges is, in general, critical for
RNA binding anity, structural features of the ligand
are expected to play an important role in RNA recog-
nition. Since dimeric aminoglycosides inhibit ribozyme
function to similar extents despite di€erent numbers of
amino groups, their common molecular shape (two
Discussion
General aspects of aminoglycoside-RNA binding
The Tetrahymena ribozyme requires magnesium or
manganese ions for catalysis and proper folding into its
active conformation.27 The crystal structure of the P4/P6
domain of the Tetrahymena group I intron revealed that
the molecule is folded around a core of ®ve magnesium
Figure 4. Ribozyme initial kinetics in the presence of selected mono-
meric and dimeric aminoglycosides: control (no aminoglycoside) (*);
kanamycin A (3), 2 mM (~); tobramycin (2), 1 mM and kanamycin A
(3), 1 mM (!); tobramycin (2), 2 mM (&); Kan±Tob (8), 1 mM (*);
Tob±Tob (5), 1 mM (^).
Figure 6. Inhibition curves obtained for natural and dimeric ami-
noglycosides at low ionic strength. Key: neomycin B 1 (*); tobramy-
cin (2) (&); kanamycin A (3) (~); Neo±Neo (4) (*); Tob±Tob (5)
(&); Kan±Kan (6) (~); Tob±Neo (7) (^); Kan±Tob (8) (!).
Figure 5. A 20% polyacrylamide gel illustrating ribozyme inhibition by neomycin B (1) and Neo±Neo (4) over an identical concentration range.
Lane 1: substrate only, lane 2: control (no aminoglycoside present), lanes 3 to 12: neomycin B (1) and lanes 13 to 22: Neo±Neo (4) at 100 mM, 30
mM, 10 mM, 3 mM, 1 mM, 600 nM, 300 nM, 100 nM, 30 nM and 3 nM concentration, respectively.

1366
K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
Increasing the ionic strength from 0 to 300 mM NaCl
drastically shifts the IC50 values of neomycin B (1),
tobramycin (2) and Tob±Tob (5) towards higher con-
centrations by factors of 100±250, while the IC50 value
of Neo±-Neo (4) only shifts by a factor of four (Fig. 7).
This indicates that the natural aminoglycosides as well
as Tob±Tob (5) can be displaced from their RNA bind-
ing site even at a relatively low ionic strength. To over-
come the binding anity of Neo±Neo (4), an ionic
strength (>500 mM NaCl) that interferes with ribo-
zyme function is required. Why is the anity of the
dimeric aminoglycoside Neo±Neo (4) relatively insensi-
tive towards increased salt concentrations? A possible
reason might be its di€erent number of amino groups
and their individual pKa values. At pH 7, Neo±Neo (4)
is likely to possess at least two more positive charges
than Tob±Tob (5). Depending on the pKa values of
individual ammonium groups at high salt concentra-
tions, the di€erence in overall positive charge between
Neo±Neo (4) and Tob±Tob (5) could become even more
pronounced. This may result in Tob±Tob (5) behaving
more like a `monomer' and displaying a higher sensitivity
towards the ionic strength.
Inhibition curves
Examining individual inhibition curves suggests di€er-
ent binding stoichiometries for monomeric and dimeric
aminoglycosides (Fig. 6). The inhibition curves of the
dimeric aminoglycosides fall in the nM concentration
range and their slopes are remarkably steep: the breadth
of the binding transitions reaches over only one to one
and a half decades (Fig. 6). It is noticeable that the
inhibitory activity of the dimers starts to develop at
approximately 50 nM. Since the ribozyme's concentra-
tion is 50 nM in each of the experiments, we speculate
that the characteristic inhibition curves observed for the
dimers result from the presence of other high anity
binding sites within the large RNA molecule. At con-
centrations below 50 nM, the dimers may be seques-
tered by high anity binding sites that do not interfere
with ribozyme function. At 50 nM dimer concentration,
these high anity sites may be saturated. Increasing
dimer concentration results in binding at a site that leads
to ribozyme inhibition. The steepness of the inhibition
curves suggests that the dimers bind strongly to this cri-
tical site (albeit with lower anity than to the `high a-
nity sites') since the concentration window between onset
and complete inhibition is rather narrow. The extremely
steep slopes exhibited by Neo±Neo (4) and Tob±Tob (5)
are retained at high ionic strengths (Fig. 7). Based on
these observations, at least two dimeric aminoglycosides
bind to the ribozyme at low and high ionic strength, each
site consisting of two subsites. Accordingly, we propose
that the Tetrahymena ribozyme possesses multiple bind-
ing sites for aminoglycosides.
Figure 7. Ribozyme inhibition curves in the presence of (top) neomy-
cin B 1; (middle) Tob±Tob (5); (bottom) Neo±Neo (4) at varying ionic
strengths: no NaCl (*); 50 mM NaCl (&); 200 mM NaCl (~); 300
mM NaCl (^).
highly charged moieties held together by a ¯exible lin-
ker) may contribute signi®cantly to their binding a-
nity. These aminoglycoside derivatives can most likely
occupy two sites in close proximity within the RNA
molecule.
Electrostatic interactions in RNA binding
Binding events involving nucleic acids are in¯uenced by
the ionic environment. Di€erent models describing the
association of charged ligands and nucleic acids exist.
