Structural differences between paroxetine and femoxetine responsible for differential inhibition of Staphylococcus aureus efflux pumps

Article abstract of DOI:10.1016/j.bmcl.2004.04.018

In this study the chemical modification of paroxetine was employed to determine which structural differences between the paroxetine-like and femoxetine-like selective serotonin reuptake inhibitors is responsible for the differential potency of these agent

Full text of DOI:10.1016/j.bmcl.2004.04.018

Bioorganic & Medicinal Chemistry Letters 14 (2004) 3093–3097
Structural differences between paroxetine and femoxetine
responsible for differential inhibition of
Staphylococcus aureus efflux pumps
a
Peng Wei, Glenn W. Kaatz and Robert J. Kerns
b
a,
*
a
Division of Medicinal and Natural Products Chemistry, The University of Iowa, Iowa City, IA 52242, USA
b
John D. Dingell VAMC & Wayne State University School of Medicine, Detroit, MI 48201, USA
Received 8 January 2004; revised 8 April 2004; accepted 10 April 2004
Abstract—In this study the chemical modification of paroxetine was employed to determine which structural differences between the
paroxetine-like and femoxetine-like selective serotonin reuptake inhibitors is responsible for the differential potency of these agents
in the inhibition of Staphylococcus aureus multidrug efflux pump systems.
Ó 2004 Elsevier Ltd. All rights reserved.
2
b;5
Antimicrobial agent resistance is a major problem in the
1
mutation-mediated resistance. The NorAprotein was
the first characterized pump to impart efflux-mediated
chemotherapeutic treatment of bacterial infections.
Recently, the action of membrane-based efflux pump
systems has been shown to play a significant role in the
multidrug resistance in Staphylococcus aureus (S.
6
aureus). Subsequently, a non-NorA-related multidrug
7
2
development of multidrug resistance to antibiotics. The
efflux phenotype has been identified in S. aureus.
Inhibitors of the NorAefflux pump system in S. aureus
development of efflux-mediated resistance to antibiotics
generally occurs through the up-regulation of genes
encoding transporters that efficiently expel the drug
from the bacterial cell. As a direct resistance mechanism,
the bacterial efflux pump affords subtherapeutic intra-
cellular antibiotic concentrations by expelling substrates
upon their entering the membrane or cytoplasm. Bac-
terial efflux pump systems also contribute indirectly to
other types of resistance because exposure of bacteria to
subtherapeutic drug concentrations promotes the selec-
tion and expression of higher-level adaptive resistance
have included variety of natural products and synthetic
agents.
5a
Recently, Kaatz et al. demonstrated that structural
variants of certain phenylpiperidine selective serotonin
reuptake inhibitors (P-SSRIs) inhibited the function of
8
multidrug efflux pumps in S. aureus (Fig. 1). The
P-SSRIs were also shown to augment the potency of
antimicrobial efflux pump substrates against strains of
S. aureus possessing different efflux-related mechanisms
and inhibited both NorAand non-Nor A- related efflux
phenotypes. For example, paroxetine (25 lg/mL) affor-
ded an 8-fold reduction in the MIC of norfloxacin (50–
6.3 lg/mL) against SA1199B (NorAoverproducing
phenotype). This activity is similar to the 4-fold reduc-
tion in norfloxacin MIC produced by 20 lg/mL reser-
pine (a known bacterial efflux pump inhibitor).
Paroxetine and reserpine each afforded an 8-fold
reduction in the MIC of norfloxacin (12.5–1.6 lg/mL)
2;3
mechanisms, such as target mutations. Constitutive
expression of such pumps in some pathogenic organisms
contributes significantly to their innate multidrug resis-
2;4
tance phenotype.
Due to the profound role of bacterial efflux pumps in the
development of antimicrobial resistance much effort has
been invested in the search for efflux pump inhibitors,
which when combined with substrate antibiotics can
restore antibiotic potency and reduce the occurrence of
8
against SA-K2068 (non-NorA phenotype). This aug-
mentation of antibiotic activity was shown to be a result
of efflux inhibition; the P-SSRIs all have poor intrinsic
antistaphylococcal activity. Evaluation of efflux pump
inhibition by the P-SSRIs showed ‘no’ significant ste-
reochemical requirement for pump inhibition, where
Keywords: SSRI; Efflux inhibitor; Antibiotic resistance.
