Enhancing the aqueous solubility of d4T-based phosphoramidate prodrugs

Article abstract of DOI:10.1016/S0960-894X(99)00701-5

A range of polyether para-substituted phosphoramidates were synthesised and found to have substantially elevated aqueous solubilities compared to the underivatised parent prodrug. A 30-fold increase in aqueous solubility could be achieved without a substantial decrease of in vitro activity against HIV-1. Replacement of the aryl (i.e. phenolic) moiety by tyrosine led to a substantial enhancement in aqueous solubility but also to a decrease in antiviral potency. A previously unobserved trend was identified, relating increased aryl substituent steric bulk to decreased antiviral activity. (C) 2000 Elsevier Science Ltd. All rights reserved.

Full text of DOI:10.1016/S0960-894X(99)00701-5

Bioorganic & Medicinal Chemistry Letters 10 (2000) 381±384
Enhancing the Aqueous Solubility of d4T-based
Phosphoramidate Prodrugs
Adam Siddiqui, a Christopher McGuigan, a,* Carlo Ballatore, a Sheila Srinivasan, a
Erik De Clercq b and Jan Balzarini b
aWelsh School of Pharmacy, Cardi€ University, King Edward VII Avenue, Cardi€ CF10 3XF, UK
bRega Institute for Medical Research, Katholieke Universiteit Leuven, B-3000 Leuven, Belgium
Received 6 September 1999; accepted 17 December 1999
AbstractÐA range of polyether para-substituted phosphoramidates were synthesised and found to have substantially elevated
aqueous solubilities compared to the underivatised parent prodrug. A 30-fold increase in aqueous solubility could be achieved
without a substantial decrease of in vitro activity against HIV-1. Replacement of the aryl (i.e. phenolic) moiety by tyrosine led to a
substantial enhancement in aqueous solubility but also to a decrease in antiviral potency. A previously unobserved trend was identi-
®ed, relating increased aryl substituent steric bulk to decreased antiviral activity. # 2000 Elsevier Science Ltd. All rights reserved.
The nucleotide prodrug (``protide'') concept continues
to be a subject of intensive research.1 Principal objec-
tives are to improve the activity of existing antiviral
nucleoside analogues and to confer activity upon those
nucleoside analogues that express poor activity due to
inecient intracellular phosphorylation.
used to optimise the pharmacological pro®le of these
types of prodrugs, without compromising antiviral
potency.
An important parameter that might be modi®ed by the
choice of aryl substituents is prodrug aqueous solubility.
Consequently, we sought to synthesize a series of poly-
ether para-substituted phosphoramidates 2a±2d (Fig. 2).
These substituents should have the dual e€ect of
enhancing prodrug aqueous solubility without compro-
mising optimum or near-optimum lipophilicity (and thus
antiviral potency). Additionally, they should allow the
investigation of the e€ects of substituent steric bulk
beyond the range of those substituents previously studied.8
We have shown that our approach using aryloxy phos-
phoramidates can lead to elevated levels of anti-HIV
potency for a range of antiviral nucleoside analogues
and does successfully by-pass dependency on nucleoside
kinases.2±4 Recently, we demonstrated that simple
mono-substitution of the aryl ring of the phosphor-
amidates based on d4T, ddA and d4A signi®cantly
enhances antiviral activity relative to the parent phos-
phoramidates.5±7 In addition to those substituents that
markedly enhanced antiviral potency, we were able to
show that for d4T-based phosphoramidates a variety of
aryl substituents could be tolerated. Antiviral data for
21 various aryl-substituted d4T-based phosphor-
amidates (1a±u, Fig. 1) were subjected to a Hansch-type
QSAR analysis, which revealed activity to be principally
dependent on lipophilicity as measured by compound
logP (eq 1).7 Compared to other parts of the phosphor-
amidate structure, chemical modi®cation of the aryl
group appears to result in no loss of activity, thus we
proposed that appropriate aryl substituents might be
log 1=EC50‰HIV 1Š ˆ 2:440 ‡ 5:157ꢀ1:117†logP
2
1:499ꢀ0:395†ꢀlogP†
ꢀ1†
Intracellular activation of the parent prodrug (1a) leads
to the release of three major by-products; l-alanine (a
naturally occurring amino acid), methanol and phenol.
