A comparative study of the hydrolysis pathways of substituted aryl phosphoramidate versus aryl thiophosphoramidate derivatives of stavudine

Article abstract of DOI:10.1016/j.ejmech.2004.04.002

A comparative study of aryl phosphoramidate and aryl thiophosphoramidate derivatives of 2′,3′-didehydro-2′,3′-dideoxythymidine (d4T) was performed. The study focused on the nature of the substituents and the influence of a thiophosphoramidate in the structure of these derivatives. The rate of alkaline hydrolysis of these two types of d4T derivatives indicated that replacement of oxygen with sulfur decreases the rate of hydrolysis by twofold. Additionally, the activation energy (E_a) for the sulfur analogs is comparatively higher than that of the oxygen analogs. Notably, an intermediate was formed in the hydrolysis reaction of the sulfur analogs of d4T that was absent in the case of the oxygen analog, and the tentative structure of the intermediate was proposed based on LC/mass spectroscopy data. Using both HPLC and ³¹P-NMR techniques, we identified the hydrolysis product of the phosphoramidate derivatives and were able to show in in vitro studies that porcine liver esterase can hydrolyze the methyl ester portion of the phosphoramidate derivatives. Aryl phosphoramidate derivatives of d4T were 1000-fold more active than the corresponding aryl thiophosphoramidate derivatives, indicating that the energy of activation of hydrolysis of these phosphoramidate derivatives plays a significant role in their biological potency.

Full text of DOI:10.1016/j.ejmech.2004.04.002

European Journal of Medicinal Chemistry 39 (2004) 665–683
www.elsevier.com/locate/ejmech
Original article
A comparative study of the hydrolysis pathways of substituted aryl
phosphoramidate versus aryl thiophosphoramidate derivatives
of stavudine
T.K. Venkatachalam a,*, G.Yu a, P. Samuel b, S. Qazi c, S. Pendergrass d, F.M. Uckun d
a Department of Chemistry, Parker Hughes Institute, 2699 Patton Road, Roseville, MN 55113, USA
b Department of Pharmaceutical Sciences, Parker Hughes Institute, 2699 Patton Road, Roseville, MN 55113, USA
c Department of Bioinformatics, Parker Hughes Institute, 2699 Patton Road, Roseville, MN 55113, USA
d Departments of Virology and Immunology, Parker Hughes Institute, 2699 Patton Road, Roseville, MN 55113, USA
Received 5 December 2003; received in revised form 28 April 2004; accepted 28 April 2004
Abstract
A comparative study of aryl phosphoramidate and aryl thiophosphoramidate derivatives of 2′,3′-didehydro-2′,3′-dideoxythymidine (d4T)
was performed. The study focused on the nature of the substituents and the influence of a thiophosphoramidate in the structure of these
derivatives. The rate of alkaline hydrolysis of these two types of d4T derivatives indicated that replacement of oxygen with sulfur decreases the
rate of hydrolysis by twofold. Additionally, the activation energy (Ea) for the sulfur analogs is comparatively higher than that of the oxygen
analogs. Notably, an intermediate was formed in the hydrolysis reaction of the sulfur analogs of d4T that was absent in the case of the oxygen
analog, and the tentative structure of the intermediate was proposed based on LC/mass spectroscopy data. Using both HPLC and 31P-NMR
techniques, we identified the hydrolysis product of the phosphoramidate derivatives and were able to show in in vitro studies that porcine liver
esterase can hydrolyze the methyl ester portion of the phosphoramidate derivatives. Aryl phosphoramidate derivatives of d4T were 1000-fold
more active than the corresponding aryl thiophosphoramidate derivatives, indicating that the energy of activation of hydrolysis of these
phosphoramidate derivatives plays a significant role in their biological potency.
© 2004 Elsevier SAS. All rights reserved.
Keywords: Phosphoramides; Thiophosphoramides; Stavudine; Activation energy; HPLC; 31P-NMR
1. Introduction
active metabolite, AZT-triphosphate, seems to be the conver-
sion of the monophosphate derivative to the diphosphate
derivative, whereas the rate limiting step for the intracellular
generation of the bioactive metabolite of 1, stavudine-
triphosphate, was reported to be the conversion of the nucleo-
side to its monophosphate derivative [10–12]. Anti-HIV ddN
derivatives primarily rely on nucleoside and nucleotide ki-
nases to convert them into the corresponding 5′-
Acquired immunodeficiency syndrome (AIDS) is the dis-
ease affecting a significant part of the population in recent
years due to the epidemic spread of human immunodefi-
ciency virus (HIV) [1–7]. Three categories of anti-retroviral
agents in clinical use are nucleoside analogs, nucleoside
reverse transcriptase inhibitors (NRTI), such as zidovudine
and stavudine 2′,3′-didehydro-2′,3′-dideoxythymidine (d4T,
1), protease inhibitors and non-nucleoside reverse tran-
scriptase inhibitors (NNRTI) [4–7]. The 5′-triphosphates of
2′,3′-dideoxynucleoside analogs (ddN) that are generated by
nucleoside and nucleotide kinases, are potent inhibitors of
HIV reverse transcriptase [8–9]. The rate limiting step for the
conversion of 3′-azido-3′-deoxythymidine (AZT) to its bio-
O
HN
N
O
HO
O
H
* Corresponding author.
1
E-mail address: jhumphrey@ih.org (T.K. Venkatachalam).
© 2004 Elsevier SAS. All rights reserved.
doi:10.1016/j.ejmech.2004.04.002

