Artificial nucleobase-amino acid conjugates: A new class of TAR RNA binding agents

Article abstract of DOI:10.1002/chem.201303664

The human immunodeficiency virus type-1 (HIV-1) Tat protein stimulates transcriptional elongation. Tat is involved in the transcription machinery by binding to the transactivation response region (TAR) RNA stem-loop structure, which is encoded by the 5 leader sequence found in all HIV-1 mRNAs. Herein, we report the rational design, synthesis, and in vitro evaluation of new RNA binding agents that were conceived in order to bind strongly and selectively to the stem-loop structure of TAR RNA and, thus, inhibit the Tat/TAR interaction. We have demonstrated that the conjugation of modified nucleobases, able to interact specifically with an RNA base pair, and various amino acids allows these motifs to bind the target RNA selectively and in a cooperative manner that leads to the inhibition of viral replication in HIV-infected cells. Kissing interaction: The rational design, synthesis, and in vitro evaluation of new RNA binding agents are reported. The conjugation of an artificial nucleobase, which is able to interact specifically with an RNA base pair (Hoogsteen pairing), and various amino acids leads to efficient binding to the target RNA and the inhibition of viral replication in HIV-infected cells .

Full text of DOI:10.1002/chem.201303664

DOI: 10.1002/chem.201303664
Full Paper
&
Cooperative Binding
Artificial Nucleobase–Amino Acid Conjugates: A New Class of TAR
RNA Binding Agents**
Jean-Patrick Joly,[a] Guillaume Mata,[a] Patrick Eldin,[b] Laurence Briant,[b] Fabien Fontaine-
Vive,[a] Maria Duca,*[a] and Rachid Benhida*[a]
Abstract: The human immunodeficiency virus type-1 (HIV-1)
Tat protein stimulates transcriptional elongation. Tat is in-
volved in the transcription machinery by binding to the
transactivation response region (TAR) RNA stem-loop struc-
ture, which is encoded by the 5’ leader sequence found in
all HIV-1 mRNAs. Herein, we report the rational design, syn-
thesis, and in vitro evaluation of new RNA binding agents
that were conceived in order to bind strongly and selectively
to the stem-loop structure of TAR RNA and, thus, inhibit the
Tat/TAR interaction. We have demonstrated that the conju-
gation of modified nucleobases, able to interact specifically
with an RNA base pair, and various amino acids allows these
motifs to bind the target RNA selectively and in a coopera-
tive manner that leads to the inhibition of viral replication in
HIV-infected cells.
Introduction
selected HIV-1 TAR RNA as a biologically relevant model to
quantify RNA–small-molecule binding, to decipher the mode
of interaction, and to examine the in vitro activity.
In recent years, RNA has been discovered to play important
roles in many cellular processes and it is now recognized as
a valid target for therapeutic intervention by drug-like small
molecules.[1] Furthermore, many RNAs are related to diseases,
either due to mutations that affect their normal function or be-
cause they are part of pathogenic organisms that invade
human cells. Hence, RNA is an important target for the design
of inhibitors directed at preventing pathological processes.[2]
One such target is transactivation response region (TAR) RNA,
a 59-nucleotide fragment from the human immunodeficiency
virus type-1 (HIV-1) genome that forms a highly conserved
stem-loop structure containing an internal bulge[3] and that is
found at the 5’ end of all nascent HIV-1 transcripts. HIV-1 TAR
RNA has a key role in viral replication because its interaction
with the Tat protein allows the formation of a complex involv-
ing cellular cofactors such as cyclin T1 and its cognate kinase
CDK9, which thus stimulates efficient transcription from the
retroviral promoter (LTR).[4] Therefore, in the present study, we
Given the global problem of HIV-1 infection and the high
frequency of drug resistance to the current HIV therapies, the
discovery of antiviral drugs that exhibit novel modes of action
are of the utmost importance. The majority of studies on RNA
targeting have focused on oligonucleotide-based antisense
strategies, in which high-binding-affinity hybridization between
the oligonucleotide and the target strand serves to alter gene
expression.[5] However, even if these RNA binders exhibit high
binding affinity and specificity, their intrinsic instability and
their poor ability to penetrate the cell membrane greatly
hamper their therapeutic application.[6] Peptides have also
been studied as potent ligands of hairpin RNAs on the basis of
their similarity with natural RNA ligands. Although a great
number of short peptides and analogues have already been re-
ported in the literature, none of them has shown sufficient
specificity in order to be used in therapy.[7] Finally, aminoglyco-
sides constitute a particularly well-studied class of RNA-binding
molecules, which are essentially pseudo-oligosaccharides con-
taining numerous amine groups within their structure. These
molecules are well known for their antibiotic properties, which
are related to their capacity to interact with bacterial ribosomal
RNA, a property that impairs protein biosynthesis.[8] However,
because their binding mechanism mainly involves nonspecific
electrostatic interactions, aminoglycosides suffer from a lack of
selectivity, which is responsible for their high toxicity.
[a] J.-P. Joly, G. Mata, Dr. F. Fontaine-Vive, Dr. M. Duca, Dr. R. Benhida
Institut de Chimie de Nice UMR7272 CNRS
Universitꢀ de Nice Sophia Antipolis
Parc Valrose, 06108 Nice cedex 1 (France)
Fax: (+33)4-92076151
E-mail: duca@unice.fr
benhida@unice.fr
[b] Dr. P. Eldin, Dr. L. Briant
Centre d’Etudes d’Agents Pathogꢁnes et Biotechnologies pour la Santꢀ
UMR5236 CNRS
Universitꢀ de Montpellier 1 and 2
Recently, based on a rational design, we discovered a new
class of modified nucleosides able to bind specifically to TAR
RNA at the stem-bulge junction.[9] These nucleosides contain
a modified nucleobase (S, Figure 1A) that is able to initiate
RNA recognition in a sequence-selective manner by forming
a triplet with an AU base pair.[10] The introduction of basic
1919 route de Mende, 34293 Montpellier cedex 5 (France)
[**] TAR: Transactivation response region.
Supporting information for this article (containing the synthetic procedures)
is available on the WWW under http://dx.doi.org/10.1002/chem.201303664.
