Bis(carbamoyloxymethyl) esters of 2′,3′-dideoxyuridine 5′-monophosphate (ddUMP) as potential ddUMP prodrugs

Article abstract of DOI:10.1007/s11095-005-1876-4

Purpose. We previously reported the synthesis of bis(pivaloyloxymethyl) 2′,3′-dideoxyuridine 5′-monophosphate (POM₂-ddUMP) (1a) as a membrane-transport prodrug formulation of the free parent nucleotide, ddUMP. Although successful at delivering ddUMP into cells in culture, POM ₂-ddUMP was rapidly degraded by plasma carboxylate esterases after intravenous administration to experimental animals, and therefore has limited therapeutic potential as a systemically administered prodrug. We now report the synthesis of bis(N,N′-dimethylcarbamoyloxymethyl)- and bis(N-piperidinocarbamoyloxymethyl) 2′,3′-dideoxyuridine 5′-m onophosphate [DM₂-ddUMP (1b) and DP₂-ddUMP (1c), respectively], analogues of POM₂-ddUMP that were designed to be more resistant to degradation by plasma esterases. Methods. After entering cell by passive diffusion, it was anticipated that loss of one of the carbamoyloxymethyl groups of 1b and 1c would occur by spontaneous chemical hydrolysis to give the intermediate phosphodiesters, 2b and 2c. Cleavage of the remaining carbamoyloxymethyl groups by cellular phosphodiesterase I would generate ddUMP. 1b and 1c were prepared by condensation of 2′,3′-dideoxyuridine (ddU) with the appropriate bis(N-alkylcarbamoyloxymethyl) phosphate in DMA in the presence of triphenylphosphine and diethyl azodicarboxylate (the Mitsunobo reagent). Results. The half-lives of 1b and 1c when incubated at a concentration of 10^-4 M in human plasma at 37°C were 3.5 h and 3.7 h, respectively, similar to the half-lives observed under the same temperature conditions in 0.05 M aqueous phosphate buffer, pH 7.4. By contrast, the half-life of the POM₂ prodrug, 1a, in plasma was only 5 min. The initial products of degradation of 1b and 1c were the phosphodiesters 2b and 2c. The latter compounds gave rise to ddUMP when incubated with snake venom phosphodiesterase I. Conclusion. These findings support the premise inherent in the design of 1b and 1c, namely that the carbamate prodrugs are far more resistant to hydrolysis by plasma carboxylate esterases than their POM counterparts and can revert to the free parent 5′-mononucletides by successive chemical and enzymatic hydrolysis. Further studies of 1b and 1c as membrane-permeable prodrugs of ddUMP are in progress.

Full text of DOI:10.1007/s11095-005-1876-4

Pharmaceutical Research, Vol. 22, No. 3, March 2005 (© 2005)
DOI: 10.1007/s11095-005-1876-4
Research Paper
Bis(carbamoyloxymethyl) Esters of 2؅,3؅-Dideoxyuridine 5؅-monophosphate
(ddUMP) as Potential ddUMP Prodrugs
Saeed R. Khan,1,3 Srinivas K. Kumar,1 and David Farquhar2
Received November 9, 2004; accepted December 13, 2004
Purpose. We previously reported the synthesis of bis(pivaloyloxymethyl) 2Ј,3Ј-dideoxyuridine 5Ј-
monophosphate (POM2-ddUMP) (1a) as a membrane-transport prodrug formulation of the free parent
nucleotide, ddUMP. Although successful at delivering ddUMP into cells in culture, POM2-ddUMP was
rapidly degraded by plasma carboxylate esterases after intravenous administration to experimental
animals, and therefore has limited therapeutic potential as a systemically administered prodrug. We now
report the synthesis of bis(N,N’-dimethylcarbamoyloxymethyl)- and bis(N-piperidinocarbamoyloxy-
methyl) 2Ј,3Ј-dideoxyuridine 5Ј-monophosphate [DM2-ddUMP (1b) and DP2-ddUMP (1c), respec-
tively], analogues of POM2-ddUMP that were designed to be more resistant to degradation by plasma
esterases.
Methods. After entering cell by passive diffusion, it was anticipated that loss of one of the carbamoy-
loxymethyl groups of 1b and 1c would occur by spontaneous chemical hydrolysis to give the intermediate
phosphodiesters, 2b and 2c. Cleavage of the remaining carbamoyloxymethyl groups by cellular phos-
phodiesterase I would generate ddUMP. 1b and 1c were prepared by condensation of 2Ј,3Ј-
dideoxyuridine (ddU) with the appropriate bis(N-alkylcarbamoyloxymethyl) phosphate in DMA in the
presence of triphenylphosphine and diethyl azodicarboxylate (the Mitsunobo reagent).
Results. The half-lives of 1b and 1c when incubated at a concentration of 10−4 M in human plasma at
37°C were 3.5 h and 3.7 h, respectively, similar to the half-lives observed under the same temperature
conditions in 0.05 M aqueous phosphate buffer, pH 7.4. By contrast, the half-life of the POM2 prodrug,
1a, in plasma was only 5 min. The initial products of degradation of 1b and 1c were the phosphodiesters
2b and 2c. The latter compounds gave rise to ddUMP when incubated with snake venom phosphodi-
esterase I.
Conclusion. These findings support the premise inherent in the design of 1b and 1c, namely that the
carbamate prodrugs are far more resistant to hydrolysis by plasma carboxylate esterases than their POM
counterparts and can revert to the free parent 5Ј-mononucletides by successive chemical and enzymatic
hydrolysis. Further studies of 1b and 1c as membrane-permeable prodrugs of ddUMP are in progress.
KEY WORDS: anticancer; 2Ј,3Ј-dideoxyuridine 5Ј-monophosphate; membrane-permeable prodrugs.
