Conceived and designed the experiments: LMM XY. Performed the experiments: XY. Analyzed the data: LMM XY. Contributed reagents/materials/analysis tools: LMM XY. Wrote the paper: LMM XY.
The authors have declared that no competing interests exist.
The ability of tumor cells to invade is one of the hallmarks of the metastatic phenotype. To elucidate the mechanisms by which tumor cells acquire an invasive phenotype, in vitro assays have been developed that mimic the process of cancer cell invasion through basement membrane or in the stroma. We have extended the characterization of the circular invasion assay and found that it provides a simple and amenable system to study cell invasion in matrix in an environment that closely mimics 3D invasion. Furthermore, it allows detailed microscopic analysis of both live and fixed cells during the invasion process. We find that cells invade in a protease dependent manner in this assay and that they assemble focal adhesions and invadopodia that resemble structures visualized in 3D embedded cells. We propose that this is a useful assay for routine and medium throughput analysis of invasion of cancer cells in vitro and the study of cells migrating in a 3D environment.
The most common method currently employed to investigate cell invasion potential is probably the commercial Boyden chamber, modified with a thin layer of Matrigel, through which cells must also crawl to reach the other side of the filter
Formation of invadopodia on a thin gelatin matrix overlaid on glass is particularly amenable to imaging and has allowed the characterization of the dynamics and protein composition of these invasive protrusions of the cytoskeleton
Focal adhesions are distinct from invadopodia and serve as the mechanical linkages to the ECM as well as hubs to integrate and direct numerous signaling proteins at sites of integrin binding and clustering
A circular invasion assay (CIA) was previously described to allow higher throughput and easier visualization of invading cells
Cell culture reagents were purchased from Invitrogen (Paisley, UK). MDA-MB-231 breast adenocarcinoma cells and CHL-1 melanoma cells were obtained from ATCC. HT1080 fibrosarcoma cells were gifts from B. Ozanne (The Beatson Institute, Glasgow, UK). These cells were routinely cultured in complete DMEM supplemented with 10% fetal bovine serum (FBS) and 2 mM L-glutamine at 37°C in a humidified incubator with 5% CO2. Transfection of DNA plasmids and siRNA into these cells was performed by using the Amaxa “Nucleofector” system (Solution V, Programme X-013) according to the manufacturer's instructions.
Antibodies were routinely used at 1∶1000 for western blotting and 1∶200 for immunofluoresence. Polyclonal rabbit anti-N-WASP was obtained from Atlas (Sigma). Monoclonal mouse anti-cortactin (4F11), polyclonal rabbit anti-p34-Arc (ARPC2) and monoclonal mouse anti-MT1-MMP were obtained from Millipore (Watford UK). Polyclonal rabbit anti-phospho-paxillin is from Cell Signalling Technology. Monoclonal mouse anti-vinculin is from Sigma-Aldrich. Monoclonal mouse anti-GAPDH is from Ambion. Rhodamine phalloidin, DAPI nucleic acid stain, DQ collagen, anti-mouse IgG and anti-rabbit IgG AlexaFluor antibody were obtained from Invitrogen (Paisley, UK). Horseradish peroxidase-conjugated secondary antibodies were obtained from Jackson ImmnoResearch Laboratories (Suffolk, UK). BD Matrigel™ basement membrane matrix is supplied by BD Biosciences.
The Cherry-MT1-MMP was a generous gift from Dr. Philippe Chavrier. The mouse GFP-constructs is a gift from Dr. Michael Way. N-WASP non-targeting (NT) control siRNA, ON-TARGETplus SMARTpool siRNA targeting MT1-MMP were purchased from Dharmacon.
For western blot analysis, cells were lysed in RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% NP-40 and 0.25% Na-deoxycholate) with protease inhibitor cocktail (Pierce). Lysate were separated by SDS-PAGE and transferred to PVDF membranes (Amersharm). Western blotting was performed with the ECL chemiluminescence detection kits (Pierce) with appropriate species-specific horseradish peroxidase-conjugated secondary antibodies. The images were recorded and processed using GeneSnap software and Bio-imaging system (Syngene). Western blots shown in figures are representative of typical knockdowns obtained on multiple occasions for each experiment shown.
Cells in CIA were fixed in 4% formaldehyde for 30 min followed by permeabilization in 0.1% Triton X-100 for 20 min and blocking in 1% BSA. Primary antibodies were used at a 1∶200 dilution in blocking buffer. After 3 hours of incubation at room temperature or 16 hours 4°C (depending on the antibody) cells were washed extensively in blocking buffer, then secondary antibody was added at 1∶400 dilution in blocking buffer (plus fluorescently labeled phalloidin and DAPI if required) for 1 hour at room temperature.
