Structure-based design of agents targeting the bacterial ribosome

Article abstract of DOI:10.1016/S0960-894X(03)00495-5

Rational structure-based drug design has been applied to the antibiotic thiostrepton, in an attempt to overcome some of its' limitations. The identification of a proposed binding fragment allowed construction of a number of key fragments, which were derivatised to generate a library of potential antibiotics. These were then evaluated to determine their ability to bind to the L11 binding domain of the prokaryotic ribosome and inhibit bacterial protein translation.

Full text of DOI:10.1016/S0960-894X(03)00495-5

Bioorganic & Medicinal Chemistry Letters 13 (2003) 2455–2458
Structure-Based Design of Agents Targeting the
Bacterial Ribosome
Justin Bower,a Martin Drysdale,a Richard Hebdon,b Allan Jordan,a,* Georg Lentzen,c
Natalia Matassova,d Alastair Murchie,d Jenifer Powlesc and Stephen Roughleya
aDepartment of Medicinal Chemistry, RiboTargets Ltd., Granta Park, Abington, Cambridge CB16GB, UK
bDepartment of Bacteriology, RiboTagets Ltd., Granta Park, Abington, Cambridge CB16GB, UK
cDepartment of Screening, RiboTargets Ltd., Granta Park, Abington, Cambridge CB16GB, UK
dDepartment of Assay Development, RiboTargets Ltd., Granta Park, Abington, Cambridge CB16GB, UK
Received 4 April 2003; revised 14 May 2003; accepted 15 May 2003
Abstract—Rational structure-based drug design has been applied to the antibiotic thiostrepton, in an attempt to overcome some of
its’ limitations. The identification ofa proposed binding fragment allowed construction ofa number ofkey fragments, which were
derivatised to generate a library ofpotential antibiotics. These were then evaluated to determine their ability to bind to the L11
binding domain ofthe prokaryotic ribosome and inhibit bacterial protein translation.
# 2003 Elsevier Ltd. All rights reserved.
The thiopeptide antibiotic thiostrepton 1 is a natural
product employed in topical veterinary medicine,1 but its
low solubility and poor bioavailability have hampered its
application in human infections. Active against Gram-
positive bacteria, thiostrepton stabilizes the binding of
the ribosomal protein L11 to a region ofthe 23S riboso-
mal RNA known as the L11 binding domain (L11BD).2
The dynamics ofthis L11–L11BD complex are critical
for stimulating the GTPase action of the elongation fac-
tors EF-G and EF-Tu, which in turn provide the energy
required to drive protein translation. Though efforts to
find novel therapeutics have recently been directed
toward the L11–L11BD interaction,3 rational exploita-
tion ofthis vital region ofRNA has been hampered by a
lack ofstructural knowledge. Recent work in our
laboratories has for the first time elucidated the structure
ofthe thiopeptide antibiotics (in particular thiostrepton)
bound to the L11BD and highlighted intimate details of
those residues in close contact with this region ofthe
ribosome (Fig. 1).4 These studies indicated a low mole-
cular weight fragment, based around the conserved
dehydrodemethylvaline-containing macrocycle, which
appears to be involved in the binding ofthe antibiotic to
the L11BD and, further, that focused libraries based
around this fragment may emulate the antibacterial
activity ofthiostrepton, or lead to simpler derivatives
which may overcome some ofits limitations.
Chemistry
The proposed binding fragment could be readily dis-
connected to three key fragments, namely a phenyl
thiazole acid, a dehydrodemethylvaline unit or suitable
mimic and a thiazole amino acid. From this premise,
the synthetic approach detailed in Scheme 1 was derived.
2-Aminobutyric acid was N-Boc protected, then con-
verted to the primary amide with ethyl chloroformate
and ammonia. The thioamide, obtained by treatment
with Lawesson’s reagent, was cyclised to the required
thiazole with ethyl bromopyruvate to give the hydroxy
thiazoline and dehydrated to the thiazole with tri-
fluoroacetic anhydride and 2,6-lutidine as per the lit-
erature method, avoiding racemisation ofthe
stereocentre.5 Deprotection ofthe amine, ofllowed by
facile coupling with the protected l-threonine 46 and
isobutyl chloroformate, gave the dipeptide mimetic.
Finally, isobutyl chloroformate-mediated amide forma-
tion with 2-phenyl-thiazole-4-carboxylic acid,7 followed
by hydrolysis ofthe terminal ethyl ester gave the desired
scaffold 2. A parallel synthesis, starting from l-valine
gave scaffold 3 in a similar manner.
*Corresponding author. Tel.: +(1)-223-895-315; fax: +(1)-223-895-
556; e-mail: allan.jordan@ribotargets.com
0960-894X/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00495-5

