Experimental and computational studies of epithelial transport of mefenamic acid ester prodrugs

Article abstract of DOI:10.1007/s11095-005-2587-6

Purpose. A series of ester derivatives of mefenamic acid were synthesized with the aim of suppressing local gastrointestinal toxicity of mefenamic acid. A computational method was used to assist the design of the prodrug and to gain insights into the stru

Full text of DOI:10.1007/s11095-005-2587-6

Pharmaceutical Research, Vol. 22, No. 5, May 2005 (© 2005)
DOI: 10.1007/s11095-005-2587-6
Research Paper
Experimental and Computational Studies of Epithelial Transport of
Mefenamic Acid Ester Prodrugs
Kamonthip Wiwattanawongsa,1 Vimon Tantishaiyakul,1,4 Luelak Lomlim,1 Yon Rojanasakul,2
Sirirat Pinsuwan,3 and Sanae Keawnopparat3
Received November 3, 2004; accepted February 7, 2005
Purpose. A series of ester derivatives of mefenamic acid were synthesized with the aim of suppressing
local gastrointestinal toxicity of mefenamic acid. A computational method was used to assist the design
of the prodrug and to gain insights into the structure relationship of these compounds as P-glycoprotein
(P-gp) substrates. The prodrugs were studied for their enzymatic stability, bidirectional permeability
across Caco-2 monolayer, and their potential as transporter modulators
Methods. Bidirectional transport studies were performed using Caco-2 cells. Compounds exhibiting an
efflux ratio of Ն2 were further examined for their potential interaction with P-gp and multidrug resis-
tance–associated protein (MRP) using verapamil and indomethacin. Calcein efflux inhibition studies
were conducted to investigate the efflux mechanism of these compounds. Geometry optimization of the
esters was performed, and the spatial separation of two electron donor groups of each prodrug was
measured.
Results. Morpholinoethyl ester (3) and pyrrolidinoethyl ester (4) of mefenamic acid showed evidence of
efflux mechanism. Inhibition by verapamil had a pronounced effect on the transport of 3 and 4. Indo-
methacin, however, completely inhibited the apical efflux of 3 but enhanced the efflux ratio of 4. Both
compounds increased the ratio of cellular calcein accumulation by 3- to 5-fold over control. Consistent
with the experimental data, the computational results suggest the involvement of P-gp or its interaction
in 3 and 4 transport.
Conclusions. Apical efflux of 3 is associated with P-gp and MRP, but the efflux of 4 involves P-gp and/or
MRP. The computational approach used in this study provided the basis for P-gp substrates of com-
pounds 3 and 4 from their electron donor subunits spatial separation.
KEY WORDS: Caco-2; MRP; P-gp; prodrugs; transport.
INTRODUCTION
Prodrugs that temporarily mask the acidic group of
NSAIDs have been reported to suppress GI toxicity due to
local action (2). Mefenamic acid (1) is a widely used NSAID
as a fever or pain reliever. In this study, various mefenamic
acid ester prodrugs (2–5) were synthesized with the aim of
decreasing the local irritation as shown in Fig. 1. Due to the
existence of active transporters such as P-glycoprotein (P-gp)
and multidrug resistance–associated proteins (MRPs) in the
intestinal epithelium and the physicochemical consideration
of the promoieties, a computational method was used to assist
the design of prodrugs and to investigate whether the de-
signed structures are P-gp substrates as described by Seelig
(3) and Penzotti et al. (4). The physicochemical and transport
properties of the prodrugs across Caco-2 monolayers were
evaluated in both the absorption and the secretion direction.
Two compounds (3 and 4) were shown to be substrates for an
efflux mechanism. Subsequently, a calcein-AM assay was con-
ducted to explore the potential of these two compounds as
P-gp and/or MRP inhibitors.
Nonsteroidal anti-inflammatory drugs (NSAIDs) are
generally used for the treatment of inflammatory diseases.
Usage of NSAIDs, either short-term or chronic, is frequently
associated with gastrointestinal (GI) adverse effects ranging
from mild gastric upset to life-threatening ulceration and
hemorrhage (1). These gastroenteropathies are generally be-
lieved to be caused by two different mechanisms. The first
one involves a local irritant produced by acidic group of the
NSAIDs. The second effect is attributed to blockage of pros-
taglandin biosynthesis in the GI tract, which inhibits its cyto-
protective effect.
1 Department of Pharmaceutical Chemistry, Faculty of Pharmaceuti-
cal Sciences, Prince of Songkla University, Hat-Yai, Songkhla, Thai-
land.
2 Department of Basic Pharmaceutical Sciences, School of Pharmacy,
West Virginia University, Morgantown, West Virginia, USA.
3 Department of Pharmaceutical Technology, Faculty of Pharmaceu-
tical Sciences, Prince of Songkla University, Hat-Yai, Songkhla,
Thailand.
MATERIALS AND METHODS
4
Caco-2 cells were obtained from the American Type
Culture Collection (Rockville, MD, USA) at passage 18. All
To whom correspondence should be addressed. (e-mail:
vimon.t@psu.ac.th)
721
0724-8741/05/0500-0721/0 © 2005 Springer Science+Business Media, Inc.

722
Wiwattanawongsa et al.
mechanics force field and subsequently using the AM1
Hamiltonian of a semiempirical method. The obtained ge-
ometries were then optimized on the basis of the ab initio
Hartree-Fock method at the 6-21G level. The spatial separa-
tion of two electron donor groups of each prodrug was sub-
sequently measured.
