First X-ray cocrystal structure of a bacterial FabH condensing enzyme and a small molecule inhibitor achieved using rational design and homology modeling

Article abstract of DOI:10.1021/jm025571b

The first cocrystal structure of a bacterial FabH condensing enzyme and a small molecule inhibitor is reported. The inhibitor was obtained by rational modification of a high throughput screening lead with the aid of a S. pneumoniae FabH homology model. This homology model was used to design analogues that would have both high affinity for the enzyme and appropriate aqueous solubility to facilitate cocrystallization studies.

Full text of DOI:10.1021/jm025571b

J . Med. Chem. 2003, 46, 5-8
5
As fatty acid synthesis is an essential process for cell
survival, the enzymes involved in the FAS pathway
represent promising targets for broad-spectrum anti-
bacterial agents. Cerulenin and thiolactomycin, natural-
product inhibitors of the FabB and FabF condensing
enzymes, both demonstrate antibacterial activity.4 In
addition, as the bacterial type II FAS is distinct from
the mammalian type I FAS, there is a high probability
that type II selective inhibitors can be identified.
F ir st X-r a y Cocr ysta l Str u ctu r e of a
Ba cter ia l F a bH Con d en sin g En zym e a n d
a Sm a ll Molecu le In h ibitor Ach ieved
Usin g Ra tion a l Design a n d Hom ology
Mod elin g
Robert A. Daines,*,† Israil Pendrak,† Kelvin Sham,†,X
Glenn S. Van Aller,§ Alex K. Konstantinidis,§,3
J ohn T. Lonsdale,§ Cheryl A. J anson,‡ Xiayang Qiu,‡,#
Martin Brandt,| Sanjay S. Khandekar,
Carol Silverman, and Martha S. Head*,‡
Departments of Medicinal Chemistry, Computational,
Analytical & Structural Sciences, Microbial Biochemistry,
Assay Methodology Development, and Protein Biochemistry,
GlaxoSmithKline, 1250 S. Collegeville Road,
A high throughput screening effort directed against
Streptococcus pneumoniae FabH identified the indole
analogue 1 (Table 1) as a potent inhibitor of the enzyme.
In addition to possessing antibacterial activity against
both Gram positive and Gram negative bacteria, 1
demonstrated good selectivity for the bacterial type II
FAS over the human type I FAS (IC50 ) 30 µM, hFAS).
While somewhat less potent, 1 was also shown to be a
good inhibitor of E. coli FabH. Structure-activity stud-
ies quickly revealed the essential nature of the 2,6-
dichlorobenzyl substituent. Analogues lacking this moi-
ety demonstrated greatly reduced activity against both
enzymes, indicating that the 2,6-dichlorobenzyl group
played a key role in enzyme-inhibitor recognition.
Recently, we reported X-ray crystal structures of E.
coli FabH for both the apo-enzyme and for the acetyl-
CoA complex.5 The enzyme is characterized by having
a catalytic triad, consisting of Cys112, His249, and
Asn279, located at the bottom of a hydrophobic tunnel.
The adenine subunit of the acetyl-CoA substrate binds
to a region on the surface of the protein adjacent to the
tunnel. The adenine-binding site consists principally of
Trp32 and Arg151. The pantetheine of acetyl-CoA
extends down the tunnel placing the acetate in proxim-
ity to Cys112. While the ACP binding site is less well
defined, the pantetheine chain of malonyl ACP must
also extend down the tunnel to the catalytic residues.
We wished to utilize this structural information to
design more potent FabH inhibitors. One specific goal
of our effort was to obtain X-ray crystal structures of
enzyme-inhibitor complexes to support the medicinal
chemistry effort. However, attempts to cocrystallize 1
with either E. coli FabH or S. pneumoniae FabH were
unsuccessful, presumably a result of the very poor
aqueous solubility of 1. Docking simulations were
investigated in an attempt to predict the bound confor-
mation of 1.6
Collegeville, Pennsylvania 19426
Received August 13, 2002
Abstr a ct: The first cocrystal structure of a bacterial FabH
condensing enzyme and a small molecule inhibitor is reported.
The inhibitor was obtained by rational modification of a high
throughput screening lead with the aid of a S. pneumoniae
FabH homology model. This homology model was used to
design analogues that would have both high affinity for the
enzyme and appropriate aqueous solubility to facilitate co-
crystallization studies.
â-Ketoacyl-acyl carrier protein (ACP) synthase III,
also known as FabH, plays an essential role in the
biosynthesis of bacterial fatty acids.1 FabH is the
initiator of the fatty acid chain elongation cycle2 and
plays a key regulatory role for the entire biosynthetic
pathway via feedback inhibition by long-chain acyl
ACPs.3 FabH, a member of the family of condensing
enzymes, catalyzes the cysteine-mediated Claisen con-
densation reaction between malonyl ACP and an enzyme-
bound acetate unit derived from acetyl-CoA. The cata-
lytic mechanism involves acetylation of cysteine by
acetyl-CoA, with release of CoA, followed by condensa-
tion with the enolate derived from enzyme-mediated
decarboxylation of malonyl ACP. The acetoacetyl ACP
reaction product then feeds into the fatty acid biosyn-
thesis chain elongation cycle. Whereas FabH is the
initiator of the fatty acid synthesis (FAS) cycle, two
additional condensing enzymes, FabB and FabF, per-
form the chain elongation reactions in subsequent cycles
leading to long-chain acyl ACP products. While FabB
and FabF are also condensing enzymes, FabH is struc-
turally distinct.
Initial attempts at docking 1 into the E. coli FabH
crystal structure were not able to produce a reliable
binding model that explained key features of the avail-
able SAR. Specifically, the calculated binding modes did
not account for the critical nature of the 2,6-dichloro-
benzyl group. This was true when using structures
derived from either the apo-enzyme or the acetyl-CoA
complex. As an alternative approach, and not wanting
to make significant alteration to the E. coli FabH crystal
* Corresponding authors. E-mail: robert.a.daines@gsk.com.
† Department of Medicinal Chemistry.
‡ Department of Computational, Analytical & Structural Sciences.
§ Department of Microbial Biochemistry.
| Department of Assay Methodology Development.
Department of Protein Biochemistry.
X Current address: Amgen Inc., Thousand Oaks, CA.
3 Current address: Abbott Laboratories, Abbott Park, IL
#
Current address: Pfizer Inc., Groton, CT.
10.1021/jm025571b CCC: $25.00 © 2003 American Chemical Society
Published on Web 12/02/2002