INTRODUCTION
penetrated facilely into cells and reverted to the parent 5Ј-
mononucleotides (11), (e.g., 3) after successive cleavage of
the POM groups by carboxylate esterases and phosphodies-
terases, respectively (Scheme 1). However, a significant short-
coming of this approach in vivo is that the POM2 prodrugs are
rapidly degraded in human plasma by carboxylate esterases
(S. Khan and D. Farquhar, unpublished results) before they
have an opportunity to reach their target therapeutic sites. To
address this problem, we have investigated the potential of
bis(carbamoyloxymethyl) esters of 5Ј-mononucleotides as po-
tential esterase-resistant prodrugs of the free parent nucleo-
tides. Carbamate groups should be less susceptible than ester
groups to esterase-mediated hydrolysis and, therefore, bis-
(carbamoyloxymethyl) 5Ј-nucleotide prodrugs should have
superior systemic bioavailability. Once administered in vivo,
cleavage of the first carbamoyloxymethyl group is anticipated
to occur by spontaneous chemical hydrolysis and, by analogy
with mono-POM 5Ј-nucleotides (10), cleavage of the second
should occur by nonspecific phosphodiesterases. The objec-
tive of this study was to prepare model bis(carbamoyloxy-
5Ј-Mononucleotides are obligatory intermediates in the
activation of most anticancer and antiviral nucleoside ana-
logues (1–4). However, their utility as chemotherapeutic
agents is severely compromised because they are charged at
physiologic pH and penetrate poorly into cells (5–6). 5Ј-
Mononucleotides are also susceptible to rapid degradation to
the corresponding nucleosides in tissues by 5Ј-nucleotidases
and other nonspecific phosphohydrolases (7–8). To overcome
these limitations, we reported a series of neutral bis(pival-
oyloxymethyl) [POM2] 5Ј-nucleotide esters (e.g., 1a (12)) as
potential membrane-permeable prodrugs of the parent 5Ј-
mononucleotides (7–10). We showed that these compounds
1 The Sidney Kimmel Comprehensive Cancer Center at Johns Hop-
kins, Baltimore, Maryland 21231, USA.
2 Department of Experimental Therapeutics, The University of Texas
M. D. Anderson Cancer Center, Houston, Texas 77030, USA.
3
To whom correspondence should be addressed. (e-mail:
khansa@jhmi.edu)
0724-8741/05/0300-0390/0 © 2005 Springer Science+Business Media, Inc.
390

Bis(carbamoyloxymethyl) Esters of ddUMP
391
Chemical Synthesis
Disilver Benzyl Phosphate (7)
Phosphorus acid (dried over P2O5) (25.0 g, 0.30 mol) was
dissolved in a solution of benzyl alcohol (500 ml) and Et3N
(150 ml, 33.75 mmol). Solid I2 (116 g, 0.45 mol) was added in
10 g portions with cooling in an ice bath at 5°C. The mixture
was stirred for 30 min. It was then poured into acetone (2000
ml) and an excess of cyclohexylamine (100 ml) was added.
The precipitate which formed was collected by filtration,
washed with acetone, and recrystallized from aqueous EtOH
(H2O, 200 ml; EtOH, 1100 ml) containing 1% (v/v) of cyclo-
hexylamine (14). The yield was 69.0 g (78%). The mother
liquor was evaporated to dryness and the residual solid was
recrystallized to give a further 17 g of product. The total yield
of cyclohexylammonium benzyl phosphate was 86.0 g (97%);
mp 229–233°C (lit. mp 233–234°C). A solution of this product
(50.0 g, 0.13 mol) in distilled H2O (500 ml) was passed
through a column (5 × 50 cm) of Bio-Rad AG (Richmond,
CA, USA) 50W × 12 cation-exchange resin (200-400 mesh, 2.3
mEq per ml of resin bed) in the Na+ form. The resin was
washed with 3 column volumes of distilled H2O. The com-
bined effluents were concentrated to ca. 300 ml on a rotary
evaporator under reduced pressure at <40°C and then added,
dropwise, to a rapidly stirred solution of AgNO3 (48.3 g, 0.29
mol) in H2O (300 ml). The copious white precipitate that
formed was filtered and washed sequentially with distilled
H2O, EtOH, and Et2O, and air-dried. The product was stored
in vacuo over P2O5. The yield was 48.9 g, 94%. Anal.
(C7H7O4PAg2) C, H, Ag.
Sch. 1. Mechanism of reversion of the prodrugs to the patient ionic
compounds.
methyl) 5Ј-nucleotide prodrugs and to study their stability
and chemical fate in aqueous buffers and in human plasma.
Bis(N,NЈ-dimethylcarbamoyloxymethyl) and bis(N-piperidino-
carbamoyloxymethyl) diesters of 2Ј,3Ј-dideoxyuridine 5Ј-
monophosphate (1b and 1c, respectively) were selected for
this purpose. A brief preliminary account of part of this work
has previously appeared (12).
MATERIALS AND METHODS
2Ј,3Ј-Dideoxyuridine (ddU) and phosphodiesterase I
(E.C 3.1.4.1) were purchased from Sigma Chemical Co. (St.
Louis, MO, USA). ddUMP was prepared from ddU by the
general method of Yoshikawa et al. (13) All reactions were
carried out in dry glassware and were protected from atmo-
spheric moisture. Solvents were dried over freshly activated
(300°C/4 h) molecular sieves (type 4A). Reactions with silver
salts were conducted in dry glassware in the dark and were
protected from atmospheric moisture. The silver salts were
dried in vacuo over P2O5 before use. The homogeneity of the
products was determined by ascending TLC on silica-coated
glass plates (silica gel 60 F 254, Merck, Darmstadt, Germany)
using mixtures of CHCl3-MeOH (typically 1% to 10%
MeOH) as the eluting solvent. Chromatograms were visual-
ized under a UV lamp (254 nm) or by placing the air-dried
plates in a tank of I2 vapor. Compounds containing pivaloyl-
oxymethyl or carbamoyloxymethyl groups were identified by
spraying the plates with a 0.25% solution of 4-amino-3-
hydrazino-5-mercapto-1,2,4-tetrazole (Purpald) in 0.5 N
NaOH solution and heating them in an oven at 85°C for 5
min. The liberated formaldehyde reacted with the Purpald
reagent to form purple spots against a white background.
