Similarities and differences in the localization, trafficking, and function of P-glycoprotein in MDR1-EGFP-transduced rat versus human brain capillary endothelial cell lines

In this study two immortalized brain capillary endothelial cell lines were used. Rat brain endothelial (RBE4) cells were kindly provided by Prof. Francoise Roux (INSERM U26, Hôpital Fernand Widal, Paris, France). The RBE4 cell line has been obtained after transfection of a primary rat brain endothelial cell culture with the plasmid pE1A-adenovirus encoding gene [27]. The human cerebral microvascular endothelial cell line hCMEC/D3 was kindly provided by Dr. Pierre-Olivier Couraud (Institute COCHIN, Paris, France). The hTERT/SV40-immortalized hCMEC/D3 clonal cell line was derived from human temporal lobe microvessels isolated from tissue resected during epilepsy surgery [28].

Doxycycline-inducible MDR1-EGFP-transduced variants (RBE4-MDR1-EGFP, hCMEC/D3-MDR1-EGFP) were derived from WT cells by lentiviral transduction as described previously for the hCMEC/D3 cell line, using an MDR1-linker-EGFP vector (with EGFP located at the C-terminus of human MDR1) kindly provided by Prof. Piet Borst (The Netherlands Cancer Institute, Amsterdam, The Netherlands) as template [31]. The linker sequence in the vector fusing EGFP and MDR1 consists of 36 bp (TCGACGGTACCGCGGGCCCGGGATCCATCGCCACC). hCMEC/D3-WT, RBE4-WT, and hCMEC/D3-MDR1-EGFP cell lines were cultivated as previously described [13, 36]. The same culture conditions were used for the newly prepared RBE4-MDR1-EGFP cell line. As described previously [13, 36], hydrocortisone (1.4 µM) was added to the culture medium to enhance the expression and functionality of Pgp and transendothelial resistance [23, 36]. All endothelial cell lines were grown on rat-tail collagen I (60 µg/mL, Gibco) coated 100 mm cultures dishes or permeable membrane supports at 37 °C and 5% CO_2. Pgp-EGFP protein expression in MDR1-EGFP transduced cells was induced by cell cultivation in medium containing 1 µg/µL doxycycline. Expression of Pgp-EGFP was regularly verified by observing the cells with an inverted fluorescence microscope. Doxycycline supplementation of the medium was maintained during the whole culturing period of Pgp-EGFP expressing cells. For experiments, hCMEC/D3 and RBE4 cells were seeded in a density of 50,000 cells/cm². Immortalized MDR1 transfected porcine kidney LLC-PK1 (LLC-MDR1) cells, which were used as a reference standard for vectorial transport experiments, were kindly provided by Prof. Piet Borst. We have used these cells previously for studying vectorial transport of Pgp substrates in the concentration equilibrium transport assay (CETA), which was also used in the present study (see below). Cells were cultivated in medium 199 (Gibco) supplemented with 10% fetal calf serum (FCS, Linaris), and penicillin (100 U/mL)/streptomycin (100 µg/mL) (Merck) and were seeded at a density of 300,000 cells/cm² for experiments. All experiments were carried out between passage 13–35 at 5-days post-confluency (equal to 7 days after seeding).

The RBE4 and hCMEC/D3 cell lines used here are immortalized, which has the advantage that these cells are stable for several passages and may yield a large number of endothelial cells with the same genetic and phenotypical characteristics. The immortalization status, however, causes several alterations as compared to the native original cell type, including low junctional tightness of cell monolayers and reduced expression of efflux transporters such as Pgp [23]. Therefore, for comparison with the MDR1-EGFP transduced rat RBE4 cells, experiments were also performed with primary cultured rat brain capillary endothelial cells (rBCECs). As described by us previously [36], rBCECs were prepared following a protocol by Régina et al. [37] and Perrière et al. [38] from 2 to 3 weeks old Wistar rats (obtained from Charles River; Sulzfeld, Germany). All animals were treated according to protocols evaluated and approved by the ethical committee of our university. In short, meninges were removed from cerebral cortices in phosphate-buffered saline (PBS) with 1% penicillin (100 U/mL)/streptomycin (100 μg/mL) (Biochrom AG) on ice. Subsequently, cortices were mechanically dissociated with scalpels. The homogenate was suspended in enzyme solution: DMEM/F-12 (Gibco®/Life Technologies) supplemented with 1 mg/mL dispase II (Roche), 0.1 mg/mL DNAse I (Roche), 270 U/mL collagenase II (Biochrom AG), and 1% penicillin (100 U/mL)/streptomycin (100 μg/mL) (Biochrom AG) and was incubated for 1.5 h at 37 °C with gentle agitation. The homogenate was centrifuged in 20% bovine serum albumin (Linaris GmbH) solution (1000g, 15 min, 4 °C) and the obtained cell pellet was incubated again in the enzyme solution for 1 h at 37 °C. The resulting homogenate was filtered and the retained capillary fragments were removed from the filter with EBM-2 (Lonza) medium supplemented with 20% FCS, 1% penicillin (100 U/mL)/streptomycin (100 μg/mL) (Biochrom), 1 mM HEPES (Gibco®/Life Technologies), 1% l-glutamine (200 mM, Sigma-Aldrich), and 0.5 μg/mL hydrocortisone (Sigma-Aldrich). The cells were seeded on collagen type IV (Sigma-Aldrich) coated 28 cm² plates and cultivated for 72 h in presence of 4 μg/mL puromycin (Sigma-Aldrich) for removal of pericytes [38] in humidified 5% CO₂/95% air at 37 °C. Thereafter, puromycin was removed and replaced by 2 ng/mL basic FGF (Gibco®/Life Technologies). Cells were cultured until confluency and then seeded for western blot, determination of TEER, Rho123 uptake assay, and vectorial drug transport assay essentially as described in the following for RBE4 and hCMEC/D3 cells.

