A methylester of the glucuronide prodrug DOX-GA3 for improvement of tumor-selective chemotherapy

Article abstract of DOI:10.1016/j.bcp.2004.08.004

The glucuronide prodrug of doxorubicin, DOX-GA3, can be selectively activated in tumors by extracellular human β-glucuronidase, resulting in a better therapeutic index than doxorubicin. DOX-GA3, however, is rapidly excreted by the kidney. We hypothesized

Full text of DOI:10.1016/j.bcp.2004.08.004

Biochemical Pharmacology 68 (2004) 2273–2281
www.elsevier.com/locate/biochempharm
A methylester of the glucuronide prodrug DOX-GA3 for improvement
of tumor-selective chemotherapy
Michelle de Graafa,1, Tapio J. Nevalainenb,1,2, Hans W. Scheerenb,
Herbert M. Pinedoa, Hidde J. Haismac, Epie Bovena,
*
aDepartment of Medical Oncology, VU University Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands
bDepartment of Organic Chemistry, University of Nijmegen, Toernooiveld, 6525 ED Nijmegen, The Netherlands
cDepartment of Therapeutic Gene Modulation, University Center for Pharmacy, University of Groningen,
Ant. Deusinglaan 1, 9713 AV Groningen, The Netherlands
Received 22 April 2004; accepted 5 August 2004
Abstract
The glucuronide prodrug of doxorubicin, DOX-GA3, can be selectively activated in tumors by extracellular human b-glucuronidase,
resulting in a better therapeutic index than doxorubicin. DOX-GA3, however, is rapidly excreted by the kidney. We hypothesized that slow
release of DOX-GA3 from its methylester, DOX-mGA3, by esterase activity in blood would result in improved circulation half-life (t1/2
)
of DOX-GA3. DOX-mGA3 was synthesized more efficiently with an overall yield of 60% as compared to 37% in the case of DOX-GA3.
We showed that DOX-mGA3 was enzymatically converted to DOX-GA3 with a t1/2 of approximately 0.5 min in mouse plasma to 2.5 h in
human plasma, which was in agreement with differences in esterase activity between species. DOX-mGA3, similar to DOX-GA3, was at
least 37-fold less potent than the parent drug doxorubicin in growth inhibition of four different human malignant cell lines in vitro.
Incubation of OVCAR-3 cells with DOX-mGA3 in combination with an excess of human b-glucuronidase (0.05 U mLꢀ1) resulted in a
similar growth inhibition to that of doxorubicin. Intravenous administration of DOX-mGA3 in FMa-bearing mice resulted in an area under
the concentration versus time curve (AUC) of DOX-GA3 in tumor and most normal tissues that was 2.5- to 3-fold higher than after the
same dose of DOX-GA3 itself. In tumor tissue, this was accompanied by a 2.7-fold increase in the AUC of doxorubicin from DOX-mGA3
than from DOX-GA3. In conclusion, an advantage of DOX-mGA3 over DOX-GA3 is that this prodrug can be produced with a higher
yield. Another important advantage is the improved pharmacokinetics of the lipophilic DOX-mGA3 as compared to that of the hydrophilic
DOX-GA3. This effect may even be more pronounced in man, because of the lower plasma esterase activity than measured in mice.
# 2004 Elsevier Inc. All rights reserved.
Keywords: Cancer chemotherapy; Doxorubicin; Esterases; Glucuronide; b-Glucuronidase; Prodrug
1. Introduction
fact, such metabolites could be useful as prodrugs. These
naturally occurring glucuronide prodrugs are less toxic
than their parent compounds due to increased hydrophili-
city, resulting in decreased cellular uptake [1]. They can be
selectively activated in the tumor by b-glucuronidase,
which is released in necrotic tumor areas. An example
of a naturally occurring glucuronide prodrug is the meta-
bolite of aniline mustard. Efficacy of treatment with aniline
mustard has directly been related to the level of b-glucur-
onidase in mouse tumors [2]. A clinical trial of aniline
mustard in patients with advanced cancer has shown a
relation between b-glucuronidase activity in aspirate and
imprint preparations of tumor cells, which was assessed by
a cytochemical technique, and patient’s response [3].
Success of that trial was limited, since very high tissue
levels of b-glucuronidase were required. The need for
An important metabolic route in the liver is the genera-
tion of drug glucuronides, which are biologically or che-
mically less reactive and exhibit higher polarity and
excretability than the corresponding parent aglycones. In
Abbreviations: AUC, area under the concentration versus time curve;
BNNP, bis(4-nitrophenyl)phosphate; BSA, bovine serum albumin; DMSO,
dimethyl sulphoxide; HPLC, high-performance liquid chromatography;
MTD, maximum tolerated dose; PBS, phosphate-buffered saline;
pNPAcp-nitrophenyl acetate
* Corresponding author. Tel.: +31 20 4444336; fax: +31 20 4444355.
E-mail address: e.boven@vumc.nl (E. Boven).
1
Both authors have equally contributed to this work.
2
Present address: Department of Pharmaceutical Chemistry, University
of Kuopio, P.O. Box 1627, Kuopio 70211, Finland.
0006-2952/$ – see front matter # 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.bcp.2004.08.004

2274
M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
these very high levels might be attributed to slow activation
of the aniline mustard glucuronide by the enzyme. Syn-
thetic glucuronide prodrugs that are more efficiently
hydrolyzed have been developed for enzyme prodrug
therapy (for review see [4]).
prodrug therapy. The methyl ester of other methylated
glucuronide prodrugs has been reported to be hydrolyzed
by carboxylesterases [12,13]. When administered in vivo,
the methyl group of DOX-mGA3 might be removed by
carboxylesterase activity in plasma to yield the original
DOX-GA3 prodrug, which could in turn be activated by
human b-glucuronidase (Fig. 1). We hypothesized that
slow release of DOX-GA3 after administration of DOX-
mGA3 might result in improved pharmacokinetics of the
glucuronide prodrug in vivo. Therefore, the use of the
lipophilic DOX-mGA3, which can be considered a pro-
prodrug, might allow lower doses to be administered to
achieve the same therapeutic effect.
