A novel prodrug strategy for β-dicarbonyl carbon acids: Syntheses and evaluation of the physicochemical characteristics of C-phosphoryloxymethyl (POM) and phosphoryloxymethyloxymethyl (POMOM) prodrug derivatives

Article abstract of DOI:10.1002/jps.22021

The C-phosphoryloxymethyl (POM) and phosphoryloxymethyloxymethyl (POMOM) prodrugs resulting from derivatization at the reactive α-carbon of β-dicarbonyl carbon acid drugs represent a unique approach for improving their chemical stability and aqueous solubility. This work evaluates the physicochemical and in vitro enzymatic bioconversion lability of selected prodrugs of phenylbutazone and phenindione. The POM and POMOM prodrug derivatives of phenylbutazone are highly water soluble (≥250 mg/mL), chemically stable with projected shelf-lives of 4.5 years (pH 3.5, 258C) and 1.1 years (pH 6.0, 25°C), respectively. Interestingly, both prodrug derivatives do not display a pH-dependency typical of many phosphate monoesters, although the similarities of their apparent thermodynamic activation parameters indicate a hydrolysis mechanism similar to other phosphates. These prodrugs undergo alkaline phosphatases catalyzed bioconversion to their respective carbon acids with an expected faster rate exhibited by the POMOM derivatives. Additionally, in marked contrast to the oxidative instability of phenindione, its POM prodrug is stable. The results from these studies reaffirm the rationale of transiently "masking" the reactive a-carbon/proton bond by covalently incorporating a POM or POMOM promoiety. This prodrug strategy presents a twofold advantage, enhancement of aqueous solubility and prevention of oxidative instability, two intrinsic formulation limitations found for β-dicarbonyl carbon acid drugs.

Full text of DOI:10.1002/jps.22021

A Novel Prodrug Strategy for b-Dicarbonyl Carbon Acids:
Syntheses and Evaluation of the Physicochemical
Characteristics of C-Phosphoryloxymethyl (POM) and
Phosphoryloxymethyloxymethyl (POMOM) Prodrug Derivatives
SUNDEEP S. DHARESHWAR, VALENTINO J. STELLA
Department of Pharmaceutical Chemistry, University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047
Received 1 April 2009; revised 30 July 2009; accepted 16 October 2009
Published online 17 December 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.22021
ABSTRACT: The C-phosphoryloxymethyl (POM) and phosphoryloxymethyloxymethyl
(POMOM) prodrugs resulting from derivatization at the reactive a-carbon of b-dicarbonyl
carbon acid drugs represent a unique approach for improving their chemical stability and
aqueous solubility. This work evaluates the physicochemical and in vitro enzymatic bioconver-
sion lability of selected prodrugs of phenylbutazone and phenindione. The POM and POMOM
prodrug derivatives of phenylbutazone are highly water soluble (ꢀ250 mg/mL), chemically
stable with projected shelf-lives of 4.5 years (pH 3.5, 258C) and 1.1 years (pH 6.0, 258C),
respectively. Interestingly, both prodrug derivatives do not display a pH-dependency typical
of many phosphate monoesters, although the similarities of their apparent thermodynamic
activation parameters indicate a hydrolysis mechanism similar to other phosphates. These
prodrugs undergo alkaline phosphatases catalyzed bioconversion to their respective carbon
acids with an expected faster rate exhibited by the POMOM derivatives. Additionally, in marked
contrast to the oxidative instability of phenindione, its POM prodrug is stable. The results from
these studies reaffirm the rationale of transiently ‘‘masking’’ the reactive a-carbon/proton bond
by covalently incorporating a POM or POMOM promoiety. This prodrug strategy presents a
twofold advantage, enhancement of aqueous solubility and prevention of oxidative instability,
two intrinsic formulation limitations found for b-dicarbonyl carbon acid drugs. ß 2009 Wiley-Liss,
Inc. and the American Pharmacists Association J Pharm Sci 99:2711–2723, 2010
Keywords: b-dicarbonyl carbon acids; phosphate prodrug derivatives; in vitro bioconversion;
pKa; pH-rate profile
INTRODUCTION
From a prodrug perspective, transient masking of
this reactive carbon/proton bond would seem to be an
attractive exercise. Incorporating an ionizable func-
tional group in the promoiety such as a phosphoryl-
oxymethyl (POM) or a phosphoryloxymethyloxymethyl
(POMOM) group might not only prevent oxidation
but also increase aqueous solubility of poorly soluble
b-dicarbonyl carbon acids (intrinsic solubility of
phenylbutazone is 8 mg/mL).3
The design of fosphenytoin (11, see Fig. 1), a
commercially successful prodrug of phenytoin (9), in
the early 1980s demonstrated the use of the POM
promoiety in prodrugs of hydantoins. This has been
extended to alcohols, phenols as recently shown with
the approval of fospropofol (Lusedra1).4 Of special
interest to us has been the evaluation of an
‘‘extended’’ POM, in which an additional molecule
of formaldehyde is incorporated into the promoiety. In
this POMOM promoiety, the phosphate group is
further extended from the parent drug giving
The acidity of the carbon–hydrogen bond in b-
dicarbonyl carbon acids is caused by a combination
of inductive electron-withdrawing effect of the
neighboring b-dicarbonyl groups and resonance
stabilization of the carbanion formed by the removal
of the a-carbon proton. Moreover, the weak carbon–
hydrogen bond enhances the reactivity of the
a-carbon to autoxidation, as observed for phenyl-
butazone, generating 4-hydroxyphenylbutazone, its
oxidative degradant both in solution and in the solid
state.1,2
Sundeep S. Dhareshwar’s present address is Technical R&D,
Pharmaceutical Development, Novartis Pharmaceutical Corpora-
tion, One Health Plaza, East Hanover, NJ 07936.
Correspondence to: Valentino J. Stella (Telephone: 785-864-
3755; Fax: 785-864-5736; E-mail: stella@ku.edu)
Journal of Pharmaceutical Sciences, Vol. 99, 2711–2723 (2010)
ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010
2711

2712
DHARESHWAR AND STELLA
Figure 1. POM (2a,b,10) and POMOM (3a,b,11) deriva-
tives of phenylbutazone (1a) and phenindione (1b), two
b-dicarbonyl carbon acids, and phenytoin (9), a cyclic imide.
improved accessibility and thus facilitating enzy-
matic bioconversion to the parent drug.
The objective of this article is to report the
syntheses and physicochemical evaluation of both
the C-POM and POMOM prodrugs as an approach to
improving chemical stability and aqueous solubility
of b-dicarbonyl carbon acid model drugs. This
strategy is illustrated in Scheme 1. The b-dicarbonyl
carbon acid drugs, 1a or 1b (see Fig. 1 for complete
structures), are chemically modified to generate
either the POM, 2a, or POMOM, 3a, prodrug
derivatives. The sequential in vitro or in vivo
bioconversion process of these prodrugs is initiated
in a rate-limiting alkaline phosphatase catalyzed
oxygen–phosphorus bond hydrolysis generating inor-
ganic phosphate, 4, and an intermediate, which then
spontaneously breaks down to release formaldehyde
and the parent drug.5 Particularly unusual here is the
fact that reversion requires the breaking of a carbon–
carbon bond.
Scheme 1. The derivatization and alkaline phosphatase
catalyzed bioconversion of the POM and POMOM prodrug
derivatives of b-dicarbonyl carbon acids.
carbonate, acetone, sodium phosphate (monobasic
and dibasic), ethyl acetate, n-hexanes, hydrochloric
acid, and methanol were purchased from Fisher
Scientific (Pittsburgh, PA). Normal phase silica gel,
particle size 32–63 mm, was obtained from Selecto
Scientific (Norcross, GA), while TLC plates (1000 mm
thickness, 20 cm ꢁ 20 cm) coated with silica gel
60 F254 were obtained from Aldrich Chemical
Company (Milwaukee, WI). Anhydrous tetrahydro-
furan and diethyl ether were freshly prepared by
distilling commercially available tetrahydrofuran
and diethyl ether over sodium hydride and benzo-
phenone under argon atmosphere. Anhydrous acet-
onitrile and dichloromethane were freshly prepared
by distilling commercially available acetonitrile and
dichloromethane over calcium hydride under argon
atmosphere. The water for buffer solutions was
deionized and charcoal filtered prior to distillation
from an all glass still. All other chemicals and
solvents were of reagent grade obtained from
commercial sources and used without purification.
EXPERIMENTAL
Materials
Anhydrous acetonitrile, acetone-d6, benzophenone,
chloroform-d3, Celite1, dibenzyl phosphate, diethyl
ether, dimethyl sulfoxide, deuterium oxide, human
placental alkaline phosphatase type XVII, hydrogen
peroxide (50% solution), glycine, methylsulfoxide-d6,
phosphorus oxychloride, 10% palladium–carbon,
phenindione, phenylbutazone, N-iodosuccinimide,
sodium chloride, sodium sulfate, tetrahydrofuran,
and tetra-butyl ammonium dihydrogen phosphate
(TBAH2P, 1.0 M solution in water) were obtained
from Sigma–Aldrich Chemical Company (Milwaukee,
WI). Sodium hydroxide (1.0 N solution), potassium
Analytical Equipment
Nuclear magnetic resonance spectra were performed
using a Bruker Avance 400 or an AM 500 instruments
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010
DOI 10.1002/jps

