LipidUNet-Machine Learning-Based Method of Characterization and Quantification of Lipid Deposits Using iPSC-Derived Retinal Pigment Epithelium

In this study, we aim to understand the age-related macular degeneration pathogenesis and find the potential treatment for it. The semi-automated pipeline for lipid droplet analysis provides a robust system to study metabolism and aging diseases. Our lab focuses on finding drug targets for the treatment of AMD, either using genetic or small molecule screens that aim at reducing the accumulation of lipid deposits.

To begin, dissolve Nile Red powder in acetone at three milligrams per milliliter concentration. Incubate the solution for 15 minutes at room temperature with periodic mixing. Filter the solution using a 0.22-micrometer filter once or twice, depending on the amount of precipitate remaining in the solution.

Then prepare the working solution by diluting the stock solution at a 1:500 ratio in DPBS. Add 200 microliters of the working solution to the paraformaldehyde fixed RPE cells, and incubate for 30 minutes at room temperature on a shaker, protecting from light. After incubation, wash the samples twice with DPBS.

Then add 200 microliters of DPBS, and store them at four degrees Celsius. For BODIPY staining, prepare a 3.8-millimolar stock concentration by dissolving BODIPY in anhydrous DMSO. Then prepare the working solution by diluting the stock solution at a 1:300 ratio in DPBS.

Add 200 microliters of the working solution of BODIPY to paraformaldehyde fixed RPE cells and incubate overnight on a rocker at room temperature. The next day, wash the cells thrice with DPBS, and add 200 microliters of DPBS before storing them at four degrees Celsius. For APOE immunostaining, prepare a buffer solution by combining DPBS with 1%BSA 0.5%TWEEN 20, and 0.5%Triton X-100.

Block and permeabilize the paraformaldehyde fixed RPE cells in 200 microliters of buffer solution for one hour. After incubation, add APOE primary antibody diluted at 1:100 in the buffer solution, and incubate overnight at room temperature. The following day, wash the samples thrice with DPBS.

Then add 200 microliters of secondary antibody solution diluted at 1:1000 for one hour at room temperature. After incubation, wash the cells thrice with DPBS, and add 200 microliters of DPBS before storing them at four degrees Celsius. Using a confocal microscope and a 40x objective, create a scan profile with the appropriate fluorescent channels for the lipid marker used.

After setting up the acquisition mode and channels, optimize the focus strategy by clicking on Software Autofocus. Then set the range of the Z-stack to 20 microns. After this, optimize tiles, and open the viewer to select the tile positions.

Next, using a batch image processing method, create maximum projections of each Z-stack with the extended depth of focus method. Before exporting the images, set the compression to none, and ensure Original data is checked. The resulting image is a maximum projection grayscale TIFF of only the fluorescence channel expressing the lipid marker.

Identify the TIFF images representing either Nile Red, BODIPY, or APOE, and move them into the Images folder within a directory named either Nile Red, BODIPY, or APOE, depending on the method being used. Open the LipidUnet software. In the Predict tab, select the relevant directory by clicking on the ellipsis and navigating to the named directory.

Confirm that LipidUnet has identified the images correctly by checking the class entry. Click Predict to observe the task progress. Then using the Mask Analysis tool, iterate through the generated masks and provide a quantitative count of the thresholded lipid deposits from the mask images.

Successful differentiation and maturation of RPE showed a confluent monolayer with pigmentation and polygonal morphology. In contrast, unsuccessful differentiation showed clusters of dying cells. In fluorescent images, Nile Red and BODIPY deposits appeared as small, bright, circular points.

A negative result shows incorrect image segmentation by mistaking background fluorescence as the deposit, either due to weak staining or high background intensity. APOE deposits varying in size, shape, and signal intensity and requiring optimization of staining and imaging methods to minimize variation are shown.