Studies in vitro suggest that replacement of the methyl
ester by an ethyl esterÐleading to the release of a nota-
bly less toxic alcoholÐis not detrimental to antiviral
activity.9 The release of phenol may represent a greater
challenge.
Although a full toxicological appraisal of the e€ects of
phenols on humans is incomplete, (particularly with
*Corresponding author. Tel.: +44-1222-874537; fax: +44-1222-874537;
e-mail: mcguigan@cardi€.ac.uk
0960-894X/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(99)00701-5

382
A. Siddiqui et al. / Bioorg. Med. Chem. Lett. 10 (2000) 381±384
Figure 1.
Figure 2.
regards to mutagenicity and carcinogenicity, for exam-
ple) many of the e€ects of phenol toxicity are well
characterized.10 In addition to non-speci®c modes of
toxicity, phenols (particularly those which are both
lipophilic and acidic) are known to be potent uncou-
plers of oxidative phosphorylation.11,12 Therefore, the
replacement of the phosphoramidate phenol moiety
with compounds that might be less toxic would seem
worthy of investigation. Thus, we replaced the phenol
moiety of 1a with the phenolic moiety of tyrosine (e.g. 3,
Fig. 2). The free amino group of tyrosine should also
lead to an enhanced prodrug aqueous solubility but,
conversely, a diminished lipophilicity.
The molar extinction coecients and aqueous solubility
of compounds 1a, 1i, 2a-d, 3 and 6 were determined by
standard methods (Table 1). Introduction and pro-
gressive elongation of the polyether side-chain on the
aryl moiety led to a clear, stepwise increase in the aqu-
eous solubility of the prodrugs. From a saturation con-
centration of 6.2 mmol/l for the parent prodrug 1a the
solubility could be increased to a maximum of 375.4
mmol/l for 2c (a ca. 60-fold increase). By contrast, the
lipophilicity of compounds 2a±c remained constant as
measured by logP (Table 1), an important result for the
retention of antiviral activity. Compound 2d showed
comparable aqueous solubility to compound 1i (having a
simple methoxy substituent) but, as expected, was less
soluble than compound 2a, which bears a smaller terminal
alkyl group on the aromatic side-chain.
The phenolic precursors for the synthesis of compounds
2a±d were prepared by re¯uxing hydroxyquinone with
the corresponding polyether alcohol in the presence of
phosphomolybdic acid. Synthesis of the corresponding
phosphoramidates was accomplished via conventional
phosphorochloridate chemistry (described elsewhere13).
Synthesis of 3 was achieved starting from commercially
available l-tyrosine methyl ester hydrochloride salt (4).
Protection of the amino function with BOC gave com-
pound (5), which was subsequently converted to the
phosphoramidate (6) via conventional phosphoro-
chloridate chemistry. Final deprotection of the amino
group, under acidic conditions, liberated the desired
compound as its tri¯uoroacetyl salt. Compounds 1a, 1i,
2a±d, 3 were isolated as diastereomeric mixtures result-
ing from mixed stereochemistry about the phosphorus
centre, with the diastereomeric ratio, determined by
both 31P NMR and HPLC, typically of the order 1:1.
Replacement of the phenolic moiety of 1a by tyrosine
methyl ester, led to at least a 100-fold increase in aqu-
eous solubility. However, the low value of the measured
logP (Table 1) suggested that the in vitro antiviral
potency might be less than observed for the parent pro-
drug, a premise that was shown to be correct (Table 2).