*
Corresponding author. Tel.: +1-3193358800; fax: +1-3193358766;
e-mail: robert-kerns@uiowa.edu
0
960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.04.018

3094
P. Wei et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3093–3097
Figure 1. Structures of phenylpiperidine selective serotonin reuptake
8
inhibitors that inhibit multidrug efflux pump systems in S. aureus.
paroxetine and corresponding stereoisomer 2 displayed
near equivalent activity, as did femoxetine and corre-
sponding stereoisomer 4 (Fig. 1). However, paroxetine
and 2 achieved 50% inhibition of the different S. aureus
efflux pumps at 2- to 10-fold lower concentrations than
femoxetine (3) and 4.
Scheme 1. Synthesis of paroxetine derivatives employed in this study
to evaluate the role of peripheral substituents on the differential pump
inhibitory activity of paroxetine-like versus femoxetine-like
In this study, we set out to determine which structural
differences between the paroxetine-like and femoxetine-
like P-SSRIs are responsible for differential inhibition of
efflux pumps in S. aureus, and thus determine which
peripheral structural features of the P-SSRIs are
important for pump inhibition and which sites on the
P-SSRI core structure are sensitive to substitution.
These data will reveal sites on the P-SSRI core structure
amenable to further chemical modification in an effort to
increase potency of pump inhibition. The non-stereo-
chemical structural differences between femoxetine-like
and paroxetine-like P-SSRIs are: (1) the fluorine atom
on the phenylpiperidine portion of paroxetine is absent
in femoxetine, (2) the piperidine of paroxetine is un-
substituted while femoxetine is N-methylated, (3) the
phenylether moiety of femoxetine bears a 4-methoxy-
substituent versus the sesamol group (methylenedioxy
substituent) of paroxetine (Fig. 1).
9–12
P-SSRIs.
and femoxetine are similarly substituted, but not
equivalently, with the methylenedioxy and methoxy
groups, respectively. To determine if this portion of the
P-SSRI structure is sensitive to substituent effects and is
important for pump inhibitory activity the methylene-
dioxy group of paroxetine was converted to catechol
derivative 8, affording an analog with a dramatic switch
in the character of substituents on the phenylether
12
ring.
Relative potency of the P-SSRI efflux pump inhibitors
was evaluated using the ethidium bromide (EtBr) efflux
13
assay as previously reported (Fig. 2). In this assay,
increasing concentrations of the P-SSRIs and derivatives
were evaluated for their inhibition of EtBr (pump sub-
strate) efflux from two strains of S. aureus that possess
either NorA(S A- 1199B) or a non-NorAefflux pump
(SA-K2068).
To this end, defluorinated paroxetine derivative 5 was
prepared to evaluate the role of aryl fluoride substitu-
tion on the differential activity of paroxetine-like and
9
femoxetine-like P-SSRIs (Scheme 1). Preparation of the
N-methyl paroxetine derivative 6 was similarly under-
taken to determine if N-methylation of the piperidine
ring in femoxetine is responsible for decreased pump
Efflux inhibition demonstrates that desfluoroparoxetine
5 and paroxetine inhibit EtBr efflux in both strains of
S. aureus with near equivalent potency, indicating the
aryl fluorine atom is not responsible for the difference in
pump inhibitory activity between paroxetine and
femoxetine (Fig. 2). Although the fluorine atom is small,
the ability to make this subtle change on the phenyl-
piperidine structure indicates it may be possible to
introduce structural changes on this ring toward iden-
tifying more potent pump inhibitors and toward pre-
paring analogs devoid of CNS activities.
10
inhibitory activity compared to paroxetine (Scheme 1).
