Analysis of Lipid Droplet Content in Fission and Budding Yeasts using Automated Image Processing

This protocol facilitates quantitative small-to large-scale analyses of cellular storage lipid content in a variety of yeast species and does so in an unbiased and standardized manner. The technique allows for the rapid processing of microscopic images and provides detailed quantitative outputs to easily compare samples such as various mutants and cells grown under diverse conditions. To begin, follow along in the accompanying text protocol to prepare the staining solutions, culture media, and cells.

Ensure that the cells are healthy and in the desired growth phase prior to running the experiment. Next, prepare a microscope coverslip for each sample to be imaged. Spread one microliter of slide coating solution onto a clean coverslip using the long side of a horizontally positioned pipette tip.

Allow the coating solution to dry completely, and then store the coverslips in a dust-free environment. Next, measure the optical densities of the cell cultures, and for cultures to be analyzed in the same growth phase, try to reach similar values among all tested cultures to ensure comparable experimental conditions. Then, pipette one milliliter of the cell culture into a 1.5-milliliter microcentrifuge tube.

For the S.cerevisiae cells only, add five microliters of the slide coating solution to the tube as well. Then, briefly vortex all of the microcentrifuge tubes, and incubate them at 30 degrees Celsius with shaking for five minutes. Next, add one microliter of the lipid staining solution to each culture aliquot, and vortex them briefly.

Then, add 10 microliters of the cell boundary visualization solution, and vortex them briefly a second time. Collect the cells by centrifuging them at 1, 000 times gravity for three minutes at room temperature. When finished spinning, remove 950 microliters from the supernatant and resuspend the cells in the remaining supernatant using a pipette.

Next, pipette two microliters of the dense cell suspension onto a lectin-coated coverslip. Then, place the coverslip onto a clean microscope slide to form a cell monolayer, and proceed to microscopy as quickly as possible to minimize artifacts in imaging. Setting up the microscope requires long exposures to strong light sources that could cause damage to the sample and skew results.

Therefore, set up the imaging conditions using a dedicated sample slide that will not be further used for lipid droplet quantification. Place the dedicated sample onto the stage of a phase contrast or differential interference contrast microscope setup, and focus on the cells. In the microscope's software, set the Z-stack settings so that they span the whole cell volume.

The total vertical distance depends on the cell size, and the number of optical slices depends on the numerical aperture of the objective. Next, set the focus to move relative to the central focal plane. To image lipid droplets, experimentally set the light intensity and exposure time in the green channel.

Be cautious when imaging, as BODIPY is a very bright fluorochrome that may be rapidly photobleached. Additionally, minimize the exposure time to avoid oversaturation. Capture the full green channel Z-stack first before switching to the blue channel to prevent blurring the mobile lipid droplets.

To image cell boundaries, experimentally set the light intensity and exposure time in the blue channel. If possible, set up and use an automated experimental workflow in the microscope control software to facilitate imaging of multiple samples under standardized conditions. Importantly, all images must be acquired using the same settings to allow for appropriate comparison between samples.

Once imaging conditions have been optimized, focus on the cells and image them in both the green and blue channels. Image multiple fields of view per sample to obtain robust, representative data. Save the blue and green channel Z-stack images as 16-bit, multilayer TIFF files.

Be sure to include the words green or blue in the corresponding file names. Open microscopic images in ImageJ, and remove any image stacks containing a considerable number of cells that moved during acquisition. These appear at different positions in individual Z sections.

Next, remove any image stacks containing highly fluorescent non-cell particles in the blue channel. This is often caused by dirt on the imaging surface or impurities in the cultivation medium and may interfere with detection of cells in their vicinity. Also, remove any image stacks containing a large proportion of dead cells.

These images will display cells with increased blue fluorescence compared to the live cells. While the presence of a few dead cells in the sample is typically not a problem and these cells are automatically discarded during analysis, some dead or dying cells may occasionally be recognized as live cells by the segmentation algorithm and thus skew the reported results. To begin analysis in MATLAB, first create a main folder and copy all MATLAB scripts to this location.

Next, create subfolders with the names of the various yeast species, and copy the respective TIFF images to these locations. Now, start MATLAB, open the script MAIN. m, and run the script.

In the menu, select the yeast species to be analyzed and start image processing. Inspect and process the output files as required using a spreadsheet editor or statistical package. The workflow produces semicolon-separated CSV files and segmented TIFF files with detected cell objects and lipid droplets.

Shown here are wild-type S.pombe cells that were grown in either the complex YES medium or the defined EMM medium. Compared to cells grown in the YES medium, fewer lipid droplets and higher staining intensity per unit of cell volume were detected in the EMM medium. Moreover, individual lipid droplets formed in EMM medium were larger and displayed an increased total staining intensity.

This is in agreement with previous findings of increased stored lipid content in cells grown in EMM. Next, these images show S.japonicus cells from exponential and early-stationary cultures grown in YES medium. Cells entering stationary phase showed a markedly decreased number of lipid droplets per unit of cell volume while volume-normalized lipid droplet fluorescence intensity decreased slightly between the two conditions.

The early stationary-phase lipid droplets were typically moderately larger in size and had moderately higher total fluorescence intensity compared to droplets from exponentially growing cells. In S.cerevisiae cells grown to exponential versus stationary phase, stationary cells contained somewhat fewer lipid droplets per unit of volume compared to exponentially growing cells. However, their volume-normalized droplet fluorescence intensity almost doubled.

This sharp increase in overall lipid droplet content was due to the much higher fluorescence intensity and volume of individual lipid droplets in stationary phase. Following this procedure, output data can be aggregated in a statistical package such as R to create plots of summarized results and run statistical tests on any observed differences in lipid droplet content. In principle, the image processing pipeline can be used for analysis of lipid droplet content in other microorganisms or to analyze other dot-like subcellular structures.