Capillaries were isolated from the gray matter of four rat brains, according to protocols kindly provided by Drs. Elena Puris (Ruprecht-Karls University, Heidelberg, Germany) and Björn Bauer (University of Kentucky, Lexington, KY, USA). Wistar rats (3 months old) were anesthetized by CO₂ inhalation and subsequently decapitated. Brains were removed immediately and placed on ice. After removing meninges, choroid plexus, and large superficial blood vessels by rolling each brain on Whatman® blotting paper, cortices were separated from the cerebellum, brain stem, and white matter. Pieces of cortical gray matter were homogenized in 5 volumes (w/v) of buffer A (101 mM NaCl, 4.6 mM KCl, 5 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 15 mM HEPES, 5 mM D-glucose, 1 mM sodium pyruvate, pH 7.4) using a dounce tissue homogenizer. After centrifugation (2000g, 10 min, 4 °C) the pellet was resuspended thoroughly in buffer B (buffer A with additional 16% dextran, Sigma-Aldrich) and capillaries were separated from the remaining brain parenchyma by centrifugation in a swinging bucket rotor at 4500g for 20 min at 4 °C without brake. The pellet was suspended in 10 mL buffer C (buffer A supplemented with 0.5% BSA) and the resulting suspension was passed sequentially through cell strainers with mesh sizes of 200 µm, 100 µm, and 30 µm. The fraction retained on the 30 µm cell strainer was collected in 40 mL buffer C and centrifuged at 1000g for 5 min at 4 °C to pellet capillaries, which were then washed twice with PBS and analyzed by western blotting.

Pgp expression in immortalized cell lines, primary cells, and brain capillaries was evaluated by western blotting. Cells were scraped in radioimmunoprecipitation assay (RIPA) buffer (20 mM Tris, 50 mM NaCl, 0.5% (w/v) sodium deoxycholate, 0.5% (v/v) Triton X-100) supplemented with a protease inhibitor cocktail (Roche) and mechanically lysed by passage through a 21 G syringe needle. Accordingly, brain capillaries were lysed in RIPA buffer. Cell debris was removed by centrifugation and protein concentrations were determined using a BCA protein assay kit (Thermo Fisher Scientific). Equal amounts of protein were loaded onto 10% handcasted SDS-polyacrylamide gels or 4–20% Mini-PROTEAN gels (Bio-Rad) and the separated proteins were then transferred to a PVDF-membrane (Roth). After blocking in 5% milk (Roth) membranes were incubated with anti-Pgp (1:100, Enzo, #ALX-801-002-C100) or anti-β-actin (1:4000, Sigma-Aldrich, #A2066) antibodies for 1 h at room temperature. Secondary antibody incubation was performed for 45 min at room temperature with HRP-conjugated anti-rabbit (1:1000, Dako, #P0448) or anti-mouse (1:1000, Dako, #P0260) antibodies. Protein bands were visualized with the SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific) and detected utilizing the ChemiDoc MP Imager (Bio-Rad). Densitometric protein band quantification was conducted with the Image Lab software version 6.1 (Bio-Rad). Expression of Pgp is reported normalized to β-actin.