Both the entropic release of numerous counterions from
the nucleic acid,28 as well as changes in interaction of
the nucleic acid with its ionic atmosphere,29 have been
discussed as driving forces for complex formation.
Why do dimeric aminoglycosides inhibit the L-21 Sca I
ribozyme more e€ectively than their natural monomeric
counterparts? At physiological pH, aminoglycosides are
highly-charged.30±32 Generally, their RNA binding
relies on electrostatic interactions.16±19,21 The doubled
set of amino groups may account for their increased
binding anity. If electrostatic interactions contribute
substantially to the free energy of aminoglycoside±ribo-
zyme association, the inhibition of ribozyme function is
expected to be in¯uenced by the ionic strength.
Similar observations have been made in investigating
tRNAPhe aminoglycoside binding.22 The inhibition of
Pb2+-mediated cleavage by Neo±Neo (4) also shows an
unusually steep inhibition curve, while neomycin B (1)
gives a shallow inhibition curve. It is probably not a
coincidence that the in¯ection point for Neo±Neo (4) is

K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
1367
close to 2 mM, the concentration of tRNAPhe used in
Conclusion
these experiments. This independent observation sup-
ports our hypothesis regarding the existence of another
binding site with higher anity in the RNA molecule.
We therefore favor this explanation over the alternative
possibility of cooperative binding.
To explore the existence and utilization of multiple
binding sites within a large RNA molecule, we have
constructed dimeric aminoglycosides by covalently
tethering two aminoglycoside antibiotics. Compared to
their natural parent compounds, dimeric aminoglyco-
sides inhibit ribozyme function much more e€ectively.
We propose that their enhanced RNA binding anities
result from their overall positive charge together with
their unique molecular architecture. The characteristic
inhibition curves obtained for dimeric aminoglycosides
suggest the existence of multiple aminoglycoside binding
sites within the ribozyme's three-dimensional fold.
In contrast to dimeric aminoglycosides, the inhibition
curves obtained for all natural aminoglycosides tested
fall in the mM concentration range and their slopes are
noticeably shallower (Fig. 6). Due to the lower binding
anity of natural aminoglycosides to the ribozyme, the
fraction of unbound aminoglycoside is always large
when compared to the ribozyme concentration. Thus
the inhibition curves do not indicate or exclude multiple
site binding of natural aminoglycosides.
Dimeric aminoglycosides expand the relatively small
group of RNA binding molecules known to date and
contribute to our understanding of RNA±aminoglycoside
interactions. The principle of recognizing two binding
sites in close proximity opens new perspectives for the
design of novel RNA ligands with enhanced binding a-
nities.13 Yet, the presence of multiple binding sites in an
RNA molecule may in¯uence binding selectively as
ligands may preferentially bind to sites not interfering
with biological function. Our results indicate that ami-
noglycosides may be sequestered by additional binding
sites in large RNA molecules. To overcome this draw-
back, novel ligands with improved binding selectivities
and stringent RNA binding assays have to be developed.13
Hypothetical model for RNA±aminoglycoside binding
It is reasonable to assume that the structurally sophisti-
cated L-21 Sca I ribozyme possesses multiple sites for
aminoglycoside binding with di€erent anities. One
binding site for a dimeric aminoglycoside could consist
of two adjacent `monomeric' sites. These sites could be
located next to each other in the primary sequence or
within close proximity in the ribozyme's tertiary struc-
ture. Each monomeric subunit of a dimeric aminogly-
coside could occupy one of the two sites. Most of the
entropic penalty is paid by binding the ®rst subunit to
the RNA molecule, which brings the covalently linked
second subunit in close proximity to its binding site.33
Experimental
General
Two adjacent RNA binding sites can be occupied by
two individual monomeric aminoglycosides. In contrast
to a dimeric molecule, individual aminoglycosides are
encumbered with an entropical disadvantage. This
view is experimentally supported by the ®nding that
one equivalent of a dimeric aminoglycoside decreases
the rate of substrate cleavage more than two equiva-
lents of the corresponding monomeric aminoglyco-
side(s) (Fig. 4).
NMR spectra were recorded on a GE QE300 spectro-
meter operating at 300 MHz (1H) and on a Varian
Unity 500 spectrometer operating at 500 MHz (1H) and
125 MHz (13C). 13C NMR spectra were calibrated rela-
tive to dioxane (d=66.66 ppm) in a separate NMR
tube. UV spectra were measured on a Hewlett Packard
8452 A diode array spectrophotometer. FAB and ESI
mass spectra were recorded at the Mass Spectrometry
Facility of the University of California, Riverside.
The inhibitory activity of an aminoglycoside can also be
expressed in dynamic terms. Monomeric aminoglyco-
sides are expected to dissociate quickly from the ribo-
zyme, in agreement with reports describing fast o€ rates
in aminoglycoside±RRE±RNA binding.21 Once the
RNA/monomer complex is dissociated, shielding by
counter ions prevents rapid re-association leading to
slow on rates. This e€ect is expected to be even more
pronounced at high ionic strength. Since Kd ˆ koff =kon,
it is expected to be relatively high for a monomeric
aminoglycoside. In contrast, the RNA/dimeric amino-
glycoside complex is likely to dissociate slowly. One of
two monomeric subunits may, on average, be `per-
manently' bound to the RNA target. Partial dissocia-
tion of one of the two subunits would be followed by
quick reassociation, because the dissociated subunit is
held in close proximity to its binding site by the tethered
subunit that remained bound. The IC50 value is, there-
fore, expected to be much lower for a dimeric aminogly-
coside versus a monomeric aminoglycoside, as indeed
observed.
