«Der Naturwissenschaftlichen Fakultät der Friedrich-Alexander-Universität Erlangen-Nürnberg zur Erlangung des Doktorgrades Dr. rer. nat. vorgelegt ...»
To quantify this observation, we performed a 3D invasion assay, where transfected cells, lipofectamine treated cells and control cells were seeded on the surface of 3D collagen gels. The results revealed that all GFP constructs, regardless if GFP was coupled to actin (directly as GFP-actin or indirectly with GFP-LifeAct) or the EGFP construct alone, severely impaired the ability of cells to invade a 3-D collagen gel (Fig.
3.2.5, A). The lipofectamine treatment alone, without DNA, did not alter the invasion behavior (Fig. 3.2.5, A, blue) compared to control cells (Fig. 3.2.5, A, black), as both of these curves showed a p5 value of ~ 200µm invasion depth (Fig. 3.2.5, B). These data demonstrate that expression of fluorescent proteins can impair the 3-D migration behavior of cells. Note that the cells were not exposed to any fluorescent excitation during the invasion period, so there was no photo-damage, which could cause altered cellular migration behavior.
5: 3D invasion of cells transfected with fluorescently tagged proteins (A) An invasion assay as described previously was performed with transfected and non-transfected cells as well as with cells treated with lipofectamine only. 40 fields of view from two different assays were pooled and analyzed. The invasion profile reveal, that all transfected cells were less motile than the non-transfected control cells while lipofectamine treatment had no damaging influence.
(B) The invasion depth that was reached by 5% of the cells (p5 value) was determined to point out the differences in invasion. Error bars indicate SE. Detected cell number was between 3,000 and 5,000 cells.
III RESULTS AND DISCUSSION 57Transfection with fluorescently tagged proteins influences 2D stiffness and binding strength To investigate if the change in cellular migration behavior in 3D after transfection is due to changes in cell mechanics, we performed magnetic tweezer measurements to evaluate cell stiffness and binding strength of cellular focal adhesion contacts. Again control cells, cells that were only treated with lipofectamine and cells that were transfected with GFP-actin, GFP-LifeAct or EGFP plasmids were compared.
We applied a force step protocol with force steps from 0.5 nN to 10 nN and monitored the bead displacement. To evaluate if the binding site, on which the beads attach to the cell and the cytoskeleton plays a role, we used RGD coated beads as well as fibronectin coated beads.
Regardless of bead coating, cell stiffness measurements showed that GFP-actin and EGFP transfected cells were softer than control cells (Fig. 3.2.6, A and 3.2.7, A).
Lipofectamine treatment alone showed a slight, but not significant difference in cellular stiffness. Interestingly, GFP-LifeAct transfection did not result in reduced stiffness.
Regardless of the bead coating, GFP-LifeAct transfected cells showed the same stiffness level as control cells.
6: Bead disruption rate and cell stiffness for RGD coated beads. (A) Cell stiffness was derived from the bead displacement data by fitting a power law to every force step.
Displayed is the stiffness at force step 1 nN. Stiffness from transfected cells was lower than stiffness from control cells, except for GFP-LifeAct. (B) While applying forces from 1 to 10 nN, it was monitored at which force level the bead disrupted from the cell. This is a measure for the binding strength. Data revealed that the binding strength of transfected cells (green, yellow and purple curve) was lower than the binding strength of non – transfected cells (red and blue curve).
58 III RESULTS AND DISCUSSIONEvaluating the bead detachment data however revealed that all cells transfected with GFP-constructs had a reduced binding strength, regardless if transfected with GFPactin, EGFP or GFP-LifeAct. Lipofectamine treatment alone had no effect on the binding strength. Again, the coating of the beads did not play a role for the differences between transfected and non-transfected cells. However, when beads were coated with fibronectin, fewer beads could be disrupted in total, suggesting that more binding sites for fibronectin were present (Fig. 3.2.6, B and 3.2.7, B).