Preparative separations were performed by flash chromatog-
raphy on silica (Merck, 230-400 mesh) using mixtures of
EtOAc/hexane or CHCl3/MeOH as eluent. Chemical re-
agents were purchased from Aldrich Chemicals Co. (Milwau-
kee, WI, USA) or from Sigma Chemical Co. (St. Louis, MO,
USA). Nuclear magnetic resonance spectra (1H NMR, 13C
NMR) were recorded at ambient temperature on an IBM-
Bruker Model NR/200 AF spectrometer (Boston, MA, USA)
(N,NЈ-Dimethyl)carbamoyloxymethyl Chloride (6b)
A solution of dimethylamine 4b (20 g, 0.44 mol) in tolu-
ene (250 ml) was added dropwise over 30 min with vigorous
stirring at 0°C to chloromethyl chloroformate (28.7 g, 20 ml,
0.22 mol) contained in a 1-L flask. The mixture was stirred
overnight at room temperature and then filtered to remove
precipitated salts. The filtrate was concentrated under re-
duced pressure, and the residue was taken up in toluene (100
ml), and filtered to remove additional precipitated salts. The
filtrate was washed successively with 5% NaHCO3 solution
(50 ml) and H2O (3 × 100 ml), and dried over anhydrous
Na2SO4. The solvent was evaporated under reduced pressure
(30 mmHg) at <30°C to give (N,NЈ-dimethyl)carbamoyloxy-
methyl chloride, 6b, as a light yellow viscous liquid in 83%
1
yield. H NMR (CDCl3): ␦ 5.78 (s, 2 H, CH2), 2.97 (s, 6 H,
N(CH3)2. MS: m/z 138 (MH+). Anal. (C4H8ClNO2) C, H, N.
N-Piperidinocarbamoyloxymethyl Chloride (6c)
A solution of piperidine 4c (9.9 g, 11.5 ml, 0.11 mol) in
in the Fourier transform mode, in CDCl3, using tetramethyl- toluene (60 ml) was added dropwise over 30 min with vigor-
silane as an internal standard. Mass spectra were obtained on ous stirring at 0°C to chloromethyl chloroformate, 5 (7.5 g, 6.0
a Finnegan Model 3300 quadruple spectrometer (Platte City, ml, 0.06 mol). The mixture was stirred overnight at room
MO, USA) in the electron impact mode, or the chemical temperature and then filtered to remove precipitated salts.
ionization mode using methane as the reagent gas. Elemental The filtrate was concentrated under reduced pressure, and
analyses were performed by Galbraith Laboratories, Inc. the residue was taken up in toluene (50 ml), and again filtered
(Knoxville, TN, USA) and, where indicated only by the sym- to remove additional precipitated salts. The filtrate was
bols of elements, the results were within 0.4% of the theo- washed successively with 5% NaHCO3 solution (50 ml) and
retical values. Melting points were determined on a Hoover H2O (3 × 100 ml), and dried over anhydrous Na2SO4. The
capillary apparatus (Pittsburgh, PA, USA) and are uncor- solvent was evaporated under reduced pressure (30 mmHg)
rected.
at <30°C to give N-piperidinocarbamoyloxymethyl chloride,

392
Khan, Kumar, and Farquhar
6c, as a light yellow viscous liquid in 85% yield. 1H NMR was filtered and the solution was evaporated under reduced
(IBM-Bruker Model NR/200; CDCl3): ␦ 5.79 (s, 2 H, OCH2), pressure to give bis(N-piperidinocarbamoyloxymethyl) hy-
2.97 (m, 4 H, N(CH2)2; piperidino ␣ protons), 2.97 (m, 6 H, drogen phosphate, 9c, as colorless viscous oil. The yield was 8
1
(CH2)3; piperidino ␤ and ␥ protons). MS: m/z 178 (MH+). g (89%). H NMR (CDCl3): ␦ 5.52 (d, 4 H, OCH2O, J ס 14
Anal. (C7H12ClNO2) C, H, N.
Hz), 3.19 (m, 8 H, N(CH2)2; piperidino ␣ protons), 1.49 (m, 12
H, N(CH2)3; piperidino ␤ and ␥ protons). MS: m/z 381
(MH+). Anal. (C14H25N2O8P) C, H, N.
Bis[(N,NЈ-Dimethyl)carbamoyloxymethyl] Benzyl
Phosphate (8b)
Bis[(N,NЈ-dimethyl)carbamoyloxymethyl]
2Ј,3Ј-Dideoxyuridine 5Ј-monophosphate (1b)
A solution of 6b (11.5 g, 0.067 mol) in dry toluene (100
ml), was added, dropwise, over 30 min to a suspension of
finely divided disilver benzyl phosphate, 7 (3,6) (13.0 g, 0.033
mol) in toluene (50 ml). An exothermic reaction ensued. The
reaction mixture was stirred vigorously at 80°C for 24 h. The
precipitated salts were filtered and the filtrate was washed
successively with 5% NaHCO3 solution (1 × 100 ml) and H2O
(3 × 50 ml), then dried over anhydrous Na2SO4. Evaporation
of the solvent yielded bis[(N,NЈ-dimethyl)carbamoyl-
oxymethyl] benzyl phosphate, 8b, as a colorless, viscous oil
Compound 9c (297 mg, 0.99 mmol), 2Ј,3Ј-dideoxyuridine
(140 mg, 0.66 mmol), and triphenylphosphine (136 mg, 0.99
mmol) were dissolved in DMA (3.0 ml) and the solution was
stirred magnetically for 10 min. A solution of diethylazodi-
carboxylate (0.16 ml, 0.99 mmol) in DMA (0.5 ml) was added
with stirring and the reaction mixture was maintained at 60°C
for 5 days under a N2 atmosphere. The DMA was evaporated
under reduced pressure. The remaining oil was triturated sev-
eral times with 10 ml portions of n-hexane/EtOAc (3:1) and
the supernatants were decanted. The contents of the flask
were maintained under high vacuum for 24 h to remove re-
sidual DMA, and the remaining product was taken up in the
minimum of CHCl3 and chromatographed on a column of
silica (70 g) using CHCl3-MeOH (95:5) as eluent. Bis[(N,NЈ-
dimethyl)carbamoyloxymethyl] 2Ј,3Ј-dideoxyuridine 5Ј-
monophosphate 1b (14) was isolated as a viscous oil. Yield,
1
(13.7 g, 89%). H NMR (CDCl3): ␦ 7.23 (s, 5 H, C6H5), 5.63
(d, 4 H, OCH2O, J ס 14 Hz), 5.10 (d, 2 H, C6H5CH2, J ס 8
Hz), 2.93 (s, 12 H, N(CH3)2). MS: m/z 391 (MH+). Anal.
