Synthesis, stability and in vitro dermal evaluation of aminocarbonyloxymethyl esters as prodrugs of carboxylic acid agents

Article abstract of DOI:10.1016/S0968-0896(01)00336-4

Aminocarbonyloxymethyl esters 3 based on (S)-amino acid carriers were synthesised and evaluated as potential prodrugs of carboxylic acid agents. In addition, the compounds were evaluated as topical prodrugs with the aim of improving the dermal delivery of two non-steroidal anti-inflammatory agents: naproxen and flufenamic acid. The lipophilicities of these compounds were determined and their hydrolyses in aqueous solutions and in human plasma were examined. Compounds 3 containing a secondary carbamate group were hydrolysed at pH 7.4 by two different routes: (i) direct nucleophilic attack at the ester carbonyl carbon leading to the release of the parent carboxylic acid and (ii) intramolecular rearrangement involving an O → N acyl migration, leading to the formation of the corresponding amide. The rearrangement pathway is highly dependent on the size of the carboxylic acid and amino acid substituents, being eliminated when the amino acid is valine or leucine. In contrast, compounds 3 decomposed in plasma exclusively through ester hydrolysis, most releasing the parent carboxylic acid quantitatively with half-lives shorter than 5 min. The permeation of selected prodrugs across excised postmortem human skin was studied in vitro. All prodrugs evaluated exhibited a lower flux than the corresponding parent carboxylic acid. The poor skin permeation observed for compounds 3 is most probably due to their low aqueous solubility and high partition coefficient.

Full text of DOI:10.1016/S0968-0896(01)00336-4

Bioorganic & Medicinal Chemistry 10 (2002) 809–816
Synthesis, Stability and In Vitro Dermal Evaluation of
Aminocarbonyloxymethyl Esters as Prodrugs of
Carboxylic Acid Agents
Eduarda Mendes,a,* Tania Furtado,a Joao Neres,a Jim Iley,b Tomi Jarvinen,c
Jarkko Rautioc and Rui Moreiraa
a
´
CECF, Faculdade de Farmacia da Universidade de Lisboa, 1699 Lisboa Codex, Portugal
bChemistry Department, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK
cDepartment of Pharmaceutical Chemistry, University of Kuopio, PO Box 1627, Fin-70211 Kuopio, Finland
Received 6 July 2001; accepted 21 September 2001
Abstract—Aminocarbonyloxymethyl esters 3 based on (S)-amino acid carriers were synthesised and evaluated as potential prodrugs
of carboxylic acid agents. In addition, the compounds were evaluated as topical prodrugs with the aim of improving the dermal
delivery of two non-steroidal anti-inflammatory agents: naproxen and flufenamic acid. The lipophilicities of these compounds were
determined and their hydrolyses in aqueous solutions and in human plasma were examined. Compounds 3 containing a secondary
carbamate group were hydrolysed at pH 7.4 by two different routes: (i) direct nucleophilic attack at the ester carbonyl carbon
leading to the release of the parent carboxylic acid and (ii) intramolecular rearrangement involving an O!N acyl migration, leading
to the formation of the corresponding amide. The rearrangement pathway is highly dependent on the size of the carboxylic acid and
amino acid substituents, being eliminated when the amino acid is valine or leucine. In contrast, compounds 3 decomposed in plasma
exclusively through ester hydrolysis, most releasing the parent carboxylic acid quantitatively with half-lives shorter than 5 min. The
permeation of selected prodrugs across excised postmortem human skin was studied in vitro. All prodrugs evaluated exhibited a
lower flux than the corresponding parent carboxylic acid. The poor skin permeation observed for compounds 3 is most probably
due to their low aqueous solubility and high partition coefficient. # 2002 Elsevier Science Ltd. All rights reserved.
Introduction
local drug concentration is needed.7À9 However,
because most NSAIDs also display limited skin per-
meation, the prodrug approach has also been suggested
as an attractive method of enhancing the skin perme-
ability of several members of this drug class such as
naproxen and ketoprofen.10À13
Many drugs containing the carboxylic acid group are
poorly absorbed from the gastrointestinal (GI) tract as
a result of unfavourable physicochemical properties.1
The prodrug strategy has been widely used to improve
oral delivery of such drugs by appropriate modulation
of properties, such as lipophilicity and aqueous solubi-
lity.2 Moreover, for non-steroidal anti-inflammatory
drugs (NSAIDs), such as naproxen and diclofenac,
which contain the carboxylic acid functional group, the
prodrug concept has also been used to minimise the GI
side effects that constitute the most frequent adverse
reactions of this class of drug.3À6 Further, for NSAIDs,
topical application can be considered as an alternative
delivery method to oral use, especially in the treatment
of local inflammatory and pain processes, where a high
A commonly used prodrug approach is to prepare acy-
loxymethyl derivatives 1. These offer the possibility of
modulating the rate of release of the parent drug and
simultaneously modifying the physicochemical proper-
ties, such as water solubility and lipophilicity, by vary-
ing the R group in the acyl pro-moiety. However,
neutral acyloxymethyl ester prodrugs 1 still present oral
and dermal delivery problems due to very low water
solubility and very high lipid solubility. For example, a-
(acyloxy)alkyl esters of neutral b-lactam antibiotics are
inefficiently absorbed from the GI tract14 and thus
taken orally are poorly active because they have very
low aqueous solubility. It has also been recently repor-
ted that acyloxymethyl esters of naproxen present low
*Corresponding author. Tel.: +351-21-7946-413; fax: +351-21-7946-
470; e-mail: ermendes@ff.ul.pt
0968-0896/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(01)00336-4

810
E. Mendes et al. / Bioorg. Med. Chem. 10 (2002) 809–816
Table 1. Structures of compounds 3
is based on the low chemical reactivity of carbamates in
neutral aqueous media.18 The present study reports the
synthesis of a series of aminocarbonyloxymethyl esters 3
(Table 1) and a kinetic study that evaluates the influence
of the aminoacid and carboxylic acid moieties on the
chemical reactivity and stability in human plasma. The
permeation of selected aminocarbonyloxymethyl esters
3 across excised human skin was also evaluated in vitro.