We have synthesized glucuronide prodrugs of daunor-
ubicin and doxorubicin, designated daunorubicin-GA3
(DNR-GA3) and doxorubicin-GA3 (DOX-GA3), respec-
tively [5], which contain a self-eliminating p-aminobenzy-
loxycarbonyl spacer. Both prodrugs can be completely
converted to doxorubicin by human b-glucuronidase in
vitro and they were shown to be relatively non-toxic in
vivo. The maximum tolerated dose (MTD) of DOX-GA3 in
tumor-bearing nude mice, based on the occurrence of 10%
weight loss, was approximately 500 mg kgꢀ1 i.v., compared
to 8 mg kgꢀ1 for doxorubicin [6]. Administration of glu-
curonide prodrugs, including DOX-GA3, to nude mice
bearing human ovarian cancer xenografts at half of the
MTD has resulted in concentrations of the parent drug in
plasma that were significantly lower, while the concentra-
tions of the parent drug in tumor tissue were higher than
after administration of the parent drug itself [6–8]. The
increased concentration of the parent drug in tumor tissue
after prodrug administration is beneficial as it enhances the
therapeutic effect [7,9–11]. The increased efficacy was
specifically observed in larger tumors containing a higher
proportion of necrotic areas.
Another disadvantage of glucuronide prodrugs, espe-
cially those based on doxorubicin, is the relatively ineffi-
cient synthesis [14–16]. A maximum yield of no more than
37% of DOX-GA3 can be obtained as final product, which
can be explained as follows. The precursor of DOX-GA3
having the glucuronic acid group protected as methyl ester
and tri-acetate, has to be hydrolyzed under strong basic con-
ditions. In this basic reaction mixture, however, extensive
side-product formation takes place due to base lability of
doxorubicin. This step accounts for the main reduction in
the yield of DOX-GA3. We anticipated that the production
of DOX-mGA3 (Fig. 2) should result in a much higher
yield. The hydrolysis of only the tri-acetate, and not the
methyl group, in the last step of the synthesis could be per-
formed under less basic conditions for a shorter time period
and, therefore, would result in less side-product formation.
In the present study, we synthesized DOX-mGA3. We
investigated its stability and the antiproliferative effects of
DOX-mGA3 in vitro and compared the data with the results
obtained with DOX-GA3. We analyzed the conversion of
DOX-mGA3 into DOX-GA3 by esterase activity in plasma
of mice, rats and humans and subsequent hydrolysis to
doxorubicin. We also determined whether DOX-mGA3
could be converted into DOX-GA3 and doxorubicin in vivo.
A disadvantage of hydrophilic glucuronide prodrugs is
their rapid elimination by the kidneys. In earlier experi-
ments in tumor-bearing nude mice, DOX-GA3 had com-
pletely disappeared from plasma within 4 h after i.v.
administration of 250 mg kgꢀ1 of the prodrug [6]. The
rapid elimination of glucuronide prodrugs poses a major
problem for their use in cancer treatment, as this means that
very high doses are required. We hypothesized that a
methyl ester of DOX-GA3, DOX-mGA3, would be more
lipophilic than DOX-GA3 and of better use in enzyme
Fig. 1. The methyl glucuronate prodrug will be hydrolyzed first to liberate its glucuronide by esterases. The formed acid should be further cleaved by b-
glucuronidase and spontaneous release of CO2 to the drug-spacer molecule (a). The electron-releasing free amino-group will trigger the 1,6-elimination process
and second molecule of CO2 resulting in release of the parent antracycline and iminoquinone methide (b) transition state that is rapidly hydrolyzed to the non-
toxic 4-aminobenzyl alcohol (c).

M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
2275
Fig. 2. The synthesis of the methylated glucuronide prodrugs of doxorubicin, DOX-mGA3. For details see Section 2.
2. Materials and methods
18.8 Hz 10eq-H), 2.99 (1H, d, J = 18.8 Hz 10ax-H), 3.14–
3.50 (4H, m, 30-H, Gluc2,3,4,-H), 3.66 (2H, s, CO2Me)
3.70 (1H, m, 40-H), 3.88 (1H, d, J = 9.7 Hz, Gluc5-H), 3.97
(3H, s, 4-OMe), 4.12 (1H, m, 50-H), 4.57 (2H, d, J = 6.1 Hz,
9-CH2OH), 4.70 (1H, d, J = 5.7 Hz, 40-OH), 4.85 (1H, d, J
= 6.1 Hz, OH), 4.87 (2H, s, ArCH2-), 4.92 (1H, t, J =
4.4 Hz, 7-H), 5.21 (1H, d, J = 2.6 Hz, 1-H0), 5.33 (1H, d, J =
4.4 Hz, OH), 5.42 (3H, m, Gluc1-H, OH, 9-OH), 6.84 (1H,
d, J = 7.9 Hz, 30-NH-), 7.24 (2H, d, J = 8.4 Hz, Ar3,5-H),
7.41 (2H, d, J = 8.4 Hz, Ar2,6-H), 7.63 (1H, t, J = 4.8 Hz,
3-H), 7.89 (2H, d, J = 4.4 Hz,1,2-H), 9.36 (1H, bs, ArNH-),
9.97 (1H, bs, 11-OH), 10.12 (1H, bs, 6-OH).
2.1. Synthesis of N-[4-(doxorubicin-N-carbonyl-
oxymethyl)phenyl] O-(methyl-O-b-glucuronyl) carbamate
(DOX-mGA3)
The synthesis of the protected glucuronide prodrug of
doxorubicin (Fig. 2) was performed using the same meth-
odology as described by Leenders et al. [16]. Methyl-2,3,4-
tri-O-acetyl D-glucuronic acid (1) was obtained as
described [17]. The b-glucuronyl carbamate spacer moiety
was prepared from TBDMS protected 4-hydroxymethyl-
benzoic acid (2) by the modified Curtius reaction [16]. The
resulting protected pro-moiety (3) was coupled to the
anthracycline after deprotection of the benzyl alcohol
functionality and transformation to the active carbonate
(4) using 4-nitrophenyl chloroformate. The resulting tri-
acetylated prodrugs (5) were deacetylated as described
below to obtain the methyl ester prodrug, DOX-mGA3.
