Inhibition of the Bacterial Enoyl Reductase FabI by Triclosan: A Structure-Reactivity Analysis of FabI Inhibition by Triclosan Analogues

Article abstract of DOI:10.1021/jm030182i

To explore the molecular basis for the picomolar affinity of triclosan for FabI, the enoyl reductase enzyme from the type II fatty acid biosynthesis pathway in Escherichia coli, an SAR study has been conducted using a series of triclosan analogues. Triclosan (1) is a slow, tight-binding inhibitor of FabI, interacting specifically with the E·NAD⁺ form of the enzyme with a K₁ value of 7 pM. In contrast, 2-phenoxyphenol (2) binds with equal affinity to the E·NAD⁺ (K₁ = 0.5 μM) and E·NADH (K₂ = 0.4 μM) forms of the enzyme and lacks the slow-binding step observed for triclosan. Thus, removal of the three triclosan chlorine atoms reduces the affinity of the inhibitor for FabI by 70 000-fold and removes the preference for the E·NAD⁺ FabI complex. 5-Chloro-2-phenoxyphenol (3) is a slow, tight-binding inhibitor of FabI and binds to the E· NAD⁺ form of the enzyme (K₁ = 1.1 pM) 7-fold more tightly than triclosan. Thus, while the two ring B chlorine atoms are not required for FabI inhibition, replacement of the ring A chlorine increases binding affinity by 450 000-fold. Given this remarkable observation, the SAR study was extended to the 5-fluoro-2-phenoxyphenol (4) and 5-methyl-2-phenoxyphenol (5) analogues to further explore the role of the ring A substituent. While both 4 and 5 are slow, tight-binding inhibitors, they bind substantially less tightly to FabI than triclosan. Compound 4 binds to both E·NAD⁺ and E·NADH forms of the enzyme with K ₁ and K₂ values of 3.2 and 240 nM, respectively, whereas compound 5 binds exclusively to the E·NADH enzyme complex with a K ₂ value of 7.2 nM. Thus, the ring A substituent is absolutely required for slow, tight-binding inhibition. In addition, pKa measurements coupled with simple electrostatic calculations suggest that the interaction of the ring A substituent with F203 is a major factor in governing the affinity of analogues 3-5 for the FabI complex containing the oxidized form of the cofactor.

Full text of DOI:10.1021/jm030182i

J . Med. Chem. 2004, 47, 509-518
509
In h ibition of th e Ba cter ia l En oyl Red u cta se F a bI by Tr iclosa n : A
Str u ctu r e-Rea ctivity An a lysis of F a bI In h ibition by Tr iclosa n An a logu es†
Sharada Sivaraman,§ Todd J . Sullivan,§ Francis J ohnson,§,# Polina Novichenok,§ Guanglei Cui,§
Carlos Simmerling,§ and Peter J . Tonge*,§
Department of Chemistry, SUNY at Stony Brook, Stony Brook, New York 11794-3400, and
Department of Pharmacological Sciences, SUNY at Stony Brook, Stony Brook, New York 11794-8651
Received April 17, 2003
To explore the molecular basis for the picomolar affinity of triclosan for FabI, the enoyl reductase
enzyme from the type II fatty acid biosynthesis pathway in Escherichia coli, an SAR study
has been conducted using a series of triclosan analogues. Triclosan (1) is a slow, tight-binding
inhibitor of FabI, interacting specifically with the E‚NAD+ form of the enzyme with a K1 value
of 7 pM. In contrast, 2-phenoxyphenol (2) binds with equal affinity to the E‚NAD+ (K1 ) 0.5
µM) and E‚NADH (K2 ) 0.4 µM) forms of the enzyme and lacks the slow-binding step observed
for triclosan. Thus, removal of the three triclosan chlorine atoms reduces the affinity of the
inhibitor for FabI by 70 000-fold and removes the preference for the E‚NAD+ FabI complex.
5-Chloro-2-phenoxyphenol (3) is a slow, tight-binding inhibitor of FabI and binds to the E‚
NAD+ form of the enzyme (K1 ) 1.1 pM) 7-fold more tightly than triclosan. Thus, while the
two ring B chlorine atoms are not required for FabI inhibition, replacement of the ring A chlorine
increases binding affinity by 450 000-fold. Given this remarkable observation, the SAR study
was extended to the 5-fluoro-2-phenoxyphenol (4) and 5-methyl-2-phenoxyphenol (5) analogues
to further explore the role of the ring A substituent. While both 4 and 5 are slow, tight-binding
inhibitors, they bind substantially less tightly to FabI than triclosan. Compound 4 binds to
both E‚NAD+ and E‚NADH forms of the enzyme with K1 and K2 values of 3.2 and 240 nM,
respectively, whereas compound 5 binds exclusively to the E‚NADH enzyme complex with a
K2 value of 7.2 nM. Thus, the ring A substituent is absolutely required for slow, tight-binding
inhibition. In addition, pKa measurements coupled with simple electrostatic calculations suggest
that the interaction of the ring A substituent with F203 is a major factor in governing the
affinity of analogues 3-5 for the FabI complex containing the oxidized form of the cofactor.
Sch em e 1
In tr od u ction
The pathway for fatty acid biosynthesis (FAS) in
Escherichia coli is a paradigm for understanding type
II FAS systems in other bacteria and has been exten-
sively studied. The differences in protein sequences and
active site organization between the human and bacte-
rial FAS systems make it possible for specific inhibitors
to be designed against the bacterial FAS enzymes, and
thus this system has been validated as an excellent drug
target for the design of antimicrobial agents.1-3
FabI, the NADH-dependent trans-2-enoyl-ACP re-
ductase (ENR) in E. coli, is the key regulator of fatty
acid biosynthesis. ENRs have been shown to be inhib-
ited by diazaborines, isoniazid, triclosan, and more
recently by other new classes of novel inhibitors.4-7
Triclosan (Scheme 1) (5-chloro-2-(2,4-dichloro-phenoxy)-
phenol) is a broad-spectrum biocide that has been in use
for over 30 years, mainly as a component of antimicro-
bial wash products in health-care settings. More re-
cently, triclosan has found extensive use in consumer
products such as toothpaste, mouthwashes, deodorants,
hand soaps, and lotions. It is also incorporated in
children’s toys, cutting boards, and the plastic film used
to wrap meat products.8 Until recently, it was thought
that triclosan, being a small hydrophobic molecule, was
absorbed via diffusion into the bacterial cell wall and
that nonspecific disruption of the cell wall was the
mechanism by which triclosan exhibited its antibacterial
† This work was supported by NIH Grant AI44639 to P.J .T. P.J .T.
is an Alfred P. Sloan Research Fellow.
* To whom correspondence should be addressed. Telephone: (631)
632 7907. Fax: (631) 632 7960. E-mail: peter.tonge@sunysb.edu.
§ Department of Chemistry.
#
Department of Pharmacological Sciences.
10.1021/jm030182i CCC: $27.50 © 2004 American Chemical Society
Published on Web 01/06/2004

510 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Sivaraman et al.
activity9,10 However, the first evidence that triclosan
inhibits fatty acid biosynthesis came when a strain of
E. coli resistant to triclosan was isolated, and the
resistance was mapped to the fabI gene which codes for
the E. coli ENR.11 Subsequently, extensive biochemical
and structural studies have been performed to substan-
tiate triclosan as a specific E. coli FabI inhibitor.12-16
Two ENR isoforms, FabK and FabL, have been discov-
ered in the gram-positive bacteria, Streptococcus pneu-
moniae and Bacillus subtillis, respectively.17,18 FabK is
resistant to triclosan, whereas FabL is reversibly in-
hibited by triclosan. Triclosan also directly inhibits the
FabI from Staphylococcus aureus,19 Haemophilus influ-
enzae,20 the ENR from Mycobacterium tuberculosis,
InhA,21-23 and the ENR from Plasmodium falciparum,
the malarial parasite.24-26 A recent report also suggests
that triclosan inhibits enoyl reductase of type I fatty
acid synthase and may be a potential candidate for
breast cancer chemotherapy.27
Sch em e 2
reactivity study of diazaborine inhibition of the E. coli
FabI, Rafferty and co-workers have shown that the loop
adopts two distinct conformations for separate mono-
mers of the FabI-NAD+-diazaborine complex. The
reordering of the loop brings certain amino acids closer
to the fused rings of the inhibitor and introduces newer
contacts with the cofactor. Similar observations have
been made by examining the A138G mutant B. napus
plant ENR-NAD+-thienodiazaborine crystal struc-
ture.34 All the above data suggest that the flexibility of
the loop (residues 192-198) plays an important role in
inhibitor binding and substrate recognition.
