Glucose promoiety enables glucose transporter mediated brain uptake of ketoprofen and indomethacin prodrugs in rats

Article abstract of DOI:10.1021/jm8015409

The brain uptake of solutes is efficiently governed by the blood-brain barrier (BBB). The BBB expresses a number of carrier-mediated transport mechanisms, and new knowledge of these BBB transporters can be used in the rational targeted delivery of drug molecules for active transport. One attractive approach is to conjugate an endogenous transporter substrate to the active drug molecule to utilize the prodrug approach. In the present study, ketoprofen and indomethacin were conjugated with glucose and the brain uptake mechanism of the prodrugs was determined with the in situ rat brain perfusion technique. Two of the prodrugs were able to significantly inhibit the uptake of glucose transporter (GluT1)-mediated uptake of glucose, thereby demonstrating affinity to the transporter. Furthermore, the prodrugs were able to cross the BBB in a temperature-dependent manner, suggesting that the brain uptake of the prodrugs is carrier-mediated.

Full text of DOI:10.1021/jm8015409

3348
J. Med. Chem. 2009, 52, 3348–3353
Glucose Promoiety Enables Glucose Transporter Mediated Brain Uptake of Ketoprofen and
Indomethacin Prodrugs in Rats
Mikko Gynther,* Jarmo Ropponen, Krista Laine, Jukka Leppa¨nen, Paula Haapakoski, Lauri Peura, Tomi Ja¨rvinen, and
Jarkko Rautio
Department of Pharmaceutical Chemistry, UniVersity of Kuopio, P.O. Box 1627, FI-70211 Kuopio, Finland
ReceiVed December 6, 2008
The brain uptake of solutes is efficiently governed by the blood-brain barrier (BBB). The BBB expresses
a number of carrier-mediated transport mechanisms, and new knowledge of these BBB transporters can be
used in the rational targeted delivery of drug molecules for active transport. One attractive approach is to
conjugate an endogenous transporter substrate to the active drug molecule to utilize the prodrug approach.
In the present study, ketoprofen and indomethacin were conjugated with glucose and the brain uptake
mechanism of the prodrugs was determined with the in situ rat brain perfusion technique. Two of the prodrugs
were able to significantly inhibit the uptake of glucose transporter (GluT1)-mediated uptake of glucose,
thereby demonstrating affinity to the transporter. Furthermore, the prodrugs were able to cross the BBB in
a temperature-dependent manner, suggesting that the brain uptake of the prodrugs is carrier-mediated.
Introduction
prodrug delivery.12 Several in vitro studies have been performed
with different drug molecules in order to determine the ability
of their glycosyl derivatives to bind to GluT1. In addition,
systemically delivered glycosyl derivatives of 7-chlorokynurenic
acid, L-dopa, and dopamine have been shown to have pharma-
cological activity in the CNS of rodents.13-16 According to the
GluT1 model published by Mueckler and Makepeace (2008),17
the hydroxyl group of glucose situated at the carbon 6-position
goes into a hydrophobic pocket in the transporter protein
substrate binding site. In addition, the hydroxyl group at the
carbon 6-position does not form a hydrogen bond with the
transporter that would be crucial for the affinity. Fernandez et
al.18 synthesized several glycosyl derivatives of dopamine and
tested the affinity of the prodrugs to GluT1 in human erythro-
cytes. Dopamine was linked to glucose with different linkers
at the carbon 1-, 3-, and 6-positions of glucose. The results of
glucose uptake inhibition showed that the glucose derivatives
that were conjugated at position 6 had the best affinity for
GluT1. Therefore, the hydroxyl group at the carbon 6-position
is likely the most potential functional group to which to attach
the drug molecule in order to maintain the affinity of the glucose
conjugate for the GluT1 transporter. These previous studies have
indeed demonstrated that glucose conjugates can bind to GluT1
and that the derivatives are centrally available, but none of these
studies verified the ability of conjugates to cross the BBB via
GluT1. Therefore, the overall aim of the present study was to
show with two model compounds ketoprofen and indomethacin,
that GluT1 can be utilized to carry nonsubstrate drugs into the
brain by conjugating a drug molecule to D-glucose with
bioreversible linkage. Four glucose prodrugs synthesized and
studied are ketoprofen-glucose produg (4), indomethacin-
glucose prodrug (6), indomethacin-glycolic acid-glucose pro-
drug (10) and indomethacin-lactic acid-glucose prodrug (14)
(Figure 1).
The development of new pharmacologically active drugs for
central nervous system (CNSa) disorders often fails because of
poor brain uptake of these molecules.1 The brain uptake of
solutes is efficiently governed by the blood-brain barrier (BBB).
The BBB segregates the CNS from the systemic circulation,
and its main physiological functions include maintaining ho-
meostasis at the brain parenchyma and protecting the brain from
potentially harmful chemicals. The BBB is primarily formed
from capillary endothelial cells, which differ from the other
tissues.2 The brain capillary endothelial cells are very closely
joined together by tight intercellular junctions that efficiently
restrict the paracellular diffusion of hydrophilic drugs.3 In
addition to being a selective structural diffusion barrier, the BBB
constitutes an efficient functional barrier for solutes crossing
the cell membrane. The high metabolic activity of brain capillary
endothelial cells,4 as well as effective efflux systems that actively
remove solutes from brain tissue and return them back to the
bloodstream,2,5-7 create a great challenge for potential neuro-
therapeutics. Furthermore, the BBB expresses a number of
specific carrier-mediated inward transport mechanisms that
ensure an adequate nutrient supply for the brain,8 and new
knowledge of these endogenous BBB transporters can be used
in the rational reformulation of drug molecules for active
transport.9 A small molecule may diffuse through 40 µm space
in about 1 s, which indicates that after passage across the BBB,
the drug is almost instantly distributed within the whole cerebral
tissue.1,10 These physiological facts indicate that the vascular
route would be very promising in drug delivery to the brain if
nutrient transporters present at the BBB could be targeted.