6
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 1
Letters
Ta ble 1. Activities vs S. pneumoniae and E. coli FabHa,b
a
b
Compounds 2-8 were tested as the corresponding disodium salts; compound 1 was tested as the free acid. IC50 values are an average
of two or more determinations.
F igu r e 1. Overlay of active site regions of the S. pneumoniae FabH homology model with the E. coli FabH apo-enzyme crystal
structure. Residues that are not common to both enzymes are highlighted in cyan (S. pneumoniae) and magenta (E. coli). The
catalytic triad (Cys112, His249, Asn279, not labeled) is shown at the lower portion of the image and the adenine binding residues
(Trp32, Arg151) are at the top left.
structure, a homology model of S. pneumoniae FabH
was constructed using the reported E. coli FabH-CoA
cocrystal as the structural template.7 S. pneumoniae
FabH was chosen for this work for several reasons: (1)
S. pneumoniae FabH was the protein used for our
primary assay, (2) the inhibitor demonstrated greater
potency against the S. pneumoniae enzyme, (3) efforts
were underway to crystallize the S. pneumoniae protein,
and (4) the two enzymes are very similar within the
active site region, possessing 80% identity (Figure 1).
In contrast to the E. coli crystal structures, docking
into the S. pneumoniae homology model provided a
rational binding mode that was consistent with the
experimental data. In this predicted binding mode
(Figure 2), the 2,6-dichlorobenzyl group of 1 binds
within the narrow active site tunnel. While this group
lies near the catalytic triad (Cys112, His249, Asn279),
there is no direct interaction with these residues. The
requirement of the 2,6-dichlorobenzyl group for potent
activity is likely due to the exquisite shape complemen-
tarity to this region of the S. pneumoniae protein.
Despite the conserved nature of the active sites of the
S. pneumoniae and E. coli enzymes, a similar binding
mode in the E. coli FabH crystal structure could not be
achieved due to space restriction within this region.
The S. pneumoniae homology model also predicted
that the carboxylic acid of 1 interacts with arginines
on the protein surface, while the 6-chloropiperonyl group
makes largely hydrophobic contacts with residues at the
mouth of the active site tunnel. This placement of the
6-chloropiperonyl group locates it in close proximity to
the arginine-rich region at the top of the tunnel. This
suggested that replacement of this hydrophobic group
with more polar side chains might be permissible. Upon
the basis of these results, a series of more hydrophilic
analogues were prepared that were predicted by the
model to retain affinity for the enzyme while also
improving compound solubility. It was believed that this
approach would produce compounds suitable for cocrys-
tallization with E. coli and S. pneumoniae FabH.
The resulting compounds 2-8 are shown in Table 1.8
Incorporation of a second acidic group was chosen
because of the possibility of interactions with the
multiple arginine residues at the protein surface sur-
rounding the active site tunnel. In addition, the double
negative charge would increase aqueous solubility,
albeit at the expense of bacterial cell penetration. With
the exception of compound 7, all analogues in Table 1
demonstrated comparable levels of inhibition to that of
lead compound 1. As predicted, these compounds lacked

