Design, synthesis and ribosome binding of chloramphenicol nucleotide and intercalator conjugates

Article abstract of DOI:10.1016/j.bmcl.2005.02.044

Molecular modelling based on X-ray structures of the antibiotic drug chloramphenicol bound in a bacterial ribosome has been used for design of chloramphenicol derivatives. Conjugates of the chloramphenicol amine through appropriate linkers to either a pyrene moiety or to a mono- or dinucleotide moiety were designed to improve binding to ribosomes by providing specific interactions in the peptidyl transferase site or to the P-loop in the ribosome. Specific binding of the conjugates were investigated by footprinting analysis using chemical modifications of accessible nucleotides in ribosomal RNA. The pyrene chloramphenicol conjugate shows enhanced binding to the chloramphenicol binding site compared to the native chloramphenicol, whereas the four nucleotide conjugates could not be shown to bind to the chloramphenicol binding site or to the P-loop.

Full text of DOI:10.1016/j.bmcl.2005.02.044

Bioorganic & Medicinal Chemistry Letters 15 (2005) 2079–2083
Design, synthesis and ribosome binding of
chloramphenicol nucleotide and intercalator conjugates
Dorte Johansson,a Carsten H. Jessen,a Jacob Pøhlsgaard,b Kenneth B. Jensen,a
Birte Vester,b Erik B. Pedersena and Poul Nielsena,*
aNucleic Acid Center , Department of Chemistry, University of Southern Denmark, 5230 Odense M, Denmark
bNucleic Acid Center , Department of Biochemistry and Molecular Biology, University of Southern Denmark,
5230 Odense M, Denmark
Received 13 October 2004; revised 10 February 2005; accepted 16 February 2005
Available online 18 March 2005
Abstract—Molecular modelling based on X-ray structures of the antibiotic drug chloramphenicol bound in a bacterial ribosome has
been used for design of chloramphenicol derivatives. Conjugates of the chloramphenicol amine through appropriate linkers to either
a pyrene moiety or to a mono- or dinucleotide moiety were designed to improve binding to ribosomes by providing specific inter-
actions in the peptidyl transferase site or to the P-loop in the ribosome. Specific binding of the conjugates were investigated by foot-
printing analysis using chemical modifications of accessible nucleotides in ribosomal RNA. The pyrene chloramphenicol conjugate
shows enhanced binding to the chloramphenicol binding site compared to the native chloramphenicol, whereas the four nucleotide
conjugates could not be shown to bind to the chloramphenicol binding site or to the P-loop.
Ó 2005 Elsevier Ltd. All rights reserved.
Cl
The presence of bacterial resistance to current antibiot-
ics is a serious and widespread problem that motivates a
search for new effective drugs. Most new antibiotics,
however, are directed to well-established targets, for
example, the bacterial ribosome. Studies of such ribo-
somal antibiotic sites are important tools for under-
standing the inhibitory mechanisms of drugs and for
drug design. Recent reports of X-ray structures of many
different antibiotics bound to the ribosomal subunits1–3
support modelling and design of new drugs with pre-
sumed enhanced binding and increased specificity.
Cl
O
NO2
NH
HO
OH
1
isms revealed two different binding sites of the drug.1,2
Thus, there might be two chloramphenicol binding sites
in the ribosome with the strongest binding site presumed
to be the clinically relevant site,7 and the site considered
in this study.
The antibiotic chloramphenicol (1) inhibits protein syn-
thesis by binding to the peptidyl transferase region of
the ribosome.4 Its clinical use is limited because of toxic
site effects. Chloramphenicol has been derivatised in
many ways in order to define its essential functional
groups and to improve its drug properties.5,6 Elucida-
tion of the X-ray structures of chloramphenicol bound
to the 50S ribosomal subunit from two different organ-
The high affinity chloramphenicol binding site is com-
posed entirely of ribosomal RNA and overlaps the bind-
ing site of the antibiotic puromycin.1,7 A transition state
inhibitor consisting of a CCdA-p-puromycin conjugate
based on a phosphoramidate linkage mimicking the pep-
tidyltransferase reaction8 was shown by X-ray crystal-
lography to bind to G2252 and G2251 in the
ribosomal P-loop by Watson–Crick base pairing to the
two adjacent cytidines and to bind to the A-site with
the pyromycin part.9 This inhibitor displayed a strong
activity as a peptidyltransferase inhibitor.8 This result
Keywords: Nucleic acids; Conjugates; Footprinting; Antibiotics.
