High Content Screen Assay of Endosome Movement
Quantitative Tracking of Intracellular Organelle Movement
- Fig 1: Characterization of U2-OS clones expressing LAMP1-GFP transgene. A) Co-localization of endogenous endosomes with exogenous endosomes. Endogenous endosomes are stained with anti-LAMP1 primary antibody and Alexa-633 secondary antibody. Exogenous endosomes are labeled with GFP. Several images of three different clones C3, C5 and C7 were analyzed. Manders coefficient shows the overlap between GFP and Alexa-633 channels. Maximum co-localization is 1, no co-localization is 0. B) Box plots with number of endosomes per cell. Number of exogenous GFP endosomes in clones C3, C5 and C7 compared to number of endogenous endosomes in wild type U2-OS cells. Clones C5 and C7 have significantly higher number of endosomes per cell compared to U2-OS. Number of exogenous GFP endosomes in clone C3 is similar to number of endogenous endosomes in U2-OS.
- Fig. 2: Calculated speed of endosome movement depends on the speed of acquisition. Density distributions of endosomes speed calculated from time-lapses movies taken with different acquisition frame rates. Slower acquisition speed (here in pink, corresponds to one image every 0.5 sec) results in slower calculated speed of movement, density distribution is shifted to left side. The endosomes speed calculated from time lapse movie with fast acquisition frame rate (here in red, corresponds to one image every 0.03 sec) is shifted toward the right side on x axis.
- Milan Esner, PhD, TDS – Technology Development Studio, Max-Planck Institute of Cell Biology and Genetics
Today, high content screens are performed mostly as end-point assays on fixed cells. With this kind of approach only static events can be described. But cells are in permanent dynamic equilibria. This dynamic state is necessary to respond to changes of the surrounding environment in both cell culture and in multi-cellular organisms. Here we developed an assay for quantitative measurement of endosome movement. This type of assay will bring new insights into cell physiology and pathophysiology and open a new field of cytotoxicity assays of drug candidates.
Endocytosis is a process by which molecules from the outside enter the cell. These molecules can be nutrients, signaling molecules, neurotransmitters, certain pathogens, drugs etc. . Endocytic pathways are very complex and consist of several different compartments - early endosomes, late endosomes and lysosomes, which are in constant motion to ensure the uptake of molecules from the environment and their delivery to the proper destination inside cells. Decreased motility leads to decreased levels of endocytosis and might cause some serious diseases [3, 4]. To better understand the physiology of endosome kinetics, to identify which motor proteins drive the movement of individual endosomes and to find new compounds modulating their movement, we developed an assay, able to track and quantitatively measure the movement of thousands of endosomes per movie. The assay was successfully used for an RNA interference screen and a chemical compounds screen. Technical aspects of developing such an assay will be discussed.
Labeling of Intracellular Organelles
Labeling of intracellular organelles is always challenging in live cells. As antibodies cannot be used one has to search for cell permeable dyes, which are organelle specific, photo-stable and non-toxic. The palette of suitable dyes is very limited and the common way of labeling organelles in live cells is expression of fluorescent proteins tagged to specific proteins. For our assay, we created stable U2-OS clones expressing Green Fluorescent Protein (GFP) fused N-terminally to LAMP-1 (Lysosome Associated Membrane Protein - 1).
LAMP-1 is located in late endosomes and lysosomes, therefore GFP will be located only in those organelles. Using this method, several parameters have to be examined carefully - level of transgene expression, its localization and its physiological function. We extensively characterized several clones (fig. 1) and chose one clone with best characteristics.
Nuclei labeling could be done either by labeling with cell permeable dyes, like Hoechst or DAPI, or by expressing fluorescent protein other then GFP, for example RFP fused to histone protein. Disadvantage of dyes is that they might be toxic at certain concentrations and that the excitation is near the UV spectrum, which is causing photo-toxicity after a certain number of acquisitions.
Image Acquisition and Analysis
Endosomes are very motile and can move with speed up to few µm/sec . Several parameters have to be examined and optimized for every kinetic assay - 1) frame rate per second - how many images per second have to be taken to track an object reliably and 2) the length of acquisition - how many images one movie has to contain to sample the dynamics for reliable statistics. These parameters are crucial to obtain correct measurements from the acquired movies. In figure 2 an example of how the distribution of speed calculation changes with changes of acquisition parameters. In our assay we were limited by the acquisition speed of our automated microscope, which is 4.5 images per second. This acquisition speed allows us to track 80% of all endosomes, the remaining 20% are too fast to be tracked at that frame rate.
The next parameter is the length of acquisition. The movie should be as short as possible (less images = shorter acquisition time, less data to store and analyze, and less photo damage to cells), but long enough to calculate statistically significant kinetic parameters. We acquired several movies of different lengths and found that 60 frames is the minimal amount of images per movie. To have a safety margin, we acquired 80 images per movie. To increase the sampling in order to have more robust statistics, we acquired four movies per well.
Because nuclei are rather static, one Hoechst image per movie was acquired to provide information about spatial distribution of the endosomes within cells and vectorial information about the direction of movement.
Altogether, the acquisition of one well requires 320 GFP images and 4 Hoechst images. The time per well for acquisition is 90 sec, 24 min for one column and 9 h to acquire an entire 384 well plate.
Tracking of objects is a very complex process. We used in house developed software MotionTracker  that is able to segment and track thousands of endosomes with great accuracy and calculate several kinetic parameters, like speed, processivity of movement, maximal displacement of vesicles and others.
It is problematic to obtain similar incubation periods for all wells due to the long time that is required for the acquisition of an entire plate. If the compounds would be added in one injection with a 384 needle pipetting head, the incubation time difference between the first well and the last well would be 9 h. To avoid this situation and to have similar incubation time for all compounds we used column-wise injection of compounds every 24 min. Using this approach, compound addition to the entire plate takes 9 h and the minimal incubation time is also 9 h +/- 24 min. A second acquisition is performed after another 24 h to obtain kinetics of drug action. Thus the assay has two temporal dimensions.
Due to the length of the assay, the incubation parameters inside the microscope have to be verified and optimized to guarantee that cells do not suffer unduly. We found that humidity inside the chamber is the biggest issue for long term cell survival due to media evaporation. To eliminate this problem, we covered all wells with 40 µl of mineral oil. This prevents the problem with evaporation and cells survive and grow very well for at least 96 h.
We successfully finished screening a chemical library of 2,500 compounds and a library of 740 short interfering RNA oligonucleotides (siRNA) targeting human motor proteins. The screen generated 4.5 TB of images and analysis is currently progressing on the 2500 core super computer of Technische Universität Dresden.
We developed a new assay for tracking endosomes in live cells. This assay is not dedicated to high throughput screening, but rather for secondary assays or screening of small libraries. We performed successfully several screens with both chemical compounds and siRNA. This type of assay is not limited only to the movement of endosomes, but could be applied to other organelles with appropriate labeling. We believe that this kind of assay could not only be of great interests for fundamental research, but also to be applied in toxicology for better prediction of adverse drug reaction (ADR) of new compounds. The dynamics of cellular organelles are likely quickly affected by toxic drugs, as cells try to cope with detoxifying and energy metabolism is compromised. Kinetic analysis could therefore be an early sensitive surrogate marker of toxicity.
 Doherty G. et al.: Annual Review of Biochemistry, 857-902 (2009)
 Matteoni R. and Kreis T E.: J Cell Biol, 1253-1265 (1987)
 Pal A. et al.: J Cell Biol, 605-618 (2006)
 Lai C. et al.: Molecular Brain, 23 (2009)