Intracerebral microdialysis probe recovery (extraction fraction) may be influenced by several mass transport processes in the brain, including efflux and uptake exchange between brain and blood. Therefore, changes in probe recovery under various experimental conditions can be useful to characterize fundamental drug transport processes. Accordingly, the effect of inhibiting transport on probe recovery was investigated for two capillary efflux transporters with potentially different membrane localization and transport mechanisms, P-glycoprotein and an organic anion transporter. Fluorescein/probenecid and quinidine/LY-335979 were chosen as the substrate/inhibitor combinations for organic anion transport and P-glycoprotein-medicated transport, respectively. Probenecid decreased the probe recovery of fluorescein in frontal cortex, from 0.21 ± 0.017 to 0.17 ± 0.020 (p < 0.01). Quantitative microdialysis calculations indicated that probenecid treatment reduced the total brain elimination rate constant by 3-fold from 0.37 to 0.12 (ml/min · ml of extracellular fluid). In contrast, the microdialysis recovery of quinidine, delivered locally to the brain via the probe perfusate, was not sensitive to P-glycoprotein inhibition by systemically administered LY-335979, a potent and specific inhibitor of P-glycoprotein. Recovery of difluorofluorescein, an analog of fluorescein, was also decreased by probenecid in the frontal cortex but not in the ventricle cerebrospinal fluid. These experimental observations are in qualitative agreement with microdialysis theory incorporating mathematical models of transporter kinetics. These studies suggest that only in certain circumstances will efflux inhibition at the blood-brain barrier and blood-cerebrospinal fluid barrier influence the microdialysis probe recovery, and this may depend upon the substrate and inhibitor examined and their routes of administration, the localization and mechanism of the membrane transporter, as well as the microenvironment surrounding the probe.
Microdialysis is a powerful technique that enables continuous monitoring of drug concentration in the extracellular space of various tissues in the same animal. It is well suited for pharmacokinetic studies and has been widely applied to assess drug distribution to target tissues (Elmquist and Sawchuk, 1997). Since microdialysis sampling is a non-equilibrium process, the drug concentration in the dialysate is some fraction of the actual extracellular free drug concentration, and the ratio of the concentration of the drug in the dialysate to the concentration of drug in the extracellular fluid is defined as the recovery (or extraction fraction) of the microdialysis probe. This microdialysis probe recovery is influenced by experimental factors such as the dialysate flow rate, dialysis membrane composition, effective surface area, and probe geometry. For microdialysis in vivo, the tissue environment can have additional effects on recovery. Processes eliminating solutes from the extracellular fluid (metabolism and brain-to-blood efflux), as well as diffusional properties (diffusion coefficient, tortuosity, extracellular volume fraction), determine the spatial concentration profile and the tissue mass transfer resistance about the probe (Amberg and Lindefors, 1989; Bungay, et al., 1990; Morrison et al., 1991). For most compounds, the tissue mass transfer resistance plays a more important role than the resistance from the probe membrane or dialysate in determining the extraction fraction in vivo. Therefore, it may be possible to use probe recovery to examine drug mass transfer processes, such as transporter-mediated drug transport across the blood-brain barrier.
Brain microdialysis has been used to study the active transport processes that influence drug distribution across the blood-brain barrier and BCSFB (Wong et al., 1993; Wang et al., 1995; Desrayaud et al., 1997; de Lange et al., 1999; Rao et al., 1999). In these studies, the concentration of the drug in the brain extracellular fluid or the cerebrospinal fluid was monitored by brain microdialysis and an apparent drug brain/plasma distribution ratio was calculated. The role of transporters at the blood-brain barrier and BCSFB in limiting the transport of the solute into the CNS was examined by comparing the brain/plasma distribution ratio when the transport protein was present or genetically depleted/pharmacologically inhibited (Wong et al., 1993; de Lange et al., 1998; Rao et al., 1999). However, the use of quantitative microdialysis to study transport processes has not been restricted to the estimation of brain extracellular fluid concentrations. Dykstra et al. (1993) have shown, using quantitative autoradiography, that the concentration-distance profile of zidovudine in brain tissue surrounding the microdialysis probe was significantly altered by the administration of probenecid. This change in the concentration-distance profile was attributed to the inhibition of zidovudine capillary efflux transport by probenecid and was accompanied by a decreased microdialysis probe recovery. Furthermore, a new application of quantitative microdialysis was also developed to estimate the microvascular transport constants from in vivo microdialysis probe recovery (Beagles et al., 1998). Therefore, these studies indicate that the in vivo microdialysis probe recovery can be used to study active transport processes present in the brain capillary. The appropriate application of this technique for this purpose needs further examination.