Chem. Eur. J. 2014, 20, 2071 – 2079
2071
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Results and Discussion
Synthesis of new modified nucleobase–amino acid conju-
gates
In a first approach, we envisioned the synthesis of a class of
conjugates in which only one amino acid is coupled to the arti-
ficial nucleobase S through an amide bond (series 1, Fig-
ure 1B). We thus synthesized the first series of conjugates (1b–
8b) by coupling S and various amino acids, as depicted in
Scheme 1. For this first series, we selected the basic amino
acids lysine (Lys), arginine (Arg), and histidine (His), the aromat-
ic amino acids phenylalanine (Phe), tyrosine (Tyr), and trypto-
phane (Trp), and the aliphatic amino acids alanine (Ala) and
threonine (Thr). The basic amino acids should allow for strong
electrostatic interactions with the bulge bases of the TAR se-
quence, as previously demonstrated for glycoconjugates con-
taining Lys and Arg.[9] The aromatic amino acids should ac-
count for more hydrophobic interactions that could be estab-
lished with the surrounding free RNA nucleobases or the base
pairs at the bulge. The aliphatic amino acids should involve hy-
drophobic interactions inside the RNA helix. In Ala, only the a-
NH2 group is available for interaction, whereas the hydroxy
group of the side chain in Thr could also participate in hydro-
gen bonding.
The synthesis of compounds 1b–8b comprises only two
steps. The first step involved the coupling of S with protected
amino acids, in the presence of 2-chloro-1-methylpyridinium
iodide in CH2Cl2, which yielded the protected compounds 1a–
8a in moderate to good yields (42–91%, Scheme 1A). The
second step consisted of the removal of the Boc and tBu pro-
tecting groups, as appropriate, in the presence of TFA in
CH2Cl2, which led to compounds 1b, 3b, 4b, 7b, and 8b. An
intermediate step for the removal of the Fmoc group (20% pi-
peridine in CH2Cl2) is necessary with compounds 5a and 6a
and led compounds 5a’ and 6a’, respectively. The final cleav-
age of the tBu and Boc groups, respectively, led to desired
compounds 5b and 6b. The arginine derivative 2a was con-
verted by hydrogenolysis (H2, Pd/C) into compound 2b. All of
the deprotected compounds, 1b–8b, were obtained in high
yields (88–99%) and were fully characterized.
Figure 1. A) Mode of interaction in an S–A·U triplet. B) General structure of
the newly synthesized RNA ligands (S–amino acid conjugates): series 1 and
2. C) The RNA target (TAR HIV-1).
amino acids, such as lysine and arginine, at the 3’ and 5’ posi-
tions of the nucleoside strengthened the interaction with tar-
geted RNA in a cooperative way. Unfortunately, we found that
this class of compounds has no activity in cellular HIV assays. A
careful study clearly pointed out their rapid degradation in bio-
logical media through cleavage of the labile ester bond (see
the Supporting Information).
In line with these observations and taking advantage of the
previously modified nucleobase S, we report herein the
design, synthesis, and biological evaluation of S–amino acid
conjugates as potent, cell-permeable, and stable TAR RNA li-
gands. In order to increase the in vitro stability of this novel
class of RNA binders and to decipher the mode of action, we
modified the three-dimensional distribution of the substituents
by replacing the 2’-deoxyribose scaffold with chemically stable
spacers between S and the amino acids (series 1 and 2, Fig-
ure 1B). We also investigated the effect of the amino acid and
the linker length on the binding affinity (series 1, Figure 1BB),
as well as the effect of introducing two amino acids in the
same ligand in order to further improve their affinity and to
maintain their specificity (series 2, Figure 1B).
After the synthesis of this first set of compounds, we decid-
ed to introduce a spacer between the two binding modules (S
and amino acids) by using aliphatic linkers of different length:
glycine in compound 9d, 4-aminobutyric acid in compound
10d, and 6-aminocaproic acid in compound 11 d (Scheme 1B).
These compounds all contain the histidine amino acid and will
allow us to study the effect of linker size on the overall effi-
ciency of the ligands in terms of affinity and selectivity. Indeed,
the best linker should orient S and the amino acid moieties for
optimal interaction with TAR RNA. The preparation of this
second set of compounds is depicted in Scheme 1B. First, S
was coupled with Boc-protected linkers (N-Boc-glycine, N-Boc-
4-aminobutyric acid, and N-Boc-6-aminocaproic acid) to afford
compounds 9a–11 a, respectively, in moderate yields (32–
55%). The subsequent removal of the Boc group in the pres-
ence of TFA in CH2Cl2 led to compounds 9b–11 b in almost
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2072
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
third set of compounds is illustrated in Scheme 2. For
these compounds, we decided to couple the lysine,
arginine, and histidine amino acids (in 13b–15b, re-
spectively), which, as mentioned above, were expect-
ed to bind strongly to RNA, thanks to their basic
nature. In the first step, nucleobase S was coupled
with N,N’-di-Boc-2,3-diaminopropanoic acid by using
a similar coupling procedure to that described above
to afford the desired compound 12a in 53% yield.
The second step consisted of Boc cleavage in the
presence of TFA and led to compound 12b in 99%
yield. The spacer’s free amino groups were then cou-
pled with protected amino acids (Boc-Lys for 13c, Z-
Arg for 14c, and Boc-His for 15c) in the presence of
HOBt, DIC, DMAP, and Et3N in CH2Cl2. which led to
compounds 13a–15a in 65–71% yields. The final de-
protection step was achieved by using TFA for com-
pounds 13a and 15a or by catalytic hydrogenation
for compound 14a to afford high yields (83–99%) of
the corresponding products 13b–15b, as a mixture
of two diastereomers.