The glucose transporter (GluT1) is present both on the luminal
and the abluminal membrane of the endothelial cells forming
the BBB.11 GluT1 transports glucose and other hexoses and has
the highest transport capacity of the carrier-mediated transporters
present at the BBB, being therefore an attractive transporter for
Results and Discussion
Chemical and Enzymatic Stability of Prodrugs. The
degradation of prodrugs 4 and 6 was studied in aqueous buffer
solution of pH 7.4 at 37 °C. The degradation of 4 and 6 followed
pseudo-first-order kinetics with the half-lives of 88.6 ( 0.0 and
73.4 ( 1.6 h (mean ( sd, n ) 2), respectively (Table 1).
* To whom correspondence should be addressed. Phone: +358-17-
162252. Fax: +358-17-162456. E-mail: mikko.gynther@uku.fi.
a Abbreviations: CNS, central nervous system; BBB, blood-brain barrier;
DMSO, dimethyl sulfoxide; GluT1, glucose transporter; IC50, half maximal
inhibitory concentration; PA, permeability-surface area; Vd, volume of
distribution.
10.1021/jm8015409 CCC: $40.75
2009 American Chemical Society
Published on Web 04/29/2009

Brain Uptake of Ketoprofen and Indomethacin Prodrugs
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10 3349
Figure 1. Chemical structures of ketoprofen prodrug 4 and indomethacin prodrugs 6, 10, and 14.
Table 1. Molecular Weight, Hydrolysis Rates in Phosphate Buffer
Determination of the Brain Uptake Mechanism for
Prodrugs 4 and 6. In the present study, modified in situ rat
brain perfusion technique was used for the determination of the
brain uptake of the prodrugs.19 In situ technique was chosen
for the determination of the prodrug uptake because it provides
insight into the accurate mechanism of the BBB penetration.20
In situ rat brain perfusion technique is described in detail in
our previous study.21
Solution, and Polar Surface Area of the Prodrugsa
molecular
weight (g/mol)
t1/2 (h) 37 °C phosphate
polar surface
area (Å2)
prodrug
buffer (pH 7.4)
4
6
10
14
416
519
577
591
88.5 ( 0.0
40.5 ( 2.5
3.0 ( 0.2
9.5 ( 0.0
133.52
145.99
172.29
172.29
a Mean ( sd, n ) 2.
The in situ rat brain perfusion technique has not been
previously used to determine the brain uptake of glycosyl
conjugates. Therefore, the suitability of the technique for
glycosyl conjugate uptake determination was studied by con-
firming the presence of functional GluT1-transporters in rat BBB
with [14C]D-glucose, which is an endogenous substrate for the
GluT1.11 The uptake of molecules across the BBB was
quantified by determining the permeability-surface area (PA)
product. The PA product of 0.2 µCi/mL [14C]D-glucose was
determined to be 0.0042 ( 0.0002 mL/s/g (mean ( sd, n ) 4)
(Figure 2). In addition, the determination of cerebrovascular
GluT1 functional expression was carried out with competition
assay by perfusing [14C]D-glucose (0.2 µCi/mL) with 20 mM
concentration of glucose. This coperfusion resulted in a brain
uptake of 0.0011 ( 0.0003 mL/s/g (mean ( sd, n ) 3) (73.8%
inhibition) of [14C]D-glucose, thereby demonstrating functional
expression of cerebrovascular GluT1. The PA product was also
determined using 5 °C perfusion medium which resulted in
76.2% inhibition of the PA product of [14C]D-glucose to 0.001
( 0.0003 mL/s/g (mean ( sd, n ) 3). This further suggests
that the uptake of [14C]D-glucose in the in situ rat brain perfusion
test method was carrier-mediated because the carrier-mediated
uptake is reduced when the temperature is lowered.21,22
However, the uptake of [14C]D-glucose is not completely
inhibited by low temperature, which suggests that some activity
of GluT1 is still present at the BBB. There is also the possibility
that a part of the uptake of [14C]D-glucose is due to passive
diffusion. However, 80 µM 6 is able to inhibit the [14C]D-glucose
uptake almost completely (96.7% inhibition), which indicates
that there is no passive uptake of [14C]D-glucose present at the
BBB.
Therefore, both 4 and 6 demonstrated sufficient chemical
stability in aqueous solutions for further evaluation.
Compound 4 was susceptible to enzymatic hydrolysis, as it
was quantitatively cleaved to ketoprofen and glucose in 20%
rat brain homogenate and in 50% rat liver homogenate. The
enzymatic hydrolysis followed pseudo-first-order kinetics, the
half-lives being 43.5 ( 3.6 and 8.6 ( 0.6 min (mean ( sd, n
) 3) in brain homogenate and liver homogenate, respectively.
Therefore, 4 undergoes bioconversion to ketoprofen and glucose
in the brain tissue. However, 4 is also highly susceptible to
enzymatic hydrolysis in liver, which may compromise its
effective brain drug delivery in vivo. Enzymatic hydrolysis
studies of 6 in neither brain nor liver homogenate resulted in
the formation of indomethacin, which makes the prodrug not
suitable for brain delivery. Because indomethacin was not
released from the prodrug, the hydrolysis products and the half-
life of 6 were not identified. In an attempt to solve the problem,
we synthesized two prodrugs (10 and 14) with glycolic acid
and lactic acid linker group between glucose and indomethacin.
The synthesis of these prodrugs is described in the Supporting
Information. The degradation of prodrugs 10 and 14 was studied
in aqueous buffer solution of pH 7.4 at 37 °C. The degradation
of 10 and 14 followed pseudo-first-order kinetics with the half-
lives of 3.0 ( 0.2 and 9.5 ( 0.0 h (mean ( sd, n ) 2).