Water was obtained from a Millipore Milli-Q water
puri®cation system. For biochemical application, the
water was treated with 0.1% diethylpyrocarbonate and
autoclaved. Tobramycin (free amine) was a generous
gift from Meiji Seika Kaisha, Ltd. Neomycin Bꢁ 3
H2SO4 (containing max 15% neomycin C) and kana-
mycin Aꢁ H2SO4 (containing ca. 5% kanamycin B) were
purchased from Sigma. Neomycin Bꢁ 3 H2SO4 and
kanamycin Aꢁ H2SO4 as well as the synthetic dimeric
aminoglycosides were converted into their free amines
by treatment with amberlite IRA-400 ion exchange
resin. Stock solutions (5Â) of the aminoglycosides in
autoclaved DEPC±water were prepared. Neomycin B,
tobramycin, and kanamycin A solutions were bu€ered
with 1 N HCl in order to guarantee pH 7.0 in the cleavage
reaction. T7 RNA polymerase and the Sca I endonuclease
were purchased from Boehringer Mannheim. Nucleotide
triphosphates were purchased from Pharmacia. Calf

1368
K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
intestinal alkaline phosphatase and T4 polynucleotide
kinase were purchased from New England Biolabs.
g-32P-ATP was purchased from NEN1.
2H), 3.20 (t, 1H, J=10.0 Hz), 2.92 (m, 2H), 2.82 (m, 2H),
2.67 (t, 2H, J=5.9 Hz), 1.95 (m, 1H), 1.42±1.48 (m, 55H);
HRMS (FAB) m/z [M+Na]+ 1357.6183, calcd for
C57H102NaN6O25S2 1357.6234.
Compound 9. A solution of neomycin B (1) (1.0 g, 1.626
mmol) in a mixture of DMF (20 mL), water (4 mL) and
triethylamine (2 mL) was treated with di-tert-butyldi-
carbonate (2.1 g, 9.756 mmol, 6.0 equiv). The reaction
solution was heated to 60ꢀC for 5 h, then cooled to 23ꢀC.
The volatiles were removed in vacuo. The residue was
partitioned between water (300 mL) and ethyl acetate (600
mL). The aqueous layer was separated and extracted with
ethyl acetate (2Â250 mL). The combined organic layer
was washed with brine, dried over Na2SO4, and con-
centrated in vacuo. Flash column chromatography (4.3%
methanol in dichloromethane) a€orded the desired pro-
duct as a white solid (1.8 g, 91%). Rf 0.36 (10% methanol
in dichloromethane); 1H NMR (500 MHz, [D4]methanol,
25ꢀC): d5.28 (br, 1H), 5.16 (s, 1H), 4.90 (s, 1H), 4.18 (s,
2H), 3.96 (s, 1H), 3.82±3.90 (m, 3H), 3.76 (s, 1H), 3.64±
3.72 (m, 4H), 3.48 (m, 6H), 3.19±3.30 (m, 5H), 1.94
(m, 1H), 1.56 (m, 1H), 1.38±1.46 (m, 54H); HRMS (FAB)
m/z [M+Na]+ 1237.6141, calcd for C54H94NaN6O25
1237.6166.
Compound 12. A solution of tobramycin (2) (0.5 g, 1.070
mmol) in 14 mL aqueous DMSO (DMSO/water, 6/1) was
treated with di-tert-butyldicarbonate (1.4 g, 6.420 mmol,
6.0 equiv). The solution was heated at 60ꢀC for 4 h, then
cooled to 23ꢀC. A solution of 30% aqueous ammonia (5
mL) was added dropwise to the mixture. The precipitated
solid was ®ltered, washed with water (3Â200 mL), and
dried in vacuo (970 mg, 94%). Rf 0.31 (7.5% methanol in
1
dichloromethane); H NMR (500 MHz, [D4]methanol,
25ꢀC) d5.10 (br, 1H), 5.07 (br, 1H), 3.93 (m, 1H), 3.78 (m,
1H), 3.70 (m, 2H), 3.60 (m, 3H), 3.30±3.50 (m, 9H), 2.11
(m, 1H), 1.99 (m, 1H), 1.64 (q, 1H, J=12.5 Hz), 1.42±1.48
(m, 46H); HRMS (FAB) m/z [M+Na]+ 990.5102, calcd
for C43H77NaN5O19 990.5110.
Compound 13. A solution of 12 (0.3 g, 0.310 mmol) in
pyridine (5 mL) was treated with 2,4,6-triisopro-
pylbenzenesulfonyl chloride (0.66 g, 2.180 mmol, 7.0
equiv). The reaction mixture was stirred at 23ꢀC for 12
h. It was neutralized by adding hydrochloric acid (1.0
N), and partitioned between water (300 mL) and ethyl
acetate (600 mL). The aqueous layer was separated and
extracted with ethyl acetate (2Â250 mL). The combined
organic layer was washed with brine, dried over
Na2SO4, and concentrated in vacuo. Flash column
chromatography (2.3% methanol in dichloromethane)
a€orded the desired product as a white solid (240 mg,
Compound 10. A solution of 9 (1.0 g, 0.823 mmol) in
pyridine (20 mL) was treated with 2,4,6-triisopro-
pylbenzenesulfonyl chloride (8 g, 26.4 mmol, 32.0
equiv). The reaction mixture was stirred at 23ꢀC for 12
h and then neutralized by adding hydrochloric acid (1.0
N) and partitioned between water (300 mL) and ethyl
acetate (600 mL). The aqueous layer was separated and
extracted with ethyl acetate (2Â250 mL). The combined
organic layer was washed with brine, dried over
Na2SO4, and concentrated in vacuo. Flash column
chromatography (3.3% methanol in dichloromethane)
a€orded the desired product as a white solid (0.8 g, 66%).