The antibacterial activity of several 2-hydroxydiphe-
nyl ethers as well as hexachlorophene and 2-hydroxy-
diphenylmethanes have been determined.12,19,26 These
studies have revealed the importance of the hydroxyl
group and the bridging oxygen in triclosan inhibition
of the E. coli FabI. We recently reported a detailed
kinetic analysis of the triclosan inhibition of triclosan-
resistant and active-site mutant FabIs. We observed
that FabI mutations alter the relative affinity of tri-
closan for the E‚NADH and E‚NAD+ forms of the
enzyme, suggesting that the oxidation state of the
cofactor, as well as the electron density of triclosan, play
an important role in the inhibition of FabI by triclosan.30
Here we investigate the SAR for triclosan inhibition of
FabI to gain a better understanding of the contributions
of the enzyme-triclosan interactions that lead to potent
inhibition of this enzyme. We have made stepwise
changes in the triclosan structure (Scheme 1) and
probed the effect of these changes on the ability of the
inhibitor to bind to the wild-type enzyme. The present
SAR analysis combined with the studies on mutant
FabIs together with X-ray structural data provides
important information about the molecular basis of
triclosan inhibition.
The common theme in the inhibition of ENRs by
triclosan is the requirement of the NAD+ cofactor. In
most cases, triclosan forms a stable ternary complex
with the enzyme-cofactor complex as shown by several
X-ray structures of the FabI-NAD+-triclosan com-
plex.14,15,23,26,28,29 Unlike the diazaborines and isoniazid,
triclosan does not form a covalent adduct with the bound
NAD+ cofactor. The interaction of triclosan with FabI
is stabilized by the π-π stacking interaction between
the hydroxychlorophenyl ring (ring A) and the nicoti-
namide ring of the NAD cofactor. The 2′-hydroxyl group
of the NAD+ ribose ring and the hydroxyl group of Y156
form hydrogen bonding interactions with the hydroxyl
group of triclosan. Ring B makes several hydrophobic
contacts with FabI. The ether oxygen of triclosan may
also be critical in the formation of the stable FabI-
triclosan-NAD+ complex, since the replacement of this
group by a sulfur atom abolishes inhibitory activity.12
A diphenylamine triclosan analogue, which carries
nitrogen as the bridging atom, has been shown to adopt
a similar conformation when bound to the malarial
ENR.26 Most of the features unique to triclosan binding
to FabI are preserved in this structure, thus revealing
the importance of these interactions. Furthermore,
X-ray crystal structure data from the two new and
chemically distinct classes of FabI inhibitors bound to
FabI in the presence of NAD+ corroborate the common
themes in the binding of ENR inhibitors to FabI.4,5
Resu lts
The key feature of the inhibition of the E. coli FabI
by triclosan is the slow, tight-binding step, which has
been determined by detailed kinetic analysis.13,30 Unlike
classical inhibitors, that act almost instantaneously,
slow-binding inhibitors take several minutes, hours, or
even days for their inhibitory effects to take place. The
longer life of the EI complex in vivo makes them very
potent drug candidates.31 The slow onset of tight-
binding inhibition likely arises from a conformational
change that occurs following formation of the initial
FabI-triclosan complex. Although no direct evidence
has been obtained for this change, the only structural
perturbation that has been seen so far is the ordering
of the substrate binding loop upon triclosan binding.
Interestingly, the crystal structures of the E. coli FabI
with NAD+ and diazaborine compounds reveal a similar
ordering of the loop.32,33 Recently, in a structure-
Da ta An a lysis for Slow , Tigh t-Bin d in g In h ibi-
tion of F a bI: Revisitin g th e In h ibition of Wild -
Typ e F a bI by Tr iclosa n . The method used to quan-
titate the slow, tight-binding inhibition of FabI by
triclosan is based on the procedure developed by Ward
et al.13 In the present study, we have modified the data
analysis described by Ward et al. and used by us in a
previous publication. The principal differences between
the present approach and that used previously are (i)
our equations describe the general case for Scheme 2
and so separate equations need not be used depending
on which cofactor is varied and (ii) no correction factors
are required for the values of K1 or K2 obtained from
data fitting. When eq 2 is used to fit data for triclosan
binding to wild-type enzyme, values of 7 ( 1 pM and
>1 M are obtained for K1 and K2, respectively, consistent

SAR Study for Triclosan Inhibition of E. coli FabI
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 511
Ta ble 1. Data for Triclosan Analogues: pKa, Minimum Inhibitory Constants Toward E. coli, and Inhibition Constants for Binding to
Wild-Type FabI
FabI inhibition
compound
pKa
MIC E. coli (µg/mL)a
K1
K2
binding preference
E‚NAD+ d
triclosan (1)
7.8 ( 0.1
0.3
7 ( 1 pMc
8.4 ( 0.1b
9.12 ( 0.03
8.13 ( 0.02
8.12 ( 0.06
2
3
4
3.7
0.07
0.6
0.50 ( 0.02 µM
1.1 ( 0.1 pM
3.2 ( 0.4 nM
68 ( 8 pM
0.40 ( 0.01 µM
NC with respect to DDCoAe
E‚NAD+ d
42 ( 1 nM
E‚NAD+ and E‚NADHd
E‚NAD+ and E‚NADHd
E‚NADHd
240 ( 40 nM
7.2 ( 0.2 nM
9.7 ( 0.4 nM
5
9.15 ( 0.09
1.0
E‚NADHd
6
7
8
NDf
NDf
NDf
2.0
>40
>100
0.15 ( 0.01 µM
53 ( 8 µM
1.7 mM
UC with respect to DDCoAg
NC with respect to DDCoAe
NC with respect to DDCoAe
30 ( 5 µM
1.9 mM
a
b
The experiment was repeated twice for each analogue and the values reported are average of two readings. pKa for triclosan bound
to FabI.30 c Data from Sivaraman et al.30 Slow binding inhibition. e Noncompetitive inhibition. Kii and Kis have been listed under K1
d
and K2, respectively. f ND: not determined. Uncompetitive inhibition. Kii has been listed under K1.
g
F igu r e 1. The effect of NAD+ on the apparent inhibition
constant for 3 binding to wild-type FabI. The data have been
fitted to eq 3 with K1 ) 1.1 ( 0.1 pM.
F igu r e 2. Triclosan analogue 4 inhibition of FabI. (A) The
effect of NAD+ on the apparent inhibition constant. The data
have been fitted to eq 2 to give K1 ) 3.2 ( 0.4 nM and K2 ) 42
( 1 nM. (B) The effect of NADH on the apparent inhibition
constant for binding of 4 to FabI. The data have been fitted to
eq 2 to give K1 ) 68 ( 8 pM and K2 ) 240 ( 40 nM.
with the knowledge that triclosan binds exclusively to
the E‚NAD+ complex. In agreement with this expecta-
tion, subsequent reanalysis of these data using eq 3
provided a K1 value of 7 ( 1 pM. This new value of K1
is approximately 3-fold smaller than the value of 23 (
1 pM determined previously,30 and results from the use
of KdNADH ) 5.6 µM13 in place of KmNADH ) 20 µM30 to
estimate the fraction of enzyme present as the E‚NADH
complex in the incubation mixtures prior to the addition
of DDCoA. Since the preincubations of enzyme, cofactor,
and inhibitor are performed in the absence of the enoyl-
CoA substrate, KdNADH must be used rather than
E‚NAD+ form of FabI, the experiment was repeated
varying [NADH], while keeping [NAD+] constant. In
Figure 2B, it can be seen that the observed K′ values
i
increased as a function of [NADH], again consistent
with a model in which the analogue is bound to both
cofactor-bound forms of the enzyme with a preference
for E‚NAD+. When these data were analyzed using eq
2, values of 68 ( 8 pM and 240 ( 40 nM were obtained
for K1 and K2, respectively.
The value of K1 obtained in the varied NADH experi-
ment with analogue 4 is significantly smaller (50-fold)
than K1 determined in the varied NAD+ experiment,
while there is a 6-fold difference in the K2 values
between the two experiments. Although the use of eq 2
enables both K1 and K2 to be extracted from a single
experiment, it is likely that greater reliability can be
attached to the dissociation constant describing inhibi-
tor binding to the form of the enzyme for which the
cofactor was being varied. In the varied NAD+ experi-
ment, this is K1 (3.2 nM), while in the varied NADH
experiment this is K2 (240 nM). In addition, since the
experimental design was established to quantitate the
binding of triclosan to FabI where the interaction is
exclusively with E‚NAD+, [NADH] in the varied NAD+
experiment is present at a much greater concentration
than its Kd, while [NAD+] in the varied NADH experi-
ment is present at a much lower concentration than its
Kd. Thus, one possible explanation for the variation in
K1 and K2 between the experiments with analogue 4
concerns the experimental precision of the dissociation
constants calculated in each experiment. However, a
second explanation is that the analogue is also binding
K
mNADH. We note that this does not affect Ward’s
analysis in which crotonyl-CoA was used as the sub-
strate since KdNADH ) KmNADH ) 5.6 µM.13
In h ibition of Wild -Typ e F a bI by Tr iclosa n An a -
logu es 3-5. Table 1 lists the inhibition constants
obtained for triclosan analogues binding to the wild-type
FabI enzyme. Analogues 3-5 are all slow-binding
inhibitors of the enzyme. Equation 2 was used to fit the
data for analogue 3 obtained when [NAD+] was varied
at saturating [NADH] giving a K1 value of 1.1 ( 0.1 pM
and K2 > 1 M. Reanalysis of these data using eq 3
provided a K1 value of 1.1 ( 0.1 pM (Figure 1).
