Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Concomitant administration of several drugs to patients has become a common practice in recent decades. Drug combinations are given frequently not only to patients undergoing cancer chemotherapy but also to the elderly (Balis, 1986; Lamy, 1986). It is possible that concomitantly administered drugs may interact and elicit unwanted side effects that are not associated with the individual use of the drugs (Balis, 1986; Lazarou et al., 1998). Drug-induced side effects may be caused by altered pharmacokinetics (Floren et al., 1997), altered drug distribution (Schinkel et al., 1996), or unpredictable toxicity (Kedderis, 1997).

One cause of drug-drug interactions may be the modulation of the natural function of P-glycoprotein (Pgp), a 170-kDa protein that is expressed in some cells on the plasma membrane and that can efflux chemically distinct xenobiotics and drugs from cells (Gottesman and Pastan, 1993). Pgp is expressed in humans at the blood-brain barrier (BBB) and in kidneys, liver, and adrenal glands (Chin et al., 1989; Sugawara, 1990; Tatsura et al., 1992). However, the function of Pgp has been studied mainly in connection with resistant cancer cells (Raderer and Scheithauer, 1993). Modulating or blocking the function of Pgp in tumors using coadministered drugs in patients with cancer may simultaneously block Pgp at organ sites, and such blocking may cause pharmacokinetic changes of the primary drug (Floren et al. 1997), or it may also cause changes in drug distribution, for example, into the central nervous system (Didier and Loor, 1995; Gianni et al., 1997), or changes in renal excretion (Balis, 1986; Hori et al., 1993).

Multiple drug treatments for elderly patients undergoing cancer chemotherapy or with infectious diseases, depression, or other indications warrant the assessment of possible drug-drug interactions at the Pgp level. Similar assessments were made for the P450 metabolic enzyme system. For many drugs, the possible interactions with the function of this enzyme system are well studied (Kim et al., 1999). It was shown recently that there is no correlation in substrate specificity between the P450 enzyme system and Pgp (Wandel et al., 1999).

We have assessed the modulating effect of three classes of drugs on Pgp in MDR1 gene-expressing cell lines. Such cell lines were used previously in related studies (Bergman et al., 1998). We have compared the Pgp modulating concentrations of the studied drugs with their therapeutic blood concentrations.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Lines. NIH3T3 and NIH3T3/MDR1 cells were grown as described previously (Currier et al., 1992; Hwang et al., 1996). Caco-2 cells were obtained from the American Type Culture Collection (Manassas, VA). These cells were transfected with the pHaMDR1/A retroviral vector carrying the human MDR1 gene as described previously by Pastan et al. (1988). The expression of Pgp was verified by flow cytometric evaluation of cells reacted and stained with the Pgp-specific mAb MRK16 (a generous gift of Dr. M. Gottesman, National Institutes of Health) plus anti-mouse IgG-fluorescein isothiocyanate (Sigma, St. Louis, MO). Pgp-expressing Caco-2 cells were also produced by stepwise selection with etoposide (VP16, Sigma) (10-60 µg/ml, 3-6 weeks for each step). The Caco-2 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Grand Island, NY) containing 10% fetal bovine serum, 5 mM L-glutamine, penicillin, and streptomycin (Life Technologies) at 37°C and 5% CO₂ atmosphere. The transfected cells were kept under selection with 60 ng/ml VP16 (Sigma). These cells were called Caco-2/MDR1. Human BBB capillary endothelial cells were obtained from Professor J. Molnar (Medial School of Szeged, Hungary). This immortalized cell line originated from Prudhomm et al. (1986) and its human endothelial cell properties have been characterized. Because it did not express Pgp, it was transduced with the MDR1 gene as follows. Retroviral producer cells were generated as described previously by transfection of PA317 packaging cells with plasmid pHaMDR1/A containing a full-length MDR1 cDNA (Sugimoto et al., 1998). Producer cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies) containing 10% fetal bovine serum, 5% glutamine, penicillin and streptomycin, and vincristine at 30 ng/ml and subcloned repeatedly. Transduction experiments were performed with minor modifications as reported by Gottesman et al. (1998). Briefly, producer cells were grown to confluency. On the day before the current transduction experiments, the supernatant from logarithmically growing endothelial cells was removed and replaced with 10 ml of filtered supernatant (0.45 µm) from retroviral producing cells. This supernatant was supplemented with 8 µg/ml polybrene (Sigma). Forty-eight hours later, the virus-containing supernatant was replaced with fresh medium supplemented with vinblastine at 15 ng/ml. This selecting medium was exchanged every 2 to 3 days. These cells were called BBB endothelial/MDR1. Endothelial cells were grown in RPMI medium supplemented with penicillin plus streptomycin, 10% fetal bovine serum (Life Technologies), and endothelial cell growth factor (Sigma) at 37°C and 5% CO₂ atmosphere. The transduced cells were kept under selection with 15 ng/ml vinblastine.