The anti-HIV data (HIV-1 and HIV-2) for compounds
2a±d, 3 and 6 are given in Table 2 (data for 1a and 1i are
also given for reference). Predicted values for activity
against HIV-1 using eq (1) are also displayed along with
substituent MR valuesÐused as an approximation of
substituent steric bulk (noting the comments of Hansch
et al.15). Compounds 2a±d and 6, all displayed activity
against HIV-1 and HIV-2 that was comparable to that
of the nucleoside analogue d4T. Additionally (and
unlike d4T), potency was retained in a thymidine kinase
de®cient (TK ) cell-line, indicating sucessful release of
the nucleoside monophosphate analogue. Compared to
1
Diastereomeric splittings were observed in the H, 13C
and 31P NMR spectra and additional splitting patterns
from phosphorus coupling, where appropriate, were
1
seen in the 13C and H NMR spectra.14

A. Siddiqui et al. / Bioorg. Med. Chem. Lett. 10 (2000) 381±384
383
Figure 3. A: Phosphomolybdic acid (cat.), re¯ux, 6 h; B: POCl3, Et3N, 78 ꢀC, 16 h; C: alanine methyl ester hydrochloride, Et3N, 78 ꢀC, 16 h; D:
d4T, NMI, RT, 16 h; E: BOC-anhydride, Et3N, RT, 1 h; F: TFA:DCM (1:1), RT, 1 h.
Table 1.
on the basis of logP alone. Taken together, these results
suggest a clear and negative correlation between sub-
stituent steric bulk and antiviral activityÐwhich is
apparently most evident when the substituent MR value
is near 20 or above. Compound 3, the most water solu-
ble of the series, is the least active, a result predicted by
the QSAR equation. This is consistent with the theory
that lipophilicity is a necessary property for adequate
di€usion of the prodrug across cellular membranes. The
numerical discrepancy between the observed and pre-
dicted values may re¯ect the far lower lipophilicity of 3
compared to those compounds used to construct the
original QSAR.
Compound
Molar
extinction
coecient
Saturation
conc.
(mmol/L)
logP
(1-octanol/
water)
Esterase
labilitya
(%)
1
(E) (l mol
1
cm
)
1a
1i
2a
2b
2c
2d
3
7640
8503
8181
8277
8846
8864
8182
9194
6.2
66
184
259
375
86
1.04
1.09
0.86
0.86
0.86
1.10
0.09
1.07
5.4
79
10
10
10
10
0
>794
6.3
6
5.4
a% of compound hydrolysed after incubation with pig liver esterase
for 24 h (pH 7.6, 37 ꢀC).5
In an attempt to explain why a substituent of large steric
bulk might display a decreased antiviral activity we
subjected all of the compounds from Table 1 to a pig
liver esterase (PLE) assay, which we have used pre-
viously to model the putative esterase-driven hydrolysis
of the amino acid ester, believed to be the ®rst step in
intracellular prodrug activation.5 The results are pre-
sented in Table 1 as the percentage of compound
hydrolysed after 24 h. It can be seen from these values
that no di€erences exist between any of the compounds
in the polyether series, 2a±d. Additionally, whilst com-
pounds 2a±d, 1a, 1i and 6 all serve as esterase substrates,
no observable hydrolysis of compound 3 occurred over
the experimental time-scale. Although we have pre-
viously seen that quantitative comparisons between
antiviral activity and the results of the PLE assay can be
unsuccessful, we have noted that processesing by PLE is
a necessary requirement for a prodrug to express in
the parent phosphoramidate (1a), only compound 2a
expressed a similar potency against HIV-1. Against
HIV-2, 2a was markedly less potent than 1a, expressing
a potency comparable to that of d4T. By a comparison
of the actual EC50 values against HIV-1 and those pre-
dicted using the Hansch-type equation (eq 1), it is
apparent that an increased substituent steric bulk leads
to a smaller antiviral activity than might be predicted
from using logP alone. Indeed, an analysis of the poly-
ether series (2a±d) reveals a step-wise deviation between
the observed and predicted activity values for each
additional ethoxy unit present in the structure. Simi-
larly, the activity displayed by the N-BOC protected
phosphoramidate 6 is also less than might be expected

384
A. Siddiqui et al. / Bioorg. Med. Chem. Lett. 10 (2000) 381±384
Table 2.