(C15H23N2O8P) C, H, N.
Bis[(N-piperidino)carbamoyloxymethyl] Benzyl
Phosphate (8c)
1
144 mg (45%). H NMR (CDCl3): ␦ 7.65 (d, 1 H, H-6, J ס 8
A solution of 6c (3.5 g, 0.02 mol) in dry toluene (20 ml),
was added, dropwise, over 30 min to a suspension of finely
divided disilver benzyl phosphate, 7 (3.8 g, 0.09 mol) in tolu-
ene (20 ml). The reaction mixture was maintained with stir-
ring at 80°C for 24 h. The precipitated salts were filtered, and
the filtrate was washed successively with 5% NaHCO3 solu-
tion (1 × 100 ml) and H2O (3 × 50 ml), then dried over
anhydrous Na2SO4. Evaporation of the solvent yielded
bis[(N-piperidinocarbamoyloxymethyl] benzyl phosphate, 8c,
Hz), 6.0 (dd, 1 H, H-1Ј), 5.77 (d, 1 H, H-5, J ס 8 Hz), 5.70 (d,
4 H, P(O)OCH2O, JPH ס 12 Hz), 4.28 (m, 2 H, H-5Ј), 2.97 (s,
12 H, N(CH3)2), 2.22 (m, 2 H, H-3Ј), 2.15 (m, 1 H, H-4Ј), 1.71
(m, 2 H, H-2Ј). MS: m/z 495 (MH+). Anal. (C17H27N4O11P):
C, H, N.
Bis[(N-piperidino)carbamoyloxymethyl]
2Ј,3Ј-Dideoxyuridine 5Ј-monophosphate (1c)
1
as a colorless, viscous oil (4.2 g, 89%). H NMR (CDCl3): ␦
The product was prepared from 9c (565 mg, 1.5 mmol),
2Ј,3Ј-dideoxyuridine (210 mg, 0.99 mmol), triphenylphos-
phine (390 mg, 1.5 mmol) and diethylazodicarboxylate (0.24
ml, 1.5 mmol) as described for 1b. The crude product was
purified on a column of silica (100 g) using CHCl3/MeOH
(9:1) as eluent to afford 1c (15) as a viscous oil. Yield, 111 mg
(35%). 1H NMR (CDCl3): ␦ 7.73 (d, 1 H, H-6, J ס 8 Hz), 6.20
(dd, 1 H, H-1Ј), 5.78 (d, 1 H, H-5, J ס 8 Hz), 5.71 (d, 4 H,
P(O)OCH2O, JPH ס 12 Hz), 4.40 (m, 2 H, H-5Ј), 3.45 (m, 8
H, N(CH2)2; piperidino ␣ protons), 1.51 (m, 12 H, N(CH2)3;
piperidino ␤ and ␥ protons), 2.22 (m, 2 H, H-3Ј), 2.15 (m, 1 H,
H-4Ј), 1.71 (m, 2 H, H-2Ј). MS: m/z 575 (MH+). Anal.
(C23H35N4O11P): C, H, N.
7.28 (s, 5 H, C6H5), 5.80 (d, 4 H, OCH2O, J ס 14 Hz), 5.20 (d,
2 H, C6H5CH2, J ס 8 Hz), 3.43 (m, 8 H, N(CH2)2; piperidino
␣ protons), 1.56 (m, 12 H, (CH2)3; piperidino ␤ and ␥ pro-
tons). MS: m/z 471 (MH+). Anal. (C21H31N2O8P) C, H, N.
Bis[(N,NЈ-dimethyl)carbamoyloxymethyl] Hydrogen
Phosphate (9b)
A solution of 8b (10.6 g, 0.03 mol) in cyclohexane (200
ml) was hydrogenated over 5% Pd/C (500 mg) at a pressure of
30 psi. After 1 h, an additional 500 mg of catalyst was added
and hydrogenolysis was continued for a further 1 h. The cata-
lyst was filtered and the solution was evaporated under re-
duced pressure to give bis[(N, NЈ-dimethyl)carbamoyloxy-
methyl] hydrogen phosphate, 9b, as a colorless viscous oil.
The yield was 7.3 g (68%). 1H NMR (CDCl3): ␦ 5.64 (d, 4 H,
OCH2O, J ס 14 Hz), 2.93 (s, 12 H, N(CH3)2). MS: m/z 301
(MH+). Anal. (C8H17N2O8P) C, H, N.