Therefore, 10 and 14 are not stable enough to be tested with in
situ rat brain perfusion technique. Although 6 did not release
the parent drug in brain tissue, it is reasonable to determine the
brain uptake of 6. Because 6 has higher molecular weight than
4, determining the uptake mechanism of both prodrugs may give
more insight into the GluT1 ability to transport molecules across
the BBB.
The ability of 4 and 6 to bind into GluT1 was studied by
coperfusing increasing concentrations of the prodrugs with

3350 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10
Gynther et al.
Figure 4. The brain uptake of 4 and 6. The uptake of 4 with 150 µM
concentration is 1.33 ( 0.18 pmol/mg/min and addition 50 mM
D-glucose to the perfusion medium decreased significantly (61.4%
inhibition) the brain uptake to 0.513 ( 0.009 pmol/mg/min (mean (
sd, n ) 3). Perfusion medium at 5 °C decreased significantly the uptake
of 4 to 0.515 ( 0.065 pmol/mg/min (mean ( sd, n ) 3) (61.3%
inhibition). The brain uptake of 150 µM 6 was 1.993 ( 0.429 pmol/
mg/min (mean ( sd, n ) 3), and after the addition of 50 mM D-glucose
to the perfusion medium, the uptake was 1.806 ( 0.353 pmol/mg/min
(mean ( sd, n ) 3). The decrease of the perfusion medium temperature
to 5 °C resulted in brain uptake of 0.599 ( 0.098 pmol/mg/min (mean
( sd, n ) 3) (69.9% inhibition). An asterisk denotes a statistically
significant difference from the respective control (**P < 0.01, *P <
0.05, one-way ANOVA, followed by Dunnett t-test). Columns 150 µM
4 and 150 µM 6 were used as controls in Dunnett t-test.
Figure 2. Mechanism of 4 and 6 rat brain uptake. The PA product of
0.2 µCi/mL [14C]D-glucose in absence or presence of D-glucose, low
temperature, 4 or 6. The control PA 0.0042 ( 0.0002 mL/s/g (mean (
sd, n ) 4) is decreased to 0.0011 ( 0.0003 (mean ( sd, n ) 3) (73.8%
inhibition) in the presence of 20 mM D-glucose and to 0.001 ( 0.0003
mL/s/g (mean ( sd, n ) 3) (76.2% inhibition), when using 5 °C
perfusion medium. The PA product of [14C]D-glucose decreased to
0.00103 ( 0.00014 mL/s/g (75.5% inhibition) and 0.00014 ( 0.00005
mL/s/g (96.7% inhibition) (mean ( sd, n ) 3) after perfusing the brain
with 4 and 6, respectively. After washing 4 from the brain capillaries,
the PA product of 0.2 µCi/mL [14C]D-glucose was 0.0033 ( 0.0001
mL/s/g (mean ( sd, n ) 3). The PA product of 0.2 µCi/mL [14C]D-
glucose was 0.0032 ( 0.0003 mL/s/g (mean ( sd, n ) 3) after 6 was
washed from the brain capillaries. An asterisk denotes a statistically
significant difference from the respective control (***P < 0.001, **P
< 0.01, *P < 0.05, Brown Fortsythe, followed by Dunnett T3-test).
compared to D-glucose could be due to the higher molecular
weight of the prodrugs. It is proposed that as the substrate binds
to the GluT1 binding site, the transporter protein cavity occludes
the bound substrate and then opens at the opposite side of the
membrane where the bound substrate can dissociate.23The steri-
cal hindrance caused by higher molecular weight of the prodrugs
could slow the conformational change of the transporter, and
as the transporters are occupied by the prodrugs, the transporters
are not able to facilitate the uptake of [14C]D-glucose. Therefore,
the low IC50 values only suggest that the prodrugs are able to
bind to GluT1 and the ability cross the BBB is not necessarily
achieved.
Brain Uptake Determination of the Prodrugs. In addition
to determining the binding of the prodrugs to GluT1, we also
determined the brain uptake of the prodrugs across the BBB.
The brain uptake of the prodrugs was determined with the in
situ rat brain perfusion technique using 150 µM concentration
at 60 s perfusion time (Figure 4). In the perfusion of 150 µM,
both 4 and 6 indeed resulted in detectable amounts of both
prodrug in brain tissue. The brain uptake of 150 µM 4 was 1.33
( 0.18 pmol/mg/min (mean ( sd, n ) 3), and an addition of
50 mM D-glucose to the perfusion medium decreased the brain
uptake 61.4% to 0.513 ( 0.009 pmol/mg/min (mean ( sd, n )
3) (Figure 4). Finally, brain uptake of 4 was determined at 150
µM concentration in a 5 °C perfusion medium. Lower temper-
ature decreased the uptake of 4 61.3% to 0.515 ( 0.065 pmol/
mg/min (mean ( sd, n ) 3). The brain uptake of 4 was also
determined with 450, 1000, 3500, and 15000 µM concentrations,
but the uptake was not saturable within this concentration range
(Figure 5). It is possible that because of poor affinity of 4 for
GluT1 and high capacity of the transporter (Km value of
D-glucose uptake is 11 mM and Vmax is 1420 nmol/min/g),12
the uptake did not saturate, although the uptake of 4 was mainly
carrier-mediated. Another explanation for the lack of uptake
saturation could be that passive diffusion of 4 is significant
enough to hide the saturation of Glut1-mediated uptake.
The brain uptake of 150 µM 6 was 1.993 ( 0.429 pmol/
mg/min (mean ( sd, n ) 3) (Figure 4) and after the addition
of 50 mM D-glucose to the perfusion medium the uptake was
1.806 ( 0.353 pmol/mg/min (mean ( sd, n ) 3). The
Figure 3. Inhibition of 0.2 µCi/mL [14C]D-glucose uptake by 4 and 6.
