Synthesis of analogs of L-valacyclovir and determination of their substrate activity for the oligopeptide transporter in Caco-2 cells

Article abstract of DOI:10.1016/S0928-0987(02)00047-7

L-Valacyclovir, a prodrug of acyclovir, is a substrate for the peptide transporter (PepT1) in the intestinal mucosa, which accounts for its higher than expected oral bioavailability. The substrate activity of L-valacyclovir for PepT1 is surprising, particularly when one considers that the molecule has the structural features of a nucleoside rather than a peptide. In an attempt to better understand the structure-transport relationships (STR) for the interactions of L-valacyclovir with PepT1, analogs of this molecule with structural changes in the guanine moiety were synthesized and their substrate activity for PepT1 in Caco-2 cell monolayers was determined. The analogs synthesized include those that had the guanine moiety of L-valacyclovir substituted with purine, benzimidazole, and 7-azaindole. All three analogs (purine, benzimidazole, and 7-azaindole) exhibited affinity for PepT1 in binding studies, but only the purine analog (as the L-valine ester) showed PepT1-associated transcellular transport across Caco-2 cell monolayers. The benzimidazole and 7-azaindole analogs (as their L-valine esters) were rapidly metabolized by esterase when applied to the apical surface of Caco-2 cells, which probably explains their low penetration as the intact prodrugs via PepT1.

Full text of DOI:10.1016/S0928-0987(02)00047-7

European Journal of Pharmaceutical Sciences 16 (2002) 1–13
www.elsevier.nl/locate/ejps
Synthesis of analogs of L-valacyclovir and determination of their substrate
activity for the oligopeptide transporter in Caco-2 cells
b
a
c
c
Gerda Marie Friedrichsena , , Weiqing Chen , Mikael Begtrup , Chao-Pin Lee , Philip L. Smith ,
*
Ronald T. Borchardtb
aDepartment of Medicinal Chemistry, The Royal Danish School of Pharmacy, Universitetsparken 2, DK-2100 Copenhagen, Denmark
bDepartment of Pharmaceutical Chemistry, The University of Kansas, Lawrence, KS 66047, USA
cGlaxoSmithKline, Collegeville, PA 19426, USA
Received 8 October 2001; received in revised form 22 March 2002; accepted 2 April 2002
Abstract
L-Valacyclovir, a prodrug of acyclovir, is a substrate for the peptide transporter (PepT1) in the intestinal mucosa, which accounts for its
higher than expected oral bioavailability. The substrate activity of L-valacyclovir for PepT1 is surprising, particularly when one considers
that the molecule has the structural features of a nucleoside rather than a peptide. In an attempt to better understand the structure–
transport relationships (STR) for the interactions of L-valacyclovir with PepT1, analogs of this molecule with structural changes in the
guanine moiety were synthesized and their substrate activity for PepT1 in Caco-2 cell monolayers was determined. The analogs
synthesized include those that had the guanine moiety of L-valacyclovir substituted with purine, benzimidazole, and 7-azaindole. All three
analogs (purine, benzimidazole, and 7-azaindole) exhibited affinity for PepT1 in binding studies, but only the purine analog (as the
L-valine ester) showed PepT1-associated transcellular transport across Caco-2 cell monolayers. The benzimidazole and 7-azaindole
analogs (as their L-valine esters) were rapidly metabolized by esterase when applied to the apical surface of Caco-2 cells, which probably
explains their low penetration as the intact prodrugs via PepT1.
2002 Elsevier Science B.V. All rights reserved.
Keywords: L-Valacyclovir; PepT1; Caco-2 cells; Structure–transport relationships
1. Introduction
carbonyl group of the first peptide bond; a hydrophobic
pocket for binding the side chain of an amino acid; and
one or two carboxylate binding sites.
The oligopeptide transporter PepT1 is localized in the
small intestine and is involved in the absorption of di- and
tri-peptides and many peptidomimetics (Walker et al.,
1996; Adibi, 1997; Dantzig, 1997). Furthermore, several
non-peptidic compounds have been shown to be substrates
for PepT1 (Han et al., 1998; Temple et al., 1998). Bailey
et al. (2000) have suggested a substrate template for PepT1
based on known structure–transport relationship (STR) for
this transporter. This template suggests four key binding
sites on PepT1 for the different structural features of a
substrate: a strong charge–charge interaction binding site
for an N-terminal amino group; a hydrogen bond to the
L-Valacyclovir is an orally active prodrug of the antiviral
drug acyclovir (Beauchamp et al., 1992; Crooks and
Murray, 1994). The oral absorption of this hydrophilic
prodrug appears to be mediated by its substrate activity for
PepT1 (Ganapathy et al., 1998; Sinko and Balimane, 1998;
de Vrueh et al., 1998). While the L-valine ester group of
L-valacyclovir provides an amino group and a carbonyl
group as recognition sites for PepT1, the presence of a
guanine moiety in this molecule makes it more closely
resemble a nucleoside than a di- or tri-peptide. This may
suggest that the guanine moiety of L-valacyclovir resem-
bles the structural features of the amino acid side chain
and/or the C-terminal carboxylate of a di- or tri-peptide,
the other recognition sites suggested by Bailey et al.
(2000) as being important in binding to PepT1.
*
Corresponding author. LEO Pharma. Products, Industriparken 55,
DK-2750 Ballerup, Denmark. Tel.: 145-4492-3800; fax: 145-4484-
5880.
In this study, a series of L-valacyclovir and acyclovir
analogs with structural alterations in the guanine moiety
E-mail address: gerda.friedrichsen@leo-pharma.com (G.M. Friedrich-
sen).
0928-0987/02/$ – see front matter
PII: S0928-0987(02)00047-7
2002 Elsevier Science B.V. All rights reserved.