Mono[(N,NЈ-Dimethyl)carbamoyloxymethyl] 2Ј,
3Ј-Dideoxyuridine 5Ј-phosphate (2b)
NH4HCO3 buffer (0.05 M), pH 8 (5 ml), was added to 1b
(15 mg, 0.03 mmol) in a round-bottomed flask, and the mix-
ture was stirred at room temperature for 30 min. The contents
of the flask were transferred to a 20 ml vial that was main-
tained at 37°C for 4 h. The progress of the formation of the
monocarbamoylmethyl derivative was monitored by high-
Bis[(N-piperidino)carbamoyloxymethyl] Hydrogen
Phosphate (9c)
A solution of 8c (3.9 g, 0.08 mol) in cyclohexane (200 ml) pressure liquid chromatography (HPLC) on a ␮Bondapak
was hydrogenated over 5% Pd/C (100 mg) at a pressure of 30 C-18 reversed-phase column (150 × 3.90 mm, i.d.; Phenome-
psi. After 1 h, an additional 100 mg of catalyst was added and nex, Torrance, CA, USA) using a stepwise gradient of
hydrogenolysis was continued for a further 1 h. The catalyst MeOH:H2O (1:1) and 0.05 M NH4OAc, pH 4.4, as described

Bis(carbamoyloxymethyl) Esters of ddUMP
393
under aqueous stability studies, above. After completion of (solution A) and 0.05 M ammonium acetate, pH 4.4 (solution
the reaction, the solution was lyophilized. H2O (10 ml) was B) at a flow rate of 1 ml per min delivered according to the
added and the solution was again lyophilized; this procedure following program: 10 min of A; 10 min of A:B, 80:20 (v/v); 10
was twice repeated. Yield: 11 mg (93%). The identity and min of A:B, 60:40 (v/v); and 10 min of B. Eluted compounds
purity of the product was confirmed by HPLC and NMR were monitored with a variable-wavelength UV detector set
analysis. 1H NMR (MeOD): ␦ 7.65 (d, 1 H, H-6, J ס 8 Hz), 6.0 at 268 and 0.005 absorbance unit full scale (AUFS) sensitivity.
(dd, 1 H, H-1Ј), 5.77 (d, 1 H, H-5, J ס 8 Hz), 5.66 (d, 2 H, The retention times for POM2-ddUMP, POM1-ddUMP, ddU,
P(O)OCH2O, JPH ס 12 Hz), 4.25 (m, 2 H, H-5Ј), 2.92 (s, 6 H, and ddUMP were 28.79, 25.10, 14.72, and 8.48 min, respec-
N(CH3) ), 2.25 (m, 2 H, H-3Ј), 2.12 (m, 1 H, H-4Ј), 1.74 (m, tively. For DM2-ddUMP, DM1-ddUMP, ddU, and ddUMP
2
2 H, H-2Ј). MS: m/z 378 (MH+). Anal. (C13H20N3O8P): C, H, N. the retention times were 28.81, 25.31, 13.88, and 8.46 min,
respectively, and for DP2-ddUMP, DP1-ddUMP, ddU, and
ddUMP the retention times were 28.92, 25.33, 13.78, and 8.16
min, respectively.
Mono[(N-piperidino)carbamoyloxymethyl]
2Ј,3Ј-Dideoxyuridine 5Ј-phosphate (2c)
This compound was prepared from 1c (30 mg, 0.06
mmol) as described for the dimethyl analogue, 2b, above.
Yield: 21 mg (80%). H NMR (MeOD): ␦ 7.73 (d, 1 H, H-6,
Stability Studies of ddUMP Prodrugs in Human Plasma
1
Human plasma was obtained from the Blood Bank (M.
D. Anderson Cancer Center, Houston, TX). The reaction was
started by adding 100 ␮l of stock solutions (1 × 10−2 M) of
POM2-ddUMP (1a), DM2-ddUMP (1b), and DP2-ddUMP
(1c) in EtOH to plasma (1900 ␮l) contained in separate 3 ml
vials such that the final concentrations of prodrug were 1 ×
10−4 M. The samples were agitated for 20 s on a Vortex
shaker to ensure thorough mixing and then incubated at 37°C.
Samples (100 ␮l) from each vial were withdrawn at 0, 1, 5, 15,
30, 60, 120, 180, 360, and 1440 min and diluted with 4 volumes
of ice-cold MeOH. The mixtures were agitated on a Vortex
shaker for 1 min and then centrifuged at 10,000 rpm for 10
min to sediment precipitated protein. The supernatants were
analyzed by HPLC for parent compounds and degradation
products as described above for aqueous stability studies.
J ס 8 Hz), 6.20 (dd, 1 H, H-1Ј), 5.78 (d, 1 H, H-5, J ס 8 Hz),
5.71 (d, 2 H, P(O)OCH2O, JPH ס 12 Hz), 4.4 (m, 2 H, H-5Ј),
3.45 (m, 4 H, N(CH2)2; piperidino ␣ protons), 1.49 (m, 6 H,
(CH2)3; piperidino ␤ and ␥ protons), 2.22 (m, 2 H, H-3Ј), 2.15
(m, 1 H, H-4Ј), 1.71 (m, 2 H, H-2Ј). MH+ ס 418.21. Anal.
(C13H20N3O8P): C, H, N. 3.19
Aqueous Solubility of ddUMP Prodrugs
Potassium phosphate buffer (0.05 M), pH 7.4 (3 ml), was
added to 10 ml round-bottomed flasks containing 15.2 mg
each of 1a, 1b, and 1c. The mixtures were stirred at room
temperature for 2 h, then the contents of each flask were
transferred to 10 ml centrifuge tubes and centrifuged at 2000
rpm for 10 min. The supernatants were decanted and passed
through a 0.22 pm Millipore filter to remove any remaining
particulate matter. The concentrations of 1a, 1b, and 1c in
solution were determined by UV adsorption using the mea-
sured value of 7838 for the molar extinction coefficient (␧) of
these compounds at wavelength of 268 nm.