Recent studies have suggested that an effective prodrug
for dermal drug delivery must possess good biphasic
solubility characteristics: that is, adequate aqueous
solubility together with enhanced lipophilicity over the
parent drug.19, 20
Compound
R1
R2
R3COOH
3a
3b
3c
3d
3e
3f
H
H
H
H
H
CH2CO2Et
CH(Me)CO2Et
CH(CH2Ph)CO2Et
CH(CHMe2)CO2Et
CH(CH2CHMe2)CO2Et
CH2CO2Et
Benzoic acid
Benzoic acid
Benzoic acid
Benzoic acid
Benzoic acid
Diclofenac
H
3g
3h
3i
3j
3k
H
H
H
Me
Me
CH2CO2Et
CH2CO2Et
CH2CO2Et
CH2CO2Et
CH2CO2Et
Clofibric acid
Nicotinic acid
Flufenamic acid
Benzoic acid
Naproxen
3l
Benzoic acid
dermal delivery through excised human skin, probably
due to their poor aqueous solubility and high partition
coefficients that are above the optimal range for skin
permeation.12 The introduction of a basic a-amino
group into the acyl pro-moiety of 1, for example by
using an a-aminoacid carrier such as 2, significantly
increased the water-solubility but led to unstable com-
pounds.13 This is due both to the strong electron-with-
drawing effect of the protonated amino group at
physiological pH and also to the ability of the unproto-
nated amino group to enhance hydrolysis via nucleo-
Experimental
Melting points were determined using a Kofler camera
Bock Monoscop M and are uncorrected. Elemental
analyses were performed either by Medac Ltd, Brunel
Science Park, Englefield Green, Egham, Surrey, UK or
Instituto Tecnologico Nuclear, Sacavem, Portugal.
FTIR spectra were recorded using a Nicolet Impact 400
spectrophotometer1H and 13C NMR spectra were
recorded in CDCl3 solutions with Me4Si as internal
standard, using a Jeol LA300 or General Electric QE-
300 spectrometers. Coupling constants, J, are expressed
in Hz. The carboxylic acids were purchased from Sigma.
All other chemicals and solvents were of reagent grade,
except buffers substances and HPLC solvents which
were analytical grade and LiChrosolv# (Merck) grade,
respectively. Column chromatography was performed
using silica gel 60mesh 70–230(Merck).
philic and general-base catalysis.2
A
further
disadvantage of aminoacyloxymethyl esters 2 is their
potential to undergo intramolecular rearrangement,
involving the amino group of the amino acid fragment,
to form an amide, as has been reported for such deri-
vatives of cefamandole.14 Even so, because they offer
enormous potential in terms of low toxicity, amino
acids have been considered the ideal carriers for the
development of prodrugs. Thus, a potentially valuable
strategy employing amino acid carriers is undone by the
presence of the free amino group derived from the
amino acid moiety.
Gogate has previously discussed the rationale behind
the use of (acyloxy)alkyl carbamates as prodrugs of
amines.15,16 Indeed, modified carbamates with an enzy-
matically hydrolyzable ester function were also sug-
gested as prodrugs for amines17 and this is now
probably one of the most versatile prodrug approaches
for an amino containing compound. By adapting and
combining the acyloxymethyl and carbamate strategies,
we now propose a novel and versatile approach for the
design and synthesis of natural-amino-acid-carrier pro-
drugs of drugs containing the carboxylic acid function-
ality. Conceptually, this approach corresponds to an
inversion of the amino acid position in the potentially
unstable prodrugs 2, and the rationale for this approach
General method for the synthesis of aminocarbonyl-
oxymethyl esters (3)
A dichloromethane solution (20mL) of the amino acid
ethyl or methyl ester hydrochloride (11.3 mmol) and
triethylamine (10.8 mmol) was added to a solution of
chloromethyl chloroformate (0.95 mL; 10.8 mmol) in
dichloromethane (25 mL) at À10 ꢀC. After stirring at
À10 ꢀC for 25 min and 75 min at room temperature, the
reaction mixture was filtered, and the filtrate washed
with water, dried and evaporated to obtain the corre-
sponding chloromethylcarbamate 4 in ca. 90% yield. A

E. Mendes et al. / Bioorg. Med. Chem. 10 (2002) 809–816
811
solution of the sodium salt of the appropriate carboxylic
acid (6.4 mmol) and tetrabutylammonium bromide (6.4
mmol) in tetrahydrofuran (10mL) was added to a
solution of the intermediate 4 (6.4 mmol) in tetra-
hydrofuran (5 mL). After stirring for 24 h at room
temperature, the solvent was evaporated and the resul-
tant solid purified by silica gel column chromatography
using diethyl ether as eluent.