666
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
triphosphates as discussed before. However, such com-
pounds were found to act as poor substrates for nucleoside
kinases [10,13–16]. Consequently, development of pro-drug
strategies were sought that could bypass the initial nucleo-
side kinase activation. In an attempt to overcome the depen-
dence of ddN analogs on intracellular nucleoside kinase
activation, we and others have prepared aryl phosphate de-
rivatives of zidovudine and 1 [11,12,17–21]. Some of these
derivatives were found to be potent anti-HIV agents with
subnanomolar IC50 values. In the present study, we prepared
thiophosphoramidate derivatives of 1 and compared their
alkaline hydrolysis rates with those of phosphoramidate de-
rivatives.
2 weeks. Solvent was evaporated under reduced pressure to
furnish viscous oil. The crude reaction product was further
purified using column chromatography using (chloroform/
methanol = 10:1). The appropriate fractions were pooled
together and once again the solvent was removed under
reduced pressure to yield a viscous oil which was subse-
quently subjected to preparative TLC (chloroform/
methanol) = 10:1 to obtain analytically pure compound in
12.3% yield. Further modifications of this procedure using
solvents such as dichloromethane, chloroform and carrying
out the reaction at room temperature did not yield the desired
product. An increase in temperature beyond 50 °C resulted in
decomposition of the product.
2. Chemistry
2.1. 2′,3′-Didehydro-3′-deoxythymidine 5′-(4-bromo-phenyl
methoxyalaninyl thiophosphate) (4)
NMR spectra were obtained at ambient temperature in
deuterated solvent. Chemical shifts are reported as d values in
parts per million and referenced to the solvent or an internal
standard. Coupling constants (J) are reported in Hz. FT-IR
spectra were obtained as KBr pellets. Melting points (m.p.)
are uncorrected. HPLCs were obtained using an analytical
RP-18 Lichrospher column, 5 µm (4.6 × 150 µm) and
acetonitrile/water as the eluent. The flow rate was maintained
at 1.0 ml min–1 and the detection wavelength was set at
266 nm. The column was maintained at room temperature
throughout the analysis. Column and TLC chromatography
was performed using silica gel 60, 230–400 mesh. Com-
pounds 7–12 were prepared according to previously pub-
lished methods [6].
1H-NMR (300 MHz, CDCl3) d 1.37 (m, 3H), 1.81 (s, 3H,
3.73 (d, 3H, J = 5.1 Hz), 4.02–4.41 (m, 4H), 5.04 (m, 1H),
5.92 (s, 1H), 6.32 (m, 1H), 6.99–7.10 (m, 3H), 7.26 (m, 1H),
7.43 (d, 2H, J = 8.4 Hz), 8.62 (m, 1H). 13C-NMR (75 MHz,
CDCl3) d 12.8, 21.3, 51.2, 53.0, 67.2, 68.1, 84.7, 90.1, 111.5,
118.7, 123.1, 127.6, 132.7, 133.6, 135.8, 149.6, 150.8, 163.6.
31P-NMR (121 MHz, CDCl3) d –14.39, –13.92. UV (MeOH)
kmax: 224 (sh), 265 nm. IR (KBr, cm–1): 3456, 2360, 1693,
1483, 1216, 1143, 1011, 912, 828. HPLC: 19.3, 19.7 min.
2.2. 2′,3′-Didehydro-3′-deoxythymidine 5′-(phenyl
methoxyalaninyl thiophosphate) (2)
Typical procedure for the preparation of aryl thiophos-
phoramide derivatives of 1,2′,3′-didehydro-3′-deoxythymi-
dine 5′-(p-bromophenyl methoxyalaninyl thiophosphate)
1H-NMR (300 MHz, CDCl3) d 1.37 (m, 3H), 1.80 (s, 3H),
3.73 (d, 3H, J = 6.6 Hz), 3.96–4.44 (m, 4H), 5.04 (m, 1H),
5.90 (s, 1H), 6.33 (m, 1H), 6.83 (d, 2H, J = 8.7 Hz), 7.02 (m,
1H), 7.12–7.34 (m, 6H), 8.36 (m, 1H). 13C-NMR (75 MHz,
CDCl3) d 12.6, 21.4, 51.1, 53.0, 67.0, 67.9, 84.9, 90.1, 111.6,
121.2, 125.6, 127.4, 129.7, 133.6, 135.9, 150.7, 163.6. 31P-
NMR (121 MHz, CDCl3) d –14.64, –14.24. UV (MeOH)
kmax: 265 nm. IR (cm–1): 3297, 1693, 1593, 1489, 1211,
1149, 1091, 923, 776, 756. LC/MS: 511 (M + Na). HPLC:
16.1, 16.5 min. Yield: 11.7%.
(4). Under
a nitrogen atmosphere, a solution of
p-bromophenol (34.60 g, 0.20 mol) and triethylamine
(27.85 ml, 0.20 mol) in anhydrous diethyl ether (200 ml) was
added dropwise to a vigorously stirred solution of thiophos-
phoryl chloride (20.31 ml, 0.20 mol) in ether (300 ml) at
0 °C. The mixture was allowed to warm to ambient tempera-
ture, with stirring for 5 h, and then heated under reflux for 2 h.
The mixture was filtered and the precipitate was washed with
diethyl ether (100 ml). The combined filtrate and washings
were evaporated to dryness under reduced pressure to yield
colored oil. The oil was subjected to vacuum distillation to
yield the product as a colorless oil (p-bromoarylthiophos-
phorylchloride). Yield: 50%; b.p.: 140–143 °C/0.7 mm) [6].
A solution of L-alanine methyl ester hydrochloride (0.28 g,
2.0 mmol) and 1.0 ml triethylamine in anhydrous acetonitrile
(20 ml) was added dropwise with vigorous stirring to a
solution of 2.0 mmol of appropriately substituted aryl thio-
phosphoryl chloride in 20 ml anhydrous acetonitrile at 0 °C.
The reaction mixture was slowly warmed to ambient tem-
perature with stirring, and 0.35 g (5.0 mmol) of 1,2,4-triazole
was added to the reaction mixture. The contents were stirred
at room temperature for 6 h, and 0.224 g (1.0 mmol) of d4T
was added and the reaction was allowed to stir at 50 °C for
2.3. 2′,3′-Didehydro-3′-deoxythymidine 5′-(p-fluorophenyl
methoxyalaninyl thiophosphate) (3)
1H-NMR (300 MHz, CDCl3) d 1.37 (m, 3H), 1.82 (s, 3H),
3.73 (d, 3H, J = 4.8 Hz), 3.96–4.42 (m, 4H), 5.04 (m, 1H),
5.93 (s, 1H), 6.33 (m, 1H), 7.01–7.32 (m, 6H), 8.30 (m, 1H).
13C-NMR (75 MHz, CDCl3) d 12.7, 21.4, 51.2, 53.0, 67.1,
68.0, 84.8, 90.1, 111.5, 116.4, 122.7, 127.5, 133.5, 135.8,
150.7, 163.5. 31P-NMR (121 MHz, CDCl3) d –13.56, –13.88.
19F-NMR (282 MHz, CDCl3) d –41.57. UV (MeOH) kmax
:
265 nm. IR (cm–1): 3261, 1693, 1509, 1457, 1196, 1149,
1089, 916, 839. LC/MS: 522 (M + Na). HPLC: 17.0,
17.3 min. Yield: 8.3%.

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
667
2.4. 2′,3′-Didehydro-3′-deoxythymidine
phosphorodichloridate. This was treated with
L-alanine me-
5′-(p-methoxyphenyl methoxyalaninyl thiophosphate) (5)
thylester hydrochloride at –78 °C yielding the corresponding
thiophosphorochloridate derivatives. In the subsequent step,
these derivatives were reacted with 1 in anhydrous acetoni-
trile in the presence of triethylamine and triazole to obtain the
target compounds, which were then purified.
1H-NMR (300 MHz, CDCl3) d 1.40 (m, 3H), 1.82 (s, 3H),
3.73 (d, 3H, J = 5.4 Hz), 3.78 (m, 3H), 3.88–4.42 (m, 4H),
5.03 (m, 1H), 5.90 (s, 1H), 6.33 (m, 1H), 6.83 (d, 2H,
J = 8.7 Hz), 6.99 (m, 1H), 7.04–7.11 (m, 2H), 7.29 (m, 1H),
8.18 (m, 1H). 13C-NMR (75 MHz, CDCl3) d 12.9, 21.4, 51.1,
53.0, 55.9, 67.0, 67.9, 84.9, 90.0, 111.6, 114.7, 122.1, 127.4,
133.6, 135.9, 144.0, 150.7, 157.1, 163.5. 31P-NMR
(121 MHz, CDCl3) d –13.69, –13.27. UV (MeOH) kmax: 225
(sh), 270 nm; IR (cm–1): 3287, 2953, 1690, 1506, 1466,
1201, 1145, 1034, 912. LC/MS: 542 (M + Na). HPLC: 16.2,
16.4 min. Yield: 16.1%.
3.1. Alkaline hydrolysis of phosphoramidates
and thiophosphoramidates of 1
In our earlier studies [19–21] we have examined the alka-
line hydrolysis of phosphoramidate derivatives of 1 and con-
cluded that the superior biological activity shown by these
compounds is due to the rapid formation of an important
metabolite, namely the alaninyl monophosphate derivative of
d4T (6, X = O) [22]. In addition, we have demonstrated that
the presence of electron withdrawing groups on the phenyl
ring enhances the activity for these phosphoramidate deriva-
tives by improving the rate of hydrolysis. By analogy we
have also used an enzyme esterase (in vitro) to further sup-
port our assertion that the biological activity of these phos-
phoramidate derivatives is directly related to their rate of
hydrolysis. We [20,22] and others [11,12,27] have reported
that the first step involves the hydrolysis of the methylester to
the corresponding acid, which simultaneously cyclize and
eliminate the phenoxy moiety of the compound that ulti-
mately results in the formation of the most active metobolite,
6 (X = O). However, we could not isolate the intermediate
cyclized product to unequivocally decipher the mechanism
of elimination of the phenoxy moiety in the structure of these
phosphoramidate derivatives.
2.5. 5′-[4-Flurophenyl methoxyalaninyl
phosphate]-2′,3′-didehydro-3′-deoxy thymidine (9)
Yield: 46%, m.p. 42–44 °C, UV (MeOH) kmax: 210 and
266 nm; IR: 3423, 3245, 3072, 2954, 1691, 1504, 1247,
1089, 1037, 939 cm–1, 1H-NMR (CDCl3) d 10.08 (bs, 1H),
7.16 (m, 1H), 7.08 (m, 2H), 6.91 (m, 3H), 6.20 (m, 1H), 5.79
(t, 1H), 4.92 (m, 1H), 4.42 (t, 1H), 4.22 (m, 2H), 3.85 (m,
1H), 3.58* (m, 3H), 1.70* (m, 3H), 1.22* (m, 3H); 13C-NMR
(CDCl3) d 173.5 (q), 163.8 (d), 160.7, 157.5, 150.7 (d), 145.7
(q), 135.3 (d), 132.7 (d), 126.9 (d), 121.3 (t), 115.8 (q), 110.8
(d), 89.2 (d), 84.2 (d), 66.8 (t), 52.2, 49.8 (d), 20.4 (d), 12.1
(d); 31P-NMR (CDCl3) d 3.80 (d), 3.22 (d); 19F-NMR
(CDCl3) d –42.8 (t); HPLC tR: 6.3, 6.6 min.
3. Results and discussion
In the present investigation, we aimed to examine the
hydrolysis of thiophosphoramidate derivatives synthesized
as shown in Scheme 1. We have determined the rate of
We have previously reported the rates of alkaline hydroly-
sis for the phenyl phosphoramidate derivatives of 1 at room
temperature [19–22]. The alkaline hydrolysis profiles of the
phosphoramidate analogs of nucleoside monophosphates
under acidic, neutral and alkaline conditions were reported
by several groups [23–28]. In the present study, we synthe-
sized the corresponding phenyl thiophosphoramidate deriva-
tives in order to evaluate their biological activity as well as
their hydrolysis profiles. The synthesis was accomplished by
a procedure represented by Scheme 1.
O
HN
X
O
N
HO P O
NH
H
O
HOOC
Me
In brief, thiophosphoryl chloride was treated with various
substituted phenols in the presence of triethylamine and anhy-
drous acetonitrile at low temperatures to produce the thio-
H
6 (X=O, S)
O
HN
S
S
a
b
O
N
R
O P O
R
OH
R
O P Cl
Cl
O
NH
Me
H
H
MeOOC
R = H (2), F (3), Br (4), OMe (5)
Scheme 1. Synthesis of substituted aryl thiophosphoramidate derivatives of d4T. (a) PSCl3/Et3N, (b) L-alanine methyl ester, 1,2,4-triazole, Et3N, 1.