The barrier integrity of hCMEC/D3 and RBE4 cells was monitored using the non-invasive TEER measurement that quantifies the electrical resistance of a cell monolayer. Cells were grown on rat-tail collagen I (60 µg/mL, Gibco) coated permeable filter inserts [ThinCert®; Greiner Bio-One, 12-well, 0.4 µm pore size, polyethylene terephthalate (PET)] and subjected to TEER measurements daily starting on day one post-seeding in at least three replicates. Manual TEER measurements were conducted with the EVOM Volt-Ohm resistance meter (World Precision Instruments) equipped with an EndOhm-12 chamber. TEER values were reported as the measured electrical resistance multiplied by the growth area of the filter inserts and were calculated by subtracting the measured TEER of a coated filter insert without cells from the TEER of a coated filter insert with cells. For comparison of TEER values measured in the ThinCert® system, TEERs were also measured in the Transwell® system from Corning (Corning Costar Corporation, Cambridge, MA) and tissue culture (TC−) inserts from Sarstedt (Nümbrecht, Germany). Furthermore, 6-well and 12-well multiwell plates and different membrane materials were compared (Additional file 1).

The functional tightness of hCMEC/D3-WT, hCMEC/D3-MDR1-EGFP, RBE4-WT, and RBE4-MDR1-EGFP cell monolayers grown on filter inserts was evaluated by their paracellular permeability to radiolabeled D-mannitol (MW = 182.2 Da, Hartmann Analytic). Seven days post-seeding, the medium in the basolateral donor chamber of the 12-well ThinCert® system was replaced by Opti-MEM (12-well: 1.5 mL, 6-well: 2.7 mL) containing d-[¹⁴C]mannitol (specific activity 2.035 GBq/mmol) in a concentration of 1.85 kBq/mL. The upper compartment was filled with mannitol-free assay buffer (12-well: 0.7 mL, 6-well: 2 mL) to measure transport in the basolateral-to-apical direction (b-A). Incubation was performed under cell culture conditions (37 °C, 5% CO₂, humidified atmosphere) on a horizontal shaker (55 rpm). After 30, 60, 120, and 180 min, samples (12-well: 50 µL, 6-well: 130 µL) were collected from the apical chamber and the amount of transported d-mannitol was quantified by liquid scintillation counting using the 1450 Microbeta Trilux liquid scintillation counter (Perkin Elmer Wallac). Basolateral volume was adjusted after each sampling to compensate for changes in the hydrostatic pressure. TEER measurements before and after the assay were carried out to confirm barrier integrity. The apparent permeability coefficients (P_app) in nm per second were calculated according to Artursson [39]. Each experiment was performed in at least three replicates. As was described for the TEER measurements above, mannitol fluxes were compared in the ThinCert®, Transwell®, and Sarstedt-TC-insert systems, using 6-well and 12-well multiwell plates and different membrane materials. Based on this comparison, all subsequent experiments were performed with ThinCert® cell culture inserts [12-well, 0.4 µm pore size, polyethylene terephthalate (PET)] from Greiner Bio-One.

The functionality of Pgp in RBE4-WT, RBE4-MDR1-EGFP, hCMEC/D3-WT, hCMEC/D3-MDR1-EGFP cells, and primary rBCECs was analyzed by measuring the uptake of the fluorescent Pgp substrate Rho123 (Sigma-Aldrich). Cells were grown on 6-well plates and cultured for seven days in the presence or absence of doxycycline (1 µg/mL, Biochrom). On the day of the assay, cells were incubated with the Pgp inhibitor tariquidar (0.5 µM, dissolved in DMSO) or the dissolvent in Opti-MEM for one hour under cell culture conditions on a horizontal shaker (55 rpm). Afterward, Rho123 (5 µM, dissolved in ethanol) was added and cells were incubated for another two hours. Rho123 transport was terminated by washing the cells twice with ice-cold PBS. Intracellular Rho123 concentration was determined by cell lysis in standard lysis buffer (25 mM Tris, 50 mM NaCl, 0.5% (w/v) sodium deoxycholate, 0.5% (w/v) Triton X-100) and measurement of fluorescence intensity with the FLUOstar OPTIMA (ex.: 485 nm, em.: 520 nm; BMG Labtech). Protein concentrations were determined using a BCA assay kit (Thermo Fisher Scientific). Rho123 uptake was obtained for each condition in triplicates and calculated as absolute fluorescence per mg protein. Additionally, Pgp functionality was expressed as multidrug resistance activity factor (MAF) as described by Huber et al. [40]. This factor describes Pgp efflux activity based on differences in intracellular Rho123 accumulation in cells after Pgp inhibition compared to cells with active Pgp-mediated efflux. Thus, higher MAF values correlate with increased Pgp activity. For better comparison, differences in Rho123 accumulation are normalized to intracellular Rho123 levels in inhibitor-treated cells. MAF values for Pgp were calculated according to Huber et al. 2012 utilizing the following equation: MAF (%) = 100 ·  (MFI_TQ − MFI₀)/(MFI_TQ) where MFI_TQ and MFI₀ represent the mean fluorescence intensity of intracellular Rho123 in tariquidar (TQ)-treated and non-treated cells, respectively.