IC50 values are 32.85 ( 8.17 µM and 0.71 ( 0.04 µM for 4 and 6,
respectively. Data are mean ( sd (n ) 2). IC50 values are calculated
with nonlinear regression analysis using GraphPad Prism 4.0 for
Windows.
[14C]D-glucose. The prodrugs were able to inhibit the uptake of
[14C]D-glucose in a concentration-dependent manner (Figure 3).
Nonlinear regression analysis was used to determine the half
maximal inhibitory concentration (IC50) values of 4 and 6. The
IC50 values were 32.85 ( 8.17 µM for 4 and 0.71 ( 0.04 µM
for 6, which indicates that 6 has higher affinity for GluT1
compared to 4.
To further study the binding kinetics of the prodrugs to GluT1,
the PA product of [14C]D-glucose was determined after perfusing
rat brain first with the prodrugs at 80 µM for 30 s, followed by
washing the prodrug from the brain capillaries with 30 s
perfusion of prodrug-free perfusion medium and finally per-
fusing the rat brain with 0.2 µCi/mL [14C]D-glucose for 30 s.
This resulted in the PA products of [14C]D-glucose 0.0033 (
0.0001 mL/s/g and 0.0032 ( 0.0003 mL/s/g (mean ( sd, n )
3) for 4 and 6, respectively, indicating that the binding of 4
and 6 to the GluT1 is reversible (Figure 2).
These results show that 6 has higher affinity for GluT1 than
4, and both of the prodrugs have higher affinity for GluT1 than
D-glucose. This higher inhibition caused by the prodrugs

Brain Uptake of Ketoprofen and Indomethacin Prodrugs
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10 3351
determined from the whole brain is higher compared to brain
parenchyma (supernatant), a fraction of the prodrugs is captured
into the endothelial cells.
Conclusion
In this study, two glucose prodrugs 4 and 6 were synthesized
and their ability to cross the BBB via GluT1 was determined
with in situ rat brain perfusion technique. Both 4 and 6
demonstrated reversible concentration dependent inhibition of
brain uptake of the radiotracer [14C]D-glucose in the in situ rat
brain perfusion model, indicating that the prodrugs bind to the
GluT1. In addition, three factors strongly suggest that the brain
uptake of 4 and 6 is GluT1-mediated. First, both prodrugs were
able to cross the BBB and gain entry into the brain tissue.
Second, the uptake of 4 and 6 was decreased significantly when
5 °C perfusion medium was used, indicating that the uptake is
carrier-mediated. Furthermore, the high polar surface areas of
4 and 6 indicate that passive diffusion across the BBB is limited.
The uptake of 4 was decreased when 50 mM concentration
of D-glucose was added in to the perfusion medium. In addition,
the uptake of 4 did not saturate even in 15 mM concentration,
which suggests that the affinity of 4 for Glut1 is low and the
capacity of the transporter is high. It is also possible that passive
diffusion of 4 is significant enough to hide the saturation of
Glut1-mediated uptake. In the case of 6, the uptake did not
decrease when 50 mM D-glucose was added into the perfusion
medium. However, the unspecific inhibition of GluT1 using
lowered temperature suggests that the uptake of 6 is, at least,
partly GluT1-mediated. The lack of inhibition of 6 brain uptake
by 50 mM D-glucose indicates that the affinity of 6 for the
transporter is much higher than the affinity of D-glucose or 4.
This hypothesis is further supported by the low IC50 value of 6.
These results strongly suggest that a hydrophilic drug can be
attached to the hydroxyl group of D-glucose at the carbon
6-position and still maintain the affinity of the D-glucose
promoiety for the GluT1 transporter. However, glucose as a
promoiety has several limitations. The structure of glucose limits
the amount of drug molecules that can be linked with biode-
gradable bonds with it. Practically only drugs that bear a
carboxyl acid group can be linked without a linker/spacer with
glucose. In addition, the stability of ester prodrugs in systemic
circulation might not be adequate for clinical use of ester
prodrugs for drug brain targeting. Therefore this prodrug
technology demands further development before being fully
applicable to oral drug delivery. In summary, drug molecules
as large as indomethacin can be conjugated with D-glucose and
GluT1 is able to mediate the uptake of the conjugate across the
BBB into the brain parenchyma.
Figure 5. Relationship between concentration of the perfusion medium
and brain uptake of prodrug 4. The uptake of 4 was 1.33 ( 0.18, 3.55
( 0.33, 7.78 ( 0.68, 20.77 ( 0.78, and 93.21 ( 5.51 pmol/mg/min
(mean ( sd, n ) 3) using 150, 450, 1000, 3500, and 15000 µM
concentrations, respectively.
Table 2. Capillary Depletion Analysis after 60 s Perfusion of 4 and 6
with 450 µM Concentration, Followed by Washing the Prodrug from the
Capillaries with 30 s Perfusiona
right cerebrum
Vd (mL/g)
4
whole brain
supernatant
pellet
0.0054 ( 0.0002
0.0050 ( 0.0006
b
6
whole brain
supernatant
pellet
0.0155 ( 0.0044
0.0135 ( 0.0052
b
a Mean ( sd, n ) 3. b Below lower limit of detection.
decrease of the perfusion medium temperature to 5 °C
resulted in brain uptake of 0.599 ( 0.098 pmol/mg/min (mean
( sd, n ) 3) (69.9% inhibition). The uptake of 6 was slightly
higher than the uptake of 4, and the uptake was not decreased
by addition of D-glucose in the perfusion medium, unlike
the uptake of 4. Because addition of D-glucose did not affect
the uptake of 6, the uptake of 6 is either due to passive
diffusion or the affinity of 6 for GluT1 is much higher than
the affinity of D-glucose.