PRODRUG STRATEGY FOR b-DICARBONYL CARBON ACIDS
2713
and J values are reported in Hertz. Chemical shifts
are reported as parts per million (ppm) downfield
from tetramethylsilane as an internal standard for
1H NMR and from 85% H3PO4 as an external
standard in 31P NMR. The following abbreviations
are used for describing 1H NMR splitting patterns: s,
singlet; d, doublet; t, triplet; q, quartet; m, multiplet;
and br, broad. UV spectra were obtained using a
Perkin Elmer Lambda 6 UV/Vis spectrophotometer
(Norwalk, CT). Mass spectral analyses were per-
formed on a VG-ZAB (VG Analytical Ltd, Manchester,
UK) by the Mass Spectral Laboratory at the
University of Kansas (Lawrence, KS). Water content
was estimated on a Metrohm AG Karl Fisher
coulometer (Herisam, Switzerland).
The HPLC system (Shimadzu Scientific Instru-
ments, Columbia, MD) used for analysis in all these
studies consisted of SCL 10A system controller, two
LC-10AT pumps, a SPD 10A UV/Vis detector, and a
SIL 10A autoinjector. The injection volume was
20 mL, and the column oven was maintained at
408C. Samples were analyzed using a Phenomenex1
(Torrance, CA) HypercloneTM C18 reverse phase
analytical column (4.6 mm i.d. ꢁ 150 mm) with a
mean particle diameter of 5 mm.
For analysis of 1a and 2a, the mobile phase
consisted of acetonitrile (45%, v/v) and a 10 mM
phosphate buffer (55%, v/v, containing 16 mM
TBAH2P) adjusted to pH 6.8, and pumped at a flow
rate of 0.9 mL/min. The detection was performed by
absorbance at 240 nm, with retention times of 4.3, 3.6,
and 7.7 min for 1a, 2a, and 13, respectively. For
experiments involving 1a and 3a, the mobile phase
consisted of acetonitrile (40%, v/v) and a 10 mM
phosphate buffer (60%, v/v, containing 16 mM
TBAH2P) adjusted to pH 6.8, and flow rate of
1.0 mL/min. The detection was performed by absor-
bance at 240 nm, with retention times of 5.5, 6.2, and
9.1 min for 1a, 3a, and 13, respectively.
For the analysis of 1b, 2b, 3b, and 12, the mobile
phase consisted of acetonitrile (35%, v/v) and a 10 mM
phosphate buffer (65%, v/v, containing 16 mM
TBAH2P) adjusted to pH 6.8, and pumped at a flow
rate of 1.0 mL/min. The detection was performed by
absorbance at 228 nm, with retention times of 8.4, 4.3,
5.4, and 4.8 min, respectively.
the cake washed with cold water and then dried in
vacuo.
4-Butyl-4-Hydroxymethyl-1,2-Diphenylpyra-
zolidine-3,5-Dione (5a). Eighty-six percent yield,
white solid. 1H NMR (CDCl3, 400 MHz): d 0.9 (t,
J ¼ 6.3 Hz, –CH3), 1.3 (d, J ¼ 10.47 Hz, –CCH2CH2
C–), 1.97 (d, J ¼ 7.9 Hz, –CCH2C–), 4.1 (s, 2H,
–CCH2O), 6.9 (m, 2H, arom.), 7.35–7.37 (m, 8H, arom).
13C NMR (CDCl3, 500MHz): d 13.54, 22.58, 26.45,
31.15, 56.96, 66.01, 122.86, 126.90, 128.93, 135.57,
172.60. MS (EI): m/z (relative intensity) 338 (Mþ, 75).
2-(Hydroxymethyl)-2-Phenylindene-1,3-
Dione (5b). Ninety-one percent yield, pinkish-white
solid. 1H NMR (CDCl3, 400 MHz): d 4.4 (s, 2H,
–CCH2O), 7.3–7.4 (m, 2H), 7.5 (m, 3H), 7.9 (dd,
J ¼ 5.6, 2.9 Hz, 2H), 8.1 (dd, J ¼ 5.55, 3.05 Hz, 2H).
13C NMR (CDCl3, 500 MHz): d 65.97, 123.54, 128.15,
128.92, 133.18, 135.88, 142.42, 200.47. MS (EI): m/z
(relative intensity) 253 (Mþ, 80).
Disodium(4-Butyl-3,5-Dioxo-1,2-Diphenylpyr-
azolidin-4-yl)Methyl Phosphate (2a). To a stirred
solution of 5a (2.0 g, 5.88 mmol) in anhydrous
methylene chloride (75 mL) maintained under argon
atmosphere in ice-bath was added dropwise pyridine
(0.52 mL, 6.46 mmol, 1.1 equiv.). Solution was stirred
at reduced temperature for 0.5 h, followed by drop-
wise addition of phosphorus oxychloride (1.76 mL,
17.64 mmol, 3 equiv.). Solution was stirred at room
temperature until complete loss of starting compound
was confirmed using a TLC plate (n-hexanes/ethyl
acetate, 60/40, v/v, for 5a, Rf ¼ 0.34). After 2 h of
stirring, the solvent was removed in vacuo and water
(10 mL) was added to the residue, followed by
dropwise addition of 1 N NaOH to raise pH to
approximately 10. The reaction mixture was parti-
tioned between ethyl acetate (25 mL) and water
(25 mL). The aqueous layer is collected and acidified
using 1 N HCl. An ethyl acetate/aqueous extraction
was performed, the organic layer was collected and
removed in vacuo. Conversion of the free acid to
its disodium salt was done by adding NaOH
(2.1 Molar Equivalents) to the solution of free acid
dissolved in approximately 5 mL of water and 1 mL of
ethanol. This was followed by the dropwise addition of
cold ether until precipitation was complete. The
resulting disodium salt was collected by filtration,
purified by recrystallization, and lyophilized to give
2a as a white powder; 57.2% yield. 1H NMR
(D2O, 400 MHz): d 0.78 (t, J ¼ 6.30, 6.30 Hz, –CH3),
1.27 (d, J ¼ 10.47 Hz, 4H, –CCH2CH2C–), 1.79 (d,
J ¼ 7.90 Hz, 2H), 4.04 (s, 2H, –CCH2OP–), 7.21–7.28
(m, 2H, p-arom.), 7.31 (t, J ¼ 7.44, 7.44 Hz, 4H), 7.37
(d, J ¼ 7.75 Hz, 4H, arom.). 31P NMR (D2O, 162 MHz):
d 2.5 (s, 1P). MS (ESþ): m/z (relative intensity) 463.2
Synthetic Procedures
General Procedure for the Hydroxymethyl Derivatives
of Carbon Acids (5a,b)
To a solution of b-dicarbonyl carbon acid (35 mmol) in
water (100 mL) containing 10% (v/v) ethanol was
added formalin (119 mmol, 3.4 equiv.) and potassium
carbonate (3.5 mmol, 0.1 equiv.), and stirred at room
temperature for 24 h. The resulting suspension is
passed through a sintered glass filter (medium), and
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010