The purpose of the present study was to investigate the effect of transport inhibition at the blood-brain barrier and BCSFB on the brain microdialysis probe recovery using the steady-state retrodialysis method. A theoretical model is proposed to describe the transport kinetics of two different transporters, such as P-glycoprotein and an organic anion transporter, by using a simple steady-state model of transport across a cell monolayer that can describe drug transport across the capillaries of the blood-brain barrier (see ). In this model, a relationship between the mechanism of transport and the effect of inhibitors on capillary efflux is presented. Coupled with the effects of capillary efflux exchange on the microdialysis recovery (Bungay et al., 1990), it is suggested that the effect of transport inhibitors on microdialysis recovery can give insights into the mechanism of transport at the blood-brain barrier.
To explore these relationships, two substrate/inhibitor pairs were used. The effect of probenecid on the probe recovery of fluorescein was determined as an example of organic anion transport and the effect of LY-335979, a potent and specific P-glycoprotein inhibitor (Dantzig et al., 1999), on the probe recovery of quinidine was determined as an example of P-glycoprotein-mediated transport. In previous studies, these inhibitors were shown to significantly enhance the brain distribution of the substrates (Wang et al., 1996; Huai-Yun et al., 1998; Sun et al., 1998). In addition, the influence of other competing elimination processes, such as brain metabolism, on the influence of transport inhibition on probe recovery was analyzed through simulations based on a mathematical model of microdialysis sampling (Bungay, et al., 1990).
Fluorescein and difluorofluorescein were purchased from Molecular Probes (Eugene, OR). Probenecid and quinidine were purchased from Sigma (St. Louis, MO) and Acros (Pittsburgh, PA), respectively. LY-335979 was a gift from Eli Lilly (Indianapolis, IN). Solvents were of HPLC grade, and all other chemicals were reagent grade or better.
Microdialysis Probe Placement.
Male Wistar rats weighing between 260 and 340 g were used in this study. The surgical procedures for implantation of the microdialysis probe guide cannula, probe placement, and the cannulation of the femoral artery and vein were similar to Yang et al. (1996) with slight modification. At all times, including the microdialysis sampling period, the rats had free access to food and water. Surgical preparation of these rats was done using aseptic techniques, and all surgical procedures were performed under anesthesia using an i.p. dose of 50 mg/kg sodium pentobarbital (Abbott Laboratories, Chicago, IL). An i.m. dose of 60,000 units of procaine penicillin G (Wyeth-Ayerst, Princeton, NJ) was given after surgery.
The stereotaxic coordinates for probe placement in the frontal cortex were 3.0 mm anterior and 1.5 mm lateral (left) to the bregma; the tip of the guide cannula was 1 mm ventral from the brain surface. For probe placement in the lateral ventricle, the coordinates were 0.8 mm posterior and 1.5 mm lateral (right) to the bregma, and 3.0 mm ventral from the brain surface. The rat was allowed to recover for 3 to 4 days after placement of the guide cannula. The femoral artery and vein were then cannulated for blood sampling and dose administration, respectively. CMA/12 microdialysis probes (CMA-Microdialysis, Acton, MA) of 3-mm and 1-mm membrane length were used for cortex extracellular fluid and ventricle cerebrospinal fluid sampling, respectively, and the probes were slowly implanted into the brain parenchyma and lateral ventricle through the guide cannula approximately 18 to 24 h before the initiation of sampling via the probe. These procedures adhered to the Principles of Animal Care outlined by National Institutes of Health publication 85-23 and were approved by the Institutional Animal Care and Use Committee of the University of Nebraska Medical Center.
Microdialysis Sampling Procedure.
For both in vitro and in vivo microdialysis sampling, CMA/12 probes were perfused with artificial cerebrospinal fluid [119.5 mM NaCl, 4.75 mM KCl, 1.27 mM CaCl2, 1.19 mM KH2PO4, 1.19 mM MgSO4, 1.6 mM Na2HPO4 (pH 7.2)] (Benveniste and Huttemeier, 1990) using a microprocessor-controlled syringe pump (Harvard 22, Harvard Apparatus, Natick, MA). The perfusate flow rate through both the ventricle probe (1 mm) and the cortical probe (3 mm) was 0.5 μl/min. Microdialysates were collected over 20-min intervals directly into the injection loops of a multiport valve (E-36; Valco, Houston, TX), controlled by a digital valve sequence programmer (DVSP2, Valco). Microdialysates were then directly injected into the HPLC for analysis (see below).
The determination of fluorescein and difluorofluorescein concentration in microdialysate was done using HPLC with fluorescence detection. The HPLC system consisted of a LC-10AD pump, RF-10A fluorometric detector, and CR501 integrator (Shimadzu, Kyoto, Japan). Separations were carried out on a BDS-Hypersil C-18 column (2.0 × 150 mm, 5 μm) (Keystone Scientific, Inc., Bellefonte, PA). The mobile phase was an acetonitrile:buffer mix (13.6:86.4, w/w) with a buffer composition of 10 mM ammonium phosphate and 10 mM sodium citrate (pH 6). The mobile phase flow rate was 0.25 ml/min, and the column eluate was monitored at excitation and emission wavelengths of 488 and 510 nm, respectively.
