Models of tumor cell invasion into three-dimensional extracellular matrix better reflect the in vivo situation than two-dimensional motility assays. Using matrix invasion assays combined with confocal imaging of fluorescently-labeled cells, detailed information on invasion modes and the distinct contributions of leading versus following cells can be obtained.
A defining characteristic of cancer malignancy is invasion and metastasis 1. In some cancers (e.g. glioma 2), local invasion into surrounding healthy tissue is the root cause of disease and death. For other cancers (e.g. breast, lung, etc.), it is the process of metastasis, in which tumor cells move from a primary tumor mass, colonize distal sites and ultimately contribute to organ failure, that eventually leads to morbidity and mortality 3. It has been estimated that invasion and metastasis are responsible for 90% of cancer deaths 4. As a result, there has been intense interest in identifying the molecular processes and critical protein mediators of invasion and metastasis for the purposes of improving diagnosis and treatment 5.
A challenge for cancer scientists is to develop invasion assays that sufficiently resemble the in vivo situation to enable accurate disease modeling 6. Two-dimensional cell motility assays are only informative about one aspect of invasion and do not take into account extracellular matrix (ECM) protein remodeling which is also a critical element. Recently, research has refined our understanding of tumor cell invasion and revealed that individual cells may move by elongated or rounded modes 7. In addition, there has been greater appreciation of the contribution of collective invasion, in which cells invade in strands, sheets and clusters, particularly in highly differentiated tumors that maintain epithelial characteristics, to the spread of cancer 8.
We present a refined method 9 for examining the contributions of candidate proteins to collective invasion 10. In particular, by engineering separate pools of cells to express different fluorescent proteins, it is possible to molecularly dissect the activities and proteins required in leading cells versus those required in following cells. The use of RNAi provides the molecular tool to experimentally disassemble the processes involved in individual cell invasion as well as in different positions of collective invasion. In this procedure, mixtures of fluorescently-labeled cells are plated on the bottom of a Transwell insert previously filled with Matrigel ECM protein, then allowed to invade “upwards” through the filter and into the Matrigel. Reconstruction of z-series image stacks, obtained by confocal imaging, into three-dimensional representations allows for visualization of collectively invading strands and analysis of the representation of fluorescently-labeled cells in leading versus following positions.
1. Retroviral labeling of cells with fluorescent proteins
2. Inverse Matrigel invasion assay
3. Staining and visualization
4. Representative Results
An example of a Z-series of optical slices is shown in Figure 3A. In this instance, cells were stained with Calcein AM and the number of cells invading up from the filter can be seen to decrease with distance. Quantification of invasion can be done by analyzing the ratio of Calcein AM positive pixels to negative pixels at each interval, or by using the fixation/staining method detailed above and counting PI positive nuclei at each position. One advantage of Calcein AM staining is that 3-dimensional reconstructions of cell invasion can be assembled using software such as Volocity, giving a visual depiction of the mode of invasion (Figure 3B). If cells are labeled by expression of fluorescent proteins, then the positions of each color cell can be visualized in 3-dimensional reconstructions, either viewed from the side (Figure 3C) or by making slices through the reconstruction (Figure 3D).
Figure 1. Schematic diagram of steps involved in setting up inverse invasion assay. A) Matrigel ECM thawed on ice. B) Matrigel is diluted 1:1 with PBS containing any drug treatments at twice final concentration. C) Transwell inserts are placed into multiwell plates, and Matrigel pipetted into each. D) Cell suspensions made at desired concentration. E) Once the Matrigel has set, the plate is inverted and removed, cells are plated onto the underside filter of the Transwell inserts. F) In the inverted position, the multiwall plate is carefully placed over Transwell inserts, making contact with the cell suspension. G) Cells are allowed to adhere to the filter for 4 hours.
Figure 2. Continuation of the schematic diagram for the inverse invasion assay. A) Once cells have adhered, dip each Transwell into serum-free media twice to remove loose cells. B) Place washed Transwell into a final well containing media plus treatments as required. C) Media containing chemoattractant (e.g. 10% fetal bovine serum) with treatments as required is carefully layered onto Matrigel.