2
G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
were synthesized in an attempt to elucidate its role (if any)
in binding to PepT1. The substrate activity of these L-
valacyclovir and acyclovir analogs for PepT1 and a
nucleoside transporter were then examined in Caco-2 cells,
an in vitro model of the intestinal mucosa (Hidalgo et al.,
1989).
in the Department of Chemistry at The University of
Kansas.
2.2.1.1. Synthesis of 2-(9-purinemethoxy)ethyl-L-valinate
(2b)
2.2.1.1.1. 9-(2-Benzoyloxyethoxymethyl)-purine To
a
stirred solution of lithium diisopropylamide (LDA) (4.16
ml, 8.33 mmol) in dry N,N-dimethylformamide (DMF) (30
ml) was added a solution of purine (1.00 g, 8.33 mmol) in
DMF (20 ml) at 0 8C under argon. After 5 min, 2-
(chloromethoxy)ethyl benzoate (1.47 ml, 8.33 mmol) was
added, and the reaction mixture was stirred for 26 h at
room temperature. The solution was quenched with water
(20 ml) and the compound was extracted with ethyl acetate
(EtOAc) (3350 ml). The organic phases collected from
the extraction were combined, dried over anhydrous
MgSO4, filtered, and evaporated in vacuo. The crude
product was purified by flash chromatography eluting with
EtOAc to yield a yellow oil (699 mg, 28%) which
solidified upon drying in vacuo. 1H NMR (CDCl3) d 9.15
(1H, s, H-4), 9.00 (1H, s, H-2), 8.25 (1H, s, H-8), 7.95
(2H, d, J58.0 Hz, H-19), 7.54 (1H, t, J57.5 Hz, H-39),
7.42 (2H, t, J57.7 Hz, H-29), 5.77 (2H, s, OCH2N), 4.46
(2H, t, J54.7 Hz, CH2O), 3.93 (2H, t, J54.7 Hz, CH2O);
FAB-MS (in TG/G) m/z: 299.1 (M1H1), calculated
molecular weight: 298.1.
2.2.1.1.2. 9-(2-Hydroxyethoxymethyl)-purine (2a) 9-(2-
Benzoyloxyethoxymethyl)-purine (625 mg, 2.09 mmol)
was stirred in 40% aq. methylamine (60 ml) for 1.5 h at
room temperature. After evaporation of the solvent, the
residue was purified by flash chromatography eluting with
EtOAc–MeOH (9:1) to yield colorless crystals (112 mg,
28%). 1H NMR (CD3OD) d 9.13 (1H, s, H-4), 8.98 (1H, s,
H-2), 8.67 (1H, s, H-8), 5.81 (2H, s, OCH2N), 3.67 (2H,
m, CH2O), 3.65 (2H, m, CH2O); FAB-MS (vscan in
PEG300/Gly) m/z: 195.0882 (M1H1), calculated molec-
ular weight: 194.0803.
2. Materials and methods
2.1. Materials
Starting materials for synthesis, solvents for HPLC, and
all other reagents employed were purchased from Aldrich
(Milwaukee, WI, USA) or Acros Organics distributed by
Fisher Scientific (Houston, TX, USA) and were used as
received. L-Valacyclovir was obtained as a gift from
GlaxoSmithKline. Caco-2 cells were obtained from Ameri-
can Type Culture Collection (ATCC, Rockville, MD,
USA). Dulbecco’s modified Eagle’s medium (DMEM) was
purchased from JRH Biosciences (Lenexa, KS, USA).
Heat-inactivated fetal bovine serum (FBS) was obtained
from Atlanta Biologicals (Norcross, GA, USA). L-
Glutamine solution, penicillin (10,000 U/ml)/strep-
tomycin (10,000 mg/ml) solution, and non-essential amino
acids solution (NEAA, 10 mM in 0.85% saline) were
purchased from Life Technologies (Grand Island, NY,
USA). Rat-tail collagen (type I) was obtained from Col-
laborative Biomedical/Becton Dickenson (Bedford, MA,
USA). Earle’s balanced salt solution (EBSS), Hanks’
balanced salt solution (HBSS), GlyPro, carnosine,
acyclovir, uridine, inosine, thymidine, dipyridamole, and
phloridzin were purchased from Sigma (St. Louis, MO,
USA). [3H]Carnosine (5 mCi/ml) was obtained from
Moravek Biochemicals (Brea, CA, USA) and [3H]uridine
(5 mCi/ml) was purchased from American Radiolabeled
Chemicals (St. Louis, MO, USA).
2.2.1.1.3. 2-(9-Purinemethoxy)ethyl-N-Boc-L-valinate
To a stirred solution of 9-(2-hydroxyethoxymethyl)-purine
(50 mg, 0.26 mmol) in DMF–dichloromethane (1:2, 9 ml)
2.2. Methods
was
added
N-(3-dimethylaminopropyl)-N9-ethylcar-
2.2.1. Synthesis of L-valacyclovir analogs
bodiimide (EDAC) (109 mg, 0.57 mmol), 4-(dimethyl-
amino)pyridine (DMAP) (32 mg, 0.26 mmol), and N-Boc-
L-valine (84 mg, 0.39 mmol) under argon. The solution
was stirred at 80 8C for 40 h. After evaporation of the
solvent, the residue was purified by flash chromatography
eluting with EtOAc to yield a clear oil (63 mg, 62%). 1H
NMR (CDCl3) d 9.20 (1H, s, H-4), 9.04 (1H, s, H-2), 8.28
(1H, s, H-8), 5.75 (1H, s, OCH2N), 5.05 (1H, d, J58.2
All reactions involving air-sensitive reagents were per-
formed under argon. All glassware was flame-dried prior to
use. The crude products were purified using flash chroma-
tography with 70–230 mesh size silica gel purchased from
1
Merck (West Point, PA, USA). H NMR spectra were
recorded on a 400-MHz instrument (Bruker DRX-400).
Chemical shifts are expressed in parts per million (d)
relative to the internal standard tetramethylsilane (TMS).
Positive fast-atom-bombardment mass spectrometry (FAB-
MS) was carried out using a Xenon gun on the matrices
thioglycerol/glycerol (TG/G), polyethylene glycol 300/
glycerol (PEG300/G), and nitrobenzyl alcohol (NBA) as
specified below. The mass spectra were recorded on a ZAB
HS mass spectrometer (VG Analytical, Manchester, UK)
Hz, NH), 4.30 (2H, m, CH2O), 4.22 (1H, m, CHNH), 3.81
]
(2H, m, CH2O), 2.08 (1H, m, CH(CH3)2), 1.45 (9H, s,
]
C(CH3)3), 0.95 (3H, d, J56.8 Hz, CH3), 0.87 (3H, d,
J56.8 Hz, CH3); FAB-MS (in TG/G) m/z: 394.2 (M1
H1), calculated molecular weight: 393.2.
2.2.1.1.4. 2-(9-Purinemethoxy)ethyl-L-valinate (2b) 2-
(9-Purinemethoxy)ethyl-N-Boc-L-valinate (50 mg, 0.13