2456
J. Bower et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2455–2458
Figure 1. Thiostrepton, highlighting observed nOes to the ribosomal RNA (marked *), and the resultant proposed binding fragment.
Scheme 1. R=H or Me. Reagents: (a) Boc2O, NaOH; (b) ethyl chloroformate, THF then NH4OH; (c) Lawesson’s reagent, toluene; (d) ethyl
bromopyruvate, KHCO3, 1,2-dimethoxyethane; (e) TFAA then 2,6-lutidine, 1,2-dimethoxyethane; (f) EtOH, AcCl; (g) 4, iso-butyl chloroformate,
N-methyl morpholine, THF; (h) TFA, DCM; (i) 2-phenyl-thiazole-4-carboxylic acid, iso-butyl chloroformate, N-methyl morpholine, THF; (j) LiOH,
H2O, THF.
Scheme 2. (a) Ethyl chloroformate, THF then NH4OH; (b) Lawesson’s
reagent, toluene; (c) ethyl bromopyruvate, KHCO3, 1,2-dimethoxy-
Scheme 3. (a) TFA, DCM; (b) 2-phenyl-thiazole-4-carboxylic acid,
iso-butyl chloroformate, N-methyl morpholine, THF; (c) LiOH, H2O,
ethane; (d) 4, iso-butyl chloroformate, N-methyl morpholine, THF.
THF; (d) mesyl chloride, triethylamine, DCM; (e) TFA, DCM then aq
HCl; (f) DBU, CHCl3; (g) 2-phenyl-thiazole-4-carboxylic acid, iso-
butyl chloroformate, N-methyl morpholine, THF; (h) LiOH, H2O,
The protected threonine derivative 4 was also employed
THF.
in the preparation ofkey intermediate 5 (Scheme 2),
which could be diversified to form scaffolds 6 and 7
(Scheme 3).
more troublesome. Literature precedent suggested that
the treatment ofa protected threonine derivative with
thionyl chloride or phosgene would generate a forma-
Thus, 4 was converted to the thioamide, via the primary
amide and cyclised with ethyl bromopyruvate and
TFAA/2,6-lutidine. The resultant thiazole acid was
N-deprotected and coupled with a second equivalent of
4 to yield 5. Deprotection ofthe N-Boc acetal unit, iso-
butyl chloroformate mediated coupling with the phenyl
thiazole acid and hydrolysis ofthe ethyl ester gave facile
access to scaffold 6 (Scheme 3).
8
tion ofa cyclic sulfamidite or cyclic anhydride,9 which
could be treated with base to yield the desired dehy-
drodemethylvaline unit and sulfur dioxide or carbon
dioxide respectively. Alternatively, it was reported that
the cyclic anhydride derived from Cbz-2-amino-2-bute-
noic acid (itself formed from 2-ketobutyric acid and
benzyl carbamate10) could be coupled directly with 4
and further elaborated to give a late-stage intermediate
in the synthesis of 7.11 In our hands, however, these
approaches failed to give satisfactory results. We also
The preparation ofthe dehydrodemethylvaline-contain-
ing unit 7, displayed in thiostrepton itself, proved to be