Rf 0.40 (methanol±dichloromethane 1:10); 1H NMR (500
MHz, [D4]methanol, 25ꢀC) d7.32 (s, 2H), 5.45 (br, 1H),
5.18 (br, 1H), 4.60 (br, 1H), 4.35 (m, 1H), 4.26 (m, 2H),
4.15 (m, 4H), 3.88 (s, 1H), 3.78 (m, 1H), 3.73 (m, 2H), 3.60
(m, 1H), 3.50 (m, 4H), 3.36±3.42 (m, 4H), 3.20 (m, 2H),
2.96 (m, 1H), 1.95 (m, 1H), 1.56 (m, 1H), 1.38±1.46
(m, 5H), 1.27 (m, 18H); HRMS (FAB) m/z [M+Na]+
1503.7507, calcd for C68H116NaN6O27S 1503.7498.
1
65%). Rf 0.33 (7.5% methanol in dichloromethane); H
NMR (500 MHz, [D4]methanol, 25ꢀC) d7.28 (s, 2H), 5.09
(br, 2H), 4.40 (m, 1H), 4.27 (m, 1H), 4.14 (m, 3H), 3.72 (t,
1H, J=10.4 Hz), 3.40±3.60 (m, 12H), 2.94 (m, 1H), 2.04
(m, 2H), 1.64 (q, 1H, J=12.0 Hz), 1.42±1.48 (m, 46H),
1.26 (m, 18H); HRMS (FAB) m/z [M+Na]+ 1256.6487,
calcd for C58H99NaN5O21S 1256.6451.
Compound 14. A solution of 13 (150 mg, 0.122 mmol) and
cesium carbonate (70 mg, 0.220 mmol, 1.8 equiv) in DMF
(7 mL) was treated with bis(2-mercaptoethyl)ether (114
mL, 0.920 mmol, 7.5 equiv). The reaction mixture was
stirred at 23ꢀC for 12 h and then partitioned between
water (300 mL) and ethyl acetate (600 mL). The aqueous
layer was separated and extracted with ethyl acetate
(2Â250 mL). The combined organic layer was washed
with brine, dried over Na2SO4, and concentrated in vacuo.
Flash column chromatography (2.3% methanol in
dichloromethane) a€orded the desired product as a white
solid (125 mg, 94%). Rf 0.33 (7.5% methanol in dichloro-
Compound 11. A solution of 10 (480 mg, 0.324 mmol)
and cesium carbonate (200 mg, 0.613 mmol, 1.9 equiv)
in DMF (10 mL) was treated with bis(2-mercaptoethyl)-
ether (320 mL, 2.580 mmol, 8.0 equiv). The reaction
mixture was stirred at 23ꢀC for 12 h and then parti-
tioned between water (300 mL) and ethyl acetate (600
mL). The aqueous layer was separated and extracted
with ethyl acetate (2Â250 mL). The combined organic
layer was washed with brine, dried over Na2SO4, and
concentrated in vacuo. Flash column chromatography
(3.5% methanol in dichloromethane) a€orded the
desired product as a white solid (400 mg, 92%). Rf 0.43
(10% methanol in dichloromethane); 1H NMR (500
MHz, [D4]methanol, 25ꢀC) d5.35 (br, 1H), 5.15 (s, 1H),
4.95 (s, 1H), 4.24 (s, 2H), 4.10 (m, 1H), 3.89 (m, 2H), 3.76
(s, 1H), 3.72 (m, 1H), 3.68 (t, 2H, J=8.3 Hz), 3.62 (m,
3H), 3.55 (m, 3H), 3.46 (m, 1H), 3.36 (m, 2H), 3.26 (m,
methane); H NMR (500 MHz, [D4]methanol, 25ꢀC):
1
d=5.14 (s, 1H), 5.06 (s, 1H), 4.10 (m, 1H), 3.74 (t, 2H,
J=6.3 Hz), 3.70 (m, 1H), 3.60 (m, 6H), 3.36±3.50 (m,
6H), 3.04 (d, 1H, J=14.5 Hz), 2.92 (t, 2H, J=6.6 Hz),
2.78 (m, 2H), 2.70 (m, 1H), 2.65 (t, 2H, J=6.6 Hz), 2.13
(m, 1H), 2.01 (m, 1H), 1.63 (q, 1H, J=12.2 Hz), 1.44
(m, 46H); HRMS (FAB) m/z [M+Na]+ 1110.5186,
calcd for C47H85NaN5O19S2 1110.5178.
