A microcarrier-based spheroid 3D invasion assay to monitor dynamic cell movement in extracellular matrix

Background Cell invasion through extracellular matrix (ECM) is a critical step in tumor metastasis. To study cell invasion in vitro, the internal microenvironment can be simulated via the application of 3D models. Results This study presents a method for 3D invasion examination using microcarrier-based spheroids. Cell invasiveness can be evaluated by quantifying cell dispersion in matrices or tracking cell movement through time-lapse imaging. It allows measuring of cell invasion and monitoring of dynamic cell behavior in three dimensions. Here we show different invasive capacities of several cell types using this method. The content and concentration of matrices can influence cell invasion, which should be optimized before large scale experiments. We also introduce further analysis methods of this 3D invasion assay, including manual measurements and homemade semi-automatic quantification. Finally, our results indicate that the position of spheroids in a matrix has a strong impact on cell moving paths, which may be easily overlooked by researchers and may generate false invasion results. Conclusions In all, the microcarrier-based spheroids 3D model allows exploration of adherent cell invasion in a fast and highly reproducible way, and provides informative results on dynamic cell behavior in vitro.


Background
Malignant tumors have the potential to metastasize from the original tissue to a distant site, which is the main cause of morbidity and mortality in tumor patients. During this process, the basic but decisive step is migration of tumor cells through the extracellular matrix (ECM) either towards lymph and blood vessels, or to a secondary site after survival in circulation [1]. To disseminate in tissue, cells require adhesion, proteolysis of ECM components and migration, which also occurs in normal physiological processes like embryonic morphogenesis and wound healing [2]. There are a diversity of strategies for cells movement, either individually (e.g. amoeboid or mesenchymal migration) or collectively (multicellular streaming, cluster, strand or sheet), which are based on cell-cell adhesion and cell-matrix interaction [3][4][5]. This activity can be simulated and observed by in vitro models and optical imaging to study cellular and molecular mechanisms. Unlike 2D migration, a 3D matrix provides both a substructure and obstacles to all surfaces of cells during movement through the surroundings, which simulates cell movement through tissues. Importantly, 3D models provide more information on the process of cell migration and invasion, including cell morphological alterations, cell-cell interaction, cell-matrix interaction, and matrix remodeling. Therefore, 3D models can serve as a supplement or an advanced alternative to 2D assays.
To examine cell invasive potential, a variety of in vitro assays are developed in a 3D format. Among them the Transwell invasion assay, or Boyden chamber assay, is widely used. Basically it includes seeding cells on a layer of ECM gel pre-coated on top of a porous membrane, and assessing cell invasion by measuring the number of cells passing through the ECM gel. The chamber invasion assay is straightforward to quantify invading cells induced by chemoattractants [6] or internal gene regulation [7]. Despite the advantages, this assay counts vertically invading cell numbers at the endpoint but neglects the whole invasion process. How cells move in matrix and interact with surroundings remains unclear. As a substitute, a matrix embedding cell culture offers more possibilities. Cell aggregates, such as multicellular spheroids, can be embedded in a 3D matrix and cells moving away from spheroids into the matrix are monitored by microscopy. This approach allows cells migrating in any direction and many migratory parameters can be detected, including cell trajectories, migration distance, and velocity. However, establishing spheroids has met with challenges such as absence of formation, lack of size and uniformity control, difficulty in manipulation, requirements of special equipment and training, and is time consuming [8,9]. Most importantly, not all cells are capable to form tight and regular-shaped spheroids, but some end up as friable and loose aggregates, or aggregation does not occur at all, which complicates manipulation and use in an invasion assay [10][11][12]. Therefore we choose microcarriers as a core to grow spheroids and to standardize the invasion assay in a simple and highly reproducible way. Adherent cells which do not aggregate spontaneously, can attach to microcarriers and thus form spheroids. Interestingly, introduction of carriers also enables co-culture of different cell types in close proximity [13]. Although microcarrierbased spheroids, because of the core, do not mimic fully the in vivo situation of solid tumors, they are faster to establish and stabilize experimental conditions allowing easy duplication compared to cell-only spheroids. In this study, we describe a microcarrierbased spheroid model to investigate dynamic cell behavior in three dimensional matrices.

Results
In this study we present a method for 3D invasion examination and introduce various measurements according to different experimental settings and requirements. The whole workflow and schematic diagram is shown in Fig. 1.