Modulation of P-glycoprotein in NIH3T3/MDR1, Caco-2/MDR1, and BBB Endothelial/MDR1 Cells by Drugs as Tested by Flow Cytometry. Modulation of Pgp function by drugs was assessed using flow cytometry and the fluorescence substrates R123, calcein AM, or daunorubicin (DN; Sigma) as described previously (Hwang et al., 1996). The latter two substrates were used to confirm R123 results. Both parental and Pgp-expressing cells were treated in parallel with a drug or with a combination of drugs. The mean FI of histograms obtained with the cells was calculated using FACScan software (Becton Dickinson, Mountain View, CA) and was expressed as percentage of inhibition. Percentage of inhibition was calculated as:
FI_MDR1 and FI_par are FI of the Pgp-expressing MDR1 and the parental cells, respectively. For the best expression of dose dependence of individual drugs, the scale of drug concentrations was varied in the tables under Results.

Effect of the Modulation of P-glycoprotein Function by Drugs on the Proliferation of BBB Endothelial Cells. Parental and BBB endothelial/MDR1 cells were used in this assay. Parental and Pgp-expressing cells were adjusted to 25 × 10⁴ cells/ml in complete medium. Cells were then treated with the potential Pgp blockers (all from Sigma). After a 15-min incubation, 80 ng/ml DN was added, and the cell suspensions were distributed into 24-well plates. Some cells were treated with the potential Pgp-modulating drugs only to assess the effect of these drugs alone. After a 3-day incubation at 37°C and 5% CO₂ atmosphere, the cells were treated with Alamar blue (Alamar, Sacramento, CA). This assay is based on the oxidation-reduction potential of the cells affected by the dye indicator in the medium, resulting in fluorescence and color changes. The developed color was read on a cytofluor system (AstroScan, Isle of Man, British Islands) after 1 h. The percentage of growth was calculated as the ratio of dye absorption intensity for cell lines treated with potential Pgp blockers over cell lines treated with both potential Pgp blockers and the Pgp substrate, DN. After appropriate dose-finding experiments for DN and drugs, two final experiments were performed, each in duplicate. One of these experiments is presented here in this article.

