Tumorigenic mesenchymal clusters are less sensitive to moderate osmotic stresses due to low amounts of junctional E-cadherin

The migration of tumorigenic cells is a critical step for metastatic breast cancer progression. Although the role of the extracellular matrix in breast cancer cell migration has been extensively described, the effect of osmotic stress on the migration of tumor breast cohorts remains unclear. Most of our understanding on the effect of osmotic stresses on cell migration comes from studies at the level of the single cell in isolation and does not take cell–cell interactions into account. Here, we study the impact of moderate osmotic stress on the migration of cell clusters composed of either non-tumorigenic or tumorigenic cells. We observe a decrease in migration distance and speed for non-tumorigenic cells but not for tumorigenic ones. To explain these differences, we investigate how osmotic stress impacts the mechanical properties of cell clusters and affects their volumes. Our findings show that tumorigenic mesenchymal cells are less sensitive to osmotic stress than non-tumorigenic cells and suggest that this difference is associated with a lower expression of E-cadherin. Using EGTA treatments, we confirm that the establishment of cell–cell adhesive interactions is a key component of the behavior of cell clusters in response to osmotic stress. This study provides evidence on the low sensitivity of mesenchymal tumorigenic clusters to moderate osmotic stress and highlights the importance of cadherin-based junctions in the response to osmotic stress.

www.nature.com/scientificreports/ of cancer cells, which enables cancer cells to invade surrounding tissue, decreases when a low osmotic pressure is applied (~ 1 kPa) 1 . In addition, osmotic pressure has an impact on stem cell fate, suggesting that changes in cell volume due to osmotic pressure alter physiological parameters like stem-cell differentiation 35 . Taken together, these observations have led to the recognition of osmotic pressure as an important parameter in physiological and pathological situations and demonstrate a clear link among cell behavior, cancer and osmotic pressure. However, most previous work used levels of osmotic stress 33,34,[45][46][47] far from the physiological and pathological ranges 14,[48][49][50][51] . Thus, the mechanism by which osmotic stress modulates cell migration and the precise impact it has on tumorigenic cells remains poorly understood. Further, most previous studies of the role of osmotic stress were performed at the single-cell level 1, 35 and do not take into account the modulation of cell-cell adhesions among tumor cells. Cell-cell interactions are crucial for tissue integrity, collective cell migration and cancer propagation.
In this paper, we determine the effect of moderate osmotic stresses on the migration of minimal cell clusters and investigate the role of cell-cell junctions. To apply hyperosmotic stress, we add 300-Da polyethylene glycol (PEG 300) directly to the medium. We demonstrate that when non-tumorigenic epithelial cells are cultured under osmotic stress, not only the cell volume but also cell motility decrease. By contrast, this behavior is not observed with tumorigenic mesenchymal cells, which are shown to be less affected by osmotic pressure. By inhibiting cell-cell adhesions in non-tumorigenic cells we show that they become less affected by osmotic stresses, similar to what is observed in tumorigenic cells which have lower cell-cell adhesions. These observations reveal a surprising and previously unidentified role of cell-cell adhesions in the regulation of cell volume and migration in response to variations of osmotic pressure.

