Synthesis and in vitro biological evaluation of valine-containing prodrugs derived from clinically used HIV-protease inhibitors

Article abstract of DOI:10.1016/j.ejmech.2007.08.016

In an approach to improve the pharmacological properties and pharmacokinetic profiles of the current protease inhibitors (PIs) used in clinics, and consequently, their therapeutic potential, we performed the synthesis of PI-spacer-valine prodrugs (PI = saquinavir, nelfinavir and indinavir; spacer = -C(O)(CH₂)₅NH-), and evaluated their in vitro stability with respect to hydrolysis, anti-HIV activity, cytotoxicity, and permeation through a monolayer of Caco-2 cells (used as a model of the intestinal barrier), as compared with their parent PI and first generation of valine-PIs (wherein valine was directly connected through its carboxyl to the PIs). The PI-spacer-valine conjugates were prepared in two steps, in good yields, by condensing an acid derivative of the appropriate protected valine-spacer moiety with the PI, followed by deprotection of the valine protecting group. With respect to hydrolysis, we found that the PI-spacer-valine prodrugs were chemically more stable than the first generation of PI-Val prodrugs. Their stabilities correlated with the low to very low in vitro anti-HIV activity measured for those prodrugs wherein the coupling of valine-spacer residue to the PIs was performed onto the peptidomimetic PI's hydroxyl. Prodrugs wherein the coupling of the valine-spacer residue was performed onto the non-peptidomimetic PI hydroxyl displayed a higher antiviral activity, indicating that these prodrugs are also to some extent anti-HIV drugs by themselves. While the direct conjugation of l-valine to the PIs constituted a most appealing alternative, which improved their absorptive diffusion across Caco-2 cell monolayers and reduced their recognition by efflux carriers, its conjugation to the PIs through the -C(O)(CH₂)₅NH- spacer was found to inhibit their absorptive and secretory transepithelial transport. This was attributable to a drastic reduction of their passive permeation and/or active transport, indicating that the PI-spacer-valine conjugates are poor substrates of the aminoacid carrier system located at the brush border side of the Caco-2 cell monolayer.

Full text of DOI:10.1016/j.ejmech.2007.08.016

Available online at www.sciencedirect.com
European Journal of Medicinal Chemistry 43 (2008) 1506e1518
http://www.elsevier.com/locate/ejmech
Original article
Synthesis and in vitro biological evaluation of valine-containing
prodrugs derived from clinically used HIV-protease inhibitors
Dominique Roche a, Jacques Greiner a, Anne-Marie Aubertin b, Pierre Vierling a,
*
a Laboratoire de Chimie des Mole´cules Bioactives et des Aroˆmes, UMR 6001, Universite´ de Nice Sophia-Antipolis, CNRS,
Faculte´ des Sciences, Institut de Chimie de Nice, 28, avenue de Valrose, 06100 Nice, France
b INSERM U74, Institut de Virologie, 3 rue Koeberle´, 67000 Strasbourg, France
Received 23 April 2007; received in revised form 13 July 2007; accepted 9 August 2007
Available online 14 September 2007
Abstract
In an approach to improve the pharmacological properties and pharmacokinetic profiles of the current protease inhibitors (PIs) used in clinics,
and consequently, their therapeutic potential, we performed the synthesis of PIespacerevaline prodrugs (PI ¼ saquinavir, nelfinavir and indi-
navir; spacer ¼ eC(O)(CH2)5NHe), and evaluated their in vitro stability with respect to hydrolysis, anti-HIV activity, cytotoxicity, and perme-
ation through a monolayer of Caco-2 cells (used as a model of the intestinal barrier), as compared with their parent PI and first generation of
valineePIs (wherein valine was directly connected through its carboxyl to the PIs). The PIespacerevaline conjugates were prepared in two
steps, in good yields, by condensing an acid derivative of the appropriate protected valineespacer moiety with the PI, followed by deprotection
of the valine protecting group. With respect to hydrolysis, we found that the PIespacerevaline prodrugs were chemically more stable than the
first generation of PIeVal prodrugs. Their stabilities correlated with the low to very low in vitro anti-HIV activity measured for those prodrugs
wherein the coupling of valineespacer residue to the PIs was performed onto the peptidomimetic PI’s hydroxyl. Prodrugs wherein the coupling
of the valineespacer residue was performed onto the non-peptidomimetic PI hydroxyl displayed a higher antiviral activity, indicating that these
prodrugs are also to some extent anti-HIV drugs by themselves. While the direct conjugation of L-valine to the PIs constituted a most appealing
alternative, which improved their absorptive diffusion across Caco-2 cell monolayers and reduced their recognition by efflux carriers, its con-
jugation to the PIs through the eC(O)(CH2)5NHe spacer was found to inhibit their absorptive and secretory transepithelial transport. This was
attributable to a drastic reduction of their passive permeation and/or active transport, indicating that the PIespacerevaline conjugates are poor
substrates of the aminoacid carrier system located at the brush border side of the Caco-2 cell monolayer.
Ó 2007 Elsevier Masson SAS. All rights reserved.
Keywords: HIV-protease inhibitors; Valine-containing prodrugs; Saquinavir; Indinavir; Nelfinavir; Transport; Caco-2 monolayer
1. Introduction
reservoirs or sanctuaries for the virus, such as the lymphatic
system and central nervous system [3e5], wherein the antivi-
rals, and more particularly the protease inhibitors (PIs), do not
penetrate at an efficient inhibitory level [6,7]. Most of these
PIs display disadvantageous physicochemical and pharmaco-
logical properties. To overcome their suboptimal pharmacoki-
netics, high daily doses must be ingested, often with food and
fluid restrictions. This complicates patient adherence, and
contributes to resistance issues and to the appearance of seri-
ous long-term metabolic complications, such as cardiovascular
disturbances, hyperlipidemia, lipodystrophy, insulin resis-
tance, osteopenia, and diabetes [8e12], and to lower the viral
Highly active antiretroviral therapies (HAART) against
AIDS-HIV infection using a combination of HIV reverse
transcriptase and protease inhibitors have been remarkably
successful, leading to a decline in morbidity and mortality
[1,2]. However, despite these HAARTs, viral replication is still
persisting, indicating, among others, the existence of
* Corresponding author. Tel.: þ33 4 92 07 61 43; fax: þ33 4 92 07 61 51.
E-mail address: vierling@unice.fr (P. Vierling).
0223-5234/$ - see front matter Ó 2007 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2007.08.016

D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
1507
treatment outcome [13]. In order to reduce viral replication in
infected patients, an alternative is to improve the pharmaco-
logical properties, safety and pharmacokinetic profiles, and,
consequently, the therapeutic potential of the PIs already
used in clinics. Aiming at this goal, we [14e20] and others
[21e31] adopted the very efficient prodrug approach. Success-
ful results were obtained with the disclosure of the FDA ap-
Ind(8)eVal (see structure in Fig. 1) exhibited extended stabil-
ity with respect to hydrolysis and a dramatic improvement of
permeation characteristics compared to the parent compound
(Fig. 1) [16,18]. Ind(8)eVal was therefore selected as an inter-
esting candidate for further in situ and in vivo investigations in
an attempt to characterize both its oral bioavailability and
CNS distribution. Unfortunately, its promising biopharmaceut-
ical characteristics suggested from the in vitro experiments
were not confirmed by the in situ and in vivo experiments
[19]. Apart from Ind(8)eVal, most of these PIeaminoacid
ester compounds released the active free drug very rapidly
(half-lives of hydrolysis in buffer at 37 ꢀC of 3e4 h) while
their carbamate analogs were too stable, thus hampering
further in vivo developments [15,18].
In continuation with this work, we designed the PIe
spacereaminoacid prodrugs wherein L-valine and the PI are
tethered together with a eC(O)(CH2)5NHe spacer. The spacer
was introduced for the modulation of the chemical and biolog-
ical stability of the prodrugs, and for the accessibility of va-
line, thus allowing a better recognition by its cellular
transporters. To preserve its recognition by the aminoacid
transporter systems, valine was conjugated via its acid func-
tion to the amine function of the spacer unit, which, via its
acid extremity, was linked to the PI through a hydrolyzable es-
ter function. As esterases are ubiquitous in cells, in vivo hy-
drolysis of the ester prodrugs is expected to release the
active parent drug [38]. Such a spacer strategy proved very
successful for increasing the resistance to hydrolysis of the
proved fosamprenavir,
a phosphate ester prodrug of
amprenavir [32e36]. However, a number of shortcomings
have still to be overcome for the PIs. Among others, the gen-
eration of PI prodrugs should display higher water solubility,
increased bioavailability (plasma concentration, blood circula-
tion time), and/or improved delivery of the parent PI into HIV
sanctuaries. One should therefore be able with these PI pro-
drugs to (i) circumvent drug inactivation resulting from their
in vivo binding to plasma proteins, (ii) limit their rapid metab-
olization and inactivation by cytochrome P450 3A4, and/or
(iii) inhibit their possible transport by the multidrug-resistant
P-glycoprotein (P-gp) responsible for their limited oral bio-
availability and brain penetration [13,37].
In our previous studies [14e20], and with the aim of im-
proving drug permeation through the intestinal and blood brain
barrier by targeting influx carrier-mediated transport systems,
we designed PIeaminoacid ester prodrugs. The benefit of this
approach was shown for increasing their permeation across
polarized monolayers formed of the human intestinal Caco-2
cells (which are widely approved models of the intestinal
epithelium). Among many tested compounds, the ester
32
23
Saquinavir prodrug
Nelfinavir prodrugs
O
24
25
31
30
33
34
22
21
3
7
15
O
14
16
29
28
4
NH2
OR
13
11
2
20
OR
a
a
35
36
26
27
H
H
N
1
4
N
5
N
6
N
N
9
5
8
18
10
8
2
3
10
12
26
17
1
9
18
R O
N
37
38
b
6
28
29
27
19
H
19
O
7
O
11
40
25
N
31
S
O
N
O
41
20
21
32
H
H
24
12
30
39
17
16
13
14
Saquinavir (Ra = H)
Ra = ValNH(CH2)5C(O)
23
Nelfinavir (Ra= Rb= H)
22
Saq-C(O)C5NVal
15
Nelf(1)-C(O)C5NVal
Nelf-[C(O)C5NVal]2(2TFA)
Ra = H; Rb= ValNH(CH2)5C(O)
Ra= Rb = ValNH(CH2)5C(O)
33
Indinavir prodrugs
32
31
34
17
26
N
24
25
35
Structure of ValNH(CH ) C(O) (Ra and/or Rb)
2 5
N
OR
27
28
16
30
a
OR
36
b
3'
H
7"
29
N
NH2
O
N
18
15
8
5'
H3C
13
11
O
9
H
5"
3"
14
12
10
4'
N
1"
19
20
7
2'
1' 6" 4"
2"
1
CH3
O
6
O
22
N
2
H
23
5
21
Previously synthesized PI prodrugs [16]
3
Val
H3C
Indinavir (Ra= Rb= H)
NH2
4
Ra = H; Rb= Val
Ra = Val; Rb= H
Ra = Val
Ind(8)-Val
Ind(14)-Val
Saq-Val
Ra = H; Rb= ValNH(CH2)5C(O)
Ra = Rb = ValNH(CH2)5C(O)
Ind(8)-C(O)C5NVal
Ind-[C(O)C5NVal]2
CH3
O
Fig. 1. Chemical structures and code names of the valine-containing protease inhibitor (PI ¼ saquinavir, indinavir, and nelfinavir) prodrugs described in this study
and atom numbering used in the description of their NMR spectra.

