Human retinal pigment epithelium cells as functional models for the RPE in vivo

Abstract

Purpose: The two most commonly used in vitro models of the retinal pigment epithelium (RPE) are fetal human RPE (fhRPE) and ARPE-19 cells; however, studies of their barrier properties have produced contradictory results. To compare their utility as RPE models, their morphologic and functional characteristics were analyzed.

Methods: Monolayers of both cell types were grown on permeable membrane filters. Barrier function and cellular morphology were assessed by transepithelial resistance (TER) measurements and immunohistochemistry. Protein expression was evaluated by immunoblotting and ELISA assays, and retinoid metabolism characterized by HPLC.

Results: Both cultures developed tight junctions. However, only the fhRPE cells were pigmented, uniform in size and shape, expressed high levels of RPE markers, metabolized all-trans retinal, and developed high TER (>400 Ωcm(2)). The net secretion of pigment-epithelium-derived factor (PEDF) was directed apically in both cultures, but fhRPE cells exhibited secretion rates a thousand-fold greater than in ARPE-19 cells. The net secretion of vascular endothelial growth factor (VEGF) was significantly higher in fhRPE cultures and the direction of this secretion was basolateral; while net secretion was apical in ARPE-19 cells. In fresh media, VEGF-E reduced TER in both cultures; however, in conditioned media fhRPE cells did not respond to VEGF-E administration, but retreatment of the conditioned media with anti-PEDF antibodies allowed fhRPE cells to fully respond to VEGF-E.

Conclusions: Properties of fhRPE cells align with a functionally normal RPE in vivo, while ARPE-19 cells resemble a pathologic or aged RPE. These results suggest a utility for both cell types in understanding distinct, particular aspects of RPE function.

Figure 1.

RPE cell monolayer development and the establishment of barrier function. RPE paracellular permeability was assessed by transepithelial resistance measurements (TER) in ARPE-19 and fhRPE cells. In both types of cells, the TER began to increase after reaching confluence. ARPE-19 cells became confluent within 5 days after plating and reached a maximum resistance of 43 ± 5 Ωcm² within 2 weeks (n = 12). fhRPE cells became confluent within 3 to 5 weeks after plating and reached a maximum resistance of 1046 ± 43 Ωcm² within 6 to 8 weeks (n = 10). Values are mean ± SE of individual wells.

Figure 2.

Epithelial morphology in established RPE cell monolayers. Epithelial morphology was assessed by phase contrast and immunofluorescence microscopy in ARPE-19 and fhRPE cells. Phase-contrast micrographs of (A) ARPE-19 and (D) fhRPE cells provide evidence of differences in morphology and pigmentation. Immunofluorescent micrographs were prepared with fluorescein-labeled secondary antibody (mouse): (B) ARPE-19 and (E) fhRPE cells with anti-ZO-1 primary antibody indicate that ARPE-19 cells have inferior structure and organization. Micrographs of (C) ARPE-19 cells and (F) fhRPE cells with anti-occludin primary antibody show that ARPE-19 cells have inferior occludin staining at the cell-cell contacts, indicating limited tight junctions (as at the same time ZO-1 staining is clearly present). The images were prepared using a 20× objective lens; the scale bar represents 50 μm.

Figure 3.

Retinoid metabolism in monolayer RPE cultures. Expression of RPE visual cycle proteins (A) showing immunoblots against RPE65 and CRALBP in both fhRPE and ARPE-19 cells. While fhRPE cells abundantly express both proteins, ARPE-19 cells express only low quantities of CRALBP. β-Actin indicated approximately equal loading of the lanes. Functional retinoid metabolism was missing from ARPE-19 cells (B); but was present in fhRPE cells (C) as shown by the corresponding retinoid profiles taken by monitoring the HPLC at 360 nm 2 days post administration of 5 μmol/L all-trans retinal (atRal) in phosphatidylcholine (PC) vesicles. The profile for ARPE-19 cells (B), was dominated by unconverted all-trans retinal, while the profile for fhRPE cells (C), was dominated by retinyl-esters and 11-cis retinal (11cRal), indicating functional retinoid metabolism. The data are representative of three independent experiments. atRol, all-trans retinol.

Figure 4.

Secretion of VEGF and PEDF from established RPE cell monolayers. Time-dependent changes in the concentration of VEGF (A, B) and PEDF (C, D) were measured by ELISA from media collected from the apical (●) and basolateral (○) sides of ARPE-19 (A, C) and fhRPE (B, D) cultures at the indicated time points of 0, 6, 24, 30, and 48 hours in all experiments except fhRPE PEDF, where it was 0, 0.5, 1, 1.5, and 2 hours. The secretion rates were determined from the slopes of curves (linear) fitted to the experimental data. Solid lines indicate apical and dashed lines basolateral secretions. Values are the mean ± SE of individual wells (n = 4).

Figure 5.

VEGF-induced barrier breakdown in RPE cell monolayers. VEGF-E (a selective VEGF-R2 agonist) was administered apically to ARPE-19 (A) and fhRPE (B) cells at a concentration of 5 ng/mL in cultures that had been maintained in either fresh, unconditioned media (●), or in conditioned media (○) for 24 hours. In the fhRPE cells, a third experiment was also performed in conditioned media, where the cells were pretreated with anti-PEDF 1 hour before the treatment with VEGF-E (□). TER was measured starting 1 hour before VEGF-E administration, to 5 hours post VEGF-E administration. To show the trend of the change, the closed circles were connected with solid lines, the open circles were connected with dashed lines, and the open squares were connected with a dotted line. Values are the mean ± SE of individual wells (n = 6) normalized to the TER at t = −60 min.