Different Cell Dispersions in Matrix Show Invasiveness
This method can be used to monitor invasiveness of adherent cells in vitro. Here we performed the 3D invasion assay Fig. 1 Workflow diagram of the whole assay with schematic drawings and example results with melanoma cell lines (BLM, M14 and MEL57) and colorectal cancer cell lines (SW480 and CACO2) in 1.6 mg/ml collagen I gel. These cell lines were chosen because of difference in cell dispersion in the matrix allowing us show typical invasiveness patterns which may be visible. Images of cell dispersion were obtained every day and the maximum migration distances were measured. Within 4 days BLM cells migrated 285 μm away from the microcarrier core. M14 and MEL57 cells migrated slower than BLM cells, with dispersion of 270 μm and 110 μm in 6 days respectively. All melanoma cells moved collectively in the matrix, but single cells were visible in the front of migrating cells. In comparison, colorectal cancer cells SW480 show less invasive and remained more connected than melanoma cell lines. CACO2 cells grew around the core to multi-layers without any sign of migration into matrices (Fig. 2). The results indicate that this 3D assay can be used to examine cell invasive capacity and the way cells move. Fig. 2 Cell invasion/dispersion in collagen I. Melanoma cells (BLM, M14 and MEL57) and colorectal cancer cells (SW480 and CACO2) were cultured on microcarrier beads and embedded in collagen I gel (1.6 mg/ml). Cell invasion was monitored and recorded daily, and three independent experiments were performed. This assay lasted for 6 days and was ended when cells started to move out of frame. a Representative pictures of cell invasion of each cell line. All three melanoma cell lines displayed invasive behavior at different levels, while two colorectal cancer cell lines appeared less invasive, especially CACO2, which showed non-invasive growth. Scale bar, 100 μm. b Line graphs show maximum migration distances measured every day of each cell line

The Content and Concentration of Matrix Influence Cell Invasion
To investigate the effect of matrix composition on cell invasion, we tried three different types of matrices. Here we use LLC cells because of the individual movement these cells show in collagen. Collagen type I and reconstituted basement membrane (Matrigel) are most commonly used matrices for 3D culture. Agar is a mixture of polysaccharides and can solidify at 32~40°C for biological use. Fluorescently labeled LLC cells disperse collectively in Matrigel, spread individually in collagen, while no migration was observed in agar (Fig. 3a). Further, to test if the concentration of matrix would influence cell invasion, we used M14 cells in a gradient of collagen matrices and monitored cell invasion in 6 days. We selected M14 cells for the moderate migration speed this cell line shows; not too fast like LLC and BLM, which would move out of the imaging field, or too slow like MEL57, SW480 and CACO2 which demand long culture time causing cell proliferation to affect the migration. The results show a visible descending of migration distances in 4 to 6 days when collagen concentration was increased (Fig. 3b, c). These data demonstrate that different content and concentration of matrix influence cell invasion, so matrix can be adjusted for different experimental design.

Evaluating the Effect of Treatment on Cell Invasion Using the Migration Index
To study the effect of a certain treatment on cell invasion, we added extra 10% FBS to a final concentration of 20% in culture medium as treatment, while using DMEM supplemented with 10%FBS as control. To reduce the interference factor of cell division, instead of using collectively migrating cells, we fluorescently labeled LLC cells, which move individually, for confocal time-lapse imaging in three dimensions. Because LLC cells move individually and are scattered in collagen, measuring maximum migration distance, i.e. the distance travelled by one cell furthest from the bead, may exaggerate the real invasiveness and may cause deviation in the data analysis. Therefore we defined a migration index considering the weights of all fast and slowly migrating cells. The migration index is calculated as the sum of all migrating cells multiplied with the distance from the bead. In this setting, fast migrating cells add more values than slowly migrating cells to the migration index, which shows the invasive capacity of the cells together. The cell number is difficult to obtain from images, so cell areas are used to represent cell numbers.
Here we used homemade macros (Additional file 1) in Fiji to measure migrating cell areas at every 10 μm away Cell coated microcarrier beads were embedded in 5 mg/ml growth factor reduced (GFR) Matrigel, 1.6 mg/ml collagen I or 0.3% agar respectively, and pictures were taken 56 h later. Scale bars, 100 μm. b Melanoma cell line M14 were grown on beads and cell invasion was monitored in a series of concentrations of collagen I gel. Five spheroids were recorded for each individual assay and migratory distance was measured in three independent experiments. Error bars represent standard deviation. c Representative pictures of M14 invasion in different concentration of collagen I for 6 days. Scale bar, 100 μm from the core. In Fig. 4a, the red circle shows the microcarrier core and green areas indicate migrating cells included in data analysis. At 72 h, cells with 20% FBS supplemented in medium seem to have larger migration areas at all distance ranges than cells in 10% medium, while the maximum distances in both groups are very close, around 350 μm (Fig. 4b). This result indicates the necessity of introducing the migration index. After computing the migration index of all time points, we found no significant difference between 10 and 20% medium, although an increasing trend was observed in 20% medium (Fig. 4c). The data reveal that the migration index calculation may be affected by increased cell proliferation, and reducing nutrients in the medium will make results of cell invasion more convincing.