Osmotic stress modulates migration distance and speed of non-tumorigenic epithelial cells but not tumorigenic mesenchymal breast cells.
To investigate whether osmotic pressure can affect the migration of breast cells, we used two different cell lines: a non-tumorigenic epithelial breast line (MCF-10A, Fig. 1A-C) and a tumorigenic mesenchymal breast line (MDA-MB-231, Fig. 1D-F). Both cell lines were cultivated on glass substrates coated with fibronectin (FN). Fibronectin was selected as a model extracellular matrix (ECM) protein because all cell types used in our study show common integrins, such as α5β1 that bind to fibronectin. After 8 h in culture with fresh media, we add three different concentrations of PEG 300 (Supplementary Fig. S1A-B) into the culture medium to exert well-controlled moderate osmotic stresses on cell cultures ranging from ~ 128 to ~ 231 kPa, as observed in physiological situations [48][49][50][51] . The motility of the different cell lines in response to osmotic stresses is then observed with time-lapse microscopy. For each cell line, experiments are started with an initial acquisition of 8 h under isotonic conditions (culture medium), representing the control. Then, a PEG solution of 1%, 1.5% or 2.5% (v/v) is added to the culture medium and an additional time-lapse acquisition of 8 h is collected (Supplementary Movies S1-S6). We estimated the viability of cells treated with PEG and do not observe that moderate osmotic pressures impact cell viability ( Supplementary Fig. S2). To avoid any artefacts and to ensure the lowest variability in our analysis, the migration distance and speed determined for each PEG concentration are normalized by the values obtained from the control experiments performed on the same culture. This normalization enables us to determine the variation of migration distance and speed from the control for each PEG concentration. Interestingly, we find that the maximum distance from the origin (Fig. 1B) and the average migration speed (Fig. 1C) of MCF-10A cells decreases significantly in response to moderate osmotic stress. The normalized distance and average speed decrease monotonically with increasing PEG concentration, reaching values of 9.84 ± 1.44 µm (n = 13) and 0.192 ± 0.026 µm/min (n = 7) respectively at a concentration of 2.5% (v/v). We observe a slight effect of PEG concentration between PEG 1% and PEG 1.5-2.5% but we don't observe an effect of PEG concentration on maximum migrating distance suggesting that MCF-10A are sensitive to very low osmotic stress (130 kPa). Surprisingly, our findings demonstrate that the migration parameters of MDA-MB-231 cells are not affected by osmotic stresses for PEG concentrations ranging from 1% to 2.5% (v/v) (Fig. 1E, F). Interestingly, the differences in migratory speed (Fig. 1G) and distance (Fig. 1H) are statistically different at 2.5% (v/v) PEG for the two cell lines. These results demonstrate that tumorigenic mesenchymal breast cells are more resistant to osmotic stress changes than non-tumorigenic breast epithelial cells.