While 6 was anticipated to establish the role of N-
methylation in the differential activity of paroxetine and
femoxetine, both the N-unsubstituted and N-methylated
P-SSRIs are expected to be ionized under assay condi-
tions. Thus, N-acetyl paroxetine derivative 7 was pre-
pared to further evaluate the role of N-substitution on
pump inhibitory activity and to potentially elucidate if
there is a requirement for an ionizable piperidine ring
11
(
Scheme 1). The phenylether substituent of paroxetine

P. Wei et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3093–3097
3095
Figure 2. Effect of P-SSRIs and derivatives on EtBr efflux. (A) S. aureus 1199B, which constitutively over expresses norA and harbors a topoiso-
merase IV Asubunit substitution (A116E); (B) S. aureus K2068, a strain demonstrating a non-NorAefflux-related multidrug resistance phenotype.
The horizontal line indicates the concentrations necessary to inhibit efflux by 50%. d, 1; , 3; +, 4; ., 5; j, 6; N, 7; r, 8.
The N-methyl paroxetine derivative 6 displays less po-
tent pump inhibition than paroxetine against both pump
systems tested (Fig. 2). Most significant is that 6 inhibits
of efflux inhibition. Further exploration of structure–
function relationships around this portion of the P-SSRI
structure is warranted toward identifying more potent
pump inhibitors.
5
5
0% EtBr efflux against SA-1199B at approximately
0 lM, which is equivalent to the activity of femoxetine
and 4 (Fig. 2). This indicates that the structural differ-
ence between paroxetine-like P-SSRIs and femoxetine-
like P-SSRIs responsible for differential inhibition of
efflux in SA-1199B is N-substitution of the piperidine
ring. Furthermore, the N-acetyl derivative of paroxetine
With respect to structural requirements of the P-SSRIs
for CNS activity, the core 4-phenylpiperidine structure
is common to many CNS active agents in addition to the
P-SSRIs, likely due to similarities with the aryl-alkyl-
amine pharmacophore of native neurotransmitters
including serotonin, norepinephrine, and dopamine. The
apparent requirement of an unsubstituted piperidine
nitrogen for improved efflux pump inhibition shown
here makes N-substitution an unlikely option to destroy
CNS activity while increasing and/or broadening efflux
pump inhibitory activity. However, remodeling the 4-
phenylpiperidine core structure by changing the distance
and/or orientation between the piperidine nitrogen and
the phenyl ring may be one approach to identify
improved efflux pump inhibitors that display decreased
CNS activity (e.g., 3-phenylpiperidine derivatives,
replacing the piperidine ring with alkyl amines of varied
lengths to break the 4-phenylpiperidine pharmacophore,
or incorporating substituents onto the piperidine ring).
Perhaps most important is that P-SSRIs and other aryl-
substituted alkylamine-based CNS agents accommodate
only limited substitution of their aryl ring moieties,
which is consistent with their requirement to bind sites
on transporters that recognize the parent neurotrans-
mitters. Thus, exploring a structurally diverse range of
aromatic groups and aryl substituents for either of the
aryl rings in the P-SSRI structure is likely a promising
approach to initially identify derivatives having in-
creased pump inhibitory activity and a concomitant
decrease or elimination of CNS activity.
7
, which can no longer carry a positive charge, also
displays nearly identical inhibition of efflux to femoxe-
tine, 4 and 6 in SA-1199B (Fig. 2). This observation that
N-methylation and N-acetylation of paroxetine both
produce femoxetine-like activity suggests either a puta-
tive requirement for the N-H group to form a positive
binding contact in pump inhibition (e.g., hydrogen
bond) or that N-substitution may be sterically less
favorable.
