The chemistry, toxicology, and identification in rat and human urine of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one: A reactive metabolite in felbamate bioactivation

Article abstract of DOI:10.1021/tx000139n

4-Hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one has been proposed to be a reactive metabolite of the anti-epileptic drug felbamate [Thompson et al. (1996) Chem. Res. Toxicol. 9, 1225-1229]. 4-Hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one exists in equilibrium with 3-oxo-2-phenylpropyl aminooate, which is known to eliminate to generate 2-phenylpropenal. Thus, this species is postulated to be a latent form of the ultimate reactive metabolite, 2-phenylpropenal. The chemistry of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one is proposed to parallel that of 4-hydroxycyclophosphamide, the bioactivated form of cyclophosphamide that undergoes ring-opening to aldophosphamide and subsequent elimination to afford 2-propenal (acrolein). The work presented here reports the chemical synthesis of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one and demonstrates that under buffered conditions it exists in equilibrium with 3-oxo-2-phenylpropyl aminooate. The rate-limiting step in the decomposition of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one is the irreversible β-elimination from 3-oxo-2-phenylpropyl aminooate to 2-phenylpropenal. We have found the half-life of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one to be 4.6 ± 0.4 h under in vitro conditions that mimic the physiological setting. As a consequence of the relatively long half-life of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one, we have sought evidence for the significance of this pathway in experimental and clinical conditions. We report here the observation of this metabolite in the urine of rats being treated with 3-hydroxy-2-phenylpropyl aminooate, the esterase-mediated metabolite of felbamate, and in the urine of patients undergoing felbamate therapy. In addition, we have shown that 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one is toxic to cultured ceils in a time-dependent manner, most likely as a result of its decomposition to 2-phenylpropenal. Taken together, the data support the hypothesis that 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one represents a "time-release" form of 2-phenylpropenal capable of traveling to distal sites from its locus of bioactivation and thereby mediates felbamate associated toxicities.

Full text of DOI:10.1021/tx000139n

958
Chem. Res. Toxicol. 2001, 14, 958-964
Th e Ch em istr y, Toxicology, a n d Id en tifica tion in Ra t a n d
Hu m a n Ur in e of
4-Hyd r oxy-5-p h en yl-1,3-oxa za p er h yd r oin -2-on e: A
Rea ctive Meta bolite in F elba m a te Bioa ctiva tion
Christine M. Dieckhaus,† Webster L. Santos,† R. Duane Sofia,‡ and
Timothy L. Macdonald*,†
Chemistry Department, University of Virginia, Charlottesville, Virginia 22901, and
Wallace Laboratories, Research and Development, Cranbury, New J ersey 08512
Received J une 30, 2000
4-Hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one has been proposed to be a reactive metabolite
of the anti-epileptic drug felbamate [Thompson et al. (1996) Chem. Res. Toxicol. 9, 1225-
1229]. 4-Hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one exists in equilibrium with 3-oxo-2-
phenylpropyl aminooate, which is known to eliminate to generate 2-phenylpropenal. Thus,
this species is postulated to be a latent form of the ultimate reactive metabolite, 2-phenylpro-
penal. The chemistry of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one is proposed to parallel
that of 4-hydroxycyclophosphamide, the bioactivated form of cyclophosphamide that undergoes
ring-opening to aldophosphamide and subsequent elimination to afford 2-propenal (acrolein).
The work presented here reports the chemical synthesis of 4-hydroxy-5-phenyl-1,3-oxazaper-
hydroin-2-one and demonstrates that under buffered conditions it exists in equilibrium with
3-oxo-2-phenylpropyl aminooate. The rate-limiting step in the decomposition of 4-hydroxy-5-
phenyl-1,3-oxazaperhydroin-2-one is the irreversible â-elimination from 3-oxo-2-phenylpropyl
aminooate to 2-phenylpropenal. We have found the half-life of 4-hydroxy-5-phenyl-1,3-
oxazaperhydroin-2-one to be 4.6 ( 0.4 h under in vitro conditions that mimic the physiological
setting. As a consequence of the relatively long half-life of 4-hydroxy-5-phenyl-1,3-oxazaper-
hydroin-2-one, we have sought evidence for the significance of this pathway in experimental
and clinical conditions. We report here the observation of this metabolite in the urine of rats
being treated with 3-hydroxy-2-phenylpropyl aminooate, the esterase-mediated metabolite of
felbamate, and in the urine of patients undergoing felbamate therapy. In addition, we have
shown that 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one is toxic to cultured cells in a time-
dependent manner, most likely as a result of its decomposition to 2-phenylpropenal. Taken
together, the data support the hypothesis that 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one
represents a “time-release” form of 2-phenylpropenal capable of traveling to distal sites from
its locus of bioactivation and thereby mediates felbamate associated toxicities.
identified N-acetylcysteine adducts of 2-phenylpropenal
which are excreted in the urine of rats and humans after
felbamate administration (6, 7). Given the reactivity of
2-phenylpropenal (t1/2 ∼ 40 s in the presence of 500 µM
glutathione), it is expected that 2-phenylpropenal will
react at the site of formation. Therefore, the bioactivation
of felbamate to 2-phenylpropenal in the liver could
explain the observed hepatotoxicity, but is unlikely to
account for the observed blood dyscrasias. We propose
that 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one rep-
resents a latent form of 2-phenylpropenal capable of
traveling to distal sites from the liver in the body.
In tr od u ction
Felbamate (1), 3-(aminocarbonyloxy)-2-phenylpropyl
aminooate, is an anti-epileptic drug useful in the treat-
ment of several forms of epilepsy (1, 2). In its first year
of widespread distribution, felbamate therapy was found
to be associated with an increase in the incidence of
aplastic anemia (3) and hepatotoxicity (4), resulting in
its limited use. Previous research from our laboratory has
shown that the primary metabolic pathway of felbamate
proceeds through 3-oxo-2-phenylpropyl aminooate (3) (5).