EC50 (mM)a
CEM (HIV-1)
EC50 (mM)
EC50 (mM)
CEM (HIV-2)
EC50 (mM)
CEM/TK (HIV-2)
CC50
MRep-X
b
Compound
Predictedc (HIV-1)
1a
1i
0.075
0.057
0.09
0.55
0.75
0.2
0.049
0.040
0.13
0.13
0.13
0.075
0.053
0.6
1
1.3
0.25
10
0.7
0.075
0.047
0.34
1.2
1.35
0.4
5
0.8
33
>100
35
47
114
64
1.03
7.87
2a
2b
2c
2d
3
19.28f
30.69f
42.10f
23.93f
26.56f
52.37f
Ð
0.04
100
d
4
97.28
0.043
Ð
6
d4T
0.65
0.651
217
174
0.770
aEC50 is the concentration required to protect CEM cells against the cytopathicity of HIV by 50%. Data are the mean of two to four independent
experiments.
bCC50 is the concentration required to inhibit CEM cell proliferation by 50%.
cUsing eq (1).
dNot determined.
eFrom Hansch and Leo.16
fCalculated from fragment values.16
7. Siddiqui, A. Q.; McGuigan, C.; Ballatore, C.; Wedgwood, O.;
vitro antiviral activity.5,17 It is unclear as to whether the
inactivity of 3 under the PLE assay conditions indicates
De Clercq, E.; Balzarini, J. Bioorg. Med. Chem. Lett. 1999, 9, 25.
8. Values in parenthesis represent standard errors of the coef-
®cients.
9. Saboulard, D.; Naesens, L.; Cahard, D.; Salgado, A.;
Pathirana, R.; Velazquez, S.; McGuigan, C.; De Clercq, E.;
Balzarini, J. Mol. Pharmacol. 1999, 56, 693.
that poor processing by cellular esterase(s) contributes
to the low observed antiviral activity, but it is clear that
3 is the least active compound in this study.
10. Bruce, R. M.; Santodonato, J.; Neal, M. W. Toxicology
and Industrial Health 1987, 3, 535.
11. Fujita, T.; Miyoshi, H. Biochim. Biophys. Acta 1988, 935, 312.
12. Fujita, T.; Nishioka, T.; Miyoshi, H. Biochim. Biophys.
Acta 1987, 891, 194.
In conclusion, we have demonstrated that the aqueous
solubility of phosphoramidate prodrugs based on d4T
can be enhanced by substitution of polyether side-
chains on the aryl ring. We have shown that replace-
ment of the phenol moiety with that of tyrosine can
substantially enhance aqueous solubility but at the cost
of antiviral activity, con®rming the importance of con-
trolling prodrug lipophilicity. Additionally, we report a
newly observed trend correlating substituent steric bulk
with decreased antiviral activity.