Nucleic Acid Center is funded by the Danish National Research
Foundation for studies on nucleic acid chemical biology.
*
Corresponding author. Tel.: +45 6550 2565; fax: +45 6615 8780;
e-mail: pon@chem.sdu.dk
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.02.044

2080
D. Johansson et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2079–2083
motivated us to explore the potential of similar but
simpler and easily available conjugates of the chloram-
phenicol core structure to nucleotides with potential
binding to the P-loop. Furthermore, in an alternative
approach we envisioned that conjugation to a stacking
moiety such as a pyrene might be another way of
improving the target affinity of chloramphenicol. Re-
lated studies with conjugates, but without specific
RNA targets have been described.6,10 In both ap-
proaches, we applied the X-ray structural information
of the ribosome in modelling experiment using the Mac-
roModel system.11
From the known SAR of chloramphenicol5 and the
knowledge of the ribosomal binding site,1 we decided
to replace the dichloroacetyl group of chloramphenicol
by alternative acyl groups as the anchor for conjugation.
We designed and prepared four chloramphenicol nucleo-
tide conjugates for improved and combined binding to
the peptidyl transferase centre and the P-loop region
of the ribosome as well as one chloramphenicol pyrene
conjugate (Fig. 1).
Figure 2. A combination of the two chloramphenicol conjugate types
are shown with the chloramphenicol part in green, the pyrene
(compound 2) in red and two cytidine residues (compound 6) in blue.
Relevant 23S RNA nucleotides are shown as labelled (E. coli
numbering). Predicted hydrogen bonds to P-loop nucleotides are
indicated.
Design of the pyrene conjugate 2 was done in Macro-
Model from the D.radidurans/chloramphenicol struc-
ture. However, no specific binding site is targeted, and
no design requirements were placed on the linker be-
tween the components. A glycine linker, combined with
the available (1-pyrenyl)acetic acid, was chosen as a
synthetically convenient linker. On the other hand, the
P-loop targeting conjugates needed a more precisely de-
signed linker. The P-loop binding CC nucleotides from
the H.marismirtui/CCdA-p-puromycin co-crystal9 were
moved into the D.radidurans/chloramphenicol structure1
in MolMol. In dynamic simulations (performed in Mac-
roModel using the amber* force field in implicit water)
the two core components, that is, the chloramphenicol
part and the dinucleotide, were fixed and joined by dif-
ferent short linkers. Four structures were subsequently
chosen for synthesis providing two different syntheti-
cally convenient linkers and mono- as well as dinucleo-
tide conjugates 3–6 (Fig. 1). The design of the two types
of chloramphenicol conjugates and the envisioned ribo-
somal binding sites is visualised in Figure 2.
C
HO
O
C
HO
O
O
_
O
P
O
C
O
O
P
O
O
_
HN
O
O
O
O
O
P
NO2
_
O
NH
O
n
HO
O
O
NO2
NH
OH
HO
n
NH
O
2
NO2
OH
HO
3 (n=1), 4 (n=2)
OH
5 (n=1), 6 (n=2)
Figure 1. The designed pyrene and nucleotide chloramphenicol conjugates. C = cytosine-1-yl.

D. Johansson et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2079–2083
2081
The commercially available precursor for chloramphen-
icol derivatives (1R,2R)-2-amino-1-(4-nitrophenyl)-1,3-
propandiol (7) was used as a starting material for the
preparation of the conjugates (Scheme 1). In order to
prepare the pyrene conjugate with a glycine linkage, 2,
two subsequent coupling steps were performed. The
appropriate Fmoc-protected glycine derivative was cou-
pled with 7 to give 8 in a high yield. The Fmoc group
was used after a similar Boc-strategy has failed, and
the Fmoc-group was easily removed by a mild basic
treatment. (1-Pyrenyl)acetic acid was converted to its
N-hydroxyl succinimide ester in situ and mixed with
the deprotected 8 to give the target conjugate 2 in a rea-
sonable yield.