Under physiological conditions, 3-oxo-2-phenylpropyl
aminooate has been shown to undergo both spontaneous
cyclization to form 4-hydroxy-5-phenyl-1,3-oxazaperhy-
droin-2-one (6) and elimination to form 2-phenylpropenal
(5), the proposed ultimate reactive metabolite in felbam-
ate bioactivation (Scheme 1) (5). Additional research has
The oxazolidine, 4-hydroxy-5-phenyl-1,3-oxazaperhy-
droin-2-one, exists in equilibrium with 3-oxo-2-phenyl-
propyl aminooate, which undergoes elimination to form
2-phenylpropenal (5). We propose that the essential
features of the oxazolidine parallel the chemistry of
4-hydroxycyclophosphamide, which can undergo ring
opening and elimination to 2-propenal (acrolein) (8).
Studies have demonstrated that human serum albumin
(HSA)1 catalyzes the decomposition of 4-hydroxycyclo-
* Correspondence should be addressed to this author at the Depart-
ment of Chemistry, University of Virginia, McCormick Rd., Charlottes-
ville, VA 22901. Phone: 804-924-7718. E-mail: tlm@virginia.edu.
† University of Virginia.
‡ Wallace Laboratories, Research and Development.
10.1021/tx000139n CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/07/2001

A Latent Form of a Reactive Felbamate Metabolite
Chem. Res. Toxicol., Vol. 14, No. 8, 2001 959
Upon concentration, the ethyl acetate fractions afforded the
product as pristine white crystals in good yield; >98% purity
by HPLC. The product was isolated as a pair of diastereomers
in a ratio of 3 (major epimer):1 (minor epimer) as determined
by 1H NMR. APCI-MS: parent [M + 1H]+ m/z ) 194. APCI-
MS/MS: daughters [M + 1H]+ m/z ) 150 (loss of carbon dioxide)
and [M + 1H]+ m/z ) 133 (2-phenylpropenal). Elemental
analysis: theoretical C, 62.17; H, 5.74; N, 7.25; found C, 62.24;
H, 5.81; N, 7.16. 1H NMR (acetone-d6): δ (major epimer) 2.72
(1H, bs), 2.96-3.04 (1H, m), 4.24-4.33 (2H, m), 5.04-5.11 (1H,
d, J ) 5.39 Hz), 6.73 (1H, bs), 7.15-7.32 (5H, m); (minor epimer)
2.68 (1H, bs), 3.26-3.34 (1H, m), 4.38-4.46 (2H, m), 5.20-5.25
(1H, d, J ) 6.16 Hz), 6.73 (1H, bs), 7.15-7.32 (5H, m). 13C NMR
(acetone-d6): δ 43.90, 67.65, 77.75, 128.00, 128.14, 129.3, 137.33,
155.27.
Sch em e 1. Th e Meta bolism of F elba m a te
Syn th esis of 3-Oxo-2-p h en ylp r op yl Am in ooa te (3). The
aldehyde metabolite was synthesized from 3-hydroxy-2-phenyl-
propyl aminooate as follows. To a solution of 3-hydroxy-2-
phenylpropyl aminooate (100 mg, 0.51 mmol) in 10 mL of CH2Cl2
was added Dess-Martin periodinane (300 mg, 0.71 mmol) in
an aluminum foil covered round-bottom flask. After stirring for
2 h at room temperature, the mixture was separated im-
mediately using silica gel (ether, Rf ) 0.7). The solvent was
removed in vacuo to provide the product as solid white crystals
(94 mg, 95%). The product is stable at -80 °C, neat or as a
solution in dry acetone. At room temperature, the product
readily decomposes to 2-phenylpropenal and its dimerization
product, which is yellow in color. The 1H NMR (acetone-d6)
agreed with the previously published data (5). The compound
purity was determined to be g95% pure as determined by
HPLC/UV at λ ) 214 nm. A slight impurity of 2-phenylpropenal
was observed in the sample and background corrections are
included in the reported data.
phosphamide (9) and may offer an explanation for the
selective tissue toxicity of the oxazolidine. We hypothesize
that the oxazolidine may reach the blood as a stable
reservoir and “time release” form of 2-phenylpropenal
and capable of meditating the observed blood dyscrasias.
There has been no direct observation of the oxazolidine
in the blood or urine of patients being treated with
felbamate. In vitro studies were able to demonstrate the
generation of the oxazolidine by incubating 4-hydroxy-
2-phenylpropyl aminooate (2) with S9 human liver frac-
tions (10). In vivo work provided evidence for this
metabolite by indirect observation of the corresponding
oxazine dione (7) (11). The purpose of this work was to
better characterize the chemistry and toxicology of 4-hy-
droxy-5-phenyl-1,3-oxazaperhydroin-2-one and provide
direct evidence for its formation in vivo.
Th e p H -Dep en d en t Decom p osit ion of 4-H yd r oxy-5-
p h en yl-1,3-oxa za p er h yd r oin -2-on e (6). The oxazolidine (6,
100 µM) was incubated at 37 °C in 100 mM potassium
phosphate buffer at various pH units. For incubations with HSA
(fatty acid and globulin free; Sigma A-3782), HSA was added
to a final concentration of 60 mg/mL. At each time point, 50 µL
of the sample was removed and mixed with 50 µL of 100 µM
felbamate (1) as an internal standard. The amount of oxazolidine
remaining was determined by HPLC [Waters Symmetry C18
column 2.1 × 150 mm; 200 µL/min flow rate; 30% CH3CN:70%-
(0.1%)HOAc; UV detection, λ ) 214 nm]. The half-life was
determined using a first-order approximation and plotting ln
[6] versus time. For each pH value, the experiments were run
in triplicate.
