Structure-based design of carboxybiphenylindole inhibitors of the ZipA-FtsZ interaction.

Article abstract of DOI:10.1039/b312016c

Structural features of two weak inhibitors of the ZipA-FtsZ protein-protein interaction which were found to bind to overlapping but different areas of the key binding site were combined in one new series of carboxybiphenyl-indoles with improved inhibitory

Full text of DOI:10.1039/b312016c

View Online / Journal Homepage / Table of Contents for this issue
Structure-based design of carboxybiphenylindole inhibitors of the
ZipA–FtsZ interaction
Alan G. Sutherland,*a Juan Alvarez,b Weidong Ding,a K. W. Foreman,b Cynthia Hess Kenny,a
Pornpen Labthavikul,c Lidia Mosyak,b Peter J. Petersen,c Thomas S. Rush III,b Alexey Ruzin,c
Desiree H. H. Tsaob and Karen L. Whelessa
a Chemical and Screening Sciences, Wyeth Research, Pearl River, NY 10965, USA.
E-mail: sutherag@wyeth.com; Fax: ϩ01 845 602 5561; Tel: ϩ01 845 602 3701
b Chemical and Screening Sciences, Wyeth Research, Cambridge, MA 02140, USA
c Infectious Diseases Biology, Wyeth Research, Pearl River, NY 10965, USA
Received 29th September 2003, Accepted 21st October 2003
First published as an Advance Article on the web 29th October 2003
Structural features of two weak inhibitors of the ZipA–
FtsZ protein–protein interaction which were found to bind
to overlapping but different areas of the key binding site
were combined in one new series of carboxybiphenyl-
indoles with improved inhibitory activity.
As the incidence of infectious bacteria with resistance to multi-
ple classes of drugs continues to increase1,2 there is a continuing
need to identify new antibiotics — particularly those that target
essential pathogen specific processes through novel mechanisms
of action. We are therefore investigating small molecule inhibi-
tors of the binding of the proteins ZipA and FtsZ as this inter-
action is an essential component of the cell division process in
Gram-negative bacteria.3–6 Briefly, these proteins are essential
components of the septal ring which forms at the site of cell
division — inhibition of the interaction between the two results
in inhibition of cell division, leading to filamentation and
ultimately cell death. While ZipA or its orthologs have only
been detected in Gram-negative organisms to date, FtsZ is
universally present in bacteria. ZipA orthologs are therefore
likely to exist in Gram-positive organisms7 so the discovery of
these inhibitors may lead to an antimicrobial agent with a
broad spectrum of activity — including against organisms with
resistance to a range of other drugs.
The structure of ZipA is well described8,9 and, in particular,
high quality X-ray and NMR structures demonstrating the
key binding interactions at and between the ZipA and FtsZ
C-termini have been solved.10–12 These show that the FtsZ bind-
ing site on ZipA is a broad, flat, predominantly hydrophobic
surface. Our initial leads were relatively weak inhibitors and
included the indoles 113 (IC50 1170 µM)14 and 215 (IC50 2060
µM), and the oxazole 316 (IC50 2750 µM). X-Ray analysis
demonstrated that all three compounds span this hydrophobic
region to some extent. While the crystal structures of the indole
ring systems essentially overlay in the binding site, the oxazole
occupies a largely distinct region (Fig. 1).
This raised the possibility that
a chimeric molecule,
incorporating structural features of both the indoles and
Fig. 1 Overlay of crystallographic orientations for indole 2 (violet)
with indole 1 (orange) and oxazole 3 (red). The protein surface is
colored according to lipophilicity, with brown indicating strong
lipophilicity, blue indicating strong hydrophilicity, and shades of green
indicating amphiphilicity.
oxazole, might have enhanced activity. Although the binding
free energies would not be additive because parts of either 1 or
2 overlap with 3 in their bound orientations, the interaction of
the combined fragments with distinct regions of the binding site
would be expected to increase the potency.
We therefore modeled several options for a chimera in an
effort to find the best starting point for potency optimization.
Compounds 2 and 3 overlap in two places (see Fig. 1). One of
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 4 1 3 8 – 4 1 4 0
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
4138