1508
D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
highly unstable PIetyrosine conjugates [14,18]. However, its
impact on the pharmacological properties could therefore not
been evaluated.
This paper is dedicated to the synthesis of various
PIespacerevaline prodrugs (Fig. 1), their chemical stability
with respect to hydrolysis under physiological conditions,
and their in vitro anti-HIV activity. Using the well-known
Caco-2 intestinal barrier model, we report also on their trans-
epithelial transport. All these features are some of the prereq-
uisites for further in vitro and in vivo investigations.
Nelfe[C(O)C5Val]2 (isolated as its di-TFA salt) prodrugs,
respectively. The TFA anion quantification in the isolated mate-
rials was assessed by 19F NMR using 3,3,3-trifluoroethanol as
internal standard.
The chemical structures of all protected and deprotected PI
1
prodrugs were unambiguously ascertained by H, 13C NMR
and mass spectrometry. That the isolated monoesters resulted
from the selective acylation of the C-26 saquinavir, C-8 indi-
navir and C-1 nelfinavir hydroxyls, respectively, was attested
1
more particularly by H and 13C NMR. As expected, the res-
onances of the corresponding H-26 proton and C-26 carbon
atoms of the saquinavir derivatives were shifted downfield
[jDj ¼ 1.25 ꢁ 1.36 ppm and jDj ¼ 6.6 ꢁ 7.3 ppm, respectively]
in comparison with those of saquinavir [d(H-26) ¼ 3.95 ppm;
d(C-26) ¼ 67.1 ppm]. For indinavir, acylation at the C-8 posi-
tion (as in Ind(8)eC(O)C5N(Boc)Val) was evidenced by the
downfield shift of the H-8 (jDj ¼ 1.10 ꢁ 1.18 ppm) and C-8
carbon resonance (jDj ¼ 3.0 ꢁ 3.8 ppm), and the upfield shift
of the vicinal C-9 carbon signal (see Table 1). Furthermore,
that the H-14/C-14 and C-13/15 vicinal b-carbon resonances
of indinavir and Ind(8)eC(O)C5N(Boc)Val were located at
very close chemical shifts indicated that the C-14 hydroxyl
group was not acylated [41,42]. By contrast, when esterifica-
tion of the indinavir C-14 position (as in Inde[C(O)C5N
(Boc)Val]2) had occurred, a significant downfield shift of the
H-14/C-14 atom resonances and upfield shift of the C-13/15
vicinal b-carbon ones was observed, as expected (see Table
1). For nelfinavir, acylation of only its aromatic C-1 hydroxyl
(as in Nelf(1)eC(O)C5N(Boc)Val) was highlighted by the
shielding (jDj ¼ 5.7 ppm) of the C-1 carbon signal and the
concomitant deshielding of its C-2/6 vicinal b-carbon atoms
(by about 8 and 6 ppm), as expected for phenyl esters. That
the remaining C-18 hydroxyl of Nelf(1)eC(O)C5N(Boc)Val
was not substituted was further supported by the H/C-18 and
C-10/19 vicinal resonances, which appeared at very close
chemical shifts to those of nelfinavir (see Table 1) [43,44].
Its esterification (as in Nelfe[C(O)C5N(Boc)Val]2) was un-
ambiguously confirmed by the downfield shift of the H-18 sig-
nal from 3.98 ppm (for free nelfinavir) to 5.36 ppm for the
diester prodrug, and of C-10/19 vicinal resonances, though
the C-18 chemical shift remains almost unaffected (see Table 1).
All these data are further in line with those reported for the
C-8/C-14 indinavir and C-1/C-18 nelfinavir prodrugs
[14,15,18,20].
2. Chemistry
The PIespacerevaline prodrugs shown in Fig. 1 were pre-
pared in two steps by condensing saquinavir, indinavir or nel-
finavir with the key (N-Boc)-valineespacer acid synthon 1,
followed by deprotection of the amino protecting group
(Scheme 1). Synthon 1 was obtained by condensing 6-amino-
hexanoic acid with N-Boc-(L)-Val in the presence of HOSu/
DCC (75% yield [39]).
Concerning saquinavir, acylation of its unique hydroxyl
was performed in nearly 70% yield using conventional EDC/
DMAP as coupling reagent [40]. Concerning indinavir and
nelfinavir, both of which contain two hydroxyls, and in line
with previous work [14,15,18,20], a stochastic esterification e
one to one equivalent e was preferred instead of a more te-
dious and time-consuming protection/deprotection strategy
(though more elegant and probably more efficient in terms
of yields) of one of these two hydroxyls. The stochastic ester-
ification led into the formation of a mixture of mainly the
monoester and diester resulting from the acylation of the
C1-hydroxyl of nelfinavir (30% yield) and C8-hydroxyl of in-
dinavir (20% yield), and of both hydroxyls (10% yield for nel-
finavir and 18% yield for indinavir), respectively (for the atom
numbering, see Fig. 1). Only traces of the monoester corre-
sponding to the acylation of the C18-hydroxyl of nelfinavir
and C14-hydroxyl of indinavir were detected by TLC.
Deprotection of the N-Boc protecting group in the saqui-
navir, indinavir or nelfinavir conjugates was achieved in a 1:9
TFA/CH2Cl2 medium and afforded in 87e100% yields
the target SaqeC(O)C5NVal (isolated as its tri-TFA salt),
Ind(8)eC(O)C5NVal (isolated as its mono-TFA salt), Inde
[C(O)C5Val]2 (isolated as its di-TFA salt), Nelf(1)e
C(O)C5Val (isolated as its mono-TFA salt), and
(i)
(ii)
Saq-C(O)C5N(Boc)Val
(69%)
Saq-C(O)C5NVal
(92%)
H3C
O
Saquinavir
CH3
CH3
(ii)
(ii)
(ii)
(ii)
Ind(8)-C(O)C5N(Boc)Val (20%)
+
Ind(8)-C(O)C5NVal (90%)
(87%)
Ind-[C(O)C5NVal]2
(i)
NH
O
H3C
H
N
(18%)
Indinavir
Ind-[C(O)C5N(Boc)ValP]
(CH2)5CO2H
CH3
O
Nelf(1)-C(O)C5N(Boc)Val (30%)
+
Nelf(1)-C(O)C5NVal (98%)
Nelf-[C(O)C5NVal]2 (95%)
1
(i)
(10%)
Nelfinavir
Nelf-[C(O)C5N(Boc)Val]2
Scheme 1. Synthetic pathway to the PIespacerevaline prodrugs (PI ¼ saquinavir, indinavir, and nelfinavir; spacer ¼ eC(O)(CH2)5NHe): (i) EDC/DMAP, CH2Cl2
(or DMF); (ii) TFA, CH2Cl2. All the deprotected prodrugs were isolated as TFA salts. For more details, see Section 6.

D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
1509
Table 1
vitro in CEM-SS and MT-4 cell cultures infected with HIV-1
LAI and HTLV IIIB, respectively, according to published
procedures [45e47]. All these data are collected in Table 2
together with those of their parent PIs [14,48] and first gener-
ation of PIeVal prodrugs [14].
Their permeation was assessed in vitro through a monolayer
of Caco-2 cells used as an intestinal barrier model, according
to the same protocol employed previously [14]. The results of
the bi-directional transport studies of the various PIespacere
valine prodrugs and of their parent PI are presented in Fig. 2
(which illustrates the transport profiles) and Table 3 (which
collects the percentages of transported (pro)drug and the ap-
parent permeability coefficients Papp calculated from the slope
of a plot of the cumulative receiver concentration with time).
13C chemical shifts of the indinavir C-8/14, nelfinavir C-1/18 carbon atoms
bearing a hydroxyl or an acyl functionality, and of their vicinal carbon atoms
(C-b) (for the atom numbering, see Fig. 1)
Compound
13C chemical shifts (in ppm)
C-14 C-13
(b)
C-15 C-8 C-7 C-9
(b) (b) (b)
Indinavir
65.8 39.2
61.5 73.0 38.1 57.5
61.6 76.0 37.6 55.1
Ind(8)eC(O)C5N(Boc)Val 66.2 38.8
Ind(8)eC(O)C5NVal
Inde[C(O)C5N(Boc)Val]2 69.8 37.5
67.1 40.6 or 40.3 62.6 76.8 38.2 56.7
58.9 75.8 39.0 55.2
58.3 75.6 38.7 55.0
Inde[C(O)C5NVal]2
69.5 37.1
Compound
13C chemical shifts (in ppm)
C-18
C-10 C-19 C-1
(b) (b)
C-6
(b)
C-2
(b)
Nelfinavir
70.5
54.2 59.6 156.8 123.4 116.8
54.7 59.8 151.1 129.6 124.7
54.4 59.6 151.1 129.4 124.6
4. Results and discussion
Nelf(1)eC(O)C5N(Boc)Val 70.7
Nelf(1)eC(O)C5NVal 70.4
4.1. Biological activity and chemical stability
Nelfe[C(O)C5N(Boc)Val]2 70.5 or 70.3 51.3 55.9 149.6 128.2 123.2
Nelfe[C(O)C5NVal]2
72.1
51.7 57.8 151.2 129.3 124.9
The saquinavir hydroxyl, the indinavir C-14 but not the C-8
hydroxyl, and the nelfinavir C-18 but not the C-1 hydroxyl, are
involved in the peptidomimetic noncleavable transition state
isostere responsible for the protease inhibitory potency of sa-
quinavir, indinavir, and nelfinavir, respectively [49]. There-
fore, it is most important that these hydroxyls be accessible
for antiviral activity. In the previous studies dedicated to the
ester prodrugs of saquinavir, indinavir and nelfinavir [14,18],
a close correlation between their anti-HIV activity and the hy-
drolysis of their acylated ‘‘peptidomimetic’’ hydroxyl, hence
the liberation of the active free drug during the time of incu-
bation, was found: the faster the hydrolysis, the closer the
anti-HIV activity level to that of the respective parent drug.
Concomitantly, the level of HIV inhibition was very low for
the prodrugs for which hydrolysis and release of this
3. Pharmacology
The in vitro stability of synthesized saquinavir, nelfinavir
and indinavir prodrugs with respect to hydrolysis (expressed
by their hydrolysis half life (t1/2)) was checked using the
same hydrolysis protocol as that described in previous studies
from our laboratory [14]. These experiments were performed
in a pH 7.3 buffer at 37 ꢀC and in the absence of serum, cells
and virus using a prodrug concentration in the 0.26e1.85 mM
range which necessitated the addition of 2e5% (v/v) of MeOH
to the medium. The HIV inhibition levels (IC50) and cytotox-
icities (CC50) of the valineespacer saquinavir, indinavir, and
nelfinavir conjugates (as their TFA salt) were evaluated in
Table 2
Anti-HIVactivity (IC50) and cytotoxicity (CC50) data for saquinavir, indinavir and nelfinavir prodrugs in CEM-SS and MT-4 cell cultures infected with HIV-1 LAI
and HTLV IIIB, respectively, together with their hydrolysis half life (t1/2
)
Compounda
IC50 (nM) CEM-SS
Rb
IC50 (nM) MT4
Rb
CC50 (M) CEM-SS
CC50 (M) MT4
c
t1/2 (h)
Saquinavirh
9
e
18
290
340
22
e
>10ꢁ5
>10ꢁ5
9 ꢂ 10ꢁ6
>10ꢁ4
>10ꢁ4
>10ꢁ4
>10ꢁ5
>5 ꢂ 10ꢁ5
e
>10ꢁ5
e
SaqeVale
140
175
ꢃ10
17
16
19
16
19
>10ꢁ5
40
60
e
46f
68
72
80
e
SaqeC(O)C5NVal
Indinavirg
Ind(8)eVale
Ind(14)eVale
Ind(8)eC(O)C5NVal
Inde[C(O)C5NVal]2
Nelfinavirh
53 ꢂ 10ꢁ7
e
e
>10ꢁ4
ꢄ1.7
ꢄ9
150
27
7
w1
10
>10ꢁ4
90
36
>10ꢁ4
ꢄ3.6
ꢄ64
e
230
2950
e
4.7 ꢂ 10ꢁ5
4.5 ꢂ 10ꢁ5
e
640
2d
134
e
Nelf(1)eC(O)C5NVal
Nelfe[C(O)C5NVal]2
a
94
15,500
47
7750
230
>CC50
>10ꢁ5
5 ꢂ 10ꢁ6
>10ꢁ5
4.8 ꢂ 10ꢁ6
72
82
Prodrugs were used as their TFA salt; see Fig. 1 for the structures of valine-derived prodrugs.
b
c
R is the ratio of the prodrug IC50 to that of its parent compound.
t
1/2, which corresponds to the time at which 50% of hydrolysis is observed (measurement of disappearance of the prodrug), has been determined from hydro-
lysis experiments performed by incubating the prodrugs in a pH 7.3 DMEM solution at 37 ꢀC.
d
Data from Ref. [48].
Data from Ref. [14].
e
f
Data from Ref. [18].
Used as its sulfate.
Used as its mesylate salt.
g
h