The Position of Spheroids in 3D Matrix Influences Cell Invasion
During experiments using this 3D assay, we observed that spheroids might settle at the bottom of the culture plate because of the softness of the gel. When spheroids touch the bottom, most cells prefer migrating along the bottom instead of invading the collagen scaffold (Fig. 5a). This is possibly due to the low resistance in the interface between gel and bottom surface. The spheroids at the bottom cannot be included in data analysis because of exaggerated cell migration distances. If this settlement of beads at the bottom of the well occurs to most spheroids, the matrix concentration might be too low. Normally, increasing the concentration by 0.1~0.2 mg/ml can improve the viscosity of matrix during gel preparation but not reduce migration distance too much (Fig.  3b). In order to avoid beads to settle at the bottom, and to keep the matrix concentration as low as required, we tried to make a sandwich gel consisting of a bottom gel without spheroids and a top gel with spheroids. Interestingly, spheroids could be found in the interface between the two layers of gel and most cells appeared to move in this interface (Fig. 5b). A possible solution could be inverting the culture plate for 1-2 min at room temperature (Fig. 5c), which can, however, only be applied to 96-well format as the well is small enough to retain the viscous liquid. Making use of fluidity of the gel at a certain temperature is another solution. When a low matrix concentration is used, the gel mixed with the cell-coated beads may be pipetted carefully at room temperature to keep the beads in the gel by increasing the viscosity. A proper position of spheroids in the matrix will allow cells to migrate evenly to all directions (Fig. 5d), which shows the innate cell invasion capacity in matrix. Here we show incorrect positions of spheroids in matrices and possible solutions to obtain proper positions for good experiments.

Discussion
This microcarrier-based spheroid invasion assay provides a powerful approach to assess cell biological behavior in a 3D format, including motility, invasion, angiogenesis, morphological changes, and cell-cell interaction. This method has been used to study the effect of specific gene on cell migration and invasion [14,15]. It can also be adapted to investigate endothelial cells sprouting and vessel formation [16][17][18]. After microscopy the gel with invading cells can be fixed for The application of microcarrier beads is a fast and highly reproducible way to make spheroids. It allows adherent cells, especially cells which cannot form aggregates with regular shape, to be embedded in matrix as spheroids for invasion study. The microcarrier beads we used in this assay are made of cross-linked dextran coated with a thin layer of denatured collagen. The coating provides a good culture surface for cells to attach and grow. Considering different cell types, beads can be coated with other attachment factors to fit demanding culture conditions.
The matrix selection may lead to different results of cell invasion. Collagen I is the main component of ECM and forms fibrillary networks to withstand stretching. Matrigel is extracted from Engelbreth-Holm-Swarm murine sarcoma and consists of laminin, collagen IV, heparin sulfate proteoglycans, entactin and a few growth factors, which simulates the ECM complex [19]. Here we used growth factor reduced Matrigel so as to decrease the impact of these factors on cell proliferation and invasion. To examine cell invasiveness both of the matrices mentioned above can be used in this method. Importantly, other types of matrices extracted from animal or human tissues can be used as an alternative as long as the matrix can solidify at 37°C [20]. Moreover, modification of the matrix by adding ECM components enables fine tuning of the conditions in which the cells reside. Our results indicate that the content and concentration of matrix will affect cell performance and therefore results. For appropriate use of this method we recommend to choose or modify the matrix according to the experimental design, and to try different concentrations or compositions if necessary.
In this study we dilute matrix with serum-free medium to generate a determined concentration. On top of the gel culture medium is added to maintain cell growth and prevent gel from drying out. To examine if agents added to the culture medium would influence cell behavior, we compared cell invasion when exposed to 10 or 20% serum. Although a higher serum concentration did not increase the outcome significantly, a positive trend was observed because of enhanced cell proliferation with or without migration. Cell proliferation is inevitable but can be reduced by decreasing the concentration of serum or other growth promoting supplements. Our results indicate that nutrients or treatments in the medium can penetrate into the gel and act on the cells. So, to test different treatments in this 3D invasion assay, growth factors, inhibitors or drugs can be supplemented either in the medium or directly in the gel. Another interesting finding is that the position of a spheroid in the matrix has an impact on cell moving paths. When spheroids sediment at the bottom of a culture vessel, most cells move along the interface between culture vessel and matrix; while if spheroids are in the middle of two gel layers as a "sandwich", most cells move between these two gel layers. These observations demonstrate that cells tend to migrate along the path of the least resistance, and researchers need to pay attention to this issue when using this method or similar 3D settings.
Although the microcarrier-based 3D invasion assay has a broad application, the presence of a carrier limits the use to study tumor cell behavior in a spheroid with an anoxic core. Moreover, to study infiltration of tumor cells into a spheroid of normal cells, or to study infiltration of immune cells into a tumor cell spheroid, the assay needs to be extended. A multilayer spheroid can be created over time for this purpose by adjusting the matrix to inhibit migration away from the bead but allow growth. Notably, the described microcarrier-based method cannot be applied to non-adherent cells.