The negative effect on pump inhibition observed for N-
substituted paroxetine derivatives 6 and 7 in SA-1199B
is also observed in the inhibition of efflux in SA-K2068,
where N-methylation and N-acetylation afford approx-
imately a 2-fold increase in concentration required to
achieve 50% pump inhibition (Fig. 2). In this latter case
however, the N-substituted paroxetine analogs still
inhibit 50% efflux at 5- to 10-fold lower concentrations
than 4 and femoxetine, respectively, indicating the subtle
structural differences in the different substituents on the
phenylether moiety are important to the inhibition of
efflux in strain SA-K2068 but not strain SA-1199B. The
overall importance of the phenylether substituents to
pump inhibitory activity is exemplified by the clear de-
crease in pump inhibition observed for phenylether-
modified paroxetine derivative 8 (Fig. 2). The effect of
modifying the phenylether group on pump inhibition
activity is more pronounced in the inhibition of efflux in
SA-K2068 than SA-1199B, which is consistent with data
discussed above where N-methyl paroxetine 6, femoxe-
tine and 4 showed similar inhibitory potency against
efflux in SA-1199B while N-methyl paroxetine is a more
potent inhibitor of efflux than femoxetine and 4 in SA-
K2068. Thus, substituents on the phenylether portion of
the P-SSRIs structure play an important role in potency
We have demonstrated that the disparity in potency of
efflux pump inhibition between paroxetine-like and
femoxetine-like P-SSRIs against NorAand non-NorA
efflux pumps in S. aureus is primarily due to differing N-
substitution of the piperidine ring. In addition, substit-
uents on the phenylether moiety have been shown to
play an important role in modulating potency of efflux
inhibition, indicating further chemical modification of
this ring may afford the opportunity to identify more

3096
P. Wei et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3093–3097
potent pump inhibitors. Continued structure–activity
relationship studies to further characterize the pharma-
cophore of the P-SSRIs required for pump inhibition
and to identify peripheral substitution of the phenyl-
ether and phenylpiperidine rings that improve potency
and remove CNS activities are expected to reveal novel
potent inhibitors of the efflux pump systems in S. aureus.
filtered, and evaporated to dryness. Purification of 5 was
achieved using semi-preparative reversed-phase (C-18)
HPLC employing a linear gradient of acetonitrile in water
containing 0.1% TFA, affording separation of the desired
product as a single peak. Acetonitrile was evaporated and
the remaining aqueous sample lyophilized affording 97%
yield of 5 a hygroscopic white solid. Purity of 5 was
confirmed by analytical HPLC showing a single peak at
1
2
20 and 254 nm. H NMR (CDCl
3
) d 2.08 (d, 1H), 2.24
(
(
1
3
1
1
t, 1H), 2.52 (t, 1H), 2.81 (t, 1H), 3.08 (m, 2H), 3.48–3.72
m, 4H), 5.86 (s, 2H), 6.14 (d, 1H), 6.35 (s, 1H), 6.24 (d,
H), 7.19–7.41 (m, 5H), 9.40 (s, 1H). C NMR (CDCl ) d
3
1
3
Acknowledgements
0.07, 39.34, 42.59, 44.47, 46.81, 67.74, 98.09, 101.18,
05.69, 107.87, 127.34, 127.46, 129.04, 141.23, 142.02,
This study was partially supported by VAresearch
funds and by the University of Iowa Biological Sciences
Funding Program.
þ
48.22, 153.75. MS, ESI, calcd (M+H ) 312, found
þ
(
M+H ) 312.
1
0. N-Methylation of paroxetine to afford 6 was performed as
follows. Methyl iodide (13.6 mg, 0.096 mmol) was added
dropwise to paroxetineÆHCl hemihydrate (30 mg,
0
.08 mmol) and 55 mg potassium carbonate in 2 mL of
References and notes
dry acetonitrile. The reaction mixture was then heated to
reflux for 5 h, cooled, the solution filtered and filtrate
evaporated to dryness. Pure 6 was isolated as a hygro-
1
2
. (a) Walsh, C. Nature 2000, 406, 775–781; (b) Prieto, J.;
Calvo, A.; Gomez-Lus, M. L. J. Antimicrob. Chemother.