An 0.1 N solution of NaOMe in methanol (1.9 mL,
0.21 mmol) at 0 8C was added to a solution of 5 [16]
(225 mg, 0.21 mmol) in anhydrous mixture of methanol
(50 mL) and tetrahydrofuran (10 mL), and the deep blue
solution was stirred for 1 h under argon. Thereafter, the
solution was neutralized by the addition of the H+ form of
Dowex 50 WX2 cation-exchange resin. Filtration followed
by evaporation under reduced pressure resulted in a crude
residue, which was purified with column chromatography
(SiO2 MeOH/CH2Cl2, 1/10), giving DOX-mGA3 as
amorphous red solid (117 mg, 0.126 mmol), mp 172–
174 8C. The overall yield was 60%. Anal. calc. for
C43H46N2O21ꢁ1.5H2O: C, 54.15; H, 5.18; N, 2.94. Found:
C, 53.97; H, 5.48; N, 3.16. NMR [300 MHz, (CD3)2SO)] d
1.12 (2H, d,0J = 7.0 Hz, 50-Me), 1.45 (1H, dd, J = 12.7 Hz, J
= 3.9 Hz, 2 eq-H), 1.83 (1H, dt, J = 12.7 Hz, J = 3.5 Hz,
20ax-H), 2.09 (1H, dd, J = 14.4 Hz, J = 5.7 Hz, 80ax-H), 2.20
(1H, d, J = 3.5 Hz, J = 14.9 Hz, 80eq-H), 2.91 (1H, d, J =
2.2. In vitro characteristics of DOX-mGA3
A stock solution of DOX-mGA3 (7.5 mM) was prepared
in DMSO. The purity and stability of DOX-mGA3
(10 mM) was determined in phosphate-buffered saline
(PBS)/0.1% bovine serum albumin (BSA) at pH 7.4 at
37 8C. Samples (20 mL) were taken at different time
points. Analysis with reversed-phase HPLC was performed
as described in the next section.
The hydrophilicity of DOX-mGA3 was analyzed by
determination of the octanol/PBS partition coefficient.
DOX-GA3 and doxorubicin were used as control drugs.
Doxorubicin was derived from Pharmacia & Upjohn. The
(pro)drugs were dissolved in 1 mL of octanol to a con-
centration of 10 mM, and then, 1 mL of PBS was added.
The mixture was incubated for 3 h at 37 8C under con-
tinuous rigorous shaking (300 rpm). Aliquots of 100 mL of
both phases were collected and diluted in 1 mL of metha-
nol. The ratio of (pro)drug in octanol and PBS was
determined by measuring the fluorescence of both phases
(excitation wavelength 480 nm and emission wavelength
of 580 nm; Perkin Elmer 3000). The octanol/PBS partition
coefficient was calculated after subtraction of the back-
ground from octanol and PBS.

2276
M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
The hydrolysis of the methyl group of DOX-mGA3
2.5. In vitro antiproliferative effects
(10 mM) in human, mouse and rat plasma was analyzed.
Samples (20 mL) were taken at different time points and
prepared for HPLC analysis. As a control, the hydrolysis of
DOX-mGA3 (6 mM) by 3 units of carboxylesterase of
porcine liver (Sigma–Aldrich) was checked in 100 mL
of PBS/0.1% BSA after incubation at 37 8C for 45 min.
In order to demonstrate that DOX-mGA3 is not a good
substrate for human recombinant b-glucuronidase, DOX-
mGA3 (100 mM in PBS) was diluted in PBS/0.1% BSA
with human b-glucuronidase (10 mg mLꢀ1) to a final con-
centration of 10 mM. At different time points, samples
(20 mL) were taken and prepared for HPLC analysis. For
measurement of the half-life of the hydrolysis of DOX-GA3
generated from DOX-mGA3, DOX-mGA3 was diluted in
mouse plasma (100 mM) and incubated for 30 min at 37 8C.
This mixture was then incubated in PBS/0.1% BSA with
human b-glucuronidase (10 mg mLꢀ1) to a final prodrug
concentration of 10 mM as described above.
The human ovarian cancer cell line OVCAR-3 [19], the
human breast cancer cell line MCF-7 [20], the human lung
cancer cell line A549 (ATCC CLL 185), and the human
colon cancer cell line WiDr (ATCC CLL 218) were grown
in Dulbecco’s modified Eagle’s medium supplemented
with 10% fetal calf serum, 50 IU mLꢀ1 penicillin and
50 mg mLꢀ1 streptomycin (Life Technologies) in a humi-
dified atmosphere containing 5% CO2 at 37 8C.
Cells were harvested and seeded in a 96-well microtiter
plate at 5000 cells per well. After 24 h, prodrug or drug was
added at different concentrations to the culture medium
with or without human b-glucuronidase (0.05 U mLꢀ1) in
triplicate. After 72 h, the wells were incubated with the cell
proliferation reagent WST-1 (Roche) for 1 h at 37 8C. The
absorbance was measured at a wavelength of 450 nm. The
antiproliferative effects were determined and expressed as
a percentage of growth of treated cells as compared to
control cell growth, which was set at 100%. The IC50 value
was expressed as the (pro)drug concentration that gave
50% cell growth inhibition.
2.3. HPLC analysis
Samples (20 mL) were treated by the addition of 80 mL
acetonitrile in order to precipitate proteins. After incuba-
tion at ꢀ20 8C for 10 min, samples were centrifuged at
16,000 ꢂ g for 10 min. The supernatant (80 mL) was
collected and 80 mL of phosphate buffer (0.5 mM triethy-
lamine/20 mM sodium phosphate) was added.