Conclusions
This study displays a highly reproducible and less timeconsuming 3D invasion assay together with practical quantifications and data analysis. Introducing microcarriers to generation of spheroids contributes to uniformity control, short experimental period and the use of a broad range of cell types. We also show time-lapse imaging of cell movement in 3D, which allows visualization of the whole process and advanced analysis. In conclusion, this microcarrierbased 3D invasion assay is a powerful tool to study cell invasion in vitro.

Imaging System and Climate Control Configuration
As time-lapse imaging may take hours to days, a screening system, e.g. confocal microscope, integrated with a cell incubation setup is indispensable. Here we show our imaging workspace setup as an example (Fig. 6). A sealed Perspex box was built on the microscope to maintain temperature. The box is heated by a heating unit Fig. 6 Climate controlled confocal microscopy configuration for time-lapse imaging. a Temperature controller. b Heating unit. c Gas wash bottle. d Motorized stage with an experimental plate on top. A tube with humidified airflow containing 5% CO 2 is connected to the plate through a ventilation duct. A sensor in the box is connected to the temperature controller normally set to 37°C. A 5% CO 2 /air mixture is supplied through a gas wash bottle for humidification, and the flow goes directly to the cell culture plate. Medium evaporation needs to be tested to optimize air flow before experiment. Since cells move in three dimensions in matrices, the microscope with z stacks scanning is recommended for continuous screening with the climate control system. A standard microscope can be used for manual image acquisition as the focus needs to be adjusted over time.

Preparation of Microcarrier Beads
Cytodex Microcarrier beads (C3275, Sigma-Aldrich) were hydrated in PBS for at least 3 h at room temperature. After beads settlement, discard the supernatant and add fresh Ca 2+ and Mg 2+ free PBS to a stock concentration of 50 ml/g. The beads in PBS are sterilized by autoclaving at 120°C for 20 min and can be stored at 4°C. Upon use, mix bead suspension in stock thoroughly and pipette 1 ml to a 15 ml Falcon tube. Centrifuge the mixture at 400 g for 5 min and aspirate the supernatant carefully. Re-suspend beads in a volume of 10 ml culture medium to make the final suspension.

Cell Culture
Human melanoma cell lines (BLM, M14 and Mel57), colorectal cancer cell lines (SW480 and CACO2) and mouse Lewis lung carcinoma (LLC) cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) under conditions of 5% CO 2 at 37°C.

Preparation of Cell Spheroids with Microcarrier Beads
Cells were suspended in culture medium at a density of 2~5 × 10 5 cells/ml. Add 1 ml cell suspension and 1 ml bead suspension to a round bottom tube with snap cap (352,059, Corning). Place the tube in a 37°C incubator with 5% CO 2 for 6 h and gently shake the tube manually every 2 h to allows cells to evenly distribute on the beads. Manually shaking cannot be replaced by a shaker as most cells will not adhere under continuous shaking. After 6 h of incubation, transfer the mixture (2 ml) to a 6-well plate or a 35 mm petri dish and incubate for 1 to 2 days until most beads are fully covered with cells. Gently clap the culture plate to let spheroids detached for further use. The cell number required to obtain a confluent coverage of beads vary for different cell lines, and should be tested beforehand.

