Anthranilate 4H-oxazol-5-ones: Novel small molecule antibacterial acyl carrier protein synthase (AcpS) inhibitors

Article abstract of DOI:10.1016/j.bmcl.2003.10.032

D-optimal design and Projection to Latent Structures (PLS) analysis were used to optimize screening hit 5 (B. subtilis AcpS IC₅₀: 15 μM, B. subtilis MIC: >200 μM) into a series of 4H-oxazol-5-one, small molecule, antibacterial, AcpS inhibitors. Specifically, 15, 16 and 18 show μM or sub-μM AcpS inhibition (IC₅₀s: 15: 1.1 μM, 16: 1.5 μM, 18: 0.27 μM) and moderate antibacterial activity (MICs: 12.5-50 μM) against B. subtilis, E. faecalis ATCC, E. faecalis VRE and S. pneumo⁺.

Full text of DOI:10.1016/j.bmcl.2003.10.032

Bioorganic & Medicinal Chemistry Letters 14 (2004) 37–41
Anthranilate 4H-oxazol-5-ones: novel small molecule antibacterial
acyl carrier protein synthase (AcpS) inhibitors
Adam M. Gilbert,a,* Matthew Kirisits,a Patrick Toy,a David S. Nunn,a Amadeo Failli,b
Elizabeth G. Dushin,c Elena Novikova,c Peter J. Petersen,c Diane Joseph-McCarthy,d
Iain McFadyend and Christian C. Fritze
aDepartment of Chemical and Screening Sciences, Wyeth Research, 401 N. Middletown Road, Pearl River, NY 10945, USA
bDepartment of Chemical and Screening Sciences, Wyeth Research, CN-8000, Princeton, NJ 08852, USA
cDepartment of Infectious Diseases Research, Wyeth Research, 401 N. Middletown Road, Pearl River, NY 10945, USA
dDepartment of Chemical and Screening Sciences, Wyeth Research, 87 Cambridge Park Dive, Cambridge, MA 02140, USA
eMillennium Pharmaceuticals Inc., Cambridge, MA 02140, USA
Received 19 August 2003; revised 13 October 2003; accepted 14 October 2003
Abstract—D-optimal design and Projection to Latent Structures (PLS) analysis were used to optimize screening hit 5 (B. subtilis
AcpS IC50: 15 mM, B. subtilis MIC: >200 mM) into a series of 4H-oxazol-5-one, small molecule, antibacterial, AcpS inhibitors.
Specifically, 15, 16 and 18 show mM or sub-mM AcpS inhibition (IC50s: 15: 1.1 mM, 16: 1.5 mM, 18: 0.27 mM) and moderate anti-
bacterial activity (MICs: 12.5–50 mM) against B. subtilis, E. faecalis ATCC, E. faecalis VRE and S. pneumo+.
# 2003 Elsevier Ltd. All rights reserved.
1.Introduction
holo-Acp accepts acetyl and malonyl groups from ace-
tyl-CoA and malonyl-CoA respectively and shuttles
these acyl groups between enzymes in the fatty acid
biosynthesis pathway.3 AcpS is required for the bio-
synthesis of important components of membrane lipids
and bacterial lipopolysaccharides, and the essentiality of
this enzyme was determined genetically in Streptococcus
pneumoniae (S. pneumo)4 and Escherichia coli (E. coli).5
The crystal structures of S. pneumo6 and Bacillus subtilis
(B. subtilis)7 AcpS have been determined but no small
molecule inhibitors of AcpS have previously been
reported.
Rapidly developing resistance of known bacterial
strains to currently prescribed pharmaceutical agents
has stimulated a vigorous search for novel mechanisms
to kill bacteria.1,2 Strains such as methicillin-resistant
Staphylococcus aureus (MRSA), vancomycin-resistant
Enterococcus (VRE), penicillin-resistant Streptococcus
pneumoniae (PRSP) and multiple drug-resistant Entero-
bacteriaceae klebsiella (MDR ESBL) do not respond to
current clinical drug therapies including quinolones and
b-lactams.
In order to overcome bacterial resistance to currently
prescribed antibiotics, we have searched for other cel-
lular processes that would serve as good antibacterial
targets. One such process is the conversion of apo-acyl
carrier protein (apo-Acp) to holo-acyl carrier protein
(holo-Acp) which is catalyzed by acyl carrier protein syn-
thase (AcpS), an enzyme that catalyzes the transfer of
the 40-phosphopantetheinyl group of coenzyme A
(CoA) to a serine residue of apo-Acp. The resulting
In this letter, we disclose the discovery of a series of
small molecule, antibacterial, anthranilate 4H-oxazol-5-
one AcpS inhibitors 1. In addition, we will show how
multivariate analysis (MVA) and experimental design
were used to rapidly optimize the biological activity of
these molecules.
2.Chemistry
Anthranilate 4H-oxazol-5-ones
according to literature procedures. Thus, acid chlorides
1
were prepared
* Corresponding author. Tel.: +1-845-602-4865; fax: +1-845-602-
5561; e-mail: gilbera@wyeth.com
2 were converted to hippuric acids 3,8 and cyclized to
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.10.032