Monolayer Studies with BBB Endothelial Cells. Monolayer studies were performed with BBB endothelial cells essentially as described previously for Caco-2 cells (Delie and Rubas, 1997; Anderle et al., 1998) but with the following modifications. Cells were plated in complete medium at 2 × 10⁴ cells per Transwell insert (Corning Costar, Cambridge, MA). After 5 days, the medium was changed, and when cells reached confluence, as determined by microscopic examination and by measuring the trans-epithelial electrical resistance, assays were carried out. Modulators of Pgp were added to both chambers in serum- and phenol red-free RPMI medium (Sigma). DN was added to the apical or the basolateral chamber depending on the direction of the measurement. Chambers were kept at 37°C during the experiment. To test for the expression of functional Pgp in monolayers, mAb MRK16 plus anti-mouse antibody-fluorescein isothiocyanate (Sigma) or R123 was added to the monolayers and after an appropriate incubation time, cells were trypsinized, and fluorescence was measured by flow cytometry as noted above. Parental BBB endothelial cells were used as control for both Pgp expression and functionality. Samples were diluted 1:1 with dimethyl sulfoxide and assayed for DN using reversed-phase HPLC with tandem mass spectrometry detection. The instrument used was an APl 2000 triple quadrapole mass spectrometer (Perkin-Elmer Life Sciences, Foster City, CA) with a TurboSpray interface. The mass spectrometer was operated in positive-ion tandem mass spectometry mode monitoring the transition 528 m/z (MH⁺) to 321 m/z using a dwell time of 350 ms. The orifice and ring voltages were set at 35 and 350 V, respectively. The nebulizer, heater, and curtain gases (nitrogen) were set at 41, 35, and 45 psi, respectively, at a temperature of 325°C. The ion spray voltage was set at 55 kV. Nitrogen (5 psi) was also used as the collision gas.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Surface Expression of P-glycoprotein. The number of Pgp molecules, products of MDR1 gene expression, on the plasma membrane of the cell lines are shown in Table 1. The MRK16 mAb recognizes one epitope on each Pgp molecule (Georges et al., 1993). As expected, the transfected and transduced cell lines showed increased numbers of Pgp molecules per cell compared with the parental cell lines. The observed fluorescence values for the parental cells were superimposable or close to the autofluorescence values of the cells, and therefore no real number of Pgp/cell (detection limit = 250 epitope/cell) could be calculated for them. These results indicate that the purchased Caco-2 cells did not have measurable amount of surface Pgp molecules. Other investigators also found that Pgp expression on Caco-2 cells depends on the number of passages and of days culturing in monolayers (Hosoya et al., 1996).

Functional Assessment of P-glycoprotein on Transfected, Transduced, and Selected Cells. Function of Pgp in transfected and selected Caco-2 cells in the parental Caco-2 cells and in the BBB endothelial cell lines was assessed by R123, calcein AM, and DN uptake assays. Assays were carried out with flow cytometry in the presence and absence of the Pgp modulators PSC833 and verapamil as described under Materials and Methods.

View larger version (32K): [in this window] [in a new window]   Fig. 1.   Fluorescence histograms of 4 × 104 cells. A, Caco-2/MDR1 cells without (a) and with 2 µg/ml PSC833 (b) using rhodamine123 as a substrate. B, same as A but using calcein AM as substrate. C, VP16-selected Caco-2/MDR1 cells using rhodamine123 as substrate without (a) and with 2 µg/ml verapamil (b); histogram for parental cells is shown under (c). D, same as C but using calcein AM as substrate. E, rhodamine123 intake into parental (a), transduced (b), and parental cells (c), and 2 µg/ml verapamil-treated BBB endothelial/MDR1 cells (d). F, same as E but using 2.5 µg/ml daunorubicin as substrate.

Modulation of P-glycoprotein Function by Some Drugs as Tested by Flow Cytometry. Modulation of Pgp by antiemetic drugs was tested in NIH3T3/MDR1, Caco-2/MDR1, and BBB endothelial/MDR1 cells. NIH3T3/MDR1 cells were used as a reference cell line because they had been used previously to evaluate Pgp modulation (Hwang et al., 1996). In general, incubations were performed using concentrations higher than the usual therapeutic blood levels. However, if the drug did not block Pgp function in each of the cell lines at a reasonably high concentration, no additional testing was performed. Variability (%CV) was lowest with the NIH3T3 cells; %CV generally did not exceed ±10% with the NIH3T3 cells and ±15% with the other cell lines.

View larger version (73K): [in this window] [in a new window]   Fig. 2.   Dose-dependent inhibition of Pgp in NIH3T3/MDR1 (), Caco-2/MDR1 (), and BBB endothelial/MDR1 () cells by antiemetic drugs using rhodamine123 as a substrate. Shaded areas represent clinical blood levels (ng/ml). Calculation of percentage of inhibition is given under Materials and Methods and based on means of fluorescence histograms (n = 4 × 104 cells). Both substrates (DN and rhodamine123) gave similar values of inhibition in these cell lines. Percentage of inhibition by the drugs varies from cell to cell line. Each point represents one typical result within the same assay (n = 2-4). Usual therapeutic blood levels are indicated with letter codes. A and B, Sridhar et al. (1994); C, Guichard et al. (1993); D and E, Physicians' Desk Reference (1997; pp 2882 and 2807, respectively); F, Ratnakar et al. (1995); G, Vogt et al. (1993); H, not found.