J. Bower et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2455–2458
2457
hoped that the aforementioned protected 2-amino-2-
butyric acid could be coupled directly to 4 and further
derivatised to 7, but again this proved unsuccessful.
which we had previously shown to display nOes to the
L11BD RNA (Fig. 1). These couplings were conducted in
parallel in a Bohdan Mini-blockTM and employed solid-
supported DCC and catalytic HOBt, followed by treat-
ment with solid-supported isatoic anhydride to capture
the excess amine used to drive the reaction to completion.
Careful optimisation of this chemistry allowed the
required products to be isolated simply by filtration and
concentration in vacuo, as single entities by LC-MS.
The required unit was, however, successfully obtained
by mesylation ofthe rfee hydroxyl of 5, which we pro-
posed could be eliminated with DBU to form the dehy-
drodemethylvaline moiety. We had assumed this would
be a facile process and was immediately attempted on
the acetal-protected material. These attempts initially
proved fruitless but we reasoned that the cyclic acetal
rigidifies the scaffold and restricts the conformation
such that E2 elimination cannot occur. Indeed, we
found that deprotection prior to treatment with DBU
allowed the desired transformation to occur smoothly.
Results and Discussion
Site-specific methylation ofthe L11BD RNA has been
shown to be a major mechanism ofthiostrepton resis-
tance,12 occurring exclusively on O-20 ofA1067. This
methylation prevents binding ofthe antibiotic, whilst
still allowing formation of the L11–L11BD complex and
resultant activation ofthe elongation factors EF-Tu and
EF-G, in turn allowing normal ribosomal function.
Methylation can be readily monitored using radio-
labeled S-adenosyl methionine, both on the whole
ribosome, and also on an artificial 58-nucleotide con-
struct ofthe L11BD region. This allows the rapid
detection ofthose compounds that interact with this
region. Those which interact with the L11BDR 58-mer
construct in a manner similar to thiostrepton will pre-
vent incorporation ofthe radiolabel and this can be
detected by scintillation counting.13 The assay can be
repeated on the entire ribosomal RNA, to ensure that
this binding is selective for L11BD and confirm that the
compounds do not bind in a non-selective manner.
Those compounds found to interact selectively with the
L11BD site can then be assessed for both their ability to
inhibit protein translation and their effect on bacterial
cell growth and proliferation.
We had assumed to this point that the dehy-
drodemethylvaline moiety (or a suitable mimic thereof)
would be crucially implicated in the interaction ofour
compounds with the L11BD. To test this hypothesis, we
also prepared the truncated scaffold 8, devoid ofthis
central unit, as shown in Scheme 4. The common inter-
mediate 4 was converted to the corresponding protected
thiazole, deprotected and coupled to the phenyl thiazole
acid and the ethyl ester hydrolysed to generate 8.
Once prepared, the scaffolds 2, 3, 6, 7 and 8 were deri-
vatised with a diverse set ofamino- and hydroxyamino-
units (Fig. 2). We reasoned these units may interact in a
similar manner to the ethanediol unit in thiostrepton,
Scheme 4. (a) Ethyl chloroformate, THF then NH4OH; (b) Law-
esson’s reagent, toluene; (c) ethyl bromopyruvate, KHCO3, 1,2-dime-
thoxyethane; (d) TFA, DCM; (e) 2-phenyl-thiazole-4-carboxylic acid,
iso-butyl chloroformate, N-methyl morpholine, THF; (f) LiOH, H2O,
THF.
From the synthesized library of93 compounds, five
were identified that prevented the incorporation ofthis
radiolabel into the L11BD 58-mer construct (Table 1)
Compound 8–e appeared initially to be the strongest
inhibitor ofL11BD methylation. However, the afilure
to inhibit radiolabel incorporation in the presence ofthe
whole ribosomal RNA indicates that this example ofthe
truncated series appears to interact with RNA in a non-
selective manner. For those representatives ofthe more
complex series, inhibition ofmethylation was reduced,
but satisfyingly similar across both the 58-nucleotide
Table 1. Inhibition ofmethylation ofA1067, on both the L11BD
58-mer construct and rRNA
Compd
% Inhibition ofL11BD
58-mer methylation
% Inhibition of
rRNA methylation
Scaffold
Amine
8
3
6
2
3
e
59
37
35
28
24
—
37
27
13
20
g
(OH)
n
n
All inhibition studies conducted at a compound concentration of
50mM.
Figure 2. Amines incorporated into scaffolds 2, 3, 6, 7 and 8.