The passive diffusion of the prodrugs across the BBB is
probably limited because the polar surface areas of the prodrugs
are over 90 Å2 (Table 1).24 In addition, the uptake of both
prodrugs was significantly decreased by low temperature, which
indicates that the uptake of the prodrugs is carrier-mediated.
The low temperature could not inhibit the uptake of the prodrugs
entirely, which may indicate that the brain uptake of the prodrugs
is partly due to passive diffusion. However, the uptake of [14C]D-
glucose was not entirely inhibited by low temperature and
leaving some transporter activity present at the BBB. Therefore,
the fraction of passive diffusion of the whole brain uptake cannot
be determined without noncompeting GluT1 inhibitor. To our
knowledge, such an inhibitor suitable for in situ rat brain
perfusion does not exist.
Experimental Section
General Synthetic Methods. Commercial reagents were ob-
tained from major chemical suppliers and used without further
purification unless otherwise noted. 1H (500.13 MHz) and 13C
(125.77 MHz) NMR spectra were recorded on a Bruker Avance
500 spectrometer in CDCl3 and CD3OD. Chemical shifts (δ) are
reported in ppm relative to CDCl3 and CD3OD (7.26 and 3.31 for
1H and 77.0 and 49.0 for 13C), respectively. Electrospray ionization
mass spectra (ESI-MS) were obtained on a LCQ ion trap mass
spectrometer equipped with an electrospray ionization source
(Finnigan MAT, San Jose, CA). Elemental analyses (CHN) were
carried out with a Thermo Quest CE Instruments EA 1110 CHNS-O
elemental analyzer. Flash chromatography was performed on J. T.
Baker silica gel for chromatography (pore size 60 Å, particle size
50 µm). The synthesis of 2 and the prodrugs 4, 6, 10, and 14 are
described in Schemes 1-3, found in the Supporting Information.
Capillary Depletion Analysis. Capillary depletion analysis
of brain samples from perfused brain showed that the prodrugs
are present in the supernatant fraction, which consists of brain
parenchyma (Table 2). The concentration of the prodrugs in
the endothelial cell enriched pellet fraction was below the lower
limit of detection. However, because the uptake of the prodrugs

3352 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10
Gynther et al.
All tested compounds possessed purity of at least 95% and was
determined using combustion analysis method.
phosphate buffer solution of pH 7.4 (0.16 M, ionic strength 0.5) at 37
°C. An appropriate amount of prodrug was dissolved in 10 mL of
preheated buffer, and the solutions were placed in a thermostatically
controlled water bath at 37 °C. At appropriate intervals, samples were
taken and analyzed for remaining prodrug by HPLC. Pseudo-first-
order half-time (t1/2) for the hydrolysis of the prodrugs was calculated
from the slope of the linear portion of the plotted logarithm of
remaining prodrug versus time.
The rate of enzymatic hydrolysis of the prodrugs were studied at
37 °C in rat brain homogenate, which was diluted to 20% (v/v) with
isotonic 0.16 M phosphate buffer of pH 7.4 and in rat liver homogenate,
which was diluted to 50% (v/v) with isotonic 0.16 M phosphate buffer
of pH 7.4. The concentration of esterase enzymes in the homogenates
was not determined. The solutions were kept in a water bath at 37 °C,
and 0.2 mL of homogenate/buffer mixture was withdrawn and added
to 0.2 mL of acetonitrile to precipitate protein from the sample. After
immediate mixing and centrifugation, the supernatant was analyzed
for remaining prodrug and released ketoprofen or indomethacin by
the HPLC. Pseudo-first-order half-time (t1/2) for the hydrolysis of the
prodrugs was calculated from the slope of the linear portion of the
plotted logarithm of remaining prodrug against time.
Synthesis of Ketoprofen-Glucose Produg (4). 3,4,5,6-Tet-
rakis(trimethylsilyloxy)tetrahydro-2H-pyran-2-yl)methyl 2-(3-ben-
zoylphenyl) propanoate (3) (0.98 g, 1.39 mmol) was dissolved in
dichloromethane (20 mL), solution was cooled to 0 °C, and TFA
(6 mL) was added. After 4 h, NaHCO3 (2.34 g, 27.8 mmol) was
added to the solution and stirred 30 min, and the mixture was
concentrated. Raw material was purified by flash chromatography
eluting with dichloromethane, gradually increasing polarity to 85:
15 dichloromethane:methanol to give product as a white solid. Yield
0.50 g (86%). 1H NMR (Methanol-d4) δ 1.51 (d, J ) 6.76 Hz, 3H,
-CH3), 3.21-3.49 (m, 3H, CH), 3.67 (q, J ) 4.71 Hz, 2H, CH),
3.91-4.49 (m, 3H, CH + CH2), 5.02 (d, J ) 3.40 Hz, 1H, -CH),
7.48-7.79 (m, 9H, ArH). 13C NMR (Methanol-d4): δ 17.8 (-CH3),
45.1 (-CH), 63.9 (-CH2), 70.4 (-CH), 72.4 (-CH), 74.0 (-CH),
76.6 (-CH), 92.6 (-CH), 128.3 (2C, ArCH), 128.5 (ArCH), 128.6
(ArCH), 128.8 (ArCH), 129.8 (2H, ArCH), 131.9 (ArCH), 132.6
(ArCH), 137.4 (ArC), 137.7 (ArC), 141.3 (ArC), 174.4 (CO), 197.2
(CO). MS: m/z calcd for C22H24O8 [M]+ ) 416.2; found 434.0 [M
+ NH4]+. Elemental anal. calcd for C22H24O8 C: 63.45; H: 5.81;
found C22H24O8 C: 63.13; H: 5.91.