Exp er im en ta l P r oced u r es
Ch em ica ls a n d In str u m en ts. The chemicals were pur-
chased from either Sigma (St. Louis, MO) or Aldrich (Milwau-
kee, WI) and were of the highest grade available. The urinalysis
was performed on a Waters 2690 Separations Module HPLC
using a Waters Symmetry C18 (2.1 × 150 mm) column. The flow
was directed into a Waters 486 Tunable absorbance detector
set at λ ) 214 nm and then into a Finnigan Mat LCQ ion trap
atmospheric chemical ionization (APCI)2 mass spectrometer.
NMR spectra were obtained on a 300 MHz General Electric
QE300 spectrometer, and chemical shifts are reported in ppm.
The cell proliferation assays were read on a Molecular Devices
Corporation Emax precision microplate reader set at λ ) 490
nm.
Syn th esis of 4-Hyd r oxy-5-p h en yl-1,3-oxa za p er h yd r oin -
2-on e (6). To a solution of 3-hydroxy-2-phenylpropyl aminooate
(2) (100 mg, 0.512 mmol) in CH2Cl2 was added Dess-Martin
periodinane (300 mg, 0.707 mmol). The mixture was stirred for
2 h and concentrated at room temperature. This reaction yielded
3-oxo-2-phenylpropyl aminooate (3), which was purified by flash
chromatography using silica gel with ether. The aldehyde (3)
was dissolved in acetone and an equal volume of 100 mM
potassium phosphate buffer (pH 8.0). The mixture was stirred
for 15 min and extracted with ethyl acetate, washed with brine,
and dried with Na2SO4. The crude mixture was purified by flash
chromatography using silica gel with ether, then ethyl acetate.
Th e Ra te-Lim itin g Step in 4-Hyd r oxy-5-p h en yltetr a h y-
d r o-1,3-oxa zin -2-on e Decom p osition . To determine the rate-
limiting step in 4-hydroxy-5-phenyltetrahydro-1,3-oxazin-2-one
decomposition, we trapped the aldehyde intermediate using
O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine hydrochloride at
pH 5. The trapping experiment was conducted at pH 5 because
of the pH optimum for the reaction. To 950 µL of phosphate
buffer, pH 5, at 37 °C was added 50 µL of 20 mM O-(2,3,4,5,6-
pentafluorobenzyl)hydroxylamine in ethanol and then 5 µL of
20 mM 4-hydroxy-5-phenyltetrahydro-1,3-oxazin-2-one. At each
time point, 50 µL of the sample was removed and mixed with
50 µL of 100 µM felbamate as an internal standard. The amount
of 6 remaining was determined by HPLC [Waters Symmetry
C18 column 2.1 × 150 mm; 200 µL/min flow rate; 30% CH3CN:
70%(0.1%)HOAc; UV detection, λ ) 214 nm]. The half-life was
determined using a first-order approximation and plotting ln
[6] versus time. For each pH value, the experiments were run
in triplicate.
Th e Decom p osition of 3-Oxo-2-p h en ylp r op yla m in ooa te
(3). The aldehyde carbamate was dissolved in acetone on dry
ice to a final concentration of 50 mM. The aldehyde carbamate
solution was then diluted into phosphate buffer at various pH
units at 37 °C to 350 µM. Immediately after diluting the
aldehyde into the buffer, an aliquot was removed, felbamate was
added as an internal standard (100 µM), and 15 µL was injected
1
HSA: human serum albumin.
APCI: atmospheric pressure chemical ionization.
2

960 Chem. Res. Toxicol., Vol. 14, No. 8, 2001
Dieckhaus et al.
onto the HPLC column. The amount of the oxazolidine and
2-phenylpropenal formed at ∼1 min was determined by HPLC
[Waters Symmetry C18 column 2.1 × 150 mm; 200 µL/min flow
rate; 30% CH3CN:70%(0.1%)HOAc; UV detection, λ ) 214 nm].
Calibration curves for the oxazolidine and 2-phenylpropenal
using felbamate as an internal standard were determined and
found to be linear. Under acidic conditions (pH 5) and about 1
min, a significant amount of 3-oxo-2-phenylaminooate remained,
and this suggests that the aldehyde and the oxazolidine had
not reached equilibrium. In contrast, under physiological condi-
tions, we did not observe the aldehyde after about 1 min of
incubation, which suggests that the aldehyde and oxazolidine
had reached equilibrium.
medium for 72 h, at which time the medium was removed and
the number of active cells was determined using the Promega
(Madison, WI) MTS cell proliferation assay. Briefly, the cells
were incubated with 100 µL of RPMI media (Life Technologies,
Gaithersburg, MD) containing 2.5 nmol of phenazime metho-
sulfate (PMS) and 66 nmol of 3-(4,5-dimethylthiazol-2-yl)-5-(3-
carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS)
for 1 h. MTS is bioreduced in the presence of PMS by dehydro-
genase enzymes in viable cells to a formazan, which absorbs
light at 490 nm. The absorbance at 490 nm is directly propor-
tional to the number of active cells. The GI50 for each drug was
determined by plotting the percent growth inhibition versus
drug concentration in the linear region, determining the linear
regression for each plot, and calculating the GI50 based on the
linear regression. For each compound, the assay was performed
in quadruplicate, and the data are reported as the average of
those assays ( the standard deviation.
Tim e-Dep en d en t Gr ow th In h ibition . The HEK293/SF
cells (as above) were plated into 96-well plates at a density of
1 × 104 cells per well and allowed to adhere for 24 h. After
24 h, the medium was removed and replaced with 50 µM drug
(6 or 5)-containing serum-free media. At each time point, the
drug-containing medium was removed, and the number of viable
cells was determined using the Promega MTS cell proliferation
assay described above. The assays were performed in quadru-
plicate.
