Published On: Thu, Jan 24th, 2019

E-cadherin, actin, microtubules and FAK dominate different spheroid formation phases and important elements of tissue integrity [RESEARCH ARTICLE]


Tissue integrity maintains the functionality of a tissue through a tight regulation of cell–cell and cell–ECM adhesion as well as growth. These cellular processes have been intensively studied in cells grown on rigid, two-dimensional surfaces. Many differences between two-dimensional and more physiological three-dimensional cell cultures have been documented in the literature (Pampaloni et al., 2007, 2010). Cells grown in a monolayer culture form extensive numbers of focal adhesions, which can have antagonising effects on the formation of cell–cell adhesion sites and cell migration (Cukierman et al., 2001). Thus, the molecular mechanisms are strongly influenced by non-physiological culture conditions.

The mechanisms that facilitate adhesion in three-dimensional cell aggregates as well as in tissues are largely unknown. Hence, we studied the contributions of different adhesion-associated proteins in a spheroid formation assay.

The aggregation of cells into multicellular structures in a non-adhesive environment has been observed for various cell types. Several studies have described the kinetics of spheroid self-assembly. Neelamegham and colleagues found that the aggregation mechanism is primarily diffusion-controlled, i.e. cells almost always adhere upon collision (Neelamegham et al., 1997). Conflictingly, Enmon et al. proposed a reaction-controlled aggregation. Here, adhesion does not necessarily occur when two cells collide (Enmon et al., 2001). We employed our recently developed three-dimensional computational model for spheroid formation (Garg et al., 2015). Fitting the model to our experimental data revealed a binding probability of at most 73% for each condition. This suggests that the spheroid formation process is reaction- rather than diffusion-controlled. Hence, aggregation is an active process, in which cells get in contact upon collision, but do not necessarily form connections.

Previous studies have suggested a general mechanism for spheroid formation (Lin and Chang, 2008; Enmon et al., 2001). We successfully identified the three phases of spheroid formation. They comprise an initial aggregation, compaction and spheroid growth. However, phase transitions do not occur abruptly. The characterisation of a phase means that a certain process dominates this phase. For example, cell division takes place at any time during the formation process (see Fig. 1E, T47D 48 h, magnified section), but it does not dominate during the aggregation and compaction phases. With our image-based spheroid formation assay, we showed that spheroid formation differs substantially amongst the three investigated cell lines: 4T1 aggregated the fastest, followed by HC11, and T47D aggregated by far the slowest. Since no obvious explanation for the observed differences can be based on the invasive phenotype of the cells, we further investigated the role of different proteins that are associated with adhesion during spheroid formation. Therefore, we inhibited E-cadherin, the actin and microtubule networks as well as FAK.

Cadherins are essential mediators for cell–cell contacts and crucial for a successful spheroid formation (Ivascu and Kubbies, 2007; Lin et al., 2006; Saias et al., 2015). Like previous studies, our results showed that spheroid formation was impaired when E-cadherin was blocked. Moreover, the computational model revealed that the inhibition of E-cadherin has no influence on the binding probability of the cells. This indicates that the presence of E-cadherin does not increase the chance of forming cell contacts when cells collide. Instead, other adhesion-facilitating proteins may be crucial for this process. A previous study has found that desmosomal proteins are important for spheroid formation (Saias et al., 2015). To which extent they influence the binding probability of cells remains elusive. In two out of the three cell lines (HC11 and 4T1 cells), the unbinding probability was strongly increased when E-cadherin was blocked. This suggests that E-cadherin contributes to stabilising cell contacts. This is further supported by the occurrence of strong fluctuations of the projected area during spheroid formation, which are caused by fragile connections between the cells. This result provides a yet unknown aspect of E-cadherin in the three-dimensional cellular context.

Although the actin cytoskeleton is crucial in spheroid formation details have never been investigated. Consistent with previous studies (Tzanakakis et al., 2001; Yoshii et al., 2011), we showed that the actin cytoskeleton is indispensable for spheroid formation. Cytochalasin D disrupted the F-actin network and affected the self-assembly of the cells. Cells did not form proper spheroids, but clusters with loose cell contacts.

