Induction of acquired drug resistance in endothelial cells and its involvement in anticancer therapy

Mouse monoclonal antibodies against human P-gp: C219 were obtained from Calbiochem, La Jolla, CA; 4E3 from Dako, Glostrup, Denmark; and 265/F4 from Abcam, Paris, France. Antibody MRK16 blocking P-gp function was obtained from Kamiya Biomedical Company (Seattle, WA). The anti-ABCG2 antibody BXP-21 came from Abcam and the anti-MRP1 antibody QCRL-1 from Santa Cruz Biotechnology Inc., CA. The antibodies against vWF, flt-1, CD31, or CD105 as well as the FITC or HRP-conjugated F (ab’) ₂ fragment of goat anti-mouse IgG were all provided by Dako. Doxorubicin chlorhydrate was purchased from Amersham Pharmacia Biotech (Uppsala, Sweden). Rhodamine 123 and Verapamil were obtained from Calbiochem and Daunorubicin, Etoposide, Vinblastine, Cyclosporine A, Fumitremorgin C, and Diethylstibesterol Terfenadine were provided by Sigma Chemical Co. (Saint Louis, MO).

Parental and resistant HMEC-1 (Dr TL Lawley, Department of Dermatology, Atlanta) lines were cultured in MCDB-131 medium supplemented with 10% fetal calf serum (FCS), 2 mM L-glutamine, 10 ng/ml EGF, 1 μg/ml hydrocortisone, 100 units/ml penicillin, and 100 μg/ml streptomycin as described elsewhere [38]. Dox-resistant HMEC cells were obtained by continuously exposing cells to escalating concentrations of Dox from 0.001 μg/ml to 0.24 μg/ml over a 12-week period. Two subcell lines of HMEC-1 cells were collected: one was maintained in a culture with 0.08 μg/ml Dox (HMECd1 cells), and another with 0.24 μg/ml Dox (HMECd2 cells). No mutagenic agents were used in the establishment of these Dox-resistant HMEC cells. In the experiments looking at the reversibility of Dox resistance, both HMECd1 and HMECd2 cell lines were cultured in complete medium without Dox for four weeks. HUVEC were isolated as reported elsewhere [39] and seeded on a 1% gelatin-coated plastic flask in MEM-199 medium supplemented with 20% FCS, 15 mM sodium bicarbonate, 15 mM hepes, 2 mM L-glutamine, 10 ng/ml EGF, 1 μg/ml hydrocortisone, 100 units/ml penicillin, and 100 μg/ml streptomycin. Human breast adenocarcinoma cells MDA-MB-435 were cultured in DMEM medium containing 10% FCS, 2 mM sodium pyruvate, 1 mM L-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin. All types of cells were digested with trypsin-EDTA once or twice a week and cultured in a 37°C incubator with a 100% humidified atmosphere of 5% CO₂.

Parental and resistant HMEC sublines were seeded at a density of 4 x 10⁴ cells per well in 48-well culture plates and exposed to a range of drug concentrations for 72 hours at 37°C in an atmosphere of 5% CO₂. After 70 hours incubation, 1 μCi ³H-thymidine (Amersham Pharmacia biotech) was added per well for 2 hours. Wells were then washed twice in PBS and successively incubated with 5% trichloroacetic acid for 20 minutes at 4°C and then 0.5 N NaOH for 90 minutes at 37°C. Radioactivity incorporated into adherent cells was recorded on a β counter (Beckman). The 50% cytotoxic concentration (IC50) values were defined as the drug concentration producing 50% inhibition of cell growth and the resistance index (RI) corresponded to the ratio of IC50 values between the resistant and parental cell lines.

Cell viability was determined using the MTS cell proliferation assay (Promega). Cells grew to a confluence of 90% in 75 cm² cell culture flasks and were passed into 96-well plates (7500 cells/well). Each well contained 100 μl of culture medium supplemented with various concentrations of drugs or with a concentration of DMSO as control. After incubation for either 24, 48, or 72 hours, 20 μl of the MTS reagent was added to each well, and the plate placed in the 5% CO₂ incubator at 37°C for an additional 2 hours. The optical density (OD) was then read at 492 nm using a microplate reader (Labsystems Multiskan MS). The IC50 values were defined as the concentration of drug producing 50% inhibition of cell growth and the RI corresponded to the ratio of IC50 values between the resistant and parental cell lines. Experiments were performed in triplicate and repeated at least three times.