Consequently, analogue 3 binds preferentially to the E‚
NAD+ form of the enzyme with an affinity that is
around 7-fold higher than that seen for triclosan (K1 )
7 pM).
For analogue 4, Ki′ decreased slightly as [NAD+]
increased (Figure 2A) suggesting that this analogue
bound to both E‚NAD+ and E‚NADH with a preference
for E‚NAD+. Data fitting using eq 2 when [NAD+] was
varied at a constant [NADH] gave K1 and K2 values of
3.2 ( 0.4 and 42 ( 1 nM, respectively (Figure 2A). To
confirm that analogue 4 had a preference for the

512 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Sivaraman et al.
Ta ble 2. Calculated Interaction Energiesa between the
Triclosan Analogues and NAD+
calculated
interaction
energy kcal/mol
experimental
binding energy
for E‚NAD+b
compound
triclosan (1)
-18.19
-15.77
-17.32
-16.69
-16.35
-15.46
-8.73
-16.56
-11.76
-7.20c
2
3
4
5
F igu r e 3. Triclosan analogue 5 inhibition of FabI. (A) The
effect of NAD+ on the apparent inhibition constant. The data
have been fitted to eq 4 to give with K2 ) 7.2 ( 0.2 nM. (B)
The effect of NADH on the apparent inhibition constant for
binding of 5 to FabI. The data have been fitted to eq 4 to give
K2 ) 9.7 ( 0.4 nM.
a
Interaction energies between the triclosan analogues and
NAD+ were calculated with a molecular mechanics approach using
the AMBER 7 modeling package and the parm99 force field.
Calculated from K1 in Table 1. c Since binding to E‚NAD+ is not
b
observed for analogue 5, it was assumed that K1 is at least 100-
fold larger than K2.
to the free enzyme so that an E‚I inhibited complex is
present in addition to the E‚NADH‚I and E‚NAD+‚I
complexes. To account for the presence of E‚I in the data
analysis, an additional term 1/K3 must be added to the
two terms in the denominator to eq 2, where K3 is the
dissociation constant of the analogue for free enzyme.
Iterative fitting of the data provided estimates of K1 )
3.2 ( 0.3, K2 ) 100 ( 400, and K3 ) 2 ( 4 nM when
[NAD+] was varied and K1 ) 6 ( 4000, K2 ) 240 ( 50,
and K3 ) 1.5 ( 0.2 nM when [NADH] was varied. Thus,
while K1 in the varied NAD experiment and K2 in the
varied NADH experiment have similar, well-defined
values to those determined by eq 2 above, the inclusion
of a term to account for binding of analogue 4 to free
enzyme with a dissociation constant of 1-2 nM has
improved the agreement between these values and K2
in the varied NAD+ experiment and K1 when NADH
was varied, but with a substantial loss of precision. The
observation that inclusion of K3 substantially increases
the error in K1 and K2 therefore argues against the
possibility that the inhibitor might also bind to the free
enzyme.
case of analogues 7 and 8, the plots of v versus
v/[DDCoA] gave parallel lines indicating noncompetitive
inhibition with Kii and Kis values of 53 and 30 µM,
respectively, for analogue 7 and 1.7 and 1.9 mM,
respectively, for analogue 8 (data not shown).
Deter m in a tion of p Ka Va lu es for Tr iclosa n a n d
An a logu es 2-5. Table 1 gives the pKa values deter-
mined for triclosan and analogues 2-5. The pKa of 7.8
for triclosan matches well with the pKa value reported
for this compound in the literature.13 When bound to
FabI, the pKa for triclosan is increased to 8.4. Compound
2, which is the triclosan analogue without chlorines, has
a pKa of 9.12 which is 1.3 units higher than triclosan.
Compounds 3 and 4 have identical pKa values of 8.13
and 8.12, which are each a unit lower than that of 2
and consistent with the inductively electron withdraw-
ing effect of the meta substitution with respect to the
hydroxyl group of 2. Analogue 5, which has a methyl
group meta to the hydroxyl group, has a higher pKa
value of 9.15.
Ca lcu la ted In ter a ction En er gies Betw een Ea ch
An a logu e a n d NAD+. The interaction between tri-
closan analogues and NAD+ was calculated with a
molecular mechanics approach using the AMBER 7
modeling package and the parm99 force field. The
results of these calculations are given in Table 2.
Triclosan gave the most favorable interaction energy
with a value of -18.19 kcal/mol, while analogue 2, in
which all the chlorine atoms have been removed, gave
the least favorable interaction energy with a value of
-15.77 kcal/mol.
Inhibition by analogue 5 showed exclusive binding
preference to E‚NADH as seen by the data which fit well
with eq 4 yielding a K2 value of 7.2 ( 0.2 nM (Figure
3A). The selectivity of analogue 5 for binding to E‚
NADH was verified in experiments in which NADH was
varied and a K2 value of 9.7 ( 0.4 nM was obtained
(Figure 3B).
In h ibition of Wild -Typ e F a bI by Tr iclosa n An a -
logu es 2 a n d 6-8. In contrast to analogues 3-5,
analogue 2, which is the triclosan analogue without any
chlorine atoms, is a classical reversible inhibitor of the
wild-type enzyme. Plots of v versus v/[DDCoA] gave
parallel lines (data not shown), indicating noncompeti-
tive inhibition of FabI by 2, and data fitting using eq 5
gave Kis and Kii values of 0.5 and 0.4 µM, respectively.
On the basis of our knowledge of how triclosan binds to
FabI and the related enzyme from M. tuberculo-
sis,13,22,28,30 we assign Kii to the dissociation constant for
binding to the E‚NAD+ complex and Kis to the dissocia-
tion constant for binding to E‚NADH. Thus, removal of
all three triclosan chlorine atoms has reduced the
affinity of the enzyme for the inhibitor by over 70 000-
fold and removed the preference for the E‚NAD+ com-
plex. Analogues 6-8 were also classical reversible
inhibitors of wild-type FabI. Plots of v versus v/[DDCoA]
for each concentration of analogue 6 intersected on the
x-axis, indicating uncompetitive inhibition with respect
to DDCoA with Kii ) 0.15 µM (data not shown). In the
Deter m in a tion of Wh ole Cell An tim icr obia l Ac-
tivity. Table 1 gives the MIC values for triclosan and
analogues 2-8. They range from 0.07 to 100 µg/mL.
Discu ssion
Tr iclosa n a n d 2-P h en oxyp h en ol (2). Triclosan is
a slow, tight-binding picomolar inhibitor of the wild-
type E. coli FabI (K1 ) 7 pM) and binds preferentially
to the E‚NAD+ form of the enzyme. To investigate the
molecular basis for the high affinity of triclosan for FabI,
we have performed a structure-activity analysis by
making systematic changes to the triclosan diphenyl
ether skeleton and examining the impact of these
changes on enzyme inhibition. Inspection of the FabI-
NAD+-triclosan crystal structure reveals several po-
tentially important interactions between the inhibitor
and the enzyme including hydrogen bonds between the
ring A hydroxyl and the hydroxyl groups of Y156 (2.7

SAR Study for Triclosan Inhibition of E. coli FabI
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 513
Å) as well as the ribose sugar of NAD+ (2.60 Å), an
interaction between F203 and the ring A chlorine and
a stacking interaction between ring A and the nicoti-
namide ring of NAD+.15 The importance of the interac-
tions involving Y156 and F203 have previously been
examined by analyzing the inhibition of the Y156F and
F203L mutants by triclosan. For both mutants, triclosan
binds to both the E‚NAD+ and E‚NADH forms of FabI,
albeit with a continued preference for E‚NAD+, and
affinity of the inhibitor for the enzyme is reduced 100-
(F203L) to 400- (Y156F) fold.30 The data for Y156F are
consistent with other studies in which it has been shown
that removal of the triclosan ring A hydroxyl decreases
affinity by more than 10 000-fold.13
placement of this residue with Leu results in a 100-fold
decrease in affinity.15,30 Consequently, the interaction
of the ring A chlorine with F203 is pivotal in controlling
the affinity of the analogue for the enzyme.
The observation that the deletion of the two chlorine
atoms on ring B increases the affinity 7-fold is signifi-
cant and suggests that the ring B chlorine atoms
participate in unfavorable steric interactions with the
enzyme. Consistent with this observation, the MIC
value for analogue 3 (0.07 µg/mL) is around 4-fold lower
than that of triclosan (0.3 µg/mL), supporting the
contention that the antibacterial activity of triclosan
results primarily from an inhibition of FabI. To probe
the role of ring B further, we studied the inhibition of
the enzyme by 3-chloro-benzene-1, 2-diol, a triclosan
analogue which has a hydroxyl group in the place of ring
B. While we did not complete a full kinetic analysis for
3-chloro-benzene-1,2-diol, we observed that at 180 µM
only 25% of the enzyme activity was inhibited (data not
shown). Thus, the replacement of ring B by a hydroxyl
group decreases affinity by more than 106-fold. Consis-
tent with this large decrease in affinity for FabI,
3-chloro-benzene-1,2-diol had no effect on cell growth
up to a concentration of 100 µg/mL. These SAR data
indicate that whereas the ring B chlorine atoms are not
important for enzyme inhibition or biological activity,
ring B itself must be retained.