2714
DHARESHWAR AND STELLA
((M þ H)þ, 100). HRMS (M þ H)þ, C20H22N2Na2
O6P: anal. calculated, 463.1011; found, 463.1003,
0.008 amu.
((M þ H)þ, 75). HRMS (M þ H)þ, C22H27N2SO3: anal.
calculated, 399.1742; found, 399.1745, 0.003 amu.
2-((Methylthiomethoxy)Methyl)-2-Phenylin-
Disodium(1,3-Dioxo-2-Phenylinden-2yl)-
Methyl Phosphate (2b). To a stirred solution of 5b
(2.0 g, 7.92 mmol) in anhydrous methylene chloride
(75 mL) maintained under argon atmosphere in ice-
bath was added dropwise pyridine (0.7 mL, 8.7 mmol,
1.1 equiv.). Solution was stirred at reduced tempera-
ture for 0.5 h, followed by dropwise addition of
dene-1,3-Dione (7b). Seventy-seven percent yield,
1
distinct odor. H NMR (acetone-d6, 400 MHz): d 1.61
(s, 3H), 4.30 (d, J ¼ 7.76 Hz, 2H), 4.52 (s, 2H), 7.39–
7.26 (m, 3H), 7.40–7.51 (m, 2H), 7.85–7.94 (m, 2H),
8.03–8.11 (m, 2H). MS (FABþ in TG/G): m/z (relative
intensity) 312.3 (Mþ, 50).
phosphorus oxychloride
(3.6 mL,
23.76 mmol,
3 equiv.). Stir solution at room temperature till
complete loss of starting compound is confirmed
using a TLC plate (n-hexanes/ethyl acetate 70/30, v/v,
for 5b, Rf ¼ 0.14). The workup for 2b is similar to
that described above for 2a. The disodium salt 2b
General Procedure for the Dibenzyl Protected
Phosphate Esters (8a,b)
To a stirred solution of 7a or 7b (15 mmol) in
anhydrous THF (50 mL) was added N-iodosuccini-
mide (2.0 equiv., 30 mmol) and dibenzyl phosphate
(3.0 equiv., 45 mmol). After 12 h, the mixture was
filtered and concentrated in vacuo. To this residue
was added dichloromethane (50 mL) and aqueous
solution of sodium thiosulfate (50 mL) and allowed to
stir at room temperature for 30 min. The organic layer
was separated, washed with brine, filtered, dried over
sodium sulfate, and then concentrated in vacuo. The
viscous reddish residue was chromatographed on
normal phase silica gel (approximately 100 g) with n-
hexanes/ethyl acetate (60/40) for eluent. Fractions
containing 8a (Rf ¼ 0.3) or 8b (Rf ¼ 0.35) were
combined and concentrated in vacuo to afford the
desired products as oil.
is obtained as
a white powder; 75.4% yield.
1H NMR (D2O, 400 MHz): d 4.45 (d, J ¼ 3.06 Hz, 2H,
–CCH2OP–), 7.13–7.22 (m, 2H), 7.27 (m, 3H), 7.93
(dd, J ¼ 5.68, 2.92 Hz, 2H), 8.01 (dd, J ¼ 5.55, 3.05 Hz,
2H). 31P NMR (D2O, 400 MHz): d 3.1 (s, 1P). MS (ESþ):
m/z (relative intensity) 377 ((M þ H)þ, 50). HRMS
(M þ H)þ, C16H11Na2O6P: anal. calculated, 377.0167;
found, 377.0170, 0.0003 amu.
General Procedure for the Methylthiomethoxymethyl
Adduct of Carbon Acids (7a,b)
To a stirred suspension of 5a,b (20 mmol) in acetic
anhydride (10 mL) and acetic acid (2 mL), maintained
under argon atmosphere, was added anhydrous
DMSO (20 mL). The reaction mixture was stirred at
room temperature for 24 h. To this solution is added
an aqueous solution of NaHCO3 dropwise, and
contents let to stir for further 0.5 h. An organic-
aqueous extraction was performed using dichloro-
methane (50 mL). The organic phase was washed
with brine, filtered, dried over sodium sulfate, and
concentrated in vacuo. The yellowish oily residue was
chromatographed on normal phase silica gel (approx.
100 g) with n-hexanes/ethyl acetate (80/20) for eluent.
Fractions containing 7a (Rf ¼ 0.58) or 7b (Rf ¼ 0.36)
were combined and concentrated in vacuo, followed
by recrystallization from n-hexanes to yield the
product as a yellowish powder.
Dibenzyl((4-Butyl-3,5-Dioxo-1,2-Diphenylpyr-
azolidin-4-yl)Methoxy)Methyl Phosphate (8a).
61.1% yield, reddish viscous oil. 1H NMR (acetone-d6,
400 MHz): d 0.87–0.99 (m, 3H), 1.39 (dd, 4H, J ¼ 7.13,
3.44 Hz), 1.83 (dd, J ¼ 9.5, 5.85 Hz, 2H), 4.09–4.16 (m,
2H), 5.12 (d, J ¼ 7.99 Hz, 4H), 7.19–7.33 (m, 2H),
7.35–7.55 (m, 18H). 13C NMR (acetone-d6, 500 MHz):
d 12.898, 22.284, 26.115, 28.398, 68.693, 68.736,
72.433, 92.320, 123.272, 126.729, 127.835, 128.189,
128.359, 128.725, 136.289, 171.626. 31P NMR (acet-
one-d6, 162 MHz): À20.04 (s, 1P). MS (FABþ in NBA):
m/z (relative intensity) 628.7 (Mþ, 60).
4-Butyl-4-((Methylthiomethoxy)Methyl)-1,2-
Dibenzyl((1,3-Dioxo-2-Phenylinden-2-yl)-
Diphenylpyrazolodine-3,5-Dione (7a). 83.7%
yield, distinct odor. H NMR (acetone-d6, 400 MHz):
Methoxy)Methyl Phosphate (8b). 59.7% yield,
reddish viscous oil. H NMR (acetone-d6, 400 MHz):
1
1
d 0.86 (d, 3H, J ¼ 5.39 Hz), 1.33 (d, 5H, J ¼ 2.94 Hz),
1.71–1.84 (m, 2H), 1.92 (d, 2H, J ¼ 5.68 Hz), 3.89 (d,
2H, J ¼ 5.49 Hz), 4.62 (d, 2H, J ¼ 5.60 Hz), 7.22 (t, 2H,
J ¼ 6.57 Hz), 7.32–7.48 (m, 8H). 13C NMR (acetone-d6,
500 MHz): d 12.572, 13.007, 22.403, 26.332, 28.383,
70.253, 74.811, 123.122, 126.76, 128.768, 136.675,
172.258. MS (CI): m/z (relative intensity) 399
d 4.48 (s, 2H), 5.09 (dd, J ¼ 9.98 Hz, 6H), 7.31–7.46 (m,
16H), 8.02–8.12 (m, 4H). 13C NMR (acetone-d6,
500 MHz): d 61.85, 68.689, 72.06, 92.391, 123.366,
126.802, 127.803, 128.193, 128.362, 128.947, 133.434,
135.994, 136.303, 136.359, 136.437, 142.16, 198.61.
31P NMR (acetone-d6, 162 MHz): À17.1 (s, 1P). MS
(CI): m/z (relative intensity) 543 ((M þ H)þ, 25).
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010
DOI 10.1002/jps