38
A. M. Gilbert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 37–41
the corresponding 4H-oxazol-5-ones 4 with triethyl
orthoformate and Ac2O.9 The target anthranilate 4H-
oxazol-5-ones 1 were prepared by heating 3 with
anthranilic acids in EtOH.10
3.Biology
AcpS inhibitory activity was determined using a B. sub-
11
tilis GST-Acp-HTRFassay.
The in vitro determin-
ation of the minimal inhibitory concentrations (MICs)
against aerobic bacteria was performed by the micro-
dilution broth method as recommended by the National
Committee for Clinical Laboratory Standards.12
4.Results and discussion
Scheme 1. Reagents and conditions: (a) Glycine, NaOH, H2O, 23 ꢂC;
(b) (EtO)3CH, Ac2O, 95 ꢂC; (c) Anthranilic acid (H2N(Ar2)CO2H),
EtOH, 80 ꢂC.
4-Cl Phenyl 4H-oxazol-5-one 5 was identified as an
inhibitor of B. subtilis AcpS in a high-throughput screen
(AcpS IC50: 15 mM) (Fig. 1). Unfortunately this com-
pound had no MIC activity against several strains of
bacteria. Compound 5 showed a low PAMPA perme-
ability (0.05ꢀ10ꢁ6 cm/s)13 and a low AlogD (0.43) sug-
gesting that this compound may not enter the bacterial
cell to reach its target. Molecular modeling of 5 into the
active site of B. subtilis AcpS showed that variation of
the R1, R2 and R3 positions could lead to optimized
contacts in the active site, thereby increasing binding
affinity. These modeling constraints were used to select
monomers for an initial synthetic array using 4 acid
chlorides and 14 anthranilic acids (chosen using a D-
optimal design14 of the t[1], t[2] and t[3] scores of the
PCA of a 27 anthranilic acid candidate set) containing
variations of the R1, and R2 substituents (Fig. 1).
using calculated MW, hydrogen bond donor (HBD)
count, hydrogen bond acceptor (HBA) count, daylight
clogP and cMR, polar surface area (PSA),16 and mea-
sured s values17 and atom counts for the whole mole-
cule as well as the R1, R2 and R3 substituents,
AcpS inhibition data (IC50) and MIC data (B. subtilis,
E. faecalis ATCC, E. faecalis VRE and S. pneumo+)
using SIMCA-P.18 Two PLS models were generated to
understand the effects of simultaneously varying the R1,
R2 and R3 substituents: a model that correlates physi-
cochemical properties with AcpS inhibitory activity (A:
2, R2X: 0.50, R2Y: 0.71, Q2(cum): 0.63, Fig. 2), and a
model that correlates physicochemical properties AcpS
inhibitory activity and MIC activity A: 2, R2X: 0.50,
R2Y: 0.44, Q2(cum): 0.29, Fig. 3).
A set of 42 (Set 1) oxazolones was synthesized using the
procedure outlined in Scheme 1 (Fig. 1). AcpS inhibi-
tory and MIC data are presented for selected com-
pounds in Table 1. MVA15 was used to summarize/
correlate the calculated and the measured molecular
physicochemical properties with in vitro AcpS potency
and MIC activity for Set 1. PLS models were generated
Figure 2 shows AcpS inhibitory activity is primarily
correlated with increasing R3clogP, R3sigma and
R3MW. This effect is seen in comparing several com-
pounds in Table 1: compare 10 (R3=Cl) to 11 (R3=CF3)
and 12 (R3=OMe) to 13 (R3=CF3). Figure 2 also shows
that AcpS inhibition can be increased by decreasing the
values of the R2cMR and R2clogP variables. The
R1PSA, R1HBD, R1HBA variables are on an orthogo-
nal axis to the R2, R3 and AcpS inhibitory activity
variables and thus have a minimal effect. The goodness
of prediction of the model is satisfactory as the Q2(cum)
term is above 0.5 (Q2(cum)=0.63).
The PLS model in Figure 3 is even more informative as
it correlates increasing AcpS inhibition and MIC activ-
ity with increasing R3clogP, R3sigma and R3MW and
decreasing values of the R1 and R2 variables (primarily
the w*c[1] loading). AcpS inhibitory activity and MIC
activity appear to be correlated— the correlation is
strongest for the B. subtilis MIC activity, and there
is correlation between increasing AcpS inhibition/MIC
activity and increasing molecular clogP and MW.
Importantly, analysis of these two PLS models indicates
that more potent antibacterial AcpS inhibitors can be
synthesized via adjustment of the R3 substituent noted
Figure 1. Retrosynthetic analysis of compound 5, monomer selection
and synthesis of 4H-oxazol-5-one Set 1.

