Time-Resolved Fluorescence Imaging and Analysis of Cancer Cell Invasion in the 3D Spheroid Model

Summary

Presented here is a protocol for the fabrication of a spheroid imaging device. This device enables dynamic or longitudinal fluorescence imaging of cancer cell spheroids. The protocol also offers a simple image processing procedure for the analysis of cancer cell invasion.

Abstract

The invasion of cancer cells from the primary tumor into the adjacent healthy tissues is an early step in metastasis. Invasive cancer cells pose a major clinical challenge because no efficient method exist for their elimination once their dissemination is underway. A better understanding of the mechanisms regulating cancer cell invasion may lead to the development of novel potent therapies. Due to their physiological resemblance to tumors, spheroids embedded in collagen I have been extensively utilized by researchers to study the mechanisms governing cancer cell invasion into the extracellular matrix (ECM). However, this assay is limited by (1) a lack of control over the embedding of spheroids into the ECM; (2) high cost of collagen I and glass bottom dishes, (3) unreliable immunofluorescent labeling, due to the inefficient penetration of antibodies and fluorescent dyes and (4) time-consuming image processing and quantification of the data. To address these challenges, we optimized the three-dimensional (3D) spheroid protocol to image fluorescently labeled cancer cells embedded in collagen I, either using time-lapse videos or longitudinal imaging, and analyze cancer cell invasion. First, we describe the fabrication of a spheroid imaging device (SID) to embed spheroids reliably and in a minimal collagen I volume, reducing the assay cost. Next, we delineate the steps for robust fluorescence labeling of live and fixed spheroids. Finally, we offer an easy-to-use Fiji macro for image processing and data quantification. Altogether, this simple methodology provides a reliable and affordable platform to monitor cancer cell invasion in collagen I. Furthermore, this protocol can be easily modified to fit the users’ needs.

Introduction

During cancer progression, cancer cells can acquire a motile and invasive phenotype, enabling them to escape the tumor mass and invade into the surrounding tissues¹. Eventually, these invasive cancer cells can reach and grow inside secondary organs, a process called cancer metastasis¹. Metastasis causes more than 90% of cancer-related deaths². One reason for this is that, while localized tumors are clinically manageable, no efficient methods exist for the elimination of invasive cancer cells once metastatic spreading has occurred. Therefore, the emergence of invasive cancer cells and the transition from a localized to an invasive disease is posing a major clinical challenge. Determining how cancer cells initiate and sustain an invasive behavior may lead to the development of novel potent therapies.

The 3D spheroid model is an ideal platform to investigate the motile behavior of cancer cells under controlled, yet physiologically relevant conditions³. Indeed, in this assay, spheroids of cancer cells are embedded inside extracellular matrix (ECM), for example collagen I, which mimics a simplified tumor. Then, imaging is used to visualize the invasion of cancer cells from the spheroid into the collagen matrix. However, multiple challenges limit this procedure.

The first challenge occurs at the embedding step, where the liquid collagen matrix can spread across the dish surface, causing the spheroid to touch the bottom of the dish. Consequently, cells from the spheroid spread on the two-dimensional (2D) surface, breaking the three-dimensional (3D) spheroid morphology. Increasing the volume of collagen is an efficient, but costly solution. To prevent cells from spreading on the 2D surface, while maintaining a minimal volume of collagen, we developed a spheroid imaging device (SID) by bounding a 1 mm-thick, 3-hole polydimethylsiloxane (PDMS) insert onto a glass bottom dish.

The second challenge of the spheroid assay is the labeling of cancer cells in spheroids, which is limited by the poor penetration of antibodies and fluorescent dyes, an effect that increases with the spheroid size. While the ideal solution for labeling cells is the establishment of cell lines stably expressing fluorescent protein(s), this option is mostly restricted to immortalized cell lines and is limited by the availability of fluorescent protein chimeras. Here, we describe an optimized protocol for immunofluorescence staining of fixed spheroids, as well as the efficient use of a cytoplasmic dye to label cells immediately before embedding the spheroid.

The third challenge of the spheroid assay is the lack of simple Fiji macros for semi-automated quantification of cell invasion over time. To address this challenge, we describe a simple methodology to analyze the spheroid area over time. We illustrate the advantages of this protocol using the 4T1 and 67NR cell lines as examples.