Ligand binding to HIV-1 TAR RNA: affinity and spe-
cificity
After the synthesis of the first series of compounds
(1b–8b and 9d–11 d, Scheme 1), we evaluated their
ability to bind HIV-1 TAR RNA. These binding studies
were performed in 20 mm 2-[4-(2-hydroxyethyl)-1-pi-
perazinyl]ethanesulfonic acid (HEPES) buffer (pH 7.2)
containing 20 mm NaCl, 140 mm KCl, and 3 mm
MgCl2 at 208C by measuring the fluorescence change
of a fluorescently labeled TAR fragment (5’-Alexa488
)
with increasing concentrations of ligands. This
method, broadly used to study the interactions be-
tween RNA and small ligands,[11] allowed us to deter-
mine the dissociation constants (Kd) for all of the
compounds (Table 1) after analysis of the binding iso-
therm diagram. Neomycin was used as a positive
control. This compound belongs to the class of ami-
noglycoside antibiotics and, besides its ability to bind
bacterial ribosomal RNA, it is also reported as a very
efficient and general RNA ligand.[12]
Scheme 1. Synthesis of ligands with one amino acid (series 1). A) Synthesis of com-
pounds 1b–8b. Reagents: a) Boc-Lys-OH for 1a, Z-Arg-OH for 2a, Boc-His-OH for 3a,
Boc-Phe-OH for 4a, Fmoc-Tyr(tBu)-OH for 5a, Fmoc-Trp(Boc)-OH for 6a, Boc-Ala-OH for
7a, Boc-Thr(tBu)-OH for 8a, 2-chloro-1-methylpyridinium iodide, Et3N, CH2Cl2, reflux, 3 h;
b) 50% TFA, CH2Cl2, RT, 24 h, for 1b, 3b, 4b, 6b–8b; c) 20% piperidine, CH2Cl2, RT, 24 h,
for 5b and 6b; d) H2, Pd/C, 24 h, for 2b. B) Synthesis of histidine conjugates with differ-
ent linker lengths (9d–11 d). Reagents: a) N-Boc-l-glycine for compound 9a, N-Boc-4-
aminobutyric acid for 10a, N-Boc-6-aminocaproic acid for 11 a, 2-chloro-1-methylpyridini-
um iodide, Et3N, CH2Cl2, reflux, 3 h; b) 50% TFA, CH2Cl2, RT, 24 h; c) HOBt, DIC, DMAP,
Et3N, CH2Cl2, RT, 4 h. Boc: tert-butoxycarbonyl; Z: benzyloxycarbonyl; Fmoc:9-fluorenyl-
methoxycarbonyl; TFA: trifluoroacetic acid; HOBt: 1-hydroxy-1H-benzotriazole; DIC: diiso-
propylcarbodiimide; DMAP: 4-dimethylaminopyridine.
As illustrated by the Kd values reported in Table 1,
compounds 1b, 2b and 6b, in which S was coupled
with lysine, arginine, and tryptophan, respectively, ex-
hibited increased affinity for TAR RNA (Kd =7.5, 2.4,
quantitative yields (97–99%). The spacer’s free amino group
was coupled with Boc-protected histidine in the presence of
HOBt, DIC, DMAP, and Et3N in CH2Cl2, which led to compounds
9c–11 c in 54–67% yields. The final deprotection step was ach-
ieved by using TFA and led to compounds 9d–11 d in quanti-
tative yields (99%).
and 3.1 mm, respectively; Table 1, entries 2, 3, and 7) relative to
that of neomycin (Kd =17.2 mm; Table 1, entry 1). Compound
3b, which featured a histidine residue, showed a binding affin-
ity very close to that of neomycin (Kd =17.5 mm; Table 1,
entry 4). The aromatic amino acids Phe and Tyr seemed to be
less favorable for the interaction than neomycin because com-
pound 4b had a moderate affinity (Kd =30.3 mm) and 5b was
not able to bind the target (Table 1, entries 5 and 6, respective-
ly). In similar way, aliphatic amino acids 7b and 8b bound TAR
RNA with low or no affinity (Table 1, entries 8 and 9, respective-
Finally, we decided to investigate the effect of introducing
a second amino acid in the same molecule to increase the af-
finity. To this end, we selected a spacer bearing two reactive
amines (2,3-diaminopropanoic acid), and the synthesis of this
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2073
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
interact with TAR RNA in a cooperative way to increase the
overall affinity.
The fact that 1b (Lys), 2b (Arg), 3b (His), and 6b (Trp) bind
with high affinity compared with 4b (Phe), 5b (Tyr), 7b (Ala),
and 8b (Thr) is probably due to additional interactions be-
tween the functional groups of the amino acid side chains
(Lys, Arg, His, and Trp) and TAR RNA. In a similar way, the in-
creased affinity observed with conjugates 3b (His) and particu-
larly 6b (Trp) could be ascribed to the possibility of additional
H-bonding involving the NH amino groups of the imidazole
(His) and indole moieties (Trp). The increased affinity of 6b rel-
ative to that of 3b (3.10 mm versus 17.5 mm, respectively) could
be ascribed to the known proptotropic tautomerism inherent
to the imidazole ring (N1–N3 H shift) compared with the
indole system.[13]
In the case of ligands 9d–11 d, each bearing a histidine resi-
due linked to the nucleobase with a spacer of 1, 3, or 5 carbon
atoms, respectively, we clearly observed that short linkers give
rise to a more favorable interaction (Table 1, entries 13–15). In
fact, a study of the Kd values demonstrated that the C1 linker
arm (glycine) of compound 9d had the best affinity (15.3 mm)
compared with those of 10d (C3, 30.5 mm) and 11 d (C5, no af-
finity). The use of a longer and more flexible linker probably
exposed the imidazole domain out of the helix and led to dis-
favored interactions. By contrast, a short linker length gives
rise to a ligand with a more rigid conformation and favorable
interactions.
Scheme 2. Synthesis of ligands featuring two amino acids (13b–15b). Re-
agents: N,N’-di-Boc-2,3-diaminopropanoic acid, 2-chloro-1-methylpyridinium
iodide. a) Et3N, CH2Cl2, reflux, 3 h; b) 50% TFA, CH2Cl2, RT, 24 h; c) Boc-Lys-
OH for 13a, Z-Arg-OH for 14a, Boc-His-OH for 15a, HOBt, DIC, DMAP, Et3N,
CH2Cl2, RT, 4 h; d) H2, Pd/C, 24 h for 14b.