1510
D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
20
15
10
5
20
20
15
10
5
A1
A2
B2
B4
A3
C1
C2
[Saquinavir] = 9.4 µM
[Saq-Val] = 21 µM
(100% hydrolysis during 3 h incubation)
[Saq-C(O)C5Val] = 26.5 µM
15
10
5
10.3
9.6
BL to AP
BL to AP
5.6
BL to AP
2.7
4.8
2.2
4.3
6.1
3.2
2.6
AP to BL
1.8
1.5
1.8
AP to BL
3.1
0.9
1.5
AP to BL
1
0
0
0
0
0
0
1
2
3
0
0
0
1
2
3
0
0
0
1
2
3
40
30
20
10
0
40
30
20
10
0
20
15
10
5
B1
38
[Indinavir] = 175 µM
[Ind(8)-Val] = 330 µM
[Nelfinavir] = 80 µM
26.2
BL to AP
BL to AP
18.1
12.2
11.5
6.3
10.7
16.6
BL to AP
3.8
AP to BL
6.1
2.7
2.4
0.5
AP to BL
2.6
2.2
1.4
0
AP to BL
3.6
0
1
2
3
1
2
3
1
2
3
40
30
20
10
0
40
30
20
10
0
20
15
10
5
B3
[Ind(8)-C(O)C5NVal] = 185 µM
[Ind-[C(O)C5NVal]2] = 150 µM
[Nelf(1)-C(O)C5NVal] = 195 µM
or
[Nelf-[C(O)C5NVal]2] = 248 µM
9
BL to AP
BL to AP
BL to AP
AP to BL
3
2.1
1.9
1.1
0.6
0
AP to BL
AP to BL
1
2
3
1
2
3
1
2
3
Fig. 2. Bi-directional transepithelial transport across a Caco-2 cell monolayer of PIespacerevaline-conjugates (panels A3, B3-4, C2) in comparison with that of
their parent drugs (panels A1, B1, C1) and/or PIeVal conjugates (panels A2, B2) (data from Ref. [16]). Absorptive translocation [apical (AP) to basolateral (BL)
compartment]: lozenges. Secretory translocation (BL to AP): squares. The results are expressed as (pro)drug transport percentages (%) vs time (in hours). The
percentage values represent the ratios of (pro)drug concentration in receiver vs donor chamber ꢂ 100. The initial concentration of the (pro)drug in the donor com-
partment is indicated on each panel. Results are means ꢅ SD from three experiments. All incubations were performed at 37 ꢀC and pH 7.4.
Table 3
Log P data of the PIs and prodrugs, their percentages of (pro)drug transport from the donor chamber after the 3 h of experiment, and apparent permeability
coefficients Papp
Compounda [concentration in
donor chamber (mM)]
Log Pd
BL to AP
AP to BL
% (ꢅSD) in receiver
10.3 (1.9)
Papp (ꢅSD) (cm/s, ꢂ1E ꢁ 8)
398 (117)
% (ꢅSD) in receiver
2.6 (0.4)
Papp (ꢅSD) (cm/s, ꢂ1E ꢁ 8)
110 (31)
Saquinavir [9.4]f
4.727
(4.73)e
5.963
6.182
3.681
(3.68)e
4.923
4.891
5.141
6.531
5.842
(5.84)e
6.377
7.436
SaqeValb,c [21]
9.6 (0.1)
4.8 (0.4)
38 (2)
343 (22)
187 (24)
1550 (234)
6.1 (0.3)
2.2 (0.1)
2.6 (0.3)
232 (22)
58 (8)
103 (10)
SaqeC(O)C5NVal [26.5]
Indinavir [175]g
Ind(8)eValb [330]
18.1 (0.3)
14.6 (0.5)
9 (0.9)
737 (26)
708 (147)
292 (17)
625 (86)
251 (68)
15.6 (0.4)
14.0 (0.4)
nd
644 (44)
513 (50)
e
Ind(14)eValb [177]
Ind(8)eC(O)C5NVal [185]
Inde[C(OC5NVal]2 [150]
Nelfinavir [80]f
2.1 (0.2)
6.3 (1.4)
nd
e
103 (66)
2.7 (1.1)
Nelf(1)eC(O)C5NVal [195]
Nelfe[C(O)C5NVal]2 [248]
nd
nd
e
e
nd
nd
e
e
nd ¼ not detected, i.e. below the detection limit which is 0.4, 0.1, and 0.1 mM for the indinavir, saquinavir, and nelfinavir derivatives, respectively.
a
Prodrugs were used as their TFA salts; see Fig. 1 for the structures of valine-derived prodrugs.
Data from Ref. [16].
b
c
100% hydrolysis during 3 h experiment (for more details, see Section 6).
Calculated using Chem 3D Ultra 8.0, CambridgeSoft, Cambridge, MA.
d
e
Data taken from Ref. [60] and calculated using CQSAR program.
Used as its mesylate salt.
Used as its sulfate.
f
g

D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
1511
peptidomimetic hydroxyl was very slow. On the other hand, no
correlation was found between the hydrolysis rate of the acyl-
ated C-8 indinavir prodrugs (C-8 hydroxyl is not part of the
transition state isostere) and their anti-HIV activity [14].
All the valineespacer prodrugs, when incubated in the
same conditions (pH 7.3 buffer at 37 ꢀC and in the absence
of serum, cells and virus) as the first generation of PIeVal pro-
drugs, were hydrolyzed within a close interval of time, their
hydrolysis t1/2 being comprised between 60 and 82 h (Table 2).
As compared with the chemical stability of the PIeVal pro-
drugs, the PIeC(O)C5NVal conjugates displayed a 50e60%
improved resistance to hydrolysis, which can be attributed to
the presence of the eC(O)(CH2)5NHe spacer. These results
are in line with those established for tyrosineePI prodrug an-
alogs containing a eC(O)(CH2)4e spacer linking the PI to the
tyrosine hydroxyl [18].
In line with previous studies, the anti-HIV efficiency of sa-
quinavir, indinavir, and nelfinavir was substantially reduced
upon conjugation of the valineespacer moiety to the peptido-
mimetic hydroxyl of these PIs (as in SaqeC(O)C5NVal,
Inde[C(O)C5Val]2 and Nelfe[C(O)C5Val]2) as well as to
the hydroxyl of indinavir and nelfinavir, which is not part of
the transition state isostere (as in Ind(8)eC(O)C5NVal,
Nelf(1)eC(O)C5NVal). This is illustrated by the R values
(from ꢄ3.6 to 7750), which correspond to the ratio of the
IC50 of the PI prodrug vs that of the parent PI (see Table 2).
The lower anti-HIV activities of these PI prodrugs reflect to
some extent their relative stabilities with respect to hydrolysis
during the 5-day time-span of the antiviral assays. In line with
previous studies, we found also a correlation between antiviral
activity and chemical stability, the most active prodrug in
a given series (i.e. SaqeVal vs SaqeC(O)C5NVal and
Ind(8)eVal vs Ind(8)eC(O)C5NVal) being the less stable
one (hence SaqeVal and Ind(8)eVal).
It should further be emphasized that the Ind(8)evaline and
Nelf(1)evaline conjugates can be seen not only as prodrugs
but also as potential anti-HIV drugs. Indeed, the C-8 hydroxyl
of indinavir and C-1 hydroxyl of nelfinavir (onto which the
valineespacer unit was conjugated) are not involved in the
noncleavable transition state isostere responsible for their pro-
tease inhibitory potency. Their hiding (as in Ind(8)eVal,
Ind(8)eC(O)C5NVal, Nelf(1)eVal, Nelf(1)eC(O)C5NVal)
should therefore not modify drastically the antiviral activity
of indinavir and nelfinavir, respectively. This was indeed
what was found for the indinavir conjugates, which display
an antiviral activity slightly lower to that of indinavir
(R w 3.6e10). However, masking of the ‘‘non-peptidomi-
metic’’ C-1 hydroxyl of nelfinavir decreased more signifi-
cantly the antiviral activity of nelfinavir (R w 47 for
Nelf(1)eC(O)C5NVal). These data indicate likely that (i)
these conjugates themselves may possess an antiviral activity,
which is lower than that of their parent drug, and/or (ii) the
connection of the L-valine substituents onto this non-peptido-
mimetic hydroxyl reduces the recognition by the protease,
thus decreasing the interactions between the peptidomimetic
noncleavable isostere and viral protease active site. This phe-
nomenon also accounts, in part, for the very low anti-HIV
activity of the indinavir- and nelfinavir-diester derivatives. It
is further noteworthy that no cytotoxicity was detected for
any of these prodrugs.
4.2. Transepithelial transport
There is currently a considerable interest in increasing the
absorptive permeability of the HIV PIs and reducing their se-
cretion by the efflux carrier systems, such as the multidrug-
resistant P-glycoprotein (P-gp) which is responsible for their
limited oral bioavailability and brain penetration [50,51]. In
this regard, esterase-rich and efflux carrier-expressing Caco-
2 cell line monolayers are widely accepted in vitro models
of the intestinal epithelium for screening drug and prodrug
candidates and thus for evaluating prodrug approaches for en-
hanced intestinal drug absorption [52e56]. Caco-2 cell line
monolayers have already been used in numerous studies to
characterize the permeation of various (pro)drugs, including
indinavir, saquinavir, nelfinavir [50,51,57,58], and those pro-
drugs, which have been developed in our laboratory
[16,18,20].
Translocation of the various valineespacerePI prodrugs
across the polarized Caco-2 cell monolayers was evaluated
at a concentration where they are soluble and in comparison
with that of their parent PI and first generation of ValePIs. In-
vestigation of the transport across the cell monolayer in the ab-
sorptive [from apical (AP) to basolateral (BL)] and secretory
(from BL to AP) directions constitutes a mean of evaluating
the influence of P-gp and related efflux carriers, which are lo-
cated on the AP side of the monolayer. Transepithelial electri-
cal resistance after confluence has been used to monitor the
integrity of the cell monolayer [59]. Translocation was initi-
ated by adding the test solution to the AP or BL side of the
monolayer (donor chamber). As Caco-2 cells are also rich in
esterases, the hydrolysis of the ester prodrugs during their per-
meation across the monolayer was carefully checked. Only the
prodrugs were detected either in the donor or acceptor cham-
ber, indicating that no hydrolysis had occurred after the 3 h of
transport experiments.
Most importantly, the transport results (Fig. 2 and Table 3)
indicate that conjugation of L-valine to the PIs through
a spacer, as in the PIeC(O)C5NVal conjugates, had detrimen-
tal effects on PI permeation across the Caco-2 cell monolayer,
by contrast with its direct conjugation, as in SaqeVal, Ind(8)e
Val and Ind(14)eVal, which was most beneficial. Indeed, in
the adsorptive AP to BL direction and in the saquinavir series,
a nearly 3-fold translocation decrease was measured for Saqe
C(O)C5NVal (Fig. 2, panel A3) as compared with SaqeVal
(Fig. 2, panel A2), permeation of SaqeC(O)C5NVal being fur-
ther comparable with that of its parent saquinavir (Fig. 2, panel
A1). No transport at all was detected neither for Ind(8)e
C(O)C5NVal (Fig. 2, panel B3) nor for Nelf(1)eC(O)C5NVal
(Fig. 2 panel C2), at least above the detection limit for those
derivatives, whereas a substantial (5- to 6-fold) improvement
of indinavir permeation was noticed with Ind(8)eVal
(Fig. 2, panel B2) and Ind(14)eVal (not shown) [16]. In the
secretory BL to AP direction, a significantly larger reduction