The rigidity of breast cell lines is not affected by osmotic stresses.
To help understand the differences in migration among non-tumorigenic (MCF10A) and tumorigenic breast cells (MDA-MB-231) in response to moderate osmotic stress, we measure the change in cell stiffness of the two cell lines in response to osmotic stress at 1.5% and 2.5% (v/v) using optical magnetic twisting cytometry (OMTC) 52 ( Fig. 2A,B). Indeed, the cellular stiffness or rigidity, which is usually expressed as an elastic modulus in Pascals, is an important cellular characteristic that characterize the tridimensional (3D) reorganization of the cell cytoskeleton in response to physico-chemical changes of the cell microenvironment. Previous works have shown that the cell stiffness matches the substrate rigidity 53,54 , cell stiffness is involved in stem cell behavior 55 and a low cell stiffness promotes cell migration 56 . Under osmotic compression, water efflux can increase leading to the buckling of cytoskeletal filaments, thereby reducing the cell stiffness and consequently, this cell rigidification due to water efflux with osmotic pressure can have an impact on cell migration 35,52,57,58 . In addition, some of us have shown that cell stiffness can increase with osmotic pressure in different cell types, based on cell volume changes through water efflux 35 and can induce glass transition because compression dramatically slows intracellular relaxation processes 52 . Based on these findings, we examined whether the cellular stiffness of normal and tumorigenic tissue can change with the osmotic pressure with regard to their migratory behavior. Similar to the migration parameters, we normalize the cell stiffness in response to 1.5% and 2.5% (v/v) PEG with the baseline stiffness measured on the same cells in the absence of PEG. We observe that the individual stiffness of MCF-10A (Fig. 2C) and MDA-MB-231 (Fig. 2D)  Moderate osmotic stress leads to a reduction in cell volume, but tumorigenic breast mesenchymal cells are less sensitive to osmotic stress. When isolated cells are subjected to an external osmotic pressure, their volume decreases due to water efflux, and their stiffness increases 35 . The internal osmotic pressure of a cell is regulated by the concentration of ions and small proteins and must adapt to changes in the external osmotic pressure to eliminate any gradient in pressure across the cell membrane. This can result in a significant change in subcellular macromolecular density. To further investigate whether moderate osmotic stresses applied to cell monolayers can lead to cellular volume changes, we measure by confocal microscopy the volume of minimal tissues composed of either MCF-10A or MDA-MB-231 cells. To standardize our measurements, we  Using these two other cell lines, our results show that osmotic stress leads to large volume decrease for all the cell lines expressing E-cadherin: non-malignant MCF-10A, malignant MCF-7 and MDCK cells which are characterized by a stronger cell-cell interactions (Fig. 4C, Supplementary Fig. S3). However, we observed a lower volume loss for the MDA-MB-231 tumorigenic breast mesenchymal cells, characterized by a low expression of E-cadherin (Fig. 4C).
To confirm our findings, we quantify the maximum normalized distance (Fig. 4D) and speed (Fig. 4E) of MCF-7 and MDCK cells treated with PEG 2.5% to compare their migrating behavior with MCF-10A (Fig. 1B,C) and MDA-MB-231 (Fig. 1E,F). Our results show that the migratory behavior of cell types establishing E-cadherin cell-cell interactions (MCF-10A, MCF-7 and MDCK) is sensitive to moderate osmotic pressure, whereas MDA-MB-231 cells characterized by a low E-cadherin interactions is less sensitive to osmotic pressure. To further understand these results, we determine the contrast of E-cadherin between cell-cell junctions and the cytoplasm by immunostaining experiments (Fig. 4F), as determined in previous studies 74    www.nature.com/scientificreports/ control, and then add EGTA solution at 5 mM and acquire an additional 8 h of time-lapse images. The migration speed increases for MCF-10A (Fig. 5A) and MDCK cells (Fig. 5C) after disruption of cell-cell adhesions by the EGTA treatment, whereas the migration speed of MDA-MB-231 cells (Fig. 5B) that contain lower amounts of cadherin at cell-cell junctions (Fig. 4E) is not affected by EGTA treatment. The effect of EGTA treatment on cadherin disruption is confirmed by immunostaining experiments (Fig. 5D). MCF-10A and MDCK tissues treated with EGTA are characterized by a very low amount of E-cadherins between cells (Fig. 5E), the amount of E-cadherins in EGTA-treated MDA-MB-231 tissues remain statistically unaffected. To strengthen our results, we add PEG at 2.5% (v/v) to EGTA-treated tissues. We observe that PEG treatment does not affect the migration speed of all three cell types (Fig. 5A-C), confirming that cadherin-based junctions are a key component of the migratory behavior of epithelial cell assemblies under moderate osmotic stress.

Discussion
Many studies have shown that osmotic pressure can impact cell volume and affect cell structure 14,35,36 . Osmotic stress can lead to a significant reduction of the cell volume due to water efflux; this results in a corresponding increase in molecular intracellular crowding 14,77 . However, most of the previous work focused on isolated cells and used high levels of osmotic stress, both of which are far from physiological conditions. Here, we study how moderate osmotic stresses can affect the migration of breast cell lines by using four different cell lines: a non-tumorigenic epithelial breast line (MCF-10A) and a malignant breast line with lack of E-cadherin (MDA-MB-231), a malignant breast line with positive E-cadherin expression (MCF-7) and a strong adhesion model of epithelial cells (MDCK). By performing OMTC experiments, we find that the mechanical properties of non-tumorigenic and tumorigenic breast cells remain constant under moderate osmotic pressure. That means that cell stiffness could not be considered as an important player in the modulation of the migratory behavior under osmotic stress.
To demonstrate the role of cell-cell adhesions, we compare the effect of osmotic pressure on cell volumes between single cells and cell clusters. Single cells are less affected than cell clusters, suggesting that cell-cell interactions have a key influence on modulation of cell volume in response to osmotic stress. To validate this hypothesis, we perform experiments with EGTA to inhibit cell-cell adhesion in cell clusters. We find that the cell volume is less affected by osmotic stress in EGTA-treated clusters, regardless of the cell type. This suggests that cell-cell adhesions have a major role in the cell-volume regulation mechanism and hence the sensitivity of epithelial cells to osmotic stress.
Our results suggest that tension across E-cadherin would change in response to osmotic conditions. Indeed, an essential aspect of epithelial homeostasis is attributed to maintenance of cell-cell junctions. In addition, changes in cell-cell junction are one of the hallmarks of epithelial-to-mesenchymal (EMT) transition which is an important regulator of cell migration. It could be therefore very interesting to use FRET-force biosensors for E-cadherin to estimate the tension in cell-cell junctions in response to the modulation of the osmotic conditions.
To conclude, this study helps unravel the variability in osmoregulatory mechanisms among different cell lines and highlights the central role of cell-cell interactions. We envision that a better understanding of the osmoregulation of cancer cells in different in vitro conditions might help to better target potential pharmacological agents.