Encouraged by the results obtained with the first series of
compounds, we synthesized ligands 13b (di-Lys), 14b (di-Arg),
and 15b (di-His) bearing two amino acid residues in order to
further increase the affinity. These conjugates were obtained
by coupling S and the appropriate amino acids with diamino-
propanoic acid as a linker (Scheme 2). Interestingly, 13b–15b
showed the best Kd values of 0.24, 0.11, and 1.82 mm, respec-
tively (Table 1, entries 16–18). These results are of great impor-
tance because the affinities of these small molecules, particu-
larly 13b and 14b, are approximately three- to eightfold
higher than that of the previously reported high-molecular-
weight Tat peptide (15-mer, Kd =0.71–0.81 mm).[14] These high
observed affinities could be ascribed to the additional coopera-
tive interactions involving the second amino acid.
Table 1. Dissociation constants for the ligand–TAR RNA interaction.[a]
Entry
Ligand
Kd (TAR) [mm]
1
2
neomycin
1b
17.2
7.5
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
2b
3b
4b
5b
6b
7b
8b
S
Lys
Arg
9d
10d
11 d
13b
14b
15b
2.44
17.5
30.3
n.b.[b]
3.10
47.4
n.b.[b]
n.b.[b]
n.b.[b]
n.b.[b]
15.3
30.5
n.b.[b]
0.24
0.11
To evaluate the specificity of this class of compounds, the
binding affinities of the best ligands were determined in the
presence and absence of a 100-fold excess of tRNA (K’d) and
a 100-fold excess of double-stranded DNA (dsDNA; K’’d), re-
spectively (Table 2). For ligands with one amino acid, we ob-
served that 1b (Lys) and 3b (His) maintained their affinity
toward TAR RNA and were highly specific (1.1<K’d/Kd <1.4 and
1.2<K’’d/Kd <1.5; Table 2, entries 2 and 4), whereas 2b (Arg)
showed very low specificity (K’d/Kd =10.7 and K’’d/Kd =7.4;
Table 2, entry 3). For ligands featuring two amino acids, we ob-
served that 13b (di-Lys) and 14b (di-Arg) exibited only a mini-
mal decrease of the specificity compared to 1b and 2b (K’d/
Kd =2.9 and 3.2, K’’d/Kd =3.0 and 3.6, respectively; Table 2, en-
tries 5 and 6). In contrast, 15b (di-His) showed a significant de-
crease of the specificity in the presence of tRNA (K’d/Kd =25;
Table 2, entry 7) and DNA competitors (K’’d/Kd =6.5) relative to
1.82
[a] Fluorescence measurements were performed in buffer A (20 mm
HEPES, pH 7.4, 20 mm NaCl, 140 mm KCl, and 3 mm MgCl2). Kd values are
given with an uncertainty of Æ10%. [b] n.b.: no binding.
ly). Interestingly, we observed that the artificial nucleobase S
alone and the free lysine and arginine amino acid residues
were not able to bind to TAR RNA (Table 1, entries 10–12).
These results clearly demonstrate that these ligands bear an
original mode of binding, in which both S and the amino acid
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2074
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
total free energy, such as nonionic hydrophobic ef-
fects driven by entropy, and specific interactions, in-
cluding H-bonds, van der Waals interactions, and p
stacking, and 2) the pure electrostatic (polyelectro-
lyte) contribution, DGoel, which reflects the ionic inter-
actions occurring between two groups of opposite
charge and is highly dependent on the salt concen-
tration. As expected, and in line with the Kd values,
we found that the interactions between all of the S–
amino acid conjugates and TAR RNA mainly involve
specific nonelectrostatic interactions because the
DGonel component represents 81–96% of the overall
free energy (DG8; Table 3), whereas the neomycin–
TAR interaction is clearly dominated by the electro-
static component (DGonel =33% and DGeol =67%). In-
terestingly, we also observed that an increase in am-
monium groups has no significant effect on the DGnoel
Table 2. Competition assays in the presence of tRNA and dsDNA.[a]
Entry
Ligand
Kd (TAR)
[mm]
K’d (TAR) with
tRNA[b] [mm]
K’d/Kd
K’’d (TAR) with
dsDNA[c] [mm]
K’’d/Kd
1
2
3
4
5
6
7
neomycin
1b
2b
17.2
7.5
78.5
10.4
26.2
18.10
0.69
0.35
45.5
4.6
1.4
10.7
1.1
2.9
3.2
25
n.b.
11.5
–
1.5
7.4
1.2
3.0
3.6
6.5
2.44
17.5
0.24
0.11
1.82
18.20
20.5
0.72
0.40
11.89
3b
13b
14b
15b
[a] Fluorescence measurements were performed in buffer A (20 mm HEPES, pH 7.4,
20 mm NaCl, 140 mm KCl, and 3 mm MgCl2). Kd values are given with an uncertainty
of Æ10%. [b] Measured in the presence of a 100-fold excess of a mixture of natural
tRNAs (tRNA mix). [c] Measured in the presence of a 100-fold excess of a 15-mer DNA.
n.b.: no binding.
the specificity of its analogue, 3b, bearing one amino acid (K’d/
Kd =1.1 and K’’d/Kd =1.2; Table 2, entry 4).
component because ligands 13b–15b, which contain twice as
many ammonium groups as 1b–6b, have very similar contri-
butions from the DGonel values to those obtained for 1b–6b
(1b/13b, 2b/14b,and 3b/15b; Table 3). This original mode of
binding could be ascribed to: 1) the high specific H-bonding
contribution of the S system and 2) the optimal position of the
amino acid, which allows additional specific interactions.[10b,15]
These results also showed that the overall free energy (DG8) is
driven by the enthalpy of the binding (DH8) and by the high
contribution of its nonelectrostatic component (DGonel; Table 3).
Finally, CD spectra were recorded in the absence and pres-
ence of ligands to show whether the involved interactions
affect the RNA structure. A typical example is shown for 13b,
which was selected for its high affinity (Figure 2). The CD spec-
trum of the TAR hairpin alone shows strong positive and nega-
tive peaks at 265 and 210 nm, respectively, and a weak nega-
tive signal at 240 nm, in accordance with the A form of RNA.