002, 50(Suppl. 2), S7–S12; (c) Normark, B. H.; Normark,
S. J. Int. Med. 2002, 252, 91–106; (d) Livermore, D. M.
Clin. Infect. Dis. 2003, 36(Suppl. 1), S11–S23.
scopic white solid in 92% yield employing HPLC separa-
1
tion and analysis as described for 5. H NMR (CDCl
3
) d
.82–1.92 (m, 2H), 2.04–2.14 (m, 2H), 2.19–2.29 (m, 1H),
2
1
2
.38 (s, 3H), 2.44 (m, 1H), 2.97 (d, 1H), 3.21 (d, 1H), 3.44
. (a) Van Bambeke, F.; Glupczynski, Y.; Pl ꢀe siat, P.;
Pech ꢁe re, J. C.; Tulkens, P. M. J. Antimicrob. Chemother.
(
(
(
9
1
m, 1H), 3.59 (m, 1H), 5.87 (s, 2H), 6.17 (m, 1H), 6.38
1
3
s, 1H), 6.65 (d, 1H), 6.96 (m, 1H), 7.19 (m, 1H). C NMR
CDCl ) d 34.20, 42.01, 43.43, 46.36, 56.11, 59.44, 69.47,
7.96, 101.06, 105.57, 107.83, 115.28, 115.56, 128.77,
28.87, 139.50, 141.57, 148.14, 154.35, 159.07, 163.24.
2
003, 51, 1055–1065; (b) Lomovskaya, O.; Watkins, W. J.
3
Curr. Med. Chem. 2001, 8, 1699–1711; (c) Ryan, B. M.;
Dougherty, T. J.; Beaulieu, D.; Chuang, J.; Dougherty, B.
A.; Barrett, J. F. Exp. Opin. Invest. Drugs 2001, 10, 1409–
422.
. (a) Markham, P. N. Antimicrob. Agents Chemother. 1999,
3, 988–1136; (b) Markham, P. N.; Neyfakh, A. A.
Antimicrob. Agents Chemother. 1996, 40, 2673.
. Poole, K.; Srikumar, R. Curr. Top. Med. Chem. 2001, 1,
9–71.
. (a) Kaatz, G. W. Exp. Opin. Emerging Drug 2002, 7, 223–
33; (b) Lomovskaya, O.; Watkins, W. J. Mol. Microbiol.
Biotechnol. 2001, 3, 225–236; (c) Renau, T. E.; Lemoine,
R. C. Drugs Future 2001, 26, 1171–1178; (d) Renau, T.;
Leger, R.; Flamme, E.; Sangalang, J.; She, M.; Yen, R.;
Gannon, C.; Griffith, D.; Chamberland, S.; Lomovskaya, O.;
Hecker, S.; Lee, V.; Ohta, T.; Nakayama, K. J. Med.
Chem. 1999, 42, 4928–4931; (e) Markham, P. N. Antimic-
rob. Agents Chemother. 1999, 43, 988–989; (f) Markham,
P. N.; Neyfakh, A. Antimicrob. Agents Chemother. 1996,
0, 2673–2674.
. (a) Kaatz, G. W.; Seo, S. Antimicrob. Agents Chemother.
997, 41, 2733–2737; (b) Kaatz, G. W.; Seo, S. Antimicrob.
Agents Chemother. 1995, 39, 2650–3579; (c) Kaatz, G. W.;
Seo, S.; O’Brien, L.; Wahiduzzaman, M.; Foster, T.
Antimicrob. Agents Chemother. 2000, 44, 1404–1406.
. Kaatz, G. W.; Moudgal, V. V.; Seo, S. M. J. Antimicrob.
Chemother. 2002, 50, 833–838.
þ
þ
MS, ESI, calcd (M+H ) 344, found (M+H ) 344.
1. Acetic anhydride (0.1 mL) was added dropwise to a stirred
solution of paroxetineÆHCl hemihydrate (30 mg,
1
1
3
0
.08 mmol) and catalytic DMAP in 2 mL dry pyridine
under N . The mixture was stirred at room temperature
for 2 h then evaporated to dryness. Pure 7 was isolated in
5% yield as a white solid employing HPLC separation as
4
2
4
5
9
5
1
described for 5. H NMR (CDCl
3
) d 1.68–2.04 (m, 3H),
.20 (s, 3H), 2.62–2.82 (m, 2H), 3.05–3.25 (m, 1H), 3.44–
2
3
4
2
.51 (m, 1H), 3.62–3.68 (m, 1H), 3.90–4.19 (m, 1H), 4.75–
.95 (m, 1H), 5.87 (s, 2H), 6.17 (m, 1H), 6.38 (s, 1H), 6.65
(d, 1H), 6.94 (m, 1H), 7.17 (m, 1H). MS, EI, calcd 371,
found 371.