Protocol

1. Fabrication of a Spheroid Imaging Device (SID) to optimize spheroid embedding (Duration 1 day)

1. Create the spacer using a 3D printer (Figure 1A, B and Supplementary File 1).
2. Weigh out a 10:1 (wt/wt) ratio of base polymer:crosslinker in a plastic cup [e.g., 20 g of ethylbenzene base polymer and 2 g of silicone resin crosslinker to create polydimethylsiloxane (PDMS)].
3. Thoroughly mix the PDMS solution in the plastic cup using a disposable pipette.
4. Place the plastic cup into a vacuum chamber to remove the air bubbles from the mixture. Quickly release the vacuum pressure to remove the small amount of air trapped at the surface of the mixture and dissipate the remaining air bubbles.
5. Incubate the 3D printed spacer at 100 °C for 5 min to increase its flexibility.
6. Clean the two glass plates that will be used to construct the PDMS mold thoroughly by wiping them with 100% isopropanol. If the glass plates were previously used to cast PDMS, be sure to remove any remaining old PDMS by gently scraping the glass plates with a razor blade and wiping them with 100% isopropanol.
7. Construct the mold by placing the 3D printed spacer flush in-between the two clean glass plates.
8. Seal the mold using large binder clips on the outside edges of the glass plates. Place two binder clips on the bottom edge and one on the top corner.
9. Inspect the top part of the mold to ensure that the spacer is flush with the glass plates. This guarantees that no deformations will be present and that an even sheet of PDMS will be created.
   NOTE: If the resulting sheet is not of uniform thickness, the clips need to be adjusted slightly.
10. Cut the tip of a disposable pipette (~2 cm from the tip) and add the PDMS mixture at a slow and constant rate to the top left corner of the mold. Pour the mixture slowly to prevent the creation of large air pockets.
11. Place the mold into a vacuum chamber to remove air bubbles that formed during the pour.
12. Cure the PDMS by incubating the mold at 100 °C for 1 h.
13. Retrieve the mold from the incubator and allow it to cool to touch.
14. Remove the binder clips and glass plates from the spacer containing the cured PDMS.
15. Use a razor blade to cut through the seal that was created on all four sides of the mold between the spacer and the glass plate. With all four sides cut into, begin pulling apart the mold to reveal the PDMS sheet in the spacer.
16. Carefully peel the new PDMS sheet off of the spacer using tweezers.
17. On a cutting mat, punch out 17.5 mm diameter PDMS disks from the sheet, and then punch three evenly distributed 5.5 mm diameter holes, using different size biopsy punches. Here these 3-hole PDMS disks are referred as “inserts” (Figure 1C).
    NOTE: Nonuniform cuts made into the PDMS, or a faulty bind via plasma treatment, may result in the future leaks.
18. Clean each insert by gently removing any dust particles using tape.
19. Stick the inserts onto a piece of double-sided tape and wrap the tape around the lid of a 10 cm Petri dish.
20. Place the lid of the Petri dish in the plasma machine, along with the open 35 mm glass bottom dishes.
21. Activate the surface of the inserts and of the glass via plasma treatment for 1 min at 300 mTorr. A hand-held plasma wand can be used.
22. Using tweezers, quickly attach the upside, i.e., treated side, of one insert (Figure 1D) to the glass part of a glass bottom dish (Figure 1E). Repeat for all dishes.
23. Use pointer finger and thumb to apply an even pressure while rotating the glass bottom dish. This will ensure stable securement of the insert to the glass bottom dish.
24. Incubate the SIDs (Figure 1F) at 60 °C for 20 min to strengthen the adhesion between glass and PDMS.
25. Perform a second round of plasma treatment on the SIDs, using the same settings as in step 1.21 or with the hand-held wand. This will render the free PDMS surface adhere to the poly-L-lysine in the coating solution (see 1.26).
26. Freshly prepare the following solutions.
    1. Coating solution: 1x PBS containing 0.01% (vol/vol) poly-L-Lysine [e.g., add 10 µL of 0.1% (vol/vol) poly-L-Lysine to 90 µL of 1x PBS].
    2. Crosslinking solution: Distilled water containing 1x (vol/vol) glutaraldehyde [e.g., add 10 µL of 10x glutaraldehyde to 90 µL of distilled water].
    3. Storage solution: 1x PBS containing 10x (vol/vol) Penicillin-Streptomycin [e.g., add 1 mL of 100x Penicillin-Streptomycin to 9 mL of 1x PBS].
27. Add 35 µL of coating solution per each hole and incubate for 1 h at room temperature.
28. Aspirate the poly-L-Lysine and rinse the entire SID 3 times with distilled water.
29. Add 35 µL of crosslinking solution per each hole incubate for 30 min at room temperature.
30. Aspirate the crosslinking solution and rinse the entire SID 3 times with distilled water.
31. Add 70% ethanol into each SID and place under ultraviolet (UV) light for 30 min.
32. Under the hood, aspirate ethanol and rinse the SID 3 times with distilled water.
33. Add 2.5 mL of storage solution per each SID.
    NOTE: At this stage, the SIDs can be stored at 4 °C for a week. It was observed that the binding strength between PDMS and glass decreases over time. Beyond a week, users are recommended to test the SIDs for potential leakage before use. To do so, aspirate the storage solution and add 35 µL of 1x PBS into each hole. After use, PDMS inserts can be peeled off the glass bottom dishes. To achieve optimal cleaning of the glass bottom dishes, several washes with isopropanol and hydrochloric acid should be used to remove PDMS residues.