Compound 15. A solution of 14 (100 mg, 0.092 mmol) in
methanol (5 mL) was treated with 2,20-dipyridyldisul®de

K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
1369
(40 mg, 0.180 mmol, 2.0 equiv). After the reaction, the
mixture was stirred at 23ꢀC for 5 h, the solvent was
removed in vacuo. Flash column chromatography (3.5%
methanol in dichloromethane) a€orded the desired pro-
duct as a white solid (93 mg, 85%). Rf 0.32 (7.5%
methanol in dichloromethane); 1H NMR (500 MHz,
[D4]methanol, 25ꢀC) d 8.39 (m, 1H), 7.95 (m, 1H), 7.84
(m, 1H), 7.23 (m, 1H), 5.13 (br, 1H), 5.05 (br, 1H), 4.10
(m, 1H), 3.70 (m, 6H), 3.64 (t, 2H, J=6.2 Hz), 3.56
(t, 1H, J=6.3 Hz), 3.36±3.51 (m, 6H), 3.01 (m, 3H), 2.90
(m, 2H), 2.69±2.75 (m, 3H), 2.14 (m, 1H), 2.01 (m, 1H),
1.64 (q, 1H, J=12.3 Hz), 1.44 (m, 46H); HRMS (FAB)
m/z [M+Na]+ 1219.5125, calcd for C52H88NaN6O19S3
1219.5164.
NMR (500 MHz, D2O, 25ꢀC) d5.44 (d, 2H, H10, J=3.0
Hz), 5.03 (d, 2H, H100, J=3.5 Hz), 4.03 (m, 2H, H500), 3.83
(m, 6H, H50, S-S-CH2CH2O), 3.72 (m, 8H, H5, H30, 600-S-
CH2CH2O), 3.59 (m, 2H, H20), 3.52 (m, 2H, H200), 3.38 (t,
2H, H400, J=9.3 Hz), 3.30 (m, 6H, H4, H6, H40), 3.16 (m,
2H, H60), 3.02 (t, 4H, S-S-CH2CH2O, J=12.5 Hz), 2.94
(m, 10H, H1, H3, H300, 600-S-CH2CH2O), 2.80 (m, 6H,
H60, H600), 1.99 (m, 2H, H2eq), 1.26 (q, 2H, H2ax, J=12.0
Hz); 13C NMR (125 MHz, D2O, 25ꢀC) d 99.7, 98.9, 87.7,
85.5, 74.3, 72.7, 72.0, 71.7, 71.6, 71.5, 71.0, 69.4, 68.1, 54.0,
50.1, 48.9, 41.0, 37.4, 35.0, 33.3, 31.6; ESI MS m/z
[M+H]+ 1208, calcd for C44H87N8O22S4 1208.
Kan±Tob (8). The compound was synthesized using a
similar procedure as described for Tob±Neo (7). 1H NMR
(500 MHz, D2O, 25ꢀC) d5.50 (m, 1H, H10 of kanamycin
A), 5.35 (m, 1H, H10 of tobramycin), 5.05 (m, 2H, H100 of
both tobramycin and kanamycin A), 4.04 (m, 2H), 3.92
(m, 1H), 3.83 (m, 4H), 3.72±3.76 (m, 8H), 3.59 (m, 4H),
3.42 (m, 2H), 3.26±3.36 (m, 7H), 3.30 (m, 6H), 2.97±3.05
(m, 15H), 2.78±2.85 (m, 6H), 2.02±2.10 (m, 3H), 1.70 (q,
1H, J=12.0 Hz), 1.32 (m, 2H); 13C NMR (125 MHz,
D2O, 25ꢀC) d99.8, 98.6, 98.2, 87.8, 87.5, 84.7, 84.1, 74.7,
74.3, 72.5, 71.9, 71.8, 71.7, 71.5, 71.4, 71.0, 70.6, 70.1, 69.4,
69.3, 68.0, 66.0, 54.1, 54.0, 53.9, 50.1, 50.0, 49.2, 48.8, 40.8,
40.7, 37.3, 34.6, 34.5, 33.8, 33.3, 33.2, 31.8, 31.6; ESI MS
m/z calcd for C44H87N9O20S4 [M+H]+ 1191, found 1191.
Compound 16. A solution of 15 (80 mg, 0.067) in metha-
nol (5 mL) was treated with a solution of 11 (80 mg, 0.060
mmol, 0.9 equiv) in methanol (5 mL). The resulting solu-
tion was stirred at 23ꢀC for 10 h. The solvent was removed
in vacuo. Flash column chromatography (4% methanol in
dichloromethane) a€orded a white solid (95 mg, 60%). Rf
0.32 (10% methanol in dichloromethane); 1H NMR (500
MHz, [D4]methanol, 25ꢀC) d 5.33 (s, 1H), 5.14 (m, 2H),
5.06 (s, 1H), 4.94 (s, 1H), anomeric protons; MS (FAB) m/
z [M+Na]+ 2445, calcd for C104H185NaN11O44S4 2445.
Tob±Neo (7). Compound 16 (100 mg, 0.050 mmol) was
treated with 99% tri¯uoroacetic acid (4 mL) for 3 min at
23ꢀC. The volatiles were removed in vacuo. The residue
was dissolved in deionized water. The resulting solution
was treated with Amberlite-400 resin (10 g, OH form).