For these experiments, a concentration equilibrium transport assay (CETA) was used, which is more sensitive to identify Pgp substrates than conventional bidirectional (concentration gradient) transport assays [41]. In the CETA, the drug is added to both (apical and basolateral) sides of the monolayer, so that the initial drug concentration is the same in both compartments, thereby minimizing the effect of passive diffusion across the cell monolayer. Particularly for lipophilic compounds, passive transcellular diffusion could form a bias in Transwell assays by concealing active transport, which we have previously demonstrated by comparing vectorial drug transport of various small lipophilic drugs in conventional bidirectional assays versus drug transport in CETA [41].

In the present study, the CETA assay was used to determine the flux of the radiolabeled Pgp substrate [N-methyl-³H]N-desmethyl loperamide ([³H]dLop) across hCMEC/D3-MDR1-EGFP and RBE4-MDR1-EGFP monolayers. Although Kannan et al. [42] have shown that [³H]dLop is trapped in lysosomes, this does not restrict its use for studying Pgp-mediated vectorial drug transport in BCECs, as demonstrated recently by us in primary cultured porcine BCECs [43]. Cells were grown on permeable ThinCert® filter supports (12-well, 0.4 µm pore size, PET, Greiner Bio-One). LLC-MDR1 monolayers were used as a reference standard for the demonstration of vectorial transport of the Pgp substrate ([³H]dLop) used in these experiments. Transport studies were performed in triplicates seven days after seeding. Before substrate exposure, the culture medium was replaced by Opti-MEM (Gibco) with or without the Pgp inhibitor verapamil (20 µM) for 1 h in the apical and basolateral chambers. Verapamil was preferred to tariquidar (which was used in the Rho123 accumulation experiments) because tariquidar has been shown to compete for lysosomal trapping of [³H]dLop [42], which may form a bias for data on vectorial drug transport at the cell membrane. To initiate transport studies, [³H]dLop was added to the upper and lower compartment in a concentration of 5 nM in the absence or presence of verapamil (20 µM). After 30, 60, 120, 180, and 240 min, samples of 50 µL and 70 µL were retrieved from the apical and basolateral chambers, respectively. Quantification of substrate transport was conducted by liquid scintillation counting using a 1450 Microbeta Trilux liquid scintillation counter (Perkin Elmer Wallac). All incubation steps were performed under cell culture conditions on a horizontal shaker (55 rpm) with 0.7 mL medium in the upper and 1.5 mL medium in the lower compartment. Barrier tightness was confirmed via TEER measurements before and after the assay. Baseline dLop levels were corrected by the amount of dLop absorbed by the filter system as measured in blank inserts without cells.

Additionally, vectorial drug transport across RBE4-MDR1-EGFP, hCMEC/D3-MDR1-EGFP, LLC-MDR1, and primary rBCEC monolayers was studied using 2 µM Rho123 as a fluorescent Pgp substrate essentially following the same protocol as described for [³H]dLop. Transport was measured using a FLUOstar OPTIMA fluorescence reader (ex.: 485 nm, em.: 520 nm; BMG Labtech), analyzing samples from basolateral and apical compartments collected at the various time points.

Cell morphology of hCMEC/D3, RBE4, rBCECs, and LLC cells was assessed via phase-contrast microscopy. Therefore, cells were seeded onto 6-well cell culture dishes and analyzed after confluence with an inverted fluorescence microscope (Olympus IX-70, Hamburg, Germany) and a 10× objective.

Intercellular Pgp transfer was studied by coculturing RBE4-WT and –MDR1-EGFP cells to equal amounts (50:50) on glass coverslips until confluency. Wildtype cells were prelabeled with either Cell Proliferation Dye eFluor 670 (5 µM, 10 min, 37 °C, eBioscience) or CellTracker red CMTPX (5 µM, 25 min, 37 °C, Thermo Fisher Scientific) for identification in the monolayer after coculturing. Since Pgp-EGFP is exclusively expressed by MDR1-EGFP transduced cells and not by WT cells (which exhibit only low intrinsic Pgp levels), Pgp transfer between transduced and WT cells can be easily traced by the green fluorescent signal of the EGFP-tagged Pgp. Cell nuclei were stained with bisbenzimide H (5 mM, 5 min, 37 °C, Sigma-Aldrich). Images were acquired by confocal fluorescence microscopy and live-cell imaging.