Letters
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 1
7
F igu r e 2. Predicted bound conformation of 1 in a homology model of S. pneumoniae FabH. Residues depicted (S. pneumoniae
FabH numbering) correspond to those of E. coli FabH shown in Figure 3.
F igu r e 3. Cocrystal structure of compound 5 with E. coli FabH (E. coli FabH numbering).
F igu r e 4. Surface diagram of cocrystal structure of 5 with E. coli FabH illustrating fit of 2,6-dichlorobenzyl group in active site
tunnel. View is perpendicular to that of Figure 3 with the catalytic residues (not shown) located at the bottom of the active site
tunnel.
antimicrobial activity but possessed suitable aqueous
solubility to permit cocrystallization with the available
FabH proteins.
gauche+ conformation in the apo-enzyme, and is in the
gauche- conformation in the cocrystal structure (com-
pare with Figure 1). Along with a small shift in the
backbone angles, this change in dihedral angle opens
enough space in the binding site pocket to accommodate
the second chlorine of the 2,6-dichlorobenzyl group. As
noted, this added space was lacking in the previous E.
coli FabH crystal structures and, therefore, prevented
the successful docking of 1.
Additional key interactions observed in the X-ray
structure include a water-mediated interaction between
the C-2 carboxylic acid and Arg36 as well as a salt
bridge between the hexanoate side chain and Arg151,
a residue that forms part of the adenine binding site of
acetyl-CoA.5 The structure further reveals that com-
pound 5 binds to the apo-form of the enzyme (Cys112
Cocrystallization studies yielded an X-ray structure
for the enzyme-inhibitor complex between 5 and E. coli
FabH (Figure 3).9 This is the first reported cocrystal
structure of a FabH condensing enzyme showing the
binding mode of a small molecule inhibitor. The crys-
tallographically determined binding mode of this inhibi-
tor is remarkably similar to that predicted by docking
1 into the S. pneumoniae FabH homology model. In
particular, the 2,6-dichlorobenzyl group fits into the
hydrophobic tunnel as predicted, with no direct interac-
tion with the catalytic residues. One key difference is
seen for Ile156 between this cocrystal structure and that
of the apo-enzyme.5 The dihedral angle Ì1 is in the