Synthesis of Indomethacin-Glucose Prodrug (6). Indometha-
cine (0.389 g, 1.09 mmol) and 1,2,3,4-tetra-O-trimethylsilyl-ꢀ-D-
glucopyranose (0.500 g, 1.07 mmol) and DMAP (0.013 g, 0.107
mmol) were dissolved in 25 mL of dichloromethane at room
temperature. DCC (0.253 g, 1.23 mmol) was added and the mixture
was stirred for 5 h at room temperature. Formed urea was filtered,
and filtrate was extracted with 5% acetic acid (20 mL), H2O (20
mL), and dried with Na2SO4 and concentrated. Formed solid (5)
was dissolved in acetonitrile (30 mL) and solution was cooled to 0
°C and TFA (0.75 mL) was added. After one hour, NaHCO3 (1.79
g, 21.3 mmol) was added to the solution and mixture was
concentrated. Raw material was purified by flash chromatography
eluting with dichloromethane, gradually increasing polarity to 80:
20 dichloromethane:methanol to give product as a white solid. Yield
0.30 g (71%).
Polar Surface Area. Polar surface areas of the prodrugs were
calculated using the method by Ertl et al.25 implemented in
Molecular Operating Environment.26
In Situ Rat Brain Perfusion Technique. Adult male Wistar
rats (200-230 g) were supplied by the National Laboratory Animal
Centre (Kuopio, Finland) for the rat brain perfusion studies. Rats
were anesthetized with ketamine (90 mg/kg, ip) and xylazine
(8 mg/mL, ip), and their right carotid artery system was exposed.
The right external carotid artery was ligated, and the right common
carotid artery was cannulated with PE-50 catheters filled with 100
IE/mL heparin. The in situ rat perfusion technique is described in
more detail in our previous study.21
Determination of the Brain Uptake Mechanism for the
Prodrugs. The ability of the prodrugs to bind into GluT1 was studied
with the in situ rat brain perfusion technique. The 100% PA product
of [14C]D-glucose was determined after 30 s perfusion of 0.2 µCi/mL
[14C]D-glucose solution. In a competition study, [14C]D-glucose (0.2
µCi/mL) was coperfused with increasing concentration of 4 or 6 for
30 s. The concentrations ranged from 0.1 to 1000 µM for 4 and 0.01
to 80 µM for 6. To study whether the binding of 4 and 6 to GluT1
was reversible, the PA product of [14C]D-glucose was determined after
perfusing rat brain first with 4 or 6 at 80 µM for 30 s, followed by
washing the prodrug from the brain capillaries with 30 s perfusion of
prodrug-free perfusion medium, and finally perfusing the brain with
0.2 µCi/mL [14C]D-glucose for 30 s.
Brain Uptake Studies of the Prodrugs. The prodrug brain
uptake studies were performed with the in situ rat brain perfusion
technique. The prodrugs were dissolved in dimethyl sulfoxide
(DMSO) and then added to the perfusion medium, resulting in 1%
(v/v) DMSO solution. After adjusting the pH to 7.4, the solution
was filtered with 0.45 µm Millex-HV filters. The rat brains were
perfused 60 s with 37 °C perfusion medium containing prodrug.
After perfusion, the remaining prodrug was washed from the brain
vasculature with cold prodrug-free perfusion medium (5 °C) for
30 s. The prodrug concentration of the perfusion medium was
analyzed by HPLC after each perfusion to confirm that prodrug
had stayed intact. The brain uptake was also determined by
inhibiting carrier-mediated uptake unspecifically with low temper-
ature. The brain capillaries were first washed for 30 s with cold
prodrug-free perfusion medium (5 °C), followed by perfusion for
60 s with cold prodrug solution (5 °C), and washed again with
cold prodrug-free perfusion medium (5 °C).
Capillary Depletion Analysis. Capillary depletion analysis was
carried out as previously described by Triguero et al.27 Brain samples
(right brain hemisphere) were weighed and homogenized in glass
homogenizer with 1.5 mL of physiological buffer. After homogeniza-
tion, 2 mL of 26% dextran solution was added and the mixture was
further homogenized. The homogenate was separated into two micro-
centrifuge tubes and centrifuged 15 min (5400g, 4 °C). The resulting
supernatant consisting of the brain parenchyma and the pellet rich in
1H NMR (Methanol-d4): δ 2.27 (s, 3H, -CH3), 3.10-3.32 (m,
2H, CH), 3.50-3.64 (m, 2H), 3.68 (s, 2H, CH2) 3.79 (s, 3H, -CH3),
3.94-4.50 (m, 2H, CH + CH2), 5.05 (d, J ) 3.00 Hz, 1H, -CH),
6.63 (dd, J ) 2.13 Hz, J ) 8.97 Hz, 1H, ArH), 6.82 (t, J ) 9.23
Hz, 1H, ArH), 6.88 (d, J ) 2.11 Hz, 1H, ArH), 7.51 (d, J ) 8.37
Hz, 2H, ArH), 7.63 (d, J ) 8.49 Hz, 2H, ArH).
13C NMR (Methanol-d4): δ 13.4 (-CH3), 30.5 (-CH2), 56.1
(-CH3), 65.2 (-CH2), 71.6 (-CH), 73.7 (-CH), 74.7 (-CH), 76.2
(-CH), 93.9 (-CH), 102.6 (ArCH), 112.5 (ArCH), 113.9 (C), 115.7
(ArCH), 130.1 (2C, ArCH), 131.8 (ArC), 131.9 (ArC), 132.2 (2C,
ArCH), 135.4 (C), 136.8 (ArC), 140.0 (ArC), 157.3 (ArC), 169.8
(CO), 172.7 (CO).
MS: m/z calcd for C25H26ClNO9 [M]+ ) 519.1; found 537.0 [M
+ H2O]+.
Elemental anal. calcd for C25H26ClNO9 C: 57.75; H: 5.04; N:
2.69; found C25H26ClNO9 ·3/2H2O C: 54.90; H: 5.34; N: 2.56.