(1H, d, J=7.2, ArH), 6.74 (1H, s, NH), 6.94–7.34 (6H,
m, ArH). Found: C, 52.7; H, 4.4; N, 5.7. Calcd for
C20H20N2Cl2O6: C, 52.8; H, 4.4; N, 6.1.
3g. Synthesised using ethyl glycinate hydrochloride and
clofibric acid (sodium salt). Yield 40%. Oil; nmax/cmÀ1
3384, 1730(v br), 1650;
dH 1.29 (3H, d, J=7.0,
CH3CH2), 1.62 (6H, s, Me2), 3.92 (2H, d, J=5.1,
NCH2), 4.20(2H, q, J=7.0, CH3CH2), 5.44 (1H, t,
J=5.1, NH), 5.84 (2H, s, OCH2O), 6.80(2H, d, J=8.8,
ArH), 7.19 (2H, d, J=8.8, ArH). Found: C, 52.0; H,
5.7; N, 3.8. Calcd for C16H20NO7Cl: C, 51.4; H, 5.4; N,
3.7.
3a. Synthesised using ethyl glycinate hydrochoride and
sodium benzoate. Yield 15%. Mp 68–70 ꢀC; nmax/cmÀ1
3312, 1740, 1720 (br), 1603; dH 1.26 (3H, t, J=7.0,
CH3CH2), 3.96 (2H, d, J=5.4, NCH2), 4.16 (2H, q,
J=7.0, CH3CH2), 5.63 (1H, t, J=5.4, NH), 5.96 (2H, s,
OCH2O), 7.36–8.16 (5H, m, ArH). Found: C, 55.3; H,
5.3; N, 4.8. Calcd for C13H15NO6: C, 55.5; H, 5.3; N,
4.9.
3h. Synthesised using ethyl glycinate hydrochloride and
nicotinic acid (sodium salt). Yield 34%. Mp 80–81 ꢀC;
n
max/cmÀ13230, 1750 (v br), 1600; dH 1.24 (3H, d, J=7,
CH3CH2), 3.97 (2H, d, J=5.7, NCH2), 4.18 (2H, q,
J=7.0, CH3CH2), 5.75 (1H, t, J=5.1, NH), 5.99 (2H, s,
OCH2O), 7.36–7.40(1H, m, ArH), 8.28–8.31 (1H, m,
ArH), 8.76–8.78 (1H, m, ArH), 9.22 (1H, s, ArH).
Found: C, 51.5; H, 5.1; N, 10.2. Calcd for C12H14N2O6:
C, 51.1; H, 5.0; N, 9.9.
3b. Synthesised using ethyl alaninate hydrochloride and
sodium benzoate. Yield 50%. Oil; nmax/cmÀ1 3348, 1741
(v br), 1600 dH 1.19 (3H, t, J=7.1, CH3CH2), 1.36 (3H,
d, J=7.1, CHCH3), 4.13 (2H, q, J=7.1, CH3CH2), 4.30
(1H, quint, J=7.5, CHCH3), 5.47 (1H, d, J=7.3, NH),
5.92 (2H, 2 Â d, J=5.86, OCH2O), 7.34–8.03 (5H, m,
ArH). Found: C, 57.0; H, 5.9; N, 4.6. Calcd for
C14H17NO6: C, 57.0; H, 5.8; N, 4.7.
3i. Synthesised using ethyl glycinate hydrochloride and
flufenamic acid (sodium salt). Yield 44%. Mp 84–85 ꢀC;
n
max/cmÀ1 3404, 3328, 1753, 1725 (br), 1696; dH 1.26
3c. Synthesised using ethyl phenylalaninate hydrochlo-
ride and sodium benzoate. Yield 35%. Mp 90–92 ꢀC;
n
J=7.2, CH3CH2), 3.17 and 3.10(2H, 2 Â dd, J=14.7,
6.0Hz, Ar–C H2), 4.17–4.12 (2H, m, CH3CH2), 4.64
(1H, dt, J=8.1; 6.0, a-CH), 5.40(1H, d, J=8.1, NH),
5.97 (2H, s, OCH2O), 7.08–8.09 (10H, m, ArH). Found:
C, 65.0; H, 6.2; N, 3.8. Calcd for C20H21NO6: C, 64.7;
H, 5.7; N, 3.8.
(3H, t, J=7.5, CH3CH2), 4.04 (2H, d, J=5.4, NCH2),
4.20(2H, q, J=7.2, CH2CH3), 5.57 (1H, t, J=5.4,
CH2NH), 6.00 (2H, s, OCH2O), 6.80–8.03 (8H, m,
ArH), 9.45 (1H, s, NH). Found: C, 54.9; H, 4.3; N, 6.0.
Calcd for C20H19N2O6F3: C, 54.5; H, 4.3; N, 6.4.
max/cmÀ1 3333, 1749, 1705 (br), 1600; dH 1.22 (3H, t,
3j. Synthesised using ethyl sarcosine hydrochloride and
sodium benzoate. Yield 58%. Oil; nmax/cmÀ1 1700 (v
br), 1600; dH (mixture of rotamers): 1.20and 1.26 (3H, 2
 t, J=7.0, CH3CH2), 3.06 (3H, s, NCH3), 3.98 and
4.05 (2H, 2 Â s, CH2N), 4.11 and 4.19 (2H, 2 Â q,
J=7.0, CH3CH2), 5.99 and 6.04 (2H, 2 Â s, OCH2O),
7.50–8.30 (5H, m, ArH). Found: C, 57.0; H, 5.7; N, 4.9.