We initiated our SAR study by examining the binding
of analogue 2, which lacks the three triclosan chlorine
atoms. Significantly, analogue 2 binds 70 000-fold less
tightly to FabI than does triclosan. In addition, unlike
triclosan, analogue 2 is not a slow-binding inhibitor and
also binds to both the E‚NAD+ and E‚NADH forms of
FabI. Thus, removal of all three chlorine atoms from
triclosan has substantially reduced the affinity of the
inhibitor for the enzyme and also removed the prefer-
ence for the E‚NAD+ enzyme complex. It has been
proposed that the structure of the enolate anion inter-
mediate of the substrate is closely mimicked by the
phenolic ring of triclosan, leading to the proposal that
triclosan would be bound as the anion to FabI.14
Consequently, the decrease in affinity observed for
analogue 2 compared to triclosan could in part be due
to an alteration in the ionization state of the bound
analogue since the pKa values for 2 and triclosan are
9.12 and 7.8, respectively (Table 1). However, we
determined the pKa of triclosan bound to FabI and
observed that the triclosan pKa increases from 7.8 for
free inhibitor to 8.4 when bound to the enzyme. Thus,
triclosan is bound to FabI as the neutral species rather
than the anion. Since the free pKa of 2 is more basic
than triclosan, we expect analogue 2 to also be bound
as the neutral species to FabI. Thus, the decrease in
affinity observed for 2 compared to triclosan likely does
not result from an alteration in the ionization state of
the analogue. Alteration of the phenolic hydroxyl pKa
value could alter the strength of hydrogen bonding
interactions involving this group, however this effect is
likely to be small compared to the 70 000-fold decrease
in affinity of 2 relative to triclosan.
5-F lu or o-2-P h en oxyp h en ol (4) a n d 5-Meth yl-2-
P h en oxyp h en ol (5). The remarkable observation that
replacement of the ring A chlorine on 2-phenoxyphenol
resulted in the reappearance of slow, tight-binding
inhibition, and a 450 000-fold increase in affinity,
indicated that the presence of a substituent at this
position was critical for binding. To probe this effect in
more detail, we extended our SAR study to include two
analogues in which the ring A chlorine atom had been
replaced with a fluorine (4) or a methyl (5) substituent.
Both analogues 4 and 5 are slow, tight-binding
inhibitors of FabI, thus confirming that a substituent
at the 5-position on ring A is critical for the slow binding
step observed in the inhibition of FabI by the phenoxy-
phenols. However, the affinity of both these analogues
for FabI is significantly reduced compared to analogue
3 and triclosan. Replacement of the chlorine in 3 with
a fluorine (4) reduces the affinity of the inhibitor for the
E‚NAD+ FabI complex 3000-fold. In addition, binding
of 4 to the E‚NADH form of the enzyme can now also
be observed, albeit 10- to 100-fold less tightly than to
the oxidized cofactor complex. While the pKa values for
3 and 4 are similar, the van der Waals radius of fluorine
is 0.5 Å smaller than that of chlorine. Thus, the
reduction in analogue affinity of 4 compared to 3 is most
likely due to alterations in the interaction of the ring A
substituent with F203. If this effect is primarily steric
in origin, then it would be predicted that analogue 5
would bind to FabI more tightly than 4, since the van
der Waals radius of the methyl substituent in 5 differs
from that of chlorine by only 0.2 Å (2 and 1.8 Å,
respectively). However, no interaction of 5 with the E‚
NAD+ complex can be detected. Instead, 5 binds exclu-
sively to the E‚NADH form of FabI with 6000-fold lower
affinity than 3 binding to the E‚NAD+ FabI complex.
Since the affinities of 4 and 5 for the E‚NADH FabI
complex only differ by ca. 20-30-fold, the major effect
5-Ch lor o-2-P h en oxyp h en ol (3). To understand the
relative importance of the chlorine atoms, we next
examined analogue 3 which has a chlorine atom at the
5-position meta with respect to the hydroxyl group on
ring A. We predicted that the inhibition constant for the
binding of 3 to FabI would be the ultimate test for the
contribution of ring A to inhibition since it is the same
as in the triclosan structure. In addition, since ring B
points away from the active site and makes few hydro-
phobic contacts with the protein,15 this analogue would
enable us to evaluate the contribution of ring B toward
binding. Analogue 3 is a slow, tight-binding inhibitor
of the enzyme, binding preferentially to E‚NAD+ with
a K1 value of 1.1 pM. Remarkably, restoring the chlorine
atom to ring A increases the affinity 450 000-fold. As
noted above, the ring A chlorine forms an edge-on π
interaction with the aromatic ring of F203 while re-

514 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Sivaraman et al.
on replacing the ring A chlorine with fluorine and
methyl substituents has been to drastically alter the
affinity of the inhibitor for the E‚NAD+ form of FabI.
In addition, both analogue 3 as well as triclosan may
also bind to the E‚NADH complex with nanomolar
affinity. However, since these compounds bind to E‚
NAD+ with picomolar affinity, then an interaction with
E‚NADH would be masked.
The MIC values of 4 and 5 are raised 2-3-fold
compared to triclosan, indicating that these analogues,
despite binding several orders of magnitude less tightly
to FabI than triclosan, have similar abilities to inhibit
cell growth. This could arise if the potency of each
inhibitor with regard to their MICs is governed by the
concentration of target in the cell. For example, if the
concentration of FabI in the cell is 1 µM, then at least
1 µM analogue will be needed to completely inhibit the
enzyme. At these concentrations of enzyme and inhibi-
tor, inhibition constants of 10, 1, and 0.1 nM will give
90, 97, and 99% enzyme inhibition, respectively.
FabIs by triclosan that whereas triclosan binds prefer-
entially to the E‚NAD+ form of the wild-type enzyme,
mutagenesis alters the relative affinity of triclosan for
the E‚NAD+ and E‚NADH forms of the enzyme.30 This
phenomenon was also observed in the inhibition of the
M. tuberculosis ENR (InhA) by triclosan22 and is again
highlighted in the present studies. Two points can be
made from the current analysis. First, alteration in the
relative affinity for E‚NAD+ and E‚NADH is observed
for both slow, tight-binding inhibitors (1, 3-5) and also
for the classical reversible inhibitors (2, 6-8). Conse-
quently, whatever structural change(s) is responsible for
slow-binding inhibition, it is not directly linked to
whether the inhibitor binds preferentially to E‚NAD+
or E‚NADH. Second, our SAR study demonstrates that
subtle changes in the electronic and steric properties
of ring A (3-5) have a dramatic effect on the affinity of
the inhibitor for the enzyme and that these effects are
dominated by alterations in affinity for E‚NAD+, while
the affinity of the analogues for E‚NADH remains
largely unchanged. Of course, the ability to detect
binding to either or both cofactor-bound forms of the
enzyme is dictated by the relative affinities of the
analogues for these two enzyme complexes. Thus, 3
binds only to E‚NAD+ (K1 ) 1.1 pM), 4 binds to both
E‚NAD+ (K1 ) 3.2 nM) and E‚NADH (K2 ) 240 nM),
while 5 only binds to E‚NADH (K2 ) 7.2 nM). For the
interaction of 5 with E‚NAD+ to be masked by binding
to E‚NADH, we estimate that K1 (binding to E‚NAD+)
must be at least 2 orders of magnitude larger than the
value of K2 (7.2 nM) observed for the interaction of 5
with E‚NADH. Thus, replacement of the ring A chlorine
with fluorine decreases the affinity of the analogue for
E‚NAD+ by 4.8 kcal/mol, while replacement with a
methyl group decreases the affinity by at least 8 kcal/
mol (K1 estimated as 0.7 µM for analogue 5).