13. McGuigan, C.; Pathirana, R. N.; Balzarini, J.; De Clercq,
E. J. Med. Chem. 1993, 36, 1048.
14. Selected data for 2a, 20,30-Didehydro-20,30-dideoxy-
thymidine-50-(4-methoxyethoxyphenyl
methoxyalaninyl)
phosphate: dP[CDCl3] 4.17, 4.71; dH[CDCl3] 1.38 (3H, m, Ala-
Me), 1.90 (3H, d, 5-Me), 3.50 (3H, d, OMe), 3.76 (6H, m,
CO2Me, CH2-200, NH), 4.03 (1H, m, Ala-CH), 4.15 (1H, m,
CH2-100), 4.38 (2H, m, H-50), 5.10 (1H, m, H-40), 5.96 (1H, m,
H-30), 6.39 (1H, m, H-20), 6.88 (2H, m, meta-Ph), 7.05 (1H, m,
H-10), 7.12 (2H, m, ortho-Ph), 7.32 (1H, m, H-6), 8.81 (1H, d,
NH); dC[CDCl3] 15.08, 15.14 (5-Me), 23.79 (t, J 3.5, Ala-Me),
52.81, 52.94 (Ala-CH), 55.39 (COOMe), 62.00 (OMe), 69.22,
69.85 (d, J 5.2, C-50), 70.42, 70.42, 70.46 (CH2-200), 73.77, 73.79
(CH2-100), 87.33, 87.38, 87.44, 87.49 (C-40), 92.34, 92.60 (C-10),
111.69, 111.82 (C-5), 118.08, 118.22 (meta-Ph), 123.66, 123.73,
123.87, 123.93 (ortho-Ph), 130.05, 130.22 (C-20), 135.84, 136.14
(C-30), 138.36, 138.64 (C-6), 146.52, 146.61, 146.75, 146.83
(para-Ph), 153.46, 153.54 (C-2), 158.77 (Ph), 166.34, 166.43 (C-
4), 176.58, 176.68, 176.75, 176.85 (Ala-CO); MS m/e FAB
540.1747 (MH+, C23H31N3O10P requires 540.1747); HPLC tR
30.24, 30.73 min (grad. I); tR 31.32, 31.91 min (grad. II).
HPLC (Shimadzu) was conducted on an SSODS2 reverse
phase column using a water/acetonitrile (Fisher: HPLC grade)
eluent; gradient I (standard gradient): 0±80% CH3CN (0±60
min), 80±0% CH3CN (60±65 min), ¯ow rate: 1 ml/min, UV
detection at 265 nm; gradient II: 0±10% CH3CN (0±5 min),
10±70% CH3CN (5±55 min), 70±0% CH3CN (55±60 min),
¯ow rate: 1 ml/min, UV detection at 265 nm.
Acknowledgements
We thank the AIDS Directed Program of the MRC, the
Biomedical Research Programme of the European Com-
mission and the Belgian Geconcerteerde Onderzoeksacties
(Vlaamse Gemeenschap Project 95/5) for ®nancial sup-
port. We also thank Mrs A. Absillis for technical assis-
tance and Ms Helen Murphy for secretarial support.
References and Notes
1. Meier, C. Synlett. 1998, 3, 233.
2. Balzarini, J.; Kruining, J.; Wedgwood, O.; Pannecouque,
C.; Aquaro, S.; Perno, C. F.; Naesens, L.; Witvrouw, M.;
Heijtink, R.; De Clercq, E.; McGuigan, C. FEBS Lett. 1997,
410, 324.
3. McGuigan, C.; Cahard, D.; Sheeka, H. M.; De Clercq, E.;
Balzarini, J. Bioorg. Med. Chem. Lett. 1996, 6, 1183.
4. McGuigan, C.; Cahard, D.; Sheeka, H. M.; De Clercq, E.;
Balzarini, J. J. Med. Chem. 1996, 39, 1748.
15. Hansch, C.; Leo, A.; Unger, S. H.; Kim, K. H.; Nikaitani,
D.; Lien, E. J. J. Med. Chem. 1973, 16, 1207.
16. Hansch, C.; Leo, A. Substitution Constants for Correlation
Analysis in Chemistry and Biology. Wiley Interscience: New
York, 1979.
17. McGuigan, C.; Cahard, D.; Ballatore, C.; Siddiqui, A.; De
Clercq, E.; Balzarini, J. Biorg. Med. Chem. Lett. 1998, 8, 2949.
5. Siddiqui, A. Q.; McGuigan, C.; Ballatore, C.; De Clercq, E.;
Balzarini, J. J. Med. Chem. 1999, 42, 393.
6. Siddiqui, A. Q.; McGuigan, C.; Ballatore, C.; Zuccotto, F.;
Gilbert, I. H.; De Clercq, E.; Balzarini, J. J. Med. Chem. 1999,
42, 4122.