Addition of one equivalent of ligand 13b slightly affects the
signal at 210 nm. This indicates that the involved interactions
do not produce a significant change in the RNA structure, and
the contribution of the S system (H-bonding) does not abolish
the overall base stacking (kissing interaction). In a similar way,
when the ligand concentration was increased to 5 or 10 equiv-
alents (high excess), a similar
From these results, we can conclude that: 1) compared to
neomycin, the S-based ligands have higher TAR affinities and
better specificities and 2) S conjugates featuring one amino
acid have slightly lower TAR affinities but better specificities
than their S–diamino acid analogues. These results also illus-
trate how the binding affinity and specificity of S–amino acid
conjugates can be tuned by optimizing the linker length and
the nature and number of amino acid residues.
To get further insights on the ligand–TAR binding mode, the
thermodynamic parameters associated with the formation of
the complexes were determined. Nonelectrostatic (DGonel, DHnoel
,
and TDSonel) and electrostatic (DGeol, DHoel, and TDSeol) parameters
were obtained by plotting the Gibbs free energy (DG8) versus
the temperature and by examining the dependency of the dis-
sociation constants on the ionic strength of the solution
(Table 3). The enthalpy of binding (DH8) is independent of the
salt concentration, so only the contribution of the Gibbs
energy to the overall binding is discussed in this section.
Indeed, the DG8 value, which represents the total energy, can
be divided into two components: 1) the DGonel value, which re-
flects the contribution of nonelectrostatic interactions to the
decrease of signal intensity at
210 nm was observed, which is
Table 3. Thermodynamic parameters for ligand–TAR interactions.
mainly due to the phosphate
(TAR)–ammonium (residual ex-
[b]
o [c]
Ligand
DG8
DH8[a]
TDS8[a]
TDS8/DH8 DGonel
TDSonel
DG
el
[kJmolÀ1
]
[kJmolÀ1
]
[kJmolÀ1
]
[kJmolÀ1
]
[kJmolÀ1
]
[kJmolÀ1
]
cess of ligand) interactions.
Moreover, as attested by NMR,
fluorescence, and CD spectra of
13b, no self-structural organiza-
tion of this ligand was observed
at high concentrations, which
completely excluded the contri-
bution of 13b alone to the ob-
served CD variations at 210 nm.
The fact that the CD spectra
only indicated minimal alteration
of the base stacking clearly indi-
neomycin À26.7
À26.7Æ1.3
À38.1Æ1.5
À38.4Æ0.62
À49.0Æ3.1
À46.7Æ2.4
À38.7Æ1.9
À58.6Æ5.1
À0.025Æ0.01 0.00093
À9.26Æ1.5 0.24
À17.8 (67%)
À8.9Æ0.9
8.9Æ0.8
1b
2b
3b
À28.8
À31.5
À25.6
À37.1
À39.1
À32.2
À25.7 (89%) À12.4Æ1.1 À3.1Æ1.0
À28.6 (91%) À9.74Æ0.9 À2.9Æ1.1
À6.84Æ0.63 0.18
À23.4Æ3.2
À9.84Æ2.9
À0.317Æ0.1
À26.8Æ4.9
0.61
0.21
0.0082
0.46
À24.7 (96%)
À0.9Æ0.3 À0.9Æ0.4
13b
14b
15b
À30.1 (81%) À16.8Æ2.2 À7.0Æ1.1
À35.9 (92%) À3.52Æ1.2 À3.2Æ1.0
À30.2 (94%) À28.8Æ1.7 À2.0Æ0.7
[a] Determined by temperature effect experiments by using the equation DGoT =DHTor +DCP-
(TÀTr)ÀTDSoTrÀTDCPln(T/Tr). See the Supporting Information for definitions and further details. [b] Determined
by salt effect experiments by using the equation log(Kd)=log(Knel)ÀZYlog[KCl]. See the Supporting Information
for definitions and further details. The percentage of nonelectrostatic interactions (DGonel/DG8) is given in paren-
theses. [c] DGoel =DG8ÀDGonel =TDSeol.
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2075
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 4. Cellular inhibition of viral proliferation.
Ligand
IC50[a] [mm]
neomycin
1b
2b
n.a.
0.63
n.a.
3b
13b
14b
0.41
1.78
1.79
[a] IC50: 50% inhibitory concentration. The obtained values are given with
an uncertainty of Æ5%. n.a: no activity up to 50 mm.
3b) were those exhibiting the highest specificity (1.1<K’d/Kd <
1.4 and 1.2<K’’d/Kd <1.5). In a similar way, compounds 13b
and 14b, which are approximately threefold less specific than
1b and 3b, retain the activity, whereas ligands 2b and 15b,
which have the lowest degree of specificity, are completely in-
active.
Figure 2. CD spectra of TAR RNA in the absence (black dotted line) and in
the presence of 1 equivalent (gark gray dotted line), 5 equivalents (gray
dotted line), and 10 equivalents (light gray dotted line) of compound 13b.
cates that the overall structure of the targeted RNA is main-
tained, even at high ligand concentrations.
Finally, cytotoxicity assays performed by incubating increas-
ing concentrations of drugs with MAGIC-5B cells for 48 h re-
vealed no modification of cell viability relative to cells incubat-
ed in the presence of culture medium (Figure 3B). These re-
sults clearly indicate that this class of ligands is not toxic in the
given cells. Moreover, the fact that these ligands inhibited HIV-
1 replication in vitro clearly attested for their capacity to pene-
trate into the cell and to interfere with the retroviral replication
cycle. Hence, this class of ligands clearly shows an interesting
profile and could be used for further investigations in RNA tar-
geting.
Furthermore, the results obtained by monitoring the fluores-
cence variation as a function of the concentration of ligand (Kd
experiments performed in triplicate) are reproducible and per-
fectly fit with 1:1 stoichiometry (ligand/RNA), even in the pres-
ence of a high excess of ligand. If nonspecific interactions took
place, this would have been visible in these fluorescence stud-
ies, which would have shown nonconstant values for the stoi-
chiometry ratio, other than 1:1.