Id en tifica tion of 4-Hyd r oxy-5-p h en yl-1,3-oxa za -p er h y-
d r oin -2-on e (6) in Ra t a n d Hu m a n Ur in e. The animal
experiments were conducted under a protocol approved by the
University of Virginia Animal Use and Care Committee (#2799-
05-98), and the patient urine was collected under a protocol
approved by the University of Virginia Human Investigations
Committee (#7310). Rat urine was collected continuously for 18
h post dose from 250 g Sprague-Dawley rats that had been
dosed chronically for 2 weeks with 70 mg kg-1 day-1 of
3-hydroxy-2-phenylpropyl aminooate or control rats that had
not been treated. The rat urine was stored at -60 °C until
thawed at 37 °C at the time of analysis. The human urine was
obtained from patients undergoing felbamate therapy and
analyzed about 18 h post collection. The control urine samples
were collected from adult volunteers not being treated with
felbamate. The human samples were stored at room tempera-
ture. The urine (3 mL for rat; 50 mL for human) was extracted
3 times with an equal volume of ethyl acetate and dried with
Na2SO4. The solvent was removed by rotary evaporation, and
the solid was dissolved immediately in 500 µL of acetone on dry
ice. The acetone solution (50 µL) was diluted into 150 µL of 0.1%
HOAc in water and analyzed by LC/MS/MS. Three MCF-dosed
rat urine samples, one blank rat urine sample, four patient urine
samples, and three blank urine samples were analyzed in total.
The reported results correspond to representative analysis. For
the coelution experiments, the oxazolidine synthetic standard
was added to the extracted sample to give a final concentration
of 10 µM, except in the reported MCF-dosed rat coelution, in
which case synthetic standard was added to a final concentra-
tion of 50 µM because of the high level of the oxazolidine present.
The sample (15 µL) was analyzed by HPLC (as above) and
detected with positive ion APCI-MS/MS (Finnigan LCQ). The
ion trap was set to trap the parent ion (m/z ) 194) with a 3.0
m/z isolation width. Fragmentation was achieved with 16%
collision energy using helium gas, and daughter ions were
detected in the full scan mode from 100 to 200 m/z. The tune
parameters were as follows: source voltage ) 4.14 kV; source
current ) 4.83 µA; vaporizer temperature ) 450 °C; sheath gas
(nitrogen) ) 34.9; auxiliary gas (nitrogen) ) 29.2; capillary
voltage ) 2.85 V; capillary temperature ) 150 °C.
Resu lts
Syn th esis of 4-Hyd r oxy-5-p h en yltetr a h yd r o-1,3-
oxa za in -2-on e. The oxazolidine was synthesized and
characterized as reported under Experimental Proce-
dures. The synthesis afforded the desired product and
provides a route to producing several grams of material.
We found the oxazolidine to be stable neat at room
temperature for several months and dissolved in aceto-
nitrile at -20 °C for several months. In aqueous solu-
tions, the oxazolidine readily equilibrates to 3-oxo-2-
phenylpropyl aminooate and eventually eliminates to
afford 2-phenylpropenal.
Th e p H-Dep en d en t Decom p osition of 4-Hyd r oxy-
5-p h en yl-1,3-oxa za p er h yd r oin -2-on e. The decomposi-
tion of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one to
2-phenylpropenal occurs spontaneously in aqueous solu-
tions. The decomposition demonstrates base catalysis and
has a half-life of 4.6 ( 0.4 h under physiological condi-
tions (Figure 1). Given time, the oxazolidine decomposes
to produce one product, 2-phenylpropenal. The addition
of physiologically relevant concentrations of HSA acceler-
ated the rate of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-
2-one decomposition to 50 ( 3 min when monitoring the
loss of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-2-one.
Formation of 2-phenylpropenal was not observed and is
thought to be binding to HSA (Table 1).
Th e Ra te-Lim itin g Step in th e Decom p osition of
4-H yd r oxy-5-p h en yl-1,3-oxa za p er h yd r oin -2-on e.
Trapping 3-oxo-2-phenylpropyl aminooate during the
decomposition of 4-hydroxy-5-phenyl-1,3-oxazaperhydroin-
2-one would allow for the determination of the rate-
limiting step in the decomposition of the oxazolidine. If
the ring opening from the oxazolidine to 3-oxo-2-phenyl-
propyl aminooate is the rate-determining step in its
decomposition, trapping the aldehyde carbamate should
not change the oxazolidine decomposition half-life. How-
ever, if the â-elimination from 3-oxo-2-phenylpropyl
aminooate to 2-phenylpropenal is the rate-limiting step
in the oxazolidine decomposition, trapping the aldehyde
carbamate should result in a decrease in the half-life of
Gr ow th In h ibition Assa ys (GI50). The cells used for the
growth inhibition studies were HEK-293/SF (human kidney)
obtained from the American Type Culture Collection (Manassas,
VA) that had been adapted to grow in serum-free media. The
cell line was selected because it was established as a serum-
free cell line. Serum culture conditions were avoided because
of the potential for serum proteins to react with the compounds
of interest. The cells were grown in Nephrigen Serum Free
Media obtained from Celox Laboratories, Inc. (St. Paul, MN),
at 37 °C in humidified air containing 5% carbon dioxide. For
the assays containing fetal bovine serum (FBS), FBS was added
to the Nephrigen Serum Free Media to a final concentration of
10% (v/v). The cells were plated into a 96-well plate at a density
of 1 × 104 cells per well and allowed to adhere to the plates for
24 h. After 24 h, the medium was removed and replaced with
media containing 25, 12.5, 10, 7.5, 5.0, or 2.5 µM of each drug
(6, 5, and 2-propenal) with less than 0.5% EtOH (v/v). Control
experiments showed that 0.5% EtOH has no effect on cell
survival. The cells were allowed to grow in the drug-containing

A Latent Form of a Reactive Felbamate Metabolite
Chem. Res. Toxicol., Vol. 14, No. 8, 2001 961
F igu r e 1. The pH-dependent decomposition of 4-hydroxy-5-
tetrahydro-1,3-oxazin-2-one (6). At pH 12, the oxazolidine half-
life was less than 15 min. Error bars are included for each data
point.