Calcd for C14H17NO6: C, 56.9; H, 5.8; N, 4.7.
3d. Synthesised using ethyl valinate hydrochloride and
sodium benzoate. Yield 30%. Oil; nmax/cmÀ1 3380, 1730
(v br), 1600. dH 0.87 (3H, d, J=7.5, CH3CHCH3), 0.94
(3H, d, J=6.9, CH3CHCH3), 1.24 (3H, t, J=6.9,
CH3CH2O), 2.10–2.22 (1H, m, MeCH), 4.17 (2H, m,
CH3CH2O), 4.27 (1H, dd, J=9.0, 4.7, a-CH), 5.48 (1H,
d, J=9.0, NH), 5.97 (2H, s, OCH2O), 7.42–8.06 (5H, m,
ArH). Found: C, 59.3; H, 6.9; N, 4.4. Calcd for
C16H21NO6: C, 59.4; H, 6.5; N, 4.1.
3k. Synthesised using ethyl sarcosine hydrochloride and
naproxen (sodium salt). Yield 45%. Oil; nmax/cmÀ1 1710
(v br), 1600; dH (mixture of rotamers) 1.20and 1.25
(3H, 2 Â t, J=7.0, CH3CH2), 1.43 and 1.46 (3H, 2 Â d,
J=6.0, CHCH3), 2.68 and 2.76 (3H, 2 Â s, NCH3), 3.91
(3H, s, OMe), 4.10and 4.17 (2H, 2 Â q, J=7.0,
CH3CH2), 3.93 (2H, m, NCH2), 5.55–5.66 (2 H, 4 Â d,
J=5.7, OCH2O), 6.95–7.27 (6H, m, ArH); dC 14.07 and
14.10( CH3CH2), 18.16 (a-CH3), 35.86 and 35.18 (N–
CH3), 45.15 and 45.13 (CH), 50.52 and 50.24 (CH3CH2),
52.23 (CH3O), 61.18 (NCH2), 80.36 and 80.65 (OCH2O),
105.5 (CHAr), 118.9–134.9 (CHAr and CAr), 157.7
(COMe), 154.3 and 154.9 (C¼O), 168.9 and 169.0
(C¼O), 173.5 and 173.5 (C¼O). Found: C, 61.5; H, 6.4;
N, 3.4. Calcd for C21H25NO7: C, 61.5; H, 6.2; N, 3.5.
3e. Synthesised using ethyl leucinate hydrochloride and
sodium benzoate. Yield 40%. Oil; nmax/cmÀ1 3370, 1740
(v br), 1600; dH 0.93 (6H, t, J=6.6, (CH3)2CH), 1.25
(3H, t, J=6.9, CH3CH2O), 1.49–1.70(3H, m,
(Me2CHCH2), 4.17 (2H, q, J=7.0, CH3CH2O), 4.37
(1H, m, a-CH), 5.37 (1H, d, J=8.1, NH), 5.98 (2 H, s,
OCH2O), 7.44–8.07 (5H, m, ArH). Found: C, 60.2; H,
7.2; N, 4.3. Calcd for C17H23NO6: C, 60.5; H, 6.9; N,
4.1.
3f. Synthesised using ethyl glycinate hydrochloride and
diclofenac (sodium salt). Yield 20%. Mp 63–65 ꢀC;
n
J=7.0, CH3CH2), 3.87 (2H, s, CH2Ar), 3.98 (2H, d,
J=5.7, NCH2), 4.22 (2H, q, J=7.6, CH3CH2), 5.39
(1H, t, J=5.2, CH2NH), 5.80(2H, s, OC H2O), 6.56
3l. Synthesised using methyl prolinate hydrochloride
and sodium benzoate. Yield 77%. Oil; nmax/cmÀ1 1740
(v br), 1607; dH (mixture of rotamers) 1.79–2.31 (4H, m,
2 Â CH2), 3.44–3.69 (2H, m, NCH2), 3.61 and 3.74 (3H,
2 Â s, CH3O), 4.39 and 4.36 (1H, 2 Â dd, J=11.1, 8.7
max/cmÀ1 3370, 3358, 1760, 1744, 1720; dH 1.28 (3H, t,

812
E. Mendes et al. / Bioorg. Med. Chem. 10 (2002) 809–816
and J=11.1, 8.4, a-CH), 5.95, 5.98 and 6.06 (2H, 4 Â
overlapping doublets, J=5.6, OCH2O), 7.42–8.10(5H,
m, ArH); dC 23.35 and 24.22 (CH2 proline), 29.85 and
30.77 (CH2 proline), 46.62 and 46.95 (NCH2 proline),
52.19 and 52.28 (CH3O), 58.79 and 59.23 (a-CH), 80.24
and 80.64 (OCH2O), 128.45 (ArCH), 129.97 and 130.03
(ArCH), 133.52 and 133.56 (ArCH), 152.58 and 153.17
(C¼O), 165.28 and 165.33 (C¼O); 172.61 and 172.65
(C¼O). Found: C, 58.6; H, 5.5; N, 4.5. Calcd for
C15H17NO6: C, 58.6; H, 5.5; N, 4.5.
samples were centrifuged and the supernatant analysed
by HPLC for the presence of substrate and products.