P-gp inhibitors Cyclosporine A at 2.5 μM or Verapamil at 1 μM and ABCG2 inhibitors Fumitremorgin C at 5 μM or Diethylstibesterol at 0.5 μM were used in these experiments. After incubation for 48 or 72 hours, cell viability was assessed by the MTS assay. The reversal fold (RF) values, as a measure of the potency of reversal, were obtained from fitting the data to RF = IC50 of cytotoxic drug alone/IC50 of cytotoxic drug in the presence of a modulator [40].

HMEC-1, HMECd1, and HMECd2 cells (10⁶/ml in PBS-BSA) were incubated with 1–2 μg/ml Rho in the dark at 37°C in 5% CO₂ for one hour. Then, the cells were washed twice with ice-cold PBS and analyzed immediately using flow cytometry at different time points. To test Rho efflux specificity, cells were incubated with 30 μM Verapamil or 10 μg/ml MRK16. Results were expressed in an arbitrary unit of the mean fluorescence intensity (MFI). The drug efflux was expressed relative to the amount of drug accumulated.

HMEC-1, HMECd1, and HMECd2 cells were treated with 2.5 μM Cyclosporine A, 1 μM Verapamil, 5 μM Fumitremorgin C, or 0.5 μM Diethylstibesterol for 24 hours. After incubation, the treated and non-treated cells were harvested and total RNA prepared using the SV total RNA isolation system kit (Promega, USA). The purity of total RNA was checked by a ratio of A260/A280 (>1.9). Total RNA (50 ng) was used to synthesize the first-strand cDNA in a 20 μl reaction solution using the GoScript Reverse Transcription System kit (Promega, USA). Then, 2 μl of cDNA was used for qPCR in triplicates using a taqman® gene expression assay, the primers for P-gp (Hs01067802_m1), ABCG2 (Hs01053790_m1), and the primers for TBP as controls (TATA box binding protein, Hs99999910_m1, Applied Biosystem). The qPCR was performed by 10 minutes of initial denaturation followed by 44 cycles of 15 s at 95°C and 60 s at 60°C in a BioRad CFX96® Real-time System. Delta Ct method was used for analyzing the qPCR results and TBP was used as an internal control for mRNA-level normalization.

Western blot was performed on whole cell lysates by incubating the cells in the lysis buffer (10 mM Tris pH 6.8, 1 mM EDTA, 10% NP40, 1 mM PMSF, 0.1% SDS) on ice for 30 minutes. Cell debris was removed by centrifugation at 16000 g for 10 minutes. Protein concentration was determined by BCA™ protein assay (Thermo Scientific, USA). A 50 μg protein of each sample was loaded on 8% SDS-PAGE, and the protein transferred to a PVDF membrane by the iBlot™ dry blotting system (Invitrogen, USA). The membranes were blocked by 5% nonfat dry milk for one hour and incubated with either anti-P-gp (Abcam ab-3364) or anti-ABCG2 antibodies (Abcam ab-3380) at 4°C overnight. They were then washed with TBS-tween buffer for one hour and incubated with appropriate HRP-conjugated secondary antibodies (Invitrogen Corp) diluted in blocking buffer for one hour at room temperature. After washing, western blotting luminol reagent (Santa Cruz Biotechnology, USA) was added to the membranes and the chemiluminescence recorded using a Fuji LAS-3000 system. The membranes were then treated with antibody stripping buffer (Gene Bio-application Ltd. Israel), and incubated with anti-actin antibody (1:4000 dilution, Sigma, USA) as control.

