One of the rare genetic disorders that we at Perlara have been working on is Mucolipidosis type IV (MLIV) – one of the Mucolipidosis group of inherited lysosomal storage disorders. For background on MLIV, and to learn more about our work in optimizing a MLIV drug repurposing screen using MLIV patient fibroblasts, read Feba’s post from last December. This post will detail how we went about conducting the screen.
Choosing a Marker
The first step was to identify a screenotype that we could scale up from 96 well plates to 384 well plates for a high throughput screen. We had a choice between LysoTracker and P62/LC3 antibody staining. LysoTracker, a fluorescent dye for labeling and tracking acidic organelles, showed a multifold difference in lysosome fluorescence in wild type (WT) cells versus patient MLIV cells (IVS3−2A→G mutation). P62/LC3 antibody staining showed a minor difference in accumulation between wild type and patient MLIV cells only after starvation. We decided on LysoTracker for the main screen because of the ease of staining, and the more significant differential between wild type cells and patient MLIV cells. Keep in mind the greater the differential, the easier it will be to distinguish when a compound brings the phenotype back to WT. We kept P62/LC3 as a secondary screen since MLIV cells display an autophagy defect.
Starting from the left you can see starved P62 staining (40x), starved LC3 staining (40x), and finally, LysoTracker (20x). Though the difference in P62 is noticeable, LysoTracker shows the greatest contrast between WT and MLIV.
Creating a Mask
The next hurdle was finding a way to analyze the LysoTracker data after collecting it. The screen layout consisted of running several thousand compounds, imaging each well, and comparing those images to controls to see if any rescued the phenotype. You can imagine doing this by eye would be both time consuming and very subjective. To get around this we use a program which recognizes fluorescence captured and can quantify it. LysoTracker fluoresces in the red, channel and our nuclear marker fluoresces in the blue channel. Therefore, we created a mask using Custom Module Editor on MetaXpress (special thanks to Paula Gedraitis for assistance!) that identified the sum of the fluorescent puncta in the red channel and divided it by the nuclear count giving us a reading of total puncta and total fluorescence per cell. We do this to control for cell density in each well. Though each well is seeded with a set number of cells, the reality is, there are differences in cell numbers that need to be accounted for. Using this mask, we had a single reading per well which could be easily compared to controls.
10x. Using the TRITC channel the mask calculates the total amount of puncta and divides that by the number of cells found in the DAPI channel. The output is the total TRITC puncta sum per cell.
Scaling to 384 Well Plates
After choosing the screenotype and having a mask we moved to optimize the LysoTracker staining in 384 well plates for higher throughput screening. We found staining to be consistent as we moved to 384 wells and saw the relationship between WT and MLIV patient cells scale. At this point our vision for the screen was to incubate MLIV patient cells with the Microsource Spectrum Collection (collection of small molecules), then stain with LysoTracker and see what molecules rescued the phenotype back to WT.
Running the Microsource Spectrum Library
Thinking we had everything ready to go, we ran the first replicate of the Microsource Spectrum Collection (about 3000 molecules) using LysoTracker on 384 well plates, and quickly realized an unexpected gradient present on the plates. The intensity of the fluorescence was diminishing from left to right, something we had not observed in 96 well plates. We attributed this to the introduction of the LysoTracker, a mild base, which causes the lysosomes to become more basic and reduce fluorescence intensity over time. Keep in mind that LysoTracker stains for acidic organelles and the actual uptake of the dye is making the cells more basic hence reducing the fluorescence of the dye.
We also tested a positive control, MK6-83, which is known to rescue MLIV cells if the TRPML mutation is a point mutant. On first glance it looked like it was working because of the reduced fluorescence, but we realized that the reduction in fluorescence was due to the lysotracker dye itself, since a lot of time passes by the time the second half of the plate is imaged. By running the plate with MK6-83 on the left of the plate instead we saw that it did not in fact rescue the phenotype, and it was not used as a control. We were relieved to note that the observed decrease upon treatment with MK6-83 just had to do with lysotracker and well location, because the cells we are working with are null for TRPML and should not respond to this compound.
Full 384 well plate dyed with LysoTracker. Starting from the left we have a full column on WT then a full column of MLIV followed by the Microsource until the last two columns which are the MK6-83. There is a clear reduction in fluorescence on the plate from left to right.
Fixing the LysoTracker gradient
Knowing what was causing the gradient, the next step was to find a way to reduce the effect of it. Through several trials we found that live cell imaging with a significantly shorter imaging time reduced the severity of the gradient to the point where it was no longer relevant. To negate any lingering gradient, we also added both MLIV and WT controls to the left and right columns on the plate allowing us to compare wells to nearby controls. Furthermore, we screened only half of each source plate with 12 columns of compound library per destination plate. The updated screen consisted of half plates with compound treated cells being incubated in LysoTracker, then imaged live without fixation with a much shorter imaging time. The wells on the left side of the plate were compared to columns 1 and 2 controls and the wells on the right side of the plate were compared to columns 15 and 16 controls.
New plate layout with WT/MLIV cells on each side. The wells to the left of the red line will only be compared the controls on the left side of the plate and vice versa. Only 12 columns in the middle are used in the half plate format vs. 20 in the full plate.
Running the Microsource Collection (Again)
With the gradient problem fixed we went ahead and ran the first replicate of the Microsource Collection again. When looking at the first replicate we could see toxic molecules, molecules that reduced LysoTracker fluorescence, and molecules that increased LysoTracker fluorescence. All signs that the screen was running smoothly. We continued with the second and third replicate of the screen and ended up with a great correlation between the replicates as well as several compounds that were hits on all three replicates.
An example of a hit from one of the microsource compounds. The blue stain is for nuclei and the yellow is LysoTracker. The left shows WT with little to no LysoTracker, the middle shows the robust puncta of MLIV, and on the right is a compound that reduced the MLIV fluorescence back to WT levels.
The next step for us is to take the hits from the LysoTracker screen and run it through our LC3/P62 secondary assay to see if they are also restoring autophagy. We will also compare the cell team’s hits with the other two teams (worm and fly) to see if there are overlapping compounds that rescue disease across multiple models. Will keep you posted!
In the meantime, you can learn more about our work using model organisms for MLIV by reading previous posts on MLIV flies and worms.
