Department of Pharmaceutics and Pharmaceutical Chemistry,
University of Utah, Salt Lake City, Utah (M.D.J., J.C.); Division of
Pharmaceutical Sciences, College of Pharmacy, University of Kentucky,
Lexington, Kentucky (B.D.A.)
Enhanced central nervous system (CNS) delivery of certain poorly
penetrating 2',3'-dideoxynucleosides has been achieved by designing
prodrugs that are substrates for enzymes, such as adenosine deaminase
(ADA), that are present at high activities in brain tissue. In this
study, the potential role of adenosine deaminase localized within the
endothelial cells of the blood-brain barrier (BBB) in providing
enhanced intracellular and CNS delivery of an ADA-activated prodrug is
assessed in vitro using cell culture models of the BBB. The kinetics of
uptake and bioconversion of 2'-
-fluoro-2',3'-dideoxyadenosine
(F-ddA), a model ADA-activated prodrug of
2'--fluoro-2',3'-dideoxyinosine, were determined in primary cultured
bovine brain microvascular endothelial cells. Model-based
simulations of CNS availability derived from in vitro estimates of
parameters for simple passive diffusion and ADA-catalyzed deamination
suggest that ADA that is localized within the BBB plays an important
role in the conversion of F-ddA to 2'--fluoro-2',3'-dideoxyinosine during its passage across the BBB. Consistent with in vivo
observations, these simulations demonstrate that elevated levels of
certain enzymes, such as ADA, in the brain microvascular endothelial
cells of the BBB may be exploited in the design of brain-targeted
prodrugs or drug-carrier conjugates, which brain tissue selectively converts.
 |
Introduction |
Rational
approaches for increasing blood-brain barrier
(BBB2) penetration may be essential for those
dideoxynucleosides that exhibit reduced efficacy in the treatment of
HIV infections in the central nervous system (CNS). Theoretically, drug
entry into the CNS may be enhanced through the design of lipophilic
analogs that cross the BBB more rapidly by simple passive diffusion.
For instance, numerous modifications of the purine and/or the sugar rings of the anti-HIV dideoxynucleosides have been explored to increase
lipophilicity (Herdewijn et al., 1987
, 1988; Roey et al., 1989; Masood
et al., 1990; Barchi et al., 1991). However, most of these lipophilic
analogs exhibit diminished anti-HIV activity or are too cytotoxic to
use in humans.
A related alternate approach involves the preparation of bioconvertible
derivatives (prodrugs) with increased lipophilicity, which allows the
prodrug to be rapidly taken up into tissues and subsequently
metabolized to release the active parent compound (Roche, 1977; Stella
and Himmelstein, 1980; Anderson, 1996). Increased lipophilicity alone,
however, does not ensure higher concentration of the active drug at the
target tissue. Efficient targeting via the prodrug approach requires
the parent drug to be formed within the target organ at a rate
sufficient to compete with its elimination from the target organ
(Stella and Himmelstein, 1980, 1982, 1985). This is seldom achieved for
prodrugs that are brain directed (Greig, 1989). For example, a series
of lipophilic 5'-ester derivatives of 2',3'-dideoxyinosine (ddI) were
evaluated for their potential to improve the CNS delivery of ddI
(Anderson, 1996). None of the compounds evaluated gave
significantly higher CNS concentrations of ddI than when ddI was
administered alone because esterase activity in plasma far exceeds that
in brain tissue, resulting in premature bioconversion of the prodrug.
Recent findings that a series of lipophilic, adenosine
deaminase-activated 6-halo-2',3'-dideoxypurine nucleosides (i.e.,
prodrugs of ddI) yielded ddI concentrations in cerebrospinal
fluid and brain parenchyma, which were 5- and >20-fold higher
than when ddI was administered alone at the same infusion rate,
provided a promising new direction for the design of prodrugs of
certain anti-HIV dideoxynucleosides for CNS delivery (Anderson et al., 1992; Morgan et al., 1992). The enhanced CNS delivery of ddI achieved by ADA activation is believed to be a consequence of the relatively high activity of ADA in brain tissue, leading to a much improved selectivity in target tissue bioconversion compared with esterase activated prodrugs. These observations suggest that new enzyme-based strategies for improving the CNS delivery of anti-HIV agents are viable
(Anderson et al., 1992).
A related anti-HIV dideoxynucleoside,
2'--fluoro-2',3'-dideoxyadenosine (F-ddA) (Roth et al., 1998),
designed as an acid stable analog of 2',3'-dideoxyadenosine (ddA) with
a reduced deamination rate (Marquez et al., 1990; Masood et al., 1990),
has been shown to produce 5- and 20-fold higher concentrations of
F-ddATP and F-ddADP in human T-cells when compared with ddA controls
(Masood et al., 1990). Its anti-HIV active deamination product,
2'--fluoro-2',3'-dideoxyinosine (F-ddI), is equally acid-stable and
is a purine nucleoside phosphorylase (PNP)-resistant isostere of ddI.