View larger version (68K): [in this window] [in a new window]   Fig. 3.   Dose-dependent inhibition of Pgp by Ca2+ channel blocking drugs in NIH3T3/MDR1 () and BBB endothelial/MDR1 () cells using rhodamine123 as substrate. Shaded areas represent clinical blood levels. Interpretation and calculation of results are as indicated in Fig. 2. a, Physicians' Desk Reference [1997; pp 2261 (Nicardipine), 1467 (Diltiazem), and 1597 (Bepridil)]; b, Gilman et al. [1990; pp 1715 (Verapamil) and 1696 (Nifedipine)]; c, Li et al. (1998).

View larger version (64K): [in this window] [in a new window]   Fig. 4.   Dose-dependent inhibition of Pgp by antipsychotic drugs in Caco-2/MDR1 cells using rhodamine123 as substrate. Interpretation and calculation of results are as indicated in Fig. 2. PDR, Physicians' Desk Reference [1997; pp 1037 (Pimozide), 2377 (Clozapine), and 819 (Clomipramine)]; GG, Gilman et al. [1990; pp 1683 (Haloperidol), 1673 (Desipramine), 1697 (Nortriptyline), 1677 (Doxepin), 1658 (Amitriptyline), 1668 (Chlordiazepoxide), and 1669 (Chlorpromazine)]; A, Dencker et al. (1994); B, Janicak et al. (1989); C, Gustarvon and Carrigan (1990); D, Bertler et al. (1980); E, Musa (1989); F, Miyake et al. (1991).

Modulation of P-glycoprotein Function by Combinations of Drugs as Tested by Flow Cytometry. Table 2 shows the effect of treatment of BBB endothelial/MDR1 cells with the combination of two Pgp blocking drugs on the uptake of daunorubicin. Both drugs were used at or lower than therapeutic blood concentrations and individually inhibited Pgp only partially at these concentrations (38-77%). Combinations of these drugs additively inhibited (up to 100%) Pgp function, i.e., caused a higher uptake of daunorubicin into the cells than did inhibition by each drug separately.

Sensitization of P-glycoprotein-Expressing Cells to Daunorubicin by Some Antiemetic Ca²⁺ Channel Blockers and Antipyschotic Drugs in Cultures. Several drugs found to modulate Pgp function by the flow cytometric method were selected for the proliferation assay (Figs. 2-4). In preliminary experiments, doses that were maximally effective for Pgp modulation but otherwise not toxic to the cells were determined (data not shown). Results with prochlorperazine and droperidol (antiemetics), bepridil and nicardipine (Ca²⁺ channel blockers), and trifluoperazine and haloperidol (antipsychotic drugs) are shown in Fig. 5. The known Pgp modulators PSC833 and ketoconazole were used as positive controls. Results are expressed as the mean of duplicate experiments. Proliferation of the BBB endothelial/MDR1 cells shows dose dependence of the Pgp modifier drug as shown for ketoconazole. For each selected drug, the indicated concentration made the BBB endothelial/MDR1 cells sensitive to daunorubicin.

View larger version (63K): [in this window] [in a new window]   Fig. 5.   Cell proliferation of BBB endothelial/MDR1 and parental cells in the presence of selected antipsychotic, antiemetic, and Ca2+ channel blocker drugs. PSC833 and ketoconazole are positive controls. Cell proliferation was measured by the Alamar blue assay. All DN- and drug-treated MDR cells differ from DN-only treated cells with statistical significance (P < .05, n = 3-6) except ketaconazole (450 ng/ml; P > .05). Growth of untreated cells was taken as 100%.