Mice were maintained under specific pathogen-free conditions in the animal facility of the Institut Universitaire d’Hématologie, Saint Louis Hospital in Paris. All experimental procedures were performed in accordance with the recommendations of the European Community (86/609/EEC) and the French National Committee (87/848) for the care and use of laboratory animals. Female athymic nude mice Nu/Nu Swiss (9 weeks of age) (Iffa-credo, France), weighing 18–22 g, were housed under controlled environmental conditions (approximately 25°C) with commercial food and water freely available. Primary results showed that the maximal tolerated dose of Dox by athymic mice for a 6 week period was 6 mg/kg/week. Dox was prepared in 0.9% sodium chloride and ip injections given twice weekly. The experimental procedure consisted of a pretreatment of the mice for 15 days with sodium chloride as a control or 6 mg/kg/week Dox. MDA-MB-435 cells (4×10⁶ cells/200 μl PBS) were then injected subcutaneously into their dorsal midline. Tumor growth was determined 25 days after cell injection and sizes monitored by measuring two diameters with a dial-caliper. Tumor volume was calculated as TV = length × (width)² × π/6.

At the end of the experiments, the mice were sacrificed and the percentage of endothelial cells expressing P-gp on the liver, kidneys, heart, and tumor measured by flow cytometry. Tissues were cut into approximately 1×1-mm² squares and rinsed in physiologic serum. The pieces were incubated with 2 mg/ml collagenase at 37°C for 20 minutes with frequent agitation. The cell suspension obtained following extensive trituration with a 5 ml pipette was filtered on a 70 μm nylon cell strainer followed by a second 40 μm filtration. The second filtrates were centrifuged at 1200 rpm for 5 minutes and the pellets washed twice in 1 ml PBS containing 0.5% BSA. Endothelial cells were isolated by immunoabsorption on magnetic beads coated with anti-mouse CD31 and CD105 IgG according to the recommended protocol (Myltenyi Biotec, France). The isolated cells were characterized by flow cytometry using anti-mouse vWF IgG or C219 antibody. Labeling was revealed by second incubation with fluorescein-conjugated goat anti-mouse IgG.

Immunohistochemical studies were carried out on 5 μm paraffin sections before and after treatment. Primary antibody against P-gp C219 antibody was used at 1:50 dilution. All the immunostainings were performed in an automated immunostainer (Ventana Medical System, France). The intensity and percentage of the cytoplasmic staining on tumor sections were noted.

Our experiments showed that HMEC-1 cells are initially sensitive to Dox treatment. In our attempt to study the induction of drug resistance in endothelial cells, we added progressively increasing doses of Dox into the culture media of the HMEC-1 cells during a period of approximately 12 weeks. When the cells had gradually adapted to the presence of higher concentrations of Dox, two conditions were then chosen to stabilize the Dox-resistant endothelial cell: one population was maintained in a culture with 0.08 μg/ml Dox (HMECd1), and another with 0.24 μg/ml Dox (HMECd2). As shown in Table 1, MTS assay indicated a 15- and 24-fold increase in drug-resistance in the stabilized subcell lines HMECd1 and HMECd2, as compared to their parental cells. ³H-thymidine incorporation assay indicated a 36- and 178-fold increase in the RI of HMECd1 and HMECd2 cells in comparison to the parental HMEC cell line (Table 2). Their cellular characteristics were close to those of the parental cells as shown by comparable morphologies and equivalent expression levels of von Willebrand factor, CD31, CD105, flt1, and E-cadherin (data not shown). When we assessed the stability of the Dox-resistant phenotype by culturing HMECd2 in the absence of drugs, we found that after 2 weeks in a drug-free medium, there was no significant change in the drug resistance phenotype or resistance index. However, when grown without selection pressure for 4 weeks, the RI to Dox decreased from 178.5 to 1.25 (p < 0.001). Therefore, endothelial cells were able to induce or reverse the expression of P-gp.

The resistance of these cells to other drugs was then tested. The use of three MDR-related drugs, Daunorubicin, Vinblastine, or Etoposide, showed that both of the Dox-resistant endothelial cell lines were also resistant to higher concentrations of these drugs compared to parental cells (Table 2). In contrast, no significant differences between parental and resistant sublines were found with Mytomycin C treatment (Table 2).