Intracellular distribution of the autofluorescent Pgp substrate doxorubicin was investigated in cocultures of RBE4-WT and -MDR1-EGFP cells seeded on glass coverslips until confluency. Coculturing of WT and MDR1-EGFP cells allowed to compare the differential intracellular distribution of the autofluorescent Pgp substrate in WT cells with low intrinsic expression of Pgp and MDR1-EGFP transduced cells with a higher Pgp expression. Cells were incubated with LysoTracker (75 nM, 1 h, 37 °C, Thermo Fisher Scientific) and doxorubicin (10 µM, 30 min, 37 °C, Enzo) before live-cell imaging and confocal microscopy. For imaging, glass coverslips were mounted into a PeCon open chamber (PeCon, Erbach, Germany), covered with serum-free Opti-MEM, and kept at 37 °C. Endolysosomal staining in MDR1-EGFP transduced RBE4 and hCMEC/D3 cell cultures was performed similarly, however, without doxorubicin incubation.

Confocal fluorescence microscopy was conducted with a Leica TCS SP5 II fluorescence microscope (Leica Microsystems, Bensheim, Germany) combined with a 63 × 1.2 water immersion objective. For excitation, wavelengths of 405 nm (bisbenzimide H, LysoTracker), 561 nm (doxorubicin, CMTPX), 633 nm (eFluor 670), and 488 nm (Pgp-EGFP) were used.

Following fixation of the cells with paraformaldehyde or ice-cold methanol (10 min, − 20 °C) indirect Pgp staining was performed as described above using a primary monoclonal antibody against Pgp (1:100, Sigma-Aldrich, #P7965) and a secondary Alexa Fluor 568 antibody (1:500, Thermo Fisher Scientific, #A11004).

Data are presented as means ± standard error of the mean (SEM) of at least three biological replicates. An unpaired t-test was applied to assess the significance of intergroup differences. Significant differences between multiple groups within one experiment were either calculated by one-way or two-way analysis of variance (ANOVA) followed by Dunnett’s or Bonferroni post hoc tests. P values < 0.05 were considered statistically significant. Statistical analyses were conducted using the PRISM 8 software (GraphPad Software Inc.).

As shown in Additional file 2 (A, B), both hCMEC/D3-WT cells and RBE4-WT cells formed monolayers and exhibited elongated spindle-shaped morphology when examined at confluence by phase-contrast microscopy, substantiating previous reports [13, 36]. The transduction with MDR1-EGFP did not alter the morphology of the cells (Additional file 2 (C, D)). As shown in Additional file 3A, primary cultured rat BCECs (rBCECs) exhibited similar morphology to that observed for the immortalized BCEC lines. Additional file 3B illustrates the morphology of the LLC-MDR1 kidney epithelial cell monolayers used for comparison with BCECs here (see below).

However, differences were observed in the localization of Pgp-EGFP in the two immortalized BCEC lines when using confocal laser scanning fluorescence microscopy and live-cell imaging (Additional file 4). In RBE4 cells, Pgp-EGFP was predominantly expressed at the cell surface (Additional file 4A), consistent with its function as an efflux transporter, whereas, as reported previously [13, 31], in hCMEC/D3 cells, the Pgp-EGFP fusion protein was visible both at the plasma membrane and in intracellular vesicles, most likely presenting endolysosomal compartments (Additional file 4B). To demonstrate that the Pgp-EGFP-positive intracellular vesicles are of endolysosomal origin, we co-stained MDR1-EGFP-transduced RBE4 and hCMEC/D3 cells with a endolysosomal marker (LysoTracker). As shown in Fig. 1B, endolysosomal vesicles were positive for Pgp-EGFP fluorescence signal in hCMEC/D3 cells, whereas intracellular Pgp-EGFP localization was hardly observed in RBE4 cells (Fig. 1A). These results further substantiated findings from our previous studies with endolysosomal markers in hCMEC/D3-MDR1-EGFP cell cultures that the Pgp-EGFP-positive intracellular vesicles are of endolysosomal origin [13, 31]. The marked difference in numbers of Pgp-EGFP-positive intracellular vesicles in RBE4 versus hCMEC/D3 cells was confirmed by counting these vesicles in both cell lines (Additional file 5).

To prove that in cells expressing the Pgp-EGFP fusion protein, EGFP is not simply cleaved and being sequestered in intracellular vesicles, we evaluated whether an antibody to Pgp colocalizes to the EGFP signal. As shown in Additional file 6, indirect staining of Pgp in MDR1-EGFP transduced RBE4 and hCMEC/D3 cells colocalized with the Pgp-EGFP fluorescence signal in intracellular vesicles (in hCMEC/D3) and in the cell membrane of both RBE4 and hCMEC/D3 cells.