In vitro skin permeation study
Samples of human skin were obtained from the
abdominal region of adult human cadavers from the
Kuopio University Hospital (Kuopio, Finland). The
epidermis was isolated from the underlying dermis by
heat separation at 60 ꢀC in distilled water for 2 min,
after which the skin specimens were dried and frozen
prior to use. The permeation studies were carried out
using the Franz-type diffusion cell (PermeGear, Inc.,
Riegel, PA, USA) as previously described.12 Skin speci-
mens were rehydrated before being mounted in the dif-
fusion cell. The receptor medium (0.05 M pH 7.4 isotonic
phosphate buffer solution) was stirred and kept at 37 ꢀC
throughout the study. The compounds were applied as
suspensions in 0.05 M pH 5.0 acetate buffer. At specified
time intervals 250 mL aliquots were withdrawn from the
receptor compartment and replaced with fresh buffer.
The drug concentrations were assayed by HPLC. The
steady-state flux for parent compounds and their pro-
drugs was determined by plotting the cumulative
amount (in nmol) of the parent drug and intact prodrug
as measured in the receptor phase against time, and
dividing the slope of the steady-state position by the
surface area of the diffusion cell (0.71 cm2).
Apparent partition coefficients
These were determined at 22 ꢀC using octanol–pH 7.4
phosphate buffer and octanol–pH 5.0acetate buffer.
Each phase was mutually saturated before the experi-
ment. The volumes of each phase were chosen so that
solute concentrations in the aqueous phase after dis-
tribution were readily measurable. The compounds were
dissolved in 1-octanol and the octanol–phosphate buffer
mixtures were shaken for 30min to reach an equili-
brium distribution. Each phase was analysed separately
by HPLC. Partition coefficients, P, were calculated from
the ratio of the peak area in octanol to the peak area in
buffer.
Hydrolysis in aqueous solution
All kinetic experiments carried out in aqueous systems
were buffered (phosphate, acetate, etc.) with the ionic
strength maintained at 0.5 M using NaClO4. Reaction
mixtures were analysed using UV spectrophotometry
and HPLC. Usually, a 10 mL aliquot of a 10À2 M stock
solution of substrate in acetonitrile was added to 10mL
of the appropriate thermostatted buffer solution. At
regular intervals, samples of the reaction mixture were
analysed by HPLC using the following conditions: a
Merck LiChrospher# 100 RP-18 5 mm 125 Â 4 mm
column; mobile phase, acetonitrile/water (the composi-
tion of which depended on the compound); flow rate 1.0
mL/min; detector wavelength, 230nm.
Statistical analysis
A one-factor analysis of variance (ANOVA factorial)
was used to test the statistical significance of differences
between parent acid and prodrugs. Significance in the dif-
ferences in the means was tested using Fisher’s protected
least significance difference (PLSD) at 95% confidence.
Results and Discussion
Synthesis
The synthesis of N-aminocarbonyloxymethyl esters 3a–l
(Table 1) involves the preparation of N-[(chloro-
methyl)oxy]carbonyl derivatives 4 by reaction of the
appropriate amino acid ester with chloromethyl chloro-
formate (Scheme 1). Reaction of intermediates 4 with
the sodium salt of the appropriate carboxylic acid agent
afforded 3 in 15–77% yield. As the main aim of the
present work was to study the stability and skin perme-
ability of these compounds, we made no attempt to
optimise these yields. However, two trends are worthy
Hydrolysis in human plasma
Human plasma was obtained from the pooled, hepar-
inised blood of healthy donors, and was frozen and
stored at À70 ꢀC prior to use. For the hydrolysis
experiments, the substrates 3 were incubated at 37 ꢀC in
human plasma that had been diluted to 80% (v/v) with
pH 7.4 isotonic phosphate buffer. At appropriate inter-
vals, aliquots were added to acetonitrile to both quench
the reaction and precipitate plasma proteins. These
Scheme 1. ClCH2OCOCl/Et3N/CH2Cl2; R3CO2Na/TBAB/THF.

E. Mendes et al. / Bioorg. Med. Chem. 10 (2002) 809–816
813
of mention. First, the yields are higher the more bulky
the amino acid moiety. This suggests that the carboxylate
salt can attack more than one site in the intermediate 4
and that steric bulk directs it to the less-hindered chloro-
methyl site. Second, tertiary carbamates also give rise
to higher yields. One possible explanation for this is
that those intermediates 4 that contain the secondary
carbamate functionality undergo a competetive E1cb
elimination reaction catalysed by the carboxylate ion.
The structure of compounds 3a–l follows from their
1
spectroscopic and analytical data. In particular, the H
NMR spectra of N-aminocarbonyloxymethyl esters 3a
and 3c–i exhibit a characteristic singlet at ca. d 6.0–6.6
ppm due to the OCH2O group. Interestingly despite the
diastereotopic nature of the OCH2O protons in com-
pounds 3b–e, only 3b revealed the presence of a pair of
Figure 1. pH-Rate profile for aminocarbonyloxymethyl benzoate 3a at
37 ꢀC.
1
doublets (J=5.9 Hz). The 13C and H NMR spectra of
derivatives 3j–l, which contain a tertiary carbamate
group, reveal them to be a mixture of E and Z rotamers
due to restricted rotation about carbamate C–N bond.
