«Der Naturwissenschaftlichen Fakultät der Friedrich-Alexander-Universität Erlangen-Nürnberg zur Erlangung des Doktorgrades Dr. rer. nat. vorgelegt ...»
We assembled collagen gels from 0.3 mg/ml to 2.4 mg/ml and fibrin gels from 0.25 mg/ml to 2 mg/ml to perform invasion assays. We observed that the 0.25 mg/ml fibrin gels completely detached from the cell culture dishes, thus, were technically difficult to use for experiments. We repeated the experiment with 0.25 mg/ml fibrin gels 3 times and although the gels polymerized normally, after 3-5 hours they detached from the cell culture dish and floated in the dishes. Therefore cells were seeded only on fibrin gels from 0.5 mg/ml to 2 mg/ml gels, which remained stable over the time of measuring.
Following the protocol for the invasion assay, 50,000 cells were seeded on top of all gels, collagen and fibrin, and placed into the incubator for 3 days. When the assays were stopped by fixation after 3 days, we observed that in 0.5 mg/ml fibrin gels, cells produced extensive holes in the gels (Fig. 3.3.2, B), which made it impossible to analyze as all cells had accumulated at the bottom of the dish. Thus we only analyzed fibrin gels of 1 mg/ml and 2 mg/ml and collagen gels of 0.3 mg/ml, 0.6 mg/ml, 1.2 mg/ml and 2.4 mg/ml. Analyzing the assays showed that MDA invasion into both fibrin gels was possible and that they invaded with similar p5 values of 150.32 ± 2.31 µm for 1 mg/ml and 165.03 ± 3.08 µm for 2 mg/ml (Fig. 3.3.3). For the collagen gels, we found the same biphasic response of cell invasion on concentration as shown in Chapter 3.1.3, although in this experiment, p5 values were higher than before. This shows on the one hand, that with every passage of the cells, the maximum invasion depth can vary.
However it also shows that the biphasic answer of MDA cell invasion on collagen concentration, with a maximum of invasion at 1.2 mg/ml, is highly reproducible. For fibrin we were not able to measure such a broad spectrum of concentrations, but we were able to show that MDA cells also invaded into fibrin and that their invasion depth
64 III RESULTS AND DISCUSSIONwas similar to those of MDAs in collagen gels. To evaluate in detail how cell invasion is dependent on the composition of an ECM however, it is necessary to repeat the invasion assays in fibrin gels, ideally with a broader range of concentrations. Moreover, invasion assays in co-polymerized networks of collagen and fibrin would help to enlighten the question of how composition of the ECM influences cell migration and invasion.
2: MDA cell invasiveness in different matrices (A-D) Invasion profiles of MDA cells in fibrin networks of different concentrations. Fibrin networks of concentration 0.25 mg/ml (A) were not stable enough in 35mm culture dishes to seed cells on top and measure invasion. In
0.5 mg/ml fibrin gels (B), MDA cells did invade, but created massive holes (red arrowheads) in the gels. All MDA cells were attached to the bottom of the dish and an analysis of invasion depth was not possible. MDA cells did invade into fibrin gels of 1 mg/ml (C) and 2mg/ml (D).
(E-H) Invasion profiles of MDA cells in collagen gels of concentrations 0.3 mg/ml (E), 0.6 mg/ml (F), 1.2 mg/ml (G) and 2.4 mg/ml (H). The collagen profiles show the same biphasic answer of invasion depth on collagen concentration as seen before (Fig. 3.1.17), with a maximum at 1.2 mg/ml. In all panels, at least 3,000 cells were analyzed.
III RESULTS AND DISCUSSION 65Figure 3.3.
3: MDA cell invasiveness in different matrices. P5 values derived from the invasion profiles in 3.3.2. Invasion of MDA MB 231 cells was similar in fibrin gels of 1.0 mg/ml and 2.0 mg/ml (left side). In collagen gels (right side), invasion of MDA MB 231 cells followed a similar biphasic answer as seen in Figure 3.1.17. For the analysis at least 25 fields of view and 3,000 cells were analyzed (mean ± SE).