The improved lipophilicity (10 times more lipophilic than F-ddI)
(Barchi et al., 1991), acid stability, and reduced deamination rate of
F-ddA, coupled with the PNP resistance of its deamination product, made
this compound a candidate for oral formulation, which is considered to
be the most practical route for chronic anti-AIDS therapy in a large patient population (Marquez et al., 1987, 1990).
F-ddA has been shown to enhance the CNS delivery of F-ddI in comparison
with results obtained by administering F-ddI alone in rats (Singhal et
al., 1997). After a 120-min infusion of F-ddA, the total
dideoxynucleoside (F-ddA plus F-ddI) brain parenchyma/plasma concentration ratio was approximately 8-fold higher than that obtained
from F-ddI after an equivalent F-ddI infusion. This is consistent with
the expected effect of lipophilicity in enhancing the CNS uptake of
F-ddA. Furthermore, the steady-state brain parenchyma/plasma concentration ratio of F-ddI was found to be 5-fold higher than that
obtained after an equivalent F-ddI infusion, suggesting that F-ddA was
indeed acting as an ADA-activated prodrug of F-ddI. Although
substantial levels of ADA exist in the brain tissue, with particularly
high activities in endothelial cells of the BBB (Nagy et al., 1984;
Mistry and Drummond, 1986; Schrader et al., 1987; Johnson and Anderson,
1996), the site of prodrug activation remains unclear. The present
study was undertaken to clarify the role of the BBB in the
bioconversion of F-ddA during its passage across the BBB, using uptake
studies into primary cultured bovine brain microvessel endothelial
cells (BBMECs). The contributions of other relevant transport
mechanisms (i.e., simple passive diffusion and carrier-mediated
transport) were also evaluated. Quantitation of processes, such as
passive diffusion, carrier-mediated transport, and prodrug metabolism
in vitro, may provide further insights into the role of the BBB in
governing CNS entry and prodrug bioconversion in vivo.
| |
Materials and Methods |
Chemicals and Reagents.
F-ddA, F-ddI, ddA, and
(+)-erythro-9-(2-hydroxy-2-nonyl)adenine (EHNA) were
provided by the National Cancer Institute (NIH, Bethesda, MD) and were
used as received. Additional quantities of F-ddI needed for the
monolayer and homogenate experiments were prepared from F-ddA using
adenosine deaminase (type VII) purified from calf intestinal mucosa
(Sigma Chemical Co., St. Louis, MO). Approximately 1 mg of ADA was
added to 25 ml of culture medium containing 20 mg of F-ddA at 25°C.
The formation of F-ddI was monitored via HPLC, as described below. When
the reaction was complete, the mixture was placed in a Centricon
Amicon Bioseparations; Millipore Corporation, Bedford, MA) concentrator
(a 100,000-mol. wt. cut-off) and centrifuged for 15 min. ADA activity
in aliquots of the ultrafiltrate was shown to be negligible. The F-ddI
solution was compared with the authentic reference standard for purity and subsequently used as prepared or otherwise diluted for use in
uptake studies.
Dulbecco's phosphate-buffered saline was purchased from Hyclone
Laboratories (Logan, UT). Minimum essential medium/F-12 Ham's nutrient
mixture (1:1) (Hyclone Laboratories) supplemented with 10 mM HEPES, 13 mM sodium bicarbonate, pH 7.4, 100 µg/ml penicillin G (Sigma Chemical
Co.), 100 µg/ml streptomycin (Sigma Chemical Co.), 100 µg/ml
heparin (170 units/mg), and 10% plasma-derived horse serum (HyClone
Laboratories) were used in the culture medium.
Isolation and Culture of Brain Microvascular Endothelial Cells.
Capillary segments were isolated from bovine cerebral gray matter (Dale
T. Smith & Sons Meat Packing Company, Draper, UT) by a two-step
enzymatic dispersion treatment followed by centrifugation over
pre-established 50% Percoll density gradients, as previously described
(Johnson and Anderson, 1996). Sequential filtering of the preparation
through 500- and 95-µm nylon mesh filters yielded a relatively
purified and homogeneous population of microcapillaries containing 5 to
20 individual endothelial cells per cluster. The isolated microvessels
were cultured immediately or stored at 
70°C in culture medium with
10% (v/v) dimethyl sulfoxide.