F igu r e 2. The pH-dependent product formation from 3-oxo-
2-phenylpropyl aminooate at ∼1 min.
aldehyde carbamate in the slightly basic pH region. The
newly reported data are consistent with general acid/base
catalysis for oxazolidine formation from 3-oxo-2-phenyl-
propyl aminooate and offer an explanation for the in-
creased half-life of the oxazolidine in the slightly alkali
region.
Ta ble 1. Th e Ha lf-Life of
4-Hyd r oxy-5-tetr a h yd r o-1,3-oxa zin -2-on e u n d er
P h ysiologica lly Releva n t Con d ition s
conditions
T1/2 (h)
100 mM K2HPO4, pH 7.5
100 mM K2HPO4, pH 7.5,
+ 60 mg of HSA/mL
4.6 ( 0.4
0.8 ( 0.1
Id en tifica tion of 4-Hyd r oxy-5-p h en yl-1,3-oxa za -
p er h yd r oin -2-on e in P a tien t a n d 3-Hyd r oxy-2-p h en -
ylp r op yl Am in ooa te Dosed Ra t Ur in e. Given the
relatively long half-life, we sought to identify 4-hydroxy-
5-phenyl-1,3-oxazaperhydroin-2-one in the urine of pa-
tients being treated with felbamate and rats that had
been dosed with the esterase-mediated felbamate me-
tabolite, 3-hydroxy-2-phenylpropyl aminooate. We sought
to identify this metabolite in rats that had been treated
with 3-hydroxy-2-phenylpropyl aminooate, as opposed to
felbamate, given the significant interspecies metabolism
differences (13). The oxazolidine metabolite was identi-
fied routinely in both the patient urine and rat urine by
LC/MS/MS coelution experiments. Under the conditions
applied, the oxazolidine metabolite elutes at 3.3 min and
produced characteristic MS/MS spectra with daughter
ions m/z ) 176 (loss of water), m/z ) 150 (loss of carbon
dioxide), m/z ) 133 (2-phenylpropenal), and m/z ) 132
(2-phenylpropenal preformed cation) (Figures 3a-c and
4a-c).
Th e Gr ow th In h ibition of 4-Hyd r oxy-5-p h en yl-
1,3-oxa za p er h yd r oin -2-on e on Cells in Cu ltu r e. The
oxazolidine metabolite, 4-hydroxy-5-phenyl-1,3-oxazaper-
hydroin-2-one, produced 50% growth inhibition at 13.0
( 1.4 µM when applied to HEK-293 cells as compared to
5.4 ( 1.8 µM for 2-phenylpropenal and 10.7 ( 1.7 µM
for 2-propenal. Control experiments showed that 0.5%
ethanol had no effect on growth inhibition.
Th e Tim e-Dep en d en t Gr ow th In h ibition of 4-Hy-
d r oxy-5-p h en yl-1,3-oxa za p er h yd r oin -2-on e a n d 2-
P h en ylp r op en a l. The growth inhibition of 4-hydroxy-
5-phenyl-1,3-oxazaperhydroin-2-one most likely results
from its decomposition to 2-phenylpropenal, and not as
a result of inherent properties. Within the first hour of
incubation on HEK-293 cells under serum-free condi-
tions, 2-phenylpropenal caused almost 100% growth
inhibition (Figure 5). The growth inhibition of the ox-
the oxazolidine. The trapping experiment caused a
decrease in the half-life of the oxazolidine at pH 5 from
172 ( 12 to 57 ( 9 h. The data support the hypothesis
that the oxazolidine and aldehyde carbamate exist in an
equilibrium that is rapid relative to the rate-limiting step
in the oxazolidine decomposition. The rate-limiting step
is the â-elimination reaction from the aldehyde carbam-
ate that affords 2-phenylpropenal. The kinetics governing
the oxazolidine decomposition mimic the kinetics ob-
served for the decomposition of 4-hydroxycylophospha-
mide in that 4-hydroxycyclophosphamide has also been
shown to exist in a rapid equilibrium with aldophospha-
mide with rate-limiting â-elimination affording 2-prope-
nal (12).
Th e Decom p osition of 3-Oxo-2-p h en ylp r op yl Am i-
n ooa te. The interesting increase in the half-life of the
oxazolidine metabolite in the slightly basic region
prompted us to address the decomposition rate of the
aldehyde carbamate, 3-oxo-2-phenylpropyl aminooate, to
afford the oxazolidine and 2-phenylpropenal. Previous
data had reported that the decomposition of the aldehyde
carbamate occurs in a constant 10:1 ratio (oxazolidine
to 2-phenylpropenal, respectively) throughout an acidic,
neutral, and basic pH region (5). We propose that the
increase in the half-life of the oxazolidine metabolite in
the slightly basic region occurs as a result of general acid/
base catalysis, which is not consistent with the previously
reported constant 10:1 ratio. Our results conflict with the
prior data and demonstrate that the decomposition of the
aldehyde carbamate does not decompose to the oxazoli-
dine metabolite and 2-phenylpropenal in a constant ratio
throughout the pH-buffered region examined (Figure 2).
The data reported here demonstrate that the oxazolidine
metabolite has a pH optimum for formation from the

962 Chem. Res. Toxicol., Vol. 14, No. 8, 2001
Dieckhaus et al.