Synthesis of Mefenamic Acid Prodrugs
2,3-Dihydroxypropyl
2-[(2,3-dimethylphenyl)amino]benzoate (2)
To a solution of mefenamic acid (5.00 g, 20.75 mmol) in
dry CH2Cl2 (150 ml) was added a solution of solketal (5.42 g,
41.07 mmol) in dry CH2Cl2; the mixture was cooled at 0°C,
followed by the addition of DMAP (0.10 g, 0.78 mmol) and
DCC (5.20 g, 25.06 mmol). The reaction mixture was stirred
at room temperature overnight. The precipitate of N,NЈ-
dicyclohexylurea (DCU) was removed by filtration. The fil-
trate was concentrated in vacuo, and the residue was purified
by column chromatography using ethylacetate:hexane (1:20)
as the eluent and recrystallization from MeOH/H2O to obtain
4.49 g (61%) of solketal ester of mefenamic acid as a white
solid: mp 58.3–61.4°C; IR (KBr) 1689 cm−1 1H NMR (60
;
MHz, CDCl3,) ␦ 1.40 (s, 3H, CH3), 1.47 (s, 3H, CH3), 2.17 (s,
3H, ArCH3), 2.29 (s, 3H, ArCH3), 4.33–3.68 (m, 5H,
CH2CHCH2), 6.56 (m, 2H, aromatic), 7.03–6.87 (m, 4H, aro-
matic), 7.90 (m, 1H, aromatic), 9.01 (s, 1H, NH, broad); ESI
MS: m/z 378 [M+Na]+, 356 [M+H]+.
Fig. 1. Structure of mefenamic acid esters.
chemicals used for Caco-2 cell culture were obtained from
Gibco BRL (Life Technologies, Grand Island, NY, USA).
Phenylmethyl sulfonylfluoride (PMSF), N,NЈ-dimethyl-
aminopyridine (DMAP), N,NЈdicyclohexyl-carbodiimide
(DCC), and N-hydroxysuccinimide were obtained from
Sigma (St. Louis, MO, USA). Solketal, 4-(2-hydroxy-
ethyl)morpholine, 1-(2-hydroxyethyl)-2-pyrrolidinone, and
1-(2-hydroxyethyl)-pyrrolidine were obtained from Fluka
Chemie (Buchs, Switzerland). Other reagents and solvents
were purchased from common suppliers and were used as
received.
Partition coefficients were calculated using ClogP for
Windows (Biobyte Corp., Claremont, CA, USA). The melt-
ing point was determined on a MEL-TEMP II capillary melt-
ing point apparatus and is uncorrected. Proton magnetic reso-
nance spectra were recorded on a Varian Unity Inova (500
MHz) or a Jeol JNM-PMX 60SI (60 MHz). Chemical shifts
are reported in parts per million (ppm, ␦ units) and peak
multiplicities are expressed as follows: s, singlet; d, doublet;
dd, doublet of doublet; t, triplet; m, multiplet. The IR spectra
A mixture of solketal ester of mefenamic acid (0.99 g,
2.78 mmol) and 100 ml of 70% acetic acid was heated to 60°C
for 1 h. The reaction mixture was cooled to ambient tempera-
ture and extracted with CH2Cl2 (3 × 20 ml); the CH2Cl2 ex-
tracts were washed in H2O and brine. Precipitated solid was
collected, washed with water, partially air dried, and dissolved
in 50 ml CH2Cl2. The solution was dried with anhydrous so-
dium sulfate, and the solvent was removed by evaporation
under reduced pressure. The residual was recrystallized from
methanol to obtain 0.30 g (34%) of 2 as white crystals: >99%
1
purity; mp 60.6–62.4°C; IR (KBr) 1680 cm−1; H NMR (500
MHz, DMSO-d6) ␦ 2.08 (s, 3H, CH3), 2.27 (s, 3H, CH3),
3.42–3.50 (m, 2H, CH2OH) 3.79–3.84 (m, 1H, CH2CH), 4.19
(dd, J ס 0.02, 0.01 Hz, 1H, OCH2), 4.33 (dd, J ס 0.02, 0.01
Hz, 1H, OCH2), 4.70 (t, J ס 0.01 Hz, 1H, OH), 5.02 (d, J ס
0.01 Hz, 1H, OH), 6.65–6.73 (m, 2H, aromatic), 7.03–7.13 (m,
3H, aromatic), 7.30–7.33 (m, 1H, aromatic), 7.94–7.96 (m, 1H,
aromatic), 9.12 (s, 1H, NH, broad); ESI MS: m/z 316 [M+H]+.
were recorded on a Perkin Elmer model 1600 FT-IR spec- 2-Morpholin-4-ylethyl
trometer. Mass spectra (ESI) were measured on a Micromass 2-[(2,3-dimethylphenyl)amino]benzoate (3)
Platform II mass spectrometer. Yields are of purified prod-
To a solution of 1 (5.00 g, 20.75 mmol) in CH2Cl2 (150
ucts and were not optimized. The purity of the esters was
assessed by analytical HPLC.
ml) was added 4-(2-hydroxyethyl)morpholine (2.86 g, 21.80
mmol) and DMAP (0.10 g, 0.83 mmol). The mixture was
cooled to 0°C, followed by the addition of DCC (5.21 g, 25.11
mmol). The reaction was stirred for an additional 4 h at 0°C
and stored overnight in the refrigerator. The precipitated
DCU was removed by filtration. The filtrate was concentrated
in vacuo, and the residue was purified by column chromatog-
raphy using ethyl acetate:hexane (20:80) as the eluent. The
product was recrystallized from hexane to afford 4.20 g (57%)
of 3 as pale yellow crystals: >99% purity; mp 84.8°C; IR (KBr)
Computational Method
The three-dimensional structures of compounds were
built using Chem3D Ultra ver. 7 software (CambridgeSoft,
Cambridge, MA, USA), and molecular calculation was di-
rectly performed using the Gaussian 03 W program (Gauss-
ian, Inc., Pittsburgh, PA, USA) on the Chem3D window. Ge-
ometry optimization was performed initially by molecular

Epithelial Transport of Mefenamic Ester Prodrugs
723
1
1678 cm-1; H NMR (500 MHz, CDCl3) ␦ 2.19 (s, 3H, CH3), quantitatively diluted with mobile phase and analyzed by
2.34 (s, 3H, CH3), 2.59 (m, 4H, N(CH2)2), 2.80 (t, J ס 5.80 Hz, HPLC as described in “Samples Analysis,” below.