Moreover, for the proline derivative, 3l, the OCH2O
signal appears as two pairs of overlapping doublets (the
central signal of which at d 5.98 is twice as intense as the
other two signals), reflecting the diastereotopic nature
of the methylene protons (J=6.0Hz) as well as the
presence of two rotamers. The spectroscopic detection
of rotamers has been reported for other prolyl carba-
mates and is consistent with a barrier to rotation for the
carbamate N-C¼O framework similar to the barrier of
acyclic tertiary amides.21,22
well as acetate, phosphate and borate buffers. The
pseudo-first-order rate constants, kobs, were found to
vary with buffer concentration. The pH-rate profile
(Fig. 1) was constructed using the intercepts of plots of
kobs versus [buffer], together with the kobs values deter-
mined in HCl and NaOH solutions. Between pH 2 and
pH 6 reactions were too slow to monitor accurately. A
similar pH-rate profile, indicative of the presence of
acid-catalysed and base-catalysed processes has been
described for the related N-acyloxyalkoxycarbonyl
derivatives of aniline, 5 (R=Ph) studied by Gogate.15
The pH-rate profile can be described by eq (1),
þÂ þÃ
Kinetics and products of chemical hydrolysis
kobs ¼ kapp ½OHÀŠ þ kH
H
ð1Þ
OHÀ
The hydrolysis of compounds 3a–l was studied in iso-
tonic pH 7.4 phosphate buffer at 37 ꢀC, and the corre-
sponding half-lives are given in Tables 1 and 2. All
reactions displayed strict first-order kinetics over at least
four half-lives. Compounds 3 are reasonably stable in
pH 7.4 buffer, with half-lives ranging between 9 and 169
h, except the diclofenac and nicotinic acid derivatives, 3f
and 3h, which hydrolyse with half-lives of ca. 1 and 2 h,
respectively.
app
OHÀ
where k
and kH are the OHÀ catalysed and H+
+
catalysed second-order rate constants, respectively. The
best computer fit (solid line) to the experimental data
(individual points) using eq (1), gave values for the cat-
alytic second-order rate constants, kH and kOHÀ, of
+
9.0Â10Â and 16.94 MÀ1
s
À1, respectively.
5
The hydrolysis of derivative 3a was also studied over a
range of pH values using HCl and NaOH solutions as
As revealed by the HPLC analysis of the reaction mix-
tures, the hydrolysis of N-aminocarbonyloxymethyl
ester 3a proceeds with quantitative formation of the
parent carboxylic acid at pH < 3 (Scheme 2). However,
for pH > 5, ethyl N-benzoylglycinate 6 (R2=EtO-
COCH2NH, R3=Ph) was also detected as a major
reaction product, presumably arising from a rearrange-
ment involving a O!N intramolecular acyl transfer
from the carbamate conjugate base (Scheme 2). As
revealed from the data in Table 2, the parent carboxylic
acid and the N-acylamine rearrangement product were
also observed in the hydrolysis of secondary carbamates
3a–c and 3f–i in pH 7.4 phosphate buffer (see also
Scheme 2.

814
E. Mendes et al. / Bioorg. Med. Chem. 10 (2002) 809–816
Table 2. Kinetic and product data for the hydrolysis of esters 3 in pH 7.4 isotonic phosphate buffer and in 80% human plasma at 37 ꢀC, together
with their apparent partition coefficients (logPapp, mean+SD; n=3)
Compound
pH 7.4 phosphate buffer
80% Human plasma
t1/2 / (min) % carboxylic acid
logPapp
t1/2 / (h)
% Rearrangement
% carboxylic acid
pH 7.4
pH 5.0
3a
3b
3c
3d
3e
3f
3g
3h
3i
9.04
30.6
15.5
70.3
29.2
0.98
37.08.8
2.24
19.0
34.3
39.1
169
45.4
34.8
2.3
ND
ND
7.0
91.2
82.0
19.4
ND
ND
ND
54.6
65.2
97.7
100
100
93.0
1.623.904
18.0
80.6
100
0.24
1.90
4.87
2.85
2.50
4.17
10
0.43
21.8
100
100
100
100
100
100
2.02ꢁ0.05
2.45ꢁ0.03
3.74ꢁ0.18
3.17ꢁ0.06
3.61ꢁ0.17
4.16ꢁ0.15
ꢁ0.01
1.88ꢁ0.04
2.26ꢁ0.03
3.57ꢁ0.09
3.02ꢁ0.03
3.63ꢁ0.01
3.92ꢁ0.05
2.91ꢁ0.03
0.63ꢁ0.09
4.29ꢁ0.30
2.07ꢁ0.02
3.35ꢁ0.17
2.14ꢁ0.08
100
100
100
100
100
0.68ꢁ0.02
3.99ꢁ0.17
2.09ꢁ0.10
3.52ꢁ0.10
2.12ꢁ0.09
3j
3k
3l
0.41
2.72
3.48
100
100
ND: Not detected
boxylic acid; no rearrangement product was detected
for any compound. This observation contrasts sharply
with the corresponding product profiles in pH 7.4 buf-
fer, especially for the benzoates 3a,b and for the nicoti-
nate 3h. These product profiles in human plasma also
contrast with those for the closely related secondary N-
acyloxyalkoxycarbonyl derivatives of aniline 5 (R=Ph)
which were reported to give between 15 and 30% of
rearrangement product, but compare with that for the
benzylamine counterpart 5 (R=CH2Ph) which hydro-
lysed in human plasma almost exclusively to the parent
amine.16
The susceptibility of compounds 3 to undergo enzy-
matic plasma activation was assessed in vitro by com-
paring the half-lives for hydrolysis in 80% human
plasma with those in pH 7.4 phosphate buffer at 37 ꢀC
(Table 2). With half-lives ranging from 0.2 to ca. 20 min,
it is clear that human plasma enzymes markedly acceler-
ated the rate of hydrolysis of aminocarbonyloxymethyl
esters 3a–l. Reactivity in human plasma appears to be
dependent on the size of the substituents in both the
amino acid carrier and the parent carboxylic acid.
