Relationship between permeability status of the blood–brain barrier and in vitro permeability coefficient of a drug

European Journal of Pharmaceutical Sciences 12 (2000) 95–102 www.elsevier.nl / locate / ejps

Relationship between permeability status of the blood–brain barrier and in vitro permeability coefficient of a drug Pieter Jaap Gaillard, Albertus Gerrit de Boer* Department of Pharmacology, Leiden /Amsterdam Center for Drug Research ( LACDR), Leiden University, P.O. Box 9503, 2300 RA Leiden, The Netherlands Received 6 March 2000; received in revised form 29 May 2000; accepted 3 July 2000 In honour of Professor D.D. Breimer, Ph.D, on his 25th anniversary as Professor in Pharmacology and Pharmacotherapeutics.

Abstract Objective: The aim was to test the hypothesis that the assessment of basal and drug-induced changes in permeability of the blood–brain barrier (BBB) during in vitro drug transport assays is essential for an accurate estimation of the permeability coefficient of a drug. Methods: An in vitro BBB model was used, comprising of brain capillary endothelial cells (BCEC) and astrocytes co-cultured on semi-permeable filter inserts. Experiments were performed under control and challenged experimental circumstances, induced to simulate drug effects. The apparent BBB permeability coefficient for two markers for paracellular drug transport, sodium fluorescein (Papp,FLU , Mw 376 Da) and FITC-labeled dextran (Papp,FD4 , Mw 4 kDa), was determined. Transendothelial electrical resistance (TEER) was used to quantify basal and (simulated) drug-induced changes in permeability of the in vitro BBB. The relationship between Papp and TEER was determined. Drug effects were simulated by exposure to physiologically active endogenous and exogenous substances (i.e., histamine, deferroxamine mesylate, adrenaline, noradrenaline, bradykinin, vinblastine, sodium nitroprusside and lipopolysaccharide). Results: Papp,FLU and Papp,FD4 in control experiments varied from 1.6 up to 17.6 (10 26 cm / s) and 0.3 up to 7.3 (10 26 cm / s), respectively; while for individual filters Papp,FLU was 4 times higher than Papp,FD4 (R 2 50.97). As long as TEER remained above 131?V cm 2 for FLU or 122?V cm 2 for FD4 during the transport assay, Papp remained independent from the basal permeability of the in vitro BBB. Below these TEER values, Papp increased exponentially. This nonlinear relationship between basal BBB permeability and Papp was described by a one-phase exponential decay model. From this model the BBB permeability status independent permeability coefficients for FLU and FD4 (PFLU and PFD4 ) were estimated to be 2.260.1 and 0.4860.03 (10 26 cm / s), respectively. In the experimentally challenged experiments, a reliable indication for PFLU and PFD4 could be estimated only after the (simulated) drug-induced change in BBB permeability was taken into account. Conclusions: The assessment of basal BBB permeability status during drug transport assays was essential for an accurate estimation of the in vitro permeability coefficient of a drug. To accurately extrapolate the in vitro permeability coefficient of a drug to the in vivo situation, it is essential that drug-induced changes in the in vitro BBB permeability during the drug transport assay are determined.  2000 Elsevier Science B.V. All rights reserved. Keywords: Blood–brain barrier; Drug transport; Tight junctions; In vitro; Permeability coefficient; Fluorescein

1. Introduction The physico-chemical properties of a drug (i.e., lipophilicity, molecular weight, shape and charge) determine if a drug permeates across the blood–brain barrier (BBB). In addition, the drug may be a substrate for active or passive transmembrane transporters, drug efflux pumps or *Corresponding author. Tel.: 131-71-5276-215; fax: 131-71-5276292. E-mail address: [email protected] (A.G. de Boer).

metabolizing enzymes (Tamai and Tsuji, 1996; Begley, 1996). Furthermore, it is important to realize that the BBB is under physiological control and that its transport characteristics can be changed by diseases, and also by drugs (Takeda et al., 1992; Borges et al., 1994; Fischer et al., 1995, 1996; Albert et al., 1997; de Boer and Breimer, 1998; Hurst and Clark, 1998). In particular, molecules involved in virtually all intracellular signaling pathways (including: tyrosine kinases, Ca 21 , protein kinase C, heterotrimeric G proteins, calmodulin, cAMP, lipid second messengers and phospholipase C) have been reported to

0928-0987 / 00 / $ – see front matter  2000 Elsevier Science B.V. All rights reserved. PII: S0928-0987( 00 )00152-4