Flow cytometric studies demonstrated a high level of P-gp expression on the cell surface of Dox-treated cells, whereas it was almost absent on parental cells (Figure 1a and b). P-gp surface expression was dependent on the Dox concentration used for cell establishment; it reached 9.2 ± 2.9 MFI for HMECd1 cells (p < 0.05) and 45.1 ± 8.4 MFI for HMECd2 cells (p < 0.005) compared with 2.8 ± 0.8 MFI for parental cells. This P-gp expression represented a 3.2- and 16-fold increase in comparison with parental cells. Interestingly, when treated with 0.16 μg/ml Dox for 15 days, a primary culture of endothelial cells isolated from the human umbilical vein also expressed a P-gp protein on their surface (Figure 1b). In contrast, we did not find any expression of MRP1 in both Dox-resistant HMEC and HUVEC (data not shown)

Western blot analysis of the levels of P-gp showed that its expression in drug-resistant HMECd1 and HMECd2 cells increased about 4- and 6- fold, respectively (Figure 1c). Furthermore, we also determined the changes of P-gp mRNA levels using qPCR. The results showed an increase in P-gp mRNA by approximately 3.4 and 7.2 folds in HMECd1 and HMECd2 cells, respectively regardless of the presence of the P-gp or ABCG2 inhibitors (Figure 1e).

Levels of ABCG2 expression on drug-resistant HMECd1 and HMECd2 cells were also evaluated using qPCR and western blot. Our results showed a 1.41- and 1.68-fold increase in ABCG2 mRNA in HMECd1 and HMECd2 cells, regardless of the presence of the ABCG2 or P-gp inhibitors (Figure 1f). The ABCG2 protein also increased about 1.5 and 2 fold, respectively (Figure 1d). Thus, our results indicate that Dox induced predominantly P-gp expression.

Transporter functionality was tested by evaluating the ability of these cells to efflux a fluorescent Rho probe. Kinetic analyses by flow cytometry showed that parental cells incorporated the fluorescent probe in a time-dependent manner, reaching a plateau of 41.2 ± 7.9 MFI at 80 minutes (result not shown). In contrast, both Dox-resistant cell lines demonstrated a significant decrease in Rho accumulation (indicative of an enhanced efflux), reaching 13.1 ± 3.9 MFI for HMECd1 and 6.9 ± 1.3 MFI for HMECd2 at 80 minutes (p < 0.025). This indicated a 68% and 83% reduction in intracellular Rho accumulation (Figure 2a). Similar experiments with Dox-treated and untreated HUVECs showed that only the former could significantly and specifically efflux Rho (p < 0.05) (Figure 2b). When incubating both Dox-resistant HMEC cells in the presence of 5 μM Rho for one hour at +4°C, to block the energy-dependent function of P-gp, the Rho uptake reached ≈ 34.5 MFI, a comparable value to that of 38.4 ± 3.3 MFI obtained for parental cells. By analyzing data obtained during the establishment of Dox resistance, we demonstrated a linear correlation between P-gp transporter expression and its Rho efflux function as confirmed by a correlation factor R² of 0.9301 (Figure 1g), indicating P-gp plays a major role in drug efflux in these cells.

We tested the effects of two functional inhibitors of P-gp, Verapamil and the MoAb MRK16, on Rho accumulation (Figure 2a-d). The presence of Verapamil did not significantly modify the Rho accumulation in parental HMEC cells (Figure 2c, d). In contrast, it effectively blocked efflux of the fluorescent dye in both Dox-resistant cell lines, raising them significantly to an intracellular Rho level comparable to that of parental HMECs (Figure 2c, d). The very low Rho accumulation (2 μM) in HMECd1 and HMECd2 cells increased to 96.1 ± 4.9 MFI (p < 0.01) and 73.45 ± 2.5MFI (p < 0.025) respectively when 30 μM Verapamil was added. Varying the concentration of Verapamil from 1 to 100 μM resulted in a progressive increase of intracellular Rho accumulation, indicating its specific effect. This reached a plateau at 30 μM (data not shown). The presence of the specific P-gp inhibitory MoAb, MRK16, reproduced the effect of Verapamil and restored a level of Rho accumulation in both HMECd1 and HMECd2 similar to that of parental cells (Figure 2d). In contrast, QCRL-1, a MoAb directed against MRP1, had no effect on Rho accumulation (data not shown). Taken together, these results indicate that the loss of Rho accumulation in Dox-resistant endothelial cells involves the P-gp function which has the property to mediate cell exclusion of drugs.