2H, CH2N), 3.72 (m, 4H, O(CH2)2), 4.47 (t, J ס 5.80 Hz, 2H,
Chemical Stability
OCH2), 6.70–6.67 (m, 1H, aromatic), 6.76–6.75 (m, 1H, aro-
matic), 7.05–7.04 (m, 1H, aromatic) 7.18–7.10 (m, 2H, aro-
matic), 7.27–7.24 (m, 1H, aromatic), 7.98–7.96 (m, 1H, aro-
matic), 9.22 (s, 1H, NH, broad); ESI MS: m/z 355 [M+H]+.
The stability of esters 2–5 was examined in 0.01 M HCl
(pH 2.0), 0.05 M acetate buffer (pH 5.0), and 0.05 M phos-
phate buffer (pH 7.4). The ionic strength was maintained at
0.1 M by the addition of NaCl. Because the esters are mostly
insoluble in aqueous solution, stock solutions of the esters
were prepared by dissolving the compounds in methanol. The
reaction samples were prepared by adding 20–100 ␮l of 1–2
mg/ml methanolic stock solution of the esters into prewarmed
buffer solution, resulting in 5.0–25.0 ␮g/ml solutions with
1–2% v/v of methanol. The solutions were then placed into a
thermostatically controlled water bath at 37°C. At appropri-
ate times, 250–500 ␮l aliquots of samples were taken and
stored at 4°C until assayed by HPLC. Upon analysis, samples
were thawed, added with 10 ␮l diclofenac solution (internal
standard), diluted with mobile phase, and analyzed by HPLC.
Pseudo-first-order rate constants (k) were determined from
the slopes of linear plots of the logarithm of residual prodrug
concentrations vs. time. Triplicate samples were analyzed,
and the mean value of the rate constant was reported. The
corresponding half-life (t1/2) was then obtained from the
equation: t1/2 ס 0.693/k.
2-Pyrrolidin-1-ylethyl
2-[(2,3-dimethylphenyl)amino]benzoate HCl (4)
To a solution of mefenamic acid (3.00 g, 12.43 mmol) and
N-hydroxysuccinimide (1.43 g, 12.43 mmol) in 1,2-dimethoxy-
ethane (50 ml) was added DCC (2.56 g, 12.45 mmol). The
mixture was stirred at 0°C for 1 h and stored overnight in the
refrigerator. The precipitated DCU was removed by filtra-
tion. The filtrate was added to a solution of 1-(2-
hydroxyethyl)-pyrrolidine (2.93 ml, 24.86 mmol) in 1,2-
dimethoxyethane (15 ml). The mixture was stirred overnight
at room temperature. The main product was purified by col-
umn chromatography using ethyl acetate:hexane (5:95) as the
eluent, which was concentrated in vacuo. The crude product
was diluted with absolute ethanol (50 ml) and treated with
gaseous hydrogen chloride to give 0.47 g (10%) of 4 as a white
solid: >99% purity; mp190–191°C; IR (KBr): 1681 cm−1; H
1
NMR (500 MHz, DMSO-d6): ␦ 1.91 (m, 2H, CH2CH2N in
pyrrolidine ring), 2.01 (m, 2H, CH2CH2N in pyrrolidone
ring), 2.09 (s, 3H, CH3), 2.28 (s, 3H, CH3), 3.11 (m, 2H,
CH2N), 3.61 (m, 4H, CH2NCH2 in pyrrolidine ring) 4.62 (t,
J ס 5.03 Hz, 2H, OCH2), 6.65–6.74 (m, 2H, aromatic), 7.05–
7.15 (m, 3H, aromatic), 7.33–7.37 (m, 1H, aromatic), 8.06 (dd,
J ס 0.016, 0.003 Hz, 1H, aromatic), 9.14 (s, 1H, NH, broad);
ESI MS: m/z 339 [M+H]+.
Enzymatic Stability
Degradation kinetics of esters 2–5 were studied at 37°C
in Caco-2 homogenate, human plasma, and rat liver homog-
enate preparations. Each ester was incubated at the concen-
tration range of 10–100 ␮M (depending on the solubility of
the ester) in prewarmed biological medium studied. Dimethyl
sulfoxide (DMSO) at a final concentration of 1–2% v/v was
added to assist in the dissolution. The mixtures were kept in
a water bath at 37°C, and at appropriate intervals, aliquots
(250–300 ␮l) of the reaction solution were withdrawn and
analyzed by HPLC as described in “Samples Analysis.” Re-
actions were followed for at least two half-lives. Apparent
half-lives for the disappearance of ester and pseudo-first-
order rate constants were calculated as previously described.
2-(2-Oxopyrrolidin-1-yl)ethyl
2-[(2,3-dimethylphenyl)amino]benzoate (5)
This compound was prepared from 1 (5.00 g, 20.75
mmol) and 1-(2-hydroxyethyl)-2-pyrrolidinone (2.67 g, 20.67
mmol) as described for compound 3 and gave 4.67 g (64%) of
5 as white crystals: >99% purity; mp 66.7°C; IR (KBr): 1682
cm−1 1H NMR (500 MHz, CDCl3): ␦ 2.05 (m, 2H,
;
NC(O)CH2CH2 in pyrrolidone ring), 2.18 (s, 3H, CH3), 2.33
(s, 3H, CH3), 2.40 (t, J ס 8.00 Hz, 2H, NC(O)CH2 in pyrrol-
idone ring), 3.56 (t, J ס 7.50 Hz, 2H, CH2N in pyrrolidone
ring), 3.71 (t, J ס 5.50 Hz, 2H, CH2N), 4.45 (t, J ס 5.50 Hz,
2H, OCH2), 6.69–6.66 (m, 1H, aromatic), 6.76–6.73 (m, 1H,
aromatic), 7.04–7.03 (m, 1H, aromatic), 7.16–7.09 (m, 2H, aro-
matic), 7.27–7.24 (m, 1H, aromatic), 7.94–7.92 (m, 1H, aro-
matic), 9.22 (s, 1H, NH, broad); ESI MS: m/z 353 [M+H]+.