Inverted invasion assays were performed as previously described
5×105 MDA-MB-231 cells were suspended in 300 µl of PureCol pepsinised collagen I (3.3 mg/ml) (Advanced BioMatrix) and placed in a well of 24-well plate and allow to gel in 37°C for 2 hours. 500 µl of growth medium was added in to the well after the collagen was set. Cells were allowed to invade in the gel for 24 hours before fixation with 4% formaldehyde for 1 hour. After fixation, the plug of collagen with cells inside were placed into a 4 ml tube and permeabilized with with 0.1% Triton X-100 for 1 hour. Samples were then washed extensively and labeled with rhodamine phalloidin (1∶50) and primary antibodies (1∶100) overnight at 4°C followed by washing with PBS five times. Appropriate secondary antibodies were applied and incubated overnight at 4°C followed by extensive washing the next day.
For the modified Circular Invasion Assay (CIA) method, a square space (0.80 cm2) devoid of cells was created by placing a biocompatible silicon self-stick cellular stopper (Thermoscience, Ibidi, 80209) in the center of a 35 mm glass bottom dish (Ibidi) before seeding 6×105 MDA-MB-231 cells. After cells adhere, the stopper is removed and 250 µl of 50% BD Matrigel™ (4.5 mg/ml) in PBS was overlaid onto the cell monolayer seeded in the inner circle of the dish to create a matrix barrier (0.8 mm high) against the cellular surface and allowed to polymerize for 2 hours prior to adding growth medium on the top of the set Matrigel. Monolayers with overlaid Matrigel, were then imaged with a Nikon time-lapse microscope or incubated in a humidified atmosphere of 5% CO2 at 37°C for 24 hours prior to fixation and immunofluorescence. Cells were tracked using ImageJ plugin Manual Tracking and the tracking results were analyzed using ImageJ plugin Chemotaxis Tool to calculate cell speed and invaded area. This quantification was done in least three independent experiments for each assay.
We have further developed and more extensively characterized the wound closure-based circular invasion assay as an effective method for visualizing cells during the invasion process
(A) Photos from a time-lapse video of MDA-MB-231 cells invading into Matrigel in CIA showing the formation of long invasion chains at the invading wound edge. Scale bar 50 µm. (B) Cells in a large finger-like chain at the leading edge of a CIA are shown fixed and stained with actin (red), and DNA (blue). Scale bar 20 µm. (C) Actin (red), N-WASP (green) and DAPI (blue) staining of a cell chain at the front of the invading area. White arrowheads indicate puncta of N-WASP co-localizing with filamentous actin. Scale bar 20 µm. (D) Image sequence showing an invading cell actively remodeling the matrix and generating what appear as micro-tunnels (white arrow). Scale bar 20 µm. (E) Staining of actin (red) and DNA (blue) of cell invasion chains in inverted invasion assay and visualized by confocal microscopy. Image showing collective cell invasion chains on single plane (left panel) and the side view of z-stack 3D projection (right panel). Scale bar 20 µm. See also
We compared the morphology of MDA-MB-231 cells in the CIA with and without Matrigel overlay. Without matrix, cells are well spread and form fan-like lamellipodia actin protrusions, while the cells in CIA invading under Matrigel assume an elongated shape that resembles cells in a 3D matrix (
(A) Cells in wound healing assay without Matrigel on 2D surface and cells in CIA with Matrigel overlay were fixed and stained for actin (green), focal adhesion marker phospho-paxillin (red) and DNA (blue). Z-stack confocal images were captured and cell side views are shown to indicate positions of FA/FCs. White arrowheads indicate adhesion complexes. (B) Cells invading in CIA were fixed and stained with actin (red), focal adhesion marker vinculin (green) and DNA (blue). Z-stack confocal images were captured and cell side views are shown to indicate positions of FA/FCs. White arrowheads indicate adhesion complexes. (C) Quantification of adhesion complexes (puncta stained with phospho-paxillin) at the bottom of the cells (within 1 µm range above the glass) and on the cell body or on top of the cells (above 1 µm) under both conditions. All error bars indicate means ± SD; **, P<0.01 by Student's t-test. See also
To test whether migration in CIA is matrix metalloprotease (MMP) dependent, we monitored cell migration with addition of MMP broad-spectrum inhibitor GM6001. Normal medium in the CIA was replaced with medium containing 25 µM GM6001 and followed by time-lapse microscopy. As shown in
(A) Addition of 25 µM GM6001 significantly impairs MDA-MB-231 cell invasion in CIA speed and progressed area (arbitrary units) versus time are shown in the graphs. Scale bar 50 µm. (B) siRNA knockdown of MT1-MMP significantly impairs invasion in CIA. Western blot shows a representative knockdown of MT1-MMP. Movies were analyzed in three independent experiments. All error bars indicate means ± SD; **, P<0.01 by Student's t-test. Scale bar 50 µm. See also Movie S7.