A. M. Gilbert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 37–41
39
Table 1. In vitro AcpS inhibitory data and minimum inhibitory concentration (MIC) data for selected Set 1 oxazolones
Compd
In vitro AcpS inhibition
MIC data (mM)b
R1
R2
R3
IC50 (mM)a
B. subtilis
E. faecalis ATCC
E. faecalis VRE
S. pneumo+
5
6
7
8
H
H
FH
Br
I
Br
Br
FH
F
H
H
Cl
H
15.0
>20
2.1
>200
—
>200
>200
200
>200
>200
>200
200
>200
—
100
100
25
>200
100
>200
100
>200
>200
—
100
200
50
—
50
—
200
100
>200
—
200
100
12.5
—
100
—
100
100
Cl
H
H
Cl
Cl
OMe
H
Cl
Cl
Cl
CF3
3.4
1.3
>20
2.0
>20
1.9
9
10
11
12
13
14
CF 3
CF3
Me
H
0.83
100
a IC50 values were determined once over a range of 7 concentrations. Standard error for each run is ꢃ10%.
b Compounds were tested 1ꢀ against each bacterial strain. Standard error for each run is ꢃ1 dilution.
Figure 2. PLS loadings (w*c[1] vs w*c[2]) of Set 1: physicochemical properties (X matrix), AcpS inhibitory data (Y matrix) (A: 2, R2X: 0.50, R2Y:
0.71, Q2(cum): 0.63). Model shows that AcpS inhibitory activity primarily increases with increasing R3clogP, R3sigma and R3MW values and
decreasing R2clogP and R2cMR values. R1HBD, R1HBA and R1PSA have little effect.
Figure 3. PLS loadings (w*c[1] vs w*c[2]) of Set 1: physicochemical properties (X matrix), AcpS inhibitory data and bacterial MIC data (Y matrix)
(A: 2, R2X: 0.50, R2Y: 0.44, Q2(cum): 0.29). Model shows that AcpS inhibitory activity and MIC activity primarily increases with increasing
R3clogP, R3sigma and R3MW values and with decreasing values for the R1 and R2 variables. AcpS inhibitory activity and MIC activity also increase
with increasing molecular clogP and MW.

40
A. M. Gilbert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 37–41
Table 2. In vitro AcpS inhibitory data and minimum inhibitory concentration (MIC) data for selected Set 2 oxazolones
Compd
In vitro AcpS inhibition
MIC datab
E. faecalis ATCC
R1
R4
R5
IC50 (mM)a
B. subtilis (mM)b
E. faecalis VRE
S. pneumo+
15
16
17
18
19
20
21
Cl
Br
NAc
Me
Cl
FCF
FCF
FCF
1.1
1.5
0.13
0.27
20
50
50
>200
25
50
50
25
>200
100
100
50
50
25
>200
100
50
12.5
12.5
200
25
25
12.5
25
3
3
3
3
FCF
OCF3
OCF3
OCF3
H
H
H
Br
Me
15
4.4
25
25
25
100
200
a IC50 values were determined 1ꢀ over a range of 7 concentrations. Standard error for each run is ꢃ10%.
b Compounds were tested once against each bacterial strain. Standard error for each run is ꢃ1 dilution.
activity as 15, 16 and 18 but have less potent enzymatic
activity. These two groups of compounds do show the
same trend in biological activity-AcpS IC50: R1=Cl.
R1=Br
>R1=Me;
Overall
MIC
activity:
R1=Me>R1=Cl. R1=Br.
Thus we have shown how MVA and experimental
design can be used to rapidly optimize both in vitro
AcpS inhibitory/antibacterial activity to produce a ser-
ies of small molecule AcpS inhibitors represented by
compound 18. MVA is a particularly powerful as it
effectively summarizes variables into simpler linear
combinations that are readily interpretable. It also
allows one to interpret the effect of simultaneously
varying multiple positions within a molecule. Additional
studies on other small molecule AcpS inhibitors will be
forthcoming.
Figure 4. Retrosynthetic analysis of compound 14, monomer selection
and synthesis of 4H-oxazol-5-one Set 2.
Acknowledgements
above and by increasing overall molecular lipophilicity/
MW. This conclusion was then used as the new starting
point for a new library synthesis.
The authors would like to thank the members of the
Pearl River Discovery Analytical Chemistry Pharma-
ceutical Profiling Group for aqueous solubility,
PAMPA permeability, AlogD and CYP inhibition
determinations on all compounds in this manuscript.
A second set of oxazolones (Set 2) was then synthesized
as shown in Figure 4. Acid chlorides were used where
the R5 substituent (R3 substituent in Set 1) was set to F,
Cl, OCF3 and CF3, and anthranilic acids were used that
gave active compounds in Set 1. AcpS inhibitory and
MIC data are presented for selected Set 2 compounds in
Table 2. Potent antibacterial AcpS inhibitors were
References and notes
1. Cohen, M. L. Science 1992, 257, 1050.
2. Spratt, B. G. Science 1994, 264, 388.
3. Lambalot, R. H.; Gehring, A. M.; Flugel, R. S.; Zuber,
P.; LaCelle, M.; Marahiel, M. A.; Reid, R.; Khosla, C.;
Walsh, C. T. Chem. Biol. 1996, 3, 923.
4. McAllister, K. A.; Peery, R. B.; Meier, T. I.; Fischl, A. S.;
Zhao, G. J. Biol. Chem. 2000, 275, 30864.
5. Takiff, H. E.; Baker, T.; Copeland, T.; Chen, S. M.;
Court, D. L. J. Bacteriol. 1992, 174, 1544.
5
obtained where R4=Fand R =CF3 as seen in 15, 16
and 18. Compound 18, having an AcpS IC50=0.27 mM
and a B. subtilis MIC=25 mM, is the most potent com-
pound in terms of both AcpS inhibition and B. subtilis
MIC activity. Compound 17 is the most potent AcpS
inhibitor obtained, but it fails to show any MIC activ-
ity. A second series of antibacterial AcpS inhibitors
emerged with the substitution pattern of R4=OCF3 and
R5=H (19–21). These compounds show similar MIC
6. Chirgadze, N. Y.; Briggs, S. L.; McAllister, K. A.; Fischl,
A. S.; Zhao, G. EMBO J. 2000, 19, 5281.