Rat tail collagen (type I) was derivatized to tissue culture dishes
(35 × 10 mm) with a cross-linking reagent,
1-cyclohexyl-3-(2-morpholinoethyl)carbodiimide-metho-p-toluenesulfonate (Aldrich Chemical Co., St. Louis, MO), to provide a uniform and durable
surface for cell attachment and growth that would withstand multiple
media changes and the rigors of the washing procedure (Macklis et al.,
1985). Brain microvessels were seeded onto the treated dishes at
approximately 50,000 cells/cm2 and cultured at
37°C under 95% humidity and 5% CO2/95% air.
Culture media were changed every other day. Monolayers reached
confluence in 8 to 10 days after initial plating.
Metabolism of F-ddA in BBMEC Homogenate.
The metabolism of F-ddA in the presence of BBMEC homogenate was
monitored to obtain preliminary information concerning the nature and
extent of F-ddA metabolism within the brain microvascular endothelial
cells. Feeding medium was removed from several dishes, and the
confluent monolayers were rinsed with Dulbecco's phosphate-buffered saline, pH 7.4, to remove any residual protein from the culture medium.
Sterile water (1 ml) was added, and the dishes were incubated at 37°C
under 95% humidity and 5% CO2/95% air for 15 min to loosen the monolayer. The monolayers were then scraped from the
dishes, collected in sterile 5-ml cryovials, and rapidly frozen at
70°C. Before use, thawed samples were further disintegrated by
mechanical homogenization with a Potter-Elvehjem tissue grinder
(0.10-0.15-mm clearance; Kimble/Kontes, Vineland, NJ) to form a
homogeneous suspension.
Adenosine deaminase activity was measured using F-ddA as a reference
substrate in pH 7.4 phosphate buffer (8.4 mM). The reaction was
initiated by adding 0.05 ml of homogenate to 0.45 ml of warm (37°C)
phosphate buffer containing F-ddA (55 µM-10 mM) and incubated at
37°C. The reaction was quenched by the addition of a 0.10-ml aliquot
of the reaction solution to 1 ml of ice-cold acetonitrile. The
acetonitrile was then evaporated under an N2
stream. The residue was then reconstituted with 0.5 ml of phosphate
buffer (8.4 mM), pH 7.4, and analyzed immediately by HPLC. Initial
rates of F-ddI formation were fit to the conventional Michaelis-Menten
equation via nonlinear least-squares regression analysis (Scientist;
MicroMath, Inc., Salt Lake City, UT) to obtain estimates of the kinetic
parameters V
and
K
.
Uptake, Efflux, and Metabolism in BBMEC Monolayers.
Monolayer uptake experiments were conducted using a previously
published method (Johnson and Anderson, 1999) to examine the relative
contributions of passive diffusion, metabolism, and possible carrier-mediated transport to the kinetics of F-ddA and/or F-ddI accumulation within endothelial cells. Briefly, F-ddA uptake was initiated by the addition of 1 ml of warm culture medium, pH 7.4, or 1 ml of endothelial assay buffer (122 mM NaCl, 25 mM
NaHCO3, 10 mM D-glucose, 3 mM KCl,
1.2 mM MgSO4, 0.4 mM
NaHPO4, 1.4 mM CaCl2, and
10 mM HEPES adjusted to pH 7.4 with NaOH) (Shah et al., 1989)
containing F-ddA (0.1-10 mM) with or without EHNA (0.7 µM), a potent
and selective adenosine deaminase inhibitor (Harriman et al., 1994),
followed by incubation at 37°C under 95% humidity and 5%
CO2/95% air. At selected times, uptake
experiments were terminated by rapidly removing donor solution followed
by sequential washes (4 × 50 ml) with ice-cold Dulbecco's
phosphate-buffered saline, pH 7.4, to remove residual extracellular- or
surface-bound permeant. Cell monolayers were then solubilized with 1 N
NaOH (0.5 ml for 15 min at room temperature) and neutralized with an equivalent volume of 1 N HCl. Aliquots were collected and analyzed for
both permeant and protein content. Intracellular concentrations (Cmonolayer) were
calculated by dividing the intracellular mass (millimoles) by the
monolayer volume (~1.2 µl) based on protein content (Johnson and
Anderson, 1999).
Extracellular ADA activity in the donor solution was periodically
measured using F-ddA as a reference substrate in endothelial assay
buffer, pH 7.4, at 37°C. In preparation for these experiments, the
culture medium was removed from the monolayer and replaced with
endothelial assay buffer containing F-ddA (0.4 mM) and incubated at
37°C under 95% humidity and 5% CO2/95% air.
After 90 min, the donor solution was removed from monolayer contact.