668
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
formation of each of the hydrolysis products to further under-
standing of the mechanism of alkaline hydrolysis.
phosphoramidate derivatives are less reactive than the phos-
phoramidate derivative, and therefore, has a higher transition
state energy.
Table 1shows the rate of formation of each of the metabo-
lites at various temperatures during alkaline hydrolysis of
stampidine 7, a bromo-substituted phosphoramidate deriva-
tive of 1. The table also gives the energy of activation {(Ea) in
kcal mol–1} calculated using theArrhenius equation. The rate
of disappearance of the starting material is found to be slower
than the rate of formation of metabolite. This suggests that
the first step involving the hydrolysis of methylester to the
corresponding acid is a slow step. In the second step the
cyclized intermediate product eliminates the phenoxy moiety
to form the metabolite and that this step may involve the build
up of the intermediate before proceeding. Since multiple
steps are involved for the formation of the metabolite and
intermediates we are unable to determine the kinetic order of
this process. However for simplicity we assumed this process
follows first order rate early in the kinetic cycle. The rate of
formation for each metabolite increased with temperature as
expected. The energy of activation of formation 6 (X = O)
was 12.8, 13.1 kcal mol–1 for 1 and 15.1 kcal mol–1 for
p-bromophenol.
We next examined the hydrolysis profiles of compounds
containing an electron donating methoxy group in the phenyl
moiety of both phosphoramidate 8 as well as thiophosphora-
midate 5 derivatives of 1. Fig. 1 shows the HPLC chromato-
If we compare the rate of formation of 1, 6, and
p-bromophenol as well as the disappearance rate of the start-
ing material from the alkaline hydrolysis of 7 with that of 4, it
is evident that the hydrolysis of the latter was almost 10-fold
slower, and the difference in Ea values indicates that thio-
Table 1
Rate constants and the corresponding energy of activation (Ea) values for the
4-bromo-substituted derivatives
O
HN
X
O
N
Br
O
P O
NH
O
H
MeOOC
Me
H
X = S (4), O (7)
Compound
Temperature 6 a
(°C)
1 a
Phenol a Initial
compound a
disappearance
rate
7
–10
0
0.23 0.30
0.88 1.14
1.58 1.69
2.49 2.14
4.31 3.07
12.8 13.1
0.06 0.04
0.27 0.19
0.70 0.46
24.5 24.9
0.05
0.70
0.82
1.33
2.19
15.1
0.02
0.09
0.16
21.2
0.19
0.46
8
1.25
15
25
1.91
4.12
Arrhenius (Ea) b NA
13.9
4
8
0.04
15
25
0.07
0.37
Arrhenius (Ea) b NA
23.4
a Rate constants were calculated using first order rate equations and are
expressed in h–1. The values are the average of three independent runs and
the error was 5%.
b Activation energies (Ea) were calculated using the Arrhenius equation
and are expressed in kcal mol–1
.
Fig. 1. Kinetics of alkaline hydrolysis for 8 at room temperature.

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
669
gram of the reaction between base and phosphoramidate
derivative at various intervals of time.
a sulfur in place of an oxygen in these phosphoramidate
derivatives apparently stabilizes this intermediate. To our
knowledge, this is the first time that formation of such an
intermediate has been reported during alkaline hydrolysis of
thiophosphoramidate derivatives of 1. Unfortunately, there is
no literature precedence for this putative intermediate of
thiophosphoramidate analogs of stavudine. The LC/mass
spectral data for the two peaks and their proposed intermedi-
ate structures are shown in Table 3. Table 2 further shows that
the rate of hydrolysis of 5 is faster than 8 at various tempera-
tures. This can be rationalized due to the fact that the thio-
phosphoramidate derivative reacts by an alternate pathway.
Additionally, this result indicates that the rate of intermediate
formation is much higher than the formation of other prod-
ucts (Table 2). Several attempts to synthesize these interme-
diates were unsuccessful due to elimination of the phenoxy
moiety. Although one would expect the phenoxy group as the
leaving group in preference over the nucleoside moiety dur-
ing alkaline hydrolysis, the intermediate products retained
the phenoxy moiety in their structure (Table 3). The differ-
ence in the nucleophilicity between the sulfur and oxygen
analogs could account for the differences in the intermediates
formed in the reaction. Also, we see the formation of d4T,
which demonstrates the tentative structure of the intermedi-
ate. Further more the p-fluoro substituted thiophosphorami-
date derivative gave an intermediate which is different in
structure as compared to the other two derivatives. Based on
the LC/mass spectral fragmentation the proposed structure
seem to be the carboxylic acid. The energy of activation
varied from 27 to 36 kcal mol–1 for most of the products.
We next synthesized the fluoro-substituted phosphorami-
date 9 as well as thiophosphoramidate derivative 3 for com-
parison. Compound 9 underwent hydrolysis slower com-
The peak at 8.49 and 9.38 min corresponds to the starting
material (distereoisomers) and the peak at 4.98 min repre-
sents the methoxyphenol formation. The other peaks in the
chromatogram at 2.7 and 1.9 min are those of 6 and 1,
respectively. The retention times of the above are confirmed
by comparison with authentic samples of appropriate com-
pounds. At 1 h methoxyphenol, 1 and 6 were detectable. At
3 h, 15 min, almost 50% of 8 was converted into the hydroly-
sis products. The rate of formation of the hydrolysis products
and the Ea values are shown in Table 2.
Overall, the temperature dependent formation of the hy-
drolysis products was slower for the methoxy-substituted
compounds 5 and 8 than for the bromo-substituted com-
pounds 4 and 7 (Tables 1 and 2). This was expected due to the
electron donating substituents on the phenyl ring should
decrease the hydrolysis rate whereas electron withdrawing
groups should increase the hydrolysis rate. As shown in
Table 2, the energy of activation for the formation of 6 from
the parent 8 was 15.7 kcal mol–1. The energies of activation
for all other products were higher than that of 6.
Next we examined the alkaline hydrolysis of 5 at room
temperature (Fig. 2). Within 30 min we observed two inter-
mediate peaks in the chromatogram, one at 8.4 min and
another at 10.0 min. Formation of these two intermediate
products during alkaline hydrolysis of the thiophosphorami-
date derivative 5 is in sharp contrast with the results obtained
from the phosphoramidate derivative 8 (Fig. 1). In our earlier
publications on the hydrolysis profiles of phosphoramidate
derivatives of 1 [21], we were unable to observe this interme-
diate, most likely due to their unstable nature preventing
detection under our experimental conditions. Introduction of
Table 2
Rate constants and the corresponding energy of activation (Ea) values for the 4-methoxy-substituted derivatives
O
HN
X
O
N
MeO
O P O
O
NH
H
H
MeOOC Me
X = S (5), O (8)
Compound
Temperature (°C)
6 a
1 a
Phenol a
Inter. a
Initial compound a
disappearance rate
8
8
0.06
0.20
0.30
15.7
0.03
0.19
0.57
29.9
0.05
0.40
0.99
29.4
0.02
0.15
0.39
26.8
0.10
0.46
1.59
26.7
0.05
0.05
0.03
NA
NA
0.01
0.06
0.19
27.1
0.05
0.14
1.34
36.1
15
25
NA
8
NA
NA
Arrhenius (Ea) b
NA
5
0.06
0.33
0.91
38.6
15
25
NA
Arrhenius (Ea) b
Inter. = Intermediate.
a Rate constants were calculated using first order rate equations and are expressed in h–1. The values are the average of three independent runs and the error
was 5%.
b Activation energies (Ea) were calculated using the Arrhenius equation and are expressed in kcal mol–1
.