8
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 1
Letters
is not acetylated) as predicted by kinetic studies (data
not shown). With the 2,6-dichlorobenzyl group filling the
active site tunnel, the acetyl-CoA substrate is prevented
access to the catalytic residues and is blocked from
acetylating Cys112, the first step in the catalytic cycle.
In comparing the predicted binding mode of 1 (Figure
2) with the actual bound structure of 5 (Figure 3), it
appears that important elements for recognition exist.
First, the 2,6-dichlorobenzyl group is required to fill a
complementary hydrophobic region within the active
site tunnel (Figure 4). Second, the presence of an acidic
group is needed in order to form an ionic interaction
with one of the arginine residues at the top of the active
site tunnel (either directly or through water). For the
newly synthesized analogues, an additional binding
element has been incorporated via a second acidic group.
The data in Table 1 suggests that the acceptable chain
length for this second acidic group is four to seven
carbon units with four appearing optimal. Compound
7, containing the acetate side chain, is likely to be too
short to make optimum contacts with one of these
arginine residues. The indole itself appears to merely
serve as a scaffold on which these three binding ele-
ments are arranged. Interestingly, the crystal structure
does not shed additional light on the nature of the
6-chloropiperonyl binding of 1.
The present study describes the use of a S. pneumo-
niae FabH homology model to design inhibitors with
favorable physical properties to facilitate cocrystalliza-
tion studies. While lead compound 1 contained a hy-
drophobic group at the 1-position of the indole, docking
calculations suggested that more polar groups would be
tolerated within this region. The resulting structural
modifications led to potent inhibitors with improved
aqueous solubility. This approach has resulted in the
first X-ray crystal structure of an enzyme-small mol-
ecule inhibitor complex for a FabH condensing enzyme.
The information obtained from this cocrystal will be
used for the rational design of more potent inhibitors
of this novel and unexploited antibacterial target.10 Of
significance is the ability to use a FabH from one
bacteria, in this case S. pneumoniae, to design inhibitors
of FabH from multiple bacteria, i.e., S. pneumoniae and
E. coli. This will be a critical element in obtaining broad-
spectrum antibacterial agents that inhibit FabH from
many different organisms.
The atomic coordinates of the cocrystal structure of
5 with E. coli FabH have been deposited in the RCSB
Protein Data Bank with accession code 1MZS.
Su p p or tin g In for m a tion Ava ila ble: Experimental de-
tails and structural characterization. This material is available
free of charge via the Internet at http://pubs.acs.org.
Refer en ces
(1) Rock, C. O.; Cronan, J . E. Escherichia coli as a model for the
regulation of dissociable (type II) fatty acid biosynthesis. Bio-
chim. Biophys. Acta 1996, 1302, 1-16.
(2) (a) Tsay, J . T.; Oh, W.; Larson, T. J .; J akowski, S.; Rock, C. O.
Isolation and Characterization of the â-Ketoacyl-Acyl Carrier
Protein Synthase III Gene (fabH) from Escherichia coli K-12. J .
Biol. Chem. 1992, 267, 6807-6814. (b) Clough, R. C.; Matthis,
A. L.; Barnum, S. R.; J aworski, J . G. Purification and Charac-
terization of 3-Ketoacyl-Acyl Carrier Protein Synthase III from
Spinach. A Condensing Enzyme Utilizing Acetyl-Coenzyme A
to Initiate Fatty Acid Synthesis. J . Biol. Chem. 1992, 267,
20992-20998.
(3) Heath, R. J .; Rock, C. O. Regulation of Fatty Acid Elongation
and Initiation by Acyl-Acyl Carrier Protein in Escherichia coli.
J . Biol. Chem. 1996, 271, 1833-1836.
(4) Magnuson, K.; J ackowski, S.; Rock, C. O.; Cronan, J . E., J r.
Regulation of Fatty Acid Biosynthesis in Escherichia coli.
Microbiol. Rev. 1993, 57, 522-542.
(5) (a) Qiu, X.; J anson, C. A.; Konstantinidis, A. K.; Nwagwu, S.;
Silverman, C.; Smith, W. W.; Khandekar, S.; Lonsdale, J .; Abdel-
Meguid, S. S. Crystal Structure of â-Ketoacyl-Acyl Carrier
Protein Synthase III. J . Biol. Chem. 1999, 274, 36465-36471.
(b) Qiu, X.; J anson, C. A.; Smith, W. W.; Head, M.; Lonsdale, J .;
Konstantinidis, A. K. Refined Structures of â-Ketoacyl-Acyl
Carrier Protein Synthase III. J . Mol. Biol. 2001, 307, 341-356.
The above references contain detailed analyses of the E. coli
FabH structure including substrate binding elements and
catalytic mechanism.
(6) All docking calculations were performed using the program Flo;
see McMartin, C.; Bohacek, R. S. QXP: powerful, rapid computer
algorithms for structure-based design. J . Comput.-Aided Mol.
Design 1997, 11, 333-344. No structural waters were included
in docking calculations in S. pneumoniae FabH.
(7) The model was built using the MOE modeling package, Chemical
Computing Group, Montreal Canada.
(8) For enzyme assay conditions see Khandekar, S. S.; Gentry, D.
R.; Van Aller, G. S.; Warren, P.; Xiang, H.; Silverman, C.; Doyle,
M. L.; Chambers, P. A.; Konstantinidis, A. K.; Brandt, M.;
Daines, R. A.; Lonsdale, J . T. Identification, Substrate Specific-
ity, and Inhibition of the Streptococcus pneumoniae â-Ketoacyl-
Acyl Carrier Protein Synthase III (FabH). J . Biol. Chem. 2001,
276, 30024-30030.
(9) Attempts at crystallizing S. pneumoniae FabH with and without
inhibitors have been unsuccessful.
(10) For two recently reported inhibitors of S. aureus FabH see Xin,
H.; Reynolds, K. A. Purification, Characterization, and Identi-
fication of Novel Inhibitors of the â-Ketoacyl-Acyl Carrier Protein
Synthase III (FabH) from Staphylococcus aureus. Antimicrob.
Agents Chemother. 2002, 46, 1310-1318.
J M025571B