Caco-2 Homogenate Preparation
Caco-2 homogenate was prepared according to the
method of Augustijns et al. (5) with slight modifications.
Briefly, Caco-2 cells grown in 75-cm2 culture flasks for 21–25
days were washed 1–2 times with ice-cold phosphate-buffered
saline (PBS), pH 7.4. The cells were scraped off, collected in
4 ml ice-cold PBS, and homogenized using a 15-ml glass-
Teflon polytron homogenizer (Thomas Scientific, Swedes-
boro, NJ, USA). Cell debris was removed by centrifugation at
12,000 rpm for 10 min at 40°C, the resultant supernatant was
used as Caco-2 homogenate in stability studies (protein con-
tent ס 0.63 mg/ml).
Aqueous Solubility
The aqueous solubility of mefenamic acid prodrugs was
determined in water and acetate buffer pH 5.0 as well as
phosphate buffer pH 7.4 at the buffer concentration of 0.05
M, ionic strength 0.1 M adjusted with NaCl. Excess amounts
of each compound were added to 1–2 ml of buffer in screw-
Rat Liver Homogenate Preparation
Rat livers were obtained from male 250–300 g Sprague-
capped glass vials. Samples were then tumbled at 20 rpm at Dawley rats (Animal Care Units, Prince of Songkla Univer-
25°C until a constant solubility value was obtained. The satu- sity, Hat-Yai, Songkhla, Thailand). After blotting to dryness,
rated solutions were filtered through 0.45-␮m cellulose ac- tissue was sliced into small pieces and homogenized in ice-
etate membrane filters (Sartorius, Goettingen, Germany), cold PBS (1 ml per g tissue). Aliquots (1.0 ml) of the tissue

724
Wiwattanawongsa et al.
homogenates were kept at −80°C until use. Before each ex- ability coefficient, Papp (cm/s), and calculated according to the
periment, the homogenates were quickly re-homogenized on equation
ice using a 15-ml glass-Teflon polytron homogenizer. Cell de-
1
dQ
bris was removed by centrifugation at 12,000 rpm for 10 min
at 40°C, and the supernatant was used in stability studies.
Liver homogenates were diluted to 50% (v/v) with phosphate
buffer, pH 7.4, in stability studies yielding protein content of
47.3 mg/ml.
Papp
=
AC0 dt
where dQ/dt is the rate of appearance of the drug in the
acceptor side (␮mol/s), C0 is the initial drug concentration on
the donor side (mM), and A is the surface area of the mono-
layer (cm2).
Human Plasma
Efflux Inhibition Studies
Human plasma was obtained from the Songklanakarin
Hospital Blood Bank (Hat-Yai, Songkhla, Thailand). For sta-
bility studies, human plasma was diluted to 80% (v/v) with
phosphate buffer, pH 7.4 (protein content ס 49.57 mg/ml).
Inhibition of polarized drug transport across Caco-2
monolayer was performed to determine whether P-gp and/or
MRPs are involved in the transepithelial transport of 3 and 4.
Verapamil, a P-gp modulator, and indomethacin, an MRP
inhibitor, were used. Both compounds were applied at 100
␮M to the AP side and included throughout the 3-h incuba-
tion period.
Transport Studies
Cell Culture
Statistical Analysis
Caco-2 cells were grown in a controlled atmosphere at
5% CO2 and 90% relative humidity at 37°C. Culture media
consisting of Dulbecco’s modified Eagle medium (DMEM)
supplemented with 10% fetal bovine serum, 1% nonessential
amino acid, 1% L-glutamine, 100 ␮g/ml streptomycin, 100
U/ml penicillin, and 0.25 ␮g/ml amphotericin B. At 80% con-
fluency, cells were detached from the plastic support by di-
gestion using trypsin/EDTA solution and transferred to a new
75-cm2 tissue culture flask. For transport studies, cells were
seeded at a density of 60,000 cells/cm2 on Transwell (Costar,
Cambridge, MA, USA) polycarbonate membrane (24.5-mm
diameter, 4.71 cm2, 3.0-␮m pore size). The culture medium
was changed every other day. Cells from a passage number
between 42 and 53 were used in transport experiments.
Caco-2 cells grown on Transwell for 21–23 days were
used in the transport studies. The transport medium (TM)
consisted of Hanks balanced salt solution (HBSS) supple-
mented with 25 mM glucose, adjusting to pH 6.5 for the apical
(AP) compartment (1.5 ml) and 7.4 for the basolateral (BA)
compartment (2.6 ml). To ensure monolayer integrity, the
transepithelial electrical resistance (TER) was measured us-
ing the Millicell-ERS system (Millipore Co., Bedford, MA,
USA). Monolayers with TER Ն 250 ⍀·cm2 were used in the
experiments. Cell monolayers were equilibrated in TM con-
taining phenylmethyl sulfoxide (PMSF, 0.5 mM), an inhibitor
of serine protease/mammalian esterase to prevent any pos-
sible ester degradation at the apical side, for 20 min at 37°C
before the experiments.
Data were analyzed for statistical significance by one-
way analysis of variance (ANOVA) followed by Scheffe’s
test. Values of p < 0.05 were considered significant.