670
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
Fig. 2. HPLC profile of compound 5 treated with sodium hydroxide at various time intervals.
pared to the bromo-substituted compound 7, which is
understandable due to lower electron withdrawing capacity
of fluorine as compared to bromine. However, we could not
identify any intermediates from the HPLC profiles, a result
similar to that observed for the bromo-substituted com-
pound. Table 4 shows the rate of formation of various prod-
ucts for the fluoro-substituted derivative at various tempera-
tures.
alkaline hydrolysis profiles of fluoro thiophosphoramidate
(3); we were able to observe the formation of an intermediate
product (Fig. 3) similar to that in the case of 5. Thus, unlike
compound 4, compound 3 yielded the intermediate product.
In order to explain various trends associated with the
substitutions on the phenyl ring of these thiophosphorami-
dates as well as phosphoramidates, we prepared and ana-
lyzed the corresponding unsubstituted analogs 2 (Fig. 4) and
10, respectively (Table 5). We observed the formation of
phenol, 1 and 6 (X = O, S) during the alkaline hydrolysis of
The energy of activation varied from 9.5 to 13.7 kcal
mol–1 for most of the products. When we examined the

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
671
of activation was approximately 25 kcal mol–1 for most of the
products of the phosphoramidate derivatives (X = O).
There are two possible pathways through which products
can result for the phosphoramidate and thiophosphoramidate
derivatives (Fig. 5).
Table 3
LC/mass spectral analysis of the intermediate product of thiophosphorami-
date derivatives
Compound Observed
mass of
Tentative structure
intermediate
Compound A forms products C and D at rate k1, or C and
D can be formed through an intermediate B with rates k2 and
k3. Therefore, the rate of formation of products relates to k1
and the rate of formation of B can be used to derive k2.
Differences will be observed between C and D formation if
the reaction scheme traverses through B. As discussed before
caution should be exercised in comparing rate of formation
of products and intermediates when the kinetic order is not
known. However, our interest was to compare the rate of
formation of intermediate and products between phosphora-
midate and thiophosphoramdiate under similar assumption
and experimental conditions.
Using this notation, we conclude that in the case of phos-
phoramidate compounds (X = O), lack of detection of inter-
mediate forces us to assume that the reaction proceeds by
forming C and D directly. On the other hand in the case of
thiophosphoramidate derivatives we were able to observe the
formation of intermediate B. These steps are further illus-
trated through the transition state diagram shown in Fig. 6.
Thus, the presence of sulfur instead of oxygen in the
structure of phosphoramidate derivatives greatly influences
the pathway through which the products are formed during
hydrolysis.
2
342 (M – 1)
S
ONa
O P
NH
+ 2Na
Me
COOMe
H
330 (M + 1)
S
ONa
+ 2Na
O
P
NH
+
COOMe
H
187 (M)
S
-
O
O P
NH
+
105 (M + 2)
351 (M + 2)
H2N
Me
COOMe
To further understand the thermodynamic factors associ-
ated with hydrolysis pathway between phosphoramidate and
thiophosphoramidate derivatives, we evaluated Gibbs free
energy as well as enthalpy for phosphoramidates and thio-
phosphoramidates. We initially obtained the energy of acti-
vation by using Arrhenius equation:
3
SH
OH
F
O P+
+3 Na
NH
COOH
Me
H
(1)
log k = log A − Ea ⁄ 2.303RT
306 (M + 1)
where, k represents the specific reaction rate, A is a constant,
Ea is the energy of activation, R is the gas constant (1.987 cal
deg–1 mol–1) and T is the absolute temperature. Plotting log k
vs. 1/T we were able to obtain the slope of the line that in turn
gives the following Eq. (2) comprising energy of activation:
SH
OH
NH
+3 Na
F
O P+
+
Me
H
(2)
Slope = −Ea ⁄ 2.303R
5
372 (M – 1)
We next invoked the transition state theory, or absolute
rate theory, according to which an equilibrium is considered
to exist between the normal reactant molecule and an acti-
vated complex of these molecules. Subsequent decomposi-
tion of the activated complex leads to product. For an el-
ementary bimolecular process, the reaction may be written
as:
S
ONa
+ 2Na
MeO
O
P
NH
Me
COOMe
H
[ A…..OH]‡
A + OH-
P
these unsubstituted compounds. The rates of hydrolysis were
slower than those of the compounds with electron withdraw-
ing substituents in the aryl moiety. We found that the energy
The double dagger is used to designate the activated state.
In principle, the transition state theory gives the influence of

672
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
Table 4
Rate constants and the corresponding energy of activation (Ea) values of the 4-fluoro substituted derivatives
O
HN
X
O
N
F
O
P
O
O
H
NH
MeOOC
Me
H
X = S (3), O (9)
Compound
Temperature (°C)
6 a
1 a
Phenol a
Inter. a
Initial compound a
disappearance rate
9
8
0.45
1.01
1.52
11.7
0.04
0.09
0.34
21.7
1.18
2.17
3.16
9.6
0.35
0.95
1.31
12.7
0.01
0.03
0.09
20.2
NA
NA
0.73
1.32
2.94
13.7
0.02
0.07
0.18
21.8
15
25
NA
8
NA
Arrhenius (Ea) b
NA
3
0.02
0.06
0.18
19.6
0.07
0.14
0.42
17.4
15
25
NA
Arrhenius (Ea) b
Inter. = Intermediate.
a Rate constants were calculated using first order rate equations and are expressed in h–1. The values are the average of three independent runs and the error
was 5%.
b Activation energies (Ea) we calculated using the Arrhenius equation and are expressed in kcal mol–1
.
Table 5
Rate constants and the corresponding energy of activation (Ea) values of the unsubstituted derivatives
O
HN
X
O P O
NH
O
N
O
H
MeOOC
H
X = S (2), O (10)
Compound
Temperature (°C)
6 a
1 a
Phenol a
Inter. a
Initial compound
adisappearance rate
10
8
0.08
0.53
1.14
24.3
0.05
0.08
0.27
16.7
0.08
0.34
1.08
25.7
0.10
0.25
0.52
15.7
0.09
0.51
1.25
25.2
0.01
0.03
0.05
18.1
NA
0.02
0.08
0.21
23.5
0.01
0.07
0.15
23.3
15
25
NA
8
NA
NA
Arrhenius (Ea) b
NA
2
0.16
0.29
0.99
18.1
15
25
NA
Arrhenius (Ea) b
Inter. = Intermediate.
a Rate constants were calculated using first order rate equations and are expressed in h–1. The values are the average of three independent runs and the error
was 5%.
b Activation energies (Ea) were calculated using the Arrhenius equation and are expressed in kcal mol–1
.
temperature on reaction rates by the general equation shown
below:
position. This constant can be considered a good approxima-
tion to the following equation:
k = (meDSt⁄R)e−DEt⁄RT
(4)
(3)
m = (RT ⁄ Nh)
In the above equation, k represents the rate of reaction at a
given temperature, m corresponds to the frequency of decom-
where R is the gas constant, T being the absolute temperature,
N is Avogadro’s number and h is Plank’s constant. Using the

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
673
Fig. 3. HPLC profile of compound 3 treated with sodium hydroxide at various time intervals.
above equations and following the chemical principles such
as:
3.2. Relationship between r and Ea for phosphoramidate
and thiophosphoramidate derivatives
We explored the relationship between Hammett sigma (r)
[29] and the energy of activation obtained for each set of the
previously discussed compounds to find out whether any
correlation exists among the substituents and Ea/TnS. The
(5)
DG = DH − TDS
we were able to calculate the Ea, DG and TDS for these
derivatives.