Furthermore, we also stained Pgp in RBE4-WT cells, showing that Pgp was exclusively expressed at the cell surface in these cells (Additional file 7). In contrast, Pgp protein expression in hCMEC/D3-WT cells has been described both in the plasma membrane and in intracellular vesicles [44, 45], substantiating the present findings in MDR1-EGFP-transduced hCMEC/D3 cells (Fig. 1B). This precludes the possibility that intracellular localization of Pgp observed in MDR1-EGFP-transduced hCMEC/D3 cells was a result of the EGFP tag. If that were the case, intracellular Pgp localization should have also been observed in the MDR1-EGFP-transduced RBE4 cells, which it was not. Moreover, the demonstration of vesicular localization of Pgp in different studies on hCMEC/D3 cells [13, 31, 32, 44, 45] argues against the possibility that the differences between hCMEC/D3 and RBE4 cells reported here were simply due to transfected or immortalized clone differences.

As shown in Fig. 2, RBE4-WT and hCMEC/D3-WT cells exhibited approximately the same low expression of Pgp, which was not affected by doxycycline. A similar low Pgp expression was observed in MDR1-EGFP transduced cells in the absence of doxycycline. In the presence of doxycycline, Pgp expression increased seven to ninefold in MDR1-EGFP transduced cells when compared to WT controls (Fig. 2B, D). The Pgp band in the induced cells was too intense to allow separating endogenous Pgp (170 kDa) and Pgp-EGFP (200 kDa). However, when a lower amount of protein was used for western blot analysis, the shift in the Pgp-EGFP fusion bands was visible (Fig. 3A).

For comparison, we determined the Pgp protein expression of primary rBCECs and freshly isolated rat brain capillaries (Fig. 3). The Pgp content of rBCECs was only about 4.5-fold higher than Pgp protein expression in RBE4-WT cells, whereas the Pgp content of freshly isolated rat brain capillaries exceeded the Pgp content of the MDR1-EGFP transduced RBE4 cells (Fig. 3B). The about 4-times higher Pgp content of freshly isolated rat brain capillaries vs. primary cultured rBCECs shown in Fig. 3 is in line with previous experiments by Régina et al. [46], who reported that the amount of Pgp was about 4-times higher in isolated rat brain microvessels than in rBCECs prepared from these microvessels. Importantly, the data shown in Fig. 3 demonstrate that transduction of RBE4 cells with MDR1-EGFP does not lead to a gross overexpression of the transporter but to intermediate Pgp levels between those of rBCECs and freshly isolated rat brain capillaries. Interestingly, the Pgp content of MDR1-transfected LLC cells was similar to that of freshly isolated rat brain capillaries (Fig. 3B).

As would be expected, transduction of RBE4 and hCMEC/D3 cells with MDR1-EGFP did not affect TEER or paracellular mannitol flux of the cells (Fig. 4, Table 1). As shown in Fig. 4A, TEER progressively increased over 7 days after seeding in RBE4 cells, whereas a plateau in TEER values was reached after 4 days in hCMEC/D3 cells. TEER values were comparable in all cell lines, reaching maximal values of ~ 13–16 Ω cm² when measured with an EndOhm chamber. For hCMEC/D3 cells, these low TEER values were comparable to values reported in the literature when using an epithelial Volt-Ohm meter (EVOM) coupled to an EndOhm chamber for TEER measurements [43, 47], whereas TEER measurements with chopstick electrodes result in significantly higher values [21, 43, 48]. When TEER was measured with an EndOhm chamber in primary rBCECs, values were in the range of 80–140 Ω cm² (Fig. 5A), i.e., about 5–10-times higher than the TEER values determined in RBE4 cells. These TEER values of rBCECs were similar to the values (123 ± 1.1 Ω cm²) reported previously for these cells [38].

As a consequence of the low TEER values of RBE4 and hCMEC/D3 cells, paracellular permeability (indicating paracellular leakage through endothelial tight junctions) was relatively high with average mannitol flux values of between 54 and 60 nm/sec (Table 1). Mannitol permeability was comparable in RBE4 and hCMEC/D3 cells and was not affected by transduction with MDR1-EGFP.

As reported previously for hCMEC/D3 cells [47] and primary murine BCECs [49], the type of tissue culture inserts and membrane materials markedly affected TEER and mannitol flux values of hCMEC/D3 and RBE4 cells as shown in Additional file 1. The highest TEER and lowest mannitol flux data were obtained with ThinCert® 12-well multiwell plates, using PET membranes with a pore density of 2 × 10⁶ and pore size of 0.4 µm (Additional file 1), providing the best growth conditions for the cells to form a tight monolayer. This system was therefore chosen for the data shown in Table 1 and Fig. 4, as well as for all subsequent experiments.