After 2 h at 23ꢀC, the mixture was ®ltered, concentrated,
and freeze-dried to a€ord a white solid (60 mg, 100%). 1H
NMR (500 MHz, D2O) d 5.62 (1H, J=2.5 Hz), 5.36 (d,
1H, J=3.0 Hz), 5.32 (s, 1H), 5.06 (d, 1H, J=3.0 Hz), 5.02
(s, 1H); anomeric protons 13C NMR (125 MHz, D2O,
25ꢀC) d109.7, 99.8, 98.5, 98.1, 87.7, 85.1, 84.0, 81.1, 79.8,
78.4, 76.6, 74.7, 73.2, 72.9, 72.4, 71.8, 71.5, 71.4, 70.5, 70.2,
69.6, 69.3, 68.5, 68.1, 66.0, 55.1, 54.0, 52.3, 50.3, 50.0, 49.8,
49.2, 48.8, 41.0, 40.9, 40.8, 37.5, 37.4, 35.1, 34.6, 34.4, 33.7,
33.3, 31.8, 31.4; ESI MS m/z calcd for C49H98N11O22S4
[M+H]+ 1321, found 1321.
UV spectra of dimeric aminoglycosides 4±8 show an
absorption at 250 nm (e=4.3Â102) indicating the for-
mation of a disul®de bond. In the symmetrical dimers,
4±6 NMR signals corresponded to half of the molecule.
In the nonsymmetrical dimers, the 1H NMR spectra
show severe overlap of the signals from the two mono-
mers. However, the appearance of the anomeric hydro-
gen signals from both aminoglycosides strongly suggest
the formation of the nonsymmetrical dimers. In addi-
tion, the mass spectra of all dimeric aminoglycosides
show satisfactory m/z signals which are in agreement
with their dimeric structure.
DNA preparation. The DNA template 50TTTTTAGA
GGGCCTATAGTGAGTCGTATTA30 and the T7 pro-
moter were synthesized by standard phosphoramidite
chemistry on a MilliGen/Biosearch Cyclone Plus DNA
synthesizer. The crude oligomers were puri®ed by gel
electrophoresis on 20% polyacrylamide/7M urea gels,
extracted with 1ÂTBE (pH 8.3) for 12 h, ®ltered and
desalted. The pT7L-21 plasmid26 was ampli®ed using a
QiagenTM maxiprep procedure. After digestion with Sca I
the linearized plasmid was phenol extracted and ethanol
precipitated and used without further puri®cation.
Tob±Tob (5). The compound was synthesized using a
similar procedure as described for Tob±Neo (7). 1H NMR
(500 MHz, D2O, 25ꢀC) d5.37 (d, 2H, H10, J=2.5 Hz), 5.05
(d, 2H, H100, J=4.0 Hz), 4.02 (m, 2H, H500), 3.84 (t, 4H, S-
S-CH2CH2O, J=5.6 Hz), 3.81 (m, 2H, H5), 3.74 (t, 4H,
600-S-CH2CH2O, J=5.6 Hz), 3.72 (m, 2H, H50), 3.58 (m,
4H, H40, H200), 3.44 (t, 2H, H400, J = 10.0 Hz), 3.34 (m,
4H, H4, H6), 3.29 (m, 2H, H300), 3.04±3.12 (m, 10H, H1,
H3, H20, S-S-CH2CH2O), 2.96 (m, 6H, one of H60, 600-S-
CH2CH2O), 2.83 (m, 4H, H600), 2.78 (q, 2H, H60, J1=15.0
Hz, J2=7.5 Hz), 2.10 (m, 2H, H30eq), 2.02 (m, 2H, H2eq),
1.78 (q, 2H, H30ax, J=12.0 Hz), 1.31 (q, 2H, H2ax,
J=12.5 Hz); 13C NMR (125 MHz, D2O, 25ꢀC) d 99.8,
97.9, 87.6, 83.8, 74.7, 71.8, 71.6, 71.2, 70.3, 69.3, 68.0, 65.9,
54.0, 50.0, 49.1, 48.7, 40.6, 37.3, 34.3, 33.5, 33.2, 31.8; ESI
MS m/z [M+Na]+ 1196, calcd for C44H88NaN10O18S4
1196.
RNA preparation. Substrate RNA (50GGCCCUCU
AAAAA30) was transcribed from the synthetic DNA
template with T7 RNA polymerase and nucleotide tri-
phosphates using the general procedure reported by
Uhlenbeck.34 The oligoribonucleotide was puri®ed by
electrophoresis on 20% polyacrylamide/7 M urea gels,
extracted with 250 mM Tris±HCl (pH 7.0) or 200 mM
KOAc, 1 mM EDTA (pH 5.45) for 12 h at 4ꢀC and
ethanol precipitated. The transcript was dephos-
phorylated with alkaline phosphatase and 50-labeled
Kan±Kan (6). The compound was synthesized using a
similar procedure as described for Tob±Neo (7). 1H

1370
K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
with g-32P-ATP and T4 polynucleotide kinase,35 gel
puri®ed as described above and ethanol precipitated.