66 III RESULTS AND DISCUSSION
3.3.3 Different cell types prefer different matrices
To evaluate if the different migration behavior in different matrices is cell type specific, we tested some of our matrices with different cell lines. The knowledge of cell type specific migration behavior in different matrices could provide new information for the development of implants and answer the questions why some tumor cells are more aggressive and malignant than others.
In our experiments we compared the following cell lines: MDA MB 23, which is an established epithelial cell line from the metastatic tissue of a 51 years old breast tumor patient . The cell line was cultured in our lab and treated as described in the Materials and Methods section. PS and EJ cells were primary mesenchymal cells, extracted from two breast tumors from cancer patients at the Frauenklinik in Erlangen.
Both women with breast cancer demonstrated a clinical and molecular phenotype where both types of cancer were rare and with poor prognosis (Tab. 3.3.1).
patient also had a poorer prognosis. MDA cells were franchised from ATCC and were also triple negative for treatment receptors.
To compare these three types of tumor cells with non-cancerous normal cells, we also extracted primary epithelial cells from healthy breast tissue as a control cell line, labeled with the letters KB. We then performed invasion assays of all cell lines in fibrin and collagen to test for differences.
Data analysis revealed that all cancerous cell lines where highly invasive into the collagen gels (Fig. 3.3.4). Especially PS cells, where 5% of the cells migrated to an invasion depth of ~ 200 µm. EJ and MDA cells migrated similarly to a value of 171.01 µm ± 2.66 µm and 169.79 µm ± 2.82µm, respectively. For Non-cancerous normal KB cells on the other hand, a significant invasion was not observed. Their p5 value was only ~ 60µm and microscopically we observed that they formed colonies which grew into a similar depth. We did not consider KB cells invasive, because when the colonies were growing on the surface of the gel, they could also indent the matrix up to 60 µm.
−2 −2 10 −1
The analysis of the fibrin gels showed a very different result (Fig. 3.3.5). All cell types demonstrated decreased invasion depths compared to collagen gels. For example, the highly invasive PS and MDA cells only invaded ~ 90 µm or ~ 100 µm, respectively, compared to collagen gels. Surprisingly, EJ, which was also a deep migrating cell line in collagen gels, had a strongly reduced and non-significant invasion depth of only ~ 35 µm. The non-cancerous KB also showed a p5 value of ~ 50 µm, which we attributed again to the colony growth and considered these cells as performing no significant invasion into the fibrin gels.
In conclusion, these data show that different cell types alter their migratory behavior and appear to prefer different types of extracellular matrix for invasion and migration.
In addition, repeating the measurements with MMP inhibitors will show which cells rely on MMPs and will further enlighten the details of their migration through a 3D ECM. These findings could be important if a patient needs a soft tissue implant, for example after a successful cancer treatment. This implant should be based on a material which prohibit the building of metastases and promote the re-cellularization of noncancerous cells.
−2 −2 10 10 −1
5: Different cell types in fibrin gels. All cells used in the experiment were breast isolated cells. All collagen concentrations were 1 mg/ml. (A-D) Invasion profiles of EJ cells (A), PS cells (B), MDA cells (C) and KB cells (D). All cell types invaded into the collagen gels, although it is clearly visible that PS cells (B) were more invasive compared to the other cell lines, whereas the non–cancerous KB cells were the least invasive. (E) For better comparison, the profiles from panels A-D are plotted into one graph. Moreover, the p5 value was determined (inset) for easier comparison. In all panels, at least 1,800 cells were analyzed.
III RESULTS AND DISCUSSION 693.3.4 New frontiers – the human amnion as a substitute material for cell migration experiments The idea to use the human amnion from the placenta, a „leftover‟ from childbirth, for implant use was first realized in the 1950s . Our aim was to use this fully natural material as a substitute material to study cell migration in vitro. To better visualize the composition of the amnion, we stained the cell nuclei with Draq5 and the cell membrane with agglutinin membrane stain and obtained image stacks with confocal microscopy. We reconstructed the image stacks with the 3D reconstruction software AMIRA. The reconstructed images confirmed the organization of the amnion, with the epithelial cuboidal cell layer on top of the basement membrane and the connective tissue housing single fibroblasts (Fig. 3.3.6 and SI Movie_5).