As shown in Fig. 6A, C, Pgp was functional in RBE4 and hCMEC/D3-WT cells, since the accumulation of Rho123 in the cells was significantly increased by the Pgp inhibitor tariquidar. Doxycycline did not alter the functionality of Pgp in the WT cells. In MDR1-EGFP transduced RBE4 cells, in the absence of doxycycline, Rho123 accumulation was similar to that determined in WT cells. In the presence of doxycycline, however, Rho123 accumulation was significantly reduced, indicating increased Pgp-mediated efflux, which could be inhibited by tariquidar (Fig. 6A). When comparing Rho123 accumulation in the absence and presence of tariquidar, Rho123 efflux was increased ~ 7-fold. The increased functionality of Pgp in doxycycline-induced RBE4-MDR1-EGFP cells is also demonstrated by calculating the multidrug resistance activity factor (MAF) according to Huber et al. [40] (Fig. 6B). Very similar findings were obtained with hCMEC/D3-MDR1-EGFP cells (Fig. 6C, D).

In primary rBCECs, intracellular Rho123 accumulation was significantly lower (Fig. 5B) than Rho123 accumulation in RBE4-WT cells (56,482 ± 2456 vs. 86,513 ± 2021; P = 0.0001), which is in line with the higher Pgp expression of the rBCECs (Fig. 3). Following Pgp inhibition by tariquidar, Rho123 accumulation was similar in rBCECs and RBE4 cells. MAF values calculated for intracellular Rho123 accumulation in rBCECs were significantly higher than MAF values in RBE4-WT cells (61.2% ± 2.7 vs. 37.5% ± 4.2; P = 0.002), again in line with the higher Pgp expression of the rBCECs.

As shown in Fig. 7A–D, asymmetrical (basolateral to apical) transport of the selective Pgp substrate dLop [50] was not observed in MDR1-EGFP transduced RBE4 and hCMEC/D3 cells. In contrast, such transport was obtained in MDR1 transfected LLC cells (which are widely used as a surrogate Pgp screening model of the BBB [25, 26]), and this transport was almost completely inhibited by verapamil (Fig. 7E, F). TEER of the LLC-MDR1 cells was 132 Ω cm², i.e., considerably higher than the TEER values of the MDR1-EGFP transduced RBE4 and hCMEC/D3 cells (Table 1). Consistent with the high junctional tightness of the LLC monolayers, paracellular mannitol flux was less than 1% of mannitol diffusion per hour (corresponding to a mannitol permeability lower than 12 nm/s) and thus much lower than the mannitol permeability observed in the immortalized BCEC lines (Table 1), which explains that vectorial (Pgp-mediated) drug transport from the basolateral to the apical Transwell chamber was observed in the MDR1-transfected LLC cells but not in the BCEC lines.

In primary rBCECs, we used the Pgp substrate Rho123 to demonstrate vectorial transport. As shown in Fig. 5C, concentrations of Rho123 were significantly higher in the apical vs. basolateral chambers in the CETA assay, indicating asymmetric transport from the basolateral to the apical side. Rho123 was also used for transport (CETA) assays in LLC-MDR1, RBE4-MDR1-EGFP, and hCMEC/D3-MDR1-EGFP cells. Similar to the data with dLop (Fig. 7), significant vectorial transport was observed in LLC-MDR1 cells (Fig. 5D) but not in MDR1-EGFP transduced RBE4 and hCMEC/D3 cells (Fig. 5E, F).

In addition to actively transporting drugs out of the cell, Pgp has been proposed to mediate lysosomal sequestration of chemotherapeutic drugs in cancer cells, thus contributing to drug resistance [10, 11, 51]. We have recently described that Pgp-mediated lysosomal sequestration of Pgp substrates, including doxorubicin, also occurs in hCMEC/D3 cells and primary cultured porcine BCECs [13]. This prompted us to evaluate whether Pgp-mediated lysosomal trapping of doxorubicin also occurs in RBE4 cells, which, as described above, exhibit markedly less Pgp-EGFP in intracellular compartments compared to hCMEC/D3 cells. For this purpose, the intracellular distribution of doxorubicin was studied in cocultures of RBE4-WT cells with low intrinsic levels of Pgp and Pgp-EGFP overexpressing RBE4-MDR1-EGFP cells. Cocultures were exposed to the autofluorescent Pgp substrate doxorubicin after labeling of endolysosomal vesicles with the vital fluorescence dye LysoTracker blue (Fig. 8). Coculturing of WT and MDR1-EGFP cells allowed to follow the differential intracellular distribution of doxorubicin in cells with different Pgp protein expression levels. The localization and fate of Pgp in cocultures were easy to trace by the EGFP-tag of Pgp expressed in the RBE4-MDR1 cells by confocal fluorescence microscopy and live-cell imaging.

Doxorubicin is one of the most widely used clinical anticancer agents and a hydrophobic weak base that intercalates as a primary target into nuclear DNA [52]. Consequently, when RBE4-WT cells (which did not overexpress Pgp) were exposed to doxorubicin, the doxorubicin bound to cell nuclei.