2.6. Animals and human xenograft model
The animal experiments were performed in accordance
with the institution guidelines after approval by the Institu-
tional Animal Experimental Committee. Female NMRI
mice (Harlan) were handled under conventional condi-
tions. Specified pathogen-free conditions were used for
female athymic nude mice (Hsd: athymic nude-nu; Har-
lan).
The HPLC apparatus consisted of an autosampler (Mara-
thon-XT), a pump (Separations, Model 300), and a fluor-
escence detector (Jasco, Model 821-FP: excitation 480 nm;
emission 590 nm). Samples (50 mL) were loaded on a
reversed-phase column (Chromsep C18, 2 mm ꢂ 100 mm
ꢂ 4.6 mm, 3 mm particle size). Elution was done with
eluens in an isocratic run (0.5 mM triethylamine/20 mM
sodium phosphate:acetonitrile = 3:2). Calibration of the
system was performed as previously described [18] and the
detection limit was 0.1 mM. In each cluster of runs, stan-
dards of doxorubicin, DOX-GA3 and DOX-mGA3 served
as references at the beginning and at the end of the cluster.
Control plasma and tissue samples spiked with known
concentrations of the different compounds were included
in the biodistribution experiments. No degradation was
observed and the recovery of the compounds was taken
into consideration in the analysis. The elution peaks in the
chromatogram were integrated using the Gyncosoft pro-
gram (Gynkotek, Version 5.3E).
The human ovarian cancer xenograft FMa has been
described earlier [21]. It is a poorly differentiated muci-
nous adenocarcinoma with a mean volume-doubling time
of 5.5 days. Xenografts from previous recipients were
transferred by implanting tissue fragments with a diameter
of 2–3 mm into both flanks of 8–10-week-old mice.
Tumors from previous recipients were transferred by
implanting tissue fragments with a diameter of 2–3 mm
into both flanks of 8–10-week-old mice. Upon growth,
tumors were measured twice a week by the same observer.
The tumor volume was calculated by the equation length ꢂ
width ꢂ thickness ꢂ 0.5, and expressed in mm3.
2.7. DOX-mGA3 in plasma of non-tumor-bearing mice
DOX-mGA3 was dissolved in Cremophor EL/ethanol
(1/1 v/v) at a concentration of 10 mg mLꢀ1 and was stable
for at least 48 h as analyzed by HPLC. Immediately before
injection, one part of this solution was diluted with four
parts of 0.9% NaCl giving a concentration of DOX-mGA3
of 2 mg mLꢀ1. Six mice were injected i.v. with DOX-
mGA3 or DOX-GA3 at a dose of 20 mg kgꢀ1. Blood
samples were collected with the use of heparinized glass
capillaries at different time points ranging from 1 to 60 min
2.4. Esterase activity assay
Human, mouse or rat plasma was incubated with 200 mL
of 100 mM Tris–HCl, pH 8.0, containing 1 mM p-nitro-
phenyl acetate (pNPAc, Sigma–Aldrich), a substrate for
esterases. After mixing, the rate of conversion to p-nitro-
phenol was measured at a wavelength of 415 nm during
10 min using an ELISA plate reader (BioRad).

M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
2277
(DOX-GA3) and to 90 min (DOX-mGA3) using three mice
per time point. The samples were centrifuged at 2700 g for
10 min to separate the plasma. From the plasma, 20 mL
was prepared for analysis by HPLC.
hydrophilic DOX-GA3 had a relatively low value of ꢀ0.15
as compared to 0.52 of doxorubicin. These values were
similar to those reported before [5]. The log P of DOX-
mGA3 was 1.27 and considerably higher than that of
doxorubicin demonstrating the very lipophilic nature of
this prodrug.
2.8. Distribution of DOX-mGA3 and DOX-GA3 in
tumor and normal tissues
We tested whether the methyl ester group of DOX-
mGA3 could be hydrolyzed by esterases in blood [23].
Therefore, we incubated DOX-mGA3 with plasma from
different species (Fig. 3A). The hydrolysis in mouse
plasma (t1/2 of approximately 0.5 min) and that in rat
plasma (t1/2 of approximately 1.5 min) were much faster
than that in human plasma (t1/2 of approximately 2.5 h).
The rate of hydrolysis was in agreement with the esterase
activity in plasma of the different species (Fig. 3B). By
measurement of the conversion of the substrate pNPAc, it
was demonstrated that the esterase activity in human
plasma was approximately 2.5-fold lower than that in
mouse and approximately 2-fold lower than that in rat
plasma. Of note: the assay, however, does not discriminate
for differences in catalytic rate or esterase content relevant
for DOX-mGA3 conversion between species. To confirm
that esterase activity present in blood was indeed respon-
sible for the hydrolysis of the methyl group of DOX-
mGA3, we incubated DOX-mGA3 with porcine liver
carboxylesterase at 37 8C. After 45 min, 93% of DOX-
mGA3 was converted into DOX-GA3, while the methyl
group was not removed in the absence of the enzyme (data
not shown).
Nude mice bearing s.c. FMa xenografts with a mean
tumor volume of 300 mm3 were injected i.v. with
20 mg kgꢀ1 DOX-mGA3 (in Cremophor EL/ethanol/
0.9% NaCl (1/1/8, v/v)) or DOX-GA3 (in water). At
different time points ranging from 1 min to 2 h, blood,
tumors, liver, heart and kidneys were removed in groups of
three mice per time point. Blood samples were collected as
described in the previous section. To plasma 1/100 volume
of 100 mM of bis(4-nitrophenyl)phosphate (BNPP) was
added to prevent hydrolysis of DOX-mGA3 to DOX-GA3
[22]. Furthermore, 1/100 volume of 100 mM D-glucaric
acid-1,4-lactone monohydrate (saccharonolactone; Fluka)
was added to prevent hydrolysis of DOX-GA3 by b-
glucuronidase [6]. Samples were prepared for HPLC ana-
lysis as described above. After blood sampling, tumor,
liver, heart and kidney tissues were collected and imme-
diately frozen in liquid nitrogen and stored at ꢀ80 8C until
preparation for HPLC analysis. Tissue samples were pul-
verized (Mikro-Dismembrator II, B. Braun Biotech Inter-
national, Nieuwegein, The Netherlands) and diluted
(230 mg tissue/770 mL PBS containing 1 mM saccharo-
nolactone and 1 mM BNNP. The homogenate (40 mL) was
diluted in 3.3% (w/v) AgNO3 (8 mL) and acetonitrile
(32 mL), shaken (15 min, room temperature), sonicated
(15 min, room temperature) and centrifuged (16,000 ꢂ
g, 10 min). To the supernatant (50 mL), 117 mL of phos-
phate buffer (0.5 mM triethylamine/20 mM sodium phos-
phate) was added. Samples were analyzed by HPLC as
described above.
