Individual cell directionality decreases in single cells compared
to cells within a group, whereas individual cell speed increases.
This, in control conditions, typically correlates with the rate of
expansion of the explant. The initial slow-paced directional
radial expansion turns into a fast disorganized dispersion of
single cells. Overall dispersion comes to a halt when local cell
density is such that cells have an equal probability of moving in
any direction. However, under experimental conditions, one
can interfere with cell–cell or cell–matrix adhesion and affect
the relationship between explant dispersion and single cell
behavior. For instance, reducing cell–cell adhesion may promote single cell behavior while overall radial expansion of the
explant is impaired. Thus analyzing single cell behavior and
explant dispersion in parallel can be very informative.
15. Triangles’ areas follow an exponential distribution, thus the
usual statistical tests for normal/Gaussian datasets do not
apply if one wants to compare the distribution of triangles’
areas from one explant to another. This problem is easily solved
when comparing populations of explants. One should plot the
mean triangle area of each explant per experimental condition.
That way each explant has the same weight in the mean of a
given condition (otherwise explants with more cells contribute
more triangles and can skew the mean of the dataset). In
addition, the distribution of mean areas under control condition follows a Gaussian distribution and allows simple statistical
tests to be performed.
16. Analyzing explant dispersion using the total explored area is
the simplest and most effective way. This technique does not
require pre- or post-labeling of cell’s nuclei. It enables the
inclusion of the impact of all cells, and can be done on simple
low resolution bright-field or phase-contrast images. Triangulation on the other hand has several caveats. First, one needs to
detect single cell nuclei. This can be done by labeling NC cells
with a nuclear tracer or post-staining them with DAPI. Automatic detection of nuclei is possible (e.g., analyze particles
plug-in in Image J/FIJI) but weaker nuclei are often not
detected, artificially increasing the area of some triangles. Similarly, areas with high cell density cannot be properly segmented
and this results in artifacts (Fig. 4). This later problem can be
partially solved by acquiring images on a confocal microscope,
although high resolution images of nuclei often confuses the
automated detection of particles, since pixel intensity within
each nucleus can vary substantially, and nuclei no longer appear
as solid structures. Alternatively, one can mark each cell’s
nucleus manually. The first option, using confocal imaging,
will never be a realistic routine procedure, even if one owns a
dedicated confocal microscope for time-lapse imaging. The
272
Nade ` ge Gouignard et al.
to cells within a group, whereas individual cell speed increases.
This, in control conditions, typically correlates with the rate of
expansion of the explant. The initial slow-paced directional
radial expansion turns into a fast disorganized dispersion of
single cells. Overall dispersion comes to a halt when local cell
density is such that cells have an equal probability of moving in
any direction. However, under experimental conditions, one
can interfere with cell–cell or cell–matrix adhesion and affect
the relationship between explant dispersion and single cell
behavior. For instance, reducing cell–cell adhesion may promote single cell behavior while overall radial expansion of the
explant is impaired. Thus analyzing single cell behavior and
explant dispersion in parallel can be very informative.
15. Triangles’ areas follow an exponential distribution, thus the
usual statistical tests for normal/Gaussian datasets do not
apply if one wants to compare the distribution of triangles’
areas from one explant to another. This problem is easily solved
when comparing populations of explants. One should plot the
mean triangle area of each explant per experimental condition.
That way each explant has the same weight in the mean of a
given condition (otherwise explants with more cells contribute
more triangles and can skew the mean of the dataset). In
addition, the distribution of mean areas under control condition follows a Gaussian distribution and allows simple statistical
tests to be performed.
16. Analyzing explant dispersion using the total explored area is
the simplest and most effective way. This technique does not
require pre- or post-labeling of cell’s nuclei. It enables the
inclusion of the impact of all cells, and can be done on simple
low resolution bright-field or phase-contrast images. Triangulation on the other hand has several caveats. First, one needs to
detect single cell nuclei. This can be done by labeling NC cells
with a nuclear tracer or post-staining them with DAPI. Automatic detection of nuclei is possible (e.g., analyze particles
plug-in in Image J/FIJI) but weaker nuclei are often not
detected, artificially increasing the area of some triangles. Similarly, areas with high cell density cannot be properly segmented
and this results in artifacts (Fig. 4). This later problem can be
partially solved by acquiring images on a confocal microscope,
although high resolution images of nuclei often confuses the
automated detection of particles, since pixel intensity within
each nucleus can vary substantially, and nuclei no longer appear
as solid structures. Alternatively, one can mark each cell’s
nucleus manually. The first option, using confocal imaging,
will never be a realistic routine procedure, even if one owns a
dedicated confocal microscope for time-lapse imaging. The
272
Nade ` ge Gouignard et al.