The L-21 Sca I RNA was synthesized by in vitro tran-
scription of the Sca I linearized plasmid with T7 RNA
polymerase and nucleotide triphosphates. The crude
transcript was puri®ed by electrophoresis on a 6%
polyacrylamide/7 M urea gel. After extraction with 250
mM Tris±HCl (pH 7.0) or 200 mM KOAc, 1 mM
EDTA (pH 5.45) for 48 h at 4ꢀC the ribozyme was eth-
anol precipitated and the concentration was determined
photometrically using e260=3.2Â106.36
for curve ®tting (x=substrate concentration, n=Hill
coecient). Concentrations at in¯ection points indicate
IC50 values. All experiments were performed at least in
triplicate. Table 1 summarizes the averaged IC50 values
obtained and their standard deviations.
Due to the diculty in determining the concentration of
unbound dimers in solution, the in¯ection points of the
inhibition curves do not reveal `real' IC50 values, but
`apparent' IC50 values. Based on these assumptions the
apparent IC50 values fall at higher concentrations than
for the hypothetical simpler system with one binding
site only. Since the inhibition curves are used to com-
pare trends in RNA anity among similar ligands, the
determination of apparent IC50 values is sucient.
Determination of IC50 values. Nonadhesive, siliconized
polypropylene microcentrifuge tubes were used. In the
absence of aminoglycosides, a similar level of substrate
cleavage is observed in nonsiliconized, and in silico-
nized, nonadhesive tubes. There are no indications for
the adsorption of the ribozyme or its substrate on the
surface of any of the tubes. In the presence of dimeric
aminoglycosides the extent of substrate cleavage
depends on the type of tube used. We have observed
lower IC50 values for all dimeric derivatives when sili-
conized tubes were used. This indicates that dimeric
aminoglycosides might be partially adsorbed on non-
siliconized tube surfaces.
Estimation of error limits. The determination of IC50
values is reproducible with variations of smaller than
50% when the ribozyme catalyzed reactions are per-
formed using the same preincubation mixtures and
substrate stock solutions. We have observed con-
siderably greater variations in experiments performed
with di€erent stock solutions and di€erent substrate
preparations. The highest variation observed in IC50
values was 2.4-fold (observed for kanamycin A (3), the
worst ribozyme inhibitor). Independent of these
observed ¯uctuations, the inhibitory activities of the
aminoglycosides relative to each other show the same
trend. Part of the error is most likely derived from
handling extremely small volumes (<10 mL) which can
result in slight variations of concentrations.
The cleavage inhibition experiments were conducted
under single-turnover conditions with a saturating
ribozyme concentration and a limiting GTP concentra-
tion.36 Multiple-turnover conditions are unfavored
because complicated reaction pro®les, partially due to
product inhibition, would be expected. A trace amount
of 32P-labeled substrate was used. Every batch of radio-
labeled substrate was initially tested in cleavage inhi-
bition experiments. The reaction rate was controlled by
adjusting the GTP concentrations in such a way that
initial kinetics could be measured within the ®rst 5 min.
L-21 Sca-I ribozyme (®nal concn 50 nM), GTP (®nal
concn 1±10 mM) and the aminoglycoside in varying
concentrations was preincubated at 50ꢀC for 10 min in
HEPES bu€er, pH 7.0 (®nal concn 50 mM) in the pre-
sence of magnesium chloride (®nal concn 10 mM) and
0.01% Nonidet P 40 (nonionic detergent). After cooling
down to room temperature and equilibration for 1.5 h
the reaction was initiated by adding a 32P-50-labeled
substrate RNA (®nal concn ca. 1±10 nM for single-
turnover conditions) to reach a typical reaction volume
of 5 or 10 mL. Incubation for 1 or 1.5 min at room
temperature was followed by quenching with twice the
reaction volume of a stop solution (7 M urea, 50 mM
EDTA and 0.1ÂTBE, pH 8.3, 0.05% xylene cyanole
and 0.05% bromophenol blue). The reaction mixtures
were resolved on denaturing 20% polyacrylamide gels.
After quanti®cation of the ratio of full length substrate/
cleavage product utilizing a Molecular Dynamics Phos-
phorimager and Image QuantTM software, semiloga-
rithmic plots give sigmoidal inhibition curves. Each data
point (x) obtained in percentage of substrate cleavage
was normalized into relative inhibition by calculating
Initial kinetics. The determination of initial kinetics in
the absence and presence of aminoglycosides was car-
ried out under the same conditions as described above,
except that the reaction was initiated by adding a mix-
ture of 32P-50-labeled substrate RNA and GTP. These
experiments were carried out in nonsiliconized micro-
centrifuge tubes. The total reaction volume was 10 mL.
After initiation, ®ve aliquots of 2 mL each were removed
every minute and the reaction was stopped by quickly
mixing the aliquot with 4 mL of a stop solution. The
samples were heated up to 70ꢀC for 3 min and loaded
on a 20% denaturing polyacrylamide gel. The ratio of
full-length substrate/cleavage product was quanti®ed as
bx
described above. The function f ˆ a‰1
e
Š ‡ c was
used for curve ®tting (a=amplitude of exponential,
b=rate constant, c=zero intercept).
Acknowledgements
This work was supported by the university-wide AIDS
Research Program, University of California (R97-SD-
036), the National Institutes of Health (AI 40315) and
the Hellman Faculty Fellowship (Y.T.) We are grateful
to Dr. Scott Strobel (Yale University) for helpful dis-
cussions and the initial donation of pT7L-21 plasmid
and L-21 Sca I RNA as well as substrate RNA. We are
indebted to Dr. Jack Kyte and Dr. Marian Mosior
(University of California, San Diego) for valuable com-
ments, and to Dr. Eric Westhof (IBMC, Strasbourg) for
communicating results prior to publication.