In order to show that DOX-mGA3 will ultimately be
converted to doxorubicin by human b-glucuronidase, we
first incubated DOX-mGA3 in mouse plasma or PBS/0.1%
BSA for 30 min at 37 8C. After this time period, 95% of the
DOX-mGA3 was converted into DOX-GA3 in mouse
plasma (Fig. 4A), while no DOX-GA3 was generated in
PBS/0.1% BSA (Fig. 4B). Thereafter, human b-glucuro-
nidase was added. DOX-GA3 generated after incubation of
DOX-mGA3 in mouse plasma was completely hydrolyzed
to doxorubicin by this enzyme at 120 min. In PBS/0.1%
BSA, only 20% of doxorubicin from DOX-mGA3 could be
detected at 240 min after addition of b-glucuronidase. This
finding can be explained by spontaneous generation of
DOX-GA3 as we demonstrated that DOX-mGA3 is not
highly stable in PBS/0.1% BSA. Therefore, it can be
concluded that DOX-mGA3 itself is not a good substrate
for human b-glucuronidase and that conversion into dox-
orubicin only occurs after hydrolysis of the methyl group.
Each HPLC sample was prepared in duplicate. The mean
fluorescence of each duplicate sample was used to calcu-
late the prodrug concentration in each animal at a given
time. The half-lives (t1/2) and area under the concentra-
tion versus time curves (AUCs) were calculated using
WinNonlin 1.5 (Pharsight).
3. Results
3.1. Characterization of DOX-mGA3
3.2. In vitro antiproliferative effects
DOX-mGA3 was first examined for its stability in PBS/
0.1% BSA at 37 8C. DOX-mGA3 appeared to be gradually
hydrolyzed into DOX-GA3 with a t1/2 of approximately
10 h, while DOX-GA3 was stable for at least 24 h (data
not shown). We then calculated the the logarithm of the
octanol/PBS partition coefficient (log P) as a measure of
the hydrophilicity of the prodrug and the drug. The highly
The antiproliferative effects of doxorubicin, DOX-GA3
and DOX-mGA3 were determined in four different human
malignant cell lines (Table 1). In all four cell lines IC50
values for DOX-GA3 and DOX-mGA3 were at least 37-
fold higher than that of the parent drug doxorubicin. When
OVCAR-3 cells were incubated with DOX-mGA3 in

2278
M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
Fig. 4. Conversion of DOX-mGA3 into doxorubicin by human b-glucur-
onidase after incubation in mouse plasma (A) or PBS/0.1% BSA (B). DOX-
mGA3 (100 mM) was first incubated in mouse plasma or PBS/0.1% BSA for
30 min. These mixtures were then diluted 10 times in PBS/0.1% BSA. After
addition of human b-glucuronidase (10 mg mLꢀ1), samples were taken at
different time points for analysis by HPLC. The peak areas of DOX-mGA3,
DOX-GA3 and doxorubicin were expressed as a percentage of the total peak
areas. DOX-mGA3 (^); DOX-GA3 (&); doxorubicin (~).
Table 1
Antiproliferative effects of DOX-mGA3, DOX-GA3 and doxorubicin
a
expressed as IC50 in human malignant cell lines
(Pro)drug
A549
MCF-7
OVCAR-3
WiDr
(mM)
(mM)
(mM)
(mM)
DOX-mGA3
DOX-GA3
Doxorubicin
a
>50
26
22
30
0.3
7
>50
13
7
0.7
0.1
0.2
(Pro)drug concentration that gives 50% growth in treated cells when
compared with control cell growth. The values are the mean from two
separate experiments performed in triplicate.
3.3. Pharmacokinetics of DOX-mGA3 in plasma of
non-tumor-bearing mice
The kinetics of DOX-mGA3 in comparison to that of
DOX-GA3 in blood was first tested in non-tumor-bearing
mice (Fig. 6). DOX-mGA3 or DOX-GA3 was injected i.v.
at a dose of 20 mg kgꢀ1. Orbita punctures were taken at
different time points after injection and analyzed by HPLC
for the presence of DOX-mGA3 and DOX-GA3. Admin-
istration of DOX-mGA3 resulted in a peak concentration of
Fig. 3. (A) HPLC analysis of the hydrolysis of the methyl ester bond of
DOX-mGA3 (6 mM) in mouse, rat and human plasma at 37 8C. At different
time points, samples were taken and analyzed by HPLC. Concentrations of
DOX-mGA3 and DOX-GA3 were expressed as the corresponding peak
areas (mV ꢂ min). DOX-mGA3 (^) was totally converted to DOX-GA3
(&) in plasma of all three species. (B) Esterase activity in mouse, rat and
human plasma in two separate samples per species. Plasma was diluted in
PBS and incubated with 1 mM pNPAc. Substrate conversion was measured
after 10 min. Esterase activity was expressed as the rate of pNPAc con-
version in mmol minꢀ1 mgꢀ1. Bars, ꢃS.D.
combination with an excess of human b-glucuronidase
(0.05 U mLꢀ1), the IC50 value was comparable to that of
doxorubicin and DOX-GA3 in combination with human
b-glucuronidase (Fig. 5). This finding can be explained
by the hydrolysis of the methyl group of DOX-mGA3 in
the culture medium during the 72-h assay, whereafter
DOX-GA3 can be activated by human b-glucuronidase
to doxorubicin.