Embedding Spheroids into Matrix Gel
Spheroid suspension was transferred to a Falcon tube and left for 5 min allowing spheroids to settle. Aspirate all culture medium carefully and add the same amount (2 ml) of DMEM to re-suspend spheroids. Prepare a certain concentration of matrix with collagen (option A), Matrigel (option B) or agar (option C). The recommended concentration of collagen type I is 1.4-2.3 mg/ ml according to the quantity of collagen I in human fresh tissue [21]. For Matrigel, the concentration that forms a solid gel and allows cells to invade properly in 2 to 3 days (e.g. 4-5 mg/ml) should be determined in pilot assays before further experimentation, as it may vary between companies and batches. Here we show the volume of reagents for duplicates preparation in a 24-well format. (iii)Add 100 μl spheroids suspension to the Eppendorf tube. Slowly add 500 μl 0.6% agar and gently pipette up and down to mix well. The final concentration of agar is 0.3%. Dispense 400 μl of the mixture in each well without air bubbles and incubate the plate at room temperature for 20-30 min until a solid gel formed. (iv) Add 500 μl warm (37°C) culture medium carefully along the side wall onto the gel. To investigate treatment effects, agents can be mixed in the culture medium before adding to the gel.

Imaging Cell Invasion in Matrix
Cell invasion can be monitored by time-lapse microscopy (option A) for several days. It requires a climate control system to keep cells alive during imaging.
Here we use a confocal microscope installed with a cell culture box. A sealed Perspex box is built on the microscope to maintain temperature. Assemble the heating unit to warm the air inside the Perspex box and the motorized stage where culture plate is placed. A 5% CO 2 /air mixture is supplied through a heated gas wash bottle for humidification, and it goes directly to the cell culture plate or chamber on the motorized stage (Fig. 6). The flow rate needs to be low to prevent evaporation of medium in the plate, and it can be adjusted based on the frequency of air bubbles in the gas wash bottle. In the absence of a climate controlled configuration, it is also possible to image cell dispersion and invasion manually (option B). Image acquisition in bright field or fluorescence can be done in this setting, and several time points were recorded.
(A) Time-lapse imaging (i) Switch heating unit on and set it to 37°C before imaging to ensure the heating is stable.

Quantification of Migratory Parameters and Data Analysis
Several methods can be performed to quantify migratory parameters under different conditions. The maximum migrating distance (option A) or the average of maximum migrating distance (option B) is applied when cells migrate cohesively and very few cells move far away from the cell cluster (Fig. 7a,b). Here we use the AxioVision image analysis module as an example to measure these parameters which can alternatively be done in Fiji [22] or similar software. Some cell lines move individually or follow a path created by front cells, thus show spotted or radiating/sprouting dispersion respectively. In this case a migration index (option C) can be applied to determine cell invasion characteristics. The migration index is defined as the sum of all migrating cells multiplied with the distance from the bead. If time-lapse imaging is conducted, moving trajectories of individual cells can be tracked manually or with tracking software from which migration distances and velocity are calculated (option D).
(A) Measuring maximum migrating distance (i) Open file at a time point in AxioVision. Under "Measure" menu select the "Circle" tool. (ii) Draw a circle matching the bead to measure size of bead (Fig. 7a, red circle). From the center draw another circle involving all migrating cells (Fig. 7a, white circle). (iii)Calculate maximum migrating distance at this time point. Max migrating distance (μm) = radius of migrating front circleradius of bead circle (B) Measuring average of maximum migrating distance (i) Open file at a time point in AxioVision.
(ii) Under "Measure" menu select the "Circle" tool. Draw a circle matching the bead to measure size of bead (Fig. 7b, red circle).
(iii)Under "Measure" menu select the "Curve" tool. Draw a curve along the migrating front to generate a convex polygon (Fig. 7b, yellow curve) to measure the perimeter. Only the perimeter of convex polygon can be used to calculate the radius with this formula [23]. A concave polygon extends the perimeter which causes incorrect result. (iv) Calculate average of maximum migrating distance at this time point (Fig. 7b, white circle). Avg. max migrating distance (μm) = (perimeter of the polygon/2π)radius of bead circle (C) Computing migration index