The HPLC system and column for the analysis of quinidine concentration in microdialysate was the same as described above. Analytes were eluted by a mixture of 20 mM ammonium monobasic phosphate and 20 mM tartaric acid, adjusted to pH 3.6 with sodium hydroxide, and acetonitrile (86.6:14.4, w/w), with the excitation wavelength of 260 nm, the emission wavelength of 430 nm, and a mobile phase flow rate of 0.25 ml/min.
Determination of Probe Recovery in Vitro.
A CMA/12 3-mm microdialysis probe was placed in a 2-ml vial containing well stirred drug-free artificial cerebrospinal fluid at 37°C. Artificial cerebrospinal fluid containing fluorescein, difluorofluorescein, or quinidine was placed into a gas tight syringe and perfused through the probe at a flow rate of 0.5 μl/min. The same perfusate and flow rate were used in vivo to allow for direct comparison of the recoveries (eqs. 3 and 4). Microdialysate samples from the microdialysis probe were collected every 20 min. The concentration of the compound of interest in the microdialysate was used to calculate in vitro microdialysis probe recovery (eq. 1).
Study Design and Drug Administration.
The effect of probenecid on the in vivo probe recovery of fluorescein and/or difluorofluorescein, a fluorinated analog of fluorescein (Fig.1), was studied using the following two study designs.
Fluorescein and difluorofluorescein (0.02 μg/ml) in artificial cerebrospinal fluid were perfused through a microdialysis probe in the cortex without (first phase) or with (second phase) systemic intravenous administration of probenecid to five rats. In a repeated control group (n = 3), animals received probenecid vehicle (8.4% NaHCO3 in saline) in the second phase.
Difluorofluorescein (0.02 μg/ml) in artificial cerebrospinal fluid was perfused through microdialysis probe in cortex and ventricle to six rats, with or without the systemic administration of probenecid. In this study, rats were equally divided into two groups. Group A received treatment with probenecid in phase one and without probenecid in phase two. Group B received the opposite treatment order. In the above two studies, probenecid was given 100 mg/kg intravenous bolus followed by an infusion of 30 mg/kg/h.
The effect of LY-335979 on the probe recovery of quinidine (n = 3).
Quinidine (0.03 μg/ml) in artificial cerebrospinal fluid was perfused through microdialysis probes without (first phase) or with (second phase) systemic administration of LY-335979. LY-335979 was given as a 10 mg/kg intravenous bolus followed by an infusion of 1.25 mg/kg/h.
Microdialysate samples from the frontal cortex or the lateral ventricle were continuously collected every 20 min for approximately 20 h. The concentration of the compound of interest in the microdialysate was used to calculate microdialysis probe recovery (eq. 1).
Probe Recovery in Vitro and in Vivo.
The recovery of the probes was determined using the method of retrodialysis (Wang et al., 1993). The solute of interest (i.e., fluorescein, difluorofluorescein, or quinidine) was added to the perfusion fluid, and their relative loss was used to determine the recovery through the following relationship. Equation 1where PA(out) and PA(in) were the HPLC chromatographic peak areas of the solute of interest in the dialysate leaving the probe and in the perfusate entering the probe, respectively.
The Estimation of Mass Transfer Resistance.
According toBungay et al. (1990), the probe recovery or the dialysate extraction fraction (Ed) at steady state is equal to Equation 2where Cout and Cindenote the outlet and inlet dialysate concentration, respectively. Ce∞ is the extracellular concentration far from the probe. Qd is the flow rate of the perfusate, and R stands for the mass transfer resistance for dialysate (Rd), microdialysis membrane (Rm) and surrounding tissue (Re). The mass transfer resistance expresses the proportionality between the driving force concentration difference and the resultant mass flow rate in this environment of three resistances in series. The mass transfer resistances associated with the probe can be determined from in vitro probe recovery. When there is no tissue resistance, Equation 3The total mass transfer resistance can be estimated from in vivo probe recovery, Equation 4Thus, Re can be obtained by subtracting the in vitro mass transfer resistance from in vivo values.
Penetration Distance and Microvascular Efflux Rate Constants.
The resistance in the tissue is defined as Equation 5where K0 and K1are the modified Bessel functions of the second kind with dimensionless argument ro/Γ, De is the diffusion coefficient in brain extracellular fluid, L is the length of dialysis membrane, ro is the outer radius of dialysis membrane, φe is the extracellular volume fraction near the probe, and Γ is the penetration distance. The microdialysis probe parameter values that were used for recovery model simulations (e.g., Fig. 2), and for the fluorescein and quinidine mass transport parameter calculations, are listed in Table 1. The penetration distance is related to the diffusion coefficient and several first order rate constants as follows Equation 6and is defined as the distance from probe surface to the point where the concentration is roughly half its far-field value (i.e., Ce∞). k is the first order rate constant representing efflux to the microvasculature (k ), irreversible extracellular metabolism (k ), and the composite of irreversible intracellular metabolism and extracellular-intracellular exchange (k ) (Bungay et al., 1990).