1
correspond to the baseline and the plateau of a sigmoi-
‰x xlow=xhigh xlowŠ. The values for xhigh and xlow
dal curve. The empirical function f ˆ ꢂx=IC50†n=‰1 ‡
n
ꢂx=IC50† Š derived from the Hill Equation was applied

K. Michael / Bioorg. Med. Chem. 7 (1999) 1361±1371
1371
References
17. Clouet-d'Orval, B.; Stage, T. K.; Uhlenbeck, O. C. Bio-
chemistry 1995, 34, 11186.
1. Davies J., Davis B. D., J. Biol. Chem. 1968, 243, 3312.
2. Moazed, D.; Noller, H. F. Nature 1987, 327, 389.
3. Zapp, M. L.; Stern, S.; Green, M. R. Cell 1993, 74, 969.
4. Mei, H.-Y.; Galan, A. A.; Halim, N. S.; Mack, D. P.;
Moreland, D. W.; Sanders, K. B.; Truong, H. N.; Czarnik,
A.W. Bioorg. Med. Chem. Lett. 1995, 5, 2755.
5. Koeda, T.; Umemura, K.; Yokota, M. In Aminoglycoside
Antibiotics. Umezawa, H., Hooper, R., Eds.; Springer±Verlag:
Berlin, 1982; Vol. 62, pp 293±356.
18. Hermann, T.; Westhof, E. J. Mol. Biol. 1998, 276, 903.
19. Tor, Y.; Hermann, T.; Westhof, E. Chem. Biol. 1998, 5, R277.
20. Wang, H.; Tor, Y. Bioorg. Med. Chem. Lett. 1997, 7, 1951.
21. Hendrix, M.; Priestley, E. S.; Joyce, G. F.; Wong, C.-H. J.
Am. Chem. Soc. 1997, 119, 3641.
22. Kirk, S. R.; Tor, Y. Bioorg. Med. Chem. 1999, in press.
23. Cech, T. Annu. Rev. Biochem. 1990, 59, 543.
24. Cate, J. H.; Hanna, R. L.; Doudna, J. A. Nature Struct.
Biol. 1997, 4, 553.
6. von Ahsen, U.; Davies, J.; Schroeder, R. Nature 1991, 353,
368.
7. von Ahsen, U.; Davies, J.; Schroeder, R. J. Mol. Biol. 1992,
226, 935.
8. Davies, J.; von Ahsen, U.; Schroeder, R. In The RNA
World; Gesteland, R. F., Atkins, J. F., Eds.; Cold Spring
Harbor Laboratory Press: New York, 1993. pp 185±204.
9. Stage, T. K.; Hertel, K. J.; Uhlenbeck, O. C. RNA 1995, 1,
95.
10. Wang, H.; Tor, Y. J. Am. Chem. Soc. 1997, 119, 8734.
11. Rogers, J.; Chang, A. H.; von Ahsen, U.; Schroeder, R.;
Davies, J. J. Mol. Biol. 1996, 259, 916.
12. Chia, J.-S.; Wu, H.-L.; Wang, H.-W.; Chen, D.-S.; Chen,
P.-J. J. Biomed. Sci. 1997, 4, 208.
13. Michael, K.; Tor, Y. Chem. Eur. J. 1998, 4, 2091.
14. Hendrix, M.; Alper, P. B.; Priestley, E. S.; Wong, C.-H.
Angew. Chem., Int. Ed. Engl. 1997, 36, 95.
15. Wong, C.-H.; Hendrix, M.; Manning, D. D.; Rosenbohm,
C.; Greenberg, W.A. J. Am. Chem. Soc. 1998, 120, 8319.
16. Wang, H.; Tor, Y. Angew. Chem., Int. Ed. 1998, 37, 109.
25. Michel, F.; Westhof, E. J. Mol. Biol. 1990, 216, 585.
26. Zaug, A. J.; Grosshans, C. A.; Cech, T. R. Biochemistry
1988, 27, 8924.
27. Grosshans, C. A.; Cech, T. R. Biochemistry 1989, 28, 6888.
28. deHaseth, P. L.; Lohman, T. M.; Record, Jr., M. T. J.
Mol. Biol. 1977, 16, 4783.
29. Misra, V. K.; Sharp, K. A.; Friedman, R. A.; Honig, B. J.
Mol. Biol. 1994, 238, 245.
30. Dorman, D. E.; Paschal, J. W.; Merkel, K. E. J. Am.
Chem. Soc. 1976, 98, 6885.
31. Botto, R. E.; Coxon, B. J. Am. Chem. Soc. 1983, 105, 1021.
32. Szilagyi, L.; Pusztahelyi, Z. S.; Jakab, S.; Kovacs, I. Car-
bohydr. Res. 1993, 247, 99.
33. Williams, D. H.; Westwell, M. S. Chem. Soc. Rev. 1998, 27, 57.
34. Milligan, J. F.; Groebe, D. R.; Witherell, G. W.; Uhlen-
beck, O.C. Nucleic Acids Res. 1987, 15, 8783.
35. Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular
Cloning (2nd ed.); Cold Spring Harbor Laboratory Press: New
York, 1989.
36. Herschlag, D.; Cech, T. R. Biochemistry 1990, 29, 10159.