Indeed, the benzoate 3a hydrolysed ca. 17, 7 and 90
times faster than the more sterically hindered esters 3f,
3g and 3i, respectively. Similarly, compound 3a, a gly-
cine derivative, hydrolysed ca. 8 and 20times faster
than its alanine and phenylalanine counterparts, 3b and
3c, respectively. The rate data in Table 2 were analysed
for quantitative relationships with several physico-
chemical parameters, including Charton’s steric para-
meter n for the amino acid carrier, the pKa of the parent
carboxylic acid and the logP values of the prodrugs. No
clear correlation between the rate data encompassing all
aminocarbonyloxymethyl esters 3a–l and the corre-
sponding physicochemical parameters emerged, although
a slight correlation (r2=0.62) was observed between log
t1/2 and logP, indicating that increasing substrate lipo-
philicity lead to a reduction in reactivity (Fig. 3). A
similar trend has been reported for the closely related
acyloxymethyl carbamates of triazenes, 7,23 as well as
for other different types of prodrugs, for example ami-
domethyl esters,24 and can be ascribed to the depen-
dence of esterase activity on substrate lipophilicity.
Figure 2. HPLC analysis of the hydrolysis of 3h in pH 7.4 phosphate
buffer: 3h (&); ethyl N-benzoylglycinate (*); nicotinic acid (~).
Fig. 2). Consistent with rearrangement arising from the
conjugate base, hydrolysis of tertiary carbamates 3j–l
proceeded with quantitative formation of the parent
carboxylic acid. Inspection of the data in Table 2 sug-
gests that the extent of the rearrangement depends on
the steric hindrance of the amino acid and carboxylic
acid moieties. Thus, intramolecular acyl transfer is an
important route for benzoates 3a and 3b, whereas for
the more sterically hindered valine, 3d, and leucine
derivatives, 3e, only direct ester hydrolysis was
observed. For the nicotinic acid derivative 3h there is a
predominance of the intramolecular acyl transfer when
compared to diclofenac, 3f, and clofibrate, 3g, deriva-
tives. Moreover, the more reactive nicotinate derivative
3h displays a greater level of rearrangement than the
corresponding benzoate 3a, as might be expected for a
more electrophilic acyl group that is the result of the
presence of the nitrogen atom in the aryl ring.
Hydrolysis in human plasma
Hydrolysis of compounds 3a–l in human plasma also
followed strict first-order kinetics for at least four half-
lives, (r2 > 0.95 in all cases). Product analysis revealed
that the plasma catalysed hydrolysis of these esters
quantitatively afforded the corresponding parent car-

E. Mendes et al. / Bioorg. Med. Chem. 10 (2002) 809–816
815
buffer. Unfortunately, all the prodrugs evaluated exhib-
ited a lower flux than the corresponding parent car-
boxylic acid. The poor skin permeation observed for
compounds 3 is most probably due to their high parti-
tion coefficient (Table 2) and, more especially, their low
aqueous solubility (Table 3). Similar poor skin permea-
tion has been reported for the neutral pivaloyloxy-
methyl prodrug of naproxen (1, R=CMe3).12 This
compound has similar logPapp (2.8) and aqueous solu-
bility (0.4 mM) values12 to the naproxen derivative 3k
(Tables 2 and 3). Not surprisingly, therefore, it displays
a similar Jss value (0.02 nmol/cm2/h) to 3k. This is con-
sistent with recent reports suggesting that an effective
topical prodrug must possess good biphasic solubility
characteristics (i.e., good balance between lipid and
aqueous solubility).20,25 Moreover, optimal logP values
ranging between 2 and 3 for maximal fluxes have been
reported for several NSAIDs.26,27
In vitro skin permeation study
In vitro skin permeation studies through excised post-
mortem human skin were carried out using suspensions
in phosphate buffer to maintain constant diffusion and
maximum flux. Each parent carboxylic acid and corre-
sponding prodrug was applied in 0.05 M pH 5.0 isotonic
phosphate buffer to avoid significant hydrolysis (see
Fig. 1). The cumulative amount (in nmol/cm2) of per-
meated parent acid or intact prodrug generally dis-
played a linear relationship with time (data not shown).
The steady-state flux (Jss) was calculated from the slope
of the linear portions of these plots and is presented in
Table 3. All prodrugs tested permeated the skin intact,
reflecting their expected stability in pH 5.0phosphate
In conclusion, aminocarbonyloxymethyl esters 3 based
on non-toxic amino acid carriers provide a new and
potentially attractive approach to the preparation of
prodrugs of carboxylic acid drugs. The present study
shows that it is possible to prepare aminocarbonyloxy-
methyl esters 3 with a wide range of lipophilicities and
that simultaneously combine a high susceptibility to
plasma hydrolysis with good stability in aqueous buf-
fers. More importantly, compounds 3 can liberate
quantitatively in a few min a parent carboxylic acid
containing bulky substituents. This is a significant
improvement over the amidomethyl esters, based on the
ethyl hippurate carrier, of sterically hindered carboxylic
acids, which are only slowly hydrolysed in human
plasma.24 Despite their potential as prodrugs, the cur-
rent series of compounds displayed poor skin permea-
tion, probably due to their high lipophilicity and low
aqueous solubility. We are currently evaluating similar
derivatives containing an ionisable functionality that
should increase aqueous solubility.