was less reactive and in a basic solution afforded 10 in a
lower but acceptable yield. The primary alcohols of 10
and 11 were used for the standard nucleotide coupling
reactions13 with a commercial cytidine phosphoramidite
that contains a DMT-protection for the 50-hydroxyl
group and a benzoyl group to protect the exocyclic
amine of the nucleobase. Thus, the multiple protected
30-nucleotide chloramphenicol conjugates 12 and 13
were obtained. Subsequent detritylation afforded com-
pound 14 and 15, respectively. A complete deprotection
using TBAF for the removal of the silyl-groups and
aqueous ammonia for the removal of the cyanoethyl
group and the N-benzoyl group afforded after an
HPLC-purification the target conjugates 3 and 4. On
the other hand, a second coupling reaction using 14
and 15 and the cytosine phosphoramidite followed by
the standard deprotection steps and HPLC-purification
afforded the target dinucleotide conjugates 5 and 6.
The constitution of the conjugates 3–6 was verified by
In order to form the nucleotide conjugates of the chlor-
amphenicol amine, the alcohol functionalities of 7 were
selectively protected as silyl ethers to give 9 in a high
yield (Scheme 1). For the preparation of the appropriate
linkers, 9 was reacted with either b-propiolactone or c-
butyrolactone to give 10 and 11,12 respectively. Thus,
c-butyrolactone was simply used as the solvent affording
11 in a high yield, whereas b-propiolactone surprisingly
1
31
MS and by H, 13C and P NMR spectroscopy.14
The interaction of the chloramphenicol derivatives
with the peptidyl transferase centre of Escherichia coli
OH
n
O
NO2
NO2
NH
NH2
TBDMSO
RO
d or e
OTBDMS
OR
10 n = 1
11 n = 2
7 R=H
c
9 R=TBDMS
f
a
Fmoc
CBz
RO
O
HN
h
j
3, 4
5, 6
O
O
NO2
NH
CEO
P
O
HO
O
OH
n
8
O
NO2
NH
TBDMSO
b
OTBDMS
2
12 n = 1 / 13 n= 2, R = DMT
14 n = 1 / 15 n= 2, R = H
g
Scheme 1. Reagents and conditions: (a) Fmoc-Gly-OC6F5, CH3CN, 95%; (b) i, Et3N, DMF; ii, (1-pyrenyl)acetic acid, DCC, NHS, DMF, 39%; (c)
TBDMS-OTf, 2,6-lutidine, CH2Cl2, 94%; (d) b-propiolactone, 2,6-lutidine, CH2Cl2, 10 24%; (e) c-butyrolactone, 11 66%; (f) dC-CE/DMT
phosphoramidite, 1H-tetrazole, CH3CN then I2, aq pyridine, 12 48%, 13 34%; (g) TFA, CHCl3, Et3SiH, CH2Cl2, 14 45%, 15 54%; (h) i, TBAF, THF;
ii, 32% aq NH3, HPLC, 3 38%, 4 58%; (j) i, dC-CE/DMT phosphoramidite, 1H-tetrazole, CH3CN then I2, aq pyridine, 30%, 28%; ii, TFA, CHCl3,
Et3SiH, CH2Cl2, 52%, 52%; iii, TBAF, THF; iv, 32% aq NH3, HPLC, 5 47%, 6 70%. CBz = N-4-benzoylcytosine-1-yl, TBDMS = tert-
butydimethylsilyl, CE = 2-cyanoethyl, DMT = 4,40-dimethoxytrityl.