The rate of formation of F-ddI in the donor solution was monitored both
before and after removal from the monolayer.
Both uptake and efflux experiments were also conducted for F-ddI. The
uptake of F-ddI was likewise initiated by the addition of warm culture
medium (1 ml), pH 7.4, containing F-ddI (0.45-4.5 mM) followed by
incubation at 37°C under 95% humidity and 5%
CO2/95% air. The efflux of F-ddI was initiated
after incubating the cells with warm culture medium (1 ml), pH 7.4, containing F-ddI (0.45-4.5 mM) for 1 h followed by a rapid
replacement of the donor solution with drug-free medium (2 ml), pH 7.4. Sequential washes (4 × 50 ml) with ice-cold Dulbecco's
phosphate-buffered saline, pH 7.4, were used to remove most residual
extracellular or surface bound permeant, and the intracellular F-ddI
concentration and protein content were analyzed.
Analytical Procedures.
Resolution of F-ddA and F-ddI was achieved using a modular
reversed-phase HPLC system (Supelcosil LC-18-S analytical column; 25-cm × 4.6-mm i.d., 5 µm) with UV detection at 254 nm.
Depending upon the experimental application, the mobile phase consisted of variable amounts of acetonitrile (typically 5-20%) in phosphate buffer (pH 7.4; I = 0.02) at a flow rate of 1.5 ml/min.
In a typical separation at 10% acetonitrile, the retention times of
F-ddI and F-ddA were 5.7 and 11.7 min, respectively. Protein
content was determined by the Lowry method (Lowry et al., 1951) using
protein standards prepared from bovine serum albumin (Sigma
Diagnostics, St. Louis, MO).
Kinetic Model for Monolayer Uptake, Efflux, and Metabolism.
The concentration versus time course for F-ddI in endothelial cells
exposed to donor solutions containing only F-ddI could be described in
terms of a single apparent rate constant,
k
, for passive entry into and efflux
from the monolayer as outlined in eq. 1:
|
(1)
|
where C
is the concentration of
F-ddI in the donor solution and
C
is the intracellular
concentration. For efflux experiments, only the second term in eq. 1
was used (C = 0).
Although several models were explored in fitting the F-ddA and F-ddI
concentration versus time profiles in monolayers exposed to donor
solutions containing only F-ddA, the simplest model consistent with
both F-ddI and F-ddA uptake data is described by eqs. 2 to 5:
|
(2)
|
where Vmax (units of molar
concentration min
1) and
Km (molar concentration) are the
Michaelis-Menten parameters for intracellular ADA-catalyzed metabolism
of F-ddA, C
is the concentration of
F-ddA in the donor solution, and
funbound represents the apparent
fraction of unbound intracellular F-ddA. The first-order rate constant
for passive diffusion, k
, is related to
the permeability coefficient of F-ddA across the collective cell
membrane of the monolayer (P
) as follows:
|
(3)
|
where Amonolayer represents the
surface area of the collective cells that constitute the monolayer and
Vmonolayer represents the monolayer
volume. In the absence of metabolism (i.e., in the presence of EHNA),
eq. 2 reduces to the following:
|
(4)
|
Likewise, a similar relationship can be derived for the rate of
intracellular accumulation of the metabolite F-ddI
(C) in donor solutions
containing only F-ddA:
|
(5)
|
The relative contributions of passive diffusion, metabolism, and
nonspecific binding to the overall transfer of F-ddA into and out of
BBMECs were quantified by simultaneously fitting the combined data sets
(n = 7) from uptake experiments using F-ddI and F-ddA,
as described below. In all cases, the monolayer uptake and/or efflux of
F-ddI was described by eq. 1, whereas the monolayer uptake and
metabolism of F-ddA were described by eqs. 2 to 5. Each concentration
versus time profile was described by its own differential equation with
shared parameter estimates for k, k, the metabolism parameters
(Vmax and Km), and the apparent unbound fraction
of F-ddA (funbound) via nonlinear
least-squares regression analysis.
When the intracellular concentrations of F-ddI and F-ddA are at steady
state in uptake experiments conducted with F-ddA-containing donor
solutions, the total dideoxynucleoside (F-ddA plus F-ddI) intracellular/donor concentration ratio can be obtained by setting eqs.
2 and 5 equal to 0:
|
(6)
|
where C can be expressed
in terms of C by solving eq. 2 with
dC/dt = 0:
|
(7)
|
Equation 6 predicts that since the enzymatic conversion of F-ddA
to F-ddI is a saturable process, the ratio of total dideoxynucleosides (F-ddA plus F-ddI) intracellular concentration versus F-ddA donor concentration should decrease with increasing F-ddA donor concentration.