Stability of POM1-ddUMP and DP1-ddUMP in the
Presence of Snake Venom (Bothrops atrox)
Phosphodiesterase I
Snake venom (Bothrops atrox) phosphodiesterase I was
obtained from Sigma Chemical Co. and used as received. The
specific activity of the preparation was 0.029 U/mg of protein
where 1 U is defined as the amount that will hydrolyze 1.0
␮mol of bis(p-nitrophenyl) phosphate per minute at pH 8.8
and 37°C. DP1-ddUMP was dissolved in 0.05 M phosphate
buffer, pH 7.4, at a concentration of 10−4 M. One-milliliter
aliquots of this solution contained in 5.0 ml screw-capped
glass vials were incubated at 37°C in the absence or the pres-
ence of the enzyme. For the enzyme studies, 1 and 2 U of
phosphodiesterase I per ␮mol of substrate were used. A con-
trol reaction was run using 2 U of enzyme that had previously
been boiled for 10 min. To begin reaction the enzyme was
added to 0.2 ml of the drug solution contained in 1.5 ml micro
centrifuge tubes that were preequilibrated at 37°C. At inter-
vals of 0, 0.17, 0.34, 0.67, 1, 2, 3, 5, 8, 25 h thereafter, aliquots
Partition Coefficient of ddUMP Prodrugs
1-Octanol (3 ml; previously equilibrated with 0.05 M po-
tassium phosphate buffer, pH 7.4) was added to solutions of
1a, 1b, and 1c (ca. 6 mg each) in 0.05 M phosphate buffer, pH
7.4 (3 ml; previously equilibrated with 1-octanol) contained in
15 ml capped centrifuge tubes. The mixtures were agitated for
5 min. The octanol and aqueous layers were separated, and
the concentration of 1a, 1b, and 1c in each phase was deter-
mined by UV absorption using a value of 7838 for the molar
extinction coefficient (␧) of each compound a wavelength of
268 nm.
Stability Studies of ddUMP Prodrugs in Aqueous Buffers
Aliquots of a stock solution of POM2-ddUMP (1a), (0.1 ml) of the reaction mixture were withdrawn and added to
DM2-ddUMP (1b), and DP2-ddUMP (1c) in H2O (10−3 M) 0.1 ml of ice cold MeOH were added to deactivate the en-
were diluted with various buffers to a final concentration of zyme. The mixtures were agitated on a Vortex mixer (Queens
10−4 M. These solutions were stirred at room temperature. At Village, NY, USA) for 20 s and then centrifuged at 2000 rpm
selected time intervals (typically, 2, 4, 8, 12, 24, 30, 50, and 100 for 5 min. The supernatants were analyzed by HPLC on a
h) aliquots (50 ␮l) were removed and analyzed immediately ␮Bondapak C-18 reversed phase column (150 × 3.90 mm, i.d.;
for parent drug by HPLC on a ␮Bondapak C-18 reversed Phenomenex) protected with a Waters Guard-Pak precolumn
phase column (150 × 3.90 mm, i.d.; Phenomenex) protected filter (Milford, MA, USA) (0.22 ␮m). The mobile phase was
with a Waters Guard-Pak precolumn filter (0.22 ␮m). The a stepwise gradient of MeOH:H2O (1:1) (solution A) and 0.05
mobile phase was a stepwise gradient of MeOH:H2O (1:1) M ammonium acetate, pH 4.4 (solution B) at a flow rate of

394
Khan, Kumar, and Farquhar
Sch. 2. Synthesis of prodrugs 1b and 1c.
Sch. 3. Mechanism of release of 2a and 2b.
Aqueous Solubility
1 ml per min as described above under aqueous stability stud-
ies. The retention times for POM1-ddUMP and ddUMP were
25.10 and 8.48 min, respectively. For DP1-ddUMP and
ddUMP, the retention times were 25.33 and 8.16 min, respec-
tively. The substrate concentrations were determined spec-
trophotometrically by UV absorption using the measured
value of 7838 for the molar extinction coefficient (␧) at a
wavelength of 268 nm. The half-lives were determined by
linear least-square regression analysis of the pseudo-first-
order reactions.
The maximum solubility of 1a, lb, and 1c in 0.05 M phos-
phate buffer, pH 7.4, determined by stirring the compounds
with the buffer for 2 h at ambient temperature, were 3.6 mg/
ml for 1a, 8.65 mg/ml for 1b, and 1.9 mg/ml for 1c. The 4-fold
greater aqueous solubility of 1b over 1c is obviously due to
the reduced lipophilicity of the dimethylamino groups com-
pared to the piperidino groups.
Partition Coefficient
RESULTS AND DISCUSSION
The partition coefficients (P) of 1a, 1b, and 1c between
l-octanol and 0.05 M potassium buffer, pH 7.4, were 15.1 (log
P ס 1.18), 0.016 (log P ס −1.78), and 5.54 (log P ס 0.72),
respectively. Compounds with positive log P values (such as
1a and 1c) generally have move favorable membrane-
permeability characteristics than compounds with negative P
values (such as 1b).
Preparation of Bis(carbamoyloxymethyl) Esters of ddUMP
1b and 1c were prepared as shown in Scheme 2. Reaction
of the secondary amine, 4, with chloromethyl chloroformate,
5, yielded the chloromethyl carbamate, 6, which was reacted
with disilver benzyl phosphate, 7, to generate the correspond-
ing phosphotriester, 8. Catalytic hydrogenation of 8 yielded
the free acid 9 which was condensed with 2Ј,3Ј-dideoxyuridine
in the presence of triphenylphosphine and diethyl azodicar-
boxylate (DEAD) (the Mitsunobo reagent) to give the target
phosphotriesters, 1b and 1c (16).
Stability in Aqueous Buffers
When stirred at a concentration of 10−4 M in 0.05 M
aqueous buffers over the pH range 1 to 9, 1b was slowly
degraded with half-lives ranging from 0.7 h to 5.0 h. The
piperidino analogue, 1c, showed similar degradation kinetics
but was significantly more stable than 1b at alkaline pH. The
only products observed were the monocarbamoyloxymethyl
compounds 2b and 2c. By comparison, the POM2 analogue 1a
was markedly more stable under all pH conditions with half-
lives ranging from 1 day to 7 days (Table I).