2082
D. Johansson et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2079–2083
ribosomes were investigated by footprinting. The drug
binding sites were studied using chemical probing, by
which accessible nucleotides are modified.15 After bind-
ing of chloramphenicol and the chloramphenicol conju-
gates to 70S ribosomes, the antibiotic–ribosome
complexes and free ribosomes were treated with CMCT
[1-cyclohexyl-3-(2-N-(N-methyl)morpholinoethyl)carbo-
diimide] to probe the N3 of uridine and N1 of guanosine
residues. Some of the complexes were also treated with
kethoxal to probe the N1 position of guanosine. Primer
extension with reverse transcriptase was then used to
identify the alterations in chemical reactivity of specific
nucleotides induced by the binding of the drug to the
ribosome. Growth of E. coli cells, isolation of ribosomes
as well as modifications and primer extensions were per-
formed essentially as described before.16,17 Figure 3
shows an autoradiogram of primer extension of the re-
gion where CMCT modification of nucleotide U2506
is known to be protected by chloramphenicol.18 The
protection effects seen from 1 (chloramphenicol) and 2
were quantified by phosphor image scanning. The effect
shows that 2 binds to the peptidyl transferase part of the
ribosome and probably with the choramphenicol part of
the molecule occupying the same site as chlorampheni-
col. Furthermore, it shows a binding relative to chlor-
amphenicol clearly enhanced by at least an order of
magnitude. Inspection of the X-ray structure of the ribo-
somal drug binding site suggests that the pyrene part of
2 could intercalate or interact by stacking with nucleo-
tide U2506, G2583 and U2584 (Fig. 2). As U2584-5
are also accessible for CMCT modification this region
was also explored by footprinting but no protection ef-
fect were observed here. Thus, any interaction between
the pyrene of 2 and nucleobases has not been clearly
demonstrated.
U2506. Apparently, the chloramphenicol part of these
molecules does not bind similarly to chloramphenicol.
However, it was still a possibility that the compounds
interact with the P-loop in ribosomal 23S RNA on the
ribosome. Thus, additional experiments with kethoxal
modifications were performed searching for base pairing
from compound 3–6 to the P-loop but, no effects on the
modification pattern were seen in this region. To explore
the possibility of other footprints, the A-loop in ribo-
somal 23S RNA were analysed by primer extension after
CMCT modification of ribosome complexes. Re-exam-
ining the model of the compounds docked into the 50S
ribosomal X-ray structure to propose explanations for
the missing footprints suggests that the linker regions
between the choramphenicol part and the nucleotide
part may be too short or too rigid to allow combined
binding at two sites. As the binding of choramphenicol
by itself is not very strong, the burden of a Ôno affinity
affording nucleotide partÕ may be too heavy.
In conclusion, the nucleotide conjugate targets 3–6 were
successfully synthesised by the application of solution
phase phosphoramidite chemistry. Footprinting were
performed to determine the binding of the conjugates
in the peptidyl transferase centre of the ribosome as
anticipated by modelling experiments. The chloram-
phenicol pyrene conjugate 2 shows protection of
U2506 indicating binding to the chloramphenicol site
and the concentration dependent protection shows an
significantly enhanced binding relative to chloramphen-
icol. In contrary, no binding was found for the chloram-
phenicol nucleotide conjugates 3–6. This might be due to
linker restrictions, however, and further adjustment of
the linker structure might be sufficient for obtaining
more efficient compounds. Thus, the binding of the pyr-
ene derivative demonstrates the possibilities for large
choramphenicol conjugates to bind to the drug site. Fur-
ther effort in the design and synthesis of chlorampheni-
col nucleic acid conjugates is in progress. Clearly, the
potential of using intercalating moieties in conjugates
with antibiotics is a promising approach.
A similar analysis of binding of the four nucleotide con-
jugates 3–6 to ribosomes showed no protection effect at
Acknowledgements
The Danish National Research Foundation is thanked
for financial support. Lykke H. Hansen, Christina B.
Sørensen and Suzy W. Lena are thanked for technical
assistance.
References and notes
1. Schlunzen, F.; Zarrivah, R.; Harms, J.; Bashan, A.; Tocil,
¨
A.; Albrecht, R.; Yonath, A.; Franceschi, F. Nature 2001,
413, 814.
2. Hansen, J. L.; Moore, P. B.; Steitz, T. A. J. Mol. Biol.
2003, 330, 1061.
3. For examples with other antibiotics, see the following and
Figure 3. Gel analysis of protection of CMCT modification at
nucleotide 2506. Lane U, C, G, A indicate the 23S RNA sequence in
the analysed region. Co denotes a ribosomal RNA control lane. Mod.
denotes RNA from CMCT modified ribosomes. The triangles are from
ribosomes complexed with compounds at 0.01, 0.05, 0.2 and 0.5 mM,
respectively, and treated with CMCT. The protections are normalised
to the indicated reference band.
references herein: (a) Harms, J.; Schlunzen, F.; Fucine, P.;
¨
Bartels, H.; Yonath, A. BMC Biol. 2004, 2, 4; (b)
Brodersen, D. E.; Clemons, W. M., Jr.; Carter, A. P.;
Morgan-Warren, R. J.; Wimberly, B. T.; Ramakrishnan,
V. Cell 2000, 103, 1143; (c) Carter, A. P.; Clemons, W. M.;

D. Johansson et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2079–2083
2083
Brodersen, D. E.; Morgan-Warren, R. J.; Wimberly, B. T.;
Ramakrishnan, V. Nature 2000, 407, 340; (d) Hansen, J.