G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
3
mmol) and 25% trifluoroacetic acid (TFA) in dichlorome-
thane (5 ml) were stirred at room temperature for 3 h.
After evaporation, azeotropic distillation, and lyophiliza-
tion, the product 2b was obtained as a colorless solid (55
mg, 100%). 1H NMR (CD3OD) d 9.18 (1H, s, H-4), 9.02
(1H, s, H-2), 8.70 (1H, s, H-8), 5.82 (2H, s, OCH2N), 4.46
(2H, m, CH2O), 4.32 (2H, m, CH2O), 3.90 (2H, m,
CH2O), 3.87 (1H, m, CHNH), 2.15 (1H, m, CH(CH3)2),
stirred at 60 8C for 1.5 h. The resulting solution was
neutralized with Amberlite IR-120 at the H1-form. The
solvent was removed in vacuo and the residue was purified
by flash chromatography eluting with EtOAc/EtOAc–
MeOH (9:1) to give a yellow oil (308 mg, 75%) which
1
crystallized upon drying in vacuo. H NMR (CDCl3) d
8.29 (1H, s, HC=N), 7.68–7.67 (2H, dd, J58.3 Hz,
H-4/H-7), 7.32 (2H, m, H-51H-6), 5.71 (2H, s, CH2N),
3.62 (2H, t, J54.5 Hz, CH2O), 3.52 (2H, t, J54.5 Hz,
CH2O); FAB-MS (in PEG300/G) m/z: 193.0977 (M1
H1), calculated molecular weight: 192.0899.
]
]
0.95 (6H, m, 23CH3); FAB-MS (vscan in PEG300/G)
m/z: 294.1566 (M1H1), calculated molecular weight:
293.1488.
2.2.1.2.5. 2-(1-Benzimidazolemethoxy)ethyl-N-Boc-L-
valinate 1-(2-Hydroxyethoxymethyl)-benzimidazole (115
mg, 0.60 mmol) was dissolved in DMF–dichloromethane
(1:2, 9 ml) by heating. To this solution were added EDAC
(252 mg, 0.90 mmol), DMAP (73 mg, 0.60 mmol), and
N-Boc-L-valine (195 mg, 0.90 mmol) under argon. The
mixture was stirred at 80 8C for 18 h. After evaporation of
the solvent, water (20 ml) was added and the mixture was
extracted with EtOAc (3320 ml). The organic phases
obtained were combined, dried over anhydrous MgSO4,
filtered, and evaporated in vacuo. The crude product was
purified by flash chromatography eluting with EtOAc to
yield a clear oil (100 mg, 85%). 1H NMR (CDCl3) d 8.00
(1H, s, HC=N), 7.83 (1H, m, H-7), 7.55 (1H, m, H-4),
7.35 (2H, m, H-51H-6), 5.61 (2H, s, CH2N), 4.99 (1H, d,
J57.4 Hz, NH), 4.27 (2H, t, J54.4 Hz, CH2O), 4.20 (1H,
2.2.1.2. Synthesis of 2-(1-benzimidazolemethoxy)ethyl-L-
valinate (3b)
2.2.1.2.1. 1-Acetylbenzimidazole A mixture of benz-
imidazole (2.00 g, 16.9 mmol), acetic anhydride (35 ml),
and acetic acid (50 ml) was heated with stirring at 140 8C
for 30 min, and the resulting solution was evaporated to
dryness. The product was obtained as orange crystals (2.70
g, 100%). 1H NMR (CDCl3) d 8.37 (1H, s, CH=N), 8.23
(1H, dd, Jortho 56.8 Hz, Jmeta 52.1 Hz, H-7), 7.80 (1H, dd,
Jortho 56.8 Hz, Jmeta 52.3 Hz, H-4), 7.41 (2H, m, H-51H-
6), 2.74 (3H, s, CH3); FAB-MS (in G) m/z: 161.1 (M1
H1), calculated molecular weight: 160.1.
2.2.1.2.2. 2-(Acetoxymethoxy)ethyl acetate Acetyl bro-
mide (50.0 g, 30.1 ml, 0.41 mol) was added dropwise to
1,3-dioxolane (26.6 g, 25.1 ml, 0.36 mol) maintaining the
temperature below 5 8C in an ice bath. The reaction
mixture was allowed to warm to room temperature and
then a solution of sodium acetate (50 g) in acetic acid (200
ml) was added to the mixture at 110 8C. The mixture was
further stirred at 110 8C for 1 h. Insoluble material (NaBr)
was filtered off. The filtrate was first distilled to remove
acetic acid, then the residue was fractionated in vacuo. The
fraction boiling at 114–118 8C/10 mm Hg was collected.
This yielded 48.6 g (77%) of the product as clear oil. 1H
NMR (CDCl3) d 5.27 (2H, s, OCH2O), 4.21 (2H, m,
OCH2), 3.81 (2H, m, OCH2), 2.08 (3H, s, CH3), 2.06 (3H,
s, CH3); FAB-MS (in DCI-NH3) m/z: 194 (M1NH14 ),
calculated molecular weight: 176.1.
2.2.1.2.3. 1-(2-Acetoxyethoxymethyl)-benzimidazole A
mixture of 2-(acetoxymethoxy)ethyl acetate (4.40 g, 25
mmol), acetyl benzimidazole (2.00 g, 12 mmol), p-
toluenesulfonic acid (297 mg, 1.6 mmol), and DMF (25
ml) was stirred at 140 8C for 18 h under argon and then
concentrated to dryness. The crude product was purified by
flash chromatography eluting with EtOAc/EtOAc–MeOH
(4:1) to yield a brown oil (2.14 g, 73%). 1H NMR (CDCl3)
d 7.98 (1H, s, HC=N), 7.82 (1H, m, H-7), 7.53 (1H, m,
H-4), 7.33 (2H, m, H-51H-6), 5.59 (2H, s, CH2N), 4.16
(2H, t, J54.7 Hz, CH2O), 3.61 (2H, t, J54.7 Hz, CH2O),
1.97 (3H, s, CH3); FAB-MS (in TG/G) m/z: 235.1 (M1
H1), calculated molecular weight: 234.1.
m, CHNH), 3.63 (2H, t, J54.4 Hz, CH2O), 2.05 (1H, m,
]
CH(CH3)2), 1.45 (9H, s, C(CH3)3), 0.92 (3H, d, J56.4
]
Hz, CH3), 0.83 (3H, d, J56.7 Hz, CH3); FAB-MS (in
TG/G) m/z: 392.2 (M1H1), calculated molecular weight:
391.2.
2.2.1.2.6.
2-(1-Benzimidazolemethoxy)ethyl-L-valinate
(3b) 2-(1-Benzimidazolemethoxy)ethyl-N-Boc-L-valinate
(160 mg, 0.41 mmol) was stirred with TFA (25% in
dichloromethane, 7.5 ml) at room temperature for 3 h. The
reaction mixture was evaporated, and lyophilization gave
1
the product 3b as colorless crystals (165 mg, 100%). H
NMR (CD3OD) d 9.46 (1H, s, HC=N), 7.98–7.90 (2H,
23m, H-71H-4), 7.68 (2H, m, H-51H-6), 5.98 (2H, s,
CH2N), 4.47 (2H, m, CH2O), 4.37 (2H, m, CH2O), 3.89
(2H, m, CH2O), 3.83 (1H, d, J54.3 Hz, CHNH), 2.16
]
(1H, m, CH(CH3)3), 0.99 (3H, d, J56.4 Hz, CH3), 0.97
]
(3H, d, J56.4 Hz, CH3); FAB-MS (in PEG300/G) m/z:
292.1661 (M1H1), calculated molecular weight:
291.1583.
2.2.1.3. Synthesis of 2-(7-azaindolemethoxy)ethyl-L-val-
inate (4b)
2.2.1.3.1. 1-(2-Benzoyloxyethoxymethyl)-7-azaindole
To a stirred suspension of sodium hydride (NaH) (60% in
mineral oil, 339 mg, 8.46 mmol) in dry DMF (30 ml) was
added a solution of 7-azaindole (1.00 g, 8.46 mmol) in
DMF (20 ml) at 0 8C under argon. After 5 min, 2-
(chloromethoxy)ethyl benzoate (1.49 ml, 8.46 mmol) was
added and the reaction mixture was stirred at room
2.2.1.2.4. 1-(2-Hydroxyethoxymethyl)-benzimidazole (3-
a) A mixture of 1-(2-acetoxyethoxymethyl)-benzimidaz-
ole (500 mg, 2.12 mol) and NH4OH (1 N, 50 ml) was