F igu r e 4. (a) Blank human urine. Above left shows the LC
chromatograph with mass filter m/z ) 150 for a characteristic
daughter ion of 4-hydroxy-5-tetrahydro-1,3-oxazin-2-one (6)
(m/z ) 194). Above right shows the MS/MS spectrum (RT ) 3.29
min from LC chromatograph on left) with no detectable daugh-
ter ions characteristic for the oxazolidine metabolite. (b) Iden-
tification of 4-hydroxy-5-tetrahydro-1,3-oxazin-2-one (6) in pa-
tient urine extract. Data are displayed as in (a). Above right
shows the MS/MS spectrum (RT ) 3.28 on left) of daughters
m/z ) 176 (loss of water), m/z ) 158 (loss of CO2), m/z ) 150
(loss of CO2 and NH3), m/z ) 133 (2-phenylpropenal), and
m/z ) 132 (2-phenylpropenal preformed cation). (c) Human urine
extract with synthetic standard. Data are displayed as in (a).
The coelution of the synthetic standard with the patient urine
extract was indicated by an increase in the total ion count and
the same retention time, as indicated in (b).
F igu r e 3. (a) Blank rat urine. Above left shows the LC
chromatograph with mass filter m/z ) 150 for a characteristic
daughter ion of 4-hydroxy-5-tetrahydro-1,3-oxazin-2-one (6)
(m/z ) 194). Above right shows the MS/MS spectrum (RT ) 3.31
min from LC chromatograph on left) with no detectable daugh-
ter ions characteristic for the oxazolidine metabolite. (b) Iden-
tification of 4-hydroxy-5-tetrahydro-1,3-oxazin-2-one (6) in rat
urine extract. Data are displayed as in (a). Above right shows
the MS/MS spectrum (RT ) 3.30 min from LC chromatograph
on left) of daughters m/z ) 176 (loss of water), m/z ) 158 (loss
of CO2), m/z ) 150 (loss of CO2 and NH3), m/z ) 133 (2-
phenylpropenal), and m/z ) 132 (2-phenylpropenal preformed
cation). (c) Rat urine extract with synthetic standard. Data are
displayed as in (a). The coelution of the synthetic standard with
the rat urine extract was indicated by an increase in total ion
count and the same retention time, as shown in (b).
travelling to sites distal from the liver requires that it
“has a life” in vivo. To this end, we studied the half-life
of the oxazolidine throughout a phosphate buffer pH
range. We expected that both the ring opening of the
oxazolidine to 3-oxo-2-phenylpropyl aminooate and â-
elimination of the 3-oxo-2-phenylpropyl aminooate to
2-phenylpropenal would undergo general base catalysis.
Our results (Figure 1) have confirmed that the oxazoli-
dine, which exists in a rapid equilibrium with 3-oxo-2-
phenylpropyl aminooate, undergoes base-catalyzed de-
azolidine occurs over time and interestingly produces
about 50% growth inhibition at about 5 h (Figure 5)
or the half-life of 6 under physiological conditions
(Figure 1).
Discu ssion
The hypothesis that 4-hydroxy-5-phenyl-1,3-oxazaper-
hydroin-2-one, an oxazolidine metabolite of felbamate,
represents a latent form of 2-phenylpropenal capable of

A Latent Form of a Reactive Felbamate Metabolite
Chem. Res. Toxicol., Vol. 14, No. 8, 2001 963
phenylpropyl aminooate. We estimate that for 3-oxo-2-
phenylpropyl aminooate, the pKa of the carbamate proton
is ∼12 and the pKa of the R-carbon proton is ∼16 (based
on reported literature values in water) (14). The differ-
ences in pKa values explain the preference for cyclization
over elimination as demonstrated. Under physiological
conditions, 3-oxo-2-phenyl aminooate rapidly cyclizes,
generating the oxazolidine metabolite almost exclusively.
Given the equilibrium between the oxazolidine and 3-oxo-
2-phenylpropyl aminooate and the irreversible formation
of 2-phenylpropenal from 3-oxo-2-phenylpropyl amino-
oate, the oxazolidine completely decomposes to 2-phen-
ylpropenal over time. Under physiological conditions, the
oxazolidine demonstrates a half-life of 4.6 h, which is
sufficient time to travel to sites distal from the liver. The
oxazolidine could “have a life” in vivo.
The comparison of the oxazolidine metabolite to 4-hy-
droxycyclophosphamide led us to consider other possible
equilibrium species, including the dehydration of the
oxazolidine to afford an imine and the hydrolysis of 3-oxo-
2-phenylpropyl aminooate to afford the hydrate (12, 15-
17). The dehydration of carbinolamines depends on both
the protonation of the leaving group and the deprotona-
tion of the amine, resulting in optimum imine formation
under slightly alkali conditions (18). Maximal imine
formation is expected to occur under slightly basic
conditions and may contribute to the increase in the half-
life of the oxazolidine in this pH region. Hydrate forma-
tion of the aldehyde carbamate would be expected to
occur under acidic conditions; we hypothesize that hy-
drate formation may contribute to the stability of the
aldehyde carbamate under acidic conditions. We suspect
that each of these equilibrium species may contribute to
the overall in vivo stability of the oxazolidine.
Incubating the oxazolidine with HSA decreased the
oxazolidine half-life to 50 min (Table 1). The accelerated
decomposition of the oxazolidine in the presence of HSA
may provide insight into the hypothesis that the oxazo-
lidine represents a latent form of 2-phenylpropenal that
may be released selectively in the blood to mediate the
observed blood dyscrasias. In the same incubations, we
were unable to observe the formation of 2-phenylpropenal
from the oxazolidine metabolite and hypothesize that
2-phenylpropenal is binding to HSA. HSA contains a free
thiol at Cys-34 that has been shown to react with several
compounds including the bio-activated forms D-peni-
cillamine, captopril, benzene, and acetaminophen (19).