674
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
Fig. 4. HPLC profile of compound 2 treated with sodium hydroxide at various time intervals.
analysis of covariance (ANCOVA) model [30] significantly
predicted TDS values (R2 = 0.85, F(2,5) = 13.68, P = 0.0094).
There was a significant effect of the covariate sigma
(P = 0.0064) and X = S/X = O categorical group (P = 0.045).
The variance explained by the sigma covariate (F-
ratio = 20.2) was greater than the effect of X = S/X = O group
(F-ratio = 7.1). Therefore, the difference between X = O and
X = S is significant when the sigma values are taken in to
account. A similar effect profile was observed with Ea values
(Fig. 7; R2 = 0.87, F(2,5) = 16.53, P = 0.0063). There was a
significant effect of the sigma covariate (P = 0.0045) and the
k1
A
C + D
k2
k3
B
Fig. 5. Pathways for alkaline hydrolysis of phosphormidates and thiophos-
phoramidates.

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
675
ramidate showed a higher energy of activation compared to
phosphoramidate derivatives. The most striking difference is
the entropy of the reaction, which is represented in the table
as TDS. For electron withdrawing substituents such as bromo
and fluoro groups a negative value with a magnitude of
–7 kcal mol–1 for the phosphoramidate derivatives was ob-
served, while the electron donating group gave a value of
+3.8 kcal mol–1. The variation of TDS values for the thio-
phosphoramidate derivatives were significantly less with the
exception of the methoxy substituent, which showed a value
of +13.9 kcal mol–1.
O
N
HN
X
O
R
O P O
NH
O
H
MeOOC
Me
H
3.3. 31P-NMR studies on the reaction between
phosphoramidate and sodium hydroxide
A
B
We next set out to find by NMR what products are ob-
served during the reaction. The sample was dissolved in a
minimal amount of deuterated methanol and to this was
added sodium hydroxide solution and the NMR spectrum
was taken after various time intervals. A capillary containing
1% phosphoric acid in water served as an internal standard,
which was adjusted to 0 ppm. Table 7 depicts selected chemi-
cal shift values observed for various compounds.
Ea2
Ea1
C+D
Ea3
C+D
X=O
X=S
Fig. 6. Transition state diagram for aryl substituted phosphoramidate and
thiophosphoramidate derivatives of 1.
We chose the phosphoramidates with electron withdraw-
ing substituents as these compounds could be hydrolyzed
faster. The phosphoramidate derivatives were obtained as
distereoisomers consisting of both “R” and “S” isomers due
to the chirality at phosphorus center. In general a 40:60
enantiomeric ratio was obtained for all the derivatives. The
two peaks corresponding to each isomer of the diastereo-
meric mixture of the 7 was observed at 4.56 and 4.4 ppm,
respectively, relative to the standard phosphoric acid posi-
tioned at 0 ppm. Within 3 h we were able to observe the
completion of the reaction and the 31P-NMR showed five
major peaks in the spectrum. In the second experiment, we
used a single isomer of the bromo-substituted phosphorami-
date derivative, which showed a peak at 4.4 ppm at 0 min. In
1 h we once again observed five peaks corresponding to five
different compounds. The chemical shift values observed
(after giving allowance for the minor change in shift values
due to sample character) were similar to those obtained for
the distereoisomeric mixture. This demonstrates that the final
product of hydrolysis in both cases loses chirality during the
reaction, otherwise we would have observed marked differ-
ences in the chemical shift values for the above two experi-
ments.
X = S/X = O group category (P = 0.029) in predicting Ea
values. Fig. 7 shows the negative correlation between Ea and
sigma for both X = O (Fig. 7a) and X = S compounds
(Fig. 7b; R2 = 0.78). High proportion of the variance was also
explained for X = O (0.88) and X = S (0.76) with respect to
TDS.
The amount of variance explained by Hammett sigma was
expressed as R2 values. Fig. 7 shows that more variance was
explained by Hammett sigma for phosphoramidates than
thiophosphoramidates. The reduced correlation between r
and Ea for the thiophosphoramidates may be due to compet-
ing pathways for these compounds.
Table 6 shows the energy of activation (Ea) and entropy
values (TDS) obtained for various substituted phosphorami-
date and thiophosphoramidate derivatives of 1 (the Gibbs
free energy values varied between 21 and 23 kcal mol–1 for
most of the derivatives, not shown).
From the table values it is evident that the free energy of
activation values varied from 14 to 36 kcal mol–1 depending
on the substituents on the phenyl ring. Also the thiophospho-
Table 6
Experimentally determined values for energy of activation (Ea), and entropy
(TDS) for the disappearance of starting material at 25 °C
To further confirm our results from the NMR studies, we
used cyano-substituted as well as 2,5-dichloro-substituted
phosphoramidate derivatives (11 and 12, respectively) and
found that the final products had similar chemical shifts as
that of 7, once again indicating that they all form the same
hydrolysis products.
Compound
Ea (kcal mol–1
13.9
)
TDS (kcal mol–1
–7.5
)
7
4
23.4
+0.5
9
13.7
–7.9
3
21.7
–1.6
10
2
8
23.5
+0.3
The peaks at 10.4 and 10.5 ppm observed in each of the
cases (see Table 7 for further details) may be attributable to
an intermediate that eventually is hydrolyzed in the medium
to give the final stable products. Considering the HPLC
23.2
–0.2
27.1
+3.8
5
36.1
+13.9

676
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
O
HN
X
N
O
O
R
P O
NH
O
H
MeOOC
Me
H
Variance(Prop)
X=O
0.87
0.88
X=S
0.78
0.76
Ea
TdeltaS
Fig. 7. The effect of sigma for sulfur and oxygen substituted compounds on the energy of activation (Ea). TheANCOVA model significantly predicted Ea values
(P = 0.0063). There was a significant effect of sigma (P = 0.0045) and P = S/P = O group substitution (P = 0.029). The effect of sigma (F-ratio = 23.9) was greater
than the effect of X = S/X = O group (F-ratio = 9.2). Relationship between Ea and sigma values are shown for X = O (a) and X = S (b) substituted compounds.
Data points are labeled according to the ‘R’ substitution.
(Fig. 8) and the NMR spectroscopy results, the products can
be proposed as d4T phosphate 13 at 1.71 ppm, d4T methoxy
alaninyl monophosphate 14 at 6.23 ppm, and 6 at 7.2 ppm.
3.8 min represents d4T alanine mono phosphate methyester.
Peaks at 2.6, 2.4 and 1.7 corresponds to d4T alanine mono-
phosphate, d4T monophosphate and d4T, respectively.As the
time progressed, the starting material depletes resulting in
the formation of products. In addition, formation of
p-bromophenol is evident (13.4 min) after 5 min of reaction.
The above results are consistent with those observed in NMR
experiments.
O
O
HN
HN
O
O
O
N
O
N
HO P
HO P O
O
OH
O
NH
O
H
3.4. 31P-NMR studies on the reaction between
thiophosphoramidate and sodium hydroxide
MeOOC
Me
H
H
13
14
We next sought to determine the structure of the interme-
diates obtained in thiophosphoramidate derivatives using
31P-NMR technique. For this purpose we chose compound 5
and treated with sodium hydroxide. The starting material
gave two distinct peaks at –11.8 and –12.3 ppm relative to
phosphoric acid as internal standard at 0 ppm. After 3 h of
treatment with sodium hydroxide the NMR spectrum showed
only starting material. Leaving the reaction for additional
14 h at room temperature yielded an additional peak at
3.53 ppm although the major peak was that of the starting
Formation of 14 could occur by a nucleophilic substitu-
tion reaction with the base. Formation of 6 could occur by
hydrolysis of the methyl ester group to the carboxylic acid
and subsequent cyclization and simultaneous elimination of
the phenoxy moiety. Further hydrolysis of the cyclic interme-
diate furnishes 6. Comparing the results obtained from
HPLC, Fig. 8 shows the HPLC profile of compound 7 treated
with sodium hydroxide at various time intervals. The peak at
7–8 min corresponds to the starting material and peak at