4
G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
temperature for 14 h. The solution was quenched with
water (20 ml) and the solvent was evaporated. The product
was dissolved in water (20 ml), extracted with EtOAc
(3350 ml), and purified by flash chromatography eluting
with petroleum ether (PE)–EtOAc (from 9:1 to 4:1) to
dd, J57.9 Hz, J55.0 Hz, H-5), 6.65 (1H, d, J53.7 Hz,
H-3), 5.75 (2H, s, OCH2N), 4.42 (2H, m, CH2O), 4.27
(2H, m, CH2O), 3.82 (1H, d, J54.5 Hz, CHNH), 3.76
]
(2H, m, CH2O), 2.15 (1H, m, CH(CH3)2), 0.95 (3H, d,
]
J56.4 Hz, CH3), 0.94 (3H, d, J56.4 Hz, CH3); FAB-MS
(in PEG300/G) m/z: 292.1661 (M1H1), calculated mo-
lecular weight: 291.1583.
1
give a clear oil (1.18 g, 47%). H NMR (CDCl3) d 8.33
(1H, dd, J54.7 Hz, J51.5 Hz, H-4), 7.98 (2H, dd, J57.0
Hz, J51.4 Hz, H-19), 7.90 (1H, dd, J57.8 Hz, J51.7 Hz,
H-6), 7.54 (2H, dt, J57.4 Hz, J51.3 Hz, H-39), 7.41 (2H,
t, J57.6 Hz, H-29), 7.35 (1H, d, J53.6 Hz, H-2), 7.10
(1H, dd, J57.8 Hz, J54.7 Hz, H-5), 6.51 (1H, d, J53.6
Hz, H-3), 5.78 (2H, s, OCH2N), 4.39 (2H, t, J54.7 Hz,
CH2O), 3.81 (2H, t, J54.7 Hz, CH2O); FAB-MS (in
TG/G) m/z: 297.1 (M1H1), calculated molecular weight:
296.1.
2.2.2. Characterization of physicochemical properties
The calculated partition coefficients in octanol/water
(cLogP) for acyclovir, L-valacyclovir, and other analogues
were calculated using a commercial product ACDlabs
(version 4.1) developed by Advanced Chemistry Develop-
ment, Canada.
2.2.1.3.2. 1-(2-Hydroxyethoxymethyl)-7-azaindole (4a)
1-(2-Benzoyloxyethoxymethyl)-7-azaindole (1.10 g, 3.7
mmol) was stirred in 40% aq. methylamine (110 ml) at
room temperature for 1.5 h. After evaporation of the
solvent, the residue was purified by flash chromatography
eluting with PE–EtOAc (from 1:1 to 1:2 to pure EtOAc).
The product was obtained as a clear oil (490 mg, 77%). 1H
NMR (CDCl3) d 8.27 (1H, dd, J54.8 Hz, J51.5 Hz,
H-4), 8.02 (2H, dd, J57.8 Hz, J51.5 Hz, H-6), 7.53 (1H,
d, J53.7 Hz, H-2), 7.16 (1H, dd, J57.8 Hz, J54.8 Hz,
H-5), 6.57 (1H, d, J53.7 Hz, H-3), 5.74 (2H, s, OCH2N),
3.60 (2H, t, J54.8 Hz, CH2O), 3.54 (2H, t, J54.9 Hz,
CH2O); FAB-MS (in PEG300/G) m/z: 193.0977 (M1
H1), calculated molecular weight: 192.0899.
2.2.3. Caco-2 cell cultures
Caco-2 cells were maintained under culture conditions
(in 150-cm2 T-flasks at 37 8C, 5% CO2 /95% air atmos-
phere with 95% humidity) as described by Gao et al.
(2000). The culture medium consisted of DMEM, 10%
FBS, 1% NEAA, 2 mM L-glutamine, 100 U/ml penicillin,
and 100 mg/ml streptomycin. The cells used for ex-
perimentation were seeded at a density of 13105 cells/
cm2 either on 24-well clusters or polycarbonate mem-
branes of Transwells coated with rat tail collagen and
grown for 21–28 days following the procedure described
previously (Gao et al., 2000). All cells used for the current
studies were between passages 24 and 50. [14C]Mannitol
flux assays were performed on each batch of Caco-2 cell
monolayers grown on Transwell membranes (n53). Cell
2.2.1.3.3. 2-[1-(7-Azaindolemethoxy)]ethyl-N-Boc-L-val-
inate To a stirred solution of 1-(2-hydroxyethoxymethyl)-
7-azaindole (290 mg, 1.51 mmol) in DMF–dichlorome-
thane (1:2, 15 ml) were added EDAC (636 mg, 3.32
mmol), DMAP (184 mg, 1.51 mmol), and N-Boc-L-valine
(492 mg, 2.26 mmol) under argon. The solution was
stirred at 80 8C for 16 h. After evaporation of the solvent,
the residue was purified by flash chromatography eluting
with PE–EtOAc (1:1) to yield 543 mg (92%) of the
monolayers with apparent permeability coefficients (Papp
)
of [14C]mannitol ,331027 cm/s were considered to be
appropriate for transport studies.
2.2.4. Uptake and transport properties of L-valacyclovir
analogs
2.2.4.1. Uptake studies
1
product as a clear oil. H NMR (CDCl3) d 8.32 (1H, d,
Uptake experiments were performed in triplicate using
Caco-2 cells grown in 24-well clusters at 37 8C in EBSS
(pH 6.0). Prior to the initiation of the uptake experiments,
the culture medium was removed and the cells were rinsed
three times followed by 20-min equilibration with HBSS
(pH 7.4). After removal of HBSS, uptake of radiolabeled
compounds was initiated by adding 150 ml of ‘assay
solution’ to the culture wells. For [3H]carnosine uptake
studies, the ‘assay solution’ contained [3H]L-carnosine (0.1
mM, 1 mCi/ml) without or with the L-valacyclovir analog
(10 mM) in EBSS. For [3H]uridine uptake studies, the
‘assay solution’ contained [3H]uridine (0.1 mM, 1 mCi/
ml) without or with the L-valacyclovir analog (10 mM) in
EBSS. Aliquots (20 ml) of each ‘assay solution’ were
taken prior to initiation of the experiment to determine the
total radioactivity added to each well. Cellular uptake was
terminated after a 20-min incubation for [3H]carnosine
experiments or after a 10-s incubation for [3H]uridine
J54.7 Hz, H-4), 7.91 (1H, d, J57.7 Hz, H-6), 7.33 (1H,
d, J53.5 Hz, H-2), 7.10 (1H, dd, J57.7 Hz, J54.7 Hz,
H-5), 6.53 (1H, d, J53.5 Hz, H-3), 5.73 (1H, s, OCH2N),
5.04 (1H, d, J58.4 Hz, NH), 4.23 (1H, m, CHNH), 4.20
]
(2H, m, CH2O), 3.67 (2H, m, CH2O), 2.05 (1H, m,
CH(CH3)2), 1.43 (9H, s, C(CH3)3), 0.91 (3H, d, J56.8
]
Hz, CH3), 0.82 (3H, d, J56.8 Hz, CH3); FAB-MS (in G)
m/z: 392.2 (M1H1), calculated molecular weight: 391.2.
2.2.1.3.4.
2-[1-(7-Azaindolemethoxy)]ethyl-L-valinate
(4b) 2-[1-(7-Azaindolemethoxy)]ethyl - N - Boc-L-valinate
(500 mg, 1.27 mmol) was stirred with TFA (25% in
dichloromethane, 20 ml) at room temperature for 3 h and
then subjected to evaporation, azeotropic distillation, and
lyophilization. The product 4b was obtained as a yellow
oil (520 mg, 100%). 1H NMR (CD3OD) d 8.31 (1H, dd,
J55.0 Hz, J51.3 Hz, H-4), 8.15 (1H, dd, J57.9 Hz,
J51.4 Hz, H-6), 7.58 (1H, d, J53.6 Hz, H-2), 7.26 (1H,

G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
5
experiments by removing the ‘assay solution’ from the
culture wells followed by rinsing with 331.5 ml ice-cold
HBSS. The cells in each well were then lysed with 0.3 M
NaOH (300 ml) and transferred to scintillation vials.
Radioactivity associated with the cells was determined by
liquid scintillation counting.
was interpreted as reflecting the binding of the test
compound to the transporter.
The apparent permeability coefficients (Papp) for the
L-valacyclovir analogs were calculated using Eq. (2):
P
app 5 (DQ/Dt)/(A 3 C0),
(2)
where DQ/Dt is the mass appearance rate of a test
compound in the receiver solution (mmol/s), which is the
slope of the linear portion of the graph when the cumula-
tive amount of a test compound transported is plotted
against time (Gao et al., 2000); A is the surface area of the
cell monolayers (cm2); and C0 is the initial concentration
of the test compound on the donor side (mmol/cm3).
2.2.4.2. Transport experiments
The transcellular transport studies were performed in
triplicate as described previously by our laboratory (Gao et
al., 2000). Briefly, 1.5 ml of the assay solution containing
the L-valacyclovir analog in EBSS (pH 6.0) was applied to
the donor compartment on the apical (AP) side of the
Caco-2 cell monolayers grown on Transwell membranes
and 2.6 ml of HBSS (pH 7.4) was added to the receiver
compartment on the basolateral (BL) side of the Caco-2
cell monolayers. Aliquots were taken from both sides (200
ml from receiver side, 20 ml from donor side) at various
times up to 180 min. The same volume of HBSS was
replenished on the receiver side after each sampling.
Samples were then transferred into ice-cold HPLC vials
containing 100 ml of 0.05 N HCl to minimize enzymatic
degradation and stored at 280 8C until analysis by HPLC.
2.2.7. Statistical analysis
All data are expressed as means6S.D. Statistical analy-
sis was performed using a paired Student’s t-test. A P
value #0.05 is considered to be statistically significant.
2.2.8. Aqueous stability studies
The aqueous stability of L-valacyclovir analogs was
determined by HPLC after storage in EBSS (pH 6.0) at
218 8C for 16 days or at 37 8C for 2 h.
2.2.5. HPLC analysis
HPLC analysis was performed using a Shimadzu gra-
dient system (Shimadzu, Tokyo, Japan) consisting of two
LC-10A pumps, an SCP-6 controller, and an SPD-10A UV
detector coupled with a PC computer. Chromatographic
separation of samples was carried out on a Vydac C18
3. Results
3.1. Synthesis of L-valacyclovir analogs
To examine the role of the guanine moiety of L-val-
acyclovir in binding to PepT1, a number of analogs lacking
possible hydrogen bonding sites were synthesized (Fig. 1).
Substitution of guanine with purine resulted in compound
2b, which lacks the most obvious hydrogen-donating (the
amino group) and hydrogen-accepting (the carbonyl oxy-
gen) groups. In the benzimidazole analog 3b and the
7-azaindole analog 4b, two additional possible hydrogen
bonding sites were removed.
To synthesize these L-valacyclovir analogs (2b–4b), we
initially needed to attach the 2-hydroxyethoxy methyl
moiety to the respective heteroaromatic system. Attach-
ment of the 2-hydroxyethoxy methyl moieties to benz-
imidazole was performed according to the procedures used
to synthesize acyclovir (Matsumoto et al., 1988). Benz-
imidazole was acetylated followed by reaction with 2-
(acetoxymethoxy)ethyl acetate and cleavage of the acetyl
group with ammonium hydroxide. This synthetic scheme
was optimized using larger amounts of catalyst, higher
temperature, and longer reaction time than those described
by Matsumoto et al. (1988) (see Section 2.2.1.2 for
details). A similar approach was attempted with purine, but
selective acetylation of N-9 was not possible. Instead, a
literature procedure using the N-9-alkyl substituted
purinesulfinamides (Hanna et al., 1994) was adopted, and
selective N-9 deprotonation was performed with lithium
˚
reversed-phase column (5 mM, 300 A, 250 mm L34.6
mm I.D.). Gradient elution was performed at a flow rate of
1.0 ml/min from 0 to 10% acetonitrile in water containing
0.1% trifluoroacetic acid as an ion-pairing agent. The UV
detection was performed at 254 nm. The temperature was
maintained at 4 8C during sample analysis.
2.2.6. Data analysis
The
%
uptake
of
radiolabeled
compounds
([3H]carnosine or [3H]uridine) was calculated using Eq.
(1):
% Uptake 5 (Q retained /Qtotal) 3 100
(1)
where Qretained is the radioactivity (dpm) retained in the
cell monolayer and Qtotal is the total radioactivity (dpm)
added to each well.
Comparisons were made between control cell mono-
layers, which were incubated with radiolabeled compound
alone, and the cell monolayers that were incubated with a
L-valacyclovir analog or a known inhibitor of PepT1
(GlyPro) or inhibitors of the nucleoside transporter
(dipyridamole and phloridzin). The uptake of the radio-
labeled compound in control cell monolayers was consid-
ered as 100%. If test compounds reduced the uptake of the
radiolabeled compound below control values, this result