Fig. 5. Antiproliferative effects of DOX-mGA3 in various concentrations
with or without human b-glucuronidase in OVCAR-3 cells. Treatment of
OVCAR-3 cells with doxorubicin or DOX-GA3 with or without human b-
glucuronidase served as controls. After 72 h, cell growth was measured with
the cell proliferation reagent, WST-1, and the growth inhibitory effects were
expressed as a percentage of control cell growth. The experiment was
performed in triplicate wells. Doxorubicin (*); DOX-mGA3 (~); DOX-
GA3 (&); DOX-mGA3 with b-glucuronidase (~); DOX-GA3 with b-
glucuronidase (&). Bars, ꢃS.D.

M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
2279
Fig. 6. The concentration-time curves of DOX-mGA3 and DOX-GA3 in
plasma as a result of the i.v. administration of 20 mg kgꢀ1 DOX-mGA3 or
DOX-GA3 to groups of 3 mice each. DOX-mGA3 (^); DOX-GA3 from
DOX-mGA3 (&); DOX-GA3 from DOX-GA3 itself (&). Bars, ꢃS.D.
this prodrug of 1766 mM (t = 1 min) in plasma and an
elimination t1/2 of 15.8 min. DOX-GA3 generation showed
a plateau phase at a plasma concentration of 103 mM
during approximately 30 min followed by an elimination
phase with a t1/2 of 25.3 min. The t1/2 of DOX-GA3 itself
after i.v. injection at a dose of 20 mg kgꢀ1 was 26.1 min.
The AUC of DOX-GA3 generated from DOX-mGA3
from 1 to 60 min was 4873 nmol minꢀ1 mLꢀ1. This was
2.7-fold higher than the AUC of DOX-GA3 itself, which
was 1965 nmol minꢀ1 mLꢀ1
.
3.4. Biodistribution of DOX-mGA3 in
FMa-bearing mice
Fig. 7. The concentration-time curves of: DOX-mGA3 (^), DOX-GA3
(&) and doxorubicin (~) as a result of the i.v. administration of 20 mg kgꢀ1
DOX-mGA3; DOX-GA3 (&) and doxorubicin (~) as a result of the i.v.
administration of 20 mg kgꢀ1 DOX-GA3 in plasma, tumors, liver, heart,
and kidney of FMa-bearing mice. Groups consisted of three mice each.
Doxorubicin was not detectable in plasma after administration of the given
dose of DOX-mGA3 or DOX-GA3 at any time. Bars, ꢃS.D.
We next investigated whether the concentrations of
DOX-GA3 and doxorubicin in plasma and tissues of
FMa-bearing mice after i.v. administration of 20 mg kgꢀ1
1 DOX-mGA3 were higher than those after the same dose
of DOX-GA3. This information would give an indication
whether a higher antitumor efficacy might be expected
from DOX-mGA3. Plasma, tumors, liver, heart and kid-
neys were collected at 0.5, 1 and 2 h after administration of
the prodrug. In addition, plasma was collected by orbita
puncture at 1 min after prodrug administration. All sam-
ples were analyzed by HPLC for the presence of DOX-
mGA3, DOX-GA3 and doxorubicin (Fig. 7). The curves of
the plasma kinetics of DOX-GA3 generated from DOX-
mGA3 and of DOX-GA3 itself were similar in FMa-
bearing mice as compared to non-tumor-bearing mice
(Fig. 6). The AUC was calculated from the limited number
of time points available between 0.5 and 2 h. In general, the
AUC of DOX-GA3 after DOX-mGA3 in tumor and normal
tissues was 2.5- to 3-fold higher than after DOX-GA3,
except for that in liver (Table 2). The generation of
doxorubicin was most pronounced in tumor tissue, in
which the AUC of doxorubicin was 50% of that of
DOX-GA3 (Tables 2 and 3). Of interest, the administration
of DOX-mGA3 resulted in a 2.7-fold higher AUC of
doxorubicin in tumor tissue when compared to the AUC
after DOX-GA3 administration.
The ratio between doxorubicin concentrations after
DOX-mGA3 and DOX-GA3 administration in FMa tumor
tissue increased from almost 1 at 0.5 h after injection to
approximately 5 at 2 h (Fig. 8). In normal mouse tissues,
the ratio was approximately 1 at all time points except for a
higher ratio in heart and kidney at the earliest time point of
0.5 h.
Table 2
AUCa of DOX-GA3 in plasma and tissues of FMa-bearing nude mice after i.v. administration of 20 mg kgꢀ1 DOX-GA3 or DOX-mGA3
Treatment
Plasma (nmol minꢀ1 mLꢀ1
)
Tumor (nmol minꢀ1 gꢀ1
)
Liver (nmol minꢀ1 gꢀ1
)
Heart (nmol minꢀ1 gꢀ1
)
Kidney (nmol minꢀ1 gꢀ1
)
DOX-mGA3 427.5
44.8
15.3
188.7
297.8
29.0
12.0
100.3
33.4
DOX-GA3
59.7
a
AUC from 0.5 h until 2 h.

2280
M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
Table 3
AUCa of doxorubicin in tissues of FMa-bearing nude mice after i.v. administration of 20 mg kgꢀ1 DOX-GA3 or DOX-mGA3
Treatment
Tumor (nmol minꢀ1 gꢀ1
)
Liver (nmol minꢀ1 gꢀ1
)
Heart (nmol minꢀ1 gꢀ1
)
Kidney (nmol minꢀ1 gꢀ1
)
DOX-mGA3
DOX-GA3
a
21.6
8.1
65.4
7.8
4.5
20.6
17.5
101.9
AUC from 0.5 h until 2 h.
mGA3 as compared to the AUC after administration of
DOX-GA3 itself in non-tumor-bearing mice as well as in
tumor-bearing nude mice. The level of DOX-GA3 gener-
ated from DOX-mGA3 showed a plateau phase of approxi-
mately 30 min followed by the elimination phase. The
duration of the plateau phase might seem to be long
considering the very short conversion t1/2 of 0.5 min that
was measured in mouse plasma in vitro. The peak con-
centration of DOX-mGA3 of 1766.4 mM in vivo, however,
was considerably higher than the 6 mM used in the in vitro
assay. The slower conversion of DOX-mGA3 in vivo might
be attributed to the saturation of the esterases in plasma.