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
677
Table 7
Based on the products of hydrolysis, we propose the
following Scheme 3 for base catalyzed reaction of phospho-
ramidate derivatives of stavudine.
31P-NMR chemical shifts obtained for P = O compounds during alkaline
hydrolysis
O
O
Scheme 2 illustrates various pathways through which
products can be formed during alkaline hydrolysis of phos-
phoramidate derivatives of stavudine. If we consider step-
wise mechanism, the ester side chain is first hydrolyzed to
carboxylic acid to form intermediate B. This intermediate
forms cyclic phosphoramidate by elimination of phenoxy
moiety intramolecularly resulting in compound C which on
subsequent hydrolysis by water converts to product E. Alter-
natively, Compound D can be formed by direct attack of
nucleophile on the phosphorus center, which subsequently
gets hydrolyzed by excess base to form compound E. A third
possibility is that simultaneous attack on the ester side chain
as well as on the phosphorus center to produce compound E.
Ultimately E gets converted into d4T monophosphate and
d4T.
HN
HN
Cl
O
P
NH
O
P
NH
O
N
O
N
O
O
NC
O
O
O
O
H
H
Cl
MeOOC
Me
MeOOC
Me
H
H
12
11
Compound
Time
0
Chemical shifts (ppm)
4.6, 4.4
7 (Both isomers)
7 (Single isomer)
11 (Both isomers)
3 h
6.0, 7.0, 10.6, 11.0, 11.3
0
4.4
1 h
6.1, 7.3, 10.6, 11.0, 11.3
4.3, 4.2
0
10 min
4 h
4.0, 4.1, 6.1, 7.3, 10.4, 10.5
4.1, 6.1, 7.3, 10.4, 10.5, 11.0 a, 11.3 a
5.9, 7.0, 10.2 a, 10.4 a, 10.9, 11.1
4.5
48 h
0
12 (Both isomers)
3.5. Enzymatic hydrolysis studies using porcine liver
esterase
10 min
1.5 h
36 h
4.5, 6.1, 10.4, 10.6
6.1, 10.4, 10.6, 11.0, 11.3
6.1, 7.3, 11.0, 11.3
In Section 1, we have stated that of the various pro-drug
approaches, amino acid phosphoramidate derivatives have
shown promise as potent antiviral agents, since in some cases
they exhibited enhanced antiviral activity and reduced cyto-
toxicity when compared to the parent nucleosides [11,12,23].
In order to explain the parameters governing the extracellular
and intracellular activation of phosphoramidates, a decom-
position pathway of these nucleosides has been proposed and
is represented by Scheme 3.
a Trace amounts observed relative to the other peaks.
material. The reaction was very slow and hence we were not
able to observe any additional peaks with prolonged time of
treatment unlike HPLC. Due to the slow reaction times we
did not proceed further to identify products of reaction. Our
main goal was to find out the differences between thiophos-
phoramidate and phosphoramidate derivatives under identi-
cal conditions and relate them to biological activity.
O
HN
O
HO P
O
N
Pathway 2
O
NH
O
H
O
N
MeOOC
H
Me
HN
O
O P
O
D
O
X
O
NH
O
HN
H
O
MeOOC
O
H
Me
Pathway 3
O
O
N
-O P
HN
O
A
O
NH
O
N
-O P
O
HO
H
Me
O
O
N
OH
O
H
+
HN
E
O
O P
O
X
O
NH
O
O
N
d4T
HO
H
Me
HN
O
O
B
O
O P
O
NH
O
O
H
Me
C
Pathway 1
Scheme 2. Possible hydrolysis pathways for phosphoramidate derivative of stavudine in presence of alkali.

678
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
7
Br
OH
Br
OH
15
15
MeOH
6
14
Alanine
MeOH
H O
OH
13
1
O
H
H3PO4
O
NH
N
H
O
Scheme 3. Enzymatic reaction products of aryl phosphoramidate derivatives of 1.
This scheme represents the pathways through, which vari-
ous products can be rationalized for these phosphoramidate
derivatives. In the first step, carboxyesterase was found to
react with 7 to furnish the carboxylic acid by the elimination
of methanol, which then cyclizes and intramolecularly elimi-
nates p-bromophenol 15 forming an unstable cyclic mixed
anhydride. The cyclic anhydride hydrolyses back to the car-
boxylic acid producing 6. Alternatively, hydrolysis of the
p-bromophenol could occur through direct nucleophilic at-
tack by hydroxide ion to furnish 14. This could undergo
further hydrolysis to furnish 6. In in vivo experiments we
have discovered that p-bromophenol converts into
p-bromophenylsulfate by the action of a sulfatransferase
[22]. Further steps involving phosphodiesterase produces the
nucleoside monophosphate 13, which in the presence of a
phosphatase finally converts into 1.
Section 5. Fig. 10 shows the 31P-NMR spectra for compound
7 in presence of esterase at different time intervals.
The chemical shift of the starting material (a diastereo-
meric mixture) was found to be 4.58 and 4.09 ppm, respec-
tively (Fig. 10A).Addition of esterase to the solution resulted
in formation of a distinct peak at 6.76 ppm relative to the
standard within 2 h (Fig. 10C). After 6 h there was no change
in the NMR spectrum, (Fig. 10D) indicating that only one
product was formed in the reaction mixture. The product was
identified independently by running the authentic sample
under similar conditions. It is interesting to note that forma-
tion of one product demonstrates that the phenolic moiety in
the compound is eliminated to furnish 6. This is in accor-
dance with the results obtained from HPLC experiments,
which further substantiates the proposed mechanism of ac-
tion of enzyme esterase towards these phosphoramidate de-
rivatives of 1.
We also examined the esterase-mediated hydrolysis of 7 to
prove the formation of the carboxylic acid product 6. For this
purpose we utilized HPLC technique. Fig. 9 shows the HPLC
profiles of reaction products of 7 after treatment with porcine
liver esterase.
3.6. Antiviral activity of aryl phosphoramidate and aryl
thiophosphoramidates of 1
By varying the concentration of the substrate and keeping
the enzyme concentration constant we were able to get evi-
dence that indeed the esterase hydrolyses the methyl ester.
This is observed by the decrease in the concentration of the
starting material. The products were identified using authen-
tic samples synthesized independently for this purpose. In
addition, for the first time, we observed that one of the
isomers was hydrolyzed faster than the other as evidenced
from the values measured for the area under the curve for
individual peaks. The implication of such a differentiation
caused by the enatioselective reactivity of this enzyme to-
wards the distereoisomers will be the subject of future inves-
tigations. Further proof of formation of 6 came from the
NMR spectroscopy studies.
In our earlier discussion, we demonstrated that thiophos-
phoramidate derivatives had a higher energy of activation for
hydrolysis. As such we were interested, if this was related to
the biological activity using the antiviral activity profiles for
both the derivatives. Fig. 11 shows a histogram of the IC50
values for four sets of compounds from each series.
All the aryl substituted phosphoramidates had IC50 values
in the 1–3 nM range, whereas the corresponding thiophos-
phoramidate analogs had IC50 values in the 1–8 µM range.
Notably, 7 and 9 were more potent than 8 and 10 implying
that electron withdrawing substituents have beneficial effect.
Based on the results obtained from their rate of hydrolysis
and energy of activation, we propose that the decrease in
potency of the sulfur analogs can be attributed to their higher
energy of activation for hydrolysis during the formation of
the active metabolites.
NMR identification of the product during reaction be-
tween esterase and 7 was achieved using 31P-NMR chemical
shifts. The experimental details for this study are given in