We also depleted MT1-MMP using siRNA and tested whether inhibition of major pericellular collagenase MT1-MMP alone is sufficient to impair cell ability to invade in CIA under Matrigel. Depletion of MT1-MMP also significantly (around 30%) impaired cell invasion in under Matrigel assay (
On thin gelatin matrix, cells make invadopodia, actin-rich protrusions that degrade ECM
Invadopodia are cytoskeletal structures enriched in filamentous actin and multiple actin-associated proteins. We probed for a subset of established components of invadopodia: N-WASP, cortactin and Arp2/3 complex and discovered that they all localized to puncta within invasive pseudopods. These puncta also contained filamentous actin and were often associated with actin spikes (white arrowheads,
(A–C) Cells invading under Matrigel in CIA are fixed and stained with proteins previously localized to invadopodia, cortactin (red), N-WASP (green), Arp2/3 component p34-Arc (green) and actin (blue) in a number of cell lines including MDA-MB-231 (A), CHL1 (B), HT1080 (C). (D) Invadopodia in CIA localize at the front of the invading pseudopods (white arrowhead) and branching sites of the pseudopods (blue arrowhead). Quantification shows there are more invadopodia in the front half of the cells than the rear half. Cells were analyzed in three independent experiments. All error bars indicate means ± SD; **, P<0.01 by Student's t-test. All scale bars 20 µm.
One of the important features of invadopodia is that they have matrix-degrading capability
MDA-MB-231 cells in the CIA assay, showing (A) Cherry-MT1-MMP (red) containing vesicles are delivered to invadopodia structures marked by GFP-N-WASP (green) in CIA. Scale bar 20 µm in the upper panel and 10 µm in the lower panel. (B) DQ collagen is mixed with Matrigel and overlaid on top of the cells in CIA. DQ collagen fluorescence (green) is visualized around some of the actin-rich puncta (blue) with co-localization of N-WASP (red), indicating that they are invadopodia structures in CIA. Scale bar 20 µm.
(A) MDA-MB-231 cells completely embedded in collagen I also show similar elongated morphology as in CIA. Staining of N-WASP or cortactin (green) and actin (red) show that N-WASP and filamentous actin containing structures resembling invadopodia (white arrowheads) are also present in cells migrating in a pure 3D environment. Scale bar 20 µm. (B) Z-stack projection showing invadopodia-like structures (arrowheads) localize to various positions, including the front of the invading pseudopods facing upward into the Matrigel and at the periphery and dorsal surface. Staining shows N-WASP (green), actin (red) and DAPI (blue) or cortactin (blue). See also Movie S8.
Many of the invadopodia-like puncta locate at the cell periphery and in bright F-actin protrusions, indicating that they interface with the matrix. A 3D reconstruction of cells invading in CIA shows that invading cells typically display a wedge shape with the widest part near the nucleus and with long thin pseudopods, which extend between the glass bottom and the Matrigel (Movie S8). Unlike invadopoda on a 2D gelatin surface, which insert downward into gelatin layer
We present a straightforward, accessible and quantifiable invasion assay for use with cancer cells in vitro, which is a modification of the CIA
Cells in CIA can be observed live in time lapse using fluorescent probes for proteins (such as GFP and mCherry,
MDA-MB-231 breast cancer cells invaded collectively into the CIA, forming chains of cells that often exceeded 4–5 cells in a row and appeared very similar to cells invading into thick Matrigel plugs in the inverted invasion assay of Hennigan
Migration in the CIA was dependent on MMP activity and in particular, it was partially dependent on MT1-MMP, a transmembrane metalloprotease. Treatment with GM6001 caused around a 50% decrease in cell speed and a significant decrease in the area covered by cells in 30–40 hours of invasion. Depletion of MT1-MMP by siRNA was somewhat less effective, with around a 30% decrease in speed and a smaller but significant decrease in area covered. The reasons for a partial rather than total dependence on protease for migration in CIA are likely multiple. Firstly, we used Matrigel for this assay and MT1-MMP is a collagenase, while the complex mixture of Matrigel also contains other components such as laminin (50–60%), collagen IV (30%), entactin (8%) (BD Bioscience website and
Conversely, there could be multiple reasons why cells need metalloproteases to crawl in CIA, including degradation of a physical matrix barrier
A previous study has shown that macrophages also assume a mesenchymal mode while invading in ECM
In summary, we have made some minor modifications to the CIA assay
(MOV)
(MOV)
(MOV)
(MOV)
(MOV)
(MOV)
(MOV)
(MOV)
We thank Heather Spence and Robert Insall for technical assistance and advice. We also thank the BAIR imaging facility and Kurt Anderson for excellent imaging support.