As aforementioned, we chose the human colon 3D organoid model to study the impact of the mitotic spindle orientation on the extracellular matrix displacements. To this end, it was essential to be able to monitor the cell divisions and the ECM displacements over time. Regarding cell division, DNA and tubulin, which forms the mitotic spindle, were visualized using Hoechst (blue channel) and a BioTracker 488 Green Microtubule Cytoskeleton Dye (green channel) respectively (figure 2a). Matrix displacement was investigated by adding fluorescent beads (red channel) into the Matrigel® (figure 2a). The organoid culture evolution was followed by image acquisition every twenty minutes over 3 hours using the Opera Phenix™ confocal microscope. We then performed 3D reconstruction of the acquired images for each of the three channels (figure 2a) and analysed firstly the mitotic spindle orientation and secondly the displacements of the nuclei as well as the matrix displacements.
Quantification of asymmetric and symmetric divisions in the human colon 3D organoid model
We first confirmed that both symmetric and asymmetric divisions occur over our epithelial organoid culture (figure 2b).
We measured the orientation of mitotic spindle during the cell divisions occurring in the human colon organoids over the three hours of observation. Mitotic spindle orientation was determined as the angle formed between the mitotic spindle axis and the organoid-Matrigel® boundary (figure 2b). In the case of asymmetrical divisions, the mitotic spindle is oriented perpendicularly, or radially, to the Matrigel. Conversely, in the case of symmetrical divisions, the spindle orientation is parallel, or orthoradial, to the support. Based on the literature 17, we chose to classify the angles measured into 3 intervals: [0°-30°], [30°-60°] and [60°-90°]. From a total of 83 observed cell divisions visualized in 30 different organoids of 4 different patients (a minimum of 4 organoids/patient and a maximum of 9 organoids/patients were analysed), we counted 45 mitoses with a mitotic spindle angle between 0° and 30°, i.e. 54% of divisions clearly classified as asymmetric. Mitotic spindle angle between 60° and 90°, defining symmetric divisions, were 28, i.e. 34% of divisions. It is also noticeable that a small fraction of mitoses (12%) cannot be assigned to any of the asymmetrical or symmetrical categories, with a mitotic spindle angle between 30° and 60°.
These data also demonstrate the relevance of our biological model for the study of mitotic spindle orientation.
Impact of asymmetric and symmetric divisions on neighbouring cells
We investigated whether asymmetric and symmetric divisions might affect the neighbouring cells by focusing on the positioning of neighbouring nuclei in relation to the dividing cell (figure 3).
Metaphase being the most easily identifiable stage of the mitosis in the images, we quantified the number of neighbouring cells directly surrounding the dividing cells and measured the distance separating the nuclei of the neighbouring cells from the cell at this step of the mitosis. The barycentre of the nuclei serves as reference point for the measurements. The measurements have been done on 11 divisions of both categories picked at random among the 45 asymmetric and 28 symmetric divisions and observed on 8 organoids from 3 patients (a minimum of 2 organoids/patient and a maximum of 3 organoids/patients were analysed).
Asymmetric and symmetric dividing cells have a mean of respectively 7,55 ± 0,37 (mean ± SEM) and 7,82 ± 0,55 neighbouring cells. No significant difference is thus observed (Mann Whitney t-test, p-value = 0,9394). In the absence of mitosis, we counted an average of 6,25 ± 0,28 neighbouring cells for a given cell, thus slightly fewer than in the presence of division (Mann Whitney t-test, p-value = 0,0285 between control group and [60°-90°] group, and p-value = 0.0150 between control group and [0°-30°] group).
Regarding the distances, our data showed that in the absence of cell division, nuclei are significantly closer together (Mann Whitney t-test, p-value < 0,0001 between control group and [60°-90°] group, and p-value < 0,0001 between control group and [0°-30°] group). Comparing the two types of division, in the case of asymmetric division, the nuclei of surrounding neighbour cells are significantly closer to the dividing cell (Mann-Whitney test, p-value = 0.0022) than in the case of symmetric division (figure 3). This result could suggest that the displacement of the matrix induced by asymmetric division could be less important than the one induced by symmetric division.
Impact of asymmetric and symmetric divisions on the 3D matrix displacements
We then studied the displacements induced by both types of division on the matrix surrounding the epithelium. The signals obtained by confocal fluorescence microscopy, thanks to the cell labels (nuclei in blue, tubulin in green, figure 2a) and the beads inserted into the matrix (in red, figure 2a) to visualize the organoids and the ECM respectively, were processed in two distinct ways.
First, we established a framework that enables the tracking of the nuclei displacement over the three hours of culture (figure 4a): on the raw images, we first performed a 3D segmentation of the nuclei signal using Cellpose, and then used the Imaris software to track the nuclei over time. Indeed, Cellpose allows to train the segmentation model according to a machine-learning principle and, if necessary, to manually correct the segmentation. Segmentation using Cellpose therefore enables customizable segmentation with a high degree of precision, unlike segmentation using Imaris, where global segmentation parameters are selected and applied to all objects. This approach allowed us to perfectly perform the nuclei segmentation in 3D for our study, facilitating the follow-up of the nuclei displacement using Imaris.
As shown in figure 4a and based on the measurements reported in table 1, the global analysis of nuclei displacements within the organoids enabled us to establish that, for a giving organoid, the nuclei display an overall uniform displacement as shown in a representative experiment (figure 4a).
Table 1. Nuclei displacements in organoids.