6
G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
Table 1
Apparent permeability coefficients (Papp) of L-valacyclovir (1b) and its
analogs 2b–4b in the AP-to-BL direction across Caco-2 cell monolayers
in the absence or presence of GlyProa and their calculated partition
coefficients in octanol/water (cLogP)
Compound
cLogPb
P
app 3106 (cm/s)c
2GlyPro
1GlyPro
1b
2b
3b
4b
0.04
20.05
2.01
4.55 (60.25)
3.0 (60.22)
25.1 (60.06)
30.2 (61.94)
0.75 (60.19)
1.68 (60.16)
21.0 (61.36)
26.1 (62.06)
P,0.01
P,0.01
P,0.05
P,0.05
2.84
a Transport experiments of L-valacyclovir and its analogs (0.5 mM)
were conducted in the presence (5 mM) or absence of GlyPro at 37 8C in
EBSS as described in the Materials and methods section. Results are the
mean6S.D. of three separate experiments.
b cLogP from ACD (version 4.1).
c All the permeability coefficients for the ‘prodrugs’, L-valacyclovir
(1b) and its analogs (2b–4b) listed in this table were the ‘apparent’
values based only on the appearance of the ‘drugs’ in their hydrolyzed
forms (1a–4a) on the basolateral side of Caco-2 monolayers.
(4a.3a42a(1a), but with a decrease in the values of
cLogP by |2 logarithms.
Fig. 1. Analogs of acyclovir (1a) and L-valacyclovir (1b).
3.2. Inhibitory effect of L-valacyclovir (1b) and analogs
2b–4b on carnosine uptake in Caco-2 cell monolayers
diisopropylamide in dimethyl formamide followed by
addition of 2-(chloromethoxy)ethyl benzoate and cleavage
of the benzoyl group with methylamine. Substitution at the
N-9 position of the side chain was verified by a heteronu-
clear multiple bond correlation experiment, which showed
coupling from the NCH2O carbon to C-8 and C-6. Finally,
selective introduction of the 2-hydroxyethoxy methyl
moiety in the N-1 position of 7-azaindole was performed
by deprotonation with NaH, addition of 2-(chloro-
methoxy)ethyl benzoate, and deprotection with methyl-
amine (Beauchamp et al., 1985).
After introduction of the 2-hydroxyethoxy methyl moie-
ty, the hydroxy group was coupled to N-Boc-protected
L-valine using EDAC as coupling reagent. One equivalent
of DMAP was necessary to obtain a reasonable yield. The
Boc-group was removed with 25% trifluoroacetic acid in
dichloromethane.
The purine (2b), benzimidazole (3b), and 7-azaindole
(4b) L-valacyclovir analogs were all stable during synthesis
and upon storage. Chiral HPLC was performed on the
L-Val-benzimidazole analog 3b to exclude the possibility of
racemization during synthesis. The results of these experi-
ments confirmed that no significant racemization had taken
place (data not shown).
The lipophilic characteristics of the nucleoside analogs
of L-valacyclovir and acyclovir were estimated based on
cLogP values as shown in Tables 1 and 3. Reducing the
hydrogen-bonding sites in the guanine moiety tends to
increase the hydrophobicity of these molecules. The order
of hydrophobicity of L-valacyclovir and its analogs (1b–
4b) is: 4b.3b42b(1b. Acyclovir and its analogs (1a–
4a) show an order similar to their L-valine ester prodrugs
The affinities of PepT1 in Caco-2 cell monolayers for
the L-valacyclovir analogs were estimated by measuring
their ability to inhibit the uptake of [3H]carnosine (0.1
mM), a known substrate for this oligopeptide transporter.
As shown in Fig. 2, L-valacyclovir (1b) and its purine (2b),
benzimidazole (3b), and azaindole (4b) analogs (all at 10
mM) significantly inhibited [3H]carnosine uptake. The
extent of inhibition (575%) was comparable to that
produced by GlyPro (10 mM), which is a known inhibitor
of PepT1. Acyclovir (1a), however, showed very little
effect on [3H]carnosine uptake at a concentration of 5 mM.
One percent DMSO, which was used to solubilize
acyclovir, had no effect on [3H]carnosine uptake (data not
shown). The acyclovir analogs 2a, 3a, and 4a at 10 mM
also showed no inhibitory effect on [3H]carnosine uptake
in the Caco-2 cells (Fig. 2).
3.3. Transport of L-valacyclovir (1b) and its analogs
2b–4b across Caco-2 cell monolayers
In order to determine PepT1-mediated transport activity
for L-valacyclovir analogs 2b–4b, the AP-to-BL flux of
these prodrugs across Caco-2 cell monolayers was de-
termined in the absence or presence of GlyPro, a known
inhibitor of PepT1. For comparative purposes, the transport
of L-valacyclovir (1b) itself across Caco-2 cell monolayers
was also investigated. The results of these experiments are
presented in Fig. 3. Since the valine ester bonds of L-
valacylovir (1b) and its analogs (2b–4b) can potentially be
hydrolyzed by esterases upon exposure to the AP side of