Antiviral activity of TAR RNA ligands in cells
The S–amino acid conjugates were evaluated for their capacity
to inhibit HIV-1 replication. To this end, we determined the
ability of the laboratory-adapted HIV-1 NL4.3 strain to replicate
in MAGIC-5B cells maintained in the presence of increasing
concentrations of the newly synthesized compounds. This indi-
cator cell line expresses the CD4 receptor, the CXCR4 corecep-
tor, and an integrated copy of the b-galactosidase gene under
control of the HIV-1 LTR promoter. The capacity of incoming
and newly synthesized Tat to transactivate the retroviral pro-
moter was therefore evaluated by quantification of the b-gal-
actosidase gene expression in the cells. The efficiency of the
TAR ligands was compared with that of azidothymidine (AZT),
a reference HIV-1 inhibitor, used at a final concentration of
20 mm (Table 4 and Figure 3A).
We found that compounds 1b (Lys) and 3b (His) inhibited
HIV-1 replication with interesting IC50 values of 0.6 and 0.4 mm,
respectively (Table 4). Compounds 13b (di-Lys) and 14b (di-
Arg) exhibited similar activities (1.7 mm), which were approxi-
mately fourfold less efficient then 3b. Most importantly, li-
gands 2b (Arg) and 15b (di-His), which exhibited good affinity
but low TAR specificity, were found to be almost ineffective
under our experimental conditions. It is noteworthy that neo-
mycin has no anti-HIV activity, even at high concentrations.
Careful analysis of these data clearly showed a strong corre-
lation between the specificity of a ligand for TAR RNA and its
anti-HIV activity. For example, the most active ligands (1b and
Conclusion
A new series of artificial nucleobase–amino acid conjugates
has been synthesized and the binding affinity of the com-
pounds for a labeled HIV-1 TAR RNA has been evaluated by
fluorescence spectroscopy. In addition to the obtained Kd
values, competitive assays with tRNA and dsDNA were per-
formed. We observed that: 1) S conjugates comprising one
basic amino acid displayed higher affinity than neomycin and
strong specificity for TAR RNA and 2) S–diamino acid conju-
gates exhibited enhanced ligand–RNA binding relative to that
of their monosubstituted analogues while retaining significant
degrees of specificity. These properties can be attributed to
the exceptional cooperative mode of binding.
Furthermore, we have unambiguously established that this
class of ligand is highly stable in biological media and active in
cell-based assays in the submicromolar range, without appar-
ent toxicity, and that the specificity of the S-based ligands for
the TAR RNA model is strongly correlated with the in vitro ac-
tivity.
Altogether, these results support the idea that amino acids
and Hoogsteen based binding can be combined to create
a new class of high-affinity, high-specificity RNA-binding li-
gands. This approach is highly promising and could be applied
for the rational targeting of different RNA structures (structure-
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2076
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 3. In vitro activity of RNA ligands. A) The capacity of various concentrations of TAR RNA ligands to inhibit replication of the HIV-1 NL4.3 strain in cul-
tures of MAGIC-5B cells was determined by quantification of the b-galactosidase reporter gene activity in the cell culture. B) The toxicity of the TAR ligands
was compared with that of AZT. MAGIC-5B cells were exposed to increasing concentrations of ligands for 48 h and the cell viability was measured as de-
scribed in the Experimental Section. The cell viability was similar to the control conditions for any ligand concentration tested. The controls consisted of
mock infected cells or HIV-1-infected cells maintained in the absence of drug (not treated, NT) or in medium supplemented with the nucleoside reverse tran-
scriptase inhibitor AZT and the results are presented in the left-hand panel.
ual 13C resonance of the solvent (CDCl3, d=77.3 ppm; CD3OD, d=
49.0 ppm; [D6]DMSO, d=39.5 ppm). Splitting patterns have been
designated as follows: s: singlet; d: doublet; dt: doublet of triplets;
t: triplet; q: quartet; m: multiplet; br: broad. Coupling constants (J
values) are listed in hertz (Hz). High-resolution mass spectra were
based ligand design), including viral and oncogenic RNAs in
tumor cells.
obtained with a LTQ Orbitrap hybrid mass spectrometer with an
electrospray ionization probe (Thermo Scientific, San Jose, CA) by
Experimental Section
Materials
direct infusion from a pump syringe to confirm the correct molar
mass and high purity of the compounds. The final products were
analyzed by HPLC on a Waters Alliance 2695 pump coupled with
a Waters 996 photodiode array detector and a Thermo Scientific
Betasil RP C18 column (250ꢁ4.6 mm, 5 mm). Solvent A (0.1% TFA
in water and solvent B (0.1% TFA in acetonitrile) were used for
HPLC studies. A gradient of A/B (100:0 to 40:60 for 30 min) was
Solvents and most of the starting materials were purchased from
Aldrich and Alfa Aesar. All reactions involving air- or moisture-sen-
sitive reagents or intermediates were performed under an argon
atmosphere. Flash column chromatography was carried out on
silica gel (Merck, SDS 60A, 40–63 mm). Analytical thin-layer chroma-
tography (TLC) was conducted on Merck precoated silica gel
60F254 plates and compounds were visualized with the ninhydrin
employed at a flow rate of 1 mLminÀ1
.
test and/or under ultraviolet light (254 nm). H and 13C NMR spec-
1
tra were recorded on a Bruker AC spectrometer (200 MHz or
500 MHz). Chemical shifts (d) are reported in parts per million
(ppm) referenced to the residual 1H resonance of the solvent
(CDCl3, d=7.26 ppm; CD3OD, d=3.31 ppm; D2O, d=4.79 ppm;
[D6]DMSO, d=2.50 ppm). 13C spectra were referenced to the resid-
Unless otherwise stated, all reagents and solvents were of analyti-
cal grade and were obtained from Sigma. HEPES and all inorganic
salts for buffers were purchased from Aldrich (molecular biology
grade). RNA, Tat peptide, and DNA oligonucleotides were pur-
chased from IBA GmbH and used without further purification.