A. M. Gilbert et al. / Bioorg. Med. Chem. Lett. 14 (2004) 37–41
41
7. Parris, K. D.; Lin, L.; Tam, A.; Mathew, R.; Hixon, J.;
Stahl, M.; Fritz, C. C.; Seehra, J.; Somers, W. S. Structure
2000, 8, 883.
8. Ingersoll, A. W.; Barcock, S. H. Org. Synth. 1932, 12, 40.
9. Bland, J. M.; Stammer, C. H.; Varughese, K. I. J. Org.
Chem. 1984, 49, 1634.
cyanin conjugate (ProZyme, San Leandro, CA), 410 mM
KF, 4.1 mM EDTA in 0.5ꢀPBS). The mixture was incu-
bated overnight before reading in a Wallac VictorTM plate
reader (excitation at 340 nm, emission at 665 nm).
12. National Committee for Clinical Laboratory Standards
1997, Approved standards M7-A.
10. All compounds were characterized by LC/(ES)MS
(>90% purity and correct (MH)+ was seen).
13. Kansy, M.; Senner, F.; Gubernator, K. J. Med. Chem.
1998, 41, 1007.
11. Enzyme activity was measured using a homogeneous
time-resolved fluorescence resonance energy transfer
(HTRF1) assay. Assays were run in 384 well plates and
consisted of compound, 15 ng/mL AcpS, 1.9 mM biotin-
CoA (prepared by reacting equimolar amounts of CoA
and biocytin maleimide (Molecular Probes, Eugene, OR)),
14. Eriksson, L.; Johnasson, E.; Kettaneh-Wold, N.; Wik-
strom, C.; Wold, S. Design of Experiments, Principles and
Applications. 2000, Umea, Sweden: Umetrics AB.
15. Eriksson, L.; Johnasson, E.; Kettaneh-Wold, N.; Wold, S.
Introduction to Multi- and Megavariate Data Analysis
using Projection Methods (PCA & PLS). 1999, Umea,
Sweden: Umetrics AB.
.
and 1.25 mg/mL GST-ACP in 50 mM Tris HCl, pH 8.0,
7.5 mM MgCl2, 3.75 mM DTT, 0.0375% Tween-20, and
37.5 mg/mL BSA in a total volume 20 mL. After 3 h at
room temperature, the reaction was terminated by adding
60 mL of the stop/detection mix (82 ng/mL europium
cryptate conjugate of GST monoclonal antibody (Pack-
ard, Meriden, CT), 28.7 mg/mL streptavidin–allophyco-
16. Ertl, P.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 43,
3714.
17. Hansch, C.; Leo, A.; Hoekman, D. Exploring QSAR:
Hydrophobic, Electronic, and Steric Constants; American
Chemical Society: Washington DC, 1995.
18. Umetrics. SIMCA-P, 2003.