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
679
Fig. 8. HPLC profile of compound 7 treated with sodium hydroxide at various time intervals.
4. Conclusions
that phosphoramidate derivatives show a better correlation as
compared to thiophosphoramidates. In addition, we have
also demonstrated using our experimental results that en-
thalpy values vary between phosphoramidate and thiophos-
phoramidate derivatives. Comparison of anti-HIV activity of
aryl phosphoramidate with aryl thiophosphoramidate deriva-
tives of 1 demonstrated that the former derivatives showed
1000-fold more potent antiviral activity. Although it is pos-
sible that the reason why the thiophosphoramidates are less
potent than the phosphoramidates in the cellular studies is
because their conversion to the triphosphate is slow. Alterna-
tively, thiophosphoramidates may function as poor substrates
for the kinase enzymes. However, our results suggest that the
energy of activation of hydrolysis of these phosphoramidate
derivatives play a significant role in their biological potency.
A comparative study between aryl phosphoramidate as
well as aryl thiophosphoramidate derivatives of 2′,3′-
didehydro-2′,3′-dideoxythymidine (d4T, 1) demonstrated
that the replacement of oxygen with sulfur in their structural
framework decreases the rate of alkaline hydrolysis by two-
fold. In addition, the activation energy (Ea) for the thiophos-
phoramidate analog was comparatively higher than that of
the phosphoramidte analogs. We have also found for the first
time that an intermediate is formed in the hydrolysis of the
thiophosphoramidate derivatives. Using both HPLC and 31P-
NMR techniques, the products of the hydrolysis reaction
were identified. We have also correlated Hammett sigma
values to the energy of activation and came to the conclusion

680
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
Fig. 9. HPLC profiles of reaction between varying concentrations of 7 and esterase. A. No enzyme; B. 50 µl enzyme + 75 µl of 7; C. 50 µl enzyme + 60 µl of 7;
D. 50 µl enzyme + 50 µl 7; E. 50 µl of enzyme + 40 µl of 7.
5. Experimental protocols
adjusted to 265 nm and the reference wavelength was kept at
400 nm.Aliquots of the sample were drawn from the reaction
vial at various time intervals and analyzed. For the fast
reactions two HPLC instruments were used simultaneously
to determine the rates. Reactions at various temperatures
were conducted by keeping the reaction vessel in a constant
temperature bath. In the case of low temperature studies, the
reaction vial was left in the refrigerator and the temperature
of the refrigerator was monitored using a thermometer. Reac-
tions conducted at 0 °C were carried out using external ice
bath and those with –10 °C were performed by keeping the
reaction vial in a constant temperature freezer.
5.1. A general procedure for the kinetic study for alkali
hydrolysis of phosphoramidate and thiophosphoramidate
derivatives of d4T as monitored by HPLC
A known amount of the phosphoramidate or thiophospho-
ramidate derivative was carefully weighed (3–4 mg) using an
analytical balance and transferred into a scintillation glass
vial, 3 ml of methanol were added and the contents were
vortexed for 2 min until a homogenous solution was ob-
tained.A 0.05 N NaOH solution was prepared in a volumetric
flask using analytical grade NaOH pellets and deionized
water. A 1 ml aliquot from the stock solution of the com-
pound was pipetted out into another glass vial. To this vial
0.350 µl of 0.05 N NaOH was added from the volumetric
flask using a pipette. The contents were mixed thoroughly by
shaking the vial. From this mixture 50 µl was used for HPLC
analysis. The column used was a Lichrospher RP-18 analyti-
cal column of 4 × 250 mm. The eluent used for HPLC was
water (0.1% TEA + 0.1% TFA)/acetonitrile (65:35). The
column was maintained at room temperature. The flow rate
was maintained at 1 ml min–1, the detection wavelength was
5.2. A general procedure for the estimation of products
from the kinetic study for alkali hydrolysis
of phosphoramidate and thiophosphoramidate derivatives
of d4T
The amount of products observed during the reaction of
the phosphoramidates and thiophosphoramidates were esti-
mated from the area obtained from the HPLC profiles. In
addition, when possible authentic samples of the products
were run to identify the peaks observed during the reaction.

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
681
Fig. 10. 31P-NMR profiles of 7 with esterase enzyme at varying intervals of time (peak at 0 ppm represents the phosphoric acid standard in a fused capillary).
A: No enzyme; B: after 2 min of addition of enzyme; C: after 2 h; D: after 6 h.
Further confirmation of the product structures were obtained
using LC/mass instrument.
5.4. General method for the enzymatic reactions
of phosphoramidate and thiophosphoramidate derivatives
of d4T as monitored by HPLC
5.3. General method for the enzymatic reactions
of phosphoramidate and thiophosphoramidate derivatives
of d4T as monitored by 31P-NMR spectroscopy
The sample was dissolved in a known quantity of aqueous
methanol and was injected into the column to obtain the
starting material HPLC peaks. For the enzymatic reactions,
the concentration of the enzyme was kept constant at 50 µl
and the substrate concentration was varied keeping the vol-
ume constant throughout the HPLC analysis. All the enzy-
matic reactions were conducted at room temperature. A Li-
chrospher analytical column was used and the conditions
were set at 1 ml min–1 isocratic flow. The eluent used was
water (0.1% TEA + 0.1% TFA)/acetonitrile (65:35). The
column was maintained at room temperature. The wave-
length of absorbance was set at 265 nm with a reference value
of 400 nm. The injection volume was maintained at 50 µl per
sample. The experiments were repeated three times.
31P-NMR studies with enzyme were conducted using a
5 mm NMR tube. The sample was dissolved in deuterated
methanol (100 µl), followed by D2O (0.4 ml) and then trans-
ferred into the NMR tube using a Pasteur pipette. The capil-
lary tube containing a 1% phosphoric acid standard was
introduced into the NMR tube. The NMR spectrum of the
substrate was taken under these conditions. Subsequently
100 µl of the esterase solution was introduced into the NMR
tube, the contents were carefully shaken for a minute. The
spectrum was then taken and periodically monitored to ascer-
tain the disappearance of the starting material.