Monolayer Studies. Monolayer studies were performed with parental and BBB endothelial/MDR1 cells because BBB endothelial cells formed monolayers in less time than did Caco-2 cells (6 days versus 25 days, respectively). Figure 6A shows the results with parental and BBB endothelial/MDR1 cells. The diffusion rate of DN (apical to basolateral and vice versa) is approximately the same in both directions. This fact may mean that transduction of the MDR1 gene into these cells resulted in a nonpolarized expression of the Pgp protein. PSC833 slowed the transport of DN in both directions. The Pgp-specific mAb MRK16 added at both sides of the monolayer also slowed the transport of DN as measured in the basolateral to apical direction. Figure 6B shows results with examples from the three drug classes studied. Drugs were added to both sides of the monolayer, and transport of DN was studied in the apical to basolateral direction. Furosemide (40,000 ng/ml), a drug that did not block the function of Pgp, also did not alter the transport of DN through monolayers and served as a negative control (data not shown).

View larger version (54K): [in this window] [in a new window]   Fig. 6.   Effect of drugs or mAb MRK16 on the diffusion of daunorubicin through BBB endothelial or BBB endothelial/MDR1 cell monolayers in Transwell cell culture chambers. A, effect of PSC833 in the apical to basolateral and the basolateral to apical directions and the effect of mAb MRK16. The maximal S.D. values for all these measurements are shown for untreated monolayer cells. B, effect of drugs in the apical to basolateral direction. Differences are significant (P < .05, n = 3) in each case except untreated cells between the two directions (A) and for astemizole at 30 min (B).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we assessed the ability of some antiemetics, antipsychotics, and Ca²⁺ channel blockers to modulate the function of Pgp. Drug concentrations able to modulate Pgp function are related to those of clinical therapeutic blood levels. Drugs were tested in human Caco-2 and BBB endothelial cells into which functional Pgp was introduced by either transfection, stepwise selection, or transduction.

The success of MDR1 gene transfection and transduction experiments was assessed by determining the number of Pgp molecules on cell surfaces and by functional assays (Table 1 and Fig. 1). Also, because the function of Pgp in BBB endothelial/MDR1 cell monolayers was not known, we ascertained that function in cells grown on the Transwell inserts (data not shown). These cells were used to assess the ability of drugs to modify the function of Pgp. To relate findings with Caco-2 and endothelial cells to cells with which many Pgp blockers were already tested, we also used parental and MDR1 gene-transfected NIH3T3 cells (Currier et al., 1992; Hwang et al., 1996). Drugs were tested for their ability to modulate the function of Pgp in more than one cell line. The reason for this multiple testing is that plasma membrane composition of different cells may influence the substrate specificity of Pgp (Zijlstra et al., 1987; Saeki et al., 1992; Callaghan et al., 1993). Naturally, this argument is also valid for the cells expressing Pgp in humans. Our studies with the different cell lines essentially point in the direction that the blocking effect may be different for one drug in different human tissues. Furthermore, the number of Pgp molecules expressed in each cell line and in human tissues can also determine the blocking effect of a drug.

For the assessment of modulation of Pgp function by the different drugs, we used primarily a flow cytometric technique. This technique was used previously to determine the effects of MDR reversing agents on the uptake of epirubicin into Caco-2 cells (Dordal et al., 1995; Lo et al., 1998). The flow cytometric method for the estimation of Pgp-mediated drug resistance was also detailed previously (Aszalos and Weaver, 1998). One should note that results of these measurements reflect apparent blocking effects of drugs on Pgp. Results of the flow cytometry studies were verified using cell proliferation studies (Fig. 5).

The flow cytometry assays revealed that four of the eight commonly tested antiemetic drugs block Pgp at concentrations used to achieve pharmacological effects in patients (Fig. 2). Chlorpromazine, prochlorperazine, droperidol, and triflupromazine block Pgp at such concentrations in all three cell lines. Promethazine blocks at a concentration of 2000 ng/ml, and droperidol blocks Pgp at a concentration of 2500 ng/ml. The other three drugs, lorazepam, trimethobenzamide, and metoclopramide, do not block Pgp even at concentrations much higher than clinical blood levels.