96

P. J. Gaillard, A.G. de Boer / European Journal of Pharmaceutical Sciences 12 (2000) 95 – 102

affect drug permeability (Mitic and Anderson, 1998). These changes often correlate with changes in actin organization or through the control of assembly and disassembly of tight junctions (Mitic and Anderson, 1998). In addition, changes in transport may be induced by disease or drug actions originating from outside the BBB (de Vries et al., 1995). The basal and drug-induced changes in permeability status of the in vitro BBB model, the (coated) filter, the unstirred water layer, together with the physico-chemical properties of a drug, determine the apparent BBB permeability coefficient (Papp ) of a drug. This parameter provides information on the permeability characteristics of a drug that is independent of the experimental design, as it is corrected for the transport surface area, the time duration of the experiment and the applied concentration (Artursson, 1990). Therefore, Papp is an excellent parameter for comparing permeability values of drugs between experiments. Unfortunately, the basal permeability status of in vitro BBB models is often not quantified or reported during in vitro drug transport assays, while considerable variation may exist within preparations but also between laboratories (de Boer et al., 1999). Furthermore, an in vitro BBB with a low basal permeability is difficult to prepare and to maintain (de Boer and Breimer, 1996; Deangelis et al., 1996). In that case, it is impossible to discriminate between the basal permeability status of the in vitro BBB model and putative drug-induced changes in the in vitro BBB permeability. In addition, due to the diversity in the currently employed in vitro systems, comparison of results and therefore the predictive value for the in vivo situation, is problematic without a quantitative characterization of the permeability status of the models (de Boer et al., 1999). The analysis of transendothelial electrical resistance (TEER) is a simple method to quantify the functionality of the tight junctions of the in vitro BBB. TEER represents the permeability of small ions through the tight junctions between brain capillary endothelial cells (BCEC). The absolute value of TEER is believed to be mainly dependent on the amount and complexity of tight junctions between the cells (Madara, 1998). Likewise, this is also the limiting factor for paracellular transport of large and hydrophilic compounds (Wong and Gumbiner, 1997; Madara, 1998). Drug transport studies with an in vitro BBB model allows simultaneous determination of the permeability status of the BBB by TEER and Papp of drugs. Therefore, we investigated the relationship between Papp for two markers for paracellular drug transport (sodium fluorescein (FLU) and FITC-labeled dextran (FD4)) and TEER across the in vitro BBB in a meta-analysis performed on data obtained from 20 experiments. This analysis was undertaken to test the hypothesis that the in vitro BBB permeability status has to be taken into account during drug transport assays, in order to accurately assess the permeability coefficient of the investigated drugs.

2. Methods

2.1. Preparation of the in vitro BBB model The preparation of the in vitro BBB model was described before (Gaillard et al., 2000). Briefly, brain capillaries were isolated from cortices of brains of bovine origin. Calf brains were obtained at the slaughterhouse (Molendijk, Nieuwekerk a / d IJssel, The Netherlands). The capillary fraction was obtained after homogenization, trapped on nylon meshes and subsequently enzymatically digested. Astrocytes were isolated from cortices of brains of newborn Wistar rats (Harlan, Zeist, The Netherlands) and used to prepare astrocyte conditioned medium and co-cultures. BCEC were cultured from brain capillaries on collagen- and fibronectin-coated culture flasks in 50% astrocyte conditioned medium. BCEC were passaged on collagen-coated Transwell polycarbonate filters (surface area, 0.33 cm 2 ; pore-size, 0.4 mm; Corning Costar) and cultured to tight monolayers in 4 days. Astrocytes were seeded on the bottom of the filters 2 days before BCEC. A schematic drawing of the BCEC1astrocyte co-culture model is depicted in Fig. 1.

2.2. Assessment of Papp for FLU and FD4 FLU and FD4 were added at the start of the experiment to the apical compartment at a concentration of 1 and 100 mg / ml, respectively. After 5 h, the concentration of FLU and FD4 in the basolateral compartment was determined by high-performance liquid chromatography (HPLC) in a single analysis, as described before (Gaillard et al., 2000). Papp for both markers was calculated, according to Artursson (1990), by the following Eq. (1): dQ 1 Papp 5 ] ? ]]] dt A ? C0 ? 60

(cm / s)

(1)

where dQ / dt is the amount of FLU or FD4 transported per

Fig. 1. Schematic representation of the in vitro blood–brain barrier co-culture model. Transendothelial electrical resistance (TEER) was measured as an indication of the tightness of the tight junctions between BCEC. FLU and FD4 were applied as two markers for paracellular drug transport in the apical chamber and the transported amount was determined after 5 h in the basolateral chamber. Abbreviations: BCEC, brain capillary endothelial cells; AC, astrocytes.