In Vivo Simulations of CNS Delivery Based on in Vitro
Observations in BBMECs.
To assess the significance of various BBB metabolic processes in the
CNS availability of F-ddA, the kinetic model illustrated in Fig.
1 provides useful insight. In this
simplified model, the initial rate of F-ddA uptake from blood into the
brain parenchyma is assumed to be governed by simple passive diffusion
(kp) and metabolic clearance within
the endothelial cells of the BBB
(kmet). Allowing the F-ddA
concentration within the BBB to be at steady state and assuming the
backflux of F-ddA into the BBB from the brain parenchyma to be
negligible (initial uptake region), the fraction of F-ddA leaving the
plasma that enters the brain parenchyma intact can be expressed as an
availability estimate.
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 1.
Compartmental diagram illustrating the
kinetic determinants of F-ddA and F-ddI entry into the central nervous
system (i.e., brain parenchymal tissue) across BBB.
|
|
Quantitatively, the CNS availability of F-ddA is defined as the ratio
of its rate of appearance in the brain parenchyma
(dXbr/dt) to its rate
of loss from plasma attributable to passage across the blood-brain
barrier, (dXpl/dt), where
the rates of entry into brain and loss from plasma can be expressed in
terms of the BBB permeability coefficient
(PBBB), BBB surface area
(ABBB), Michaelis-Menten kinetic
constants (Vmax and
Km), and BBB and plasma concentrations
(CBBB and
Cpl), respectively:
|
(8)
|
|
(9)
|
Assuming steady state within the BBB and sink conditions within
the brain parenchyma, the F-ddA concentration within the BBB
(CBBB), at any given concentration in
the plasma (Cpl), can be estimated via
the following relationship:
|
(10)
|
Inserting values of CBBB into
eqs. 8 and 9 and then taking their ratio leads to estimates of CNS
availability, as defined in eq. 11:
|
(11)
|
To determine whether the bioconversion of F-ddA to F-ddI within
the BBB could enhance the concentration of F-ddI in the brain parenchymal tissue (i.e., the classical prodrug effect), the simulated flux of F-ddI into the brain tissue, after F-ddI infusion, was compared
with that obtained after an equivalent infusion of F-ddA. In the
kinetic model depicted in Fig. 1, the efflux of F-ddI is presumed to be
primarily governed by passive diffusion. The flux of F-ddI into the
brain (dX
/dt) is assumed to
be equivalent to the intracellular efflux and can therefore be
expressed in terms of the permeability-area product
(P
A) and BBB concentration
(C):
|
(12)
|
Assuming steady state within the BBB and sink conditions within
the brain parenchyma, the F-ddI concentration within the BBB, at any
given F-ddI concentration in the plasma
(Cpl), can be evaluated by the
following relationship by setting
dX/dt =0:
|
(13)
|
Inserting values of C into eq.
12 then provides an estimate of F-ddI flux into the brain. In a similar
fashion, the flux of F-ddI after an equivalent infusion of F-ddA can be
estimated from the kinetic considerations of the model presented in
Fig. 1 and the previously derived relationships with their accompanying assumptions.
Statistical and Regression Analysis.
Statistical significance was determined using a two-tailed Student's
t test for unpaired data. Values were determined to be significantly different at P 
0.05. Nonlinear
least-squares regression analysis was performed using a computer and
commercially available software (SCIENTIST; MicroMath, Inc.). The model
selection criterion of SCIENTIST, a modification of the Akaike
information criterion (Akaike, 1976), was used to select the most
appropriate model (highest model selection criterion).
| |
Results |
Metabolism of F-ddA in BBMEC Homogenate and Extracellular Media.
Illustrated in Fig. 2 are the initial
rates of F-ddI formation from various concentrations of F-ddA at 37°C
in BBMEC homogenates from primary cultures. Michaelis-Menten behavior
was observed over the entire range of the F-ddA concentration. The
kinetic parameters Km (mean ± S.D.) and Vmax (mean ± S.D.)
obtained from the best fits of the initial velocity versus F-ddA
concentration data to the Michaelis-Menten equation were 1505 ± 153 µM and 39 ± 1 µM min1
mg1 of protein, respectively. The rate of F-ddI
formation from F-ddA added to medium after 90 min of exposure of the
medium to cell monolayers followed by separation from the cells was
~15% of the rate obtained in the monolayer's presence (data not
shown).
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 2.
Initial rates of F-ddI formation from
varying concentrations of F-ddA at 37°C in cellular homogenate
obtained from primary cultures of bovine brain microvessel endothelial
cells.
|
|
Uptake, Efflux, and Metabolism in BBMEC Monolayers.