Samples Analysis
The HPLC system consisted of a 600 Pump, a 717 plus
Autosampler, a 486 Tunable Absorbance Detector, and a 746
Data Module (Waters, Milford, MA, USA). The analytical
column was Rexchrom octyl, 15 cm × 4.6 mm, 5 ␮m (Regis
Technologies, Morton Grove, IL, USA) equipped with a pre-
column packed with Novapak RP-8, 3.9 × 20 mm, 5 ␮m (Wa-
ters). The mobile phase consisted of acetate buffer (0.05 M)
and acetonitrile and/or methanol. The systems for analysis of
1–5 were acetate buffer (pH 5.0):methanol (40:60) for 2; ac-
etate buffer (pH 4.5):acetonitrile (57:43) for 1 and 3; acetate
buffer (pH 4.1):acetonitrile (50:50) for 4; and acetate buffer
(pH 4.1):acetonitrile (55:45) for 5. The eluents were detected
at 280 nm at a flow rate of 1.0 ml/min. Separation of the esters
2–5 from parent mefenamic acid was obtained with a total
analysis time of less than 20 min.
For determination in biological media, samples were de-
proteinized by adding 3 volumes of acetonitrile. After mixing,
and in the case of determining the accumulated amount of
ester in Caco-2 cell monolayers, sonicating for 10 min, the
samples were centrifuged for 10 min at 12,000 rpm at 4°C.
Aliquots of clear supernatant were then injected into the
HPLC system and analyzed as described above.
Transport studies were performed in triplicate starting
with the ester dissolved in HBSS (with 0.5–2% DMSO) yield-
ing the final concentrations of 15, 12, 42, and 41 ␮M for esters
2, 3, 4, and 5, respectively. The concentration of each ester
was chosen according to their solubility. Transepithelial trans-
port of each ester was determined in both directions (AP-BL
and BL-AP). Bidirectional transport was performed at 37°C
in a shaking water bath (85 strokes min−1). Aliquots of the
receiver solution (0.2 ml) were taken from the receiver com-
partment at various time points and were replaced with equal
volumes of fresh medium. Samples were kept at –80°C until
analysis. Permeation, assessed when flux was linear related to
time under sink condition, was expressed as apparent perme-
Protein content was determined using a Coomassie pro-
tein assay reagent (Pierce, Rockford, IL, USA) according to
the method of Bradford. Bovine serum albumin (BSA) was
used as a standard for protein content measurements.
Calcein-AM Assay
Caco-2 cells were seeded at 200,000 cells/well in 48-well
plate (Costar) and were incubated for 48 h before the assay.
Test compounds including prodrugs 3 and 4, P-gp inhibitors
verapamil and cyclosporin A, and an MRP inhibitor indo-
methacin were prepared at 50 ␮M in PBS pH 7.4. Caco-2 cell
monolayers were washed twice with PBS before the experi-

Epithelial Transport of Mefenamic Ester Prodrugs
725
Table I. Biological Stability of Esters of Mefenamic Acida
Half-lives in biological media (min) (mean SD)
ment. Ten microliters of the test compounds were added to
each well, and the plate was subsequently incubated at 37°C
for 30 min. Cells with the addition of PBS (containing 1%
methanol) served as a control. One microliter of calcein AM
(Molecular Probes, Eugene, OR, USA), prepared at 1.0 mM
in DMSO, was subsequently added into each well to yield a
final concentration of 2.0 ␮M. Fluorescence was immediately
measured at 30-min intervals for 180 min, at wavelengths 485-
nm excitation and 535-nm emission. The ratio of calcein ac-
cumulation was calculated based on the ratio of fluorescent
intensity with P-gp or MRP inhibitors or prodrugs to the fluo-
rescent intensity of control (medium only) at any time point
as described by Olson et al. (6).
Caco-2
Rat liver
Compounds
Plasma
homogenate
homogenate
2
3
4
5
46.1 1.7
59.9 7.2
37.2 3.1
15.5 1.2
58.3 1.0
736.0 36.0
1274.0 119.0
5.3 1.5
28.7 2.8
50.3 6.1
1362.0 103.0
19.0 1.3
a Each value represents the mean SD of three determinations.
in all transport studies. Table II shows the apparent perme-
ability coefficients across Caco-2 monolayers and the efflux
ratio (Papp AP-to-BL/Papp BL-to-AP) for mefenamic acid (1) and
its ester prodrugs (2–5). According to the efflux ratios, 1, 2,
and 5 (Table II) did not prove to be substrates for efflux
transporters, as their efflux ratios are close to 1. However, the
apically polarized efflux mechanisms are demonstrated for
compounds 3 and 4 with efflux ratios of 3.1 and 10.4, respec-
tively. The efflux transport is probably a key factor limiting
the permeation of these compounds, as indicated by their low
RESULTS AND DISCUSSION
Most ester prodrugs were synthesized in one step by di-
rectly coupling mefenamic acid with appropriate alcohol, and
the acquired esters were purified using column chromatogra-
phy. However, to obtain the prodrug 2, the solketal ester was
subsequently hydrolyzed, and in the case of 4, HCl was added
to the ester to form HCl salt.
Solubility
P
app AP-to-BL values.
The log p values for the prodrugs were estimated using
Prodrug 4, which was prepared as a hydrochloride salt, is
highly soluble in water. Its solubility is about 15 times higher
than mefenamic acid (1). The other prodrugs are less soluble
than mefenamic acid. The solubilities of 2 and 5 in water are
6.0 and 2.6 ␮g/ml, respectively. The solubility of 3 and 4,
which are basic compounds, were shown to be pH dependent;
the solubility was higher at the lower pH in which the com-
pounds are protonated. The solubility of 3 was increased from
2.6 ␮g/ml at pH 7.4 to 6.2 ␮g/ml at pH 5, and that of 4 was
enhanced from 109.3 ␮g/ml at pH 7.4 to 1045.6 ␮g/ml at pH 5.
the ClogP program, which calculates the value directly from
its molecular structure. The calculated logP values for 1, 2, 3,
4, and 5 are 4.8, 3.5, 5.3, 6.0, and 4.8, respectively. It has been
suggested that logP value may be used as a predictive tool for
transepithelial transport of drugs (7). As shown in Table II,
the Papp values in the AP-to-BL direction for 1, 2, and 5 are
comparable but somewhat increase in the following order 1 Ӎ
5 > 2. The Papp in AP-to-BL direction for 1, 2, and 5 are
positively correlated with their logP values. For prodrugs 3
and 4, which are substrates for efflux transporters, logP value
cannot be used for predicting their permeability. Therefore, it
is not straightforward to use the relationship between lipo-
philicity and permeability when the active or efflux mecha-
nism is involved in the transport system.
