Human pluripotent stem cells for the modelling of retinal pigment epithelium homeostasis and disease: A review.

Human pluripotent stem cells (hPSCs), which include induced pluripotent stem cells and embryonic stem cells, are powerful tools for studying human development, physiology and disease, including those affecting the retina. Cells from selected individuals, or specific genetic backgrounds, can be differentiated into distinct cell types allowing the modelling of diseases in a dish for therapeutic development. hPSC-derived retinal cultures have already been used to successfully model retinal pigment epithelium (RPE) degeneration for various retinal diseases including monogenic conditions and complex disease such as age-related macular degeneration. Here, we will review the current knowledge gained in understanding the molecular events involved in retinal disease using hPSC-derived retinal models, in particular RPE models. We will provide examples of various conditions to illustrate the scope of applications associated with the use of hPSC-derived RPE models.

A semi-automated pipeline for quantifying drusen-like deposits in human induced pluripotent stem cell-derived retinal pigment epithelium cells.

Age-Related Macular Degeneration (AMD) is a highly prevalent form of retinal disease amongst Western communities over 50 years of age. A hallmark of AMD pathogenesis is the accumulation of drusen underneath the retinal pigment epithelium (RPE), a biological process also observable in vitro. The accumulation of drusen has been shown to predict the progression to advanced AMD, making accurate characterisation of drusen in vitro models valuable in disease modelling and drug development. More recently, deposits above the RPE in the subretinal space, called reticular pseudodrusen (RPD) have been recognized as a sub-phenotype of AMD. While in vitro imaging techniques allow for the immunostaining of drusen-like deposits, quantification of these deposits often requires slow, low throughput manual counting of images. This further lends itself to issues including sampling biases, while ignoring critical data parameters including volume and precise localization. To overcome these issues, we developed a semi-automated pipeline for quantifying the presence of drusen-like deposits in vitro, using RPE cultures derived from patient-specific induced pluripotent stem cells (iPSCs). Using high-throughput confocal microscopy, together with three-dimensional reconstruction, we developed an imaging and analysis pipeline that quantifies the number of drusen-like deposits, and accurately and reproducibly provides the location and composition of these deposits. Extending its utility, this pipeline can determine whether the drusen-like deposits locate to the apical or basal surface of RPE cells. Here, we validate the utility of this pipeline in the quantification of drusen-like deposits in six iPSCs lines derived from patients with AMD, following their differentiation into RPE cells. This pipeline provides a valuable tool for the in vitro modelling of AMD and other retinal disease, and is amenable to mid and high throughput screenings.

Integrating deep learning and unbiased automated high-content screening to identify complex disease signatures in human fibroblasts.

Drug discovery for diseases such as Parkinson's disease are impeded by the lack of screenable cellular phenotypes. We present an unbiased phenotypic profiling platform that combines automated cell culture, high-content imaging, Cell Painting, and deep learning. We applied this platform to primary fibroblasts from 91 Parkinson's disease patients and matched healthy controls, creating the largest publicly available Cell Painting image dataset to date at 48 terabytes. We use fixed weights from a convolutional deep neural network trained on ImageNet to generate deep embeddings from each image and train machine learning models to detect morphological disease phenotypes. Our platform's robustness and sensitivity allow the detection of individual-specific variation with high fidelity across batches and plate layouts. Lastly, our models confidently separate LRRK2 and sporadic Parkinson's disease lines from healthy controls (receiver operating characteristic area under curve 0.79 (0.08 standard deviation)), supporting the capacity of this platform for complex disease modeling and drug screening applications.