Abstract
Purpose :
The emergence of primary cilium (PC) on human retinal pigment epithelium (RPE) cells regulates maturation of RPE as demonstrated by polarized expression of surface markers, tight junction formation, enhanced phagocytosis of photoreceptor outer segments (POS), and improved transepithelial potential/resistance (TEP/TER). We hypothesized that by analyzing the location of the PC as well as morphometric data from induced pluripotent stem cell (iPSC) derived RPE we could predict RPE maturation as defined by phagocytosis of POS, TER and TEP.
Methods :
Primary human iPSC derived RPE were cultured in porous transwell membranes over 6 weeks in the presence of PC inducers/suppressors (Aphidicolin, PGE2, and HPI-4). At the end of 6 weeks cells were stained for PC markers ARL13B, and GT335, tight junction marker ZO-1 and actin cytoskeleton stain phalloidin. Phagocytosis of labeled POS was quantified by flow cytometry. TER and TEP was recorded using a modified Ussing chamber. Cell morphometric quantification and location of PC, cellular center of mass (COM) and cell centroid were assessed via a novel ImageJ plugin. Linear and logistic regressions correlating POS level, TER and TEP with mean distance of cellular PC from cell centroid and COM as well as various morphological features were performed.
Results :
Enhanced induction of PC in RPE cells improved cell compactness and the number of neighbors per cell to a level similar to that of native human RPE. The level of cellular phagocytosis of POS, TER, and TEP were significantly correlated to PC distance from both RPE centroid and COM. Additionally, models including cell area, number of neighbors and aspect ratio of cell major and minor axis were found to be significantly predictive of RPE functionality.
Conclusions :
Using linear and logistic regression we show that iPS cell-derived RPE maturation, as assessed by phagocytosis of POS, monolayer TER and TEP can be predicted with high statistical certainty using four metrics: either PC distance to RPE centroid or COM, and cell area, number of neighbors, and the ratio of the cell major and minor axis. Due to the significance of the predictive ability of the model we conclude that analysis of fluorescent imaging of cells can adequately predict functional cell phenotype in lieu of, or in conjunction with, physiological tests.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.