1
2. Cleavage of the methylenedioxy group utilized the method
of Brooks et al.: Brooks, P. R.; Wirtz, M. C.; Vetelino, M.
G.; Rescek, D. M.; Woodworth, G. F.; Morgan, B. P.;
Coe, J. W. J. Org. Chem. 1999, 64, 9719–9721. Paroxe-
tineÆHCl hemihydrate (20 mg, 0.053 mmol) and n-butyl
ammonium iodide (49 mg, 0.132 mmol) were stirred in
4
6
2
1
mL dry CH
M solution in CH
2
Cl
2
at )78 °C under N
2 3
. BCl (0.132 mL of a
2
Cl ) was added dropwise and the
1
2
mixture stirred for 2 h. The reaction was quenched by
addition of ice water and stirred at room temperature for
3
organic layer dried over Na SO , filtered, and evaporated
0 min. The mixture was extracted with CH
2 2
Cl , the
7
8
2
4
to dryness. Pure 8 was isolated in 70% yield as a
hygroscopic solid employing HPLC separation as de-
. Kaatz, G. W.; Moudgal, V. V.; Seo, S. M.; Bondo Hansen,
J.; Kristiansen, J. E. Int. J. Antimicrob. Agents 2003, 22,
1
scribed for 5. H NMR (CDCl
3
) d 1.43 (m, 2H), 1.82 (m,
H), 2.20 (m, 1H), 2.71 (m, 1H), 2.82–2.92 (m, 2H), 3.28–
254–261.
2
3
6
9
. Defluorination of paroxetine to afford 5 was achieved
using Pd catalyzed hydrodehalogenation as previously
described for the hydrodehalogenation of aromatic
halides: Ukisu, Y.; Miyadera, T. J. Mol. Catal. A 1997,
.42 (m, 4H), 5.97 (m, 1H), 6.24 (d, 1H), 6.56 (d, 1H),
.76–6.86 (m, 2H), 6.96–7.04 (m, 2H). MS EI, calcd 317,
found 317.
3. Kaatz, G. W.; Seo, S. M.; O’Brien, L.; Wahiduzzaman,
M.; Foster, T. J. Antimicrob. Agents Chemother. 2000, 44,
1
1
25, 135–142. Briefly, a solution of paroxetineÆHCl hemi-
hydrate (20 mg, 0.053 mmol) and 10.6 mg NaOH in 2 mL
-propanol was treated with a catalytic amount of pre-
dried 10% Pd/C under a N atmosphere. The mixture was
stirred at 82 °C for 3 h, cooled to room temperature,
1
404–1406. Briefly, cells were grown in Mueller–Hinton
broth (MHB; Difco, Detroit, MI) to late log phase (OD660
.8), at which time 25 lM ethidium bromide (EtBr) and
2
2
0

P. Wei et al. / Bioorg. Med. Chem. Lett. 14 (2004) 3093–3097
3097
1
(
00 lM carbonyl cyanide m-chlorophenyl hydrazone
CCCP) were added to load cells with EtBr. After 20 min
incubation at room temperature the OD660 was adjusted to
inhibitors. The fluorescence of the suspension was mon-
itored continuously for 5 min (excitation and emission
wavelengths, 530 and 600 nm, respectively). Percent inhi-
bition of efflux was calculated by determining the differ-
ence in total efflux versus that observed in the presence of
an inhibitor.
0
1
.4 using fresh MHB containing EtBr and CCCP. Cells in
mL of this dilution were pelleted and then resuspended in
fresh MHB with or without various concentrations of