The relative affinity of analogues 3-5 for the E‚NAD+
FabI complex is likely governed by modulation of the
three principal interactions that exist in the complex
with ring A, namely, the interaction of the 5-chloro
group with F203, hydrogen bonding interactions involv-
ing the hydroxyl group and the stacking interaction
between ring A and the oxidized cofactor. In an attempt
to gain insight into the importance of the electrostatic
interaction between the aromatic ring A and the oxi-
dized nicotinamide ring, we performed a simple elec-
trostatic calculation, the results of which are given in
Table 2. These experiments revealed that 3 interacts
more strongly with NAD+ by 0.63 kcal/mol compared
to 4, 1 kcal/mol compared to 5 and 1.55 kcal/mol
compared to 2. While the calculated energies largely
follow the observed affinities of the analogues for FabI,
the differences between the experimental binding ener-
gies for E‚NAD+ vary by much larger amounts (4.8-
9.4 kcal/mol; Table 2). Differences between the calcu-
lated and experimental energies will be present since
the calculations do not account for changes in desolva-
tion energies nor differences in binding mode or con-
formational changes in either the triclosan analogues,
NAD+ or FabI. Given these approximations, however,
the computational data suggest that changes in the
stacking interaction between ring A and NAD+ resulting
from differences in electron donation/withdrawal at the
5-position are small compared to the overall alteration
2-(2-Hyd r oxyp h en yl)-p h en ol (6), 2-(2-h yd r oxy-
ben zyl)-p h en ol (7), a n d 4-(4-h yd r oxyp h en yl)-p h e-
n ol (8). Analogue 6, differing from 2-phenoxyphenol (2)
by the addition of a 2-hydroxy group to ring B, is a
classical, reversible inhibitor of the FabI enzyme and
shows uncompetitive inhibition with respect to DDCoA
(Kii ) 0.15 µM), indicating that it binds after the varied
substrate. Analogue 6 therefore binds with similar
affinity to FabI compared to 2. The major difference
between these two analogues is that 6 is an uncompeti-
tive inhibitor while 2 is a noncompetitive inhibitor.
Uncompetitive inhibition is assumed to result from
binding only to the E‚NAD+ form of FabI while non-
competitive inhibition results from binding to both E‚
NAD+ and E‚NADH.22 Thus, addition of the second
hydroxyl to 2-phenoxyphenol results in an altered
preference for the two cofactor-bound forms of the
enzyme, a theme that runs through interaction of the
ENRs with triclosan analogues.22,30 In addition, the MIC
value for 6 is 7-fold higher compared to triclosan,
consistent with the reduced affinity of this analogue for
FabI.
To examine the role of the bridging oxygen in FabI
binding and the inhibition of cell growth, this atom was
replaced by a methylene group (7). Previous studies
have indicated that such a replacement would impair
binding to the enzyme.19 Our data show that the affinity
for the enzyme is reduced more than 106-fold compared
to triclosan and 200-fold compared to 6. In addition the
MIC is raised more than 20-fold compared to 6. In
agreement with previous studies,12 these data clearly
show that the bridging oxygen is critical for enzyme
inhibition and biological activity. The substantial dif-
ference in the interaction of 7 with FabI compared to 6,
likely results primarily from removal of the hydrogen
bonding interaction between the bridging oxygen and
the 2′-hydroxyl group of the nicotinamide ribose ring.
Finally, comparison of 6 and analogue 8, in which the
two hydroxyl groups are now para to the bridging
oxygen, substantiates the requirement for an ortho
hydroxyl group in the interaction of the inhibitors with
FabI and in preventing cell growth.
In ter a ction of th e In h ibitor s w ith E‚NAD+ a n d
E‚NADH. We concluded from the inhibition of mutant

SAR Study for Triclosan Inhibition of E. coli FabI
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 515
Sch em e 3a
a
Conditions: (I) Cs2CO3, (CuOTf)2‚PhH, 5 mol % EtoAc,
1-naphthoic acid, molecular sieves, toluene, 110 °C; (II) dry CH2Cl2,
-70 °C, 1 M BBr3, RT.
5-chloro group reduces binding 450 000-fold. In addition,
subtle changes in both the steric and electronic proper-
ties of the ring A substituent have dramatic effects on
affinity and the relative preference for the two cofactor-
bound forms of the enzyme. The position of the ring A
hydroxyl and the presence of a bridging oxygen atom
are also critical for maximizing the interaction of
triclosan with FabI and controlling biological activity.
Exp er im en ta l P r oced u r es
F igu r e 4. Visual representation of the electron density in the
ring A of analogues 2-5. The calculated electrostatic potential
of each analogue has been mapped to its size-density surface
(electron density contour value ) 0.002 au). A second electron
density surface (contour value ) 0.1 au) is shown to facilitate
differentiation of the four ligands.
Gen er a l. Triclosan (Irgasan DP 300) was a gift from Ciba
Specialty Chemicals Corp (High Point, NC). Compounds 6-8
were purchased from TCI America (Portland, OR). Solvents
and reagents were used without further purification. The
progress of all reactions was monitored by TLC on precoated
silica gel plates (Merck Silica gel 60 F254). The developed
chromatograms were viewed under UV light at 254 nm. For
column chromatography Merck silica gel (70-230 mesh) was
in affinity of these analogues for FabI. The small
changes in the calculated stacking interactions through
the analogue series reflect the relatively modest changes
in electron density between analogues 2-5. This is
shown visually in Figure 4, where the calculated
electrostatic potential of each analogue has been mapped
to its size-density surface. In addition, our pKa mea-
surements indicate that all three analogues are neutral
when bound to FabI and thus the large changes in
affinity through the analogue series also likely does not
arise from alteration in hydrogen bonding interactions
involving the ring A hydroxyl group. These data there-
fore suggest that the interaction of the ring A substitu-
ent with F203 is a major determinant in controlling the
affinity of the analogues for the E‚NAD+ FabI complex.
Inspection of Figure 4 supports this hypothesis, as the
largest differences in electron densities are localized to
the 5 substituent itself. Further insight into the interac-
tion of the triclosan analogues with FabI will be
provided by a more rigorous computational analysis in
which binding free energies will be calculated using free
energy perturbation formalism coupled with molecular
dynamics simulations of the protein-inhibitor complex.
1
used. H and 13C NMR spectra were obtained in CDCl3 using
Varian spectrometers at the indicated field strengths. Chemi-
cal shifts are reported as ppm relative to TMS as an internal
standard. Mass spectra were obtained at the UIUC mass
spectrometry facility.
Syn th esis of Tr iclosa n An a logu es (2-5). The 5-substi-
tuted phenoxyphenols were prepared by coupling the appropri-
ate 4-substituted guaiacol with iodobenzene in the presence
of a copper complex35 followed by treatment with BBr3 at low
temperatures (Scheme 3).36
Gen er a l Meth od s. P r oced u r e I. The aryl halide (7.35
mmol), phenol (14.7 mmol), Cs2CO3 (32.3 mmol), (CuOTf)2‚PhH
(0.735 mmol, 5.0 mol % Cu), ethyl acetate (0.125 mmol, 5.0
mol %), 1-naphthoic acid (32.3 mmol), molecular sieves 4 Å
(1.8 g), and toluene (15 mL) were added to an oven-dried 50
mL two-necked round-bottom flask sealed with a septum and
purged with nitrogen. Subsequently, the flask was heated to
110 °C under nitrogen until the reaction was shown to be
complete by TLC. After being cooled to RT, the reaction
mixture was extracted with dichloromethane and the extract
filtered and washed with 5% NaOH. The aqueous layer was
then extracted with dichloromethane, and the combined
organic layers were washed with brine. The organic layer was
dried over MgSO4 and concentrated under vacuum to give the
crude 4-substituted-2-methoxy-1-phenoxybenzene product which
was subsequently purified by flash chromatography on silica
gel.
P r oced u r e II. A solution of 455 mg (1.8 mmol) of boron
tribromide in 1.8 mL of dichloromethane (1.0 M solution) was
added to a solution of the 4-substituted-2-methoxy-1-phenoxy-
benzene (1.4 mmol) in 3 mL of dry dichloromethane main-
tained at -70 °C under nitrogen. The reaction mixture was
stirred at -70 °C for 1 h and then at RT for 3 h while
monitoring by TLC. Following completion, the reaction was
quenched with methanol at -70 °C and concentrated under
vacuum to yield an oil. A solution of this oil in dichloromethane
was washed with 10% aqueous sodium bicarbonate solution,
and the organic layer was separated and washed with water
and then with brine. The organic layer was then dried over
MgSO4, concentrated under vacuum, and the crude product
purified by flash chromatography on silica gel.
Con clu sion
The SAR of triclosan analogue inhibitors with the
wild-type FabI enzyme provides valuable information,
which, when coupled to the analyses of inhibitor inter-
actions with the FabI mutants reported previously30 and
to the X-ray crystal structure data,15 significantly
advances our understanding of ENR inhibition by
triclosan. Triclosan is a slow, tight-binding, reversible
picomolar inhibitor of the wild-type E. coli FabI, binding
preferentially to the E‚NAD+ complex. The 5-chloro
substituent on ring A is crucial for tight-binding and
essential for slow-binding inhibition. Removal of the

516 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Sivaraman et al.
1-Meth oxy-2-p h en oxyben zen e. Procedure I was used to
convert 2-methoxyphenol and iodobenzene to the target prod-
uct. Purification by flash chromatography (5% ethyl acetate/
hexane) gave the analytically pure product as a yellow solid
(4.4 g, 79%). 1H NMR (250 MHz) (CDCl3) δ 7.32-7.25 (m,3H),
7.10-6.88 (m,6H), 3.83 (s,3H); 13C NMR (250 MHz, CDCl3) δ
157.92, 151.44, 145.05, 129.46, 124.75, 122.42, 121.08, 121.05,
117.18, 112.80, 55.94; HRMS [M+] Calculated for C13H12O2:
200.0837; Found: 200.0836.
