We have developed 3D coculture models for live-cell imaging in real-time of interactions among breast tumor cells and other cells in their microenvironment that impact progression to an invasive phenotype. These models can serve as preclinical screens for drugs to target paracrine-induced proteolytic, chemokine/cytokine and kinase pathways implicated in invasiveness.
We have developed 3D coculture models, which we term MAME (mammary architecture and microenvironment engineering), and used them for live-cell imaging in real-time of cell:cell interactions. Our overall goal was to develop models that recapitulate the architecture of preinvasive breast lesions to study their progression to an invasive phenotype. Specifically, we developed models to analyze interactions among pre-malignant breast epithelial cell variants and other cell types of the tumor microenvironment that have been implicated in enhancing or reducing the progression of preinvasive breast epithelial cells to invasive ductal carcinomas. Other cell types studied to date are myoepithelial cells, fibroblasts, macrophages and blood and lymphatic microvascular endothelial cells. In addition to the MAME models, which are designed to recapitulate the cellular interactions within the breast during cancer progression, we have developed comparable models for the progression of prostate cancers.
Here we illustrate the procedures for establishing the 3D cocultures along with the use of live-cell imaging and a functional proteolysis assay to follow the transition of cocultures of breast ductal carcinoma in situ (DCIS) cells and fibroblasts to an invasive phenotype over time, in this case over twenty-three days in culture. The MAME cocultures consist of multiple layers. Fibroblasts are embedded in the bottom layer of type I collagen. On that is placed a layer of reconstituted basement membrane (rBM) on which DCIS cells are seeded. A final top layer of 2% rBM is included and replenished with every change of media. To image proteolysis associated with the progression to an invasive phenotype, we use dye-quenched (DQ) fluorescent matrix proteins (DQ-collagen I mixed with the layer of collagen I and DQ-collagen IV mixed with the middle layer of rBM) and observe live cultures using confocal microscopy. Optical sections are captured, processed and reconstructed in 3D with Volocity visualization software. Over the course of 23 days in MAME cocultures, the DCIS cells proliferate and coalesce into large invasive structures. Fibroblasts migrate and become incorporated into these invasive structures. Fluorescent proteolytic fragments of the collagens are found in association with the surface of DCIS structures, intracellularly, and also dispersed throughout the surrounding matrix. Drugs that target proteolytic, chemokine/cytokine and kinase pathways or modifications in the cellular composition of the cocultures can reduce the invasiveness, suggesting that MAME models can be used as preclinical screens for novel therapeutic approaches.
1. Prepare DQ-substrates
It may be necessary to agitate DQ-substrate in an ultrasonic water bath for ~5 min and heat to 50 °C to facilitate dispersion.
2. Prepare MAME cocultures
See Figure 1 for a schematic diagram of cocultures.
3. Perform 4D (3D + time) imaging of live MAME cocultures by confocal microscopy
4. Representative Results
The MAME cocultures that we developed are a tractable experimental model for live-cell imaging in real-time of interactions among the various cellular constituents that comprise a breast tumor and its microenvironment. Here we demonstrate the capability of MAME models to recapitulate cell:cell interactions during breast cancer progression and of live-cell imaging to follow those interactions over time. We illustrate the ability to image interactions between a preinvasive MCF10.DCIS human breast cell line and a human breast cancer-associated fibroblast line, WS-12Ti, over a period of more than three weeks in culture. The cocultures were imaged at 3, 16 and 23 days. Optical sections through the entire volume of the cocultures were reconstructed in 3D and representative examples are shown for each of the three timepoints in Figure 2. The red fluorescence in Figure 2 represents the MCF10.DCIS and WS-12Ti cells. To discriminate between the two cell types and for long-term cocultures, one can transfect or transduce individual cell types with fluorescent proteins prior to establishing the cultures. A MAME coculture of MCF10.DCIS and WS-12Ti cells that are differentially labeled with fluorescent proteins for identification is illustrated in Figure 3.
By incorporating substrates, in this case DQ-collagens, into the MAME cocultures, we can assess and quantify changes over time in proteolysis associated with transition to an invasive phenotype. We have established methods for quantifying the fluorescent cleavage products that result from DQ-collagen degradation2,3. We normalize degradation products throughout the entire 3D volume of the MAME cocultures on a per cell basis by labeling and counting cell nuclei. Furthermore, we can localize the degradation to extracellular and intracellular compartments and quantify the total, intracellular and extracellular degradation products as we have described previously2. The green fluorescence in Figures 2 and 3 represents cleavage products of the two DQ-collagen substrates. At this time, differentially-labeled collagens are not available that would allow us to distinguish between degradation of type IV collagen and degradation of type I collagen.
The interactions in the MAME cocultures are dynamic and can change over time. To obtain temporal and dynamic information, the same field of view can be imaged and optical sections taken through the entire specimen of the volume repeatedly over a period of time from minutes to weeks, thus collecting data in 4-D. In Figure 2, we illustrate rather dramatic changes that occur over 23 days: changes in cell number, cell migration and cell interactions, structure volumes, structure shapes, deviations of the structures from sphericity, invasive outgrowths and proteolysis. Furthermore, we also demonstrate an overall change in volume due to the MAME cocultures degrading their surrounding extracellular matrix over this time period. At three days as illustrated in Figure 2A, the volume is 110,000 μm3. In contrast, the volume of the MAME cocultures at 16 and 23 days is only 46,250 μm3, as illustrated in Figures 2B and C.