The reactive Cys-34 resides in a hydrophobic crevice of
depth 9.5-10 Å (19). We estimate that 2-phenylpropenal
is small enough to fit into this binding pocket and propose
that 2-phenylpropenal may undergo a Michael addition
to HSA. The alkylation of HSA by 2-phenylpropenal
represents a potential hapten that could illicit immune-
mediated toxicity, which is consistent with the hapten
hypothesis of idiosyncratic drug reactions. In cases of
other drug toxicities including nitrofurantoin and penicil-
lin, albumin-related antibodies have been observed (19).
Given the significant half-life of the oxazolidine, we
sought to identify this metabolite in the urine of rats
being treated with 3-hydroxy-2-phenylpropyl aminooate
and patients being treated with felbamate. The pH of
urine was found to be about 6 for rats and slightly more
acidic, or about 5, for humans. The oxazolidine was not
detected in control rat urine (Figure 3a) or control human
urine (Figure 4a) but was observed routinely in the urine
of rats treated with the esterase-mediated felbamate
F igu r e 5. The time-dependent cytotoxic effect of 4-hydroxy-
5-tetrahydro-1,3-oxazin-2-one (6) probably as a result of its
decomposition to 2-phenylpropenal (5). The compound 2-phen-
ylpropenal (5) demonstrates its maximal toxicity within 1 h of
drug exposure.
composition to 2-phenylpropenal. To determine the rate-
limiting step in the decomposition of the oxazolidine to
2-phenylpropenal, we studied the effect of trapping 3-oxo-
2-phenylpropyl aminooate on the half-life of the oxazo-
lidine decomposition. We trapped 3-oxo-2-phenylpropyl
aminooate using O-(2,3,4,5,6-pentafluorobenzyl)hydrox-
ylamine hydrochloride under acidic conditions, pH 5.
Under these conditions, trapping 3-oxo-2-phenylpropyl
aminooate decreased the half-life of the oxazolidine from
172 ( 12 to 57 ( 9 h. The observed decrease in the half-
life of the oxazolidine by trapping 3-oxo-2-phenylpropyl
aminooate is consistent with the proposal that the
oxazolidine and 3-oxo-2-phenylpropyl aminooate exist in
an equilibrium that is rapid relative to the â-elimination,
and that the â-elimination affording 2-phenylpropenal is
rate-limiting.
The increase in the oxazolidine half-life in the slightly
basic region was thought to result from the cyclization
kinetics of 3-oxo-2-phenylpropyl aminooate as governed
by general acid/base catalysis (Figure 1). However, given
this proposal, the cyclization should demonstrate a pH
optimum in the slightly basic region. This proposal was
inconsistent with previous data, which demonstrated that
upon incubation throughout a pH-buffered range, 3-oxo-
2-phenylpropyl aminooate cyclizes to form the oxazolidine
and eliminates to form 2-phenylpropenal in a constant
10:1 ratio, respectively (5). Since the previous data were
inconsistent with our hypothesis regarding the kinetics
of the oxazolidine decomposition, we repeated the study
to monitor the formation of the oxazolidine and 2-phen-
ylpropenal from 3-oxo-2-phenylpropyl aminooate at vari-
ous pH units. Our results conflict with the previously
published data and demonstrate that the ratio of the
oxazolidine to 2-phenylpropenal formation is not a con-
stant ratio (Figure 2). The data demonstrate general acid/
base catalysis for the oxazolidine formation from 3-oxo-
2-phenylpropyl aminooate and general base catalysis for
the â-elimination of 3-oxo-2-phenylpropyl aminooate to
2-phenylpropenal. The increase in the half-life of the
oxazolidine in the slightly basic region reflects a pH
optimum for the oxazolidine formation from 3-oxo-2-

964 Chem. Res. Toxicol., Vol. 14, No. 8, 2001
Dieckhaus et al.
metabolite, 3-hydroxy-2-phenylpropyl aminooate (Figure
3b,c), and in felbamate patient urine samples (Figure
4b,c). While another laboratory has demonstrated the
formation of the oxazolidine in vitro (10), these data
represent the first direct evidence for its formation in
vivo.
Ack n ow led gm en t. We thank Wallace Laboratories
for generous financial support for this research.
Refer en ces
(1) Fraser, A. D. (1996) New drugs for the treatment of epilepsy. Clin.
Biochem. 29, 97-110.
To further characterize the oxazolidine, we examined
its toxicity in vitro. We determined a GI50 for the
oxazolidine on HEK-293 cells cultured under serum-free
conditions. Serum-free conditions were employed due to
the high reactivity of 2-phenylpropenal and the possibil-
ity of protein binding. While not quite as toxic as
2-phenylpropenal (GI50 ) 5.4 ( 1.8 µM) or 2-propenal
(GI50 ) 10.7 ( 1.7µM), the oxazolidine (GI50 ) 13.0 (
1.4 µM) demonstrates toxicity within the same order of
magnitude. It is worth noting that 2-phenylpropenal
appears to be slightly more toxic than acrolein probably
as a consequence of increased electrophilicity introduced
by the addition of the phenyl ring. The increased toxicity
of 2-phenylpropenal as compared to acrolein is consistent
with previously published results on HA1 cells (5). Since
kidney cells are metabolically active, the slight increase
in GI50 for the oxazolidine is likely to result in its
oxidation to tetrahydro-1,3-oxazin-2,4-dione (7) or oxida-
tion of 3-oxo-2-phenylpropyl aminooate to 3-carbamoyl-
2-phenylpropionic acid. To study the possible effect of
protein binding, the same experiments were conducted
with media supplemented with fetal bovine serum (FBS)
to 10% (v/v). Interestingly, the addition of FBS had no
effect on the GI50 of either the oxazolidine or the 2-phen-
ylpropenal.