1512
D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
of permeation of the PIs was also found for the PIe
C(O)C5NVal prodrugs as compared with the PIeVal conju-
gates. In comparison with that of its parent drug, secretory
efflux of SaqeC(O)C5NVal (respectively IndeC(O)C5NVal)
was 2-fold (respectively 6-fold) lower, whereas that of Saqe
Val (respectively IndeVal) was similar (respectively only 2-
to 3-fold lower).
the coupling of valineespacer residue to the PIs was
performed onto the peptidomimetic PI’s hydroxyl (as in
SaqeC(O)C5NVal, Inde[C(O)C5NVal]2 and Nelfe[C(O)
C5NVal]2). Ind(8)eC(O)C5NVal and Nelf(1)eC(O)C5NVal
prodrugs wherein the coupling of the valineespacer residue
was performed onto the non-peptidomimetic PI hydroxyl
displayed a higher antiviral activity, indicating that these
prodrugs are also to some extent anti-HIV drugs by
themselves.
While the conjugation of L-valine (through its carboxyl) to
the PIs constituted a most appealing alternative, which im-
proved their absorptive diffusion across Caco-2 cell mono-
layers used as a model of the intestinal barrier, and reduced
their recognition by efflux carriers, its conjugation to the PIs
through the eC(O)(CH2)5NHe spacer was found to inhibit
their absorptive and secretory transepithelial transport.
This was attributable to a drastic reduction of their passive
permeation and/or active transport, indicating that the PIe
spacerevaline conjugates are poor substrates of the aminoacid
carrier system located at the brush border side of the Caco-2
cell monolayer.
These substantial PI absorptive and secretory transport
decreases resulting from the conjugation of valine to the PIs
through the eC(O)(CH2)5NHe spacer are attributable to a dras-
tic reduction of either (i) the active transport, indicating that the
PIeC(O)C5NVal conjugates are, as compared to the PIeVal
conjugates, poor substrates of the aminoacid carrier system lo-
cated at the brush border side of the Caco-2 cell monolayer,
and/or (ii) more likely the passive permeation. It can also be
assigned to an increase of affinity of the apically localized efflux
P-gp carriers for those PIeC(O)C5NVal conjugates, as
supported by their asymmetric permeation profile [50,51].
Moreover, the lower secretory transport for the valinee
spacerePI conjugates as compared to that of their parent PI
and PIeVal conjugates indicates further that conjugation of
thevalineespacer moiety to the PIs reduces their passive perme-
ation to a larger extent than the direct conjugation of valine.
However, the decrease of passive permeation is not reflected
by the resulting increase of lipophilicity when going from
the PIs, to the PIeVal conjugates then to the PIespacereVal
derivatives, as expressed by the increase of their log P coeffi-
cient (Table 3).
6. Materials and methods
6.1. Chemistry
6.1.1. General methods, reagents and starting materials
All the reactions were performed under anhydrous nitrogen
using dry solvents and reagents. Anhydrous solvents were pre-
pared by standard methods. N,N0-Dicyclohexylcarbodiimide
(DCC), 1-(3-dimethylaminopropyl)-3-ethyl-carbodiimide hy-
drochloride (EDC), 4-dimethylaminopyridine (DMAP), N-
hydroxysuccinimide (HOSu) were purchased from Aldrich,
and 6-aminohexanoic acid, N-Boc-L-valine, and trifluoroace-
tic acid (TFA) from Fluka. All these materials were used
without further purification. Saquinavir, indinavir and nelfina-
vir (as their methanesulfonate salt or sulfate salt) were a gift
from Hoffmann-La Roche, E. Merck, and Agouron, respec-
tively, and were deprotonated prior to their use in the syn-
thetic processes (CHCl3 or EtOAc extraction of the free
base from a NaHCO3 or Na2CO3 10% solution of the prote-
ase inhibitor).
Such poor absorptive and secretory transport profiles were
also observed for PIespaceretyrosine conjugates [18]. How-
ever, it could not be attributed to the presence of the spacer
unit as the PIetyrosine derivatives were too unstable to allow
transport experiments to be performed.
5. Conclusion
This study, aimed at improving the pharmacological proper-
ties and pharmacokinetic profiles of the PIs and of their first gen-
eration of valineePI prodrugs wherein valine (through its
carboxyl) was directly connected to the PIs, was dedicated to
the synthesis of PIespacerevaline prodrugs (PI ¼ saquinavir,
indinavir, and nelfinavir; spacer ¼ eC(O)(CH2)5NHe), and
the evaluation of their chemical stability, anti-HIV activity,
and permeation across monolayers of Caco-2 cells as models
of the intestinal barrier. The PIespacerevaline conjugates
were prepared in two steps, in good yields, by condensing an
acid derivative of the appropriate protected valineespacer moi-
ety with the PI, followed by deprotection of thevaline protecting
group. With respect to hydrolysis, we found that these PIe
spacerevaline prodrugs are chemically more stable than the first
generation PIeVal prodrugs. Their chemical stabilities with
half-life times in the 60e82 h range are further compatible
with an in vivo utilization aimed at improving the absorption/
penetration or accumulation of the prodrug in specific cells/
tissues and liberation of the active free drug.
Column chromatography purifications were carried out on
Silica Gel 60 (E. Merck, 70e230 mesh). The purity of all
new compounds was checked by TLC, NMR, MS and
HPLC. TLC analyses were performed on precoated Silica
Gel F254 plates (E. Merck) with detection by UV and ninhy-
drin. HPLC analyses of the synthesized prodrugs (flow of
1 mL/min) were performed using a HP1100 apparatus using
a Lichrospher 100 RP-18 (5 mm) column (250 ꢂ 3.2 mm)
with gradient A: H2OeCH3CN (v:v) 0.1% TFA gradient as el-
uent (from 80:20 to 0:100) over 30 min; UV detection at
210 nm. With these conditions, retention times (tR) of indina-
vir, nelfinavir, and saquinavir are of 10.4, 10.8, and 15.3 min,
1
respectively. H, 13C, and 19F NMR spectra were recorded
These stabilities correlate with the low to very low in
vitro anti-HIV activity measured for those prodrugs wherein
with a Brucker AC 200 or AC 500 spectrometer at 200
(or 500), 50.3 (or 125.8), and 188.3 MHz, respectively.