PRODRUG STRATEGY FOR b-DICARBONYL CARBON ACIDS
2715
General Procedure for Synthesis of POMOM
Prodrugs (3a,b)
in vacuo. The solid residue was chromatographed on
normal phase silica gel (approximately 100 g) with n-
hexanes/ethyl acetate (60/40) for eluent. Fractions
containing 12 (Rf ¼ 0.34) were collected and concen-
trated in vacuo to afford a pale yellowish powder;
5.24 g, 97.9% yield. 1H NMR (acetone-d6, 400 MHz): d
7.31–7.33 (m, 3H, phenyl), 7.43–7.45 (m, 2H, phenyl),
8.08–8.09 (m, 4H). MS (EI): m/z (relative intensity)
238.3 (Mþ, 10). HRMS (M þ Na)þ, C15H10O3Na: anal.
calculated, 261.0528; found, 261.0529, 0.0001 amu.
To a stirred solution of 8a or 8b (5 mmol) in ethanol
(25 mL) was added approximately 0.5 g of 10%
palladium–carbon. Hydrogenolysis was performed
at room temperature for approximately 6 h. The
reaction content was filtered through a sintered glass
funnel packed with Celite1 (Sigma–Aldrich, St.
Louis, MO) which was prewashed with ethanol.
The filtrate was concentrated in vacuo to yield
viscous oil. The phosphate free acid in this oil was
dissolved in approximately 5 mL of water and 1 mL
of ethanol to which was added dropwise NaOH
(2.1 Molar Equivalents) solution. This was followed
by the dropwise addition of cold ether until precipita-
tion was complete. The resulting disodium salt was
collected by filtration and lyophilized.
4-Butyl-4-Hydroxy-1,2-Diphenyl-Pyrazoli-
dine-3,5-Dione (13). 4-Butyl-4-hydroxy-1,2-diphe-
nyl-pyrazolidine-3,5-dione was synthesized from 1a
as described above for 12. 13 (quant.) as a white solid;
1H NMR (acetone-d6, 400 MHz): d 0.86–0.9 (t, 3H),
1.3–1.42 (d, 4H), 1.97 (d, 2H), 7.2 (m, 2H, arom.), 7.3–
7.41 (m, 8H, arom.). HRMS (M þ H)þ, C19H21N2O3:
anal. calculated, 325.1552; found, 325.1554.
Disodium((4-Butyl-3,5-Dioxo-1,2-Diphenyl-
pyrazolidin-4-yl)Methoxy)Methyl Phosphate
(3a). 61.4% yield, white amorphous powder.
1H NMR (D2O, 400 MHz): d 0.77 (t, J ¼ 7.02,
7.02 Hz, 3H), 1.18–1.34 (m, 4H), 1.80 (dd, J ¼ 8.62,
6.93 Hz, 2H), 4.01 (s, 2H), 4.89 (d, J ¼ 8.86 Hz, 2H),
7.30–7.36 (m, 10H). 13C NMR (D2O, 500 MHz): d
12.95, 21.64, 23.24, 26.08, 30.33, 66.21, 71.26, 90.72,
125.16, 127.54, 129.4, 138.41, 173.84. 31P (D2O,
162 MHz): 1.92 (s, 1P). MS (ESþ): m/z (relative
intensity) 493.1 ((M þ H)þ, 30). HRMS (M þ H)þ,
C21H24N2Na2O7P: anal. calculated, 493.1116; found,
493.1100, 0.0016 amu.
pKa Determination Using 31P NMR
2
A 4 mM (ꢂ2 mg/mL) solution of prodrug was prepared
by dissolving the prodrugs in 10% D2O in isotonic
solution to prepare stock solutions of 10 mL volume.
Samples of varying pH were prepared by adding small
volumes (mL) of 1 N HCl or 1 N NaOH solution and
recording the resulting pH. Aliquots (600 mL) were
withdrawn from the stock solution after each change
of pH and transferred to a NMR tubes. Twelve to 15
such samples for analysis were prepared in the pH
range of 3–10. The change in chemical shift was
recorded as a function of pH. Solutions were analyzed
by 31P NMR spectroscopy using a Bruker Avance AV
400 NMR spectrometer operating at 161.98 MHz for
31P. A total of 256 scans were collected on each sample
with a 50 ppm window centered at 0 ppm, a 2-s
acquisition time, a 0.15 s delay between scans, and a
308 excitation pulse. The 31P probe was calibrated
using 85% H3PO4 as an external standard for the
chemical shift.
Disodium((1,3-Dioxo-2-Phenyl-2,3-Dihydro-
1H-Inden-2-yl)Methoxy)Methyl Phosphate
(3b). 53.8% yield, white amorphous powder. 1H
NMR (D2O, 500 MHz): d 4.34 (s, 2H), 4.66–4.69 (m,
2H), 7.06 (dd, J ¼ 7.44, 7.14 Hz, 2H), 7.12–7.23 (m,
3H), 7.77–7.92 (m, 4H). 13C NMR (D2O, 500 MHz): d
62.86, 81.66, 90.64, 123.96, 126.5, 128.02, 128.69,
129.28, 132.93, 137.97, 141.91, 202.72. MS (FABþ in
NBA): m/z (relative intensity) 406.5 (Mþ, 40).
31P (D2O, 162 MHz): 2.2 (s, 1P). HRMS (M þ H)þ,
C17H14Na2O7P: anal. calculated, 407.0273; found,
407.0273, 0.004 amu.
In solution, depending on the pH and the dissocia-
tion constant (Ka), these prodrugs can exist in their
diacidic, monobasic, or dibasic fractions. The diacidic
fraction of a prodrug may be ignored given the low
and unlikely pH of such a solution physiologically.
The fraction of prodrug in the dibasic form ( fdb) can be
2-Hydroxy-2-Phenyl-Inden-1,3-Dione (12). To
5 g (22.49 mmol) of 1b in an aqueous solution of
sodium bicarbonate (10 mL) was added acetonitrile
(5 mL). The solution turns reddish. To this solution
was added dropwise 30% (w/w) aqueous solution of
hydrogen peroxide until the red color disappears,
indicating loss of delocalization in the carbanion due
to an adduct formation at the a-carbon. The reaction
mixture was partitioned between methylene chloride
(25 mL) and brine (25 mL), organic layer collected,
filtered, dried over sodium sulfate, and concentrated
expressed by Eq. (1), ignoring the Ka value:
1
Ka
½HþŠ þ Ka
2
fdb
¼
(1)
2
where [Hþ] represents the molar hydrogen ion
concentration.
The observed chemical shift (dobs) of the 31P signal
can be defined by Eq. (2) is a product of the mole
fraction of the prodrug species multiplied by its
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010

2716
DHARESHWAR AND STELLA
chemical shift and is expressed by Eq. (2)
(v/v) acetonitrile, the calculated d pH was þ0.8.
Accordingly, the pH of the 10 mM phosphate buffer
was adjusted to 6.6 with intent to maintain an
apparent pHÃ of 7.4 after acetonitrile addition. A total
volume of 100 mL of ‘‘reaction solvent’’ was prepared
½HþŠdmb þ Ka ddb
2
dobs
¼
(2)
½HþŠ þ Ka
2
where dmb and ddb represent the chemical shifts for the
monobasic and the dibasic fraction of the prodrug,
respectively.
d pH ¼ pHÃ À pH
(3)
Stock solutions of 1b (1.17 mM) and 2b (2.25 mM)
were freshly prepared in reaction solvent (see above).
The hydrogen peroxide is a 50% (w/w) (14,700 mM)
commercially available aqueous solution (Sigma–
Aldrich, MO). From stock solution of 1b, triplicate
working solutions (10 mL) of concentration ꢃ122 mM
were prepared in glass tubes (Kimble Glass, Inc.,
Vineland, NJ) maintained at 378C in a shaker bath
set at 60 oscillations/min. Samples (200 mL) were
removed from this vial at predetermined time points
(15, 30, 45, and 60 min). At the 60th minute, after
withdrawing sample for HPLC analysis, 37.34 mL of
1470 mM hydrogen peroxide solution (50-fold molar
excess) was added to glass tube containing 1b and
vortexed briefly. Thereafter, samples (200 mL) were
withdrawn at predetermined time points (75, 90, 120,
150, and 180 min) for analysis by HPLC. The stress
oxidation of 2b was performed similarly, including
selection of time points, using working solution of
128 mM prepared in reaction solvent. The only
difference was that 39.18 mL of 1470 mM hydrogen
peroxide solution was added after 60th minute
sample for HPLC analysis. The HPLC set up and
retention times for analytes of interest have been
described previously.
The pKa (or Ka ) of the prodrugs can be estimated
2
2
from nonlinear curve fitting using SigmaPlot graph-
ing software (Version 9.0, Systat Software, Inc.,
San Jose, CA) of the data to Eq. (2), which relates
the observed chemical shift (dobs) as a function of
pH ([Hþ]).
In Vitro Enzymatic Lability Evaluations
Experiments involving human placental alkaline
phosphatase catalyzed prodrug bioconversion were
performed in pH 7.4, 50 mM Tris buffer containing
1 mM of ZnCl2 and MgCl2 at 378C. An alkaline
phosphatase solution was prepared at a concentration
of 0.2 mg/mL (enzyme activity 0.52 U/mg estimated
using p-nitrophenol phosphate) in the aforemen-
tioned buffer. A prodrug stock solution was prepared
in this buffer, which was then divided into three
working solutions. The alkaline phosphatase stock
solution was brought to 378C and maintained at this
temperature for 30 min. The reaction was initiated by
adding a known volume of enzyme solution to the
prodrug solutions maintained in a shaking water
bath at 378C, to give an initial prodrug concentration
of ꢃ200 mM. Aliquots (300 mL) were removed from
each glass test tube at predetermined time points (0,
5, 10, 15, 20, 25, 30, 35, 45, and 60 min) and spiked
with glacial acetic acid (6 mL) to quench the enzymatic
reaction. To demonstrate that the prodrugs are
chemically stable during the course of the experi-
ment, the above-described procedure was repeated in
the absence of alkaline phosphatase. All samples
were subsequently analyzed by HPLC to estimate the
time-dependent change in the concentration of
prodrug and drug.
To confirm that phenindione indeed oxidizes at its
a-carbon, and to serve as an analytical standard
during HPLC analysis of the reaction samples, the
oxidized derivative 2-hydroxy-2-phenylinden-1,3-
dione (12) was synthesized in the laboratory.
Although there are procedures reporting the synth-
esis of this derivative,6,7 a much simpler and high
yielding synthetic procedure has been described (see
the Synthetic Procedures Section).
Chemical Stability Evaluations
Stress Oxidation Studies of Phenindione (1b) and Its
POM Prodrug 2b
The chemical stability study of 2a and 3a, as
representative prodrug derivatives, was performed
in aqueous solutions of 50 mM buffers adjusted to an
ionic strength of 0.154 M with NaCl. The pH values
along with their corresponding buffer compositions
were: sodium formate with formic acid (for pH 3.0–
4.5), sodium acetate with acetic acid (for pH 4.5–5.5),
sodium phosphate monobasic with sodium phosphate
dibasic (for pH 6.0–7.4), and Tris–HCl with Tris base
(for pH 8.0–8.5). Stock solutions for each pH were
prepared by dissolving approximately 0.2 mg of
prodrug in 10 mL of buffer. The solution was filtered
using a syringe filter (0.22 mm, nylon filters, Fisher
Scientific), assayed for initial prodrug concentration,
Stress oxidation studies of 1b and 2b were performed
at 378C using excess of hydrogen peroxide as oxidant
in 10 mM phosphate buffer and acetonitrile with
apparent pH 7.4 (pHÃ, see below). The reaction
solution contained 122 mM of 1b or 128 mM of 2b
each independently dissolved in 40% (v/v) acetonitrile
and 60% (v/v) 10 mM phosphate buffer. The pHÃ value
of these solutions in the presence of acetonitrile was
monitored (glass electrode, Fisher Scientific Accumet
Research AR 15 pH meter), and was stable for 1 h.
This cosolvent induced pH shift, d pH, was estimated
using Eq. (3). For the solvent system containing 40%
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010
DOI 10.1002/jps