Figure 3. Representative images of results of inverse invasion assay. A) Optical sections of cells stained with Calcein AM invading Matrigel, taken at 15 μm intervals by confocal microscopy. B) Reconstruction of a 3-dimensional reconstruction of cell invasion from a stack of confocal Z-series images, viewed from the side. Reprinted from reference 10. C) Three-dimensional reconstruction of GFP and RFP-labeled cells invading Matrigel viewed from the side. D) Optical slice through the 3-dimensional reconstruction of GFP and RFP-labeled cells. Reprinted from reference 10.
Matrigel invasion assays have traditionally been set up with cells placed onto a layer of extracellular matrix protein with chemoattractant-induced motility towards and through a filter at the bottom. Invasiveness was scored as a function of how many cells could be counted on the underside of the filter. Although practically there is little difference with the “inverse” invasion assay described above, there is considerably more information about the process of invasion that can be determined by visualizing invading cells as they move up and through the Matrigel. For example, in addition to being able to visualize differently labeled cells to examine leading versus following cells in collective invasion (Figure 3C and 3D), using this method allows cells to be fixed, permeabilized, stained and then imaged for protein levels or subcellular distribution. The resolving power of this method would be limited by the baseline signal intensity of the fluorescent protein expressing cells and the sensitivity of the imaging apparatus, especially if two or more colors were to be imaged. Mixing Matrigel with dye-quenched (DQ) collagen (which is so heavily conjugated with fluorescein that the fluorescent signal is quenched until proteolysis separates collagen fragments thereby lowering the local fluorescein concentration and permitting fluorescence) is another multi-color application of this method that would allow for visualization of ECM degradation by invading cells. Using high-content screening devices fitted with confocal imaging and environmental chambers would enable this procedure to be transformed into a 4-d real time invasion assay, which could also be multiplexed with additional measures such as ECM degradation. Combined with automation, it would also be possible to design high throughput cell-based assays that could be used to screen chemical libraries for novel anti-invasion targets.
A number of factors should be carefully controlled when using this inverse invasion assay. Cell density can affect the apparent efficiency of invasion, so equal numbers of cells should be used in each replicate. Treatments that alter cell number (e.g. cytotoxic drugs) may appear to change the extent of invasion, but may also reflect this cell density influence. Although numerous cell lines work well in this assay, not all invade well enough to be useful, so each should be evaluated on a case-by-case basis. The length of time required to achieve a level of invasion adequate to provide a good signal window should be empirically determined for each cell line, the MDA MB 231 cell line works well at 4 days after the start of the assay, for example. In addition, some of the conditions suggested (e.g. 8 micron pore size for Transwell inserts) may not suit every cell line, and some optimization may be required. The extra value derived from this method when compared to the standard Matrigel invasion assay make it worthwhile establishing in any lab seriously interested in studying the processes involved in 3-D invasion.
The authors have nothing to disclose.
Funding for this research is from Cancer Research UK.
Name of the reagent | Company | Catalogue number |
---|---|---|
DMEM (Dulbecco’s Modified Eagle’s Medium) | GIBCO | 21969 |
Fetal Bovine Serum | PAA | A15-101 |
Penicillin Streptomycin | GIBCO | 15140 |
200 mM L-Glutamine (100x) | GIBCO | 25050-032 |
Puromycin | Sigma-Aldrich | P8833 |
0.05% Trypsin EDTA | GIBCO | 25300 |
Polybrene | Sigma-Aldrich | AL-118 |
Lipofectamine 2000 Reagent | Invitrogen | 11668019 |
6.5mm Transwells, 8.0 µm pore size | Corning | 3422 |
Complete Matrigel | BD Biosciences | 354234 |
Calcein AM | Invitrogen | C1430 |
RNase | Qiagen | 19101 |
Propidium Iodide | Sigma-Aldrich | P4864 |
Confocal microcope | Leica | SP2MP |