D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
1513
19F NMR with 3,3,3-trifluoroethanol as internal standard was
used to assess the TFA anion quantification in the isolated
valine-deprotected target compounds. Chemical shifts (d) are
given in ppm with respect to tetramethylsilane measured indi-
(1H, m, H-12), 2.83 (1H, t, J 8.6 Hz, H-18a), 2.70e2.47
(7H, m, H-15, H-16, H-17a, H-18b, H-30b), 2.41 (1H, t, J
8.2 Hz, H-17b), 1.97 (4H, m, H-40, H-200, H-13a), 1.50 (1H,
m, H-13b), 1.40 [13H, m, H-300, H-500, C(CH3)3], 1.32 (9H,
m, H-23), 1.18 (1H, m, H-400a), 1.11 (1H, m, H-400b), 0.88,
0.85 (6H, 2d, J 6.6 Hz, H-50), 13C NMR (CDCl3): d 174.9
(C-11), 172.5 (C-10), 172.0 (C-100), 169.8 (C-20), 155.9
[OC(O)NH], 150.4 (C-29), 149.0 (C-28), 140.9 (C-31),
139.8, 139.4 (C-1, C-6), 136.8 (C-26), 132.7 (C-25), 128.9
(C-32, C-36), 128.3 (C-33, C-35), 128.0 (C-2), 127.0, 126.3
(C-3, C-4), 125.0 (C-34), 123.7 (C-5), 123.5 (C-27), 79.8
[C(CH3)3], 76.0 (C-8), 66.2 (C-14), 65.0 (C-19), 61.6 (C-
15), 60.3 (C-20), 60.1 (C-24), 55.1 (C-9, C-16), 52.6 (C-17),
51.1 (C-22), 48.7 (C-18), 46.1 (C-12), 39.4 (C-30), 38.8 (C-
13), 38.7 (C-600), 37.6 (C-7), 33.9 (C-200), 30.7 (C-40), 29.0
(C-23), 28.6 (C-500), 28.4 [C(CH3)3], 25.8 (C-400), 24.0 (C-
300), 19.3, 18.1 (C-50).
1
rectly (i) to CHCl3 (d 7.27) for H and (ii) to CDCl3 (d 76.9)
for 13C. Concerning the description of the prodrug NMR spec-
tra, the atoms of the PI part are depicted as C-x and H-y
whereas those of the valine part are depicted as C-x0 and H-
y0 and those of the linker part are depicted as C-x00 and H-y00
(see Fig. 1 for numbering). COSY 1H/1H, 1H/13C NMR corre-
lation (on Brucker AC 500 spectrometer), 13C DEPT, and/or
mass spectrometry data fully confirm the signal assignments
and structure of the isolated materials. Electron-spray ioniza-
tion mass spectra (ESI-MS) were run on a Finnigan MAT
TSQ 7000 apparatus equipped with an atmospheric pressure
ionization source. This method used in positive mode gives
either Mþ, [M þ H]þ and/or [M þ Na]þ signals.
The synthesis of 6-[2(S )-tert-butoxycarbonylamino-3-
methyl-butyrylamino]-hexanoic acid, (Boc)ValNC5C(O) 1
was performed from N-Boc-(L)-valine, 6-amino-hexanoic
acid using DCC/HOSu activation, according to literature
[39]: Rf 0.36 (98:2 CH2Cl2eMeOH); 1H NMR (CDCl3):
d 9.76 (1H, bs, OH), 6.98 (1H, m, H-700), 5.72 (1H, d, J
9.1 Hz, H-30), 3.86 (1H, m, H-20), 3.19 (2H, m, H-600), 2.27
(2H, t, J 7.3 Hz, H-200), 1.96 (1H, m, H-40), 1.62e1.30 (6H,
m, H-300eH-500), 1.37 [9H, s, C(CH3)3], 0.87 (6H, d, J
6.1.2.1.2. Inde[C(O)C5N(Boc)Val]2. Rf 0.52 (98:2
CH2Cl2eMeOH); tR ¼ 20.6 min (gradient A); 1H NMR
(CDCl3): d 8.48 (2H, m, H-28, H-29), 7.63 (1H, d, J 7.7 Hz,
H-26), 7.21e7.18 (11H, m, H-2eH-5, H-21, H-27, H-32eH-
36), 6.92 (2H, bs, H-700), 6.77 (1H, d, J 8.5 Hz, H-10), 5.66
(1H, dd, J 5.3, J 8.5 Hz, H-9), 5.45e5.10 (4H, m, H-30, H-8,
H-14), 3.88 (2H, t, J 7.7 Hz, H-20), 3.47 (2H, m, H-24),
3.13e2.04 (26H, m, H-40, H-200, H-600, H-7, H-12, H-13, H-
15eH-19, H-30), 1.43e1.11 [39H, m, H-23, H-300eH-500,
C(CH3)3], 0.91 (12H, m, H-50); 13C NMR (CDCl3): d 174.3
(C-11), 173.6, 172.6 (C-10), 171.8 (C-100), 170.3 (C-20),
156.1 [OC(O)NH], 150.2 (C-29), 148.7 (C-28), 140.8 (C-
31), 139.3, 139.1 (C-1, C-6), 136.8 (C-26), 133.0 (C-25),
128.8 (C-32, C-36), 128.4 (C-33, C-35), 128.0 (C-2), 127.1,
126.5 (C-3, C-4), 124.9 (C-34), 124.0 (C-5), 123.4 (C-27),
79.6 [C(CH3)3], 75.8 (C-8), 69.8 (C-14), 67.0 (C-19), 60.1
(C-20), 59.9 (C-24), 58.9 (C-15), 55.2 (C-9), 55.7, 52.1 (C-
16, C-17), 50.9 (C-22), 50.1 (C-18), 45.2 (C-12), 39.5
(C-30), 39.0 (C-600, C-7), 37.5 (C-13), 34.3, 33.9 (C-200),
30.9 (C-40), 29.0 (C-23), 28.3 (C-500), 28.4 [C(CH3)3], 26.4,
26.2 (C-400), 24.6, 24.1 (C-300), 19.3, 18.0 (C-50).
13
6.4 Hz, H-50); C NMR (CDCl3): d 177.4 (C-100), 172.4 (C-
10), 156.4 [OC(O)NH], 79.9 [C(CH3)3], 60.7 (C-20), 39.6 (C-
600), 34.3 (C-200), 31.2 (C-40), 29.4 (C-500), 28.7 [C(CH3)3],
26.7 (C-400), 24.8 (C-300), 19.7 and 18.7 (C-50).
6.1.2. Synthesis of protected Boc-valine prodrugs of
protease inhibitors (condensation step)
6.1.2.1. Ind(8)eC(O)C5N(Boc)Val and Inde[C(O)C5N
(Boc)Val]2. General esterification method: Compound
1
(270.3 mg, 0.82 mmol) and DMAP (39.4 mg, 0.32 mmol)
were added to indinavir (506 mg, 0.75 mmol) in CH2Cl2
(10 mL). Then, EDC (372 mg, 1.95 mmol) was added to the
mixture at 0 ꢀC and the solution was stirred for 15 min at
0 ꢀC then for 24 h at room temperature. The reaction mixture
was diluted with CH2Cl2 and washed. The organic layer dried
over Na2SO4 was filtered, and evaporated under reduced pres-
sure. Then, the residue was chromatographed on silica gel
(10:0 to 9.5:0.5 CH2Cl2eEtOH) to give Ind(8)eC(O)C5N
(Boc)Val (300 mg, 20%) and Inde[C(O)C5N(Boc)Val]2
(360 mg, 18%) as white solids.
6.1.2.1.1. Ind(8)eC(O)C5N(Boc)Val. Rf 0.44 (98:2
CH2Cl2eMeOH); tR ¼ 13.2 min (gradient A); 1H NMR
(CDCl3): d 8.50 (2H, m, H-28, H-29), 7.61 (1H, d, J 7.7 Hz,
H-26), 7.45 (1H, bs, H-21), 7.27 (1H, m, H-27), 7.21e7.11
(9H, m, H-2eH-5, H-32eH-36), 6.63 (1H, d, J 8.5 Hz,
H-10), 6.52 (1H, bs, H-700), 5.64 (1H, dd, J 5.3, J 9.0 Hz,
H-9), 5.25 (1H, m, H-8), 5.18 (1H, d, J 8.9 Hz, H-30), 3.86
(1H, m, H-14), 3.78 (1H, t, J 8.0 Hz, H-20), 3.47 (2H, s, H-
24), 3.12 (1H, dd, J 4.8, 17.1 Hz, H-7a), 3.06e3.00 (4H, m,
H-600, H-19, H-30a), 2.90 (1H, d, J 17.1 Hz, H-7b), 2.92e2.88
6.1.2.2. SaqeC(O)C5N(Boc)Val. The general esterification
method applied to 1 (250 mg, 0.75 mmol), DMAP (39.4 mg,
0.32 mmol), saquinavir (506 mg, 0.75 mmol), and EDC
(172 mg, 0.9 mmol) in 10 mL CH2Cl2 afforded after work-
up, and purification by chromatography on silica gel (10:0
to 9.8:0.2 CHCl3eMeOH) SaqeC(O)C5N(Boc)Val (510 mg,
69%) as a white solid. Rf 0.50 (98:2 CH2Cl2eMeOH);
tR ¼ 19.7 min (gradient A); 1H NMR (CDCl3eCD3OD):
d 8.33 (1H, d, J 8.6 Hz, H-6), 8.04 (2H, m, H-1, H-7), 7.87
(1H, d, J 7.5 Hz, H-4), 7.71 (1H, m, H-2), 7.56 (1H, m, H-
3), 7.14 (2H, m, H-21, H-25), 6.92 (2H, t, J 7.4 Hz, H-22,
H-24), 6.77 (1H, m, H-23), 5.20 (1H, m, H-26), 4.87 (m, H-
12, partially hidden by the signal of water), 4.34 (1H, m, H-
18), 3.76e3.30 (2H, m, H-20, H-37), 3.22e2.52 (8H, m,
H-13, H-19, H-27, H-29), 2.66e2.52 (4H, m, H-30, H-35,
H-36), 2.32e1.48 (19H, m, H-40, H-200eH-600, H-31eH-34),
1.33 (9H, s, H-41), 1.24 [9H, s, C(CH3)3], 0.81 (6H, m, H-
13
50); C NMR (CDCl3eCD3OD): d 175.4 (C-100), 174.9,