PRODRUG STRATEGY FOR b-DICARBONYL CARBON ACIDS
2717
transferred to 1-mL prescored, type-1 borosilicate
glass ampule (Wheaton Science Products, Millville,
NJ), and flame sealed. Two controls were also
prepared at each pH, one was the parent carbon acid
phenylbutazone (1a) while the other was 4-hydro-
xyphenylbutazone (13), the oxidized degradant of
phenylbutazone (see the Synthetic Procedures Sec-
tion). These controls are poorly water-soluble; there-
fore, 3% (v/v) ethanol was added to the buffer solution
to facilitate their solubilization. The sealed ampules
were then placed into temperature-controlled ovens
(Stable Ther Constant Temperature Cabinet, Blue
Island, IL) at 50, 60, and 708C until analysis. At
predetermined time points, the ampules of 2a and 3a
and the two controls were opened and directly
injected into the HPLC system without dilution to
assay for time-dependent change in analyte content.
Individual estimations of kobs for the chemical
degradation of the prodrug were obtained by curve
fitting the experimental data to Eq. (4) (SigmaPlot
version 9.0, Systat Software, Inc.)
obstÞ
A ¼ A0eÀðk
(4)
where kobs is the pseudo-first-order rate constant for
the loss of the prodrug in time interval (t) and A0 is the
initial concentration of the prodrug.
RESULTS AND DISCUSSION
Syntheses
Scheme 2. The overall syntheses scheme for the POM
(2a,b) and POMOM (3a,b) prodrug derivatives of
b-dicarbonyl carbon acids.
The POM and POMOM prodrug derivatives of
phenylbutazone (1a) and phenindione (1b), two
pharmaceutically relevant carbon acids, and of
phenytoin (9), a hydantoin, are shown in Figure 1.
These prodrugs were synthesized in moderate to high
yields by the general synthesis route summarized in
Scheme 2. As is evident, both prodrug series were
synthesized using a common hydroxymethyl inter-
mediate 5a,b of the respective b-dicarbonyl carbon
acid. Near quantitative incorporation of the hydro-
xymethyl adduct selectively at the reactive a-carbon
was favored by the low enol tautomer content of both
phenylbutazone (1.8%)8 and phenindione (2.6%)9 in
water. Inasmuch as this hydroxyl group offered a
synthetic handle that would not have been possible
with the –CH group, it also avoided the more complex
and previously attempted route of using the carba-
nion as a nucleophile (data not presented).
The optimized synthesis route for the C-POM
prodrugs requires the use of freshly distilled methy-
lene chloride and anhydrous phosphorus oxychloride.
Phosphorylation of the hydroxymethyl intermediate
followed by aqueous work up of the resulting
phosphorodichloridate afforded the prodrugs in their
free acid form, which was then converted to the
disodium salt by neutralization, followed by lyophi-
lization. For the C-POMOM prodrug series,
methylthiomethylation of the 5a,b intermediate via
the Pummerer reaction10 facilitated the incorporation
of the methylene acetal required for the POMOM
promoiety. Addition of a few milliliters of acetic
acid into the reaction flask is a modification of the
Pummerer reaction which favors formation of
the (methylthio) methyl ether instead of a ketone.11
The iodonium ion promoted coupling of dibenzyl
phosphate to the (methylthio) methyl adduct offered a
convenient precursor to the phosphate free acid,
which was subsequently converted to the desired
disodium salt form by neutralization using 1 N
NaOH, followed by lyophilization to afford the
prodrugs as solid white powders.
Aqueous Solubility and pKa Estimation
2
For these prodrug derivatives to be useful for
eventual parenteral delivery, and possibly other
routes of delivery, of a therapeutically active parent
drug, they must possess sufficient aqueous solubility
preferably in the physiologically acceptable pH range.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010