The BBMEC permeability characteristics of F-ddA were assessed in the
presence of the potent ADA inhibitor EHNA. An EHNA concentration of 0.7 µM was sufficient to completely suppress the intracellular formation
of F-ddI. Profiles of the intracellular uptake of F-ddA, in the
presence of EHNA, are displayed in Fig.
3. The solid lines represent computer
fits of the individual data sets to eq. 4
(V = 0). The rates of influx
and efflux of F-ddA are both rapid and concentration-independent over a
>10-fold concentration range. At steady state, the intracellular F-ddA
concentrations exceeded the donor concentration by nearly 40%.
In the absence of EHNA, intracellular degradation of F-ddA occurred to
form F-ddI. The concentration dependence of F-ddA uptake and F-ddI
accumulation in cells incubated in F-ddA containing donor solutions is
displayed in Fig. 4. The solid lines
represent the best fit when all data sets were fit simultaneously to a
model containing simple passive diffusion, metabolism, and nonspecific binding of F-ddA and passive efflux of F-ddI as expressed in eqs. 2 and
5. The parameters generated from these fits are listed in Table
1. Rates of F-ddI uptake and efflux were
also determined in separate experiments in the presence of donor
concentrations of F-ddI ranging from 0.45 to 4.5 mM (curves not shown).
The value of k generated from
regression analyses of the combined data from these experiments is also
listed in Table 1. The uptake data obtained at 60 min in the presence
of F-ddA-containing donor solutions were used to plot
(C + C)/C
versus C as shown in Fig.
5.
View this table:
[in this window]
[in a new window]
|
TABLE 1
Kinetic parameters obtained from nonlinear regression analyses of the
concentration versus time data for F-ddA uptake (0.3-10 mM) and F-ddI
accumulation in monolayers exposed to F-ddA-containing donor solutions
and F-ddI uptake/efflux in monolayers exposed to F-ddI (3 mM)-containing donor solutions
Mean ± S.D. obtained from computer fits of the data.
|
|
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 5.
The intracellular total dideoxynucleoside
(F-ddA + F-ddI)/donor F-ddA concentration ratio at steady state versus
the donor concentration of F-ddA.
The solid line is the fitted line using eq. 6 and Table 1 parameters.
|
|
The concentration-dependent profiles of F-ddI intracellular
accumulation in monolayers exposed to donor solutions containing either
F-ddA (at 0.3, 3, or 9.8 mM) or F-ddI (3 mM) are highlighted in Fig.
6. Nearly equal intracellular levels of
F-ddI were obtained from a 3 mM F-ddI donor concentration or from a
10-fold lower donor concentration of F-ddA (0.3 mM). Identical
half-lives (t1/2 = 8 min) for the
various fitted curves representing F-ddI accumulation despite a nearly
10-fold change in intracellular concentration of F-ddI reflect the
model assumption that k is
concentration-independent.
Kinetic parameters obtained from in vitro monolayer uptake experiments
were used in conjunction with eqs. 8 to 11 to conduct in vivo
simulations of CNS availability of F-ddA as a function of F-ddA
concentration in plasma. The permeability-surface area product [PBBBABBB = (k
Vdmonolayer)/2] was estimated to be 7.9 × 104 ml min1 based on
the k listed in Table 1 and Vdmonolayer, the apparent BBMEC distribution
volume in the monolayer (i.e., Vmonolayer/funbound 
2 × 103 ml). A simulation of the CNS
availability of F-ddA versus its plasma concentration is shown in Fig.
7.
Figure 8 displays the results of
simulations of F-ddA and F-ddI flux into brain parenchyma normalized to
the plasma concentration of dideoxynucleoside versus the steady-state
plasma concentrations of either F-ddA or F-ddI. As shown in Fig. 8,
F-ddA provides ~15-fold higher flux of total dideoxynucleoside into
the brain when compared with equivalent infusions of F-ddI. In the
lower concentration regions, deamination of F-ddA by ADA located within
the BBB reduces brain levels of intact F-ddA but provides nearly
10-fold higher brain concentrations of F-ddI relative to F-ddI
controls.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 8.
Simulation of in vivo dideoxynucleoside
fluxes into brain parenchymal tissue, normalized to steady-state plasma
concentration, versus steady-state plasma concentrations after i.v.
infusions to steady state.
F-ddI after F-ddI infusion (solid line), F-ddI after F-ddA
infusion (short-dashed line), F-ddA after F-ddA infusion (long-dashed
line), and total dideoxynucleoside (F-ddI and F-ddA) after F-ddA
infusion (medium-dashed line).
|
|
| |
Discussion |
Role of the BBB in the Enhanced CNS Delivery of ADA-Activated
Prodrugs
Evidence from Monolayer Experiments.