In contrast, this was not seen in Pgp-overexpressing RBE4-MDR1-EGFP cells (#2 in Fig. 8). Instead, in the latter cells doxorubicin was trapped in endolysosomal vesicles and colocalized with Pgp, as visible by colocalization of Pgp-EGFP (green) with LysoTracker (blue) and doxorubicin (red) (Fig. 8, white frames). Thus, lysosomal trapping contributed to protecting the nuclei from the cytotoxic doxorubicin.

In cocultures of hCMEC/D3-WT and hCMEC/D3-MDR1-EGFP cells, we have previously shown that the Pgp-EGFP fusion protein was transferred from donor to recipient (WT) cells by cell-to-cell contact and Pgp-EGFP enriched vesicles, which were exocytosed by donor cells and endocytosed by adherent recipient cells [32]. In these previous experiments, such intercellular trafficking was not only studied by confocal laser scanning microscopy (as done here) but also by flow cytometry with the Pgp substrate eFLUXX-ID Gold, by which we demonstrated that the transferred Pgp is functional in the recipient cells [32]. In the present study, we examined whether a cell-to-cell transfer of Pgp-EGFP also occurs in RBE4 cells. For this purpose, RBE4-WT cells were labeled with the vital fluorescence dye eFluor 670 (Fig. 9A) or CMTPX red (Fig. 9B) before coculturing with RBE4-MDR1-EGFP cells. The pre-labeling of WT cells before setting up the coculture allows differentiating WT and MDR1-EGFP-transduced cells in the monolayer. In the absence of cytotoxic compounds, Pgp was mainly localized at the plasma membrane of RBE4-MDR1-EGFP cells (#2 in Fig. 9A, B). The fate of Pgp-EGFP, which is exclusively expressed by MDR1-EGFP and not by WT cells, in the coculture could be followed by the green EGFP tag. Thus, Pgp-EGFP fluorescence signal in WT cells (#3 in Fig. 9A, B) visualized by live-cell imaging 2 days after coculturing, indicates intercellular Pgp protein transfer from Pgp-EGFP overexpressing RBE4-MDR1-EGFP cells (#2 in Fig. 9A, B) to RBE4-WT cells (#1 in Fig. 9A, B).

As further proof that Pgp is functional in the MDR1-EGFP transduced cells, cells were exposed to the Pgp inhibitor tariquidar (0.5 µM), followed by the addition of doxorubicin (10 µM). As shown in Fig. 10, inhibition of Pgp led to doxorubicin-induced cell death 24 h after exposure (B,D) as visible by defragmentation of cell nuclei and break down of the monolayer, whereas in the absence of Pgp inhibition, cells showed a confluent monolayer without any signs of toxicity upon exposure to doxorubicin (A,C). Interestingly, in the absence of tariquidar, yellow aciniform vesicle aggregates were observed in both cell lines that—as indicated by the yellow color—contained both Pgp-EGFP (green) and doxorubicin (red). Such drug-sequestering vesicular (lysosomal) aggregates were recently described and characterized by us in more detail and termed “barrier bodies”, as they are attached to the apical side of the plasma membrane and phagocytized by neutrophils, thus constituting a new mechanism of drug disposal at the BBB [13].

Doxorubicin is known to exert lysosomotropic effects [12], which prompted us to examine such effects in MDR1-EGFP transduced RBE4 and hCMEC/D3 cells. For this purpose, we performed live-cell image acquisition with a fluorescence microscope (Lumascope 620) within the cell culture incubator before and after exposure to doxorubicin. These results indicated that doxorubicin increased the number of Pgp-EGFP-positive intercellular vesicles in both cell lines (Fig. 11). This was substantiated by counting these vesicles of likely endolysosomal origin (Additional file 5). Exposure to doxorubicin (10 µM) increased the average number of Pgp-EGFP-positive vesicles per cell by 220% in RBE4 cells and 170% in hCMEC/D3 cells, respectively. In addition to the number, the size of the Pgp-EGFP-positive vesicles seemed to increase (Fig. 11). The increase in number and size of EGFP-positive intracellular vesicles after treatment with doxorubicin may either indicate lysosomal biogenesis or fusion as reported previously for exposure of other cell types to chemotherapeutic drugs such as doxorubicin [12]. Alternatively, the increased number of EGFP-positive intracellular vesicles may indicate increased transport of Pgp-EGFP into the vesicles, which, however, is considered less likely. As noted above, we have previously shown for hCMEC/D3 cells that the EGFP-positive intracellular vesicles are endolysosomal vesicles [13, 31], which was confirmed in the present experiments for RBE4 cells (Fig. 8).