Indeed, we have observed a longer conversion t1/2 of 5 min
at a concentration of DOX-mGA3 of 800 mM in mouse
plasma in vitro (data not shown). Another possible expla-
nation for the 30-min plateau phase observed in mice will
be the rapid uptake of the lipophilic DOX-mGA3 in tissues,
whereafter conversion to DOX-GA3 will result in the
release of the prodrug in the circulation. The pharmaco-
kinetic behaviour of DOX-mGA3 might even be more
favorable in man considering the much lower conversion
t1/2 of DOX-mGA3 of approximately 2.5 h that we mea-
sured in human plasma in vitro (Fig. 3).
Fig. 8. Ratio of doxorubicin (DOX) concentrations obtained after DOX-
mGA3 and DOX-GA3 administration (20 mg kgꢀ1 i.v.) in mouse plasma,
normal tissue and FMa tumor tissue at 0.5, 1 and 2 h after injection. The
ratio was calculated by dividing the doxorubicin concentration obtained
after DOX-mGA3 administration by the doxorubicin concentration after
DOX-GA3 administration.
4. Discussion
DOX-GA3 is a prodrug with a higher therapeutic
index than doxorubicin as shown in human ovarian cancer
xenografts, because of tumor-selective activation by b-
glucuronidase [6]. The production process of DOX-GA3,
however, is very inefficient. The advantage of the synthesis
of DOX-mGA3 instead of DOX-GA3 is that in the last
deprotection step only the tri-acetate, and not the methyl
group, has to be chemically removed. We showed that
under less basic conditions for a shorter time period the
overall yield of DOX-mGA3 was 60%. For comparison,
we have previously reported that complete deprotection
yielded 37% of DOX-GA3 [16].
In tumor tissue, an increased AUC of DOX-GA3 was
reached after administration of DOX-mGA3 as compared
to that after administration of DOX-GA3. The AUC of
doxorubicin was 50% of the DOX-GA3 AUC in tumor
tissue, most probably as a result of the presence of extra-
cellular b-glucuronidase. Therefore, DOX-mGA3 might
be expected to give a better tumor growth inhibition than
DOX-GA3 at the same dose. In contrast, the higher con-
centration of DOX-GA3 after the administration of DOX-
mGA3 did not lead to a substantially increased concentra-
tion of doxorubicin in normal tissues, in which extracel-
lular b-glucuronidase should be absent (Table 3). Since
doxorubicin generation in heart tissue after DOX-mGA3 as
well as after DOX-GA3 [6] was proportionally lower than
that in tumor tissue, we expect that this prodrug will be less
cardiotoxic than is normally associated with doxorubicin
treatment.
We anticipated that the methyl group of the methylated
glucuronide prodrug could be enzymatically removed by
esterase activity in blood. Indeed, we showed that DOX-
mGA3 was completely hydrolyzed to DOX-GA3 in plasma
at a rate dependent on the level of esterase activity. We
demonstrated that DOX-mGA3 was not a substrate for
human b-glucuronidase. After removal of the methyl
group by esterases for formation of DOX-GA3, complete
conversion to doxorubicin could be achieved by human
b-glucuronidase. Moreover, the antiproliferative effect of
DOX-mGA3 in the presence of b-glucuronidase was simi-
lar to that of doxorubicin in OVCAR-3 cells in vitro.
We hypothesized that the administration of lipophilic
DOX-mGA3 would result in a prolonged circulation of
DOX-GA3 in mice as compared to direct administration
of the hydrophilic DOX-GA3. Despite the high esterase
activity in mouse plasma, we were able to show a 2.7-fold
increase of the plasma AUC of DOX-GA3 from DOX-
Treatment with DOX-mGA3 is hampered by its low
solubility precluding effective doses to be administered.
The lipophilic DOX-mGA3 is not soluble in aqueous
solutions such as water, phosphate buffer or ethanol. A
solution of DOX-mGA3 could be prepared in Cremophor
EL/ethanol (1:1) at a concentration of 10 mg mLꢀ1 and
further diluted in 0.9% NaCl to 2 mg mLꢀ1. In the clinic,
some drugs have to be given as a solution in Cremophor
EL. For example, administration of paclitaxel in Cremo-

M. de Graaf et al. / Biochemical Pharmacology 68 (2004) 2273–2281
2281
and efficacy in experimental human ovarian cancer. Br J Cancer
2001;84:550–7.
phor EL is a feasible approach when precautions are taken
to prevent hypersensitivity reactions [24]. The highest dose
of DOX-mGA3 that could be given in a single injection
i.v. to mice was 20 mg kgꢀ1. For comparison, we have
previously shown that DOX-GA3 gave good tumor growth
inhibition in FMa-bearing nude mice at a dose of
250 mg kgꢀ1 [6]. Even considering the fact that a 2.7-fold
higher concentration of doxorubicin can be reached in
tumor tissue after DOX-mGA3 than after the same dose
of DOX-GA3, efforts should be taken to formulate the
prodrug in different solvents.
[7] Bosslet K, Straub R, Blumrich M, Czech J, Gerken M, Sperker B, et al.
Elucidation of the mechanism enabling tumor selective prodrug
monotherapy. Cancer Res 1998;58:1195–201.
[8] Houba PH, Boven E, Van der Meulen-Muileman IH, Leenders RG,
Scheeren JW, Pinedo HM, et al. Distribution and pharmacokinetics of
the prodrug daunorubicin-GA3 in nude mice bearing human ovarian
cancer xenografts. Biochem Pharmacol 1999;57:673–80.
[9] Bosslet K, Czech J, Hoffmann D. A novel one-step tumor-selective
prodrug activation system. Tumor Targeting 1995;1:45–50.