P. J. Gaillard, A.G. de Boer / European Journal of Pharmaceutical Sciences 12 (2000) 95 – 102

minute (ng / min), A is the surface area of the filter (cm 2 ), C0 is the initial concentration of FLU or FD4 (ng / ml) and 60 is the conversion from minutes to seconds.

2.3. Assessment of TEER Electrical resistance across BCEC1astrocyte co-cultures was measured using an electrical resistance system (ERS) with a current-passing and voltage-measuring electrode (Millicell-ERS, Millipore, Bedford, MA, USA). TEER (V? cm 2 ) was calculated from the displayed electrical resistance on the readout screen by subtraction of the electrical resistance of a collagen-coated filter without cells and a correction for filter surface area. The initial value of TEER was measured and at every subsequent hour.

2.4. Experimental setup Data (Papp,FLU , Papp,FD4 and TEER) were obtained from 20 experiments which were performed according to an identical protocol. The control group consisted of filters taken from the untreated control groups of these experiments and were considered as the reference for the other groups. The tightened group consisted of filters in which a tightening in the in vitro BBB was experimentally induced by exposure to histamine (1 mM), deferroxamine mesylate (10 mM), adrenaline (10 mM) or noradrenaline (100 mM). The opened group consisted of filters in which an opening of the in vitro BBB was experimentally induced by exposure to bradykinin (10 mM), vinblastine (100 nM), sodium nitroprusside (100 mM) or lipopolysaccharide (50 ng / ml). Papp,FLU , Papp,FD4 and the relationship with the permeability status of the in vitro BBB was determined assuming different levels of knowledge and analyzed accordingly. First, only Papp,FLU and Papp,FD4 were evaluated (i.e., no prior knowledge of the permeability status of the in vitro BBB and no prior knowledge of the simulated effects on the permeability of the in vitro BBB). Second, Papp,FLU and Papp,FD4 were grouped according to expected pharmacodynamic effects (e.g., known effects from literature, pilot or previous experiments), without actually determining the effects during the experiment. Third, the initial value of the permeability status of the in vitro BBB was actually determined by means of TEER (TEER (at 0h) ). Fourth, the effect of the experimental circumstances (i.e., simulated pharmacodynamic effects) on the permeability status of the in vitro BBB was determined by TEER, measured every hour during the transport experiment. This effect was quantified by means of the geometric mean of TEER during the transport assay (TEER ( 1 – 5h) ). TEER (at 0h) and TEER ( 1 – 5h) were related to Papp,FLU and Papp,FD4 for each individual filter, and categorized based on the known experimental circumstances (i.e., control, tightened and opened group). The relationship between TEER and Papp

97

was characterized by an one phase exponential decay model, according to the following Eq. (2): Papp 5 P‘leaky‘ ? e 2K ?TEER 1 P‘barrier’

(2)

where Papp is Papp,FLU or Papp,FD4 (10 26 cm / s) for each individual filter, TEER is the corresponding TEER (at 0h) or TEER ( 1 – 5h) (V cm 2 ), K is the estimated tightness constant ((V?cm 2 )21 ) for the changes in Papp , from which the 50% permeability reduction value (i.e., TEER1 / 2 (V?cm 2 )) was determined by ln 2 /K. P‘barrier’ (10 26 cm / s) is the estimated permeability coefficient which is independent from the value of TEER, and P‘leaky’ (10 26 cm / s) is the Papp P‘leaky’ when TEER50 Papp starts out equal to P‘leaky’ 1 P‘barrier’ and decreases to P‘barrier’ with a tightness constant K. The curves were fitted with weighted Papp values (1 /Y 2 ) for the most accurate estimation of P‘barrier’ . The relationship was judged by the goodness of fit parameter R 2 (value between 0 and 1) and the standard deviation of the residuals (Sy? x, expressed in 10 26 cm / s).

2.5. Data analysis The data were analyzed using Microsoft Excel version 5.0c and GraphPad Prism version 2.01. Methods used for statistical analysis are indicated in each individual experiment, but were non-parametric, unpaired, two-tailed tests with the level of significance set at 5%.

2.6. Materials Disposable sterile plasticware was purchased from Corning Costar (Cambridge, MA, USA), Micronic (Lelystad, The Netherlands) and Greiner (Alphen a / d Rijn, The Netherlands). Instant sterile endotoxin free cell culture media, supplements and PBS were obtained at BioWhittaker Europe (Verviers, Belgium). Collagen, histamine, deferroxamine mesylate, adrenaline, noradrenaline, bradykinin, vinblastine, lipopolysaccharide, FLU and FD4 were obtained at Sigma (St. Louis, MO, USA). FCS (from P.A. Biologicals (Sydney, Australia)) was obtained at Greiner (Alphen a / d Rijn, The Netherlands) and fibronectin at Boehringer-Mannheim (Almere, The Netherlands). Inorganic salts and all other reagents were of analytical grade.

