Cell micropatterning is becoming a powerful tool for controlling the cellular microenvironment and shape and for studying the effects of physical cues on cellular behavior. Cell micropatterning offers many advantages over growing cells in flasks or in culture dishes according to standard cell culture methods. Micropatterning cells, so as to force them to fit in a certain geometric pattern and size, enables to analyze cellular behavior and to facilitate reliable and reproducible statistical analyses in small scales (~10-50 cells). The objective of this work was to establish a cell micropatterning method for patterning adhesion proteins like FN in predefined geometries on substrates and to analyze the cytoskeleton of single cells grown on these micropatterns.
4.1 Stencil patterning as the protein patterning method of choice
Two methods were testes for patterning FN on substrates: µCP on glass substrates and stencil patterning on silicone elastomer substrates. Stencil patterning showed a significantly better performance than µCP in fabricating accurately shaped micropatterns with a uniform and homogeneous distribution of FN, which constituted the basis for subsequent experiments in the scope of this work and for future experiments.
4.2 Small micropatterns are suitable for single cell analyses
In the next step, the behavior of cells on micropatterned silicone elastomer substrates was analyzed. On average more single cells were found on micropatterns with an average cell spreading area of 700 µm² than on larger micropatterns. This is due to the fact that an attached cell tends to migrate within a large micropattern without fully adapting to its shape and size (Huang et al., 2005; Théry, et al., 2006; Tseng et al., 2012). In the worst case, with time the cell divides, resulting in multiple cells per micropattern. For quantitative analyses of the cytoskeleton of single cells micropatterns with an average cell spreading area of 700 µm² are to be preferred for the cell lines used within this work.
4.3 KIFs show a perinuclear distribution in normalized cells
For quantitatively assessing the distribution of KIFs, heat maps were generated in a number of cases. Since in most cases only a small number of cells of exactly the same size and shape existed the distribution was assessed based of individual cell images. Analyses of the KIF network showed a perinuclear distribution and a decrease toward the corners, irrespective of cell shape and size. Heat maps of selected cell shapes confirmed the observation that the KIF localization was highest around the nucleus and lowest at the edges. These findings indicate that KIFs do not respond to geometrical cues. Similar results were shown for vimentin intermediate filaments (Shabbir et al., 2014). The VIF network of NIH/3T3 cells patterned on differently shaped micropatterns (700 µm²) showed a mostly perinuclear distribution, a decrease toward the cell periphery and an absence in regions marked by sharp edges.
However, a large-scale analysis with more normalized single cells is essential to confirm these first results and to produce a robust statistic, which will provide a more comprehensive view of the biological role of geometrical cues on the behavior of KIFs.
Irrespective of the meaning of geometrical influences on the KIF network, normalized cells enable to build density maps of protein distribution and hence to construct a reference cell, which can serve as a template for identifying cytoskeletal alternations in mutant cells to decipher the role of different molecular components in regulating the cytoskeleton. Future analyses with micropatterned substrates should explore the interplay of KIFs with microfilaments and microtubules in single, dual and triple filament depleted cells.
Even smallest changes in the keratin motion and turnover in response to regulatory cues can now be quantitatively assessed in great detail with less effort than before (Moch et al., 2013).
Furthermore, the use of micropatterns can be an efficient setup to analyze the role of cell-ECM interactions. Take, for example, Y-shaped micropatterns that show ECM-rich regions and regions devoid of ECM, which providing stringent and stationary boundary conditions to cells. Previous studies showed that keratin filament nucleation starts in the cell periphery in close vicinity to focal adhesions (Windoffer et al., 2006), which provide structural links between the ECM and the cytoskeleton. Therefore, it would be interesting to analyze the distribution of focal adhesions in cells spread on Y-shaped micropatterns in order to find out whether the distribution of focal adhesions is restricted to the cell periphery that co-localizes with the edges of the Y-shaped micropattern or if it focal adhesions are also localized in peripheral regions that are in no contact with the ECM of the Y-shaped micropattern. Since focal adhesions connect the ECM to the cytoskeleton, localization is only to be expected in the cell periphery that co-localizes with the ECM-rich regions of the Y-shaped micropattern. Thus, the interplay between focal adhesions, keratin filament nucleation sites and the keratin cycle dynamics can be explored.
4.4 Challenging tasks for future analyses
For analyzing single keratin filaments and keratin bundles, a high resolution and an appropriate spreading of the KIF network have to be achieved. For this purpose, cells have to reach a minimum size in diameter that would allow an appropriate spreading of the cytoskeleton and hence an improved discrimination of single filaments within the meshwork. This means that cells have to be cultured on large micropatterns. This will be a challenging task as the current work demonstrated that an increase in micropattern size led to an increase in the number of cells per micropattern or to incomplete cell spreading of single cells. Epithelial cells tend to clump together since adhesion between cells through cadherins is favored over adhesion to fibronectin through integrins, resulting in multiple cells occupying a single micropattern. In order to enable a high yield of single cells attached and fully spread on large micropatterns, new approaches have to be explored and implemented. The focus is to overcome a few major obstacles. First, special attention has to be given to isolating single cells during the first steps of dispersing the initial cell suspension. Secondly, single cells tend to migrate within large micropatterns without reaching a steady-state position. Furthermore, most single cells on large micropatterns do not spread completely in an appropriate time before cell division. A possible solution to these problems is to design and to test new patterns, which would guarantee the attachment of a single cell and a stable positioning of cells to prevent cell migration. Many previous studies already encountered this problem (Théry et al., 2006; Huang et al., 2005; Tseng et al., 2012). E.g. Tseng et al. (2012) explored the influence of different micropattern configurations on the degree of stability of intercellular junction positioning of cell doublets. Among other findings, the study demonstrated that migration of cell doublets on particular micropattern configurations was impaired. Interestingly, micropattern configurations, which consist of different combinations of adhesive and non-adhesive areas, seem to be promising for controlling cell positioning. Additionally, these micropatterns offer less area for cell attachment thus minimizing the probability of the attachment of multiple cells. Designing micropatterns with different configurations of Y-shaped and X-shaped micropatterns ought to be seriously considered if cells with a substantial diameter are needed.
Although bowtie shaped patterns were designed for the analyses of cell doublets they were not tested as it lay beyond the scope of the present work. Nonetheless, several cell doublets were observed and a few cell doublets were documented on H-shaped micropatterns of the CYTOOchips™ (Fig. 17). It was observed that H-shaped micropatterns provided a highly stable conﬁgurations in which each cell was positioned on each side of the gap.
In conclusion, as a result of this work a setup for cell micropatterning has been successfully established. It was determined that the cell lines used in the scope of this work reached a steady-state position on micropatterns with a small cell spreading area and a micropattern configuration consisting of different combinations of adhesive and non-adhesive areas. First results indicate that the KIF network does not respond to local geometrical cues since the distribution of the keratin filaments looks strikingly similar on different micropatterns. A more extensive analysis of the keratin filament distribution is necessary for a robust statistic and for building a reference cell. The results of this work paves the way for analyzing the interplay of the KIF network with other cellular components under controlled microenvironmental parameters.