Chemical and Enzymatic Stability
The hydrolysis of prodrugs was studied in aqueous buff-
ers pH 2, 5, and 7.4 at 37°C. After the prodrugs were incu-
bated in these buffer solutions up to 24 h, only 5 underwent
chemical hydrolysis and released mefenamic acid with an ap-
parent half-life of 5.2, 8.1, and 8.7 h at pH 2, 5, and 7.4,
respectively. The others are stable to chemical hydrolysis in
the aqueous buffer solutions; the release of mefenamic acid
cannot be detected during 24-h incubation.
According to ANOVA analysis, the Papp AP-to-BL for 1
Table II. Permeability of Esters of Mefenamic Acid Across Caco-2
Cell Monolayersa
Papp (×10−6 cm/s)
Efflux
The stabilities of these prodrugs in various biological me-
dia including 80% human plasma, Caco-2 homogenate, and
rat liver homogenate, were determined and their apparent
half-lives are listed in Table I. All prodrugs were degraded in
these biological media, especially in human plasma. The ap-
parent half-life of prodrugs in human plasma ranged from
15.5 to 59.9 min. Prodrugs 2 and 5 underwent rapid hydrolysis
in all biological media; however, 2 had a longer half-life.
These prodrugs have been shown to be resistant toward
chemical hydrolysis in aqueous buffer solutions; however,
they can undergo enzymatic cleavages in biological media and
release the parent compound. This indicates that compounds
2–5 can be used as prodrugs.
Compound
AP to BL
BL to AP
ratio
ClogP
1
2
3
18.6 2.0*,NS
12.7 0.3
2.0 0.7
22.2 2.1
16.5 0.3
6.1 1.9NS
7.9 0.5NS
5.4 0.8NS
24.1 3.5NS
28.7 0.8NS
41.7 7.2**
13.0 2.6
5.3 0.7
1.2
1.3
3.1
1.3
0.7
10.4
8.5
14.7
0.8
4.8
3.5
5.3
3 + verapamil
3 + indomethacin
6.5 1.8**,NS
8.1 0.9**,NS
2.3 0.3NS
3.4 0.6**
2.8 0.6NS
17.2 5.0NS
4.7 0.1
4
6.0
4 + verapamil
4 + indomethacin
5
4.8
6.5
6b
1.1
a Each value represents the mean SD of three independent deter-
minations. One-way ANOVA and Scheffe’s multiple comparison
test. *p < 0.05 for comparison of 1, 2, and 5. **p < 0.05 within the
same group. NS, not significant (p > 0.05 within the same group and
for 1, 2, and 5).
Transport Studies
Because some prodrugs are degradable in Caco-2 ho-
mogenates, PMSF was used to inhibit the degradation process b From Ref. 23.

726
Wiwattanawongsa et al.
and 5 as well as for 2 and 5 are not statistically different, but and MRP modulators as well as prodrugs 3 and 4, increase the
that for 1 is different from 2 (p < 0.05). This suggests that the calcein fluorescence. The accumulation of calcein in the pres-
prodrugs with the potential to decrease local toxicity of the ence of all test compounds was about 3- to 5-fold higher than
drug exhibit comparable permeabilities to the parent com- the control, as indicated by the calcein accumulation ratio
pound.
(Table III). These results provide evidence for the inhibitory
Occasionally, substrates for efflux systems are identified effect of prodrugs 3 and 4 on P-gp and/or MRP. In view of the
by the efflux ratio and by inhibition of the efflux-mediated fact that 3 can inhibit P-gp and/or MRP, the addition of ve-
transporters. Because two efflux systems, P-gp and MRP, are rapamil or indomethacin together with 3 could result in a
expressed in Caco-2 cells (8,9), both verapamil and indometh- complete inhibition of the efflux of 3, which was observed in
acin were therefore used as inhibitors. The P-gp inhibitor this study (Table II).
verapamil caused an increase in the permeability in the ab-
Traditionally, P-gp inhibitors can either be transported
sorptive (AP-to-BL) direction and an insignificant change in or not transported by the efflux transporter (17). Those that
the secretory direction for both 3 and 4. Accordingly, the can act as both inhibitor and substrate include verapamil and
efflux ratio was decreased for both compounds (from 3.1 to cyclosporin A, while progesterone represents a non-
1.3 for 3 and from 10.4 to 8.5 for 4). The concentration of transported inhibitor (18–20). This information along with the
verapamil used (100 ␮M) may be sufficient to inhibit the observed bidirectional transport of the compounds suggests
efflux of 3, but it partially inhibited the efflux of 4. The dis- that 3 and 4 may belong to the type of P-gp inhibitors that can
crepancy in efflux inhibition of the two compounds was prob- be transported. It has been reported that P-gp inhibitors do
ably due to the affinity of each compound to the efflux pumps not inhibit MRP-dependent drug efflux, and MRP-mediated
and the concentration-dependent effect of the inhibitor on drug efflux cannot be inhibited by P-gp modulators (21).
the permeability of each compound, as previously observed Therefore, the nonspecific reversal agents that can inhibit
by Petri et al. (10). Because of the limited solubility of the both protein functions may have an advantage over the spe-
ester and its mixture with verapamil as well as the detection cific inhibitors, and some of these agents have recently been
limit in decreasing ester concentrations, the concentration of investigated (21). These prodrugs may be useful as leads for
verapamil was not increased in this study. Consequently, the designing novel nonspecific chemosensitizers. Further studies
concentration-dependent permeability of each prodrug was are required to investigate whether the prodrugs 3 and 4 can
not investigated. Nevertheless, these results could indicate function as inhibitors of both P-gp and MRP.
that P-gp may be involved in the active transport of 3 and 4.