682
T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
100000
10000
1000
100
molecule. In this analysis the effect of the categorical vari-
X = O
X = S
ables (X = S/X = O groups) were compared controlling for
Hammett sigma values as the covariate. The amount of vari-
ance explained by Hammett sigma and the groups were
deemed significant if the P-values were less than 0.05. Re-
gression analysis was performed to investigate the effect of
Hammett sigma on the energy of activation (Ea) and entropy
(TDS) for substrate disappearance. The amount of variance
explained by Hammett sigma was expressed as R2 values.
5.6. In vitro assays of anti-HIV-1 activity
10
Normal human peripheral blood mononuclear cells (PB-
MNC) from HIV-negative donors were cultured for 72 h in
RPMI 1640 supplemented with 20% (v/v) heat-inactivated
fetal bovine serum (FBS), 3% interleukin-2, 2 mM
L-glutamine, 25 mM HEPES, 2 g l–1 NaHCO3, 50 µg ml–1
gentamicin, and 4 µg ml–1 phytohemagglutinin prior to expo-
sure to HIV-1 at a multiplicity of infection (MOI) of 0.1 dur-
ing a 1 h adsorption period at 37 °C in a humidified 5% CO2
atmosphere. Subsequently, cells were cultured in 96-well
microtiter plates (100 µl per well; 2 × 106 cells per ml) in the
presence of various concentrations of the compounds and
aliquots of culture supernatants were removed from the wells
on the seventh day after infection for p24 antigen assays, as
previously described [31]. The applied p24 enzyme immu-
noassay (EIA) was the unmodified kinetic assay commer-
cially available from Coulter Corporation/Immunotech, Inc.
(Westbrooke, ME), which utilizes a murine mAb to HIV core
protein coated onto microwell strips to which the antigen
present in the test culture supernatant samples binds. Percent
viral inhibition was calculated by comparing the p24 values
from untreated infected cells (i.e. virus controls).
1
4-Br
4-OMe
H
4-F
O
HN
X
O
N
R
O P O
NH
O
H
MeOOC
Me
H
Substituent
R
X = S
X = O
Sulfur/Oxy
Mean (nM) SEM (N=3)
Mean (nM) SEM (N=3)
Ratio
4-Br
4-OMe
H
3635.3 3527.3
8177.7 7646.2
2613.7 2233.8
997.7 726.4
1.0 0.0
2.7 1.7
1.7 0.7
1.0 0.0
3635
3067
1568
998
4-F
Fig. 11. Anti-HIV activity of aryl phosphoramidate and aryl thiophosphora-
midate derivatives of 1. IC50 values (IIIB virus) were compared for similarly
substituted groups (4-Br, 4-OMe, H, 4-F) on the phenyl ring on thiophos-
phoramidates (horizontal line pattern) and phosphoramidates (diamond pat-
tern) compounds. The table shows the mean and standard error calculated for
each group of compounds (N replicates) and the relative potency. All
thiophosphoramidate compounds showed much reduced activity with IC50
values >900-fold as compared to phosphoramidate derivatives.
References
5.5. Calculation of the rate of reaction
[1] R.H. Goldschmidt, B.J. Dong, J. Am. Board Fam. Pract 10 (1997)
144–167.
[2] S.G. Deeks, P.A. Volberding, AIDS Clin. Rev (1997–98) 145–185.
[3] P.M. Grob, J.C. Wu, K.A. Cohen, R.H. Ingraham, C.K. Shih,
K.D. Hargrave, T.L. Mctague, V.J. Merluzzi, AIDS Res. Hum. Retro-
virus 8 (1992) 145–152.
[4] F.W. Bell, A.S. Cantrell, M. Hogberg, S.R. Jaskunas, N.G. Johansson,
C.L. Jordan, M.D. Kinnick, P. Lind, J.M. Morin Jr, R. Noreen,
B. Oberg, J.A. Palkowitz, C.A. Parrish, P. Pranc, C. Sahlberg,
R.T. Ternansky, R.T. Vasileff, L. Vrang, S.J. West, H. Zhang,
X.X. Zhou, J. Med. Chem 38 (1995) 4929–4936.
[5] A.S. Cantrell, P. Engelhardt, M. Hogberg, S.R. Jaskunas,
N.G. Johansson, C.L. Jordan, J. Kangasmetsa, M.D. Kinnick, P. Lind,
J.M. Morin Jr, M.A. Muesing, R. Noreen, B. Oberg, P. Pranc, C. Sahl-
berg, R.J. Ternansky, R.T. Vasileff, L. Vrang, S.J. West, H. Zhang, J.
Med. Chem 39 (1996) 4261–4274.
[6] R. Vig, T.K. Venkatachalam, F.M. Uckun,Antiviral Chem. Chemother
9 (1998) 445–448.
[7] C. Mao, R. Vig, T.K. Venkatachalam, E.A. Sudbeck, F.M. Uckun,
Bioorg. Med. Chem. Lett 8 (1998) 2213–2218.
[8] P.A. Furman, J.A. Fyfe, M.H. St. Clair, K. Weinhold, J.L. Rideout,
G.A. Freeman, L.S. Nusinoff, D.P. Bolognesi, S. Broder, H. Mitsuya,
D.W. Barry, Proc. Natl. Acad. Sci. USA 83 (1986) 8333–8337.
The rate of reaction was computed by using first order rate
constants and an average of eight to nine time points were
used for this estimate. Energy of activation values for indi-
vidual products was obtained using the standard Arrhenius
equation. The rate constants were obtained for each of the
products at various temperatures and plotted and expressed
in kcal mol–1. The rate constants refer to rate per h since some
of the reactions were too slow even at high temperatures.
We used theANCOVA to study the influence of a categori-
cal predictor variable (chemical substitution group) and a
continuous predictor variable (Hammett sigma values) on a
response variable (Ea values/TDS). In this analysis, the vari-
ance explained by the categorical and the continuous vari-
ables can be compared. Also, a significant effect of the
categorical variable can be assigned taking into account the
relationship between the continuous variable and the re-
sponse variable. ANCOVA model was used to test for signifi-
cant effects of Hammett sigma or X = S/X = O groups on the

T.K. Venkatachalam et al. / European Journal of Medicinal Chemistry 39 (2004) 665–683
683
[9] Z. Hao, D.A. Cooney, N.R. Hartman, C.F. Perno, A. Fridland, A.L. De
Vico, M.G. Sarngadharan, S. Border, D.G. Johns, Mol. Pharmacol 34
(1988) 431–435.
[10] J. Balzarini, D.A. Cooney, M. Dalal, G.J. Kang, J.E. Cupp, E. De
Clercq, S. Broder, D.G. Johns, Mol. Pharmacol 32 (1987) 798–806.
[11] C. McGuigan, D. Cahard, H.M. Sheeka, E. De Clercq, J. Balzarini, J.
Med. Chem 39 (1996) 1748–1753.
[12] C. McGuigan, D. Cahard, H.M. Sheeka, E. De Clercq, J. Balzarini,
Bioorg. Med. Chem. Lett 6 (1996) 1183–1186.
[13] M.G. Starnes, Y.C. Cheng, J. Biol. Chem 262 (1987) 988–991.
[14] Z. Hao, D.A. Cooney, D. Farquhar, C.F. Perno, K. Zhang, R. Masood,
Y. Wilson, N.R. Hartman, J. Balzarini, D.G. Johns, Mol. Pharmacol 37
(1990) 157–163.
[15] M.A. Johnson, G.Ahluwalia, M.C. Connelly, D.A. Cooney, S. Broder,
D.G. Johns, A. Fridland, J. Biol. Chem 263 (1988) 15354–15357.
[16] M.A. Johnson, A. Fridland, Mol. Pharmacol 36 (1989) 291–295.
[17] C. McGuigan, R.N. Pathirana, J. Balzarini, E. De Clercq, J. Med.
Chem 36 (1993) 1048–1052.
[18] C. McGuigan, R.N. Pathirana, N. Mahmood, K.G. Devine, A. Hay, J.
Antiviral Res 17 (1992) 311–321.
[19] R. Vig, C. Mao, T.K. Venkatachalam, L. Tuel-Ahlgren, E.A. Sudbeck,
F.M. Uckun, Bioorg. Med. Chem 6 (1998) 1789–1797.
[21] F.M. Uckun, P. Samuel, S. Qazi, C.L. Chen, S. Pendergrass, T.K. Ven-
katachalam, Antiviral Chem. Chemother 13 (2002) 197–203.
[22] C.L. Chen, G. Yu, T.K. Venkatachalam, F.M. Uckun, Drug Metab.
Dispos 30 (2002) 1523–1531.
[23] R. Lohrmann, L.E. Orgel, J. Mol. Evol 7 (1976) 253–267.
[24] M. Oivanen, S. Kuusela, H. Lonnberg, Chem. Rev 98 (1998) 961–
990.
[25] S. Mikkola, M. Kosonen, H. Lonnberg, Curr. Org. Chem 6 (2002)
523–538.
[26] P. Jarvinen, M. Oivanen, H. Lonnberg, J. Org. Chem 56 (1991)
5396–5401.
[27] M. Ora, K. Mattila, T. Lonnberg, M. Oivanen, J. Am. Chem. Soc 124
(2002) 14364–14372.
[28] N. Mesplet, Y. Saito, P. Morin, L.A. Agrofoglio, Bioorg. Chem 31
(2003) 237–247.
[29] J. March (Ed.),Advanced Organic Chemistry, Wiley, NewYork, 1992.
[30] J.D. Annette, Texts in Statistical Science: An Introduction to General-
ized Linear Models, second ed, Chapman and Hall/CRC. A CRC
Press Company, Boca Raton, London, New York, Washington, DC,
2002.
[31] F.M. Uckun, L.M. Chelstrom, L. Tuel-Ahlgren, I. Dibirdik, J.D. Irvin,
M. Chandan-Langlie, D.E. Myers, Antimicrob. Agents Chemother 42
(1998) 383–388.
[20] T.K. Venkatachalam, T.L. Tai, R. Vig, C.L. Chen, S.T. Jan,
F.M. Uckun, Bioorg. Med. Chem. Lett 8 (1998) 3121–3126.