1514
D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
174.7 (C-14, C-16), 174.2 (C-10), 172.2 (C-38), 165.9 (C-10),
157.7 [OC(O)NH], 150.1 (C-8), 147.7 (C-9), 139,1 (C-20),
138.8 (C-6), 131.4 (C-1), 130.7 (C-2, C-4), 130.1 (C-21, C-
25), 129.4 (C-5), 129.2 (C-22, C-24), 128.9 (C-3), 127.1 (C-
23), 119.5 (C-7), 80.4 [C(CH3)3], 74.4 (C-26), 70.6 (C-37),
61.6 (C-20), 59.8 (C-29), 56.8 (C-27), 52.9 (C-12), 51.9 (C-
40), 51.4 (C-18), 40.0 (C-600), 37.9 (C-13), 36.8 (C-30), 35.5
(C-200), 35.0 (C-19), 34.6 (C-35), 31.9 (C-40), 29.0 (C-41),
28.7 [C(CH3)3], 31.7, 31.3, 27.3, 27.0, 25.9, and 25.3 (C-
300eC-500, C-31eC-34, C-36), 19.8, 18.5 (C-50).
1.02 (9H, s, H-32), 0.90, 0.86 (6H, 2d, J 6.7 Hz, H-50), 0.80
(6H, 2d, J 7.0 Hz, H-50); 13C NMR (CDCl3): d 174.2 (C-8),
172.6, 171.8, and 171.7 (C-10, C-29), 171.4 169.4 (C-100),
156.0, 155.9 [OC(O)NH], 149.6 (C-1), 138.9 (C-5), 135.8 (C-
12), 129.8 (C-13, C-17), 128.9 (C-14, C-16), 128.2 (C-6),
126.3 (C-3), 126.2 (C-15), 124.9 (C-4), 123.2 (C-2), 79.7,
79.4 [C(CH3)3], 70.5, 70.3 (C-18, C-28), 60.1, 59.9 (C-20),
59.5 (C-20), 55.9 (C-19), 51.3 (C-10), 51.1 (C-31), 39.1 (C-
600), 36.0 (C-21), 35.7 (C-11), 34.2, 34.0 (C-200), 33.8 (C-26),
31.1 (C-40), 31.0 (C-27), 29.2 (C-500), 28.3 (C-32), [C(CH3)3],
26.4, 26.3 (C-400), 24.5, 24.4 (C-300), 29.0, 25.9, and 20.4 (C-
22eC-25), 19.3, 18.0, 17.8 (C-50), 13.0 (C-7).
6.1.2.3. Nelf(1)-C(O)C5N(Boc)Val and Nelf-[C(O)C5N(Boc)
Val]2. The general esterification method was applied to 1
(247 mg, 0.75 mmol), DMAP (135 mg, 0.90 mmol), nelfinavir
(427 mg, 0.75 mmol) and EDC (173 mg, 0.90 mmol) in 10 mL
DMF. After evaporation of DMF and extraction with AcOEt,
the residue was purified by chromatography on silica gel
(100:0e98:2 CHCl3eEtOH) giving Nelf(1)eC(O)C5N
(Boc)Val (198 mg, 30%) and Nelfe[C(O)C5N(Boc)Val]2 (89
mg, 10%) as white solids.
6.1.3. Synthesis of valine prodrugs of protease inhibitors
(deprotection step)
6.1.3.1. Ind(8)eC(O)C5NVal. General deprotection method:
Ind(8)eC(O)C5N(Boc)Val (250 mg, 0.27 mmol) in a 1:9
TFAeCH2Cl2 v/v mixture (12 mL) was stirred for 4 h at
0 ꢀC. The residue obtained after evaporation of the solvents
was purified by chromatography (9:1 to 8:2 CH2Cl2e
MeOH) to give Ind(8)eC(O)C5NVal (as its mono-TFA salt)
(228 mg, 90%) as a white solid. Rf 0.34 (9:1 CH2Cl2e
MeOH); tR ¼ 9.3 min (gradient A); 1H NMR (CD3OD):
d 8.57 (1H, m, H-29), 8.52 (1H, m, H-28), 7.96 (1H, d, J
7.9 Hz, H-26), 7.51 (1H, dd, J 5.1, J 7.9 Hz, H-27), 7.34e
7.09 (9H, m, H-2eH-5, H-32eH-36), 5.52 (1H, d, J
10.1 Hz, H-9), 5.33 (1H, td, J 1.8, J 5.1 Hz, H-8), 4.05e3.45
(4H, m, H-20, H-14, H-24), 3.26e2.60 (16H, m, H-600, H-7,
H-12, H-15eH-19, H-30), 2.11 (3H, m, H-40, H-200), 1.85
(1H, m, H-13a), 1.45e1.35 (7H, m, H-13b, H-300eH-500),
1.25 (9H, m, H-23), 0.97 (6H, d, J 6.8 Hz, H-50); 13C NMR
(CD3OD): d 177.5 (C-11), 174.5 (C-10, C-100), 169.3 (C-20),
150.0 (C-29), 148.9 (C-28), 141.5 (C-31), 140.9, 140.5 (C-1,
C-6), 140.8 (C-26), 133.5 (C-25), 130.1 (C-32, C-36), 129.5
(C-33, C-35), 129.3 (C-2), 128.2, 127.5 (C-3, C-4), 125.9
(C-34), 125.1 (C-5, C-27), 76.8 (C-8), 67.1 (C-14), 67.0 (C-
19), 62.6 (C-15), 59.9 (C-20), 59.0 (C-24), 56.7 (C-9), 54.4
(C-16, C-17), 52.7 (C-22), 50.9 (C-18), 45.8 (C-12), 40.6,
40.3 (C-600, C-13, C-30), 38.2 (C-7), 34.7 (C-200), 31.5 (C-
40), 29.9 (C-500), 28.7 (C-23), 27.4 (C-400), 25.4 (C-300), 18.þ8,
18.1 (C-50); ESI-MS (positive mode): m/z 826.7 [M þ H] ,
848.6 [M þ Na]þ, in agreement with the calculated mass for
[M] ¼ C47H67N7O6 (825.52).
6.1.2.3.1. Nelf(1)eC(O)C5N(Boc)Val. Rf 0.27 (98:2
CHCl3eEtOH); tR ¼ 21.95 min (gradient A); 1H NMR
(CD3OD): d 7.54 (2H, d, J 7.5 Hz, H-13, H-17), 7.41 (1H,
dd, J 7.6, J 0.9 Hz, H-4), 7.36e7.30 (2H, m, H-14, H-16),
7.31 (1H, t, J w 7.9 Hz, H-3), 7.25e7.22 (1H, m, H-15),
7.13 (1H, d, J 8.0 Hz, H-2), 4.51 (1H, m, H-10), 4.15 (1H,
m, H-18), 3.88 (1H, d, J 6.7 Hz, H-20), 3.60e3.49 (2H, m,
H-11), 3.38e3.30 (1H, m, H-600a), 3.29e3.23 (1H, m, H-
600b), 3.08 (1H, d, J 10.0 Hz, H-20a), 2.70e2.63 (4H, m, H-
200, H-19a, H-28), 2.30 (3H, s, H-7), 2.60e2.23 (2H, m,
H19b, H-20b), 2.08e2.02 (1H, m, H-40), 1.51 [9H, S,
C(CH3)3], 1.88e1.25 (18H, m, H-300eH-500, H-21eH-27),
1.21 (9H, s, H-32), 1.00, 0.99 (6H, 2d, J 6.9 Hz, H-50); 13C
NMR (CD3OD): d 176.2 (C-8), 174.3, 173.4 (C-10, C-29),
172.3 (C-100), 157.9 [OC(O)NH], 151.1 (C-1), 140.1 (C-5),
137.6 (C-12), 130.8 (C-13, C-17), 130.1 (C-14, C-16), 129.6
(C-6), 127.7 (C-3), 127.3 (C-15), 126.3 (C-4), 124.7 (C-2),
80.8 [C(CH3)3], 71.1 (C-28), 70.7 (C-18), 61.7 (C-20), 60.6
(C-20), 59.8 (C-19), 54.7 (C-10), 52.1 (C-31), 40.3 (C-600),
37.6 (C-21), 35.9 (C-11), 35.1 (C-26), 35.0 (C-200), 32.2 (C-
40), 31.9 (C-27), 30.1 (C-500), 29.2, 29.1 [C-32, C(CH3)3],
27.6 (C-400), 25.8 (C-300), 30.1, 27.2, and 21.9 (C-22eC-25),
18.8, 18.7 (C-50), 13.9 (C-7).
6.1.2.2.2. Nelfe[C(O)C5N(Boc)Val]2. Rf 0.36 (98:2
CHCl3eEtOH); tR ¼ 26.60 min (gradient A); 1H NMR
(CDCl3): d 7.77 (1H, d, J 8.9 Hz, NH), 7.40 (2H, d, J 7.5 Hz,
H-13, H-17), 7.31 (1H, d, J 7.4 Hz, H-4), 7.24e7.21 (2H, m,
H-14, H-16), 7.16e7.11 (2H, m, H-3, H-15), 6.97 (1H, d, J
8.0 Hz, H-2), 6.79 (1H, m, NH), 6.54 (1H, bs, NH), 5.82 (1H,
s, NH), 5.36 (1H, m, H-18), 5.25 (2H, m, NH), 4.60 (1H, m,
H-10), 3.84 (1H, m, H-20), 3.62 (1H, m, H-11a), 3.42 (1H, m,
H-11b), 3.29e3.03 (4H, m, H-600), 2.96 (1H, d, J 10.6 Hz, H-
20a), 2.59 (1H, m, H-19a), 2.52 (2H, t, J 7.5 Hz, H-200), 2.40
(1H, m, H-28), 2.21 (2H, m, H-200), 2.16 (3H, s, H-7), 2.11e
1.89 (4H, m, H-40, H-19b, H-20b), 1.39, 1.37 [18H, 2s,
C(CH3)3], 1.74e1.12 (24H, m, H-300eH-500, H-21eH-27),
6.1.3.2. Inde[C(O)C5NVal]2. Deprotection of Inde[C(O)
C5N(Boc)Val]2 (300 mg, 0.24 mmol) afforded after purifica-
tion by chromatography (9:1 to 8:2 CH2Cl2eMeOH) Inde
[C(O)C5NVal]2 (as its di-TFA salt) (288 mg, 87%) as a white
solid. Rf 0.23 (9:1 CH2Cl2eMeOH); tR ¼ 9.1 min (gradient
1
A); H NMR (CDCl3eCD3OD): d 8.91 (2H, bs, H-28, H-
29), 8.20 (1H, d, J 8.1 Hz, H-26), 7.94 (1H, s, H-21), 7.61e
7.86 (10H, m, H-2eH-5, H-27, H-32eH-36), 6.07 (1H, d, J
5.2 Hz, H-9), 5.83 (1H, m, H-8), 5.58 (1H, m, H-14), 3.98
(1H, m, H-20), 3.81e3.60 (4H, m, H-7, H-24), 3.57e2.45
(23H, m, H-40, H-200, H-600, H-12, H-13, H-15eH-19, H-30),
2.12e1.66 (23H, m, H-300eH-500, H-23), 1.44e1.34 (12H, m,

D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
1515
H-50); 13C NMR (CDCl3-CD3OD): d 175.2 (C-11), 173.7 (C-
10), 173.2 (C-100), 170.8 (C-20), 149.4 (C-29), 147.8 (C-28),
140.3 (C-31), 139.2, 138.7 (C-1, C-6), 137.5 (C-26), 133.2
(C-25), 128.6 (C-32, C-36), 128.2 (C-33, C-35), 127.9 (C-2),
126.9, 126.3 (C-3, C-4), 124.7 (C-34), 123.7 (C-5, C-27),
75.6 (C-8), 69.5 (C-14) 66.8 (C-19), 59.6 (C-20), 59.4 (C-
24), 58.3 (C-15), 55.0 (C-9), 55.3, 51.7 (C-16, C-17), 50.8
(C-22), 49.9 (C-18), 44.2 (C-12), 39.3 (C-30), 38.7 (C-600,
C-7), 37.1 (C-13), 33.9, 33.6 (C-200), 31.0 (C-40), 28.7, 28.6
(C-500), 28.2 (C-23), 26.1, 26.0 (C-400), 24.2, 23.9 (C-300),
18.7, 16.6 (C-50); ESI-MS (positive mode): m/z 1039.0
[M þ H]þ, 1061.7 [M þ Na]þ, in agreement with the calcu-
lated mass for [M] ¼ C58H87N9O8 (1037.67).
19), 54.4 (C-10), 51.9 (C-31), 40.2 (C-600), 37.6 (C-21), 35.3
(C-11), 35.2 (C-26), 34.7 (C-200), 32.5 (C-40), 32.1 (C-27), 30.0
(C-500), 28.8 (C-32), 27.5 (C-400), 25.6 (C-300), 31.7, 27.1, and
21.7 (C-22eC-25), 19.3, 17.9 (C-50), 13.5 (C-7); ESI-MS (posi-
tive mode): m/z 780.6 (M þ H)þ, in agreement with the calcu-
lated mass for [M] ¼ C43H65N5O6S (779.47).
6.1.3.5. Nelfe[C(O)C5NVal]2. Deprotection of Nelfe
[C(O)C5N(Boc)Val]2 (60 mg, 0.05 mmol) afforded after puri-
fication by chromatography (1:0 to 85:15 CH2Cl2eMeOH)
Nelfe[C(O)C5NVal]2 (as its di-TFA salt) (58 mg, 95%) as
a white solid. Rf 0.22 (9:1 CHCl3eEtOH); tR ¼ 13.47 min
1
(gradient A); H NMR (CD3OD): d 7.58 (2H, m, H-13, H-
17), 7.32e7.20 (6H, m, H-3, H-4, H-14eH-16), 7.06 (1H,
m, H-2), 5.52 (1H, m, H-18), 4.53 (1H, m, H-10), 3.75 (1H,
m, H-20), 3.53 (2H, dd, J 6.1, J 8.2 Hz, H-11), 3.40e3.08
(6H, m, H-600, H-19a, H-20a), 2.93 (2H, m, H-200), 2.60 (1H,
t, J 7.2 Hz, H-28), 2.38 (2H, m, H-200), 2.16 (3H, s, H-7),
2.20e1.20 (28H, m, H-40, H-300eH-500, H-19b, H-20b, H-
21eH-27), 1.27 (9H, s, H-32), 0.96, 0.99 (12H, 2d, J
6.9 Hz, H-50); 13C NMR (CD3OD): d 174.3 (C-8), 173.3,
172.6 (C-10, C-29), 169.4, 168.6 (C-100), 151.2 (C-1), 139.6
(C-5), 136.1 (C-12), 131.8 (C-13, C-17), 130.3 (C-14, C-16),
129.3 (C-6), 128.2 (C-3), 127.8 (C-15), 125.8 (C-4), 124.9
(C-2), 72.1 (C-18, C-28), 59.9 (C-20), 57.8 (C-19, C-20),
52.8 (C-31), 51.7 (C-10), 40.4 (C-600), 35.5 (C-21), 35.2 (C-
11), 34.7, 34.6 (C-200), 33.0 (C-26), 31.5 (C-40), 29.9 (C-500,
C-27), 28.7 (C-32), 27.5 (C-400), 25.6 (C-300), 31.5, 29.9,
27.5, and 25.1 (C-22eC-25), 18.8, 18.0 (C-50), 13.5 (C-7);
ESI-MS (positive mode): m/z 992.8 (M þ H)þ, in agreement
with the calculated mass for [M] ¼ C54H85N7O8S (991.62).
6.1.3.3. SaqeC(O)C5NVal. Deprotection of SaqeC(O)C5N(-
Boc)Val (350 mg, 0.36 mmol) afforded SaqeC(O)C5NVal (as
its tri-TFA salt) (400 mg, 92%) as a white solid. Rf 0.59 (9:1
1
CH2Cl2eMeOH), HPLC: tR ¼ 14.4 min; H NMR (CD3OD):
d 8.45 (1H, d, J 8.5 Hz, H-6), 8.14 (1H, d, J 8.5 Hz, H-7),
8.10 (1H, d, J 7.1 Hz, H-1), 7.97 (1H, d, J 8.2 Hz, H-4),
7.81 (1H, td, J 1.5, J 7.7 Hz, H-2), 7.67 (1H, td, J 7.5, J
1.20 Hz, H-3), 7.16 (2H, d, J 7.2 Hz, H-21, H-25), 6.98 (2H, t,
J 7.4 Hz, H-22, H-24), 6.83 (1H, m, H-23), 5.31 (1H, m, H-
26), 4.87 (m, H-12, partially hidden by the signal of water),
4.45 (1H, m, H-18), 3.62e3.52 (2H, m, H-20, H-37), 3.33e
2.90 (8H, m, H-13, H-19, H-27, H-29), 2.74e2.65 (4H, m, H-
30, H-35, H-36), 2.36 (2H, t, J 7.2 Hz, H-200), 2.18e1.45 (17H,
m, H-40, H-300eH-600, H-31eH-34), 1.33 (9H, s, H-41), 0.81
13
(6H, m, H-50); C NMR (CD3OD): d 174.9 (C-100), 174.7 (C-
16), 173.3 (C-14), 172.8 (C-10), 169.3 (C-38), 166.2 (C-10),
150.2 (C-9), 147.9 (C-20), 139.0 (C-7), 139.3 (C-8), 131.6 (C-
2), 130.9 (C-5), 130.7 (C-1), 130.1 (C-21, C-25), 129.5 (C-3),
129.3 (C-22, C-24), 129.1 (C-4), 127.4 (C-23), 119.6 (C-6),
73.7 (C-26), 66.8 (C-37), 59.9 (C-20), 57.3 (C-27, C-29), 53.2
(C-12), 52.4 (C-40), 51.6 (C-18), 40.4 (C-600), 37.8 (C-13),
36.3 (C-19), 36.1 (C-30), 34.8 (C-200), 34.0 (C-35), 31.5 (C-40),
28.9 (C-41), 30.8, 30.0, 27.4, 26.9, 26.6, and 25.3 (C-300eC-500,
C-31eC-34, C-36), 18.9, 18.0 (C-50); ESI-MS (positive mode):
m/z 883.8 (M þ H)þ, 905.6 (M þ Na)þ, in agreement with the
calculated mass for [M] ¼ C49H70N8O9 (882.54).
6.2. Biological assays
6.2.1. General methods, reagents and starting materials
1-Pentanesulfonic acid sodium salt, sodium acetate trihy-
drate, and acetonitrile were high-performance liquid chroma-
tography (HPLC) grade. Foetal bovine serum (FBS),
Dulbecco’s phosphate buffered saline (DPBS), penicillin:-
streptomycin solution (5000 U/mL:5000 mg/mL), trypsine
EDTA 0.25% solution (2.5 g trypsin and 0.2 g EDTA in 1 L
DPBS), D-(þ)-glucose, and N-[2-hydroxyethyl]piperazine-N-
ethanesulfonic acid (HEPES) were purchased from Sigma
Chemical (St Quentin Fallavier, France). Dulbecco’s modified
eagle’s medium (DMEM) and nonessential aminoacids
DMEM 100 ꢂ (NEAA) were purchased from Gibco-Life
Technologies (Cergy-Pontoise, France). Transport Medium
(TM) consisted of DPBS containing 25 mM glucose and
10 mM HEPES (pH 7.4). Cell culture medium consisted of
DMEM supplemented with 20% foetal bovine serum, 1%
NEAA, and 2% penicillinestreptomycin.
6.1.3.4. Nelf(1)eC(O)C5NVal. Deprotection of Nelf(1)e
C(O)C5N(Boc)Val (150 mg, 0.17 mmol) afforded after
purification by chromatography (1:0 to 9:1 CH2Cl2eMeOH)
Nelf(1)eC(O)C5NVal (as its mono-TFA salt) 150 mg (98%) as
a white solid. Rf 0.32 (9:1 CHCl3eMeOH); tR ¼ 14.44 min (gra-
dient A); 1H NMR (CD3OD): d 7.45 (2H, d, J 7.5 Hz, H-13, H-
17), 7.31e7.00 (6H, m, H-2eH-4, H-14eH-16), 4.40 (1H, m,
H-10), 4.05 (1H, m, H-18), 3.60e2.95 (6H, m, H-20, H-600, H-
11, H-20a), 2.64e2.48 (4H, m, H-200, H-19a, H-28), 2.17 (3H,
s, H-7), 2.30e1.88 (3H, m, H-40, H-19b, H-20b), 1.80e1.14
(18H, m, H-300eH-500, H-21eH-27), 1.10 (9H, s, H-32), 0.95,
0.92 (6H, 2d, J 6.8 Hz, H-50); 13C NMR (CD3OD): d 176.2 (C-
8) 173.2, 173.1 (C-10, C-29), 172.3 (C-100), 151.1 (C-1), 140.2
(C-5), 137.7 (C-12), 130.0 (C-13, C-17), 130.0 (C-14, C-16),
129.4 (C-6), 127.6 (C-3), 127.0 (C-15), 126.2 (C-4), 124.6 (C-
2), 71.0 (C-28), 70.4 (C-18), 60.8 (C-20), 60.4 (C-20), 59.6 (C-
6.2.2. HPLC conditions
All HPLC analyses used for chemical stability and trans-
port experiments were performed using a HP1100 apparatus