The brain microvascular endothelial cells of the cerebral
microvasculature seem to be enriched in a number of enzymes that may
play an important role in governing the entry of many chemicals from
the blood into CNS (Pardridge, 1983). In a recent report, we quantified
the activity of two key purine-metabolizing enzymes, PNP and ADA, and
found their activities to be significantly higher in brain
microvascular endothelial cells relative to whole cerebral gray matter
(Johnson and Anderson, 1996). As a key enzyme in purine metabolism,
adenosine deaminase catalyzes the irreversible hydrolytic deamination
of adenosine to produce inosine and ammonia (Zielke and Suelter, 1971).
In addition, this enzyme also catalyzes the hydrolysis of many analogs
of adenosine altered in either the purine ring or sugar moiety (Chassy
and Suhadolnik, 1967; Barchi et al., 1991; Shirasaka et al., 1991; Ford
et al., 1995). Previous studies conducted in these laboratories have
confirmed that adenosine deaminase-activated prodrugs, such as
6-chloro-2',3'-dideoxypurine and F-ddA significantly enhance the CNS
delivery of ddI and F-ddI, respectively, due to higher brain tissue
activities of ADA compared with plasma (Morgan et al., 1992; Singhal et
al., 1997). However, the site of prodrug activation (i.e., the brain
microvascular endothelial cells versus the brain parenchymal tissue) is
yet to be determined. The results provided here demonstrate that
dideoxynucleoside uptake in the CNS can be enhanced not only through
the design of prodrugs that are activated in brain parenchyma, the
classical prodrug approach, but also via prodrugs that are activated by enzymes localized within brain microvascular endothelial cells.
Figure 2 confirms previous findings of substantial ADA activity in
blood-brain barrier endothelial cells (Johnson and Anderson, 1996).
This activity is completely inhibited by EHNA at a donor concentration
of 0.7 µM. As shown in Fig. 3, in the absence of metabolism catalyzed
by ADA, the F-ddA intracellular concentration versus time profiles are
superimposable over a concentration range of 0.1 to 6.3 mM. This
observation is highly suggestive of a membrane permeation mechanism
that is governed primarily by passive diffusion, as supported by
numerous studies indicating that passive diffusion seems to be the
major mechanism for membrane transport of ddA and F-ddA (Ahluwalia et
al., 1987; Agarwal et al., 1989; Plagemann and Woffendin, 1989; Masood
et al., 1990; Domin et al., 1993). The finding that the intracellular
concentration of F-ddA at steady state exceeds the donor concentration
by approximately 40% was taken as evidence for nonspecific (i.e.,
nonsaturable) binding to intracellular components and was accounted for
in the model equations by the parameter
funbound. However, other nonsaturable processes that produced unequal clearances into and out of the cells
cannot be ruled out based on these data.
The fitted curves in Figs. 3 to 6 representing parameter values from
Table 1 are close to the experimental data, consistent with the model
assumptions. The total dideoxynucleoside (F-ddA + F-ddI)
intracellular/donor concentration ratio at steady state decreases with
increasing donor F-ddA concentration (Fig. 5), consistent with eq. 6.
Since the ratio
C/C at steady state was nearly the same with the increasing donor concentration in the presence of EHNA (Fig. 3), saturation of ADA at
high F-ddA concentrations in the absence of EHNA seems to be the
primary factor leading to the reduced ratio of total intracellular/extracellular dideoxynucleoside.
The rate of F-ddI formation in media to which F-ddA was added after the
medium was incubated with cell monolayers for 90 min and then removed
was ~15% of that obtained in the monolayer's presence. These
results suggest that although some conversion of F-ddA to F-ddI occurs
in the donor solution after extended exposure to cells, presumably due
to ADA that has been released from damaged cells, this rate of
hydrolysis is negligible over the time of a typical uptake experiment
(30-60 min) and cannot account for the intracellular accumulation of
F-ddI.
Figure 6 illustrates the advantage of the prodrug approach for
increasing intracellular delivery of F-ddI. The intracellular concentration of F-ddI attained from the incubation of cells with 0.3 mM F-ddA is comparable to that obtained from a 3 mM donor concentration
of F-ddI itself. As shown in subsequent computer simulations, higher
CNS delivery of F-ddI would also be expected due to conversion of F-ddA
by ADA localized within the endothelial cells of the BBB.
In Vivo Simulations of the Role of BBB Bioconversion in Enhancing
CNS Delivery.