The greater lability of 1b and 1c over 1a is most likely
due to the greater nucleophilicity of the carbamoyl group
compared to the acyl group. It is likely that DM2-ddUMP and
DP2-ddUMP rearrange by attack of the carbamoyl oxygen
atom on the phosphoryl group with cleavage of the P-OCH2
Preparation of Monocarbamoyloxymethyl Esters
of ddUMP
Mono(N,NЈ-dimethylcarbamoyloxy-methyl)- and mono(N-
peridinocarbamoyloxymethyl) 2Ј,3Ј-dideoxyuridine 5Ј-mono-
phosphate (2b and 2c, respectively) were prepared by selec-
tive hydrolysis of the parent diesters, 1b and 1c, with 0.05 M
NH4HCO3 buffer at pH 7.0.
Table I. Half-livesa of 1a, 1b, and 1c in Aqueous Buffers
Table II. Half-livesa of 1a, 1b, and 1c in Human Plasmab
Half-life (h)
Half-life
Buffer
1a
88
176
157
25
1b
1c
Compound
(h)
0.05 M potassium chloride, pH 1.0
0.05 M sodium acetate, pH 4.0
0.05 M potassium phosphate, pH 7.4
0.05 M Tris-HCl, pH 9.0
2.7
4.2
5.0
0.7
2.9
3.3
3.4
3.6
POM2-ddUMP, 1a
DM2-ddUMP, 1b
DP2-ddUMP, 1c
0.08
3.7
3.5
a Half-lives represent the average of triplicate determinations.
b Initial drug concentration: 10−4 M.
a Half-lives represent the average of triplicate determinations.

Bis(carbamoyloxymethyl) Esters of ddUMP
395
Fig. 3. Stability of DP2-ddUMP, 1c, in human plasma at 37°C.
Fig. 1. Stability of POM2-ddUMP, 1a, in human plasma at 37°C.
disappeared to give rise to two other products. The initial
product was identified as POM1-ddUMP, 2a, and the two
derived products as ddUMP and ddU. The comparatively
slow rate of hydrolysis of the second ester group compared to
the first suggests the involvement of an enzyme other than
carboxylate esterase. The most likely candidate is 5Ј-
nucleotide phosphodiesterase I (EC 3.1.4.1), an enzyme abun-
dant in mammalian tissues (17,18) and known to catalyze the
hydrolysis of a wide range of naturally occurring and synthetic
phosphodiesters (16). Although the mammalian enzyme was
not available to test this hypothesis, 2a was rapidly hydro-
lyzed to ddUMP when incubated with phosphodiesterase I
derived from snake venom (Bothrops atrox), demonstrating
that it is an excellent substrate for this enzyme (Fig. 4). Con-
version of the derived ddUMP to ddU most likely occurs by
nonspecific phosphatases present in human plasma.
linkage and release of formaldehyde as shown in Scheme 3.
The intermediate mixed anhydride is then hydrolyzed to give
the phosphodiester 2 and a carbamic acid, the latter sponta-
neously breaking down to give carbon dioxide, and the cor-
responding secondary amine. This type of reaction is likely to
be less facile with POM groups because the electron density
at the carbonyl oxygen atom in carboxylate esters is lower
than that of the corresponding carbamates.
Stability in Human Plasma
When incubated at a concentration of 10−4 M in human
plasma, 1a was degraded rapidly. Fifteen minutes after the
start of the incubation, none of the parent compound was
detectable (Fig. 1). By comparison, the half-life of 1a in phos-
phate buffer, pH 7.4, was 157 min (Table II). The degradation
of 1a was accompanied by the formation of a product that
reached a peak concentration at 45 min and then gradually
The stabilities of the two bis(carbamoyloxymethyl) com-
pounds, 1b and 1c, in human plasma were markedly different
from that of 1a (Figs. 2 and 3). Thus, the half-lives of 1b and 1c
were 3.7 h and 3.5 h, respectively, similar to that observed in
0.05 M phosphate buffer, pH 7.4. These findings demonstrate
that, unlike the POM2 compound, the two bis(carbamoyloxy-
Fig. 4. Stability of POM1-ddUMP, 2a, in the presence of phosphodi-
esterase I (0.001 U enzyme/mmol of 2a).
Fig. 2. Stability of DM2-ddUMP, 1b, in human plasma at 37°C.

396
Khan, Kumar, and Farquhar
2. N. W. Bischofberger, R. J. Jones, M. N. Arimilli, K. Y. Lin, M. S.
Louie, L. R. McGee, E. J. Prisbe, W. A. Lee, and K. C. Cundy.
Preparation of viricidal nucleotide analogs. PCT Int. Appl. 154
pp. (1995).
3. P. D. Cook, G. Wang, T. W. Bruice, V. Rajappan, K. Sakthivel,
K. D. Tucker, J. L. Brooks, J. M. Leeds, M. E. Ariza, and P. C.
Fagan. Preparation of nucleotide 5Ј-monophosphate mimics and
their prodrugs useful as antiviral, antibacterial, anticancer, and
immunomodulatory agents. PCT Int. Appl. 75 pp. (2003).
4. S. Warren. Targetable nucleotide-based prodrugs with cleavable
ester linkages. PCT Int. Appl. 137 pp. (1999).
5. K. C. Lieberman and C. Heidelberger. The metabolism of 32P-
labeled ribonucleotides in tissue slices and cell suspensions. J.
Biol. Chem. 316:823–830 (1955).
6. P. M. Roll, H. Weinfeld, E. Carroll, and G. B. Brown. The utili-
zation of nucleotides by the mammal. IV. Triply labeled purine
nucleotides. J. Biol. Chem. 220:439–454 (1956).
7. D. Farquhar, D. N. Srivastava, N. J. Kuttesch, and P. P. Saunders.
Biological reversible phosphate protective groups. J. Pharm. Sci.