Figure 3. Plot of the plasma hydrolysis half-lives of compounds 3
against the corresponding pH 7.4 logPapp values.
Acknowledgements
Table 3. Steady-state fluxes (Jss) (meanꢁSE; n=2–12) for the deliv-
ery of total parent compound species through excised human skin in
vitro in 0.05 M pH 5.0 isotonic phosphate buffer at 37 ꢀC
This project was supported by Fundacao para a Ciencia
e Tecnologia (POCTI) and by the Academy of Finland.
Compound
Aqueous solubility
(pH 5.0)/(mM)
JSS
(nmol cmÀ2 hÀ1
)
Naproxen
3k
460ꢁ201.6
2ꢁ0.3
ꢁ0.2a
0.06ꢁ0.01b
References and Notes
Flufenamic acid
3i
41ꢁ0.6
0.83ꢁ0.16
1. Bundgaard, H. In Bioreversible Carriers in Drug Design:
Theory and Application; Roche, E. B., Ed.; Pergamon: New
York, 1987; p. 13.
2. Bundgaard, H. Drugs of the Future 1991, 16, 443.
3. Shanbhag, V. R.; Crider, A. M.; Gokhale, R.; Harpalani,
A.; Dick, R. M. J. Pharm. Sci. 1992, 81, 149.
4. Tammara, V. K.; Narurkar, M. M.; Crider, A. M.; Khan,
M. A. Pharm. Res. 1993, 10, 1191.
5. Tammara, V. K.; Narurkar, M. M.; Crider, A. M.; Khan,
M. A. J. Pharm. Sci. 1994, 83, 644.
6. Khalaj, A.; Amanlou, M.; Jorjani, M. Pharm. Pharmacol.
Commun. 1998, 4, 201.
d
c
—
—
Benzoic acid
39,000ꢁ5000
1301ꢁ106
3ꢁ0.5
51.24ꢁ4.40
3.54ꢁ0.71b
3b
3c
3d
3e
c
—
.50
166ꢁ200
34ꢁ2
ꢁ0.15b
0.91ꢁ0.28b
aRef 20.
bp<0.05 compared to parent compound (ANOVA, Fisher’s PLSD
test).
cNo detectable amount was found in the receptor chamber.
dCompound bellow the limit of quantitation in the solution. Calcu-
lated solubility 0.56 mM (ALOGPS v2.0available at http://
www.lnh.unil.ch/appl/cchem00.html).
7. Suh, H.; Jun, H. W.; Dzimiansky, M. T.; Lu, G. W. Bio-
pharm. Drug Disp. 1997, 18, 623.

816
E. Mendes et al. / Bioorg. Med. Chem. 10 (2002) 809–816
8. McNeill, S. C.; Potts, R. O.; Francoeur, M. L. Pharm. Res.
1992, 9, 1422.
17. Alexander, J.; Cargill, R.; Michelson, S. R.; Schwam, H.
J. Med. Chem. 1988, 31, 318.
9. Singh, P.; Roberts, M. S. J. Pharmacol. Exp. Ther. 1994,
268, 144.
18. Hegarty, A. F.; Frost, N. J. Chem. Soc., Perkin Trans. 2
1973, 1719.
10. Bonina, F. P.; Montenegro, L.; de Craparis, P.; Bousquet,
E.; Tirendi, S. Int. J. Pharm. 1991, 77, 21.
11. Bonina, F. P.; Montenegro, L.; Trapani, G.; Franco, M.;
Liso, G. Int. J. Pharm. 1995, 124, 45.
19. Taylor, H. E.; Sloan, K. B. J. Pharm. Sci. 1998, 87, 15.
20. Rautio, J.; Nevalainen, T.; Taipale, H.; Vepsalainen, J.;
Gynther, J.; Pedersen, T.; Jarvinen, T. J. Med. Chem. 2000, 43,
1489.
12. Rautio, J.; Taipale, H.; Gynther, J.; Vepsalainen, J.;
Nevalainen, T.; Jarvinen, T. J. Pharm. Sci. 1998, 87, 1622.
13. Rautio, J.; Nevalainen, T.; Taipale, H.; Vepsalainen, J.;
Gynther, J.; Pedersen, T.; Jarvinen, T. Pharm. Res. 1999, 16,
1172.
21. Cox, C.; Lectka, T. J. Org. Chem. 1998, 63, 2426.
22. Lopes, F.; Moreira, R.; Iley, J. Bioorg. Med. Chem. 2000,
8, 707.
23. Carvalho, E.; Francisco, A. P.; Iley, J.; Rosa, E. Bioorg.
Med. Chem. 2000, 8, 1719.
14. Wheeler, W. J.; Preston, D. A.; Wright, W. E.; Huffman,
G. W.; Osborne, H. E.; Howard, D. P. Med. Chem. 1978, 22,
657.
15. Gogate, U. S.; Repta, A. J.; Alexander, J. Int. J. Pharm.
1987, 40, 235.
16. Gogate, U. S.; Repta, A. Int. J. Pharm. 1987, 40, 249.
24. Iley, J.; Moreira, R.; Mendes, E. Pharm. Res. 1997, 14, 1634.
25. Taylor, H. E.; Sloan, K. B. J. Pharm. Sci. 1998, 87, 15.
26. Yano, T.; Nakagawa, A.; Masayoshi, T.; Noda, K. Life
Sci. 1986, 39, 1043.
27. Singh, P.; Roberts, M. S. J. Pharmacol. Exp. Ther. 1994,
268, 144.