To support the hypothesis that the toxicity of the
oxazolidine results from its decomposition to 2-phenyl-
propenal, we conducted time-dependent growth inhibition
experiments. The time-dependent growth inhibition study
was able to demonstrate that 2-phenylpropenal causes
almost 100% growth inhibition within the first hour of
drug exposure (Figure 5). The results indicate that
2-phenylpropenal-induced growth inhibition is not cell
cycle dependent. The growth inhibition of the oxazolidine
was found to be time-dependent and demonstrates 50%
growth inhibition between 4 and 5 h of incubation.
Furthermore, the time-dependent growth inhibition data
correlate nicely with the observed half-life of the oxazo-
lidine under physiological conditions (∼4.6 h) (Figure 1).
While the data are not conclusive, they are consistent
with the hypothesis that the oxazolidine-mediated growth
inhibition results as a consequence of its decomposition
to 2-phenylpropenal.
(2) Leppik, I. E. (1996) Felbamate. in The Treatment of Epilepsy
(Shorvon, S., Dreifuss, F., Fish, D., and Thomas, D., Eds.) pp 421-
428, Blackwell Science, New York.
(3) Pennell, P. B., Ogaily, M. S., and Macdonald, R. L. (1995) Aplastic
anemia in
a patient receiving felbamate for complex partial
seizures. Neurology 45, 456-460.
(4) O’Neil, M. G., Perdun, C. S., Wilson, M. D., McGown, S. T., and
Patel, S. (1996) Felbamate-associated fatal acute hepatic necrosis.
Neurology 46, 1457-1459.
(5) Thompson, C. D., Kinter, M. T., and Macdonald, T. L. (1996)
Synthesis and in vitro reactivity of 3-carbamoyl-2-phenylpro-
pionaldehyde and 2-phenylpropenal: putative reactive metabo-
lites of felbamate. Chem. Res. Toxicol. 9, 1225-1229.
(6) Thompson, C. D., Gulden, P. H., and Macdonald, T. L. (1997)
Identification of modified atropaldehyde mercapturic acids in rat
and human urine after felbamate administration. Chem. Res.
Toxicol. 10, 457-462.
(7) Thompson, C. D., Barthen, M. T., Hopper, D. W., Miller, T. A.,
Quigg, M., Hudspeth, C., Montouris, G., Marsh, L., Perhach, J .
L., Sofia, R. D., and Macdonald, T. L. (1999) Quantification in
patient urine samples of felbamate and three metabolites: acid
carbamate and two mercapturic acids. Epilepsia 40, 769-776.
(8) Sladek, N. E. (1988) Metabolism of oxazphosphorines. Pharmacol.
Ther. 37, 301-355.
(9) Kwon, C., Maddison, K., LoCastro, L., and Borch, R. F. (1987)
Accelerated decomposition of 4-hydroxycyclophosphamide by hu-
man serum albumin. Cancer Res. 47, 1505-1508.
(10) Kapetanovic, I. M., Torchin, C. D., Thompson, C. D., Miller, T.
A., McNeilly, P. J ., Macdonald, T. L., Kupferberg, H. J ., Perhach,
J . L., Sofia, R. D., and Strong, J . M. (1998) A potentially reactive
cyclic carbamate metabolite of the anti-epileptic drug felbamate
produced by human liver tissue in vitro. Drug Metab. Dispos. 26,
1089-1085.
(11) Thompson, C. D., Miller, T. A., Barthen, M. T., Dieckhaus, C. M.,
Sofia, R. D., and Macdonald, T. L. (2000) The synthesis, in vitro
reactivity, and evidence for formation in humans of 5-phenyl-1,3-
oxazinane-2,4-dione, a metabolite of felbamate. Drug Metab.
Dispos. 28, 434-439.
(12) Borch, R. F., Hoye, T. R., and Swanson, T. A. (1984) In situ
preparation of cis- and trans-4-hydroxycyclophosphamide with
aldophosphamide and its hydrate in aqueous solution. J . Med.
Chem. 27, 490-494.
(13) Dieckhaus, C. M., Miller, T. A., Sofia, R. D., and Macdonald, T.
L. (2000) A mechanistic approach to understanding species
differences in felbamate bioactivation: relevance to drug-induced
idiosyncratic reactions. Drug Metab. Dispos. 28, 814-822.
(14) Lide, D. R., Ed. (1999) CRC Handbook of Chemistry and Physics,
80th ed., pp 8-46-8-56, CRC Press, New York.
(15) Fenselau, C., Lehman, J . P., Myles, A., Brandt, J ., Yost, G. S.,
Friedman, O. M., and Colvin, O. M. (1982) Iminocyclophospha-
mide as a chemically reactive metabolite of cyclophosphamide.
Drug Metab. Dispos. 10, 636-640.
The chemistry, toxicology, and identification of the
oxazolidine in vivo support our hypothesis that it repre-
sents a latent precursor of 2-phenylpropenal capable of
travelling to sites distal from the liver in the body. We
believe these characteristics of the oxazolidine may
contribute importantly to the observed felbamate-induced
blood dyscrasias. The commonalities between the chem-
istry of the oxazolidine and 4-hydroxycyclophosphamide
offer insights into the reactivity of compounds with
similar structural motifs and possibly to their selective
blood cell toxicity.
(16) Borch, R. F., and Getman, K. M. (1984) Base-catalyzed hydrolysis
of 4-hydroxycyclophosphamide: evidence for iminocyclophospha-
mide as an intermediate. J . Med. Chem. 27, 485-490.
(17) Borch, R. F., and Millard, J . A. (1987) The mechanism of
activation of 4-hydroxycyclophosphamide. J . Med. Chem. 30, 427-
431.
(18) Sayer, J . M., Peskin, M., and J encks, W. P. (1973) Imine-forming
elimination reactions. I. General base and acid catalysis and
influence of the nitrogen substituent on rates and equilibria for
carbinolamine dehydration. J . Am. Chem. Soc. 95, 4278-4287.
(19) Peters, T. (1996) All About Albumin, Academic Press, New York.
TX000139N