The MRP inhibitor indomethacin increased the perme-
ability of 3 in the absorptive direction but had an insignificant
Computational Method
effect in the secretory direction. At the concentration of in-
domethacin used (100 ␮M), the efflux of 3 was completely
inhibited, and the efflux ratio was reduced from 3.1 to 0.7.
This data suggest the involvement of MRP in the transport of
3. In contrast to the effect of indomethacin on the efflux of 3,
the efflux ratio of 4 was increased in the presence of indo-
methacin. This MRP inhibitor did not statistically change the
permeability in the absorptive direction but highly enhanced
the permeability in the secretory direction. Various MRPs
including an apically localized MRP2 and basolateral MRPs
(MRP1, MRP3, MRP4, and MRP5) (9,11) probably influ-
enced the efflux of 3 and 4. Prodrug 3 may not be a good
substrate for basolateral MRPs. The inhibition of an apical
MRP2 thus reduces the efflux ratio of 3. In contrast to 3, 4
may be a better substrate for basolateral MRPs than apical
MRP2. By inhibiting these MRPs, the permeability of 4 in the
secretory direction is increased, resulting in a significant en-
hancement of the efflux of 4 over control (i.e., in the absence
of inhibitor).
It has been identified that the recognition elements for
P-gp are formed by two (type I unit) or three electron donor
groups (type II unit) with distinct spatial arrangements (3).
For the type I pattern, two electron donors are separated by
2.5 0.3 Å. The type II patterns contain either two electron
donor groups separated by 4.6 0.6 Å or three electron do-
nors separated by 2.5 0.3 Å with a 4.6 0.6 Å separation of
the outer two groups. A molecule with at least one type I or
type II unit is predicted to be transported by P-gp.
As demonstrated in Fig. 2, the spatial separation of two
electron donor groups, oxygen from CסO group of ester and
nitrogen from morpholine ring (3) and nitrogen in pyrrolidine
ring (4), is 4.9 Å, consistent with the type II pattern definition.
Furthermore, this compound also contains hydrophobic
groups (two aromatic rings) that have been recognized as an
important descriptor for P-gp substrate (4,22). In agreement
with the bidirectional transport studies, 3 and 4 are P-gp sub-
Table III. Ratio Accumulation of Calcein in Caco-2 Cell Monolayers
in the Presence of Known P-gp and MRP Modulators and Prodrugs
3 and 4 at 50 ␮M of Each Compound
Calcein-AM Assay
Calcein-AM is a nonfluorescent substrate for both MRP
and P-gp (11–13). It is cleaved in the cell to form a fluorescent
product, calcein. Calcein is not a P-gp substrate, but being a
negatively charged molecule, it can be a substrate for MRP
(14–16). However, calcein is much less efficiently transported
by MRP than calcein-AM (15). Essentially, the accumulation
of calcein may be lower due to calcein-AM and calcein efflux.
Nevertheless, the accumulation of the fluorescent calcein
should be higher in the presence of P-gp and MRP modula-
Time (min)
Compounds
30
60
90
120
Verapamil
Indomethacin
Cyclosporin A
4.9 1.8
4.5 1.6
4.2 0.9
5.8 0.9
4.6 1.5
4.2 1.0
4.0 1.0
3.6 0.6
4.1 0.4
3.4 1.4
3.6 0.8
3.3 0.8
3.3 0.5
3.6 0.4
3.1 0.8
3.1 0.6
2.9 0.7
2.8 0.6
3.0 0.2
2.6 0.6
3
4
tors than in their absence. All tested compounds, both P-gp Each value represents the mean SD of 3 determinations.

Epithelial Transport of Mefenamic Ester Prodrugs
727
5. P. Augustijns, P. Annaert, P. Heylen, G. Van den Mooter, and R.
Kinget. Drug absorption studies of prodrug esters using the
Caco-2 model: evaluation of ester hydrolysis and transepithelial
transport. Int. J. Pharm. 166:45–53 (1998).
6. D. P. Olson, B. J. Taylor, and S. P. Ivy. Detection of MRP func-
tional activity: calcein AM but not BCECF AM as a multidrug
resistance-related protein (MRP1) substrate. Cytometry 46:105–
113 (2001).
7. S. Benzaria, H. Pelicano, R. Johnson, G. Maury, J. L. Imbach,
A. M. Aubertin, G. Obert, and G. Gosselin. Synthesis, in vitro
antiviral evaluation, and stability studies of bis(S-acyl-2-
thioethyl) ester derivatives of 9-[2-(phosphonomethoxy)ethyl]ad-
enine (PMEA) as potential PMEA prodrugs with improved oral
bioavailability. J. Med. Chem. 39:4958–4965 (1996).
8. T. Hirohashi, H. Suzuki, X. Y. Chu, I. Tamai, A. Tsuji, and Y.
Sugiyama. Function and expression of multidrug resistance-
associated protein family in human colon adenocarcinoma cells
(Caco-2). J. Pharmacol. Exp. Ther. 292:265–270 (2000).
9. L. M. S. Chan, S. Lowes, and B. H. Hirst. The ABCs of drug
transport in intestine and liver: efflux proteins limiting drug ab-
sorption and bioavailability. Eur. J. Pharm. Sci. 21:25–51 (2004).
10. N. Petri, C. Tannergren, D. Rungstad, and H. Lennernas. Trans-
port characteristics of fexofenadine in the Caco-2 cell model.
Pharm. Res. 21:1398–1404 (2004).
11. H. M. Prime-Chapman, R. A. Fearn, A. E. Cooper, V. Moore,
and B. H. Hirst. Differential multidrug resistance-associated pro-
tein 1 through 6 isoform expression and function in human intes-
tinal epithelial Caco-2 cells. J. Pharmacol. Exp. Ther. 311:476–484
(2004).