72:324–325 (1983).
8. S. Khan, B. Nowak, W. Plunkett, and D. Farquhar. Nucleotide
bis(pivaloyloxy methyl)esters bypass nucleoside kinase defi-
ciency. Proc. Am. Assoc. Cancer Res. 31:425–426 (1990).
9. K. J. Sastry, P. N. Nehete, S. Khan, W. Plunkett, R. B. Arlinhaus,
and D. Farquhar. A newstrategy for the chemotherapy of ac-
quired immune immunodeficiency syndrome: membrane perme-
able dideoxyuridine monophosphate analogs as a potent inhibi-
tors of human immunodeficiency virus infection. Mol. Pharma-
col. 41:441–445 (1992).
Fig. 5. Stability of DP1-ddUMP, 2c, in the presence of phosphodies-
terase I (0.001 U enzyme/mmol of 2c).
methyl) prodrugs are essentially unaffected by carboxylate
esterases and other hydrolytic enzymes present in human
plasma. As with 1a, the phosphophodiester intermediates 2b
and 2c were formed initially and then gradually disappeared
to give rise to ddUMP and ddU. When incubated with snake
venom phosphodiesterase I (19–21), 2c was rapidly hydro-
lyzed to ddUMP demonstrating that, like 2a, it is an excellent
substrate for this enzyme (Fig. 5).
10. D. Farquhar, S. Khan, D. N. Srivastva, and P. P. Saunders. Syn-
thesis and antitumor evaluation of Bis[(pivaloyloxy)methyl] 2Ј-
Deoxy-5-fluorouridine 5Ј-Monophosphate (FdUMP): A strategy
to introduce nucleotides into cells. J. Med. Chem. 37:3902–3909
(1994).
11. D. N. Srivastva and D. Farquhar. Bioreversible phosphate pro-
tective groups: synthesis and stability of model acyloxymethyl
phosphates. Bioorg. Chem. 12:118–129 (1984).
12. O. Mitsunobo. The use of diethyl azadicarboxylate and triphe-
nylphosphine in the synthesis and transformation of natural prod-
ucts. Synthesis (Mass.) 1:1–28 (1981).
13. A. J. Kirby. New synthesis of monoalkyl phosphates. Chem. Ind.
(London) 47:1877–1878 (1963).
14. S. Khan. Design, synthesis and biological evaluation of 5Ј-
mononucleotide prodrugs: Strategies to introduce nucleotides
into cells. Ph.D. Dissertation UT M.D. Anderson Cancer Center,
Houston, TX (1997).
15. S. R. Khan and D. Farquhar. Bis(N,N-dimethylcarbamoyloxy-
methyl) 2Ј,3Ј-dideoxyuridine 5Ј-monophosphate (DM2-ddUMP):
a potential ddUMP prodrug. Tetrahedron Lett. 40:607–610
(1999).
16. H. F. Haugen and S. Skrede. Nucleotide pyrophosphate and
phosphodiesterase I. Organ distribution and activities in body
fluids. Clin. Chem. 23:1531–1537 (1977).
17. M. Land, R. A. Everard, and L. G. Butler. 5Ј-Nucleotide phos-
phodiesterase: features of substrate binding site as deduced from
specificity and kinetics of some novel substrates. Biochemistry
19:138–143 (1980).
CONCLUSIONS
DP2-ddUMP and DM2-ddUMP were synthesized as po-
tential membrane permeable prodrugs of ddUMP, and their
stabilities were evaluated in human plasma and aqueous buff-
ers. The findings clearly demonstrate that these compounds
are far more resistant to degradation by plasma enzymes than
the POM2 analogue reported earlier. However, we have not
yet ascertained that these bis(carbamoyloxymethyl) prodrugs
give rise to ddUMP within cells. This requires the availability
of radiolabeled formulations of DM2-ddUMP and DP2-
ddUMP, the synthesis of which are in progress. Success in this
prodrug approach requires that the diester generated within
cells by spontaneous hydrolysis of the first carbamoyloxy-
methyl group is further hydrolyzed by cellular phosphodies-
terases to ddUMP. Although this biotransformation cannot
be assured, the excellent facility with which hydrolysis occurs
in the presence of snake venom phosphodiesterase I lends
confidence that it will also occur within cells. Further studies
of bis(carbamoyloxymethyl) 5Ј-nucleotides as potential pro-
drugs of the free parent nucleotides are underway and will be
the subject of a future communication.
18. H. G. Khorana and W. E. Razell. Studies on polynucleotides. III.
Enzymic degradation; substrate specificity and properties of
snake venom phosphodiesterase. J. Bio. Chem. 234:2105–2113
(1959).
19. H. Fukazawa, T. Nishimura, N. Tanaka, and H. Suzuki. 5Ј-
Nucleotide phosphodiesterase and alkaline phospatase in tumor
cells: evidence for existence of novel species in the cytosol. Bio-
chim. Biophys. Acta 966:99–106 (1988).
20. H. G. Boman. Snake venom phosphodiesteras exonuclease ac-
tion. Nature 180:1181–1184 (1957).
21. M. Yoshikawa, T. Kato, and T. Takenishi. A novel method for
phosphorylation of nucleosides to 5Ј-nucleotides. Tetrahedron
Lett. 8:5065–5068 (1967).
ACKNOWLEDGMENTS
We gratefully acknowledge the financial support of grant
NIH CA 89386 (D.F.). This work was supported by grants
from FAMRI and CRF for S.R.K.
REFERENCES
1. J. Beres, W. G. Bentrude, G. Kruppa, P. A. McKernan, and R. K.
Robins. Synthesis and antitumor and antiviral activities of a series
of 1-beta-D-ribofuranosyl-5-halocytosine (5-halocytidine) cyclic
3Ј,5Ј-monophosphates. J. Med. Chem. 28:418–422 (1985).