Figure 1. Schematic diagram of MAME cocultures. Plastic coverslip is coated with collagen I, containing DQ-collagen I and fibroblasts (magenta). A 2nd layer of reconstituted basement membrane (rBM) containing DQ-collagen IV is added. Tumor cells (red) are plated on top and 2% rBM in culture media is overlaid (white dots) with every change of media. Green represents the fluorescent cleavage products of DQ-collagen I and IV.
Figure 2. MAME cocultures of MCF10.DCIS human breast cells and WS-12Ti human breast fibroblasts. The fibroblasts were embedded in collagen I/DQ-collagen I and overlaid with rBM containing DQ-collagen IV. The DCIS cells were then seeded on the rBM and overlaid with media containing 2% rBM. Over the 23 days in coculture illustrated here, there was cell proliferation, degradation of DQ-collagen IV and I (green) and collagen matrices as indicated by the reduction in volume of the cocultures. Cells were localized by labeling with CellTracker Orange in situ before imaging (red). The same live cocultures were observed over the 23 days with images captured by confocal microscopy at 3, 16 and 23 days of coculture. The resulting optical slices were reconstructed in 3D with Volocity software. Magnification, 10X.
Figure 3. MAME 16 day cocultures of MCF10.DCIS human breast cells and WS-12Ti human breast fibroblasts that express RFP (red) and YFP (white), respectively. Cocultures were established and analyzed as in Figure 2. The two cell types shown here had, however, been differentially labeled so that they could be distinguished from one another. The higher magnification images in this figure, as compared to those in Figure 2, illustrate proteolysis of DQ-collagen IV at the surface of the DCIS cells and diffuse proteolysis of DQ-collagen I in areas near the fibroblasts. Magnification, 20X.
As demonstrated, the MAME cocultures can be used for live-cell imaging in real-time of interactions among the various cellular constituents that comprise a breast tumor and its microenvironment. In ongoing studies in our laboratory, we have used MAME cocultures to identify proteolytic pathways associated with transition from DCIS to invasive ductal carcinoma, as well as the interactions between proteolytic pathways and other pathways involved in this transition such as chemokine/cytokine/growth factor pathways. We further showed that MAME models could be used as a tool for screening the effects of small molecule inhibitors, blocking antibodies or shRNAs that modulate proteolytic/chemokine/cytokine/growth factor pathways3-5. The MAME models can be used for live-cell imaging of tumor growth, invasion and proteolysis over times ranging from minutes and hours, as we have previously shown6,7, to weeks as illustrated here in Figures 2 and 3. The interactions being observed are between cells of one species, i.e., human tumor cells and tumor-associated cells, rather than between human tumor cells and mouse stromal cells as in intravital imaging8. The MAME models are a tractable system. Molecules of interest can be downregulated with shRNAs in individual cell types prior to coculturing so that their contribution in that cell type to degradation of DQ-collagens can be imaged and quantified in 3D2. Conditioned media from MAME models can be sampled to measure changes in secretion of proteases, cytokines, etc. The information obtained provides an experimental basis for testing the effects of small molecule inhibitors, blocking antibodies, etc.
The cellular composition of MAME cocultures can be readily manipulated so that one can use them to assess the contribution of other cell types found in the tumor microenvironment (e.g., myoepithelial cells, monocytes, endothelial cells of blood vessel and lymphatic origin) to malignant progression6,7. The number of the various cell types seeded, the ratio of one cell type to another and the length of time in culture will depend on the specific cell types used and the experimental question. MAME models can also be extended to assess the effects of changes in the microenvironment surrounding the breast tumors, e.g., pH, oxygen tension. For example, we have shown that when MAME cultures are maintained at a slightly acidic pH, i.e., pH 6.8, there is an increase in proteolysis and invasiveness. In addition, we have also shown that we can use similar cocultures to model other cancers like prostate cancer.
The authors have nothing to disclose.
This work was supported, in part, by National Institutes of Health R01 CA131990 (BFS and RRM). Live-cell imaging was performed in the Microscopy, Imaging and Cytometry Resources Core supported, in part, by NIH Center grant P30 CA22453 to the Karmanos Cancer Institute, Wayne State University and by the Perinatology Research Branch of the National Institutes of Child Health and Development, Wayne State University.
Reagent/Equipment | Company | Catalogue Number | Comments |
Reconstituted basement membrane (Cultrex) | Trevigen | 3445-005-01 | Comparable to Matrigel (BD Biosciences) |
Collagen I | Cohesion Laboratories | 5005-B | |
DQ-substrates (collagen I, IV) |
Invitrogen | DQ-col I-D12060 DQ-col IV D12052 |
|
22-mm plastic coverslips | Fisher Scientific Co. | 12-547 | Cut in half, leave in 70% ethanol for 10 min and air-dry before use. |
Cell culture medium | Lonza | MEBM-PRF CC-3153 MEGM CC-4136 |
Phenol red–free |
ibiTreat μ-Dish | Ibidi | 80136 | |
Volocity software | Perkin-Elmer | Version 5.5 | Other brands of imaging software can be used to generate 3D reconstructions and movies. |