3. Results

3.1. Papp only The group statistics of Papp,FLU and Papp,FD4 are summarized in Table 1 (pooled data). The coefficient of variation (CV) for Papp,FLU and Papp,FD4 was 60 and 77%, respectively. In contrast, a high positive correlation was found between the individual Papp values for FLU and FD4 (Pearson r50.986; R 2 50.97). On average, Papp,FLU was

P. J. Gaillard, A.G. de Boer / European Journal of Pharmaceutical Sciences 12 (2000) 95 – 102

98

Table 1 Groups statistics (mean6S.D. and (min–max value)) of Papp,FLU , Papp,FD4 , TEER (at 0h) and TEER ( 1 – 5h) of the different groups

n Papp,FLU (10 26 cm / s) Papp,FD4 (10 26 cm / s) TEER (at 0h) (V?cm 2 ) TEER ( 1 – 5h) (V?cm 2 )

Pooled data

Control

Tightened

Opened

245 4.562.7 (1.1–18.9) 1.361.0 (0.2–7.3)

88 4.562.9 (1.6–17.6) 1.361.2 (0.3–7.3) 98.2650.2 (21.7–226.7) 90.7648.9 (19.0–247.8)

58 3.461.8* (1.1–8.9) 0.960.6* (0.2–2.7) 102.4660.3 (27.7–238.3) 118.0661.7 (30.1–237.2)

99 5.062.7* (2.0–18.9) 1.561.0* (0.4–7.0) 165.7648.7* (48.3–251.7) 90.3630.9 (14.0–173.1)

*Significant difference from the control group (P,0.05, Mann–Whitney (rank sum) test).

4.060.8 (mean6S.D.) times higher than Papp,FD4 , for each individual filter.

3.2. Papp , categorized on the expected effect on BBB permeability The group statistics of Papp,FLU and Papp,FD4 are summarized in Table 1 (control, tightened and opened). The CVs for Papp,FLU and Papp,FD4 in the control group were 64 and 92%, respectively. In addition, the CVs for Papp,FLU and Papp,FD4 in the tightened group were 53 and 67%, respectively, where for the opened group they were found to be 54 and 67%. The averaged Papp,FLU and Papp,FD4 of the tightened and opened group were, however, significantly different from the control group (Mann–Whitney (rank sum) test).

3.3. Papp , categorized on the expected effect on BBB permeability and related to TEER( at 0 h) The group statistics of TEER (at 0h) are summarized in Table 1 for each group. The averaged TEER (at 0h) of the opened group was significantly higher compared to the control and tightened groups (Mann–Whitney (rank sum) test). In the control groups of the 20 different experiments, the between day CV was 47.5% and the within day CV was

Table 2 Fitted curves parameters (estimates195% confidence intervals) and goodness of fit parameters for the Papp –TEER (at 0h) profiles of the different groups for FLU and FD4 (displayed in Figs. 2 and 3)

P‘leaky’ (10 26 cm / s)

FLU FD4

P‘barrier’ (10 26 cm / s)

FLU FD4

TEER1 / 2 (V?cm 2 )

FLU FD4

R2 Sy? x (10 26 cm / s)

FLU FD4 FLU FD4

Control

Tightened

Opened

23.2 (15.0–31.4) 11.0 (6.1–16.0) 2.2 (1.9–2.4) 0.48 (0.42–0.54) 19.3 (16.0–24.2) 15.8 (13.1–20.0) 0.84 0.85 1.17 0.46

9.0 (6.1–11.8) 4.4 (2.6–6.2) 1.4 (1.2–1.7) 0.27 (0.23–0.32) 32.3 (24.6–47.0) 21.8 (17.4–29.2) 0.82 0.85 0.76 0.23

181.1 (264.6–426.8) 142.8 (2131.9–417.6) 3.5 (3.2–3.8) 0.81 (0.71–0.90) 13.6 (9.8–22.2) 10.6 (7.3–19.8) 0.41 0.26 2.08 0.90

28.5% for TEER (at 0h) . In addition, within one experiment the absolute value of TEER was not always homogeneously distributed over the filters in the 24-well culture plates (not shown). However, this variation allowed the analysis of Papp,FLU and Papp,FD4 in a wide range of TEER (at 0h) . Fig. 2 shows the individual X–Y pairs (TEER (at 0h) –Papp,FLU and Papp,FD4 ) and fitted curve profiles for the control, tightened and opened groups. The (apparently nonlinear) relationship between Papp,FLU and Papp,FD4 and TEER (at 0h) was described by the one-phase exponential decay model (Eq. (2), see Table 2 for the estimates of the parameters). For the control and the tightened groups, this relationship was found to be reasonably well described by the model (see Table 2 for the goodness of fit parameters); however, the fitted curves of the tightened group were just below the curves of the control group (Fig. 3). In the opened group, the Papp –TEER (at 0h) relationship was poorly fitted by the model (see Table 2 for the goodness of fit parameters) and resulted in curves that were way above those of the control group (Fig. 3).

Fig. 2. The Papp –TEER (at 0h) relationship and fitted curve profiles for FLU and FD4 of the individual X–Y pairs of the control, tightened and opened group, respectively. See text for more details.

P. J. Gaillard, A.G. de Boer / European Journal of Pharmaceutical Sciences 12 (2000) 95 – 102

99

Table 3 Fitted curves parameters (estimates195% confidence intervals) and goodness of fit parameters for the Papp –TEER ( 1 – 5h) profiles of the different groups (displayed in Figs. 4 and 5) for FLU and FD4

P‘leaky’ (10 26 cm / s)

FLU FD4

P‘barrier’ (10 26 cm / s) Fig. 3. The combined curves of the Papp –TEER (at 0h) relationship for FLU (left panel) and FD4 (right panel) of the control group (solid line), the tightened group (dashed line, high frequency) and the opened group (dashed line, low frequency). See text for more details.

3.4. Papp , categorized on the expected effect on BBB permeability and related to TEER(1 – 5 h) The group statistics of TEER ( 1 – 5h) are summarized in Table 1 for each group. No significant difference between the averaged TEER ( 1 – 5h) of the control, tightened and opened groups was found (Mann–Whitney (rank sum) test). The relative effect of the experimental circumstances (i.e., the simulated drug effects) on each individual filter was quantified by TEER ( 1 – 5h) and expressed as percentage of TEER (at 0h) . These effects (mean %6S.D.) were found to be 92.0611.3, 117.8613.2 and 55.9617.4 for the control, tightened and opened groups, respectively, and were significantly different from the control group (Mann– Whitney (rank sum) test). Fig. 4 shows the individual X–Y pairs (TEER ( 1 – 5h) –Papp ) and fitted curve profiles for the control, tightened and opened groups. The relationship between Papp,FLU and Papp,FD4 and TEER ( 1 – 5h) was described by the one-phase exponential decay model (Eq. (2), see Table 3 for the estimates of the parameters). For each group, this relationship was found to be satisfactorily described by the model (see Table 3 for the goodness of fit parameters). The CVs of the estimated P‘barrier’ for FLU and FD4 (estimated from the control group) were 5 and 6%, respectively. Moreover, the fitted curves of the tightened and opened groups were now superimposable to

FLU FD4

TEER1 / 2 (V?cm 2 )

FLU FD4

R2 Sy? x (10 26 cm / s)

FLU FD4 FLU FD4

Control

Tightened

Opened

22.3 (16.9–27.6) 9.6 (6.6–12.6) 2.1 (1.9–2.3) 0.46 (0.40–0.51) 19.0 (16.5–22.4) 16.3 (14.1–19.4) 0.92 0.88 0.86 0.40

9.3 (6.6–12.1) 4.6 (2.9–6.3) 1.3 (0.9–1.6) 0.26 (0.22–0.31) 40.6 (30.5–60.7) 26.9 (21.7–35.3) 0.86 0.91 0.67 0.18

19.4 (15.6–23.3) 7.6 (5.2–9.9) 2.0 (1.6–2.4) 0.41 (0.26–0.55) 28.3 (24.1–34.2) 25.1 (20.7–32.0) 0.91 0.85 0.81 0.40

those of the control group (Fig. 5). From the control group it was calculated that, in order to stay within the upper 95% confidential interval of P‘barrier’ , TEER ( 1 – 5h) should have stayed above 131?V cm 2 for FLU and 122?V cm 2 for FD4 at any time during the transport assay (Eq. (2)).

Fig. 5. The combined curves of the Papp –TEER ( 1 – 5h) relationship for FLU (left panel) and FD4 (right panel) of the control group (solid line), the tightened group (dashed line, high frequency) and the opened group (dashed line, low frequency). See text for more details.

Fig. 4. The Papp –TEER ( 1 – 5h) relationship and fitted curve profiles for FLU and FD4 of the individual X–Y pairs of the control, tightened and opened groups, respectively. See text for more details.