The availability estimates obtained from computer simulations plotted
in Fig. 7 predict a wide variety of possible in vivo outcomes. Of
particular interest is the very notable effect of BBB metabolism
catalyzed by ADA. At low plasma concentrations similar to those
explored in vivo (6.4 × 105 M) (Singhal
et al., 1997), the solid line in Fig. 7 indicates that the percentage
of F-ddA arriving intact in brain tissue is approximately 20% of that
absorbed across the BBB, a prediction consistent with the low CNS
availability of F-ddA observed in vivo in rats. In vivo, the CNS uptake
of F-ddA after a 120-min i.v. infusion was low, with only 5% of the
plasma concentration of F-ddA present in the brain tissue at steady
state (Singhal et al., 1997).
The simulations of F-ddA and F-ddI brain fluxes normalized to the
plasma dideoxynucleoside concentration (Fig. 8) also seem to agree
remarkably well with in vivo observations (Singhal et al., 1997),
suggesting that F-ddA may function in part as a BBB-activated prodrug
of F-ddI due to the localization of ADA within the endothelial cells of
the BBB. In vivo, Singhal et al. (1997) found that after a 120-min i.v.
infusion of F-ddA, the concentration of total dideoxynucleoside (F-ddA
and F-ddI) in rat brains was ~8-fold higher when compared with an
equivalent infusion of F-ddI. Furthermore, deamination of F-ddA by
brain tissue ADA to form F-ddI reduced CNS levels of intact F-ddA but
provided higher brain parenchymal (5-fold) ratios of F-ddI relative to
F-ddI controls.
In the kinetic model depicted in Figure 1, which served as the basis
for the simulations in Figs. 7 and 8, the CNS availability of F-ddA
(eq. 11) is determined in large part by the relative contributions of
BBB permeability (PBBB) and
intracellular conversion catalyzed by ADA. In the absence of
ADA-catalyzed metabolism, however, eq. 11 predicts that the CNS
availability of F-ddA should be 100%. It is interesting to note that
although the brain tissue/plasma concentration ratios of F-ddA in rats
pretreated with the ADA inhibitor 2'-deoxycoformycin increased 1.4-fold
in comparison to rats infused with only F-ddA (Singhal et al., 1997),
this enhanced ratio was still 10-fold lower than that for a
nonmetabolizable polar permeant, urea (Morgan et al., 1996).
Furthermore, the apparent kout/kin
ratios obtained in vivo for both F-ddA and F-ddI (12 and 111, respectively) were nearly 8- and 70-fold greater than that of urea,
respectively. These observations suggest that active transport
processes that were not evident in the monolayer experiments may also
be operative in vivo.
In conclusion, this study has demonstrated the versatility of the
monolayer uptake model for probing the relative contributions of
biological processes governing prodrug transport and bioconversion. Simulations of brain availability have helped clarify the role of ADA
localized within the BBB in the overall efficiency of targeting the
ADA-activated prodrug F-ddA to the brain. Consistent with in vivo
observations, these simulations, based upon in vitro parameter estimations, have demonstrated that elevated levels of certain enzymes,
such as ADA, in the brain microvascular endothelial cells may be
exploited in the design of brain-targeted prodrugs or drug-carrier conjugates that undergo brain tissue-selective bioconversion.
This work was supported by National Institutes of Health Grants
AI-34133 and NS-39178. Mark Johnson also received support as a
recipient of an Advanced Predoctoral Fellowship in Pharmaceutics from
the Pharmaceutical Research and Manufacturers of America Foundation.
Dr. Bradley D. Anderson,
Division of Pharmaceutical Sciences, College of Pharmacy, University of
Kentucky, 907 Rose Street, Room 327G, Lexington, KY 40536. E-mail:
bande2{at}pop.uky.edu
Abbreviations used are:
BBB, blood-brain
barrier;
HIV, human immunodeficiency virus;
CNS, central nervous
system;
ddI, 2',3'-dideoxyinosine;
ADA, adenosine deaminase;
F-ddA, 2'--fluoro-2',3'-dideoxyadenosine;
ddA, 2',3'-dideoxyadenosine;
F-ddI, 2'--fluoro-2',3'-dideoxyinosine;
PNP, purine nucleoside
phosphorylase;
BBMEC, bovine brain microvascular endothelial cell;
EHNA, (+)-erythro-9-(2-hydroxy-2-nonyl)adenine;
HPLC, high-performance liquid chromatography.
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
-Fluoro-2',3'-Dideoxyinosine Via a Blood-Brain
Barrier Adenosine Deaminase-Activated Prodrug
Top
Abstract
Introduction
), 0.6 mM F-ddA (
), and 6.3 mM F-ddA
(
). The single value obtained at 0.1 mM F-ddA (mean ± S.D.)
represents duplicate measurements at the respective time point.