NMR (250 MHz, CDCl3) δ 157.16, 147.22, 140.74, 134.91,
129.71, 123.15, 121.08, 119.09, 117.37, 116.69; HRMS [M+]
Calculated for C13H12O2: 200.0837; Found: 200.0835.
P r ep a r a t ion of Tr iclosa n a n d Tr iclosa n An a logu e
Solu tion s. Triclosan and analogues were added to the final
concentrations indicated from serially diluted stock solutions
in Me2SO. The Me2SO concentration in all the assays was
maintained at 0.5%. Control experiments demonstrated that
this concentration of Me2SO did not affect FabI activity.
Su bstr a tes a n d En zym es. trans-2-Dodecenoyl-coenzyme
A (DDCoA) was synthesized from trans-2-dodecenoic acid using
the mixed anhydride method as described previously.37 Wild-
type FabI was expressed and purified as described previously.30
Deter m in a tion of In h ibition Con sta n ts for th e Bin d -
in g of th e Tr iclosa n An a logu es to Wild -typ e F a bI. All
experiments were carried out on a Cary 100 Bio (Varian)
spectrophotometer at 25 °C in 30 mM PIPES and 150 mM
NaCl (pH 8.0). Kinetic parameters were determined spectro-
photometrically by following the oxidation of NADH to NAD+
at 340 (ꢀ ) 6.3 mM-1 cm-1) or 370 nm (ꢀ ) 2.4 mM-1 cm-1).
Slow , Tigh t-Bin d in g In h ibitor s. The inhibition of FabI
by analogues 3, 4, and 5 was performed using the procedure
developed by Ward et al.13 and previously used to quantitate
the binding of triclosan to wild-type and mutant FabIs.30 This
method was developed since triclosan is a slow, tight-binding
inhibitor of FabI and interacts specifically with the E‚NAD+
form of the enzyme. Consequently, since formation of the
enzyme-inhibitor complex occurs slowly with respect to
substrate reduction and since triclosan binds preferentially to
an enzyme-product complex, typical progress curve analysis
of slow, tight-binding inhibition cannot be used to analyze
inhibitor binding. Instead, the enzyme was preincubated with
triclosan and NAD+ in the presence of saturating NADH so
that only a small fraction of the enzyme was in the E‚NAD+
form, thus reducing the affinity of triclosan to measurable
levels.13,30 Following this protocol, wild-type FabI (7 nM) was
preincubated in the presence of a fixed concentration of NAD+
at an NADH concentration of 250 µM and with varying
concentrations of analogue (0, 0.03, 0.08, 0.1, 0.15, and 0.2 µM)
for 5 h at 4 °C. The mixture was warmed to room temperature
and the assay was initiated by the addition of DDCoA (80 µΜ).
The initial velocity obtained, which was proportional to the
amount of active enzyme present in the assay, was fitted to
eq 1, where v0 was the rate in the absence of inhibitor and [I]
was the inhibitor concentration, to generate an apparent
2-P h en oxyp h en ol (2). Procedure II was used to convert
1-methoxy-2-phenoxybenzene to the target product. Purifica-
tion by flash chromatography (5% ethyl acetate/hexane) gave
1
the analytically pure product as a white solid (2.6 g, 73%). H
NMR (300 MHz) (CDCl3) δ 7.37-7.31 (m,2H), 7.14-7.01 (m,
5H), 6.90-6.83 (m,2H), 5.56 (s,1H); 13C NMR (250 MHz,
CDCl3) δ 156.75, 147.48, 143.45, 129.84, 124.74, 123.56, 120.60,
118.88, 117.96, 116.18; HRMS [M+] Calculated for C12H10O2:
186.0681; Found: 186.0683.
4-Ch lor o-2-m eth oxy-1-ph en oxyben zen e. Procedure I was
used to convert 4-chloro-2-methoxyphenol and iodobenzene to
the target compound. Purification by flash chromatography
(5% ethyl acetate/hexane) gave the analytically pure product
as a white solid (2.1 g, 53%). 1H NMR (250 MHz) (CDCl3) δ
7.35-6.94 (m, 8H), 3.87 (s,3H); 13C NMR (250 MHz, CDCl3) δ
157.55, 151.88, 143.73, 129.53, 122.69, 121.66, 120.79, 118.80,
116.40, 113.33, 56.06; HRMS [M+] Calculated for C13H11ClO2:
234.0448; Found: 234.0443.
5-Ch lor o-2-p h en oxyp h en ol (3). Procedure II was used to
convert 4-chloro-2-methoxy-1-phenoxybenzene to the target
compound. Purification by flash chromatography (5% ethyl
acetate/hexane) gave the analytically pure product as a white
1
solid (627 mg, 67%). H NMR (250 MHz) (CDCl3) δ 7.38-7.31
(m,2H), 7.16-7.1 (m,1H), 7.05-6.98 (m,3H), 6.80-6.79 (d,2H),
5.66 (s,1H); 13C NMR (250 MHz, CDCl3) δ 156.36, 148.01,
142.36, 129.98, 129.39, 123.97, 120.56, 119.39, 118.01, 116.55;
HRMS [M+] Calculated for C12H9ClO2: 220.0291; Found:
220.0293.
4-Flu or o-1-ph en oxy-2-m eth oxyben zen e. Procedure I was
used to convert 4-fluoro-2-methoxyphenol and iodobenzene to
the target compound. Purification by flash chromatography
(5% ethyl acetate/hexane) gave the analytically pure product
as a white solid (1.3 g, 83%). 1H NMR (250 MHz) (CDCl3) δ
7.36-7.23 (m,2H), 7.12-6.87 (m,4H), 6.76-6.58 (m,2H), 3.80
(s,3H); 13C NMR (250 MHz, CDCl3) δ 161.73, 158.03 d, 152.48
d, 140.63 d, 129.49, 122.30, 122.06 d, 116.45, 106.80 d, 100.97,
56.08; HRMS [M+] Calculated for C13H11FO2: 218.0743;
Found: 218.0745.
inhibition constant K′.
i
v ) v0/ (1 + [I]/K′)
(1)
i
5-F lu or o-2-p h en oxyp h en ol (4). Procedure II was used to
convert 4-fluoro-1-phenoxy-2-methoxybenzene to the target
compound. Purification by flash chromatography (5% ethyl
acetate/hexane) gave the analytically pure product as a white
solid (1.0 g, 92%). 1H NMR (600 MHz) (CDCl3) δ 7.35-7.32
(t,2H), 7.13-7.10 (t,1H), 6.99-6.98 (d,1H), 6.86-6.84 (m,1H),
6.80-6.78 (dd,1H), 6.58-6.54 (m.1H); 5.64 (s,1H); 13C NMR
(250 MHz, CDCl3) δ 161.47, 157.31 d, 148.54 d, 139.39 d,
129.94, 123.61, 119.86 d, 117.42, 107.01 d, 103.89 d; HRMS
[M+] Calculated for C12H9FO2: 204.0587; Found: 204.0589.
2-Meth oxy-4-m eth yl-p h en oxyben zen e. Procedure I was
used to convert 4-methyl-2-methoxyphenol and iodobenzene
to the target compound. Purification by flash chromatography
(5% ethyl acetate/hexane) gave the analytically pure product
as a white solid (847 mg, 53%). 1H NMR (250 MHz) (CDCl3) δ
7.29-7.23 (m,2H), 7.03-6.71 (m,6H), 3.80 (s,3H), 2.35 (s,3H);
13C NMR (250 MHz, CDCl3) δ 158.33, 151.23, 142.43, 134.84,
129.39, 122.08, 121.42,121.21, 116.72, 113.73, 55.91, 21.29;
HRMS [M+] Calculated for C14H14O2: 214.0994; Found:
214.0997.
The Ki′ determinations were repeated at different concentra-
tions of NAD+ (7.4-200 µM) to generate a series of K′ values
as a function of NAD+. For analogues 4 and 5, a seriies of K′
i
values determined as a function of NADH were also obtained
by repeating the inhibition experiments at a fixed concentra-
tion of NAD+ (50 µM) and varying the NADH concentration
from 6.8 µM to 1 mM (4) or from 6.8 to 350 µM (5).
Subsequently, the dependence of K′ on the concentration of
i
the varied cofactor is analyzed using eq 2.
K′i ) (1 + [NADH]/KdNADH + [NAD+]/KdNAD)/(([NADH]/
(KdNADHK2)) + ([NAD+]/(KdNADK1))) (2)
K1 and K2 in eq 2 are defined in Scheme 2 and represent the
inhibition constant for inhibitor binding to the E‚NAD+ and
E‚NADH forms of the enzyme, respectively. KdNADH and KdNAD
are the dissociation constants for NADH and NAD+ binding
to free enzyme, respectively. A value of 5.6 µM was used for
13
30
5-Meth yl-2-p h en oxyp h en ol (5). Procedure II was used to
convert 2-methoxy-4-methyl-phenoxybenzene to the target
compound. Purification by flash chromatography (5% ethyl
acetate/hexane) gave the analytically pure product as white
KdNADH, while a value of 1.1 mM was used for KdNAD.