These data suggest that the reduced stiffness of the cells was not the reason why transfected cells invade less effective since cells transfected with GFP-LifeAct were not softer but still less invasive. The reason why GFP-LifeAct constructs did not lead to a reduced stiffness when transfected into cells might be due to the interaction of GFPLifeAct plasmids with the cytoskeleton. As mentioned above GFP-LifeAct is not coupled to the G-actin molecules that still have to form the F-actin fibers. Rather, GFPLifeAct reacts with the F-actin that was already built in the cell. Therefore GFP-LifeAct might either have less disturbing effects on the building of actin fibrils or it might stabilize the actin fibrils which were weakened from the transfection.
However, the reduced binding strength of transfected cells might be an explanation for the reduced invasion in 3D. It is known that cells attach and pull on collagen fibers when they migrate through the networks and that invasion is also reduced, when proteins of the focal adhesion complex are knocked out . Therefore, if the transfected cells are altered in their binding behavior it might change the migration behavior. This effect seems to have more impact when cells are migrating in 3D where they have to overcome the steric hindrance of the surrounding network. However, if the reduced binding strength is the reason why 3D cell migration was impaired it remains unclear why 2D migration behavior was not affected.
III RESULTS AND DISCUSSION 59Figure 3.2.
7: Bead disruption rate and cell stiffness for fibronectin coated beads. (A) Cell stiffness was derived from the bead displacement data by fitting a power law to every force step.
Displayed is the cell stiffness at force step 1 nN. Stiffness from transfected cells was again lower than stiffness from control cells, except again for GFP-LifeAct. (B) Again, while applying forces from 0.5 to 10 nN, it was monitored at which force level the bead disrupted from the cell.
For fibronectin coated beads, the data also show that transfected cells (green, yellow and purple curve) had a lower binding strength as non – transfected cells (red and blue curve). However, the differences were less clear to see, because the overall binding strength with FN coated beads is much higher due to the increased number of binding partners for fibronectin on the cell surface.
The above data demonstrate that transfection with fluorescent proteins can impair the 3D migration behavior of cells, even if measurements in 2D are not pointing towards an alteration. The appearance in 3D might be due to the fact, that 3D migration is much more complicated for a cell, because it has to overcome the additional steric hindrance of the matrix. The reason why fluorescently tagged proteins alter cellular behavior can only be speculated. Although a fluorescent cell label may eventually be found that does not cause such behavioral changes in a 3-D collagen gel, it is nonetheless an advantage if cell labeling is not required. If however it is necessary to label the cells, it should be evaluated if the transfection has influence on the behavior that will be observed. If so, these changes should be mentioned and discussed, too.
60 III RESULTS AND DISCUSSION
3.3 The co-dependence of matrices and cell types 3.3.1 Introduction – the variety of matrix and implant materials The biochemical and biophysical characteristics of the extracellular matrix, in particular ligand density, matrix pore size and fiber orientation, as well as matrix stiffness and viscosity, are currently discussed as important parameters to control cell migration [106, 107]. Widely used 2D culture systems can only supply a poor resemblance to an in vivo situation. Therefore, designing a material mimicking the extracellular is a crucial step to be able to study and understand cell migration in vivo. Engineering the ideal ECM – substitute material is a challenging task because the overall scaffold structure should possess all biochemical and biophysical properties of a real tissue but should be easy to control and be highly reproducible at the same time. Development of such scaffolds is not only important for cell culture experiments in order to understand the very basics of 3D cell migration but also for the development of sophisticated implant material for patients. Implant material should be stable and biocompatible on the one hand but biodegradable on the other hand. Moreover they should be designed in a way that cells are supported to migrate in, build original tissue as fast as possible and at the same time degrade the implant material, before immune rejection reactions begin. Ideally, the designed implant would not have to be removed from the patient‟s body in a subsequent operation.