[10] Houba PH, Boven E, Erkelens CA, Leenders RG, Scheeren JW, Pinedo
HM, et al. The efficacy of the anthracycline prodrug daunorubicin-
GA3 in human ovarian cancer xenografts. Br J Cancer 1998;78:
1600–6.
In conclusion, in vivo administration of DOX-mGA3 led
to improved pharmacokinetics of this prodrug and 2.7-fold
higher doxorubicin concentrations in tumor tissue as com-
pared to administration of DOX-GA3 itself. Therefore,
better antitumor effects might be expected from a lower
dose of DOX-mGA3 than from DOX-GA3. Furthermore,
the synthesis of DOX-mGA3 is much more efficient
than that of DOX-GA3. The therapeutic usefulness of
DOX-mGA3, however, should be improved aiming at
better solubility in clinically suitable solvents.
[11] Woessner R, An Z, Li X, Hoffman RM, Dix R, Bitonti A. Comparison
of three approaches to doxorubicin therapy: free doxorubicin, lipo-
somal doxorubicin, and beta-glucuronidase-activated prodrug (HMR
1826). Anticancer Res 2000;20:2289–96.
[12] Ghosh AK, Farquhar D. A methyl glucuronate prodrug of phosphor-
odiamidic mustard. Tetrahedron Lett 1997;38:8795–8.
[13] Papot S, Combaud D, Gesson JP. A new spacer group derived from
arylmalonaldehydes for glucuronylated prodrugs. Bioorg Med Chem
Lett 1998;8:2545–8.
[14] Bakina E, Wu Z, Rosenblum M, Farquhar D. Intensely cytotoxic
anthracycline prodrugs: glucuronides. J Med Chem 1997;40:4013–8.
[15] Florent JC, Dong X, Gaudel G, Mitaku S, Monneret C, Gesson JP, et
al. Prodrugs of anthracyclines for use in antibody-directed enzyme
prodrug therapy. J Med Chem 1998;41:3572–81.
Acknowledgements
[16] Leenders RG, Damen EW, Bijsterveld EJ, Scheeren HW, Houba PH,
Van der Meulen-Muileman IH, et al. Novel anthracycline-spacer-beta-
glucuronide,-beta-glucoside, and -beta-galactoside prodrugs for appli-
cation in selective chemotherapy. Bioorg Med Chem 1999;7:1597–
610.
We thank Prof. Dr. W.J.F. van der Vijgh for helpful
discussions. We thank D. Oosterhoff for the esterase
activity measurements and M. Kedde for excellent tech-
nical assistance.
[17] Mikano M. Facile 1-O-deacylation of per-O-acylaldoses. Carbohydr
Res 1989;191:150–3.
[18] De Jong J, Guerand WS, Schoofs PR, Bast A, van der Vijgh WJ.
Simple and sensitive quantification of anthracyclines in mouse atrial
tissue using high-performance liquid chromatography and fluores-
cence detection. J Chromatogr 1991;570:209–16.
References
[1] Haisma HJ, Boven E, Van Muijen M, De Jong J, Van der Vijgh WJ,
Pinedo HM. A monoclonal antibody-beta-glucuronidase conjugate as
activator of the prodrug epirubicin-glucuronide for specific treatment
of cancer. Br J Cancer 1992;66:474–8.
[19] Hamilton TC, Young RC, McKoy WM, Grotzinger KR, Green JA, Chu
EW, et al. Characterization of a human ovarian carcinoma cell line
(NIH:OVCAR-3) with androgen and estrogen receptors. Cancer Res
1983;43:5379–89.
[2] Connors TA, Whisson ME. Cure of mice bearing advanced plasma cell
tumours with aniline mustard: the relationship between beta-glucur-
onidase activity and tumour sensitivity. Nature 1966;210:866–7.
[3] Young CW, Yagoda A, Bittar ES, Smith SW, Grabstald H, Whitmore
W. Therapeutic trial of aniline mustard in patients with advanced
cancer. Comparison of therapeutic response with cytochemical assess-
ment of tumor cell beta-glucuronidase activity. Cancer 1976;38:
1887–95.
[20] Soule HD, Vazguez J, Long A, Albert S, Brennan M. A human cell line
from a pleural effusion derived from a breast carcinoma. J Natl Cancer
Inst 1973;51:1409–16.
[21] Kievit E, Pinedo HM, Schluper HM, Haisma HJ, Boven E. Compar-
ison of monoclonal antibodies 17-1A and 323/A3: the influence of
the affinity on tumour uptake and efficacy of radioimmunotherapy
in human ovarian cancer xenografts. Br
J Cancer 1996;73:
457–64.
[4] De Graaf M, Boven E, Scheeren HW, Haisma HJ, Pinedo HM. Beta-
glucuronidase-mediated drug release. Curr Pharm Des 2002;8:
1391–403.
[22] Yoshigae Y, Imai T, Taketani M, Otagiri M. Characterization of
esterases involved in the stereoselective hydrolysis of ester-type
prodrugs of propranolol in rat liver and plasma. Chirality 1999;
11:10–3.
[5] Houba PH, Leenders RG, Boven E, Scheeren JW, Pinedo HM, Haisma
HJ. Characterization of novel anthracycline prodrugs activated by
human beta-glucuronidase for use in antibody-directed enzyme pro-
drug therapy. Biochem Pharmacol 1996;52:455–63.
[23] Krisch K. Carboxylic ester hydrolases. In: Boyd PD, Boyd PD,
editors. The Enzymes. New York: Academic Press; 1971. p. 59.
[24] Foa R, Norton L, Seidman AD. Taxol (paclitaxel): a novel anti-
microtubule agent with remarkable anti-neoplastic activity. Int J Clin
Lab Res 1994;24:6–14.
[6] Houba PH, Boven E, Van der Meulen-Muileman IH, Leenders RG,
Scheeren JW, Pinedo HM, et al. A novel doxorubicin-glucuronide
prodrug DOX-GA3 for tumour-selective chemotherapy: distribution