Simulation for the Effect of First Order Elimination on Probe Recovery.
Simulations according to the above microdialysis probe recovery mathematical model (eqs. 2-6) were performed using the Microsoft Excel spreadsheet software. There are existing subroutines in the Excel program to calculate the modified Bessel functions of an argument. A parameter sensitivity analysis was performed using estimated and known parameters of the microdialysis recovery experiments to determine the relative importance of various model parameters on the resultant microdialysis recovery.
Sensitivity Analysis for the Effect of First Order Elimination on Probe Recovery.
Except for certain limiting cases, such as solutes subject to very rapid tissue elimination, the tissue resistance (Re) is the largest contributor to the overall mass transfer resistance. As a result, Re is important in determining in vivo probe recovery (Ed). Key factors affecting the level of Re and Ed are the elimination rate constants (k , k and k parameters in eq. 6), since these vary over several orders of magnitude among solutes. Simulations were performed to better understand the influence of the efflux rate constant, k , on the microdialysis behavior of drugs transported across the blood-brain barrier. The simulation considered hypothetical drugs with the same interstitial diffusion coefficient but differing efflux and non-efflux elimination rates. The diffusion coefficient was fixed at the value estimated for fluorescein (332.3 daltons) from a measured value for sucrose (342.3 daltons) in brain interstitium (Patlak and Fenstermacher, 1975). The extracellular volume fraction, φe, was set at the value of 0.35 previously estimated for brain parenchyma in the vicinity of acutely implanted probes (Dykstra et al., 1992). Probe geometric parameters were chosen to represent a CMA/12 probe with a 3-mm membrane length.
Inhibiting any of the elimination processes (efflux transport or intra- and extracellular metabolism), will reduce the corresponding rate constant (k , k , or k ). This would result in an increased tissue resistance (eqs. 5 and 6) and therefore a decreased in vivo probe recovery (eq. 2). In examining the influence of capillary efflux on the in vivo probe recovery, the effect of efflux transport inhibition (i.e., reduced k ) alone can be estimated by holding k and k constant, assuming no effects on metabolism by efflux inhibition or deletion (as in a transport gene knockout model). From Fig. 2 we can see that Edis most sensitive to variation in k when there are no other competing mechanisms for removal from the extracellular fluid (kr = k + k = 0) and when the transport process itself is slow (i.e., k is small). In this case, Ed drops significantly with a slight decrease in k . On the other hand, there are limiting situations in which Ed is insensitive to inhibition of capillary efflux. In these situations, Re is too low to contribute significantly to the total diffusional resistance. One is the situation in which competing elimination processes (metabolism) are rapid and dominant (i.e., kr is large), so that Edloses sensitivity to any changes in k . The other is the limit of rapid efflux (i.e., k is large) for which small changes in k do not appreciably affect Ed.
The Influence of Probenecid on in Vivo Probe Recovery of Fluorescein and Its Microdialysis Calibrator Difluorofluorescein in the Longitudinal Study.
Previous studies have shown that the in vitro and in vivo probe recovery of difluorofluorescein was similar to that of fluorescein (Sun et al., 1998). Therefore, difluorofluorescein has been used as the retrodialysis calibrator in the study of the transport of fluorescein across the blood-brain barrier and BCSFB. The influence of probenecid on probe recovery of both fluorescein and difluorofluorescein was then examined. Fluorescein and difluorofluorescein were perfused through the probe implanted in rat frontal cortex for 8 h before the start of the systemic administration of probenecid. The in vivo microdialysis probe recovery of fluorescein and difluorofluorescein significantly decreased following the administration of probenecid. A representative recovery versus time profile is shown in Fig. 3A. To see if there is a treatment order effect, a repeated control was performed and a representative recovery versus time profile is shown in Fig. 3B. After 10 to 12 h of probenecid infusion, the average probe recovery for both fluorescein and difluorofluorescein was significantly less than control, p < 0.01 (Fig.4). Importantly, this reduction in difluorofluorescein recovery was not significantly different than the reduction in fluorescein recovery, indicating that this would be a suitable retrodialysis calibrator under the condition of efflux transport inhibition. There was no significant difference in the in vivo recoveries for both fluorescein and difluorofluorescein between the two phases in the repeated control study.
The Influence of Probenecid on Probe Recovery of Difluorofluorescein in the Balanced Crossover Study.
A balanced crossover study was performed with a 24-h washout between the control and probenecid-treated phases. In this study, the retrodialysis probe recovery of difluorofluorescein in both the frontal cortex and lateral ventricle was monitored simultaneously. Figure5 shows that in the probenecid treatment phase, the recovery of difluorofluorescein was significantly reduced (p < 0.01) in frontal cortex. However, no difference was observed in the ventricle probe recovery between the control and probenecid-treated phases.