View Online
Table 1 Inhibition of the binding of ZipA and FtsZ14
the phenyl rings of the oxazole overlaps nearly completely with
the phenyl in the indole ring while the other oxazole phenyl
approaches the saturated carbons in the fused ring system.
The clear choice was to eliminate not only the phenyl from
the oxazole scaffold that overlaps with the indole but also the
saturated carbons from the fused ring indole scaffold (Fig. 2).
IC50/µM
1
2
1170
2060
2750
1060
619
286
192
3
8
9
12
14
Fig. 2 a) Potential attachment points for the fragments from 2 and 3.
Connections were considered at either the indole 4-position (A) or the
3-position (B); b) a “proof of concept” analogue.
Two potential attachment points were available for connect-
ing the resulting fragments: the indole 3- and 4-positions. Since
the binding site features a largely non-specific hydrophobic
interface, considering other (larger) rings besides the oxazole
became desirable for increased potency. Modeling suggested
that the indole 4-position is modestly close to the protein and
that increasing bulk at that position (even simply substituting a
phenyl for the oxazole) would perturb the binding mode con-
siderably.17 As expected from the crystal structures, a phenyl at
the indole 3-position is well tolerated. Attachments at the
indole C4 were thus lower in priority relative to the C3 position.
We therefore set out to prepare a series of ligands with the
general structure 4.
Our initial synthetic strategy was to construct the 2,3-
diarylindole ring system via a Fischer indole synthesis then
introduce the benzoic acid moiety via a Suzuki coupling. In
practice we ultimately prepared the ortho-substituted systems 8
and 9 by the reverse sequence (Scheme 1) as couplings to the
3-(2Ј-bromophenyl)indole system gave negligible yields of the
required products. Thus, after preparation of the aryl benzyl
ketone 5,18 Suzuki coupling (where use of aqueous 1-propanol
as solvent proved particularly advantageous)19 followed by sub-
jecting the crude product to a one-pot hydrazone formation-
Fischer cyclization provided rapid access to the esters 6 and 7.
Simple hydrolysis yielded the final compounds 8 and 9.
Scheme 2 Reagents and conditions: a) 1. PCl3, 2. AlCl3, benzene (61%);
b) PPh3, Pd(OAc)2, Na2CO3, C3H7OH, H2O, HO2CC6H4B(OH)2, ∆;
c) HCl, EtOH, ∆ (14% overall) d) PhNHNH2, HCl, EtOH, ∆ (11, 75%;
13, 98%); e) NaOH, aq. EtOH, ∆ (12, 96%; 14, 97%); f ), Pd(PPh3)4,
Na2CO3, C3H7OH, H2O, HO2CC6H4B(OH)2, ∆; g) HCl, EtOH, ∆ (25%
overall).
to give bromophenylindole 13, followed by a Suzuki coupling as
illustrated in the synthesis of 14 (although it proved pragmatic
to insert an esterification with subsequent hydrolysis in this
sequence to permit ready purification of the intermediate).
All the target compounds displayed an improved inhibition
of the ZipA–FtsZ interaction relative to our initial hits (Table
1) and were demonstrated to be interacting with the binding site
on ZipA by 1H and 15N NMR chemical shift perturbation
analysis.11,12 Although the ortho-substituted systems displayed a
relatively small improvement, molecular modeling suggested
that these systems were too sterically congested to achieve a
more planar conformation that would allow better binding with
the relatively flat protein surface. Such constraints play a far
smaller role with the meta-substituted systems and here we saw
a consequent marked improvement in activity.
The activity of the compounds against a variety of micro-
organisms (Table 2) also shows a marked improvement. Since
the ultimate goal of this project was the design of broad-
spectrum inhibitors of bacterial cell division, both Gram-
negative and Gram-positive bacteria were used for determin-
ation of minimal inhibitory concentrations (MIC). A yeast,
C. albicans, was used as a control to determine non-specific
activity as a means of indicating general cytotoxicity of the
compounds. Since none of the compounds inhibited growth of
E. coli wt (“wild-type”), likely due to an inability to effectively
penetrate the outer membrane, an outer membrane permeable
strain E. coli imp was utilized to assess penetration of the com-
pounds. Comparison of the data presented in Tables 1 and 2
shows that improvement in IC50 generally tracks with improve-
ment in MICs. The exception is 2 where the activity against C.
albicans suggests a general cytotoxicity likely ascribable to
detergent effects stemming from the amine side chain. The
inhibitory activity of compounds against Gram-positive
Scheme 1 Reagents and conditions: a) 1. PCl3, 2. AlCl3, benzene (87%);
b) PPh3, Pd(OAc)2, Na2CO3, C3H7OH, H2O, HO2CC6H4B(OH)2, ∆;
c) PhNHNH2, HCl, EtOH, ∆ (overall yield 6, 22%; 7, 24%); d) NaOH,
aq. EtOH, ∆ (8, 87%; 9, 98%).
The preparation of the meta-substituted systems (Scheme 2)
proved more forgiving, allowing two possible approaches. Thus,
starting from the common intermediate aryl benzyl ketone 10
we were able to use the same sequence as above to give target 12
via the intermediate ester 11. Alternatively we could also return
to our original plan of performing the Fischer cyclization first,
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 4 1 3 8 – 4 1 4 0
4139