We further analysed the dynamics of spheroid formation of cytochalasin D-treated cells computationally. The binding probability was mainly unaffected, suggesting that actin is less important for cells to form a connection. The increased unbinding probability in all cell lines suggests that cells tend to lose contacts when the actin cytoskeleton is disrupted. Hence, a role of the actin cytoskeleton during spheroid formation is to reinforce contacts.

Upon cytochalasin D treatment of 4T1 cells, we observed a continuous increase in spheroid size and also an increase in cell volume. The actin cytoskeleton is involved in cell volume regulation (Mills et al., 1994) and the disruption of actin filaments abolishes this process (Blase et al., 2009). Although HC11 cells were treated at the same drug concentration, no cell swelling was observed, showing that cell swelling is cell type specific (Pedersen et al., 2001).

The role of microtubules in cell aggregation has been investigated only rarely and no quantitative analysis of spheroid formation dynamics has been performed. Both effects, an inhibition of spheroid formation and no impact on spheroid formation upon microtubule impairment, have been described (Tzanakakis et al., 2001; Yoshii et al., 2011). We showed that nocodazole-treated cells sustain the ability to form spheroids. Our experiments provided detailed and temporally resolved results, which show that the formation is slowed down upon nocodazole treatment in all cell lines.

The growth phase in 4T1 spheroids was inhibited by a depolymerisation of microtubules in the long term. Further, the cell viability was markedly reduced.

According to the computational model, the binding and unbinding probabilities were differently affected in the three cell lines upon nocodazole treatment. In HC11 and 4T1 cells, the binding probability was not affected, but the unbinding probability was increased. This indicates that microtubules are important for the stabilisation of already formed contacts. Microtubule-mediated vesicle transport is necessary to translocate adhesion molecules to the plasma membrane (Mary et al., 2002). We conclude that upon microtubule destabilisation, cells fail to accumulate adhesion molecules at the plasma membrane to reinforce connections. This leads to the decreased spheroid formation speed.

In T47D cells, the aggregation speed was similarly impaired compared to the other cell lines. The computational model revealed that the unbinding probability was not affected. T47D cells exhibit strong cell–cell contacts and strongly express adhesion molecules at the surface (Holliday and Speirs, 2011; Ivascu and Kubbies, 2007). Consequently, transport of adhesion-associated proteins to the plasma membrane might not be that important in these cells. Moreover, the binding probability was increased when microtubules were depolymerised. This strongly suggests that single cells assemble and form contacts in one plane at the bottom of the wells. These strong cell contacts restrict the cells ability to move on top of each other against gravity. Consequently, cells are unable to arrange into the third dimension (i.e. into a proper, spherical spheroid). This agrees well with the increased projected area of nocodazole-treated cells during spheroid formation. The correlation of binding probability and microtubule depolymerisation might be due to the influence of microtubules on the activity of E-cadherin. A recent study has unveiled that a depolymerisation of microtubules activates E-cadherin by mediating the dephosphorylation of p120-catenin (Maiden et al., 2016). From this, we draw the conclusion that an intensified contact formation negatively affects the rearrangement of cells to move on top of each other. It further shows that microtubules are important to maintain a balance between formation and the rearrangement of contacts.

A recent study has described the importance of FAK phosphorylation at Y397 during spheroid formation of human dental follicle cells. Spheroid formation has been induced by culturing the cells under serum-free medium culture conditions on dishes coated with poly-L-lysine (Beck et al., 2014). Thus, a large portion of these cellular aggregates remained attached to the surface of the dish, which understandably results in FAK phosphorylation events caused by integrin adhesion to the coated surface. Our approach circumvents the attachment to any non-natural surface and provides controlled and true three-dimensional culture conditions.

We observed that the spheroid formation process itself is not heavily affected by FAK inhibition in all cell lines. However, our results showed cell type specific effects. The adhesion to the substratum activates FAK (Mitra et al., 2005). An increased expression of FAK and its activity is associated with tumours (McLean et al., 2005). We found that the basal protein level as well as the tyrosine phosphorylation at position 397 varies between the cell types. In HC11 and T47D cells, FAK phosphorylation decreased in spheroids compared to monolayer cultures, which indicates that integrin-mediated signalling via FAK is less important in non-invasive cells in three-dimensional aggregates. In line with this, HC11 spheroid formation was also found to be the least impaired. Conversely, the highly invasive 4T1 cells showed a strong effect when FAK was inhibited. These cells showed high FAK phosphorylation in spheroids. Consequently, FAK activity correlates with the invasiveness of cells and influences growth only if a high amount of active FAK is present.