Microdialysis Model Parameters for Fluorescein in the Cortex.
The mass transport parameters of fluorescein in rat brain cortex are shown in Table 2. These parameters were calculated using the experimentally determined in vitro and in vivo extraction fractions (probe recovery). The in vitro extraction fraction of fluorescein at 37°C (well stirred) and under the same microdialysis conditions as used in vivo, was 0.47. Using eq. 3, the calculated in vitro mass transfer resistance from both the microdialysate and probe membrane (Rd + Rm) was 3.2 min/μl. In the control group, the average in vivo recovery of fluorescein was 0.21 ± 0.017, resulting in a total mass transfer resistance (∑R) of 8.5 min/μl. Using eq. 4, after subtraction of Rd + Rm, the calculated tissue mass transfer resistance (Re) was 5.3 min/μl. In the probenecid-treated phase, the average in vivo recovery of fluorescein was 0.17 ± 0.02, which resulted in an increase in the tissue mass transfer resistance to 7.9 min/μl. The fluorescein penetration distance in the cortex (Γ), was calculated using eq. 5. In this calculation, the values for the argument of the Bessel functions (i.e., ro/Γ) are determined by trial and error iteration to achieve equality between the experimentally determined Re and the right-hand side of eq. 5. As would be anticipated when there is an increased mass transfer resistance, the probenecid treatment resulted in an increase in the penetration distance, Γ. Table 2 also shows that probenecid caused a 3-fold decrease in the sum of the elimination rate constants (∑k) of fluorescein from the cortex, as calculated from eq. 6.
The fluorescein effective diffusion coefficient (De) in the brain extracellular fluid was estimated from the previously determined brain extracellular fluid diffusion coefficient of sucrose (Patlak and Fenstermacher, 1975), assuming that diffusivity is inversely related to the square root of solute molecular weight (Table 1). This scaling of the diffusivity has been previously used to estimate the microdialysis-derived mass transport parameters of quinolinic acid (Beagles et al., 1998).
The Influence of LY-335979 on Probe Recovery of Quinidine.
Perfusion of a 30 ng/ml solution of quinidine through the microdialysis probe in rat cortex resulted in an in vivo retrodialysis probe recovery (extraction fraction) of 0.26 ± 0.067 (control phase). After 8 h, the quinidine probe perfusion continued, and the rat received a LY-335979 bolus followed by an infusion for 12 h (treated phase). Figure 6 shows the in vivo quinidine recovery versus time profile in a representative rat. The systemic administration of LY-335979 did not significantly change the in vivo recovery of quinidine. The in vivo probe recovery of quinidine in the treated phase is 0.24 ± 0.054. The in vitro recovery of quinidine was determined to be 0.68, using the same microdialysis conditions (i.e., probe geometry, perfusate flow rate, temperature) as in the in vivo case. Therefore, the mass transfer parameters of quinidine in rat cortex were calculated using eqs. 2through 6 in the same manner as described above for fluorescein, and these parameters are listed in Table 2. The total elimination rate constant (∑k) of quinidine is 0.46 and 0.33 ml/(min · ml extracellular fluid) in the control and treated phases, respectively, indicating that LY-335979 has a minimal effect on the elimination of quinidine from the brain to plasma.
The results outlined in this report further illustrate the relationship between the in vivo microdialysis probe recovery (extraction fraction) and efflux transport processes in the CNS. The influence of the inhibition of an efflux transporter on the probe recovery will vary depending on the different transport substrates and inhibitors examined, the localization and mechanism of an efflux transport system, the magnitude of the elimination process in the brain, as well as the microenvironment surrounding the probe (e.g., the cerebrospinal fluid versus the cortical extracellular fluid).
Efflux transport systems at the blood-brain barrier affect the brain distribution of a variety of compounds into the brain. The membrane location and the mechanisms of these transporters are likely to have important pharmacological and toxicological consequences. Efforts to understand these consequences can be aided by taking advantage of the differences in transport kinetics conferred by such restrictions in transporter location and mechanism. In this study, we proposed a theoretical transport kinetic model to study the characteristics of the transport behavior of P-glycoprotein and an organic anion transport system. P-glycoprotein is an efflux transport protein located in the apical membrane of the blood-brain barrier endothelial cells (Cordon-Cardo et al., 1989) and is well known as an effective efflux transport system in the brain (Fromm, 2000). Our theoretical model (see ) of solute flux across the blood-brain barrier indicates that, for certain substrates, the inhibition of P-glycoprotein increases brain-to-blood distribution ratio of the substrate but has no effect on the efflux rate constant (eqs. T-12 and T-13) and, therefore, no effect on the microdialysis probe recovery. On the other hand, for a transporter localized to the basolateral membrane of the blood-brain barrier, the inhibition of the transporter will result in an increase in the brain-to-blood ratio as well as a decrease in efflux rate constant (eqs. T-18 and T-19), resulting in a decrease in microdialysis probe recovery.