2718
DHARESHWAR AND STELLA
The measured solubility of 2a and 3a was ꢀ250 mg/
mL with a final pH of 7.9–8.2 at 258C. Precise aqueous
solubility estimation was not performed given the
well-documented solubility enhancement of such
prodrug derivatives, and to conserve material. Of
significance, however, is the fact that at pH 7.9, the
estimated total solubility of 1a (pKa 4.3, intrinsic
solubility 8 mg/mL) is approximately 8 mg/mL, which
implies that these prodrugs offer up to a 30-fold
higher aqueous solubility in comparison to the carbon
acid parent drug phenylbutazone.
Table 1. Optimized Parameters Obtained in the
Determination of pKa of the Phosphate Prodrug
2
Derivatives Using 31P NMR at 258C
Prodrug
dmb, ppm (ÆSE)
ddb, ppm (ÆSE)
pKa (ÆSE)
2
2a
3a
2b
3b
10
11
À1.07 (Æ0.02)
À0.01 (Æ0.009)
À0.93 (Æ0.01)
À0.04 (Æ0.01)
À1.67 (Æ0.02)
À1.47 (Æ0.09)
2.72 (Æ0.03)
2.51 (Æ0.01)
2.90 (Æ0.01)
2.25 (Æ0.01)
1.47 (Æ0.02)
2.17 (Æ0.14)
6.39 (Æ0.01)
6.75 (Æ0.01)
6.24 (Æ0.009)
6.64 (Æ0.01)
6.15 (Æ0.02)
6.55 (Æ0.09)
In solution, depending on the pH and the dissocia-
tion constant (Ka), these prodrugs would exist in their
Comparing 2b and 3b, the increase in pKa value of
2
diacidic, monobasic, or dibasic fractions. The pKa
has limited pharmaceutical relevance, as pH of most
physiologically acceptable solutions would ensure
the latter can be attributed to a greater separation of
the phenyl substituent and the two carbonyl func-
tional groups from the phosphate group. At pH 7.4, all
these prodrugs would exist predominantly as dia-
nions, thus ensuring high solubility at this pH value.
1
ionization of the diacidic species. However, pKa is
2
significant given its influence on aqueous solubility,
chemical stability, and effectiveness as an enzyme
substrate to ensure bioconversion.
In Vitro Enzymatic Bioconversion
Estimates of pKa values for the prodrugs studied
2
here were determined using 31P NMR. In Figure 2, an
illustrative example, the experimental data for the
change in observed phosphorous chemical shift of 2a
in solutions of varying pH was fit to Eq. (2), which in
Establishing alkaline phosphatase lability of the
POM and POMOM prodrugs of these carbon acid
drugs in vitro would aid in the assessment of their
in vivo behavior. Human placental alkaline phospha-
tase was chosen because it is a standard source for
such studies and is one of the few sources of a human
alkaline phosphatase available.15–17 As summarized
in Scheme 1, the prodrugs are designed to undergo
a two-step bioconversion process, resulting in the
formation of the parent drug, inorganic phosphate,
and formaldehyde. The first step (k1) in this
sequential process is rate determining, followed by
a fast nonenzymatic chemical breakdown of the
hydroxymethyl/hemiacetal intermediate (k2).5
addition to the helping determine pKa , also permits
2
one to estimate the chemical shift of the monobasic
and dibasic fractions of the prodrug. The pH depen-
dency of this observed chemical shift measurement
describes a sigmoid curve, which is typical of an acid–
base titration. A summary of pKa values of these
2
phosphate monoesters can be found in Table 1.
These values compare favorably to estimates from
the literature.12–14
The values seen in Table 1 show the inductive effect
of the carbon acid substituent on pKa values of its
prodrugs. In comparison to 2a, an electron with-
drawing phenyl substituent in 2b favors ionization, in
Figure 3 shows the loss of 2a and 3a in the presence
of alkaline phosphatase at pH 7.4 and temperature of
378C and also shows the prodrugs to be chemically
stable under the experimental conditions when the
solution was devoid of enzyme. A fit of the experi-
mental data to a pseudo-first-order model (similar to
Eq. 4) provides an estimate for the apparent first-
order rate constant, kobs. While the calculated half-
life (t1/2) for bioconversion of 2b and 3b are more or
less similar (ꢃ4 min; Tab. 2), a shorter half-life for
3a compared to 2a is consistent with less steric
hindrance to conversion. Steric hindrance has been
noted in the bioconversion of a number of phosphate
and POM prodrugs.18,19 The similar bioconversion
rates of 2b compared to 3b was somewhat unantici-
pated. It suggests that the relative rate may be
influenced by a better leaving group (lower pKa of
the hydroxymethyl) in 2b versus a not so favorable
leaving group (hemiacetal) in case of 3b. Incorporat-
ing an additional formaldehyde spacer in the
POMOM derivative is validated by the faster rate
of in vitro bioconversion in comparison to the POM
2
doing so enhances acidity and lowers the pKa value.
2
Figure 2. Plot used to estimate the second dissociation
constant (pKa ) for 2a using 31P NMR.
2
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010
DOI 10.1002/jps

PRODRUG STRATEGY FOR b-DICARBONYL CARBON ACIDS
2719
The choice of acetonitrile as an organic cosolvent
is justified, as its nitrile functional group reacts
with hydrogen peroxide in neutral to moderately
alkaline pH to form peroxycarboximidic acid, a strong
oxidizing agent.20,21 The reactivity of peroxycarbox-
imidic acid is due to the marked electrophilic
character of the oxygen atom closer to the carbox-
imidic group, which favors nucleophilic attack by
either the enolate or the carbanion of phenindione
(see Scheme 3).
The time profile for loss of phenindione (1b) and its
POM prodrug 2b over 200 min before and after adding
hydrogen peroxide is presented in Figure 4. In the
first 60 min of reaction (no added peroxide) at pH 7.4,
there is no appreciable change in concentration of 1b,
indicating that the possible formation of the oxidized
degradant in the absence of added peroxide is below
the limits of detection. The addition of hydrogen
peroxide catalyzes the oxidation of 1b to 2-hydro-
xyphenindione (12), the oxidized degradant.
Figure 3. In vitro alkaline phosphatase (0.52 U/mg,
0.104 U/mL) catalyzed bioconversion of 2a (&) and 3a
(*) in 50 mM Tris buffer containing 1 mM ZnCl2 and MgCl2
at 378C and pH 7.4. In a control devoid of alkaline phos-
phatase (upper point sets), the unaltered concentration
profile of the prodrug is indicative of their chemical stability
within experimental time frame (n ¼ 3 Æ SD).
A loss of 1b to 12 is the predominant reaction seen
on the addition of hydrogen peroxide, although with
longer time additional degradation is seen but not
identified. This is confirmed by the time-dependent
derivative for one of the substrates but this effect may
be more relevant for more hindered substrates.
The in vitro assay for phosphate monoester
hydrolysis in the presence of alkaline phosphatase
is typically performed at pH 10.4, a pH optimum for
the activity of alkaline phosphatases; however, pH 7.4
was chosen here for its physiological relevancy.
Clearly, if these prodrugs are administered parent-
erally in humans, the overwhelming presence of
alkaline phosphatases in vivo should guarantee
bioconversion.
Stress Oxidation Studies Using Hydrogen Peroxide
Stabilization of the oxidatively labile1,2 –CH bond in
the parent b-dicarbonyl carbon acids was an objective
of this study. To that extent, the carbon–hydrogen
bond was transiently replaced by a carbon–carbon
bond in the prodrugs, and stress oxidation studies
were performed to confirm stability.
Table 2. Summary of the Pseudo-First-Order Rate
Constant and Estimated Half-Life for the Alkaline
Phosphatase (0.52 U/mg, 0.104 U/mL) Catalyzed
Bioconversion of Various POM and POMOM Prodrugs to
Their Respective Parent Carbon Acids in 50 mM Tris Buffer
Containing 1 mM ZnCl2 and MgCl2 at 378C and pH 7.4
Prodrug
kobs, minÀl (ÆSE)
t1/2, min (ÆSE)
Scheme 3. The in situ generation of peroxycarboximidic
acid from acetonitrile, and proposed hydrogen peroxide
catalyzed oxidative degradation of either the enolate or
carbanion tautomer of 1b to 2-hydroxy-2-phenylinden-1,3-
dione (12) (adapted from Ref. 34).
2a
2b
3a
3b
0.05 (Æ0.001)
0.17 (Æ0.017)
0.09 (Æ0.004)
0.15 (Æ0.01)
12.8 Æ 0.2
4.1 Æ 0.4
7.0 Æ 0.3
4.5 Æ 0.3
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010

2720
DHARESHWAR AND STELLA
Figure 5. pH-rate profiles, uncorrected for buffer cataly-
sis, for the chemical degradation of POM 2a (~) and
POMOM 3a (*) prodrug derivatives of 1a as a function
of pH (708C, I ¼ 0.154 M).
in Figure 5. Although the shape of the profile for 2a
is somewhat unique among phosphate monoester
prodrugs, the pH of maximum stability is in the
neutral pH range, which is perfectly suited for
formulation delivery from a physiological perspective.
Depending upon pH, both 2a and 3a may exist as
either the monoanion or the dianion, except at very
Figure 4. (A) Concentration–time profile of phenindione
(1b, &) (122 mM) in the absence (0–60 min) and presence of
50-fold molar excess of hydrogen peroxide at 378C. Addition
of hydrogen peroxide oxidizes 1b to 2-hydroxy-2-phenylin-
den-1,3-dione (12, !), its oxidative degradant by the
mechanism shown in Scheme 3. (B) Concentration–time
profile of the POM prodrug 2b (128 mM, *) under identical
experimental conditions shows no oxidative degradation
due to the absence of a labile –CH bond (n ¼ 3 Æ SD).
low pH values, at or below pKa , where the diacidic
1
species predominates. A simplified model which
accounts for the dephosphorylation of 2a, including
mathematically equivalent rate constants, is
described by Eq. (5), which includes the rate
constants for water-catalyzed or spontaneous hydro-
lysis of the monoanion ðk0OÞ and the dianion ðkO00 Þ
fraction, respectively
formation of unknown peaks on the HPLC chromato-
gram, which interfered with later mass balance
calculations.
The probable mechanism for oxidation of 1b is
presented in Scheme 3. At a pH of 7.4, phenindione is
deprotonated and may form its conjugate carbanion
that is in equilibrium with its enolate, either of them
can react with peroxycarboximidic acid or hydrogen
peroxide to generate 12, the common oxidized
derivative. In contrast, the profile of 2b, the POM
prodrug, as expected, clearly indicates that no
oxidative degradation occurs under these conditions.
That this prodrug strategy presents a twofold
advantage should be apparent, since incorporating
an ionizable promoiety at the reactive a-carbon
enhances both aqueous solubility (ꢀ250 mg/mL)
and prevents oxidation, two intrinsic limitations in
the formulation of b-dicarbonyl carbon acid drugs.
k0 ½HþŠKa
1
O
kobs
¼
2
½HþŠ þ ½HþŠKa þ Ka Ka
1
1
2
k00 Ka Ka
1
2
O
þ
(5)
2
½HþŠ þ ½HþŠK þ K K
a1
a1 a2
where kobs is the observed rate constant for the loss
of 2a while Ka and Ka are the first and second
1
2
dissociation constants for the phosphate monoester,
respectively. The parameters used to generate
the solid line for 2Àa1 in Figure 5 were (ÆSE):
,
k0O ¼ 0.027 Æ 0.001 day
,
k0O0 ¼ 0.016 Æ 0.0007 dayÀ1
pKa ¼ 5.4 Æ 0.2, while the value for Ka was fixed
2
1
at 0.032 (pKa of 1.5).
1
Likewise, the pH-rate profile for 3a is best
described by Eq. (6), which includes the rate
constants for hydronium-ion catalyzed hydrolysis of
the monoanion ðk0HÞ, water-catalyzed or spontaneous
hydrolysis of the monoanion ðk0OÞ and the dianion ðkO00 Þ
fraction, respectively, and the hydroxide-ion cata-
lyzed hydrolysis of the dianion ðk0OHÀ Þ fraction
Chemical Stability Evaluation
The pseudo-first-order loss of 2a and 3a were followed
in the pH range of 3–9 using 50 mM buffer solutions
(I ¼ 0.154 M). The pH-rate profiles, not corrected for
any buffer catalysis, for 2a and 3a at 708C are shown
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010
DOI 10.1002/jps