Equation 2 represents the general case in which the
inhibitor binds to both cofactor-bound forms of the enzyme.
For the triclosan analogues in which data fitting returned a
large (>1 M) value for K2, it was concluded that the inhibitor
bound to E‚NAD+ but not to E‚NADH, and vice versa when
fitting gave K1 > 1 M. In such cases, to confirm that binding
1
solid (600 mg, 71%). H NMR (600 MHz) (CDCl3) δ 7.34-7.30
(t,2H), 7.10-7.09 (t,1H), 7.00-6.99 (d,2H), 6.87 (s,1H), 6.80-
6.78 (d,1H), 6.66-6.64 (d,1H), 5.43 (s,1H), 2.31 (s,3H); 13C

SAR Study for Triclosan Inhibition of E. coli FabI
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 517
to a particular cofactor-bound form of the enzyme was negli-
gible, data were reanalyzed using eqs 3 or 4, which assume
binding of inhibitor to only the E‚NAD+ or E‚NADH forms of
the enzyme, respectively.
triclosan to provide the spectrum of the ionized inhibitor.
Buffer solutions used were (pH 6.0-8.0), 30 mM PIPES, 150
mM NaCl; (pH 8.2-8.8), 30 mM Tris.HCl, 150 mM NaCl; (pH
9.0-10.0), 30 mM K2CO3, 150 mM NaCl.
Inhibitor binds only to E‚NAD+:
Ca lcu la ted In ter a ction En er gies Betw een An a logu es
1-5 a n d NAD+. The interaction between triclosan analogues
1-5 and NAD+ was calculated with a molecular mechanics
approach using the AMBER 7 modeling package and the
parm99 force field. The gas phase electron density and
electrostatic potential of each analogue was calculated quan-
tum mechanically with Gaussian 98 (HF/6-31G*). The partial
charges of each analogue were derived by a two-stage fitting
procedure (RESP) to the quantum mechanically calculated
electrostatic potential. All analogues were assumed to adopt
the same conformation and binding mode as observed in FabI/
NAD+ and triclosan tertiary complex. Each analogue was
subjected to local energy minimization with positional re-
straints on all atoms except the modified substituents. The
net interaction between analogues and NAD+, including van
der Waals and electrostatic energies, was then calculated
without constraints.
Deter m in a tion of Wh ole Cell An tim icr obia l Activity.
Whole-cell antimicrobial activity was determined by a broth
microdilution procedure. Stock solutions of triclosan and
analogues 2-5 were made in 1 M KOH. These solutions were
serially diluted into 50 mM Na2CO3 pH 8.5 buffer. The stock
solutions were serially diluted into cation-adjusted Mueller-
Hinton broth (Becton Dickenson, Cockeysville, MD) on a 96-
well microtiter plate. After dilution, 20 µL of E. coli (BL21
(DE3) pLysS) cells was added to each well of the microtiter
plate. The final test concentrations ranged from 0.01 to 64 µg/
mL. Inoculated plates were incubated at 37 °C for 24h. The
minimum inhibitory concentration (MIC) was determined as
the lowest concentration of compound that inhibited visible
growth as estimated by OD600 readings using a 96-well plate
reader.
K′ ) (1 + [NADH]/KdNADH + [NAD+]/KdNAD)/([NAD+]/
i
(KdNADK1)) (3)
Inhibitor binds only to E‚NADH:
K′i ) (1 + [NADH]/KdNADH + [NAD+]/KdNAD)/([NADH]/
(KdNADHK2)) (4)
In all of these cases, the fit to eqs 3 or 4 was comparable to
that found for eq 2 and returned essentially the same values
for the remaining variables.
Stea d y-Sta te In h ibition Stu d ies. For analogues 2 and
6-8, initial inhibition experiments demonstrated that these
compounds were not slow, tight-binding inhibitors. Conse-
quently, conventional steady-state inhibition experiments were
used to determine the affinity of these compounds for FabI,
as well as the type of inhibition. For analogue 2 the inhibition
constants were calculated by determining kcat and KmDDCOA at
a fixed, saturating concentration of NADH (250 µM) and by
varying the concentration of DDCoA (2.5-50 µM) and 2 (0,
0.26, 0.9 µM). For analogues 6-8, the inhibition constants were
determined in the same manner. Analogue concentrations used
were as follows: 6 (0, 0.1, 0.35 and 0.75 µM); 7 (0, 7, 14, and
70 µM); 8 (0, 484, 1635, 2920 µM). Each initial velocity was
determined in duplicate and at least five different DDCoA
concentrations were examined. Initial velocity data obtained
at various substrate concentrations in the presence of varying
amounts of an analogue were initially analyzed by plotting v
versus v/[S], to determine whether the inhibition was competi-
tive, uncompetitive, or noncompetitive with respect to the
varied substrate. For analogues 2, 7, and 8 the v versus v/[S]
plots indicated noncompetitive inhibition in which the inhibitor
binds both before and after the varied substrate. Consequently,
the data were analyzed using eq 5, where [S] is the concentra-
tion of the varied substrate, [I] is the concentration of the
inhibitor, V is the maximum velocity, and Km is the Michaelis-
Menten constant. By convention, Kis and Kii in eq 5 are the
dissociation constants for binding of the inhibitor to the
enzyme before and after the varied substrate, respectively. For
analogue 6, the v versus v/[S] plot indicated uncompetitive
inhibition in which the inhibitor binds after the varied
substrate, and the experimental data were analyzed using eq
6, where the parameters have the same meaning as above.
Ack n ow led gm en t. High resolution mass spectra
were obtained in the Mass Spectrometry Laboratory,
School of Chemical Sciences, University of Illinois,
supported in part by a grant from the National Institute
of General Medical Sciences (GM 27029). The authors
thank Dr. Adrian Whitty and Prof. Liz Hedstrom for
helpful discussions.
Ap p en d ix
Abbr evia tion s: FAS, fatty acid biosynthesis; KAS,
â-ketoacyl-ACP synthase; TLM, thiolactomycin; CER,
cerulenin; ENR, enoyl reductase; InhA, the enoyl re-
ductase from Mycobacterium tuberculosis; FabI, the
enoyl reductase from Escherichia coli; KasA, the â-ke-
toacyl-ACP synthase from M. tuberculosis; CoA, coen-
zyme A (lithium salt); DDCoA, trans-2-dodecenoyl-CoA.
vi ) (V([S]/Km))/(1 + [I]/Kis + [S]/Km + ([S][I]/Kii + Km))
(5)
vi ) (V[S]/([I]/Kii))/([S] + Km/(1 + [I]/Kii))
(6)
Deter m in a tion of p Ka Va lu es for Tr iclosa n a n d An a -
logu es 2-5. To determine the effect that binding has on the
electronic properties of the compounds, the pKa of triclosan
and analogues 2-5 were determined by the measurement of
the UV-vis absorbance of the conjugate base in several buffer
solutions at different pH values ranging from pH 6.8 to pH
10.8. Stock solutions of the compounds were made in DMSO
and diluted into buffer solutions such that the concentration
of DMSO was maintained at 0.4-0.5%. The absorbance values
were used to determine the concentration of the A- form and
a plot of fraction of the [A-] versus pH yielded the pKa value.
To determine the pKa of triclosan bound to FabI, UV-visible
absorption spectra were acquired at various pH values (6.1-
9.3) of solutions containing 18 µM FabI, 54 µM NAD+, and 18
µM triclosan that had been incubated for 5 h. Spectra of
triclosan bound to FabI were obtained by first subtracting a
spectrum of FabI and NAD+ from that of the ternary complex
and then interactively subtracting the spectrum of neutral
Refer en ces
(1) Heath, R. J .; White, S. W.; Rock, C. O. Lipid biosynthesis as a
target for antibacterial agents. Prog. Lipid Res. 2001, 40, 467-
497.
(2) 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.
(3) Campbell, J . W.; Cronan, J . E., J r. Bacterial fatty acid biosyn-
thesis: targets for antibacterial drug discovery. Annu. Rev.
Microbiol. 2001, 55, 305-332.
(4) Seefeld, M. A.; Miller, W. H.; Newlander, K. A.; Burgess, W. J .;
Payne, D. J . et al. Inhibitors of bacterial enoyl acyl carrier
protein reductase (FabI): 2,9-disubstituted 1,2,3,4-tetrahydro-
pyrido[3,4-b]indoles as potential antibacterial agents. Bioorg.
Med. Chem. Lett. 2001, 11, 2241-2244.
(5) Heerding, D. A.; Chan, G.; DeWolf, W. E.; Fosberry, A. P.;
J anson, C. A. et al. 1,4-Disubstituted imidazoles are potential
antibacterial agents functioning as inhibitors of enoyl acyl
carrier protein reductase (FabI). Bioorg. Med. Chem. Lett. 2001,
11, 2061-2065.

518 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Sivaraman et al.
(6) Miller, W. H.; Seefeld, M. A.; Newlander, K. A.; Uzinskas, I. N.;
Burgess, W. J . et al. Discovery of aminopyridine-based inhibitors
of bacterial enoyl-ACP reductase (FabI). J . Med. Chem. 2002,
45, 3246-3356.