This theoretical model (see ) was supported by the study of two substrate/inhibitor combinations using in vivo quantitative microdialysis, quinidine/LY-335979 for P-glycoprotein, and fluorescein/probenecid for an organic anion transport protein. Quinidine is a substrate of P-glycoprotein, and previous studies have shown that LY-335979, a potent and specific P-glycoprotein inhibitor (Dantzig et al., 1996, 1999), increased quinidine brain distribution as much as 7-fold when quinidine and the inhibitor were given intravenously (Wang et al., 1996; Starling et al., 1997). According to the microdialysis model, a reduction in the in vivo quinidine recovery is anticipated if the increase in brain-to-blood ratio of quinidine is also accompanied by a reduced efflux constant from brain-to-blood. However, the microdialysis results show no change in the quinidine in vivo recovery (Fig. 6). The quantitative microdialysis calculation (eqs. 2-6) suggests a minor change of the efflux transport constant of quinidine from brain-to-blood (Table 2). This result is consistent with the work ofKusuhara et al. (1997). In their study, when quinidine was given intravenously, systemically administered PSC-833, another potent P-glycoprotein inhibitor, increased the net brain uptake of quinidine by approximately 15-fold. In contrast, when quinidine was given intracerebrally, systemically administered PSC-833 had no significant effect on the efflux rate of quinidine as measured by the brain efflux index (Kusuhara et al., 1997). This apparent contradiction could be explained by the theoretical model of P-glycoprotein-mediated flux across the blood-brain barrier proposed in this study, where the inhibition of P-glycoprotein will enhance the brain distribution of systemically administered substrates but will not influence the efflux rate of substrates directly introduced into the brain extracellular fluid.
The in vivo microdialysis study of fluorescein showed that the in vivo recovery of fluorescein was reduced by probenecid. Fluorescein is an organic anion that is actively transported by organic anion transport systems (Engler et al., 1994; Huai-Yun et al., 1998). In a previous study in our laboratory (Sun et al., 1998), probenecid, well known as an organic anion transport inhibitor, inhibited the active efflux of fluorescein at the blood-brain barrier and BCSFB. Probenecid treatment resulted in an enhanced net transport of systemically administered fluorescein into brain tissue, i.e., the apparent tissue (cortical extracellular fluid and ventricle cerebrospinal fluid) to plasma equilibrium distribution coefficient of fluorescein was increased by approximately 2-fold (Sun et al., 1998). In the present study, quantitative microdialysis (eqs. 2-6) revealed that the reduced microdialysis recovery was associated with a decrease in the overall cortical elimination rate constant of fluorescein by approximately 3-fold (Table2). These results are in agreement with the predications from the present theoretical model of transporter-mediated flux. The model indicates that the inhibition of the transporter located at the basolateral membrane of the barrier should result in a similar magnitude of decrease in the efflux rate constant and increase in the apparent tissue-to-plasma equilibrium distribution coefficient (eqs. T-18 and T-19).
Using the microdialysis recovery approach to study the transport kinetics in CNS has a significant advantage, because in certain circumstances, the influence of a systemically administered inhibitor on the blood concentrations of a substrate can be excluded. In this study, a simulation according to the quantitative microdialysis model of Bungay et al. (1990) was performed to have a better understanding of the relationship between the change of the efflux transport constant and the change of recovery. For most compounds, the mass transfer resistance from tissue probably accounts for the major part of the resistance in vivo, and thus it has been assumed that factors that influence the tissue mass transfer resistance, such as elimination rate constant, will be important in determining the in vivo probe recovery. However, it is unlikely that this is a valid assumption for all compounds. Simulations from the microdialysis mathematical model (eqs. 2-6) show that the recovery is not sensitive to small changes in the elimination rate of the compound from the brain, if the overall elimination rate is fast (Fig. 2). This indicates that either the overall elimination rate needs to be sufficiently slow for the recovery to be influenced by an altered elimination process or the alteration must be large. Therefore, in the case where there is a fast elimination process from the brain, which can be expected from some rapidly cleared compounds, such as neurotransmitters, significant changes in recovery may be anticipated only when the clearance mechanism is greatly inhibited (Smith and Justice, 1994;Cosford et al., 1996; Vinson and Justice, 1997). In addition, during the quantitative microdialysis calculation, ignoring the probe mass transfer resistances could result in erroneous conclusions. In the case of fluorescein, the contribution of mass transfer resistances from dialysate and probe membrane accounts for more than 35% of the total resistance in the control condition in vivo (Table 2). A change in the in vivo probe recovery, caused by an increased tissue mass transfer resistance (for instance, when an efflux process is inhibited), will be attenuated when the mass transfer resistances from dialysate and probe membrane account for a large proportion of the total resistance. It should be noticed that the elimination rate of both fluorescein and quinidine from brain extracellular fluid to plasma are in the range where the recovery should be sensitive to an inhibition of the efflux process, suggesting that for both compounds, a change in the probe recovery would be observed if there would be a decrease in the efflux transport constant.
