There is a need for predictive assays amenable to medium to high-throughput screening for the assessment of brain penetration of drug leads. To this date, only a few immortalized human brain capillary endothelial cell lines have been developed and used for the establishment of human in vitro BBB models . The most extensively characterized human cell line is the hCMEC/D3 cell line, which has been reported to represent a promising in vitro human BBB model for drug transport studies [45, 46]. TY10 cell line, transduced with a temperature-sensitive SV40 large-T antigen, has furthermore been reported to be a promising and advantageous cell line with excellent expression of TJ proteins such as claudin-5, occludin, and ZO-1 . Whereas at 33°C, TY10 cells can be cultivated for more than 50 passages without undergoing morphological changes, a temperature shift from 33°C to 37°C results in the exclusion of the SV40 large-T antigen as a cancer gene [33, 34].
However, monolayers of immortalized human brain capillary endothelial cell lines are known to form only moderately restrictive barriers, with TEER values in the range of 20–200 Ωcm2[17, 20, 21]. To increase barrier tightness of four currently available human brain capillary endothelial cell lines, hCMEC/D3, hBMEC, TY10, and BB19 (Figure 1), we tested co-culture models with immortalized human astrocytes (SVG-A cell line) and pericytes (HBPCT cell line). Interestingly, we did not observe an increase of TJ resistance in immortalized human brain capillary endothelial cells under co-culture conditions. All co-culture models produced TEER values that were comparable or lower than those recorded with mono-cultures (for hBMEC cell line, see Figures 6 and 8A). This suggested that the investigated endothelial cell lines were unable to respond positively to stimuli from immortalized astrocytic or pericytic cells.
These findings are in accord with previous reports in which no additional benefit in terms of TEER was observed when culturing hCMEC/D3 cells with ACM or when co-culturing the endothelial cells with human astrocytes, respectively [24, 47]. In contrast, our results do not support recent findings which showed that TJ resistance of hCMEC/D3 and TY08 cells was significantly increased in co-cultures with human brain astrocytes/pericytes and HBPCT cells, respectively [29, 48]. This divergent effect might possibly be due to the difference of culture conditions and the nature of cell types.
We observed that membrane material of inserts and pore size had a significant impact on barrier tightness, and we optimized mono-culture systems by systematically screening a large set of 24-well tissue culture inserts from different providers for each cell line (Table 1). Highest TEER values (28.4 ± 2.47 Ωcm2) were observed with hBMEC mono-cultures on 24-well tissue culture inserts from Greiner Bio-one® with transparent PET membrane and 3.0 μm pore size (Table 1). Our findings clearly show that the selection of an appropriate tissue culture insert is critical when establishing a BBB model using these immortalized human brain capillary endothelial cell lines, corroborating previous findings in which a substantial impact of material characteristics on the adherence of cells and barrier tightness was demonstrated [31, 32]. hCMEC/D3 and TY10 cells produced TEER values in the range of 10 Ωcm2 (Table 1). BB19 cells were not included into the screening of tissue culture inserts, since these cells yielded extremely low TEER values (around 5 Ωcm2, Figure 3E). These findings are in agreement with previous studies suggesting that the use of BB19 cells as an in vitro model of the human BBB is limited due to a high sucrose permeability .
Surprisingly, we consistently observed lower TEER values as compared to literature (hCMEC/D3 and hBMEC: TEER ranging from 40–200 Ωcm2[21, 24, 25], TY10: TEER in the range of 40 Ωcm2). A possible explanation may be that we used a different system (automated CellZscope) for assessment of TEER . Low TEER values might also arise from a high concentration of serum and growth factors in the growth medium, which has been reported to prevent TJ formation between endothelial cells . However, hCMEC/D3 cells cultured with growth medium containing 5% FBS instead of 20% FBS, and hBMEC cells cultured in growth medium without growth factor supplementation did not result in increased TEER values or reduced paracellular permeability (for hCMEC/D3, see Figures 3A, 3B, Figures 7A, and 7B, for hBMEC see Figure 8). Furthermore, the selection of the well format might affect TEER values. Because we aimed to establish an in vitro BBB model suitable for higher throughput, we miniaturized the assay to a 24-well format that was selected previously for a bovine in vitro BBB model .
Subsequent optimization of the hBMEC mono-culture system resulted in TEER values in the range of 30–40 Ωcm2 (Table 2). We found that the cell seeding density is critical, since highest TEER values were obtained when hBMEC cells were seeded at a density ranging between 4.5 × 104 and 9.0 × 104 cells/cm2 onto coated inserts (Table 2). These conditions seem to be best for cell growth and TJ development between adjacent cells on the surface area of the Greiner Bio-one® inserts (0.336 cm2). As hBMEC cells are of human origin, we moreover investigated the effect of HS on barrier tightness. However, no increase in TEER could be observed (Table 2). These results do not confirm previous findings that the permeability of sucrose through hCMEC/D3 monolayers could significantly be reduced by HS supplementation . One may speculate that this could be due to a different type of HS used in these experiments. An individual HS batch as used for our experiments is prone to higher batch-to-batch variation in soluble factors and proteins than pooled HS from commercial sources.
In the evaluation of paracellular permeability, mean Papp values for Na-F and LY were significantly lower with hBMEC monolayers (5.08 ± 0.220 × 10-6 cm/s and 5.39 ± 0.364 × 10-6 cm/s, respectively) than with hCMEC/D3 and TY10 monolayers (Figures 7A and B), corroborating our measurements of TEER values. Permeability values in the order of 10-6 cm/s were obtained previously in various in vitro BBB models [7, 20, 39, 51].
In the biochemical and immunocytochemical characterization of cellular junctions, VE-cadherin was detected in all cell lines, albeit at varying levels (Figure 4). This confirmed their endothelial lineage. Interestingly, the TJ protein claudin-5 was expressed at similar levels as VE-cadherin, confirming that VE-cadherin controls claudin-5 expression . The cellular junction marker protein ZO-1 showed the same level of expression in hCMEC/D3 and hBMEC cells, but was expressed only at very low levels in BB19 and TY10 cells (Figure 4). Furthermore, white arrows in Figure 5 point to ZO-1 signal at the leading edge of migrating hBMEC cells, confirming findings of previous studies .
We used an automated CellZscope system  in order to obtain highly standardized data on-line. This system has several advantages over other methods of TEER measurement: TEER values are recorded in real-time every hour in the incubator, thereby reducing workload and avoiding any damage of the cell layer during growth. Also, a disruption of the cell layer is immediately visible from the recording. In addition, information about confluency (CCL values) is obtained simultaneously, reducing the risk of false interpretation of TEER values . Experiments with Caco-2 cells on 24-well inserts demonstrated that the CellZscope is likewise an efficient tool for evaluating barrier tightness in other cell lines, i.e. those used for the study of intestinal drug absorption. Again, independent of the membrane surface area, TEER values of Caco-2 monolayers measured with the CellZscope (24-well format) correlated to off-line TEER values measured manually with an EVOM using a 6-well format (data not shown). One limitation of the CellZscope system may be its design which does not allow the seeding of cells on the bottom of the plate. Investigation of triple co-culture model systems is hence not possible.