6: The human amnion as an extracellular matrix. Confocal image stacks of an amnion were obtained (512 x512 pixel, 599 nm x 599 nm x 1007 nm). Amnion was stained with Draq5 for nuclei and membrane stain wheat germ agglutinin 488. (A) Side view of a reconstructed image stack. The basement membrane is imaged in reflection mode and illustrated in light gray, whereas the dark gray areas indicate the connective tissue. Red color indicates cell nuclei of the epithelial layer of cells. Draqu5 also diffused through the epithelial layer of cells and stained the nuclei of fibroblasts in the connective tissue. Cell membrane of epithelial layer is indicated in green. (B) Magnification of the image area marked with a black box in (A). It is clearly visible that the membrane was only specific for the epithelial layer, indicating a monolayer of epithelial cells. (C) Top view of the same stack as in (A) and (B). Again the epithelial layer is clearly visible. All image stacks were reconstructed with AMIRA 5.3.
70 III RESULTS AND DISCUSSIONWhen monitoring the amnion in reflection mode, the basement membrane and the extracellular matrix appeared as a dense and homogenous material. However, it was possible to find deep areas where the tissue represented a fibrillar network structure (Fig. 3.3.7). We estimated a preliminary pore size of ~ 50 µm.
7: Fibrilar structure of the human amnion. Confocal image stacks from human amnion (512 x 512 pixels, 450 nm x 450 nm x 465 nm). (A) Image slice from the obtained stack. The light gray areas on the left side indicate the thick basement membrane, where no measurement of the pore size was possible, due to the dense material. (B) Magnification and brightening of the area marked with the red box in (A). The connective tissue shows areas where more fibrillar structures could be found. We analyzed these structures and found a preliminary pore size of ~50 µm. Total thickness of both slices was 465 nm.
To analyze the invasive behavior of breast cells on the amnion, we seeded 300,000 cells of either MDA MB 231 cells, PS cells, EJ cells or KB cells on top of the amnion.
Similar to the invasion assay we treated the amnion along with the cells with 2.5 % glutaraldehyde to fix the cells in the state they were. We stopped the assays after 3, 6, 10 and 13 days. While EJ and PS cells formed large colonies over the days (Fig. 3.3.8 and 3.3.9) MDA cells did not form colonies (Fig. 3.3.10). In normal cell culture flasks or dishes none of these cell lines form colonies. Interestingly, EJ cells built colonies on the amnion that grew in length and width but also in height. At day 13, they did not invade deeper than 20µm into the basement membrane. PS on the other hand, built colonies that were quite large in width and length (30 µm and 70 µm, respectively) and that grew deep into the basement membrane (80 µm). We conclude that although both
III RESULTS AND DISCUSSION 71breast tumor cells can breach through the epithelial layer into the basement membrane and beyond, PS cells are more invasive into the amnion compared to EJ.
8: EJ cells on amnion at day 3. (A) Top view of a reconstructed stack from confocal images (512x512 pixel, 486 nm x 486 nm x 495 nm, total stack height 110 µm). The agglomeration of cells is the colony of EJ cells, the cell nuclei in the monolayer are the epithelial cells. Although only the EJ cells were stained with Draq5, the staining diffused out of the EJ cells and stained the nuclei of the epithelial cells. (B) Side view of the same stack as in (A). EJ cells already piled up in their colony but did not invade into the basement membrane.
9: EJ cells on amnion at day 13. (A) Top view of a reconstructed stack from confocal images (512x512 pixel, 526 nm x 526 nm x 494 nm, total stack height 150 µm). The EJ colony is clearly visible in the middle as well as the monolayer of epithelial cells. (B) Side view of the same stack as in (A). The colony of EJ cells grew in height but also invaded ~20 µm into the basement membrane. Note that the colony represents many cells densely arranged, where we could not resolve single cell nuclei.
The single MDA cells however did not invade into the basement membrane although they proliferated on top of the epithelial layer. To test if the epithelial layer was inhibiting the invasion of MDA cells we denuded parts of the amnion by scratching the epithelial layer off with a scalpel. Indeed, MDA accumulated in the “wounded” area by day 3 (Fig. 3.3.10) and began to grow into the basement membrane (Tab. 3.3.2).