This study shows that the recovery of the retrodialysis calibrator, difluorofluorescein, was also decreased by probenecid (Figs. 3 and 4) in parallel with the decrease in fluorescein recovery. A corresponding change in the probe recovery of a retrodialysis calibrator, when compared to the solute of interest, is important for accurate estimation of brain extracellular fluid concentrations in the microdialysis sampling when efflux transport is altered through pharmacological inhibition. These results also indicate that, if carefully chosen, a retrodialysis calibrator can reflect changes in microdialysis probe recovery induced by a variety of interventions, such as the coadministration of an efflux transport inhibitor, and therefore the calibrator would be suitable for the study of transport systems at the blood-brain barrier for the estimation of the concentration of solute of interest in brain extracellular fluid.
Probenecid caused no change in the probe recovery of difluorofluorescein in the ventricle cerebrospinal fluid (Fig. 5), even though there is a parallel change of the probe recovery of difluorofluorescein with the probe recovery of fluorescein in the cortical extracellular fluid. Our previous study (Sun et al., 1998) indicated that probenecid significantly enhanced fluorescein distribution to the cerebrospinal fluid in the lateral ventricle. The probenecid-induced increase in the cerebrospinal fluid concentration of fluorescein is probably due to the inhibition of efflux processes at the choroid plexus. However, the microenvironment surrounding the probe in the ventricle is very different from that in the cortex. If efflux transport at a capillary interface in close proximity to the probe is an important determinant of microdialysis recovery, then the recovery of probes placed in the fluid environment of the ventricular cerebrospinal fluid would not be affected by the inhibition of capillary efflux.
The current study was performed under steady-state conditions with respect to the concentration in the extracellular fluid. This could be important because during transient conditions, i.e., the solute extracellular fluid concentration is changing with time, other factors such as the intracellular-to-extracellular partition coefficient and overall systemic elimination rate also influence probe recovery (Morrison et al., 1991; Bungay et al., 2001). In the study by Morrison et al. (1999), even though probenecid increased the accumulation of quinolinic acid in the brain by inhibiting a microvascular acid transporter, no significant changes were detected in the quinolinic acid microdialysis recovery in the striatum. It is important to recognize that, in this study, quinolinic acid brain concentrations changed significantly within the 3-h experimental period, therefore, factors other than capillary efflux may have had a dominant influence on determining the microdialysis recovery of quinolinic acid. With respect to retrodialysis under transient conditions, in general, the relative loss of an analyte from the tissue and the relative recovery of a marker solute from the perfusate follow different time courses, even though the steady-state Ed may be the same for both (Bungay et al., 2001). Thus, transient calibration by retrodialysis is problematic.
In summary, the use of the microdialysis extraction fraction (recovery) to make conclusions about the mass transport processes in the CNS requires a considerable understanding of the various processes that can influence the probe recovery in the brain tissue and/or the cerebrospinal fluid in the ventricles. In certain circumstances, the recovery of the solute of interest may change with alterations in capillary efflux inhibition, depending on the elimination rate of the compound from the brain, the mechanism and location of the membrane transporter, and the environment surrounding the probe. If carefully chosen, a retrodialysis calibrator could monitor changes in steady-state recovery brought about by various experimental treatments and thus represent the true recovery under different situations. With an understanding of the various influences on recovery, the use of changes in microdialysis probe extraction fraction brought about by experimental intervention can give important information about drug mass transport in the CNS. This may be especially useful considering these experiments can be performed without any systemic effect on the disposition of the compound of interest.
Mathematical Models for Efflux Transporters across the Blood-Brain Barrier.