View Online
Table 2 Minimal inhibitory concentrations (µg mLϪ1) against various microorganisms20
S. aureus
S. pneumoniae
E. faecalis
E. coli wt
E. coli imp
B. subtilis
C. albicans
1
128–256
8–16
512
16
16
4–8
4–8
1->128
64–128
8
256
4–64
16–32
4–32
8–32
2
128–256
16
512
16–32
32
4–8
256
512
512
128
128
128
128
2
512
512
512
16
8
16
128
nd
256
8
nd
4
512
8
2
3
9
512
128
128
128
128
>128
10
12
14
8–16
2–4
8
4
0.5
Piperacillin
<0.12
9 T. Ohashi, C. A. Hale, P. A. J. de Boer and H. P. Erickson,
J. Bacteriol., 2002, 184, 4313–4315.
10 L. Mosyak, Y. Zhang, E. Glasfeld, S. Haney, M. Stahl, J. Seehra and
W. S. Somers, EMBO J., 2000, 19, 3179–3191.
11 F. J. Moy, E. Glasfeld, L. Mosyak and R. Powers, Biochem., 2000,
39, 9146–9156.
12 F. J. Moy, E. Glasfeld and R. Powers, J. Biomol. NMR, 2000, 17,
275–276.
bacteria might be considered as a pointer to the existence of
ZipA ortholog(s) in these organisms. Alternatively it may sug-
gest that the compounds are not entirely specific for ZipA–FtsZ
interaction and may also be acting on other molecular targets.
Further research is needed to explore these possibilities.
We have demonstrated that a chimeric strategy does offer an
approach to improved inhibition of the ZipA–FtsZ interaction.
Future publications will describe the characterization of the
inhibitor–protein complex and the synthesis and activity of
analogues substituted at the indole 4-position.
13 P. Rosenmund, W. Trommer, D. Dorn-Zachertz and U. Ewerd-
walbesloh, Liebigs Ann. Chem., 1979, 1643–56.
14 All IC50 values were determined by a fluorescence polarization assay
which is described in depth in C. H. Kenny, W. Ding, K. Kelleher,
S. Benard, E. G. Dushin, A. G. Sutherland, L. Mosyak, R. Kriz and
G. Ellestad, Anal. Biochem., in press. Briefly, the assay determines
the inhibition of the binding of a fluorescein-labeled analogue of the
C-terminal residues (367–383) of FtsZ with the the C-terminal
domain of ZipA (residues 185–328). This assay has proved robust
for a wide variety of classes of inhibitor with close agreement upon
repetition of analyses.
Acknowledgements
We would like to thank the members of the Discovery
Analytical Chemistry department for their assistance in this
work and Russell G. Dushin and Janis Upeslacis for helpful
discussions.
15 M. E. Freed, E. Hertz and L. M. Rice, J. Med. Chem., 1964, 7, 628–
32.
16 K. Brown, J. F. Cavalla, D. Green and A. B. Wilson, Nature
(London), 1968, 219, 164.
Notes and references
17 Typically we used constrained minimizations using the Tripos force-
field in Sybyl with Kollman charges on the protein, Gasteiger–
Marsilli charges on the ligand, and a distance dependent dielectric
of 4r. The X-ray conformation of the protein with 2 bound was fixed
then 2 was then modified to an analog, which was allowed to relax in
all degrees of freedom.
18 J. E. Mulvaney and L. J. Carr, J. Org. Chem., 1968, 33, 3286–3291.
19 B. E. Huff, T. M. Koenig, D. Mitchell and M. A. Staszak, Org.
Synth., 1997, 75, 53–60.
1 C. W. Stratton, Emer. Infect. Dis., 2003, 9, 10–16.
2 A. P. Johnsonn and N. Woodford, J. Antimicrob. Chemother., 2002,
50, 621–623.
3 C. A. Hale and P. A. J. de Boer, Cell, 1997, 88, 175–185.
4 C. A. Hale and P. A. J. de Boer, J. Bacteriol., 1999, 181, 167–176.
5 Z. Liu, A. Mukherjee and J. Lutkenhaus, Mol. Microbiol., 1999, 31,
1853–1861.
6 S. Pichoff and J. Lutkenhaus, EMBO J., 2002, 21, 685–693.
7 S. Haney, PCT Int. Appl., 2002, WO 0251977.
8 S. A. Haney, E. Glasfeld, C. A. Hale, D. Keeney, Z. He and P. A. J.
de Boer, J. Biol. Chem., 2001, 276, 11980–11987.
20 Methods for Dilution of Antimicrobial Susceptibility Tests for Bac-
teria That Grow Aerobically, Approved Standards: M7-A5, vol. 20,
National Committee for Clinical Laboratory Standards, Wayne, PA.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 4 1 3 8 – 4 1 4 0
4140