HPLC Assay. The prodrug concentrations were analyzed by
Agilent 1100 HPLC system (Agilent Technologies Inc., Waldbronn,
Karlsruhe, Germany) that consisted of a binary pump G1312A, a
vacuum degasser G1379A, an automated injector system autosam-
pler Hewlett-Packard 1050, an UV-detector Hewlett-Packard 1050
variable wavelength detector, and an analyst software Agilent
ChemStation for LC Systems Rev. A.10.02. The detector wave-
length was set at 256 nm when 4 was analyzed and at 222 nm
when indomethacin prodrugs were analyzed. A mixture of aceto-
nitrile (50%) and a 0.02 M phosphate buffer solution of pH 6.0
(50%) at a flow rate of 1 mL/min was used as a mobile phase.
Ketoprofen samples were analyzed using a mixture of acetonitrile
(50%) and a 0.02 M phosphate buffer solution of pH 2.5 (50%).
Reversed-phase HPLC was conducted with a Zorbax RP-18 column
(150 mm × 4.6 mm, 5 µm, Agilent Technologies, Little Falls
Wilmington, DE).
Chemical and Enzymatic Stability of the Prodrugs. The rate
of chemical hydrolysis of the prodrugs were studied in aqueous

Brain Uptake of Ketoprofen and Indomethacin Prodrugs
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 10 3353
cerebral capillaries were separated and prepared for analysis with
HPLC. Volume of distribution (Vd) values for the homogenate,
supernatant fraction, and the capillary pellet were calculated.
References
(1) Pardridge, W. M. Blood-brain barrier drug targeting: the future of
brain drug development. Mol. InterV. 2003, 51, 90–105.
(2) Bradbury, M. W. The structure and function of the blood-brain barrier.
Fed. Proc. 1984, 43, 186–190.
Brain Sample Preparation. The quantification of 4 from the
rat brain samples was performed by analyzing the total concentration
of formed ketoprofen after its enzymatic release from 4 during 24 h
of incubation in 37 °C. Ketoprofen was isolated from the rat brain
samples by protein precipitation and solid phase chromatography.
A complete brain hemisphere was homogenated with 2.5 mL of
water to produce 3.0 mL of homogenate. The samples were acidified
with 300 µL of 2 M hydrochloric acid and vortexed for 5 min.
Then 4.0 mL of acetonitrile was added and the homogenates were
vortexed for 2 min. The homogenates were centrifuged for 10 min
(7500g, 20 °C), and the supernatants were collected. The super-
natants were diluted with 6.0 mL of water and applied to the
preconditioned and equilibrated C18 solid phase extraction car-
tridges (Discovery DSC-18; Supelco, Bellefonte, PA). The car-
tridges were first washed with 2.0 mL of 1% (v/v) acetic acid
solution to maintain ketoprofen in the un-ionized form and then
washed with 2.0 mL of water. The analytes were eluted with 6.0
mL of acetonitrile, and evaporated to dryness under a nitrogen
stream at 40 °C. Prior to analysis, samples were reconstituted in
50% (v/v) acetonitrile in water and filtrated. External standards were
used for the brain samples. The calibration curve of the brain
method was linear over a range of 0.2-10 nmol/brain hemisphere,
and samples containing high concentration of ketoprofen were
diluted prior to analysis. The lower limit of quantification for spiked
samples was 0.2 nmol of prodrug/brain hemisphere.
The quantification of 6 from the rat brain samples was performed
by isolating 6 from the rat brain samples by protein precipitation
and solid phase chromatography. A complete brain hemisphere was
homogenated with 2.5 mL of water to produce 3.0 mL of
homogenate. The samples were acidified with 300 µL of 2 M
hydrochloric acid and vortexed for 5 min. Then 4.0 mL of
acetonitrile was added and the homogenates were vortexed for 2
min. The homogenates were centrifuged 10 min (7500g, 20 °C)
and the supernatants were collected. This procedure was repeated.
Acetonitrile was evaporated from the supernatants under a nitrogen
stream at 40 °C, after which the supernatants were diluted with
6.0 mL of water and applied to the preconditioned and equilibrated
C18 solid phase extraction cartridges. The solid phase extraction
was performed like for 4. Prior to analysis, samples were
reconstituted in 50% (v/v) acetonitrile in water and filtrated. External
standards were used for the brain samples. The calibration curve
of the brain method was linear over a range of 0.3-5.0 nmol/brain
hemisphere. The lower limit of quantification for spiked samples
was 0.3 nmol of prodrug/brain hemisphere.
(3) Rubin, L. L.; Staddon, J. M. The cell biology of the blood-brain
barrier. Annu. ReV. Neurosci. 1999, 22, 11–28.
(4) Begley, D. J. The blood-brain barrier: principles for targeting peptides
and drugs to the central nervous system. J. Pharm. Pharmacol. 1996,
48, 136–146.
(5) Schinkel, A. H. P-Glycoprotein, a gatekeeper in the blood-brain
barrier. AdV. Drug DeliVery ReV. 1999, 36, 179–194.
(6) Begley, D. J. ABC transporters and the blood-brain barrier. Curr.
Pharm. Des. 2004, 10, 1295–1312.
(7) Loscher, W.; Potschka, H. Blood-brain barrier active efflux transport-
ers: ATP-binding cassette gene family. NeuroRx 2005, 2, 86–98.
(8) Pardridge, W. M.; Oldendorf, W. H. Transport of metabolic substrates
through the blood-brain barrier. J. Neurochem. 1977, 28, 5–12.
(9) Rautio, J.; Laine, K.; Gynther, M.; Savolainen, J. Prodrug approaches
for CNS delivery. AAPS J. 2008, 10, 92–102.
(10) Pardridge, W. M. Drug and gene delivery to the brain: the vascular
route. Neuron 2002, 36, 555–558.