1516
D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
with a Lichrospher 100 RP-18 (5 mm)-packed column
(250 ꢂ 3.2 mm) with a flow rate of 1 mL/min. The iso-
cratic mobile phase consisted of CH3COONa$3H2O
(15 mM) and CH3(CH2)4SO3Na (15 mM), 5 mM pH 6 buffer
and CH3CN: 59:41 for indinavir, Ind(8)eC(O)C5NVal and
Inde[C(O)C5NVal]2, 41:59 for saquinavir and Saqe
C(O)C5NVal, and 45:55 for nelfinavir, Nelf(1)eC(O)C5NVal
and Nelfe[C(O)C5NVal]2. The prodrugs and/or drugs were
detected by measuring their UV absorption at 210 nm (indi-
navir, nelfinavir and their derivatives) or 254 nm (saquinavir
and its derivative) and the signals (peak integration) were in-
terpreted by the software provided. Under their respective
HPLC conditions, the retention times measured for the differ-
ent compounds were of 10.4 min for indinavir, 10.0 min for
Ind(8)eC(O)C5NVal, 7.0 min for Inde[C(O)C5NVal]2,
5.7 min for saquinavir, 4.3 min for SaqeC(O)C5NVal,
12.5 min for nelfinavir, 10.9 min for Nelf(1)eC(O)C5NVal,
8.2 min for Nelfe[C(O)C5NVal]2. The prodrug and/or drug
concentration was determined from HPLC calibration curves.
These curves were established under the same HPLC condi-
tions, using standard calibrated prodrug and drug solutions
that were prepared in the same hydrolysis medium as the
sample under investigation. The calibration curves were lin-
ear (correlation coefficients from 0.9993 to 0.9998) in a con-
centration range of 0.4 to 1000 mM for indinavir and its
prodrugs, 0.1 to 200 mM for saquinavir and its prodrugs,
and 0.1 to 310 mM for nelfinavir and its prodrugs, the lower
limit corresponding to the limit of detection that can be
quantified with accuracy. Above the lower concentration
limit, the analytical method was reproducible.
1.5 ꢂ 106 cells/flask. For the transepithelial transport experi-
ments, Caco-2 cells (passage 60e65) were harvested with
trypsineEDTA and seeded on Anopore membrane inserts
(0.2 mm pore diameter, 25 mm diameter; Nunc, Roskilde,
Denmark) at a density of 5 ꢂ 105 cells/insert (cm2). Apical
(AP) and basolateral (BL) chamber volumes were maintained
at 2 mL. Culture medium was changed every 3 to 4 days and
cells were used for the experiments between days 14 and 27
post-seeding. Monolayer formation was monitored by mea-
surement of transepithelial electrical resistance (TEER) using
a Millicel ERS apparatus (Millipore).
6.2.5. Transport experiments
Before the transepithelial transport experiments, the Caco-2
monolayers were rinsed twice with the transport medium (TM)
(both chambers) and pre-incubated for 30 min in the TM. Af-
ter this equilibration period, the monolayer integrity was
checked by measuring its TEER. For the transport experi-
ments, only monolayers displaying TEER values above
176 U cm2 and for which TEER values fell by less than 15%
from the value measured at the end of the equilibration period
were used. Under these conditions, the age of the cell mono-
layer within the 14 to 27 days post-seeding range did not affect
the transport results. Transport was initiated by replacing the
TM in the AP or in the BL ‘‘donor’’ compartment with
2 mL of the PI or prodrug solution. AP and BL chamber
volumes were maintained at 2 mL. The test solutions were
prepared by mixing a known amount of TM with a concen-
trated MeOH stock solution of the PI or prodrug under inves-
tigation to reach a final concentration of the PI or prodrug in
the 150e185 mM, 26.5 mM, and 195e248 mM range for the in-
dinavir, saquinavir, and nelfinavir derivatives, respectively (the
exact concentrations used for each derivative are given in the
figures showing their transepithelial transport). The final
MeOH concentration in the PI or prodrug solutions in contact
with the monolayers never exceeded 3% MeOH (TEER mon-
itoring and transport experiments have shown that MeOH at
concentrations up to 5% neither affected cell monolayer integ-
rity during the 3-h period of transport experiment nor the
amount of PI and/or prodrug transported, respectively). Two
hundred microliters of samples were withdrawn from the
‘‘acceptor’’ compartment (opposite to the addition ‘‘donor’’
chamber) every 1 h over a period of 3 h and replaced by the
same amount of fresh transport medium in order to maintain
the same volume. The dilution was taken into account for
the calculations. To prevent hydrolysis of the prodrugs, all
these samples were stored at 4 ꢀC awaiting for prodrug and/
or parent PI HPLC-analysis (see Section 6.2.2). At the end
of the experiments (3 h), samples were also taken from the
‘‘donor’’ compartment for HPLC analysis and the monolayers
were checked for integrity by measuring TEER values. The
concentrations of the prodrug and parent PI that were mea-
sured in the donor and acceptor chamber at the end of the
transport experiment indicated that no non-specific adsorption
on glass or on plastic had occurred. The experiments for which
TEER has decreased by more than 8% from the value
6.2.3. Hydrolysis
Hydrolysis experiments were performed by incubating
20 mL of a DMEMeMeOH solution (pH 7.3) of the prodrug
(250 mg/mL) at 37 ꢀC with stirring. The amount of MeOH
(v/v) in these solutions was 3% for Ind(8)eC(O)C5NVal,
2% for Inde[C(O)C5NVal]2, 4% for SaqeC(O)C5NVal, 5%
for Nelf(1)eC(O)C5NVal and 3% for Nelfe[C(O)C5NVal]2.
Hydrolysis was followed by HPLC monitoring of the disap-
pearance of the prodrug and appearance of the parent PI by
injecting 40 mL of the solution onto the HPLC column (for
HPLC conditions see above). Plots of ln([prodrug]o ꢁ [pro-
drug(t)]) and of ln[PI(t)] against time were linear within the
concentration range studied, indicating that the hydrolysis is
a first order process with respect to the prodrug. The half-lives
of hydrolysis (t1/2) were measured, when possible, or calcu-
lated from these plots; t1/2 is related to the slope, K, of these
curves by the relation t1/2 ¼ (ln 2)/K.
6.2.4. Caco-2 cell culture
Caco-2 cells, clone TC7, were kindly provided by Dr. A.
Zweibaum (INSERM U178, Villejuif, France). The cells
were routinely maintained in 75 cm2 culture flasks at 37 ꢀC
in an atmosphere containing 5% CO2 and 95% relative humid-
ity. Cells were split every 7 days at
a density of