G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
7
higher than that of 1a (Fig. 3, panel A, 12%) and 2a (Fig.
3, panel B, 9%). Secondly, the appearance of 3a (Fig. 3,
panel C) and 4a (Fig. 3, panel D) on the BL side of the
cell monolayers was not linear with time but instead
appeared to reach steady-state levels after |100 min of
incubation. Thirdly, the levels of 3a (Fig. 3, panel C) and
4a (Fig. 3, panel D) appearing on the BL side were only
slightly affected (from 50 to 42% in 3a and 42 to 39% in
4a) by inclusion of GlyPro with 3b and 4b, respectively,
on the AP side of the Caco-2 cell monolayers. Table 1
shows a summary of the Papp values of all compounds
tested. The differences between the Papp values in the
presence and the absence of GlyPro are statistically
significant (1b and 2b, P,0.01; 3b and 4b, P,0.05).
3.4. Stability of L-valacyclovir (1b) and its analogs 2b–
4b upon exposure to the AP side of Caco-2 cell
monolayers
Fig. 2. Effects of acyclovir and L-valacyclovir analogs on [3H]carnosine
uptake in Caco-2 cells. [3H]Carnosine (0.1 mM, 5 mCi/ml) uptake into
Caco-2 cell monolayers was studied in the presence or absence of GlyPro,
acyclovir (1a) and its analogs 2a–4a, and L-valacyclovir (1b) and its
analogs 2b–4b at the indicated concentrations. Incubations were con-
ducted for 20 min at 37 8C in EBSS as described in the Materials and
methods section. Results are the means6S.D. of quadruplicate determi-
nations from one of two independent experiments. The asterisks (*)
indicate that the reduction from the control experiment (no addition) was
statistically significant (P,0.05) according to paired Student’s t-test.
During the transport experiments described above, the
stabilities of L-valacyclovir (1b) and its analogs 2b–4b on
the AP side of Caco-2 cell monolayers were monitored.
Although the purine (2b), benzimidazole (3b), and 7-
azaindole (4b) analogs were all stable in EBSS (pH 6.0)
and HBSS (pH 7.4) at 37 8C up to 2 h (data not shown),
exposure of these analogs in EBSS (pH 6.0) to the AP side
of Caco-2 cells resulted in their rapid hydrolysis to the
corresponding acyclovir analogs 2a–4a (Table 2). Such a
rapid hydrolysis did not occur when L-valacyclovir and its
analogs 2b–4b were directly added to the HBSS on the BL
side of Caco-2 cell monolayers (data not shown). It is
interesting to compare the stability of analogs 2b–4b to
that of L-valacyclovir (1b) (Table 2). The order of stability
was 1b.2b43b44b when these compounds were
incubated in EBSS (pH 6.0) on the AP side of Caco-2 cell
monolayers.
Caco-2 cell monolayers, HPLC analytical methods were
developed that permit the detection of both the ‘prodrugs’
(1b–4b) and the ‘drugs’ (1a–4a). As shown in Fig. 3,
incubation of L-valacyclovir and its analogs (1b–4b) on
the AP side of Caco-2 cell monolayers resulted in a
time-dependent appearance of acyclovir (1a) (panel A) and
its analogs (2a–4a) (panels B–D), respectively, on the BL
side of the cell monolayers. No detectable levels of the
‘prodrugs’ (1b–4b) were observed on the BL side of these
cell monolayers. The amount of 1a appearing on the BL
side after application of 1b to the AP side exhibited a short
lag time (|20 min) followed by a linear appearance of 1a.
At 180 min the level of 1a on the BL side reached |12%
of the initial amount of 1b that was added on the AP side
of the cell monolayer. The appearance of acyclovir (1a) on
the BL side of the cell monolayer was significantly
reduced (from 12 to 2%) when GlyPro, a known inhibitor
of PepT1, was included along with L-valacyclovir (1b) on
the AP side (Fig. 3, panel A). When the purine analog 2b
was applied to the AP side of the Caco-2 cell monolayers,
the appearance of 2a on the BL side was also linear with
time (9% at 180 min) and partially inhibited by inclusion
of GlyPro on the AP side (Fig. 3, panel B). However, the
profiles for appearance of 3a (from 3b) and 4a (from 4b)
on the BL side of the Caco-2 cell monolayers were very
different from those observed for 1a (from 1b) and 2a
(from 2b) in three respects. First, the levels of 3a (Fig. 3,
panel C, 50%) and 4a (Fig. 3, panel D, 42%) appearing on
the BL side of the cell monolayers after application of 3b
and 4b, respectively, to the AP side were significantly
3.5. Transport of acyclovir (1a) and its analogs (2a–
4a) across Caco-2 cell monolayers
To better understand the ‘apparent’ transport characteris-
tics of L-valacyclovir (1b) and its analogs 2b–4b (Fig. 3),
we determined the transport characteristics of acyclovir
(1a) and its analogs (2a–4a) in the AP-to-BL direction
across Caco-2 cell monolayers. Consistent with their
reported oral bioavailability (de Miranda and Blum, 1983),
the Papp value for acyclovir (1a) (Table 3) was found to be
significantly lower than that for L-valacyclovir (1b) (Table
1). In contrast, the values for the acyclovir analogs 2a–4a
(Table 3) were significantly greater than the Papp values for
their L-valine esters 2b–4b (Table 1). The relative Papp
values for the benzimidazole and azaindole analogs of
acyclovir (4a53a.2a41a) are consistent with their
lipophilic characteristics as estimated by cLogP values
(Table 3).
Since the Papp values for the more lipophilic acyclovir

8
G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
Fig. 3. Time course of the appearance of acyclovir (1a) or its analogs 2a–4a on the BL side when L-valacyclovir (1b) or its analogs 2b–4b were applied
on the AP side of Caco-2 cell monolayers. Experiments were done in the presence (h) or absence (♦) of 5 mM GlyPro using a 0.5-mM concentration of
L-valacyclovir (1b) or its analogs 2b–4b at 37 8C in EBSS. Results are the means6S.D. of three separate experiments. Panel A: % acyclovir (1a) from
L-valacyclovir (1b); panel B: % 2a from 2b; panel C: % 3a from 3b; panel D: % 4a from 4b.
analogs 3a and 4a were quite high, we conducted experi-
ments to show that their permeation across Caco-2 cells
was not mediated by a transporter system. One possibility
is that the acyclovir analogs 2a–4a are themselves sub-
strates for PepT1. However, this possibility can be ruled
out because inclusion of GlyPro has no effect on the
transport of 2a–4a in the AP-to-BL direction across Caco-
2 cell monolayers (Table 3).
Another possibility is that the acyclovir analogs 2a–4a
are substrates for a Na1-independent nucleoside transpor-
ter that has been reported by Ward and Tse (1999) to be
present in Caco-2 cells. To evaluate this possibility an
Table 2
Time course of disappearance of L-valacyclovir (1b) and its analogs 2b–4b and the appearance of acyclovir (1a) and its analogs 2a–4a on the AP side of
Caco-2 cells
Time
(min)
% Of total 1b–4b (0.5 mM) added to AP sidea
1b
1a
2b
2a
3b
3a
4b
4a
0
30
60
120
180
100
84
82
76
60
0
2
6
16
26
100
99
84
55
32
0
100
60
25
5
0
27
43
41
38
100
23
0
–
–
0
46
49
38
39
0.5
2.0
3.7
5.4
2
a The amounts of L-valacyclovir (1b) and its analogs (2b–4b) and acyclovir (1a) and its analogs (2a–4a) on the AP side of Caco-2 cell monolayers were
monitored with times up to 180 min. The incubation medium consisted of EBSS (pH 6.0) as described in the Materials and methods section. Results are
expressed as the percentage of the total amounts of L-valacyclovir (1b) and its analogs (2b–4b) added to the AP incubation mixture.