Organoid from
|
Patient 1
|
Patient 2
|
Patient 3
|
Patient 4
|
Number of nuclei per organoid
|
176
|
291
|
154
|
220
|
Length of nuclei displacement
(Mean (µm) ± SEM)
|
14.91 ± 0.62
|
7.93 ± 0.29
|
9.34 ± 0.27
|
5.55 ± 0.17
|
Next, the signal allowing to follow the displacement of the matrix is reported by the tracking of the fluorescent beads dispersed within the Matrigel® between the time-points acquired every 20 minutes over the three hours of acquisition. It is used to perform image correlation, and more precisely DVC, using the VicVolume software. DVC is an experimental technique based on the use of two volumetric images (3D), one in a reference state, and the other in a deformed state. The principle of DVC is to slice the reference volumetric image using a grid to obtain sub-volumes (composed of voxels). Then, the correlation is performed to search individually each of the sub-volumes obtained from the reference image in the deformed image. The result is the displacement field of the material under study. According to our observations, mitosis is completed within one hour. In our analysis, we used as initial time (T0) the initiation of the mitosis (prophase) and performed the correlation between the four images taken every 20 minutes up to T0+60min once the mitosis is completed.
Dense volume correlation shows that the extracellular matrix undergoes global displacements, in each of the three spatial directions (figure 4b). However, unlike nuclei displacements, matrix displacement is not uniform. In line with our hypothesis, this suggests that specific localized biological events may affect the matrix differently, and in particular cell divisions and their orientation. Indeed, the colour map used, where green corresponds to zero displacement, clearly shows that the matrix is stressed as a whole, but not homogeneously. For example, from the first column onwards, we can see the appearance of a displacement along the x axis (“u displacements”), which propagates over time and gains in amplitude. In addition, we observed that nuclei displacements and matrix displacements do not occur at the same scale. Indeed, nuclei exhibit displacements of the order of ten micrometers (Table 1), while matrix displacements occur over greater distances. Indeed, as seen in figure 4b, the matrix undergoes a global displacement of up to several tens or even hundreds of micrometers around the organoid.
We postulated that local displacements of the matrix could correspond to cell division events and hypothesized that monitoring the 3D ECM displacements could be used as a read out for mitoses impact onto the matrix.
Symmetric and asymmetric divisions impact the matrix differently
Penetrating the volume correlation along the z axis, we found that the presence of cell division generated a local displacement of the matrix, as exemplified in figure 5a. Indeed, on the right-hand side of the images, the green colour indicates a matrix displacement of 0 µm (so no displacement), while on the left-hand side we observed a wave of displacement whose intensity is strongest at the level of mitosis, identified in the first photo, and propagates throughout the (x,y) plane.
Having established that cell divisions participate in local ECM movements, we set out to determine the role of mitotic spindle orientation in these displacements. To solely investigate the impact of independent mitosis on matrix displacements and not of combined mitosis events, we only studied cases where mitosis was isolated, in both time and space. Here, 8 divisions of each type, observed in 5 organoids from 2 patients, were studied using these criteria. For each mitosis, matrix displacements in each of the three directions were measured at the cell-ECM frontier, and sorted relative to the two others under the labels of “longest displacement”, “intermediate displacement” and “smallest displacement” with values between 0 (i.e. no displacement) and 10 µm. For asymmetric and symmetric divisions, the longest displacement is significantly greater than the intermediate one (paired t test; asymmetric: p-value = 0.0146; symmetric: p-value = 0.0152). The intermediate displacement is significantly greater than the smallest one only for symmetric divisions (paired t test; asymmetric: p-value = 0.1109; symmetric: p-value = 0.0474) as shown in figure 5b.
Only 1 out of 8 asymmetric cell divisions causes the matrix to stain in the three directions, 2 cause a displacement in two directions while 5 cause a movement in the matrix in only one direction. In this case, the axis of matrix displacement corresponds to the axis of mitotic spindle orientation in the same plane. The longest displacement observed is an average of 4.08 ± 1.90 µm (mean ± SEM) for this type of mitosis. We can see that displacements in the other two directions are much smaller, averaging 1.11 ± 1.39 µm (mean ± SEM) for the intermediate displacement and 0.16 ± 0.28 µm (mean ± SEM) for the smallest displacement (figure 5b).
Symmetric cell divisions, on the other hand, always impact the matrix in several directions, either in two (3 out of 8 symmetric mitosis) or three directions (5 out of 8). In fact, the mean of the longest displacements is 5.44µm ± 2.31 µm (mean ± SEM), slightly greater than for asymmetric divisions, but without any significant difference. On the other hand, the mean of the intermediate displacement is 3.76 ± 2.21 µm (mean ± SEM). Similarly, the smallest displacement is 1.7 ± 1.48 µm (mean ± SEM) (figure 4b).
From a qualitative standpoint, we can conclude that asymmetric divisions involve a rather uniaxial displacement, whereas symmetric mitoses cause a multiaxial one. In addition, from a quantitative standpoint, the displacements induced in the case of symmetric division are systematically greater. Considering the longest displacement, there is no significant difference between the one observed in symmetric divisions and the one in asymmetric division (respectively of 5.44µm ± 2.31 and 4.08 ± 1.90 µm, unpaired t test, p-value = 0.3421). Nevertheless, considering the intermediate displacement, we can note that it is 3 times greater in the case of symmetric divisions than the one observed in asymmetric divisions (3.76 ± 2.21 and 1.11 ± 1.39 µm, unpaired t test, p-value = 0.0373). At last, we observed a factor of 10 between the smallest displacements of symmetric and asymmetric cell divisions (1.7 ± 1.48 and 0.16 ± 0.28 µm, unpaired t test, p-value = 0.0453).