FAK is essential for tumour growth (Wendt et al., 2011; Tanjoni et al., 2010; Tancioni et al., 2015). The influence of FAK on cell proliferation can occur in an integrin-dependent manner but also independent from integrin-ECM association (Lim, 2013). Integrin-independent effects on proliferation exhibit a nuclear localisation of FAK. In the spheroids we did not observe a nuclear translocation of FAK. Thus, we suggest that the observed effects on spheroid growth originate from integrin-mediated signalling.

4T1 spheroid growth is reduced upon FAK inhibition (Tanjoni et al., 2010), which is consistent with our observations. The strong reduction in growth upon FAK inhibition in 4T1 spheroids might be due to the strong FAK activity in these cells as indicated by a high phosphorylation at Y397 in untreated spheroids. We could not observe an altered cell viability in 4T1 monolayer cultures, which is consistent with previous findings (Tanjoni et al., 2010). This demonstrates that cells can show opposite effects when cultured in a two-dimensional monolayer compared to three-dimensional culture.

The computational model fitted to our experimental data showed that pharmacological inhibition of pFAKY397 influenced the binding as well as the unbinding probability in all three cell lines differently, underpinning the assumption that cell aggregation is differently influenced by FAK in different cell types. This can have two possible causes: (1) FAK activity affects cytoskeleton remodelling (Mitra et al., 2005), or (2) FAK modulates cell–cell adhesion (Koenig et al., 2006; Playford et al., 2008; Yano et al., 2004).

The aggregation of cells is mediated by cell–cell adhesion and may be influenced by cell–ECM adhesion. The spheroids formed from the three cell lines expressed ECM proteins such as collagen I and fibronectin, which is in line with previous studies (Bjerkvig et al., 1989). In addition, we showed that the expression of ECM proteins occurs very early in the forming spheroids (during the first 24 h). The severity of this effect may depend on the general ability of the cell to associate with the ECM. We showed that the capability of the cells to attach to ECM differs between cell lines. It further indicates that cells not only aggregate by spontaneous cell–cell attachment, but also by cells incorporating ECM into the aggregate, which may further facilitate spheroid formation.

We dissected spheroid formation into three temporally distinct phases. For these phases, we identified the importance of the four proteins E-cadherin, actin, microtubules and FAK (Fig. 6). We conclude that spheroid formation is a process that is not solely driven by cell–cell adhesion but also by the rearrangement of the cells, which might be driven by migration on the ECM.

Fig. 6.

Fig. 6.

Model showing the phase-dependent roles of intracellular components in three-dimensional cell aggregations. Spheroid formation on a non-adhesive surface is divided in the three phases: aggregation, compaction and growth. The involvement of the three intracellular components E-cadherin, actin, microtubules and FAK is indicated for each phase. The size of the capital letters indicates the importance in the according phase. The phases of spheroid formation are extrapolated to the tissue level. The tissue integrity is high when cellular rearrangement and maintenance of adhesion by reinforcement occur. Uncontrolled proliferation counteracts tissue integrity and correlates with the metastatic potential. Developmental processes in vivo, such as mammary gland development and cancer development, follow both directions of an integrated tissue.

We propose that the phases, which are present during spheroid formation, can be extrapolated to the tissue level. Tissue integrity increases if adhesion is facilitated, if the cells rearrange to their determined position or if adhesion contacts are maintained by reinforcement. Depending on the developmental needs, tissue integrity may increase or decrease. Mammary gland ductal outgrowth during puberty is directed towards an integrated tissue, whereas post-lactational regression disintegrates the tissue to accomplish remodelling processes. In cancer, increasing metastatic potential counteracts tissue integrity by destabilisation of adhesion and uncontrolled proliferation (Fig. 6).

Our previous and current studies contribute to an overall picture of cellular aggregates. A physiological understanding of cell adhesion and growth can only be achieved with three-dimensional dynamic models. Our results encourage future research on how tissues develop, how they are maintained and what causes their disruption.

About the Author

Leave a comment

XHTML: You can use these html tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>