Nicardipine and bepridil block Pgp at concentrations below therapeutic blood levels, 50 and 100 ng/ml, respectively (Fig. 3). Verapamil blocks Pgp at a concentration of 250 ng/ml (5% blocking). Nifedipine, diltiazem, and nimodipine block Pgp at concentrations higher than their therapeutic blood levels. It is worthy to mention that for complete blocking of Pgp function, a concentration of verapamil higher than 2 µg/ml is necessary (Fig. 3). For this reason, early clinical trials with resistant tumors were conducted with verapamil as the Pgp blocker at this high blood level (Raderer and Scheithauer, 1993), with the known side effect of cardiotoxicity.

Twenty antipsychotic drugs were tested with NIH3T3 and Caco-2 cell lines. Drugs with a diazolam-type structure do block Pgp function but only at higher doses. Contrary to this, phenothiazines block Pgp function at concentrations below 500 ng/ml (Fig. 4).

Some of the tested drugs block the function of Pgp incompletely at concentrations that are within their therapeutic range. One should consider, however, that the simultaneous use of more than one such drug could result in additive or synergistic Pgp modulation and greater inhibition of the Pgp function. To demonstrate this possibility, we treated BBB endothelial/MDR1 cells with the combination of ketoconazole and triflupromazine. These two Pgp-affecting drugs may be coadministered to patients. Ketoconazole was used at and below therapeutic blood concentrations with therapeutic blood concentrations of triflupromazine (Caron and Walsh, 1992) (Table 2). Also, in our experiments, we used DN, the anticancer drug, as a fluorescent substrate. As expected, combinations of these two Pgp-inhibiting drugs blocked Pgp to a greater extent than did either drug used singly. Such additive effects were shown previously for PSC833, polyoxyethylated castor oil, and verapamil (Hwang et al., 1996). It should be noted here that levels of Pgp expression in human tissues are known only qualitatively (Van de Vrie et al., 1998) and that these levels may vary from patient to patient. Therefore, our studies can predict only possible interactions of Pgp in humans. In addition, the relative binding of individual drugs to Pgp and to serum proteins is not known. Blocking Pgp in humans may be influenced by its relative binding affinities. However, we believe that the results shown with our cell lines expressing high levels of Pgp could indeed be relevant in human tissues.

Cell proliferation assays were used to confirm the results of the flow cytometric assays (Fig. 6). Two antipsychotic, two antiemetics, and two Ca²⁺ channel blockers as well as positive controls PSC833 and ketoconazole all sensitized BBB endothelial/MDR1 cells to the cytotoxic effect of DN.

Monolayer studies (Fig. 6) indicated that drugs that alter the function of Pgp as determined by the flow cytometric technique also alter the transport of DN through monolayers. However, this alteration is approximately equal in both directions because BBB endothelial/MDR1 cells express Pgp in a nonpolarized manner, unlike Caco-2 cells. Therefore, alteration of the function of Pgp by a drug in Caco-2 monolayers results in unequal diffusion rate changes in an apical to basolateral versus a basolateral to apical direction (Delie and Rubas, 1997; Anderle et al., 1998).

Diltiazem inhibits Pgp at a higher concentration (500 µg/ml) than the usual therapeutic blood concentration (200 µg/ml; Fig. 3). When this drug was coadministered with tacrolimus to a liver transplant patient, it resulted in atrial fibrillation and neurotoxicity. It was suggested that inhibition of Pgp may be one of the reasons for this drug-drug interaction-induced symptom (Hebert and Lam, 1999). This hypothesis is supported by the fact that tacrolimus was shown previously to inhibit Pgp (Weaver et al., 1993). In many cases, the exact cause of the unwanted side effects resulting from coadministration of several drugs was reported to be unknown. We believe that the results reported in this article may shed light on some of these clinically observed side effects. Our studies continue with different classes of drugs, and the need for such studies is highly recommended in a recent article by Yu (1999).