PRODRUG STRATEGY FOR b-DICARBONYL CARBON ACIDS
2721
!
k0 ½HþŠKa
k00 Ka Ka
k
½OHÀŠKa Ka
k0 ½HþŠ2Ka
00
H
O
O
OHÀ
1
1
2
1
2
1
kobs
¼
þ
þ
þ
ð6Þ
2
2
2
2
½HþŠ þ ½HþŠKa þ Ka Ka
½HþŠ þ ½HþŠKa þ Ka Ka
½HþŠ þ ½HþŠKa þ Ka Ka
½HþŠ þ ½HþŠKa þ Ka Ka
1
1
2
1
1
2
1
1
2
1
1
2
where kobs is the observed rate constant for loss of
prodrug 3a while Ka and Ka are the first and second
mathematicallyþequivalent rate constants for hydro-
nium ion (k0H [H ]) and spontaneous hydrolysis ðk0OÞ of
the monoanion fraction of 3a, respectively.
1
2
dissociation constants for the phosphate monoester,
respectively. The parameters used to generate the
solid line for 3a in Figure 5 were as follows:
Further elucidation of the dephosphorylation
mechanisms of 2a and 3a were not pursued although
the similarities of their Arrhenius activation energies
and apparent thermodynamic activation parameters
indicate a hydrolysis mechanism similar to other
phosphate monoesters.29 Figure 6 shows an Eyring
plot for 2a and 3a degradation at pH 3.5 and 6.0,
respectively, at temperatures of 50, 60, and 708C.
From the slopes of these plots, the apparent enthalpy
of activation DHz was calculated. The apparent
entropy of activation, DSz, at 258C was also estimated.
Table 3 summarizes the thermodynamic activation
parameters for 2a or 3a at 258C. These values are
comparable to other phosphate monoesters, suggest-
ing that the mechanism to initiate dephosphorylation
is similar. An Arrhenius plot (not shown) resulted in
Arrhenius activation energies of 26.73 Æ 0.25 and
27.27 Æ 2.87 kcal/mol for 2a and 3a, respectively.
k0H ¼ 1366.1 Æ 229.8 MÀ1 dayÀ1, k0O ¼ 0.14 Æ 0.06 dayÀ1
,
k0O0 ¼ 0.03 Æ 0.04 dayÀ1
,
k
¼ 3.67E À 12 Æ 1.5E À
00
À
OH
11 MÀ1 dayÀ1, and pKa ¼ 6.37 Æ 1.1, while the value
2
for Ka was fixed at 0.032 (pKa of 1.5).
1
1
Upon closer examination of the pH-rate profiles for
2a and 3a, one may observe a marked difference in
reactivity of the two phosphate monoester deriva-
tives, more so in the acidic pH range. The shape of the
pH-rate profile of 2a, in contrast to other phosphate
monoesters, suggests that in the pH range of 3.5–9.0,
the reactivity of the monoanion and the dianion is
comparable. This was somewhat unexpected, as the
rate constants for water-catalyzed or spontaneous
hydrolysis of the monoanion and dianion, respec-
tively, typically differs by a factor of 100, if not
greater. The relative lack of a pH dependency on the
dephosphorylation of 2a may be interpreted as an
unexpected increase in the stability of the monoanion
compared to other phosphate monoesters. This is
evident in the half-lives (t1/2) for dephosphorylation at
708C (pH 3.5) for the POM prodrug of phenytoin
(86.3 h)13 and a loxapine POM prodrug (11.5 h)22
whereas the corresponding value for 2a is 28.8 days.
However, the half-life for dephosphorylation of the
dianion fraction for the above three phosphate
monoesters at 708C (658C in case of loxapine) in
pH 7.4 is around 42 days. This suggests that the
flattening of the pH-rate profile is perhaps due to a
substantial decrease in reactivity of the monoanion,
to a level marginally more reactive than the dianion.
On the other hand, the pH dependency of the
POMOM derivative, 3a, is neither consistent with
relative rates of hydrolysis of the ionic species present
in solution nor with the position of a maximum in the
pH-rate profile.23–25 Instead its pH dependency is
analogous to that reported for aromatic monopho-
sphoramides,26 glucose-1-phosphate,27 and monoben-
zyl phosphate.28 A common feature among these
specific phosphate monoesters is the marked increase
in reactivity of their neutral (diacid) species in
comparison to conventional alkyl esters, resulting
in an indistinct maximum in their pH-rate profile.
This increased reactivity of 3a in acidic pH range can
be accounted either by incorporating into Eq. (6) the
rate constants for spontaneous (kO) and hydroxide-ion
These values were then used to calculate the kobs
,
half-life (t1/2) and shelf life (t90) for hydrolysis of 2a
and 3a at 258C summarized in Table 4. Arrhenius or
Eyring studies were, unfortunately, not carried out at
the pH values of maximum stability. Nevertheless,
projections could be made based on the profiles and
the recorded activation parameters from Table 3.
Even if ready-to-use solution formulations are not
possible, freeze-dried formulations are clearly an
option.
Figure 6. Eyring plot for the hydrolysis of 2a (~, pH 3.5,
50 mM) and 3a (*, pH 6.0, 25 mM) in buffers (I ¼ 0.154 M).
The temperatures studied were 50, 60, and 708C.
catalyzed hydrolysis (kOH [OHÀ]) of the diacid
À
species or as is the case by incorporating their
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010