D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
1517
measured at the beginning of the experiment were discarded.
Acknowledgements
Transport was expressed as a percentage of the initial amount
added to the donor compartment. All flux experiments were
conducted at least in triplicate in the AP to BL and BL to
AP directions. If possible, a concentration of the prodrug in
the donor chamber the closest to that of the parent PI was priv-
ileged but solubility issues of the prodrug (which needed in
some cases the use of MeOH) and detection issues of the pro-
drug and/or parent PI in the acceptor chamber was also consid-
ered for the selection of the prodrug concentration.
´
We thank the Fondation de la Recherche Medicale (Sidac-
tion), the Agence Nationale de la Recherche sur le SIDA
(ANRS), and the Centre National de la Recherche Scientifique
´
(CNRS) for financial support, and the Conseil Regional de la
Martinique for grant (D.R.).
References
[1] R.M. Gulick, Infect. Dis. Ther. 25 (2002) 179e236.
[2] A. Winston, M. Boffito, J. Antimicrob. Chemother. 56 (2005) 1e5.
[3] T.-W. Chun, L. Stuyver, S.B. Mizell, L.A. Ehler, J.A.M. Mican,
M. Baseler, A.L. Lloyd, M.A. Nowak, A.S. Fauci, Proc. Natl. Acad.
Sci. U.S.A. 94 (1997) 13193e13197.
6.2.6. Data analysis for the transport experiments
[4] T.-W. Chun, A.S. Fauci, Proc. Natl. Acad. Sci. U.S.A. 96 (1999)
10958e10961.
[5] J.N. Blankson, D. Persaud, R.F. Siliciano, Annu. Rev. Med. 53 (2002)
557e593.
The apparent permeability coefficients Papp (cm/s) were
calculated by linear regression analysis on the time course
plot of amount of (pro)drug transported from equation
Papp ¼ k*(VD/A), where k is the slope of the linear curve
ln(Ct/Co) ¼ ꢁkt, Ct being the (pro)drug concentration in the
receiver chamber at time t, Co the initial concentration in the
donor chamber, A the membrane surface area (4.52 cm2) and
VD the volume of the donor chamber (2 cm3). These calcula-
tions were performed for each transport experiment and the
values reported in Table 3 represent the means (ꢅSD) of at
least three independent experiments.
[6] X. Li, W.K. Chan, Adv. Drug Deliv. Rev. 39 (1999) 81e103.
[7] S. Taylor, A. Pereira, HIV Med. 1 (2000) 18e22.
[8] R.G. Jain, E.S. Furfine, L. Pedneault, A.J. White, J.M. Lenhard, Antiviral
Res. 51 (2001) 151e177.
[9] S.K. Gan, K. Samaras, A. Carr, D. Chisholm, Diabetes Obes. Metab. 3
(2001) 67e71.
[10] P.W. Hruz, H. Murata, M. Mueckler, Am. J. Physiol. 280 (2001)
E549eE553.
[11] M.J. Glesby, Infect. Dis. Ther. 25 (2002) 237e256.
[12] H. Murata, P.W. Hruz, M. Mueckler, Curr. Drug Targets: Infect. Disord. 2
(2002) 1e8.
[13] C.W. Flexner, Infect. Dis. Ther. 25 (2002) 139e159.
[14] A. Farese-Di Giorgio, M. Rouquayrol, J. Greiner, A.-M. Aubertin,
P. Vierling, R. Guedj, Antivir. Chem. Chemother. 11 (2000) 97e110.
[15] M. Rouquayrol, B. Gaucher, J. Greiner, A.-M. Aubertin, P. Vierling,
R. Guedj, Carbohydr. Res. 336 (2001) 161e180.
6.2.7. In vitro antiviral and cytotoxicity assays
The in vitro antiviral activity and cytotoxicity assays were
performed according to published procedures [45e47].
Briefly, CEM-SS cells were infected with a dose of HIV-1
(LAI strain) infecting 50% of the cells. Four days later, the
growth of HIV-1 was evaluated by measuring the reverse tran-
scriptase (RT), which expresses the presence of the virus in the
supernatant culture medium. The tested compounds were
added to the cell cultures after viral infection. RT inhibition%
was measured in comparison with the non-treated cells. The
growth of HIV-1 [HTLV-I (IIIB)] was followed by the cyto-
pathogenic effect induced by the virus in MT4 cells. MT4 cells
were infected with a virus dose allowing 4 days later the death
of 90%. The tested compounds were added in the cell culture
medium after viral infection and cell viability was measured
by the colorimetric MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-di-
phenyl-2H-tetrazolium bromide] test. The percentage of pro-
tection was calculated as the ratio D.
[16] M. Rouquayrol, B. Gaucher, D. Roche, J. Greiner, P. Vierling, Pharm.
Res. 19 (2002) 1704e1712.
[17] P. Vierling, J. Greiner, Curr. Pharm. Des. 9 (2003) 1755e1770.
[18] B. Gaucher, M. Rouquayrol, D. Roche, J. Greiner, A.-M. Aubertin,
P. Vierling, Org. Biomol. Chem. 2 (2004) 345e357.
[19] M. de Oliveira, J.-C. Olivier, C. Pariat, D. Roche, J. Greiner, P. Vierling,
W. Couet, J. Pharm. Pharmacol. 57 (2005) 453e458.
[20] D. Roche, J. Greiner, A.M. Aubertin, P. Vierling, Bioconjug. Chem. 17
(2006) 1568e1581.
[21] G. Bold, A. Faessler, M. Lang, Eur. Pat., 594,540, 1994.
[22] K.Y. Hostetler, D.D. Richman, E.A. Forssen, L. Selk, R. Basava,
M.F. Gardner, S. Parker, C. Basava, Biochem. Pharmacol. 48 (1994)
1399e1404.
[23] J. Alexander, D. Bindra, B.D. Dorsey, A.J. Repta, J.P. Vacca, PCT Int.
Appl., 9514016, 1995.
[24] T. Cynkowski, G. Cynkowska, P.A. Crooks, H. Guo, P. Ashton, Abstr.
Pap.- Am. Chem. Soc. 216 (1998) (MEDI-346).
[25] R.D. Tung, M.R. Hale, C.T. Baker, E.S. Furfine, I. Kaldor, W.W. Kaz-
mierski, A. Spaltenstein, PCT Int. Appl., 9933815, 1999.
[26] W.M. Kazmierski, P. Bevans, E. Furfine, A. Spaltenstein, H. Yang, Bio-
org. Med. Chem. Lett. 13 (2003) 2523e2526.
D ¼ [(OD of treated infected cells ꢁ OD of untreated in-
fected cells)/(OD of non-infected cells ꢁ OD of untreated in-
fected cells)] ꢂ 100.
[27] T. Calogeropoulou, A. Detsi, E. Lekkas, M. Koufaki, Curr. Top. Med.
Chem. 3 (2003) 1467e1495.
[28] S. Gunaseelan, O. Debrah, L. Wan, M.J. Leibowitz, A.B. Rabson,
S. Stein, P.J. Sinko, Bioconjug. Chem. 15 (2004) 1322e1333.
[29] R. Jain, S. Agarwal, S. Majumdar, X. Zhu, D. Pal, A.K. Mitra, Int. J.
Pharm. 303 (2005) 8e19.
The prodrug EC50 values were determined from the curves
of the RT inhibition % (CEM-SS cells) or the protection
percentage D (MT4 cells) against prodrug concentration.
The effect of the prodrugs on cell viability was measured on
non-infected cells using the colorimetric MTT test after 5
days of incubation at 37 ꢀC with various concentrations of
the tested product.
[30] D.A. Degoey, W.J. Flosi, D.J. Grampovnik, L.L. Klein, D.J. Kempf, X.C.
Wang, PCT Int. Appl., 2006014282, 2006.
[31] S. Luo, V.S. Kansara, X. Zhu, N.K. Mandava, D. Pal, A.K. Mitra, Mol.
Pharm. 3 (2006) 329e339.

1518
D. Roche et al. / European Journal of Medicinal Chemistry 43 (2008) 1506e1518
[32] J.M. Gatell, J. HIV Ther. 6 (2001) 95e99.
[33] L.A. Sorbera, L. Martin, J. Castaner, R.M. Castaner, Drugs Future 26
(2001) 224e231.
[46] C. Genu-Dellac, G. Gosselin, F. Puech, J.C. Henry, A.M. Aubertin,
G. Obert, A. Kirn, J.L. Imbach, Nucleosides Nucleotides 10 (1991)
1345e1376.
[34] T.M. Chapman, G.L. Plosker, C.M. Perry, Drugs 64 (2004) 2101e2124.
[35] S. Becker, L. Thornton, Expert Opin. Pharmacother. 5 (2004) 1995e
2005.
[47] C. Moog, A. Wick, P. Le Ber, A. Kirn, A.-M. Aubertin, Antiviral Res. 24
(1994) 275e288.
[48] M.L. West, D.P. Fairlie, Trends Pharmacol. Sci. 16 (1995) 67e74.
[49] F. Lebon, M. Ledecq, Curr. Med. Chem. 7 (2000) 455e477.
[50] R.B. Kim, M.F. Fromm, C. Wandel, B. Leake, A.J.J. Wood, D.M. Roden,
G.R. Wilkinson, J. Clin. Invest. 101 (1998) 289e294.
[36] E.S. Furfine, C.T. Baker, M.R. Hale, D.J. Reynolds, J.A. Salisbury,
A.D. Searle, S.D. Studenberg, D. Todd, R.D. Tung, A. Spaltenstein, Anti-
microb. Agents Chemother. 48 (2004) 791e798.
[37] M.F. Fromm, Int. J. Clin. Pharmacol. Ther. 38 (2000) 69e74.
[38] B. Testa, J.M. Mayer, Hydrolysis in Drug and Prodrug Metabolism:
[51] B.J. Aungst, N.H. Nguyen, J.P. Bulgarelli, K. Oates-Lenz, Pharm. Res.
17 (2000) 1175e1180.
Chemistry, Biochemistry, and Enzymology, Wiley-VCH, Zurich, Swit-
¨
zerland, 2003.
[52] P. Artursson, Crit. Rev. Ther. Drug Carrier Syst. 8 (1991) 305e330.
[53] P. Augustijns, P. Annaert, P. Heylen, G. Van den Mooter, R. Kinget, Int. J.
Pharm. 166 (1998) 45e53.
[39] A. Losse, Tetrahedron 29 (1973) 1203e1208.
[40] M. Bodanszky, A. Bodanszky, The Practice of Peptide Synthesis,
Springer Verlag, Berlin, Germany, 1984.
[54] P. Annaert, G. Gosselin, A. Pompon, S. Benzaria, G. Valette,
J.L. Imbach, L. Naesens, S. Hatse, E. De Clercq, G. Van Den Mooter,
R. Kinget, P. Augustijns, Pharm. Res. 15 (1998) 239e245.
[55] M.-C. Gres, B. Julian, M. Bourrie, V. Meunier, C. Roques, M. Berger,
X. Boulenc, Y. Berger, G. Fabre, Pharm. Res. 15 (1998) 726e733.
[56] P. Artursson, K. Palm, K. Luthman, Adv. Drug Deliv. Rev. 46 (2001) 27e43.
[57] J. Alsenz, H. Steffen, R. Alex, Pharm. Res. 15 (1998) 423e428.
[58] S.D. Ucpinar, S. Stavchansky, Biomed. Chromatogr. 17 (2003) 21e25.
[59] S. Lu, A.W. Gough, W.F. Bobrowski, B.H. Stewart, J. Pharm. Sci. 85
(1996) 270e273.
[41] E. Breitmaier, W. Voelter, Carbon-13 NMR Spectroscopy, , In: Mono-
graphs in Modern Chemistry, vol. 5, Verlag Chemie, Weinheim,
Germany, 1974.
[42] M. Hesse, H. Meier, B. Zeeh, Spectroscopic Methods in Organic Chem-
istry, Thieme, Stuttgart, Germany, 1995.
[43] U. Sequin, Helv. Chim. Acta 64 (1981) 2654e2664.
[44] I.I. Schuster, J. Org. Chem. 50 (1985) 1656e1662.
[45] R. Pauwels, J. Balzarini, M. Baba, R. Snoeck, D. Schols, P. Herdewijn,
J. Desmyter, E. De Clercq, J. Virol. Methods 20 (1988) 309e321.
[60] B. Bhhatarai, R. Garg, Bioorg. Med. Chem. 13 (2005) 4078e4084.