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2077
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Buffers
Antiviral activity
All buffers were filtered through 0.22 mm Millipore filters (GP Ex-
pressPLUS membrane). A small aliquot (100 mL) was first filtered
and then discarded to avoid any contaminants that might be
leached from the filter. The solutions to be used in the fluores-
cence experiments were prepared by diluting the concentrated
stocks in Milli-Q water and filtering again as described above. All
standard fluorescence measurements were performed in buffer A
(20 mm HEPES, pH 7.4, containing 20 mm NaCl, 140 mm KCl, and
3 mm MgCl2, at 258C). For competitive experiments in the pres-
ence of tRNAs, a mixture of pre- and mature yeast tRNAs (contain-
ing over 30 different species from baker’s yeast (Saccharomyces
cerevisiae, Sigma, type X-SA)) was added to buffer A to obtain
a 100-fold nucleotide excess with regard to TAR RNA. Stock solu-
tions of tRNAs were prepared in water and quantified by using an
extinction coefficient of 9640 cmÀ1 mÀ1 per base. For competitive
experiments in the presence of a dsDNA, a 15-mer sequence (5’-
CGTTTTTATTTTTGC-3’) and its complement, annealed beforehand,
were added to buffer A to obtain a 100-fold nucleotide excess with
regard to TAR RNA (900 nm duplex; 5 nm RNA).
The CD4+ CXCR4+ MAGIC-5B indicator cell line, which stably ex-
presses the b-galactosidase reporter gene cloned downstream of
the HIV-1 LTR, was cultured in Dulbecco’s modified Eagle’s medium
(DMEM) supplemented with 10% fetal calf serum, 1% glutamax,
and 1% penicillin–streptomycin antibiotic mixture (Life Technolo-
gies, Inc.) at 378C in a 5% CO2 atmosphere. Each drug was pre-
pared in four serial tenfold dilutions with culture medium. MAGIC-
5 cells were seeded (12500 per well) and cultured in a 96-well
tissue culture plate for 24 h. The cells were preincubated with the
diluted compounds for 2 h at 378C and then challenged with the
HIV-1 NL4.3 strain at a multiplicity of infection (m.o.i.) of 1. After
48 h, the cells were washed and lysed, and infection of the cell cul-
ture was monitored by measurement of the b-galactosidase activi-
ty from the total cell lysate by using the Galacto-Star chemilumi-
nescent assay kit according to the manufacturer instructions (Ap-
plied Biosystem, USA). Luminescence was recorded by using
a Centro XS3 LB960 luminometer (Berthold, France). Values were
normalized with respect to the exact protein content of each cell
lysate, as determined with the BCA protein assay (Thermo Scientif-
ic). The susceptibility of HIV-1 to the various compounds was de-
termined by using triplicate samples of infected MAGIC-5B cells in
each assay. The IC50 value for each drug was estimated from plots
of luciferase per mg of protein reduction versus drug concentra-
tion.
Binding studies and Kd determination
Ligand solutions were prepared as serial dilutions in buffer A at
a concentration four times higher than the desired final concentra-
tion to allow for the subsequent dilution before the addition of
the RNA solution. An automated pipetting system (epMotion 5075,
Eppendorf) was used in order to perform these binding studies on
384-well plates (Greiner bio-one). Refolding of the RNA was per-
formed by using a thermocycler (ThermoStat Plus, Eppendorf) as
follows: the 5’-Alexa488 TAR RNA (0.2 nmol) was diluted in buffer A
(1 mL), denatured by heating to 908C for 2 min, cooled to 48C for
10 min, then incubated at 258C for 15 min. After refolding, the
RNA was diluted to a working concentration of 20 nm through ad-
dition of the appropriate amount of buffer A. After addition of
each ligand (30 mL) on the 384-well plates in 15 dilutions (from
1 mm to 61 nm as final concentrations) and in duplicate, the RNA
solution (30 mL) was added to each well containing ligand to give
a final concentration of 10 nm. The fluorescence was measured on
a GeniosPro apparatus (Tecan) with an excitation filter of (485Æ
10) nm and an emission filter of (535Æ15) nm. Each point was
measured five times with a 500 ms integration time and averaged.
Binding was allowed to proceed overnight at 58C to achieve equi-
librium. To study the temperature dependence, the plates were in-
cubated after overnight equilibrium at different temperatures rang-
ing from 58C to 358C. Neomycin was used as a control because its
binding to TAR has already been studied by using several meth-
ods.[12] Its Kd value of (12.0Æ3.6) mm is in good agreement with
previously reported values.
Cell viability assays
The viability of cells exposed to the drugs was assessed by incu-
bating MAGIC-5B cells (12500 per well in a 96-well tissue culture
plate) in the presence of increasing concentrations of the drugs.
After 48 h of culture, the cell viability was determined by using the
CellTiter 96 AQueous One Solution cell proliferation assay (Prome-
ga). Determinations were performed in quadruplicate.
Acknowledgements
This work was supported by CNRS, ANR, Universitꢂ de Nice
Sophia Antipolis, Conseil Rꢂgional PACA, MAE, ANRT, and Ori-
base Pharma (grant to J.-P.J.).
Keywords: amino acids
· antiviral agents · cooperative
binding · nucleobases · RNA recognition
[1] For recent reviews, see: a) L. Guan, M. D. Disney, ACS Chem. Biol. 2012,
7, 73–86; b) R. Moumne, M. Catala, V. Larue, L. Micouin, C. Tisne, Bio-
chimie 2012, 94, 1607–1619; c) Y. Tor, Pure Appl. Chem. 2009, 81, 263–
272; d) J. R. Thomas, P. J. Hergenrother, Chem. Rev. 2008, 108, 1171–
1224; e) N. Foloppe, N. Matassova, F. Aboul-Ela, Drug Discovery Today
2006, 11, 1019–1027.
Data analysis
[2] For recent examples, see: a) K. Limmer, D. Aschenbrenner, H. E. Gaub,
Nucleic Acids Res. 2013, 41, e69; b) T. Lombꢃs, R. Moumnꢂ, V. Larue, E.
Prost, M. Catala, M. T. Lecourt, F. Dardel, L. Micouin, T. Carine, Angew.