The multidrug resistant transporter, P-glycoprotein, is reported to be present in the apical membrane of brain endothelial cells (Cordon-Cardo et al., 1989). Stein (1997) proposed a kinetic model for substrate transport by P-glycoprotein in tumor cells expressing this transporter. Figure7A is a schematic for the incorporation of Stein's model of combined P-glycoprotein and passive transport in the apical/luminal membrane of a cellular monolayer, together with passive transport in the basal/abluminal membrane. For the present, we neglect diffusional resistance in the adjacent external medium and cytosolic contributions. We also assume all transport processes exhibit linear concentration dependence, although nonlinear characteristics could be readily incorporated. For simplicity, the passive processes are assumed to be symmetric and the same in both membranes and, hence, can be represented by a single permeability, Ps. In accordance with the hypothesis that P-glycoprotein functions to expel solutes from within the leaflets of the plasmalemma bilayer membrane, Stein suggested a model with separate contributions from the inner and outer half-bilayer. These contributions are represented in the model schematic by the permeabilities, Paipand Paop, for pumping from the inner and outer apical half-bilayers to plasma, respectively. He assumed equilibrium partitioning between each half-bilayer and the adjacent medium. If the substrate concentrations in the plasma/apical, cytosolic and tissue extracellular fluid/basal mediums are denoted by Cp, Ci, and Ce, respectively, then the steady-state mass flux, q, across the two plasmalemma membranes is given by Equation ET-1For physically meaningful behavior, the outer half-bilayer permeability must be less than passive permeability (Paop < Ps). This is because solute must enter the outer half-bilayer through passive permeation prior to leaving the outer half-bilayer through the P-glycoprotein pumping mechanism. Eliminating Ci from the above two expressions yields Equation ET-2To relate the membrane permeabilities in these expressions to the rate constants of microdialysis theory, we begin by rewriting eq. T-2 as the local flux of solute from the extracellular fluid to plasma Equation ET-3in which the unidirectional endothelial cell permeabilities have been defined as Equation ET-4and, Equation ET-5From the above equations we obtain that the extracellular fluid-to-plasma free concentration ratio at equilibrium is Equation ET-6The extracellular fluid-to-plasma uptake (k ) and efflux (k ) rate constants are given in the Appendix of Bungay et al. (1990) as Equation ET-7and Equation ET-8where φe is the extracellular fluid volume per unit mass of tissue, Qb and S are the blood flow per unit mass of tissue and the microvascular surface area, respectively, and Φb is the ratio of blood-to-plasma solute concentrations for the fraction of solute that is exchangeable in a single pass through the capillaries.
In a previous study, the specific P-glycoprotein inhibitor LY-335979 increased the flux of quinidine across the bovine brain microvessel endothelial cell monolayer, an in vitro model of the blood-brain barrier, from the apical to basolateral side (blood to brain), but had no effect on the quinidine flux from basolateral to apical side (brain to blood) (unpublished data). This lack of sensitivity of the unidirectional basal-to-apical quinidine flux to inhibition by LY-335979 suggests from eqs. T-3 and T-4 that, for quinidine, Paip ≪ Ps, which simplifies eqs. T-4, T-5, T-7, and T-8 to Equation ET-9and Equation ET-10The predictions from these expressions are easier to interpret from their limiting forms for (Ps − Paop)S ≪ 2Qb · Φb. By truncating their Taylor expansion, eqs. T-9 and T-10 reduce to Equation ET-11and Equation ET-12and substituting these simplifications into eq. T-6 yields Equation ET-13Ppe < Pep corresponds to a lower apical-to-basal than basal-to-apical flux for a given concentration difference imposed across the blood-brain barrier monolayer. These expressions indicate that inhibition of P-glycoprotein corresponding to decreasing Paop will increase uptake and increase the extracellular fluid-to-plasma concentration ratio at equilibrium (eq. T-13). Inhibition of P-glycoprotein will not affect the efflux rate constant (eq. T-12) and, hence, should not affect microdialysis probe recovery (see Experimental Procedures, eqs. 4-6). This model is consistent with the observations regarding the quinidine transport across the blood-brain barrier by microdialysis in vivo.
For a transporter restricted to the basal/abluminal membrane (Fig. 7B), the net steady-state flux in the basal-to-apical direction is described by, Equation ET-14The transporter contribution to the flux, Pbi · Ce, is undoubtedly an oversimplification since it may also be a function of Ci. Combining the flux expressions to eliminate Ci yield, Equation ET-15The resulting uptake and efflux permeabilities are then Equation ET-16and the corresponding rate constants, following simplification to their limiting forms as before are Equation ET-17and Equation ET-18Therefore, the equilibrium extracellular fluid-to-plasma concentration ratio is Equation ET-19In this case, the inhibition of the transporter corresponds to a reduction in the permeability, Pbi, which produces a decrease in the efflux rate constant and an increase in the equilibrium concentration ratio. Similar qualitative results were obtained for an analysis of efflux transport at the apical membrane where the transport system pumps from the intracellular phase only. In each of these cases, a reduction in the efflux rate constant would predict a decrease in the microdialysis probe recovery (seeExperimental Procedures, eqs. 4-6). These predictions are consistent with the observations regarding fluorescein transport across the blood-brain barrier and its inhibition by probenecid.
Send reprint requests to: Dr. William F. Elmquist, Dept. of Pharmaceutical Sciences, 986025 Nebraska Medical Center, Omaha, NE 68198-6025. E-mail:
This work was partially supported by Grant NIH-NCI CA-75466 (to W.F.E.) from the National Institutes of Health. H.S. was supported by Graduate Fellowships from the University of Nebraska Medical Center and a Presidential Fellowship awarded by the University of Nebraska.
- blood-cerebrospinal fluid barrier
- central nervous system
- high performance liquid chromatography
- U.S. Government