12. L. Homolya, Z. Hollo, U. A. Germann, I. Pastan, M. M. Gottes-
man, and B. Sarkadi. Fluorescent cellular indicators are extruded
by the multidrug resistance protein. J. Biol. Chem. 268:21493–
21496 (1993).
13. O. Legrand, G. Simonin, J. Y. Perrot, R. Zittoun, and J. P. Marie.
Pgp and MRP activities using calcein-AM are prognostic factors
in adult acute myeloid leukemia patients. Blood 91:4480–4488
(1998).
Fig. 2. Structure of mefenamic acid ester prodrugs (2–6) and calcu-
lated distance between two-electron donor groups for each prodrug.
strates; they exhibit apical efflux that is inhibited by P-gp
inhibitor.
For prodrug 2, even though the spatial separation of elec-
tron donor groups is 4.69 Å, this compound does not seem to
be a substrate for efflux mechanism or P-gp according to the
transport experiment. This compound contains an electron
donor group (OH), which also has potential hydrogen bond-
ing donor capacity. According to Seelig (3), these groups are
not specifically favorable for an interaction with P-gp. Fur-
thermore, the predictability of the compounds as non–P-gp
substrates using the computational method was also demon-
strated for prodrug 5 and mefenamic acid guaiacol ester 6
(23), whose spatial separation of electron donor was esti-
mated to be 5.96 and 3.3 Å, respectively (Fig. 2). These com-
pounds were experimentally shown to be non–P-gp substrates
with an efflux ratio of 0.8 and 1.1, respectively (Table II).
These results also confirm the concurrence of transport stud-
ies with this computational method.
14. C. H. M. Versantvoort, T. Bagrij, K. A. Wright, and P. R. Twen-
tyman. On the relationship between the probenecid-sensitive
transport of daunorubicin or calcein and the glutathione status of
cells overexpressing the multidrug resistance-associated protein
(MRP). Int. J. Cancer 63:855–862 (1995).
In conclusion, ester prodrugs of mefenamic acid were
synthesized for reduced gastric toxicity of this NSAID. A
computational method was successfully used to design pro-
drugs that avoid the efflux process by P-gp. In addition, bidi-
rectional transport studies indicated that two prodrugs were
substrates of active efflux transporters. These prodrugs were
able to increase cellular calcein accumulation, suggesting that
they may function as P-gp and/or MRP inhibitors. These com-
pounds may be useful as leads for designing new inhibitors of
efflux transporters.
15. M. Essodaigui, H. J. Broxterman, and A. Garnier-Suillerot. Ki-
netic analysis of calcein and calcein-acetoxymethylester efflux
mediated by the multidrug resistance protein and P-glycoprotein.
Biochemistry 37:2243–2250 (1998).
16. N. Feller, H. J. Broxterman, D. C. R. Wahrer, and H. M. Pinedo.
ATP-dependent efflux of calcein by the multidrug resistance pro-
tein (MRP): no inhibition by intracellular glutathione depletion.
FEBS Lett. 368:385–388 (1995).
17. S. Kajiji, J. A. Dreslin, K. Grizzuti, and P. Gros. Structurally
distinct MDR modulators show specific patterns of reversal
against P-glycoproteins bearing unique mutations at serine939/
941. Biochemistry 33:5041–5048 (1994).
18. I. Tamai and A. R. Safa. Azidopine noncompetitively interacts
with vinblastine and cyclosporin A binding to P-glycoprotein in
multidrug resistant cells. J. Biol. Chem. 266:16796–16800 (1991).
19. I. Tamai and A. R. Safa. Competitive interaction of cyclosporins
with the vinca alkaloid-binding site of P-glycoprotein in multi-
drug resistant cells. J. Biol. Chem. 265:16509–16513 (1990).
20. T. Saeki, K. Ueda, Y. Tanigawara, R. Hori, and T. Komano.
P-glycoprotein mediated transcellular transport of MDR-
reversing agents. FEBS Lett. 324:99–102 (1993).
ACKNOWLEDGMENTS
Financial support for this research was provided by the
Thailand Research Fund through the Royal Golden Jubilee
Ph.D. Program (grant no. PHD/5GPS/43/D.1 to K.W. and
V.T.) and Prince of Songkla University.
21. S. P. Cole, K. E. Sparks, K. Fraser, D. W. Loe, C. E. Grant, G. M.
Wilson, and R. G. Deeley. Pharmacological characterization of
multidrug resistant MRP-transfected human tumor cells. Cancer
Res. 54:5902–5910 (1994).
22. T. R. Stouch and O. Gudmundsson. Progress in understanding
the structure-activity relationships of P-glycoprotein. Adv. Drug
Deliv. Rev. 54:315–328 (2002).
REFERENCES
1. E. L. Semble and W. C. Wu. Anti-inflammatory drugs and gastric
mucosal damage. Semin. Arthritis Rheum. 16:271–286 (1987).
2. V. K. Tammara, M. M. Narurkar, A. M. Crider, and M. A. Khan.
Morpholinoalkyl ester prodrugs of diclofenac: synthesis, in vitro
and in vivo evaluation. J. Pharm. Sci. 83:644–648 (1994).
3. A. Seelig. A general pattern for substrate recognition by P-
glycoprotein. Eur. J. Biochem. 251:252–261 (1988).
4. J. E. Penzotti, M. L. Lamb, E. Evensen, and P. D. J. Grootenhuis.
A computational ensemble pharmacophore model for identifying
substrates of P-glycoprotein. J. Med. Chem. 45:1737–1740 (2002).
23. V. Tantishaiyakul, K. Wiwattanawongsa, S. Pinsuwan, S. Kasi-
wong, N. Phadoongsombut, S. Kaewnopparat, N. Kaewnopparat,
and Y. Rojanasakul. Characterization of mefenamic acid-
guaiacol ester: stability and transport across Caco-2 cell mono-
layers. Pharm. Res. 19:1013–1018 (2002).