(7) Payne, D. J .; Miller, W. H.; Berry, V.; Brosky, J .; Burgess, W. J .
et al. Discovery of a novel and potent class of FabI-directed
antibacterial agents. Antimicrob. Agents Chemother. 2002, 46,
3118-3124.
(8) Schweizer, H. P. Triclosan: a widely used biocide and its link
to antibiotics. FEMS Microbiol. Lett. 2001, 202, 1-7.
(9) Regos, J .; Zak, O.; Solf, R.; Vischer, W. A.; Weirich, E. G.
Antimicrobial spectrum of triclosan, a broad-spectrum antimi-
crobial agent for topical application. II. Comparison with some
other antimicrobial agents. Dermatologica 1979, 158, 72-79.
(10) Vischer, W. A.; Regos, J . Antimicrobial spectrum of Triclosan, a
broad-spectrum antimicrobial agent for topical application.
Zentralbl. Bakteriol. [Orig A] 1974, 226, 376-389.
(23) Kuo, M. R.; Morbidoni, H. R.; Alland, D.; Sneddon, S. F.; Gourlie,
B. B. et al. Targeting tuberculosis and malaria through inhibi-
tion of Enoyl reductase: compound activity and structural data.
J . Biol. Chem. 2003, 278, 20851-20859.
(24) Surolia, N.; Surolia, A. Triclosan offers protection against blood
stages of malaria by inhibiting enoyl-ACP reductase of Plasmo-
dium falciparum. Nat. Med. 2001, 7, 167-173.
(25) Kapoor, M.; Dar, M. J .; Surolia, N.; Surolia, A. Kinetic deter-
minants of the interaction of enoyl-ACP reductase from Plas-
modium falciparum with its substrates and inhibitors. Biochem.
Biophys. Res. Commun. 2001, 289, 832-837.
(26) Perozzo, R.; Kuo, M.; bir Singh Sidhu, A.; Valiyaveettil, J . T.;
Bittman, R.; J acobs, W. R., J r.; Fidock, D. A.; Sacchettini, J . C.
Structural elucidation of the specificity of the antibacterial agent
triclosan for malarial enoyl ACP reductase. J . Biol. Chem. 2002,
277, 13106.
(27) Liu, B.; Wang, Y.; Fillgrove, K. L.; Anderson, V. E. Triclosan
inhibits enoyl-reductase of type I fatty acid synthase in vitro
and is cytotoxic to MCF-7 and SKBr-3 breast cancer cells. Cancer
Chemother. Pharmacol. 2002, 49, 187-193.
(28) Heath, R. J .; Rubin, J . R.; Holland, D. R.; Zhang, E.; Snow, M.
E. et al. Mechanism of triclosan inhibition of bacterial fatty acid
synthesis. J . Biol. Chem. 1999, 274, 11110-11114.
(11) McMurry, L. M.; Oethinger, M.; Levy, S. B. Triclosan targets
lipid synthesis. Nature 1998, 394, 531-532.
(12) Heath, R. J .; Yu, Y. T.; Shapiro, M. A.; Olson, E.; Rock, C. O.
Broad spectrum antimicrobial biocides target the FabI compo-
nent of fatty acid synthesis. J . Biol. Chem. 1998, 273, 30316-
30320.
(13) Ward, W. H.; Holdgate, G. A.; Rowsell, S.; McLean, E. G.;
Pauptit, R. A.; Clayton, E.; Nichols, W. W.; Colls, J . G.; Minshull,
C. A.; J ude, D. A.; Mistry, A.; Timms, D.; Camble, R.; Hales, N.
J .; Britton, C. J .; Taylor, I. W. Kinetic and structural charac-
teristics of the inhibition of enoyl (acyl carrier protein) reductase
by triclosan. Biochemistry 1999, 38, 12514-12525.
(29) Qiu, X.; J anson, C. A.; Court, R. I.; Smyth, M. G.; Payne, D. J .
et al. Molecular basis for triclosan activity involves a flipping
loop in the active site. Protein Sci. 1999, 8, 2529-2532.
(30) Sivaraman, S.; Zwahlen, J .; Bell, A. F.; Hedstrom, L.; Tonge, P.
J . Structure-Activity Studies of the Inhibition of FabI, the Enoyl
Reductase from Escherichia coli, by Triclosan: Kinetic Analysis
of Mutant FabIs. Biochemistry 2003, 42, 4406-4413.
(31) Morrison, J . F.; Walsh, C. T. The behavior and significance of
slow-binding enzyme inhibitors. Adv. Enzymol. 1988, 61, 201-
301.
(32) Baldock, C.; Rafferty, J . B.; Sedelnikova, S. E.; Baker, P. J .;
Stuitje, A. R.; Slabas, A. R. A mechanism of drug action revealed
by structural studies of enoyl reductase. Science 1996, 274,
2107-2110.
(33) Levy, C. W.; Baldock, C.; Wallace, A. J .; Sedelnikova, S.; Viner,
R. C.; Clough, J . M.; Stuitje, A. R.; Slabas, A. R.; Rice, D. W.;
Rafferty, J . B. A study of the structure-activity relationship for
diazaborine inhibition of Escherichia coli enoyl-ACP reductase.
J . Mol. Biol. 2001, 309, 171-180.
(34) Roujeinikova, A.; Sedelnikova, S.; de Boer, G. J .; Stuitje, A. R.;
Slabas, A. R.; Rafferty, J . B.; Rice, D. W. Inhibitor binding
studies on enoyl reductase reveal conformational changes related
to substrate recognition. J . Biol. Chem. 1999, 274, 30811-30817.
(35) Marcoux, J .-F.; Doye, S.; Buchwald, S. L. J . Am. Chem. Soc.
1997, 119, 10539-10540.
(36) Neumeyer, J . L.; Baindur, N.; Yuan, J .; Booth, G.; Seeman, P.
et al. Development of a high affinity and stereoselective photo-
affinity label for the D-1 dopamine receptor: synthesis and
resolution of 7-[125I]iodo-8-hydroxy-3-methyl-1-(4′-azidophenyl)-
2,3,4,5-tetrahydro- 1H-3-benzazepine. J . Med. Chem. 1990, 33,
521-526.
(37) Parikh, S.; Moynihan, D. P.; Xiao, G.; Tonge, P. J . Roles of
tyrosine 158 and lysine 165 in the catalytic mechanism of InhA,
the enoyl-ACP reductase from Mycobacterium tuberculosis.
Biochemistry 1999, 38, 13623-13634.
(14) Levy, C. W.; Roujeinikova, A.; Sedelnikova, S.; Baker, P. J .;
Stuitje, A. R. et al. Molecular basis of triclosan activity. Nature
1999, 398, 383-384.
(15) Stewart, M. J .; Parikh, S.; Xiao, G.; Tonge, P. J .; Kisker, C.
Structural basis and mechanism of enoyl reductase inhibition
by triclosan. J . Mol. Biol. 1999, 290, 859-865.
(16) Roujeinikova, A.; Levy, C. W.; Rowsell, S.; Sedelnikova, S.;
Baker, P. J .; Minshull, C. A.; Mistry, A.; Colls, J . G.; Camble,
R.; Stuitje, A. R.; Slabas, A. R.; Rafferty, J . B.; Pauptit, R. A.;
Viner, R.; Rice, D. W. Crystallographic analysis of triclosan
bound to enoyl reductase. J . Mol. Biol. 1999, 294, 527-535.
(17) Heath, R. J .; Rock, C. O. A triclosan-resistant bacterial enzyme.
Nature 2000, 406, 145-146.
(18) Heath, R. J .; Su, N.; Murphy, C. K.; Rock, C. O. The enoyl-[acyl-
carrier-protein] reductases FabI and FabL from Bacillus subtilis.
J . Biol. Chem. 2000, 275, 40128-40133.
(19) Heath, R. J .; Li, J .; Roland, G. E.; Rock, C. O. Inhibition of the
Staphylococcus aureus NADPH-dependent enoyl-acyl carrier
protein reductase by triclosan and hexachlorophene. J . Biol.
Chem. 2000, 275, 4654-4659.
(20) Marcinkeviciene, J .; J iang, W.; Kopcho, L. M.; Locke, G.; Luo,
Y. et al. Enoyl-ACP reductase (FabI) of Haemophilus influen-
zae: steady-state kinetic mechanism and inhibition by triclosan
and hexachlorophene. Arch. Biochem. Biophys. 2001, 390, 101-
108.
(21) McMurry, L. M.; McDermott, P. F.; Levy, S. B. Genetic evidence
that InhA of Mycobacterium smegmatis is a target for triclosan.
Antimicrob. Agents Chemother. 1999, 43, 711-713.
(22) Parikh, S. L.; Xiao, G.; Tonge, P. J . Inhibition of InhA, the enoyl
reductase from Mycobacterium tuberculosis, by triclosan and
isoniazid. Biochemistry 2000, 39, 7645-7650.
J M030182I