Chem. 2012, 124, 9668–9672; Angew. Chem. Int. Ed. 2012, 51, 9530–
9534; c) S. Umemoto, S. Im, J. Zhang, M. Hagihara, A. Murata, Y. Harada,
T. Fukuzumi, T. Wazaki, S. Sasaoka, K. Nakatani, Chem. Eur. J. 2012, 18,
9999–10008; d) R. Parkesh, J. L. Childs-Disney, M. Nakamori, A. Kumar,
E. Wang, T. Wang, J. Hoskins, T. Tran, D. Housman, C. A. Thornton, M. D.
Disney, J. Am. Chem. Soc. 2012, 134, 4731–4742; e) C. L. Beisel, Y. Y.
Chen, S. J. Culler, K. G. Hoff, C. D. Smolke, Nucleic Acids Res. 2011, 39,
2981–2994; f) P. Lopez-Senin, I. Gomez-Pinto, A. Grandas, V. Marchan,
Chem. Eur. J. 2011, 17, 1946–1953.
Binding data were analyzed by using Graphpad Prism 5 software.
Unless otherwise stated, binding profiles were well modeled by
using a simple model and assuming one-to-one stoichiometry. A
higher initial fluorescence value is observed in the presence of
tRNA, which is consistent with the modification of the polarity of
the solvent and a small amount of fluorescence from the tRNA
mixture.
[3] S. Feng, E. C. Holland, Nature 1988, 334, 165–167.
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2078
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[4] S. Bannwarth, A. Gatignol, Curr. HIV Res. 2005, 3, 61–71.
[5] L. M. Alvarez-Salas, Curr. Top. Med. Chem. 2008, 8, 1379–1404.
[6] a) R. S. Geary, Expert Opin. Drug Metab. Toxicol. 2009, 5, 381–391; b) R.
Juliano, J. Bauman, H. Kang, X. Ming, Mol. Pharm. 2009, 6, 686–695;
c) X. Zhao, F. Pan, C. M. Holt, A. L. Lewis, J. R. Lu, Expert Opin. Drug Deliv-
ery 2009, 6, 673–686.
2759; d) Y. Wang, D. A. C. Powers, S. O. Lack, D. Osborne, K. R. Fox, T.
Brown, Biochemistry 2005, 44, 5884–592.
[11] a) V. Bonnard, S. Azoulay, A. Di Giorgio, N. Patino, Chem. Commun. 2009,
2302–2304; b) N. W. Luedtke, Q. Liu, Y. Tor, Biochemistry 2003, 42,
11391–11403; c) J. R. Thomas, X. Liu, P. J. Hergenrother, J. Am. Chem.
Soc. 2005, 127, 12434–12435.
[12] a) T. Tran, M. D. Disney, Biochemistry 2010, 49, 1833–1842; b) Y. Xie, A. V.
Dix, Y. Tor, J. Am. Chem. Soc. 2009, 131, 17605–17614; c) S. Freisz, K.
Lang, R. Micura, P. Dumas, E. Ennifar, Angew. Chem. 2008, 120, 4178–
4181; Angew. Chem. Int. Ed. 2008, 47, 4110–4113; d) J. Kondo, B. Fran-
cois, A. Urzhumtsev, E. Westhof, Angew. Chem. 2006, 118, 3388–3392;
Angew. Chem. Int. Ed. 2006, 45, 3310–3314; e) F. Zhao, Q. Zhao, K. F.
Blount, Q. Han, Y. Tor, T. Hermann, Angew. Chem. 2005, 117, 5463–5468;
Angew. Chem. Int. Ed. 2005, 44, 5329–5334.
[13] S. Li, M. Hong, J. Am. Chem. Soc. 2011, 133, 1534–1544.
[14] a) D. I. Bryson, W. Zhang, P. M. McLendon, T. M. Reineke, W. L. Santos,
ACS Chem. Biol. 2012, 7, 210–217; b) H. Suryawanshi, H. Sabharwal, S.
Maiti, J. Phys. Chem. B 2010, 114, 11155–11163.
[7] a) A. Davidson, K. Patora-Komisarska, J. A. Robinson, G. Varani, Nucleic
Acids Res. 2011, 39, 248–256; b) Z. Athanassiou, R. L. Dias, K. Moehle, N.
Dobson, G. Varani, J. A. Robinson, J. Am. Chem. Soc. 2004, 126, 6906–
6913; c) S. Chirayil, R. Chirayil, K. J. Luebke, Nucleic Acids Res. 2009, 37,
5486–5497.
[8] a) S. Kumar, P. Kellish, W. E. Robinson, Jr., D. Wang, D. H. Appella, D. P.
Arya, Biochemistry 2012, 51, 2331–2347; b) S. Magnet, J. S. Blanchard,
Chem. Rev. 2005, 105, 477–498; c) M. J. Belousoff, B. Graham, L. Spiccia,
Y. Tor, Org. Biomol. Chem. 2009, 7, 30–33; d) M. D. Disney, L. P. Labuda,
D. J. Paul, S. G. Poplawski, A. Pushechnikov, T. Tran, S. P. Velagapudi, M.
Wu, J. L. Childs-Disney, J. Am. Chem. Soc. 2008, 130, 11185–11194.
[9] M. Duca, V. Malnuit, F. Barbault, R. Benhida, Chem. Commun. 2010, 46,
6162–6164.
[15] A. L. Smith, J. Kassman, K. J. Srour, A. M. Soto, Biochemistry 2011, 50,
9434–944.
[10] a) V. Malnuit, M. Duca, R. Benhida, Org. Biomol. Chem. 2010, 8, 326–
336; b) D. Guianvarc’h, R. Benhida, J-L. Fourrey, R. Maurisse, R. J.-S. Sun,
Chem. Commun. 2001, 1814–1815; c) D. Guianvarch, J.-L. Fourrey, R.
Maurisse, J.-S. Sun, R. Benhida, Bioorg. Med. Chem. 2003, 11, 2751–
Received: September 17, 2013
Published online on January 15, 2014
Chem. Eur. J. 2014, 20, 2071 – 2079
www.chemeurj.org
2079
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim