Abstract
In vitro inhibition of P-glycoprotein (P-gp) expressed in cells is routinely used to predict the potential of in vivo P-gp drug interactions at the human blood-brain barrier (BBB). The accuracy of such predictions has not been confirmed because methods to quantify in vivo P-gp drug interactions at the human BBB have not been available. With the development of a noninvasive positron emission topography (PET) imaging method by our laboratory to determine P-gp-based drug interactions at the human BBB, an in vitro-in vivo comparison is now possible. Therefore, we developed a high throughput cell-based assay to determine the potential of putative P-gp inhibitors [including cyclosporine A (CsA)] to inhibit (EC50) the efflux of verapamil-bodipy, a model P-gp substrate. LLCPK1-MDR1 cells, expressing recombinant human P-gp, or control cells lacking P-gp (LLCPK1) were used in our assay. Using this assay, quinine, quinidine, CsA, and amprenavir were predicted to be the most potent P-gp inhibitors in vivo at their respective therapeutic maximal unbound plasma concentrations. The in vitro EC50 of CsA (0.6 μM) for P-gp inhibition was virtually the same as our previously determined in vivo unbound EC50 at the rat BBB (0.5 μM). Moreover, at 2.8 μM CsA (total blood concentration), our in vitro data predicted an increase of 129% in [11C]verapamil distribution into the human brain, a value similar to that observed by us (79%) using PET. These data suggest that our high throughput cell assay has the potential to accurately predict P-gp drug interactions at the human BBB.
The in vivo importance of P-glycoprotein (P-gp) at the BBB has been well demonstrated by studies in mdr1a/b (-/-) mice. For example, compared with the wild-type mouse, in the mdr1a/b (-/-) mice, the brain/plasma concentration ratio (or the brain uptake) of the anti-human immunodeficiency virus protease inhibitors is increased 7- to 36-fold and anti-cancer taxanes, paclitaxel, or docetaxel are increased 6- to 28-fold, whereas that of verapamil is increased 8.5-fold (Endres et al., 2006). Similar data have been obtained in mice and rats where P-gp has been chemically ablated with selective inhibitors of P-gp such as PSC833, GF120918, and LY335979 (Lin and Yamazaki, 2003; Endres et al., 2006). For example, the brain/plasma ratio of verapamil is increased 24.1-fold when the rat is pretreated with cyclosporine A (CsA) (Hendrikse and Vaalburg, 2002). Based on these data and others, it has been widely postulated that P-gp plays a vital role in limiting drug distribution at the human BBB and that P-gp-based drug interactions will result in a profound increase in brain concentrations of the affected drugs and, therefore, their CNS efficacy or toxicity.
Although rodent studies make a compelling case for the importance of P-gp at the BBB in the CNS distribution of drugs, their ability to predict the magnitude of P-gp-based drug interactions at the human BBB has not been investigated. Due to safety and ethical reasons, it has not been possible to measure in vivo human BBB P-gp activity. With the development by our laboratory of a noninvasive positron emission topography (PET) imaging method to measure P-gp-based drug interactions at the human BBB, a quantitative comparison of drug interactions at the brain P-gp barrier is now possible (Sasongko et al., 2005; Hsiao et al., 2006).
In our PET imaging study, the brain distribution of [11C]verapamil, a well established P-gp substrate, was quantitatively monitored in healthy subjects in the presence and absence of the P-gp inhibitor CsA. The results showed that P-gp at the human BBB limits the entry of [11C]verapamil into the brain. In the absence of CsA, 11C radioactivity AUCbrain/AUCblood ratio was 0.55 ± 0.03, and it increased to 1.02 ± 0.05 in the presence of CsA. CsA almost doubled the entry of 11C radioactivity into the brain by inhibiting P-gp (Sasongko et al., 2005). Consistent with this result, in rats and at the same blood concentrations of CsA as those achieved in humans (∼3 μM), P-gp inhibition at the rat BBB was modest with excellent quantitative correlation with that obtained in humans [75 versus 79% increase in total 3H radioactivity distribution into the brain for rats and humans, respectively (Hsiao et al., 2006)].
Although the above correlation is excellent, it is a correlation of in vivo rodent data with in vivo human data. Therefore, we asked whether such interactions could be predicted from in vitro cell-based assay, which is high-throughput, simple, and cost-effective. In this communication, we report the inhibition of P-gp by CsA and several other inhibitors, in stable LLCPK1 cells expressing the recombinant MDR1 gene (LLCPK1-MDR1) (Woodahl et al., 2004), using verapamil-bodipy as a substrate. We determined whether the EC50 of P-gp inhibition by CsA in this cell system was predictive of the EC50 previously observed by us at the rat BBB (Hsiao et al., 2006). In addition, we asked whether the in vitro LLCPK1-MDR1 cells could quantitatively predict the inhibition of P-gp observed by us in vivo at the human BBB using PET and [11C]verapamil (Sasongko et al., 2005).
Materials and Methods
Materials. BODIPY FL verapamil, hydrochloride, culture media, fetal calf serum, medium supplements, and antibiotics were purchased from Invitrogen (Carlsbad, CA). All other reagents were of the highest grade available from commercial sources.
Cell Culture. Stable LLCPK1 cells expressing recombinant MDR1 (LLCPK1-MDR1) or control cells (no detectable P-gp expression) were grown in complete media consisting of RPMI 1640 medium supplemented with 10% (v/v) fetal calf serum and 1% (v/v) antibiotic-antimycotic and grown at 37°C in the presence of 5% CO2. The characteristics of these cells, including P-gp expression in the LLCPK1-MDR1 and the control cells, have been described elsewhere (Woodahl et al., 2004).
Verapamil-Bodipy Accumulation Assay. Cells with passage numbers 12 to 40 were plated at a density of 1 × 105 cells/well (100 μl per well) on a 96-well plate (Corning Life Sciences, Acton, MA) and grown overnight. The cells were washed with phosphate-buffered saline followed by incubation for 1 h at 37°C in serum-free media in the presence of 5% CO2 containing 0.3 μM verapamil-bodipy, 0.2% dimethyl sulfoxide, and varying concentrations of P-gp inhibitors (10 concentrations of each inhibitor with each concentration conducted in quadruplicate). After washing and replacing with phosphate-buffered saline, the verapamil-bodipy content of cells was measured by a fluorescence plate reader (ex/em: 485/535 nm).
Data Analysis. Using nonlinear regression (WinNonlin; Pharsight Corporation, Mountain View, CA), the Hill equation was fit to the fold increase in intracellular fluorescence (relative to that observed in the absence of the inhibitor) as a function of increasing inhibitor concentration. The mean EC50 of each inhibitor was determined from at least 3 independent experiments. Unless otherwise stated, data are presented as mean ± S.D.
Comparison with Previously Published in Vivo Rat and Human PET Data. We have previously published studies (Sasongko et al., 2005; Hsiao et al., 2006) where we determined the inhibition of P-gp-mediated radiolabeled-verapamil efflux across the BBB by CsA. In the rat study, we determined the EC50 of CsA P-gp inhibition at increasing pseudo steady-state CsA blood concentrations, whereas in the human study we determined P-gp inhibition at the BBB at a single steady-state blood CsA concentration of 2.8 μM.
In the rat study, anesthetized male Sprague-Dawley rats were administered i.v. infusion of CsA to achieve increasing pseudo steady-state blood concentrations until maximal P-gp inhibition was achieved. An i.v. bolus dose of [3H]verapamil (∼14 μCi) was administered when pseudo steady-state blood CsA concentration was achieved (n = 5 or greater per each concentration group). The animals were sacrificed 20 min after [3H]verapamil dose administration to determine blood, plasma, and brain 3H radioactivity by scintillation counting. Details of the sampling scheme, specimen processing techniques, and data analysis procedures are described thoroughly in our previously published study (Hsiao et al., 2006).
For the human study, experimental conditions similar to those in the rat study were used, except that the distribution of 11C radioactivity in the brain was measured at a single pseudo steady-state blood CsA concentration of 2.8 μM using PET. [11C]Verapamil (∼0.2 mCi/kg) was administered to healthy volunteers (n = 12, 6 females and 6 males) as an i.v. infusion over ∼1 min before and after at least 1 h of infusion of CsA (2.5 mg/kg/h). Arterial blood samples and brain PET images were obtained at frequent intervals over 45 min. The brain uptake of 11C radioactivity (brain/blood at 45 min) was determined in the presence and absence of CsA. For additional details on the methods, the reader is referred to Sasongko et al. (2005).
Results and Discussion
We had previously determined that a seeding density of 1 × 105 cells/100 μl medium/well with 24 h of incubation in the Corning Life Sciences plates was necessary to form a confluent monolayer. The monolayer confluence was critical in minimizing well-to-well variability because the fluorescence plate reader only measures a small region inside each well. Based on a prior optimization study carried out over a period of 4 h with 15 min measuring time intervals, 1 h of incubation time was necessary for the verapamil-bodipy and the inhibitors to achieve maximal inhibition of efflux. Similarly, based on a study using concentrations ranging from 0.1 to 5 μM, the optimal verapamil-bodipy incubation concentration was determined to be 0.3 μM to allow the range of intracellular fluorescence accumulations to fall within the linear range of the fluorescence plate reader (0–1.2 × 106 fluorescence relative units).
The EC50 of each inhibitor was determined based on 3 or more independent experiments (Table 1). The Hill equation was fit to the fold increase in intracellular fluorescence as a function of increasing inhibitor concentration to estimate the Emax, the EC50, and the Hill coefficient (γ). The observed EC50 values were specific to P-gp, as no or minimal change in intracellular fluorescence was observed in the control cells. Although the EC50 of many of these drugs has been previously determined, the reported values (Table 1) have a huge variability. This makes it impossible to use these values for in vitro-to-in vivo prediction of drug interactions. Hence, we determined the EC50 of these drugs in our own laboratory using a single methodology and a single transfected cell line.
The in vivo potency of an inhibitor will be determined by the ratio of the maximal therapeutic plasma concentration of the drug and EC50. Therefore, this ratio was calculated to provide a rank order of “potency” of the drugs to produce P-gp-based drug interaction at the human BBB. This ratio can be computed for either the total or unbound therapeutic plasma concentration of the inhibitor. Because it is not clear which should be used (Endres et al., 2006), we computed both these ratios. When the total plasma concentrations were used, the drugs predicted to be potent inhibitors of P-gp at the human BBB ranked as follows: tipranavir > quinine > quinidine > lopinavir > ketoconazole = itraconazole > amprenavir (Table 1). When the unbound therapeutic plasma concentrations were used, the rank order changed and became quinine > quinidine > amprenavir > CsA > itraconazole > lopinavir. In addition, the potential for these drugs to inhibit P-gp in vivo was considerably reduced. Nevertheless, quinine and quinidine were predicted to be potent inhibitors of P-gp activity in vivo.
The in vitro EC50 of CsA (0.6 ± 0.3 μM) (obtained using protein-free media) was remarkably consistent with the in vivo unbound EC50 of 0.47 ± 0.004 μM at the rat BBB (Table 1; Fig. 1) [the unbound EC50 value was computed using the reported CsA fraction unbound of 6% in the rat (Bernareggi and Rowland, 1991; Hsiao et al., 2006)]. In the presence of complete inhibition of P-gp, the maximal increase in verapamil accumulations in the LLCPK1-MDR1 cells is much less (∼150% increase, Fig. 1) than the maximal increase in the brain distribution of [3H]verapamil observed in the rat (Endres et al., 2006; Hsiao et al., 2006). This difference is not surprising, as the in vitro system does not incorporate binding of verapamil to brain tissue. Therefore, the ratio of the Emax of the two systems can be used as a scaling factor to conduct in vitro-to-in vivo predictions of verapamil-CsA interaction at the human BBB. Such a scaling factor assumes that the Emax at the human BBB is similar to that observed at the rat BBB.
We then asked whether this in vitro data as well as the in vitro EC50 of CsA, adjusted by the scaling factor, would have predicted the magnitude of in vivo P-gp inhibition obtained in humans using PET. In our human PET study (Sasongko et al., 2005), a steady-state CsA blood concentration of 2.8 μM (0.2 μM unbound) resulted in a 79% increase in the distribution of 11C radioactivity into the brain. At the same unbound CsA blood concentration, our in vitro studies predicted a similar increase of 129% in human brain/blood radioactivity (Fig. 2).
The above data have demonstrated the utility of our high throughput in vitro assay to predict verapamil-CsA interaction at the rat and human BBB. We recognize that these excellent correlations are all based on a single P-gp substrate-inhibitor combination, verapamil-CsA. Similar studies with additional P-gp inhibitors are needed to test whether these excellent in vitro-in vivo correlations can be extended to other inhibitors (i.e., generalized). For this reason we have determined the potency (EC50) of other drugs to inhibit P-gp at the human BBB (Table 1). Based on these data, the drugs that should be tested in humans are quinine and quinidine. Indeed, quinidine has been shown to increase the CNS effects of loperamide in humans (Sadeque et al., 2000). Such validation with other inhibitors is particularly important for P-gp because P-gp has multiple substrate/inhibitor binding sites (Martin et al., 2000). Thus, it is possible that the [11C]verapamil-CsA inhibition in our human study might have been more profound if another inhibitor had been used.
In conclusion, this is the first in vitro study to quantitatively predict the in vivo inhibition of P-gp transport activity at the human BBB. The remarkable agreement between the in vitro and the in vivo data suggests that our in vitro cell culture method has the potential to be an excellent model to predict the in vivo inhibition of P-gp at the human BBB. The utility of this high throughput in vitro assay, in conjunction with the rat, to predict P-gp-based drug interactions at the human BBB seems to be promising. As discussed above, additional in vitro and in vivo human studies with other inhibitors, such as quinine or quinidine, are needed to further validate this excellent in vitro-to-in vivo prediction.
Footnotes
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.107.018176.
-
ABBREVIATIONS: P-gp, P-glycoprotein; BBB, blood-brain-barrier; CsA, cyclosporine A; PET, positron emission tomography; PSC833, valspodar; GF120918, elacridar, N-(4-[2-(1,2,3,4-tetrahydro-6,7-dimethoxy-2-isoquinolinyl)ethyl]-phenyl)-9,10-dihydro-5-methoxy-9-oxo-4-acridine carboxamide; LY335979, zosuquidar; CNS, central nervous system.
- Received August 14, 2007.
- Accepted December 4, 2007.
- The American Society for Pharmacology and Experimental Therapeutics