100

P. J. Gaillard, A.G. de Boer / European Journal of Pharmaceutical Sciences 12 (2000) 95 – 102

4. Discussion Preferably, the in vitro Papp of a drug represents a drug-specific value, that provides a reliable prediction for the in vivo Papp of the drug and hence is unrelated to the basal permeability status of the in vitro BBB model. However, the transference of results between different in vitro BBB models is problematic because the basal permeability status of the in vitro BBB is often unknown (or unreported) and varies considerably between laboratories (de Boer et al., 1999). Moreover, pharmacologically active substances may change the permeability status of the in vitro BBB, thereby changing the in vitro Papp of the drug. To proof this notion, we investigated the relationship between the permeability status of the in vitro BBB and the in vitro Papp for two markers for paracellular drug transport. The transport of the markers FLU and FD4 across the in vitro BBB is likely to occur by the paracellular route (i.e., size-dependent transport through the tight junctions), as for each individual filter Papp,FLU was about 4 times higher than Papp,FD4 (R 2 50.97). However, Papp,FLU and Papp,FD4 , analyzed without information about TEER (Table 1, pooled data), resulted in values with a large degree of variation, even when the values were categorized according to the expected pharmacodynamic effect (Table 1, control, tightened and opened). Since this variation was also observed in the control group, the expected pharmacodynamic effects on the permeability of the in vitro BBB were not likely to be the only reason for the variation in Papp,FLU and Papp,FD4 . In fact, it could be explained by the variation in the basal BBB permeability status (i.e., TEER (at 0h) , Fig. 2, control). In particular, Papp,FLU and Papp,FD4 could be divided into a part where the transport was independent of TEER (at 0h) (estimated by P‘barrier’ in Eq. (2)) and into a part where the transport was exponentially dependent of TEER (at 0h) (described by P‘leaky’ and K in Eq. (2)). Accordingly, Papp,FLU and Papp,FD4 could theoretically range from the value of P‘barrier’ up to the value of P‘leaky’ , depending on TEER (at 0h) . When the in vitro BBB was exposed to challenging experimental circumstances (serving as a simulation for pharmacodynamic effects), the relationship between TEER (at 0h) and Papp,FLU and Papp,FD4 did no longer follow the control group (Figs. 2 and 3 and Table 2). This relationship was only valid when Papp,FLU and Papp,FD4 were related to the actual pharmacodynamic effects on the permeability of the in vitro BBB (quantified by TEER (1 – 5h) , Figs. 4 and 5 and Table 3). By these means, a distinction could be made between the effect on Papp mediated by the basal permeability status of the in vitro BBB and the effect on the permeability of the in vitro BBB that was induced by pharmacodynamic effects. We consider the permeability coefficient estimated by P‘barrier’ in Eq. (2) as the ‘true’ (i.e., permeability status

independent) in vitro Papp for a drug, and therefore the most reliable predictive estimate for the in vivo Papp of the drug. As Madara (1998) reasoned and exemplified for inulin and mannitol transport in his review on the regulation of the movement of solutes across (epithelial) tight junctions, the relationship between electrical resistance and transport of solutes is nonlinear. This nonlinearity can be explained by the fact that solute transport (i.e., Papp ) is essentially dependent on the sum of transport across all junctional pathways, whereas total electrical resistance (i.e., TEER) is essentially dependent on areas with the lowest electrical resistance between single cells, even when these low resistance areas are present at a low density. Therefore, when changes in tight junction permeability are studied, large changes in Papp will be measured with a leaky in vitro BBB (i.e., when TEER values become below 150?V cm 2 ). While, with a tight in vitro BBB, this results in the situation that hardly any effect on Papp will be measured (i.e., when TEER values remained above 150?V cm 2 ). Since TEER across the in vivo BBB was determined to be approximately 2000?V cm 2 (Butt, 1995), no effect is likely to be seen on the in vivo Papp of a drug through (direct, moderate) pharmacodynamic effects on the permeability of the tight junctions with in vivo drug transport assays. Only when the BBB is disrupted (e.g., by disruption of the tight junctions, or by transendothelial channel or pore formation, induced by drugs or diseases), changes in the in vivo Papp of a drug are expected to be observed. Nevertheless, pharmacodynamic effects may still affect other processes involved in the BBB, permeability for the drug (i.e., amount of anionic sites, pinocytotic activity, influx or efflux transporter activity, cerebral or endothelial biotransformation, CSF clearance, pharmacokinetics and cerebral blood flow). The Papp value of a drug, given without the estimation of the permeability status of the in vitro BBB (by, e.g., TEER, FLU or FD4) during the drug transport assay, is less useful because it is unknown whether the value is a result from the initial permeability status of the in vitro BBB model or from a pharmacodynamic effect on the permeability of the in vitro BBB, or both. If FLU or FD4 are used as markers to quantify the permeability status of the in vitro BBB during a drug transport assay, then Papp,FLU and Papp,FD4 may not have exceeded Papp (10 26 cm / s) estimated by the upper 95% CI of P‘barrier’ (i.e., 2.3 and 0.53 for FLU and FD4, respectively, Table 3, control group). Particularly, with higher Papp,FLU and Papp,FD4 , the permeability status independent transport across the in vitro BBB was no longer guaranteed. Conversely, from these values it was calculated (with Eq. (2)) that TEER may not have been (at any time) below 131?V cm 2 for drugs comparable to FLU, or below the 122?V cm 2 for drugs comparable to FD4, in order to guarantee permeability status independent transport of the drugs. In agreement with these findings, Tewes et al. (1997)