2458
J. Bower et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2455–2458
Table 2. Inhibition of E. coli in vitro protein translation (EcIVT)14
ribosomal RNA and generated rationally designed
ligands which have been shown to interact selectively
with the desired region. With the recent appearance of
high-resolution crystal structures ofthe prokaryotic
ribosome,15ꢀ17 we are confident that this integrated
approach will herald a new era in anti-infective
research, generating a variety ofnovel antibacterial
agents with strong clinical potential.
and antibacterial activity (MIC95
Compd
)
% EcIVT
inhibition
MIC95
Scaffold
Amine
Thiostrepton 1
98 (0.1)
—
48
—
31
0.015
>200
>200
>200
>200
>200
8
3
6
2
3
e
g
(OH)
n
n
16
References and Notes
EcIVT percentage inhibition measured at 200 mM. Figures in par-
entheses represent IC50 in mM.
1. Dennis, S. M.; Nagaraja, T. G.; Dayton, A. D. Res. Vet.
Sci. 1986, 41, 251.
2. Xing, Y.; Draper, D. E. Biochemistry 1996, 35, 1581.
3. Swayze, E. E.; Jefferson, E. A.; Sannes-Lowery, K. A.;
Blyn, L. B.; Risen, L. M.; Arakawa, S.; Osgood, S. A.; Hof-
stader, S. A.; Griffey, R. H. J. Med. Chem. 2002, 45, 3816.
4. Lentzen, G.; Klink, R.; Matassova, N.; Aboul-Ela, F.;
Murchie, A. Chem. Bio., submitted for publication
5. Moody, C. J.; Bagley, M. C. J. Chem. Soc., Perkin Trans. 1
1998, 601.
6. Garner, P.; Park, J. M. J. Org. Chem. 1987, 52, 2361.
7. Erlenmeyer, H.; Buchmann, P.; Schenkel, H. Helv. Chim.
Acta 1944, 27, 1432.
8. Stohlmayer, M. M.; Tanaka, H.; Wandless, T. J. J. Am.
Chem. Soc. 1999, 121, 6100.
construct and whole ribosomal RNA. This finding
indicates that these simplified mimics ofthiostrepton do
indeed bind selectively to the L11BD region ofthe
ribosome, albeit with reduced potency compared to the
natural antibiotic.
Furthermore, it was found that inhibition of protein
translation correlated closely with the results ofthe
methylation assay, indicating that these specific binders
were indeed interfering with the dynamics of L11–
L11BD binding in a manner similar to Thiostrepton.
Disappointingly, however, these compounds failed to
display any in vitro antibacterial activity (Table 2).
9. Shin, C.-G.; Yonezawa, Y.; Yamada, T. Chem. Pharm.
Bull. 1984, 32, 3934.
10. Yonezawa, Y.; Shin, C.-G.; Ono, Y.; Yoshimura, J. Bull.
Chem. Soc. Jpn. 1980, 53, 2905.
11. Nakamura, Y.; Shin, C.-G.; Umemura, K.; Yoshimura, J.
Chem. Lett. 1992, 1005.
12. Thompson, J.; Schmidt, F.; Cundliffe, E. J. Biol. Chem.
1982, 257, 7915.
13. Lentzen, G. F; Murchie, A. I. H. WO0179543, 2001.
14. Inhibition ofprotein translation was performed essentially
as described in Zubay, G. Annu. Rev. Genet. 1973, 7, 267.
Greenberg, W. A.; Priestley, E. S.; Sears, P. S.; Alper, P. B.;
Rosenbohm, C.; Hendrix, M.; Hung, S.-C.; Wong, C.-H. J.
Am. Chem. Soc. 1999, 121, 6527.
15. Ban, N.; Nissen, P.; Hansen, J.; Moore, P. B.; Steitz, T. A.
Science 2000, 289, 905.
16. Wimberly, B. T.; Brodersen, D. E.; Clemons, W. M.;
Morgan-Warren, R.; Carter, A. P.; Vonrhein, C.; Hartsch, T.;
Ramakrishnan, V. Nature 2000, 407, 327.
Conclusions
These studies have yielded a number ofnovel ligands
that have been shown to interact specifically with a vital
functional region of the bacterial ribosome and inhibit
protein translation. In conjunction with our earlier
4
results elucidating the structure ofthis key region, the
work described here is the first reported example of
rational, structure-based drug design targeted toward
this complex but vital cellular machine. Employing
computational techniques, NMR, molecular biology,
biochemistry and medicinal chemistry in a unified man-
ner has allowed the investigation ofa ufnctional site
with under-exploited therapeutic potential, established
critical details ofthe interaction ofthiostrepton 1 with
17. Baucom, A.; Lieberman, K.; Earnest, T. N.; Cate, J. H. D.;
Noller, H. F. Science 2001, 292, 883.