We then checked cell survival after Dox treatment in the presence of Cyclosporine A and Verapamil in both HMECd1 and HMECd2 cells. The cells were treated with a series of Dox concentrations in the presence of 2.5 μM Cyclosporine A or 1 μM Verapamil (that blocks the P-gp function). The results clearly show that the blockage of the P-gp function restored the sensitivity of HMECd1 and HMECd2 cells to Dox (Table 1). In contrast, the ABCG2 inhibitors Fumitremorgin C and Diethylstibesterol had no such effect (data not show). Therefore, our results suggest that P-gp plays a major role in the acquisition of Dox resistance in HMECd1 and HMECd2.

To evaluate the role of endothelial P-gp in tumor protection, we also tested its influence on tumor growth in vivo. Two groups of athymic nude mice were pretreated with intraperitoneal (ip) Dox injection of 6 mg/kg/week over a 15-day period (groups II and IV) whereas groups I and III were injected with physiologic serum. Groups V and VI had no pretreatment. The Dox prescription corresponded to the maximum well-tolerated dose of Dox and resulted in barely 4-8% body weight loss during the experiment and no deaths. MDA-MB-435 cells were then subcutaneously inoculated in the dorsal midline (groups III to VI). The posttreatment began, corresponding to the physiologic serum (groups I, III, V) or Dox (II, IV, VI) injections. The data in Figure 3a and 3b show the tumor growth evolution for the different treatments. When injected after tumor implantation, Dox effectively inhibited tumor growth, reaching 16.4 ± 13.9 mm³ at 25 days, a 3.7-fold decrease in tumor size compared to mice receiving physiologic serum (60.8 ± 13.5 mm³, p < 0.025) (Figure 3b). In contrast, when the mice had been sensitized by Dox injection for 15 days before tumor implantation, the tumors responded poorly to Dox posttreatment and we found a comparable tumor growth between mice receiving Dox (55.9 ± 16.1 mm³) and physiologic serum (53.2 ± 10.3 mm³) throughout the experiment (Figure 3a). To better understand the ineffectiveness of Dox pretreatment on the blockage of tumor growth, we sacrificed animals and performed histological and flow cytometric studies after cell dissociation of liver, kidneys, heart, and tumor for ten mice in each group. This showed that the Dox treatment given either post- or pre- and posttreatment did not significantly modify the morphology of these organs. In particular, no sign of cardiotoxicity was observed across the different groups.

To better quantify the P-gp positive endothelial cells in the mice, we removed their organs and tumors. Following cell dissociation, the isolated endothelial cells were characterized by flow cytometric analysis and the percentage of endothelial cells positive for P-gp labeling (C219) was measured (Figure 3c). Our results show that for the liver and kidney, 10-40% of endothelial cells became positive for P-gp expression following Dox pre- and/or posttreatment (groups II, IV, and VI). Endothelial cells within the tumor acquired the resistant phenotype when the animals had been treated with Dox (group VI, 44.74 ± 3.55%) in comparison with buffer administration (groups III and V, ~1%). When Dox was administered as a pretreatment in group IV, the percentage of P-gp positive endothelial cells within the tumor reached the remarkable level of 78.01 ± 6.39%. Immunohistological observation of these tumor sections at the end of the experiments demonstrated an induced P-gp staining on the endothelial cells, and no evident induced P-gp staining in the surrounding tumor cells (Figure 3d). These data suggest that endothelial cells participate in the resistant phenotype of tumors by serving as an initial barrier between chemotherapeutics and tumor cells.