Material and methods
Stamp fabrication. Stamps for micropatterning were fabricated by soft lithography using polydimethylsiloxane (PDMS; Sylgard 184 Silicone Elastomer Kit; Dow Corning). Briefly, a thin layer of SU8-3010 (Micro-Chem) was spin coated onto the surface of a silicon wafer. After baking, a photomask was placed on top of the wafer for UV exposure. Propylene glycol methyl ether acetate (PGMEA) (Sigma-Aldrich, 537,543) was then used to remove the unexposed SU8 from the wafer. After baking, the wafer was placed in a petri-dish and served as a mold for downstream PDMS fabrication. PDMS was well mixed with curing agent at a 10:1 ratio, degassed, poured onto the wafer previously passivated for 30 min with fluorosilane (tridecafluoro-1,1,2,2-tetrahydrooctyl-1-trichlorosilane). Finally, the wafer was placed in a 65 °C oven for at least 1.5 h. The PDMS was then cut and carefully peeled off from the mold. Cell volume. Stained cells were observed by using a × 63/1.2-NA water immersion lens and tissue by using a × 20 dry lens on a Leica TSC SP5. Optical cross-sections were recorded at 0.15-μm z-axis intervals to show intracellular, nuclear, and cortical fluorescence. By using theoretical point spread function, a stack of gray-level images (8 bits) were subjected to deconvolution before 3D visualization. The 3D visualization was carried out by using ImageJ software. The volume was calculated by counting voxel number, after thresholding the stack. Confocal measurements were previously compared with AFM data and the results from the two techniques agreed.
Osmotic stress. Hyperosmotic stresses were applied by adding PEG 300 to isotonic culture medium. The actual osmotic pressure applied to cells was calculated by adding the osmotic pressure due to PEG to that of isotonic solution (330 mOsm/kg) and was further confirmed through a selective measurement by using a microosmometer (model 3300; Advanced Instruments, Inc.). Before doing experiment, cells were allowed to equilibrate in PEG solution for 10 min at 37 °C and 5% CO 2 . The cell size decreases within 30 s and maintains the small volume for hours.

Confocal microscopy imaging.
To observe cell-cell adhesions, cells were seeded onto cell culture membrane assemblies and cultured in a cell incubator ( Stiffness measurements. The cell mechanical properties were probed by using OMTC, which is a highthroughput tool for measuring adherent cell mechanics with high temporal and spatial resolution. Measurements were performed at 37 °C. In brief, functionalized ferrimagnetic beads (4.5 µm) coated with PLL (4 kDa) were incubated with cells for 20 min in the incubator. For stiffness measurements, beads were first magnetized horizontally by a large magnetic field. A much weaker magnetic field (oscillating at 0.75 Hz) was then applied vertically, which applied a sinusoidal torque to the beads. Motion of the beads was optically recorded. The ratio between the torque and bead motion thus defined an apparent cell stiffness (Pa/nm). A series of geometric factors, based on finite element models that take into account the cell thickness and bead-cell coupling, can be used to convert the apparent stiffness into shear modulus of the cell.

E-cadherin contrast calculation.
On the E-cadherin immunofluorescence images, an average intensity was measured at the cell−cell junction (i jct ) and in the cytoplasm (i cyt ) using ImageJ. These values where averaged over 20-25 cells (I jct = ⟨i jct ⟩ and I cyt = ⟨i cyt ⟩). We then defined J, the junction to cytoplasm ratio of E-cadherin intensity : J = I jct /I cyt .