Possible implant materials for very soft tissues are biopolymer network based materials, like collagen or fibrin. These biopolymers build fiber networks that are stable with a defined pore size, stiffness and ligand density. Collagen and fibrin have the additional advantage that they both originate in mammalian tissue, where collagen is the most abundant protein in connective tissue and fibrin plays an important role in wound stabilization and wound healing . However, it is challenging to find the optimal protein concentration, the optimal pore size, stiffness and composition of the matrix for distinct implant sites as well as provide the perfect material for healthy cells to resettle.
On the other hand these same parameters should be specifically optimized to avoid invasion of cancerous cell types into these materials.
Here we tested possible materials for breast tissue implants and show that different types of tumor cells and different types of breast cells prefer different tissues. Our findings can possibly help to develop different, patient and application specific implant
III RESULTS AND DISCUSSION 61materials. In addition, understanding how different matrices influence various breast cancer cell types will enhance our knowledge of metastasis formation.
Although semi-natural materials like collagen and fibrin have the advantage of being highly reproducible in terms of morphological and mechanical properties, they still do not provide the complex structure and interaction of various proteins from natural material, especially on a molecular level. Therefore, together with the laboratory of Molecular Medicine in Erlangen, we developed a method to using of a fully natural material – the human amnion. The amnion is the innermost membrane which covers the developing placenta and which forms a sac containing the developing fetus. It is attached via the chorion to the maternal portion of the placenta. During a pregnancy the amnion produces the amniotic fluid, which together with the amniotic membrane protects the fetus. The amnion itself consists of a monolayer of epithelial cells, which is facing the fetus, and a thin layer of basement membrane filled with Laminin 5 and collagen IV and a thick layer of connective tissue, where all types of collagen and fibroblasts are found (Fig. 3.3.1).
1: The hierarchy of an amnion. (A) The layer of epithelial cells is linked to the basement membrane via integrins. Laminin 5 connects integrins on the basal surface of epithelial cells to the type IV collagen network of the basement membrane. Anchoring ﬁbrils composed of type VII and type XV collagen link the basement membrane to the interstitial matrix where type I collagen, type III collagen and fibroblasts are found (rebuild from ).
(B) Confocal reflection image of an amnion (512 x 512 pixels, 553 nm x 553 nm, bar represents 50µm). The image slice was a single z-slice from a z-stack of the amnion (total stack size ~50 µm) but because the amnion was not perfectly flat, all layers from the epithelial line to the connective tissue can be imaged simultaneously.
In its organization of an epithelial layer adjacent to a basement membrane it resembles very closely the hierarchy of a fibrotic tissue surrounding a tumor [110, 111]. Therefore it provides a perfect model system to study tumor cell migration and the building of
62 III RESULTS AND DISCUSSIONmetastasis. Moreover the amnion is routinely used as an implant material for wound healing, the treatment of COPD and as a source for human stems cells [109, 112, 113].
Here we analyzed the physical structure of the amnion and additionally studied it‟s influence on growth and invasion of different breast cell types. We used the amnion as an ECM substitute material for 3D cell invasion experiments and show our preliminary results, as this study is still in its beginnings. As an outlook we show that this complex human membrane can be used for cell migration experiments to study cell behavior in a completely natural system.
III RESULTS AND DISCUSSION 63
3.3.2 Cell migration in different matrices
We previously demonstrated that cell migration was sensitive with changes in morphological and mechanical properties of collagen gels. But cell migration is also likely to be influenced by the composition of the ECM . As mentioned before, cells have the ability to remodel and digest the matrix, but this also depends on the proteins that are present in the ECM . Therefore we tested the invasive behavior of MDA MB 231 cells not only in collagen but also in fibrin, another biopolymer. Fibrin networks self-assemble in a similar way as collagen gels. We analyzed the fibrin gels in terms of pore sizes as described above. To be able to compare the invasion of MDA cells in collagen and fibrin, we used similar concentrations in both networks.