(11) Farrell, C. L.; Pardridge, W. M. Blood-brain barrier glucose
transporter is asymmetrically distributed on brain capillary endothelial
lumenal and ablumenal membranes: an electron microscopic immu-
nogold study. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 5779–5783.
(12) Anderson, B. D. Prodrugs for improved CNS delivery. AdV. Drug
DeliVery ReV. 1996, 19, 171–202.
(13) Fernandez, C.; Nieto, O.; Fontenla, J. A.; Rivas, E.; de Ceballos, M. L.;
Fernandez-Mayoralas, A. Synthesis of glycosyl derivatives as dopam-
ine prodrugs: interaction with glucose carrier GLUT-1. Org. Biomol.
Chem. 2003, 1, 767–771.
(14) Halmos, T.; Santarromana, M.; Antonakis, K.; Scherman, D. Synthesis
of glucose-chlorambucil derivatives and their recognition by the human
GLUT1 glucose transporter. Eur. J. Pharmacol. 1996, 318, 477–484.
(15) Battaglia, G.; La Russa, M.; Bruno, V.; Arenare, L.; Ippolito, R.; Copani,
A.; Bonina, F.; Nicoletti, F. Systemically administered D-glucose conju-
gates of 7-chlorokynurenic acid are centrally available and exert anti-
convulsant activity in rodents. Brain. Res. 2000, 860, 149–156.
(16) Bonina, F.; Puglia, C.; Rimoli, M. G.; Melisi, D.; Boatto, G.; Nieddu,
M.; Calignano, A.; La Rana, G.; De Caprariis, P. Glycosyl derivatives
of dopamine and L-dopa as anti-Parkinson prodrugs: synthesis,
pharmacological activity, and in vitro stability studies. J. Drug Target.
2003, 11, 25–36.
(17) Mueckler, M.; Makepeace, C. Transmembrane segment 6 of the Glut1
glucose transporter is an outer helix and contains amino acid side chains
essential for transport activity. J. Biol. Chem. 2008, 283, 11550–11555.
(18) Fernandez, C.; Nieto, O.; Rivas, E.; Montenegro, G.; Fontenla, J. A.;
Fernandez-Mayoralas, A. Synthesis and biological studies of glycosyl
dopamine derivatives as potential antiparkinsonian agents. Carbohydr.
Res. 2000, 327, 353–365.
(19) Foster, K. A.; Roberts, M. S. Experimental methods for studying drug
uptake in the head and brain. Curr. Drug Metab. 2000, 1, 333–356.
(20) Summerfield, S. G.; Read, K.; Begley, D. J.; Obradovic, T.; Hidalgo,
I. J.; Coggon, S.; Lewis, A. V.; Porter, R. A.; Jeffrey, P. Central
nervous system drug disposition: the relationship between in situ brain
permeability and brain free fraction. J. Pharmacol. Exp. Ther. 2007,
322, 205–213.
(21) Gynther, M.; Laine, K.; Ropponen, J.; Leppanen, J.; Mannila, A.;
Nevalainen, T.; Savolainen, J.; Jarvinen, T.; Rautio, J. Large neutral
amino acid transporter enables brain drug delivery via prodrugs.
J. Med. Chem. 2008, 51, 932–936.
(22) Kageyama, T.; Nakamura, M.; Matsuo, A.; Yamasaki, Y.; Takakura,
Y.; Hashida, M.; Kanai, Y.; Naito, M.; Tsuruo, T.; Minato, N.;
Shimohama, S. The 4F2hc/LAT1 complex transports L-DOPA across
the blood-brain barrier. Brain. Res. 2000, 879, 115–121.
(23) Blodgett, D. M.; Carruthers, A. Quench-flow analysis reveals multiple
phases of GluT1-mediated sugar transport. Biochemistry 2005, 44,
2650–2660.
(24) Pajouhesh, H.; Lenz, G. R. Medicinal chemical properties of successful
central nervous system drugs. NeuroRx 2005, 2, 541–553.
(25) Ertl, P.; Rohde, B.; Selzer, P. Fast calculation of molecular polar
surface area as a sum of fragment-based contributions and its
application to the prediction of drug transport properties. J. Med. Chem.
2000, 43, 3714–3717.
Data Analyses. The results from the brain uptake experiments
are presented as mean ( sd of at least three independent experi-
ments. Statistical differences between groups were tested using
Brown Fortsythe, followed by Dunnett T3-test (Figure 2) and one-
way ANOVA, followed by two-tailed Dunnett t-test (Figure 4).
The normality of the data was tested using Shapiro-Wilk test. The
IC50 values (Figure 3) were determined by nonlinear regression
analysis using GraphPad Prism 4.0 for Windows. All statistical
analyses were performed using SPSS 14.0 for Windows.
Acknowledgment. This study was supported by the National
Technology Agency of Finland (TEKES), Graduate School in
Pharmaceutical Research, the Finnish Parkinson Foundation,
Finnish Cultural Foundation, North Savo Regional fund, and
the Academy of Finland (projects 108569, 214334 and 114721).
We also thank Tuomo Kalliokoski for polar surface area
calculations and Helly Rissanen and Miia Reponen for their
technical assistance.
(26) Molecular Operating EnVironment (MOE), Version 2008.10; Chemical
Computing Group Inc.; Montreal. Available at http://www.chemcomp.
com (accessed Feb 27, 2009).
(27) Triguero, D.; Buciak, J.; Pardridge, W. M. Capillary depletion method
for quantification of blood-brain barrier transport of circulating
peptides and plasma proteins. J. Neurochem. 1990, 54, 1882–888.
Supporting Information Available: General synthetic proce-
dures for compounds 1-3 and 7-14. This material is available
free of charge via the Internet at http://pubs.acs.org.
JM8015409