2722
DHARESHWAR AND STELLA
products of decomposition of phenylbutazone sodium salt. Diss
Pharm Pharmacol 20:653–658.
Table 3. Summary of the Apparent Thermodynamic
Activation Parameters for the Chemical Hydrolysis of 2a
and 3a (258C)
2. Pawelczyk E, Wachowiak R. 1970. Kinetics of drug decomposi-
tion. XVII. Autoxidation and hydrolysis of sodium phenylbu-
tazone solutions. Diss Pharm Pharmacol 22:475–484.
3. Mooney KG, Rodriguez-Gaxiola M, Mintun M, Himmelstein
KJ, Stella VJ. 1981. Dissolution kinetics of phenylbutazone.
J Pharm Sci 70:1358–1365.
4. Stella VJ, Zygmunt JJ, Georg IG, Safadi MS. Water soluble
prodrugs of hindered alcohols. Patent 2000008033. February 17,
2000.
Prodrug
DHz (kcal/mol)
DSz (e.u.)
2a
3a
26.10 Æ 0.25
26.54 Æ 2.87
À12.7 Æ 3.2
À8.1 Æ 2.5
Table 4. Arrhenius Plot Extrapolations to Estimate kobs
(ÆSD); Half Life, t1/2; and Shelf Life, t90, for Chemical
Hydrolysis of 2a and 3a at 258C
5. Dhareshwar SS, Stella VJ. 2008. Breakdown kinetics of C-
hydroxymethyl derivatives of b-dicarbonyl carbon acids: Impli-
cations in the bioconversion rate of C-phosphoryloxymethyl
prodrugs of carbon acids. J Pharm Sci 98:1804–1812.
6. Shchukina LA. 1961. The mechanism of formation of 2-phenyl-
2-hydroxy-1,3-indandione from a,3-dibromo-3-benzylphtha-
lide. Zh Obshch Khim 31:3041–3045.
7. Song HN, Seong MR, Son JS, Kim JN. 1998. A study on the
Friedel-Crafts type reaction of ninhydrin with arenes. Syn
Commun 28:1865–1870.
Prodrug
pH
kobs (DayÀ1
)
t1/2 (Years)
t90 (Years)
2a
3a
3.5
6.0
6.4 (Æ0.3) ꢁ 10À5
29.7 Æ 1.3
7.3 Æ 0.7
4.4 Æ 0.2
1.1 Æ 0.1
2.59 (Æ0.23) ꢁ 10À4
8. Stella VJ, Pipkin JD. 1976. Phenylbutazone ionization kinetics.
J Pharm Sci 65:1161–1165.
9. Stella VJ, Gish R. 1979. Ionization kinetics of the carbon acid
phenindione. J Pharm Sci 68:1042–1047.
CONCLUSIONS
This report demonstrates the feasibility of C-POM
and POMOM prodrug derivatives as a unique
strategy to improve aqueous solubility and prevent
oxidative instability of pharmaceutically relevant b-
dicarbonyl carbon acid drugs. One may ask if
synthesizing the POMOM prodrug derivative has
any appreciable advantage over the conventional
POM prodrug derivatives. Although the POM deri-
vative showed greater chemical stability, the
POMOM derivative show the potential for an increase
in enzymatic bioconversion rates, thus suggesting
some benefit for the POMOM derivative for sterically
hindered substrates.
The concern with release of formaldehyde, espe-
cially from the POMOM derivative, as a by-product
during in vivo bioconversion is unwarranted,30 as the
human body can effectively incorporate exogenous
formaldehyde into the pool of endogenously generated
formaldehyde. The fundamental issue is the amount
of formaldehyde that is released from such a prodrug,
which ought to be compared with the exposure from
all other sources, some of which include diet,31 from
N- or O-demethylation of antioxidants32 and a few
clinically administered drugs.33
10. Pummerer R. 1910. Phenylsulphoxyacetic acid. II. Chemische
Berichte 43:1401–1412.
11. Pojer PM, Angyal SJ. 1978. Methylthiomethyl ethers: Their use
in the protection and methylation of hydroxyl groups. Aust J
Chem 31:1031–1040.
12. Juntunen J, Huuskonen J, Laine K, Niemi R, Taipale H,
Nevalainen T, Pate DW, Jarvinen T. 2003. Anandamide pro-
drugs 1. Water-soluble phosphate esters of arachidonylethano-
lamide and R-methanandamide. Eur J Pharm Sci 19:37–43.
13. Kearney AS, Stella VJ. 1993. Hydrolysis of pharmaceutically
relevant phosphate monoester monoanions: Correlation to an
established structure-reactivity relationship. J Pharm Sci 82:
69–72.
14. Kumler WD, Eiler JJ. 1943. The acid strength of mono and
diesters of phosphoric acid. The n-alkyl esters from Me to Bu,
the esters of biological importance and the natural guanidine-
phosphoric acids. J Am Chem Soc 65:2355–2361.
15. Ueda Y, Matiskella JD, Golik J, Connolly TP, Hudyma TW,
Venkatesh S, Dali M, Kang S-H, Barbour N, Tejwani R, Varia
S, Knipe J, Zheng M, Mathew M, Mosure K, Clark J, Lamb L,
Medin I, Gao Q, Huang S, Chen C-P, Bronson JJ. 2003. Phos-
phonooxymethyl prodrugs of the broad spectrum antifungal
azole, ravuconazole: Synthesis and biological properties. Bioorg
Med Chem Lett 13:3669–3672.
16. Nicolaou MG, Wolfe JL, Schowen RL, Borchardt RT. 1996.
Facilitated intramolecular conjugate addition of amides of 3-
(30,60-dioxo-20,40-dimethyl-10,40-cyclohexadienyl)-3,3 dimethyl-
propionic acid. 2. Kinetics of degradation. J Org Chem 61:
6633–6638.
17. Krise JP, Zygmunt JJ, Georg IG, Stella VJ. 1999. Novel pro-
drug approach for tertiary amines: Synthesis and preliminary
evaluation of N-phosphonooxymethyl prodrugs. J Med Chem
42:3094–3100.
18. Kearney AS, Stella VJ. 1992. The in vitro enzymic labilities of
chemically distinct phosphomonoester prodrugs. Pharm Res
9:497–503.
19. DeGoey DA, Grampovnik DJ, Flosi WJ, Marsh KM, Wang XC,
Klein LL, McDaniel KF, Liu Y, Long MA, Kati WM, Molla A,
Kempf DJ. 2009. Water-soluble prodrugs of the human immu-
nodeficiency virus protease inhibitors lopinavir and ritonavir.
J Med Chem 52:2964–2970.
ACKNOWLEDGMENTS
The research was supported in part by the tuition
scholarship for SSD from the Department of Phar-
maceutical Chemistry, The University of Kansas,
Lawrence, KS.
REFERENCES
20. Payne GB, Deming PH, Williams PH. 1961. Reactions of hydro-
gen peroxide. VII. Alkali-catalyzed epoxidation and oxidation
using a nitrile as a coreactant. J Org Chem 26:659–663.
1. Pawelczyk E, Wachowiak R. 1968. Chemical characterization of
decomposition products of drugs. III. Identification of some
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010
DOI 10.1002/jps

PRODRUG STRATEGY FOR b-DICARBONYL CARBON ACIDS
2723
21. Sawaki Y, Ogata Y. 1981. Mechanism of the reaction of nitriles
with alkaline hydrogen peroxide. Reactivity of peroxycarbox-
imidic acid and application to superoxide ion reaction. Bull
Chem Soc Japan 54:793–799.
22. Krise JP, Narisawa S, Stella VJ. 1999. A novel prodrug
approach for tertiary amines. 2. Physicochemical and in vitro
enzymatic evaluation of selected N-phosphonooxymethyl pro-
drugs. J Pharm Sci 88:922–927.
28. Kumamoto J, Westheimer FH. 1955. The hydrolysis of
mono- and dibenzyl phosphates. J Am Chem Soc 77:2515–
2518.
29. Kirby AJ, Varvoglis GA. 1967. Reactivity of phosphate esters.
Monoester hydrolysis. J Am Chem Soc 89:415–423.
30. Dhareshwar SS, Stella VJ. 2007. Your prodrugs release for-
maldehyde: Should you be concerned? No! J Pharm Sci 97:
4184–4193.
23. Barnard PWC, Bunton CA, Llewellyn DR, Oldham KG, Silver
BL, Vernon CA. 1955. The hydrolysis of organic phosphates.
Chem Ind (Lond), 760–763.
31. WHO. 1989. Formaldehyde (Environmental Health Criteria
89). International Programme on Chemical Safety. Geneva:
WHO.
24. Bunton CA. 1970. Hydrolysis of orthophosphate esters. Acc
Chem Res 3:257–265.
32. El-Rashidy R, Niazi S. 1983. A new metabolite of butylated
hydroxyanisole in man. Biopharm Drug Dispos 4:389–396.
33. Vree TB, Verwey-van Wissen CPWGM. 1992. Pharmacoki-
netics and metabolism of codeine in humans. Biopharm Drug
Dispos 13:445–460.
25. Cox JR, Ramsay OB. 1963. Mechanisms of nucleophilic sub-
stitution in phosphate esters. Chem Rev 64:317–352.
26. Chanley JD, Feageson E. 1958. Hydrolysis of phosphonamides.
I. Aromatic phosphonamides. J Am Chem Soc 80:2686–2691.
27. Bunton CA, Llewellyn DR, Oldham KG, Vernon CA. 1958.
Reactions of organic phosphates. II. Hydrolysis of a-D-glucose
1-(dihydrogen phosphate). J Chem Soc, 3588–3594.
34. Hovorka SW, Hageman MJ, Schoneich C. 2002. Oxidative
degradation of
a
sulfonamide-containing 5,6-dihydro-4-
hydroxy-2-pyrone in aqueous/organic cosolvent mixtures.
Pharm Res 19:538–545.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 6, JUNE 2010