L.; Ippolito, J. A.; Ban, N.; Nissen, P.; Moore, P. B.;
Steitz, T. A. Mol. Cell. 2002, 10, 117.
Hendrickson, T.; Still, W. C. J. Comput. Chem. 1990, 11,
440.
12. Compound 11 was crystallised and X-ray data are
available: Johansson, D.; Larsen, F. B.; Bond, A. D.;
Nielsen, P. Acta Crystallogr. 2004, E60, o907.
4. Vazquez, D. Inhibitors of Protein Synthesis; Springer:
Berlin, 1979.
13. Caruthers, M. H. Acc. Chem. Res. 1991, 55, 5831.
14. Selected data for the target compounds 2–6; 31P NMR dP
(D2O, 121.5 MHz, 85% H3PO4 as external standard): 3
0.29; 4 0.63; 5 0.15, 0.24; 6 0.13, 0.55. HR MALDI-MS m/
z [MNa]+ (found/calcd): 2 (534.1613, 534.1636). ESI-MS
m/z: 3 596.3 [MHNa]+, 618.4 [M2Na]+; 4 588.4 [M2H]+,
610.5 [MHNa]+; 5 430.1 [M]2ꢀ, 883.2 [MNa]ꢀ; 6 450.4
5. As for example by: Bhuta, P.; Chung, L. C.; Hwang, J.-S.;
Zemlicka, J. J. Med. Chem. 1980, 23, 1299, and reviewed
in Gale, E. F.; Cundliffe, E.; Reynolds, P. E.; Richmond,
M. H.; Waring, M. J. In The molecular basis of antibiotic
action; Wiley: London, 1981; pp 460–462; For a later
example, see: Drainas, D.; Mamos, P.; Coutsogeorgopa-
ulos, C. J. Med. Chem. 1993, 36, 3542.
[M3HNa]2+, 461.4 [M2H2Na]2+, 472.4 [MH3Na]2+
.
6. Aminoglycoside chloramphenicol conjugates have been
examined: Lee, J.; Kwon, M.; Lee, K. H.; Jeong, S.; Hyun,
S.; Shin, K. J.; Yu, J. J. Am. Chem. Soc. 2004, 126, 1956.
7. Long, K. S.; Porse, B. T. Nucleic Acid Res. 2003, 31, 7208.
8. Welch, M.; Chastang, J.; Yarus, M. Biochemistry 1995, 34,
385.
9. Nissen, P.; Hansen, J.; Ban, N.; Moore, P. B.; Steitz, T. A.
Science 2000, 289, 920.
10. Aminoglycoside dinucleotide conjugates have been exam-
ined: Charles, I.; Xue, L.; Arya, D. P. Bioorg. Med. Chem.
Lett. 2002, 12, 1259.
15. Ehresmann, C.; Baudin, F.; Mougel, M.; Romby, P.; Ebel,
J.-P.; Ehresmann, B. Nucleic Acids Res. 1987, 15, 9109.
16. Poulsen, S. M.; Karlsson, M.; Johansson, L. B.; Vester, B.
Mol. Microbiol. 2001, 41, 1091.
17. Kethoxal modification: 5 pmol of the antibiotic–50S
ribosomal complexes were incubated 15 min at 37 °C in
50 lL modification buffer (50 mM HEPES–KOH
(pH 7.8), 10 mM MgCl2, 270 mM KCl) and then modified
with 50 lL kethoxal (40 mg/mL) for 20 min at 37 °C. The
reactions were stopped by adding 25 lL stop buffer (0.3 M
sodium acetate, 0.25 M H3BO3 270 mM KCl) and then
precipitated with ethanol.
11. Mohamadi, F.; Richards, N. G. J.; Guida, W. C.;
Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.;
18. Moazed, D.; Noller, H. F. Biochemie 1987, 69, 879.