G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
9
Table 3
Table 4
Apparent permeability coefficients (Papp) of acyclovir (1a) and its analogs
2a–4a in the AP-to-BL direction across Caco-2 cell monolayers in the
absence or presence of GlyPro and their calculated partition coefficients
in octanol/water (cLogP)
Apparent permeability coefficients (Papp) of acyclovir analogs 2a–4a in
the AP-to-BL direction across Caco-2 cell monolayers in the absence and
presence of nucleoside transporter inhibitors
Compound
P
app 3106 (cm/s)a
Compound
P
app 3106 (cm/s)a
cLogPb
2Inhibitors
1Inhibitors
2GlyPro
1GlyPro
2a
3a
4a
12.8 (60.21)
65.3 (62.28)
82.5 (66.24)
10.2 (60.73)
56.9 (61.16)
77.6 (68.8)
1a
2a
3a
4a
0.43 (60.01)
9.17 (60.79)
39.8 (60.06)
47.9 (60.79)
–
21.76
21.85
0.21
9.42 (60.96)
42.1 (61.5)
44.4 (62.0)
a Transport studies with acyclovir analogs 2a–4a (0.5 mM) were
conducted in the presence or absence of 10 mM dipyridamole and 1 mM
phloridzin at 37 8C in EBSS (pH 6.0) as described in the Materials and
methods section. Results are the mean6S.D. of three separate experi-
ments.
1.37
a Transport experiments of acyclovir (1a) and its analogs 2a–4a (0.5
mM) were conducted in the presence (5 mM) or absence of GlyPro at
37 8C in EBSS (pH 6.0) as described in the Materials and methods
section. Results are the mean6S.D. of three separate experiments.
b cLogP from ACD (version 4.1).
suggest that acyclovir analogs 2a–4a may be substrates for
the nucleoside transporter, which could then be in part
involved in their transcellular transport across Caco-2 cell
monolayers. However, when the transcellular transport
characteristics of the acyclovir analogs 2a–4a were de-
termined in the presence of dipyridamole and phloridzin,
these inhibitors of the nucleoside transporter had no effect
on their permeation (Table 4).
assay for the nucleoside transporter using [3H]uridine was
developed. The uptake of [3H]uridine into Caco-2 cells
from the AP side is time-dependent, reaching steady-state
at 4 min. The uptake of [3H]uridine into Caco-2 cells can
be inhibited by dipyridamole and phloridzin, known in-
hibitors of the nucleoside transporters (Plagemann and
Aran, 1990), and by a purine (inosine) and a pyrimidine
(thymidine) nucleoside (Fig. 4). Acyclovir (1a) does not
inhibit the uptake of [3H]uridine, whereas its analogs
2a–4a exhibit inhibitory activities comparable to those of
dipyridamole and phloridzin, inosine and thymidine on
[3H]uridine uptake. The order of potency for inhibition of
[3H]uridine uptake was 4a43a52a41a. These results
4. Discussion
L-Valacyclovir (1b) was designed as a prodrug of
acyclovir (1a) in an attempt to improve the oral bioavail-
ability of this antiviral agent (Beauchamp et al., 1992;
Crooks and Murray, 1994). In subsequent studies, several
laboratories (Ganapathy et al., 1998; Sinko and Balimane,
1998; de Vrueh et al., 1998) demonstrated that L-val-
acyclovir (1b) was a substrate for PepT1 in the intestinal
mucosa, which provided a rational explanation for the
good oral bioavailability of this hydrophilic prodrug. In
retrospect, L-valacyclovir (1b) is an example of an almost
‘perfect’ prodrug for enhancing oral absorption. First, L-
valacyclovir (1b) is stable in the lumen of the intestine
(Sinko and Balimane, 1998), and it is relatively stable to
the enzymes on the surface of intestinal mucosal cells (e.g.
Caco-2 cells; Table 2). Secondly, it is a substrate for
PepT1, which mediates its uptake into intestinal mucosal
cells (e.g. Caco-2 cells; Figs. 2 and 3). Thirdly, it appears
to be a substrate for intracellular esterases that catalyze the
bioconversion of L-valacyclovir (1b) to acyclovir (1a).
Finally, acyclovir (1a), not L-valacyclovir (1b), appears in
blood (Soul-Lawton et al., 1995) and on the BL side of
intestinal mucosal cells (e.g. Caco-2 cells; Fig. 3). These
transport and metabolic stability characteristics of L-val-
acyclovir (1b) are summarized in Fig. 5 (panel A).
Fig. 4. Effect of known inhibitors of the nucleoside transporter
(dipyridamole and phloridzin), nucleosides (inosine and thymidine) and
acyclovir (1a) and its analogs 2a–4a on uptake of [3H]uridine in Caco-2
cell monolayers. [3H]Uridine (0.1 mM, 5 mCi/ml) uptake into Caco-2
cell monolayers was studied in the absence or presence of these potential
inhibitors at the concentrations indicated. Incubations were conducted for
10 s at 37 8C in EBSS as described in the Materials and methods section.
Results are the means6S.D. of quadruplicate determinations from one of
two independent experiments. The asterisks (*) indicate that the reduction
from the control experiment (without addition) was statistically significant
(P,0.05) according to paired Student’s t-test.
In an attempt to better understand the role of the guanine
moiety in the binding of L-valacyclovir to PepT1, we
synthesized in this study a series of analogs having
different heteroaromatic rings (purine, benzimidazole, and
7-azaindole) replacing the guanine moiety. The biological

10
G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
Fig. 5. Model for the transport and metabolic stability characteristics of L-valacyclovir (1b) and its analogs (2b–4b).

G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
11
Fig. 5. (continued)
properties of the purine (2a–b), benzimidazole (3a–b),
and 7-azaindole (4a–b) analogs were characterized and the
pathways by which they permeate across Caco-2 cell
monolayers were elucidated.
(1b) has many other characteristics (e.g. luminal stability,
intracellular instability) that allow it to effectively deliver
acyclovir (1a) across the intestinal mucosa.
The transport characteristics of the purine analog 2b are
similar to those of L-valacyclovir (1b), whereas the trans-
port characteristics of the benzimidazole (3b) and 7-azain-
dole (4b) analogs are very different. The purine analog 2b
delivers the acyclovir analog 2a to the BL side of a Caco-2
monolayer in a time-dependent manner that is partially
inhibited by GlyPro. However, the total amount of 2a
delivered to the BL side after application of 2b to the AP
side of the cell monolayer is less than that observed with
acyclovir (1a) delivered from L-valacyclovir (1b). This
may suggest that the purine analog 2b is a poorer substrate
for PepT1; thus, transport of the purine analog 2b by
PepT1 is less efficient than that of L-valacyclovir (1b)
and/or that the bioconversion of the 2b to 2a intracellular-
ly is less efficient than the bioconversion of L-valacyclovir
(1b) to acyclovir (1a). The latter possibility may explain
the low recovery of 2a on the BL side of the cell
monolayers when total mass balance on both sides were
compared, i.e. the purine analog 2b may be trapped in the
cell monolayers. It is also worth noting that, in the
presence of GlyPro, the purine analog 2b delivers more 2a
than L-valacyclovir (1b) delivers acyclovir (1a). This
would suggest that the permeation of analog 2a across the
Caco-2 monolayers is greater than acyclovir (1a), which
The data shown in Fig. 2 clearly show that all of the
L-valine esters (2b–4b) of the acyclovir analogs (2a–4a)
bind to PepT1. The data presented in Fig. 2 also clearly
show that acyclovir (1a) and its analogs 2a–4a do not bind
to this transporter. The purine (2b), benzimidazole (3b),
and 7-azaindole (4b) analogs of L-valacyclovir (1b) were
designed to have reduced sites for potential hydrogen-
bonding interactions. It is also noteworthy that these
heteroaromatic moieties in 2b–4b exhibited quite different
lipophilicity characteristics (order of hydrophobicity:
4a(3a..2a.1a; Table 3). The PepT1 binding data
shown in Fig. 2 would suggest that the heteroaromatic
moieties in L-valacyclovir (1b) and its analogs 2b–4b are
not contributing significantly to their binding to this
transporter. These results would suggest that the primary
recognition sites for PepT1 are the amino group and the
carbonyl group provided by the valine. Finally, it is
important to mention that simply binding the analogs
2b–4b to PepT1 (Fig. 2) does not ensure that these
prodrugs will be as effective at delivering the acyclovir
analogs 2a–4a as L-valacyclovir (1b) is at delivering
acyclovir (1a) across a cell culture model (Caco-2 cells) of
the intestinal mucosa. As described above, L-valacyclovir