P. J. Gaillard, A.G. de Boer / European Journal of Pharmaceutical Sciences 12 (2000) 95 – 102

recommended that filters with endothelial monolayers used in transport experiments exhibit a TEER of at least 100?V cm 2 or a permeability of less than 1% clearance for FD4 in 2 h. Furthermore, Madara (1998) and Wong and Gumbiner (1997) described comparable threshold values of TEER (i.e., .200?V cm 2 ) in epithelial cells, where the transport of mannitol and inulin was almost independent on the absolute value of transepithelial electrical resistance. PFLU and PFD4 estimated by P‘leaky’ is not equal to Papp,FLU and Papp,FD4 of an empty coated filter, because drug transport without functional tight junctions is still restricted by the physical presence of the cells. In our experiments, there was no indication that the cells had detached from the filters. In particular, from previous experiments Papp,FLU and Papp,FD4 of an empty coated filter were determined to be 131 and 93 (10 26 cm / s), respectively (Gaillard et al., in press), and these were 7 and 13 times higher than the highest Papp,FLU and Papp,FD4 found in our experiments, respectively. In addition, the observed variation in TEER (at 0h) (Table 1 and Fig. 2) was most likely due to differences in the starting material of the (primary) isolations of the brain capillaries and astrocytes (caused by biological variation and tissue quality), but also to handling procedures of the cells during the model preparation (Gaillard et al., in press). On the other hand, TEER across the in vitro BBB was readily increased to values way above the lower limit of permeability status independent transport by the addition of cAMP and / or phosphodiesterase inhibitors or glucocorticoids (Rubin et al., 1991; Hurst and Clark, 1998; Gaillard et al., in press). We have focused on the assessment of the permeability status and functionality of tight junctions between BCEC, as we are convinced that tight junctions are a prerequisite for the maintenance of all essential properties of the BBB that restrict the transport of drugs. In particular, tight junctions prevent the passive diffusion of hydrophilic drugs between blood and brain, but they also maintain cell polarity, by preventing movement of membrane proteins (e.g., transporters and enzymes) away from specific locations of the cell membrane (Cereijido et al., 1998). Moreover, when tight junctions are present, drugs are forced to penetrate the cell-membrane or to pass through the narrow subjunctional space, where they meet cytosolic, membrane-bound or extracellular metabolizing enzymes (Ghersi-Egea et al., 1995) or outwardly directed drug transporters like P-glycoprotein. Therefore, an in vitro BBB model without demonstrable tight junction functionality is neither a good model for the transport of drugs across the BBB, nor for studies on the functionality of other BBB properties. In conclusion, the assessment of basal BBB permeability status during drug transport assays is essential for an accurate estimation of the in vitro permeability coefficient of a drug. In order to make a reliable prediction for the in vivo permeability coefficient of a drug, it is also essential

101

to determine the drug-induced changes in the BBB permeability during the drug transport assay.

Acknowledgements The contribution of Professor D.D. Breimer to this paper is very much appreciated.

References Albert, J.L., Boyle, J.P., Roberts, J.A., Challiss, R.A., Gubby, S.E., Boarder, M.R., 1997. Regulation of brain capillary endothelial cells by P2Y receptors coupled to Ca 21 , phospholipase C and mitogen-activated protein kinase. Br. J. Pharmacol. 122 (5), 935–941. Artursson, P., 1990. Epithelial transport of drugs in cell culture. I: a model for studying the passive diffusion of drugs over intestinal absorptive (Caco-2) cells. J. Pharm. Sci. 79 (6), 476–482. Begley, D.J., 1996. The blood–brain barrier: principles for targeting peptides and drugs to the central nervous system. J. Pharm. Pharmacol. 48, 136–146. Borges, N., Shi, F., Azevedo, I., Audus, K.L., 1994. Changes in brain microvessel endothelial cell monolayer permeability induced by adrenergic drugs. Eur. J. Pharmacol. 269 (2), 243–248. Butt, A.M., 1995. Effect of inflammatory agents on electrical resistance across the blood–brain barrier in pial microvessels of anaesthetized rats. Brain Res. 696, 145–150. Cereijido, M., Valdes, J., Shoshani, L., Contreras, R.G., 1998. Role of tight junctions in establishing and maintaining cell polarity. Annu. Rev. Physiol. 60, 161–177. de Boer, A.G., Breimer, D.D., 1996. Reconstitution of the blood–brain barrier in cell culture for studies of drug transport and metabolism. Adv. Drug Deliv. Rev. 22 (1-2), 251–264. de Boer, A.G., Breimer, D.D., 1998. Cytokines and blood–brain barrier permeability. In: Sharma, H.S., Westman, J. (Eds.), Progress in Brain Research. Elsevier, Amsterdam, pp. 425–451. de Boer, A.G., Gaillard, P.J., Breimer, D.D., 1999. The transference of results between blood–brain barrier cell culture systems. Eur. J. Pharm. Sci. 8 (1), 1–4. de Vries, H.E., Eppens, E.F., Prins, M., Kuiper, J., van Berkel, T.J., de Boer, A.G., Breimer, D.D., 1995. Transport of a hydrophilic compound into the cerebrospinal fluid during experimental allergic encephalomyelitis and after lipopolysaccharide administration. Pharm. Res. 12 (12), 1932–1936. Deangelis, E., Moss, S.H., Pouton, C.W., 1996. Endothelial cell biology and culture methods for drug transport studies. Adv. Drug Deliv. Rev. 18, 193–218. Fischer, S., Renz, D., Schaper, W., Karliczek, G.F., 1995. In vitro effects of fentanyl, methohexital, and thiopental on brain endothelial permeability. Anesthesiology 82 (2), 451–458. Fischer, S., Renz, D., Schaper, W., Karliczek, G.F., 1996. Effects of barbiturates on hypoxic cultures of brain derived microvascular endothelial cells. Brain Res. 707, 47–53. Gaillard, P.J., Voorwinden, L.H., Nielsen, J.L., Ivanov, A., Atsumi, R., Engman, H., Ringbom, C., de Boer, A.G., Breimer, D.D., Establishment and functional characterization of an in vitro model of the blood–brain barrier, comprising a co-culture of brain capillary endothelial cells and astrocytes. Eur. J. Pharm. Sci., in press. Ghersi-Egea, J.F., Leininger-Muller, B., Cecchelli, R., Fenstermacher, J.D., 1995. Blood–brain interfaces: relevance to cerebral drug metabolism. Toxicol. Lett. 82–83, 645–653. Hurst, R.D., Clark, J.B., 1998. Alterations in transendothelial electrical

102

P. J. Gaillard, A.G. de Boer / European Journal of Pharmaceutical Sciences 12 (2000) 95 – 102

resistance by vasoactive agonists and cyclic AMP in a blood–brain barrier model system. Neurochem. Res. 23 (2), 149–154. Madara, J.L., 1998. Regulation of the movement of solutes across tight junctions. Annu. Rev. Physiol. 60, 143–159. Mitic, L.L., Anderson, J.M., 1998. Molecular architecture of tight junctions. Annu. Rev. Physiol. 60, 121–142. Rubin, L.L., Hall, D.E., Porter, S., Barbu, K., Cannon, C., Horner, H.C., Janatpour, M., Liaw, C.W., Manning, K., Morales, J., 1991. A cell culture model of the blood–brain barrier. J. Cell Biol. 115 (6), 1725–1735. Takeda, T., Yamashita, Y., Shimazaki, S., Mitsui, Y., 1992. Histamine decreases the permeability of an endothelial cell monolayer by stimulating cyclic AMP production through the H2-receptor. J. Cell Sci. 101 (Pt 4), 745–750.

Tamai, I., Tsuji, A., 1996. Drug delivery through the blood–brain barrier. Adv. Drug Deliv. Rev. 19 (3), 401–424. Tewes, B., Franke, H., Hellwig, S., Hoheisel, D., Decker, S., Griesche, D., Tilling, T., Wegener, J., Galla, H.J., 1997. Preparation of endothelial cells in primary cultures obtained from the brains of 6-month-old pigs. In: de Boer, A.G., Sutanto, W. (Eds.), Drug Transport Across the Blood Brain Barrier: in Vitro and in Vivo Techniques. Harwood Academic Publishers, Amsterdam, pp. 91–97. Wong, V., Gumbiner, B.M., 1997. A synthetic peptide corresponding to the extracellular domain of occludin perturbs the tight junction permeability barrier. J. Cell Biol. 136 (2), 399–409.