12
G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
was confirmed in separate studies (Table 3). The purine
analog 2a was shown to have affinity for the nucleoside
transporter in Caco-2 cells (Fig. 4), which might contribute
in part to its higher permeation across Caco-2 cell mono-
layers than that of acyclovir (1a). However, inclusion of
nucleoside transporter inhibitors seemed to have limited
effect on permeation of 2a across Caco-2 cell monolayers.
Based on its cLogP value, 2a seems to be hydrophilic
rather than lipophilic and transcellular passive diffusion is
unlikely for 2a. These observations suggest that transport
of 2a across cell monolayers may also involve a transpor-
ter system in the Caco-2 cells that has not yet been
identified. In summary, the purine analog 2b appears to
deliver 2a across Caco-2 cell monolayer by a PepT1-
dependent mechanism mediating the uptake of 2b (path-
way a, Fig. 5B) and a nucleoside transporter-dependent
mechanism mediating the uptake of 2a (pathway c, Fig.
5B). In addition, a third but unknown transporter system
may also be involved in transport of 2a formed by
hydrolysis of 2b on the AP surface of Caco-2 cells.
The transport characteristics of the benzimidazole (3b)
and 7-azaindole (4b) analogs are similar to each other but
very different from that of L-valacyclovir (1b). Analogs 3b
and 4b deliver their acyclovir analogs 3a and 4a, respec-
tively, across Caco-2 cell monolayers in a time-dependent
manner that rapidly reaches steady-state and is only
marginally affected by the inclusion of GlyPro on the AP
side of the monolayers (Fig. 3C,D). The total amounts of
3a and 4a that ultimately appear on the BL side of the cell
monolayers when 3b and 4b are applied to the AP side are
significantly greater than the amounts of acyclovir (1a)
delivered from L-valacyclovir (1b). This enhanced delivery
of 3a and 4a can be attributed to the instability of 3b and
4b to esterases on the AP side of Caco-2 cells (Table 2)
and the excellent permeation characteristics of 3a and 4a
across Caco-2 cell monolayers (Table 3). Like 2a, the
acyclovir analogs 3a and 4a also appear to be substrates
for the nucleoside transporter (Fig. 5), which may also
contribute to this enhanced permeation of 3a and 4a across
Caco-2 cell monolayers. In summary, the benzimidazole
(3b) and 7-azaindole (4b) analogs appear to deliver 3a and
4a, respectively, across Caco-2 cell monolayers primarily
via their rapid hydrolysis by esterases on AP side of
Caco-2 cells and then the rapid passive diffusion of 3a and
4a across the cell monolayers (Fig. 5C). There may also be
a small contribution by a PepT1-dependent mechanism
mediating the uptake of 3b and 4b (pathway a, Fig. 5C)
and/or a nucleoside transporter-dependent mechanism
mediating the uptake of 3a and 4a (pathway c, Fig. 5C),
but these appear to be minor pathways.
hydrogen-bonding sites in the guanine moiety of L-val-
acyclovir may contribute to its substrate activity for
PepT1. Reducing these hydrogen-bonding sites in the
guanine moiety of L-valacyclovir may affect the transport
efficiency of the molecule via PepT1. On the other hand,
the potential hydrogen-bonding sites in the guanine moiety
of acyclovir play a significant role in the permeation
characteristics of this molecule across Caco-2 cell mono-
layers. Reducing these potential hydrogen-bonding sites in
the guanine moiety of acyclovir as shown in a series of
acyclovir analogs in this study greatly enhances the
permeation of these molecules across Caco-2 cell mono-
layers due to their increased lipophilicity as well as the
substrate activity of these molecules for the nucleoside
transporter in Caco-2 cells.
Acknowledgements
This work was supported by grants from the National
Institutes of Health and GlaxoSmithKline. The authors
would like to thank Richard Lloyd (GlaxoSmithKline,
Ware, UK) for providing cLogP data for acyclovir, L-
valacyclovir and its analogues.
References
Adibi, S.A., 1997. The oligopeptide transporter (Pept-1) in human
intestine: biology and function. Gastroenterology 113, 332–340.
Bailey, P.D., Boyd, C.A.R., Bronk, J.R., Collier, I.D., Meredith, D.,
Morgan, K.M., Temple, C.S., 2000. How to make drugs orally active:
a substrate template for peptide transporter PepT1. Angew. Chem., Int.
Ed. Engl. 39 (3), 505–508.
Beauchamp, L.M., Dolmatch, B.L., Schaeffer, H.J., Collins, P., Bauer,
D.J., Keller, P.M., Fyfe, J.A., 1985. Modifications on the heterocyclic
base of acyclovir: syntheses and antiviral properties. J. Med. Chem. 28
(8), 982–987.
Beauchamp, L.M., Orr, G.F., de Miranda, P., Burnette, T., Krenitsky,
T.A., 1992. Amino acid ester prodrugs of acyclovir. Antiviral Chem.
Chemother. 3 (3), 157–164.
Crooks, R.J., Murray, A., 1994. Valaciclovir—a review of a promising
new antiherpes agent. Antiviral Chem. Chemother. 5 (S1), 31–37.
Dantzig, A.H., 1997. Oral absorption of b-lactams by intestinal peptide
transport proteins. Adv. Drug Deliv. Rev. 23, 63–76.
Ganapathy, M.E., Huang, W., Wang, H., Ganapathy, V., Leibach, F.H.,
1998. Valacyclovir: a substrate for the intestinal and renal peptide
transporters PEPT1 and PEPT2. Biochem. Biophys. Res. Commun.
246 (2), 470–475.
Gao, J., Hugger, E.D., Beck-Westermeyer, M.S., Borchardt, R.T., 2000.
Estimating intestinal mucosal permeation of compounds using Caco-2
cell monolayers. In: Enna, S.J., Williams, M., Ferkany, J.W., Kenakin,
T., Porsolt, R.D., Sullivan, J.P. (Eds.), Current Protocols in Pharma-
cology. Wiley, New York, pp. 721–723.
In conclusion, the results presented in this study indicate
that the guanine moiety of L-valacyclovir does not contrib-
ute significantly to its recognition at PepT1, and the
primary recognition sites of this molecule for PepT1 are
the amino group and the carbonyl group provided by the
valine ester of L-valacyclovir. However, the potential
Han, H., de Vrueh, R.L.A., Rhie, J.K., Covitz, K.-M.Y., Smith, P.L., Lee,
´
C.-P., Oh, D.-M., Sadee, W., Amidon, G.L., 1998. 59-Amino acid esters
of antiviral nucleosides, acyclovir, and AZT are absorbed by the
intestinal PEPT1 peptide transporter. Pharm. Res. 15 (8), 1154–1159.
Hanna, N.B., Bhattacharya, B.K., Robins, R.K., Avery, T.L., Revankar,
G.R., 1994. Sulfinosine congeners: synthesis and antitumor activity in

G.M. Friedrichsen et al. / European Journal of Pharmaceutical Sciences 16 (2002) 1–13
13
mice of certain N9-alkylpurines and purine ribonucleosides. J. Med.
Chem. 37, 177–183.
Hidalgo, I.J., Raub, T.J., Borchardt, R.T., 1989. Characterization of the
human colon carcinoma cell line (Caco-2) as a model system for
intestinal epithelial permeability. Gastroenterology 96, 736–749.
Matsumoto, H., Kaneko, C., Yamada, K., Takeuchi, T., Mori, T., Mizuno,
Soul-Lawton, J., Seaber, E., On, N., Wootton, R., Rolan, P., Posner, J.,
1995. Absolute bioavailability and metabolic disposition of valaci-
clovir, the L-valyl ester of acyclovir, following oral administration to
humans. Antimicrob. Agents Chemother. 39 (12), 2759–2764.
Temple, C.S., Stewart, A.K., Meredith, D., Lister, N.A., Morgan, K.M.,
Collier, I.D., Vaughan-Jones, R.D., Boyd, C.A.R., Bailey, P.D., Bronk,
J.R., 1998. Peptide mimics as substrates for the intestinal peptide
transporter. J. Biol. Chem. 273 (1), 20–22.
Walker, E., Kissel, T., Amidon, G., 1996. The intestinal peptide carrier: a
potential transport system for small peptide derived drugs. Adv. Drug
Deliv. Rev. 20, 33–58.
Ward, J.L., Tse, C.M., 1999. Nucleoside transport in human colonic
epithelial cell lines: evidence for two Na1-independent transport
systems in T84 and Caco-2 cells. Biochim. Biophys. Acta 1419 (1),
15–22.
Y.,
1988.
A
convenient
synthesis
of
9-(2-hydroxy-
ethoxymethyl)guanine (Acyclovir) and related compounds. Chem.
Pharm. Bull. 36 (3), 1153–1157.
de Miranda, P., Blum, M.R., 1983. Pharmacokinetics of acyclovir after
intravenous and oral administration. Review. J. Antimicrob.
Chemother. 12 (Suppl. B), 29–37.
Plagemann, P.G.W., Aran, J.M., 1990. Na1-dependent, active nucleoside
transport in mouse spleen lymphocytes, leukaemia cells, fibroblasts
and macrophages, but not in equivalent human or pig cells:
dipyridamole enhances nucleoside salvage by cells with both active
and facilitated transport. Biochim. Biophys. Acta 1025 (1), 32–42.
Sinko, P.J., Balimane, P.V., 1998. Carrier-mediated intestinal absorption of
valacyclovir, the L-valyl ester prodrug of acyclovir: 1. Interactions
with peptides, organic anions and organic cations in rats. Biopharm.
Drug Dispos. 19 (4), 209–217.
de Vrueh, R.L.A., Smith, P.L., Lee, C.-P., 1998. Transport of L-valine-
acyclovir via the oligopeptide transporter in the human intestinal cell
line, Caco-2. J. Pharmacol. Exp. Ther. 286, 1166–1170.