2. Spheroid formation and embedding into collagen (Duration 4 days)

NOTE: For live imaging of spheroids, longitudinally or in time-lapse videos, use a cell line expressing a cytoplasmic and/or nuclear fluorescent protein. If such a cell line is available, follow the steps described in this section. Alternatively, in the section 3, a protocol is proposed to label cancer cells in spheroids using a cytoplasmic dye.

1. Form spheroids of 4T1 and/or 67NR cells using the hanging drop technique, as previously described^4,5,6,7, with 3,000 cells/40 µL droplet and a 3-day incubation time. Add the bovine atelocollagen I solution last and maintain all solutions on ice, at all time.
2. Identify the correctly formed spheroids using a bright-field microscope.
3. Fill a 15 mL conical tube with 8 mL of pre-warmed complete medium.
4. Collect spheroids with a P1000 pipette and transfer to the 15 mL conical tube. Wet the pipette tip by pipetting some complete medium in and out to prevent spheroids from sticking to the inside walls of the pipette tip and limit spheroid loss.
5. Allow spheroids to sink to the bottom of the tube and carefully wash the spheroids by exchanging the medium. Repeat twice.
6. Prepare a 5 mg/mL collagen I solution according to the manufacturer’s recommendations (alternate gelation procedure, see exemplary calculation below). Replace the distilled water with complete medium. When calculating the volume of medium to be used, take into account that spheroids will be added in 20 µL of complete medium [e.g., for one SID, 3 x 30 + (3 x 30) x 20 % = 108 µL of 5 mg/mL collagen I is needed (prepare 20 % extra to account for pipetting loss when handling viscous fluids); 22 µL of complete medium + 10.8 µL of 10x PBS + 1.2 µL of 1 M NaOH + 54 µL of 10 mg/mL collagen I stock].
7. Collect all spheroids in 20 µL of medium, using a P200 pipette, and add them to the collagen I solution prepared in step 2.6.
8. Slowly mix the solution by pipetting up and down to avoid heterogeneity in the collagen I concentration, limiting bubble formation. Keep the solution on ice.
9. Remove the storage solution from the SID(s) and wash 3 times with 3 mL of 1x PBS. After the final wash, leave the SID(s) dry so not to dilute the collagen I solution.
10. Start a timer and dispense 30 µL of the collagen I solution containing one spheroid into one of the three holes of the SID. Visually ensure that a single spheroid is contained in the 30 µL.
11. Repeat step 2.10 twice more to fill all the 3 holes of a SID.
12. Use a 10 µL pipette tip to re-center the spheroid if it is located close to the PDMS border. If two or three spheroids end up dispensed in one of the holes, same pipette tip can be used to separate the spheroids from each other. Stop the timer.
    NOTE: Frequently inverting of the SID upside-down and upside-up, throughout the period of the collagen I polymerization and solidification, ensures that the spheroid is positioned at the vertical center of the ECM layer, and prevents invasion of cells in 2D. The frequency of inverting (flipping) should be maximized. As the flipping frequency is controlled by the spheroid “dispensing time” measured in 2.10-2.12, the dispensing time should be minimized. In our lab, dispensing time is 2 min on average.
13. To vertically center the spheroid in the collagen layer, flip the SID upside-down and incubate at 37 °C for dispensing time.
14. Flip the SID upside-up and incubate at 37 °C for dispensing time.
    NOTE: The length of time that the spheroids spend in the upside-down orientation should be equal to the time spent in the upside-up orientation.
15. Repeat steps 2.13 and 2.14 for 30 min, until the collagen I polymerizes.
16. Add 2.5 mL of complete medium/SID and if needed, acquire an image of the spheroids for the initial timepoint.
17. Repeat steps 2.10-2.16 if multiple SIDs are used.

3. Fluorescence labeling of spheroids

1. Live imaging (Duration 6-7 days)
   NOTE: If a cell line expressing a cytoplasmic and/or nuclear fluorescent protein is available, follow the steps described in the section 2. Alternatively, for cytoplasmic labeling, the following protocol is proposed.
   1. Follow steps 2.1-2.5 of the protocol described in the section 2.
   2. Dilute the cytoplasmic dye to 25 µM in 200 µL of serum-free medium.
      NOTE: While a red cytoplasmic dye is used in this experiment, any other available color should be suitable. Cancer cells were labeled inside spheroids using nuclear dyes diluted to 20 µM in 200 µL of serum-free medium (see Table of Materials).
   3. After the final wash, resuspend spheroids in the cytoplasmic or nuclear dye solution and incubate at room temperature for 20 min, protected from light.
      NOTE: With this approach, labeling of the cells in the spheroid center will not be efficient. To achieve labeling of all cells, incubation can be extended overnight, by placing the dish on a rocker. Alternatives are described in Discussion.
   4. Wash spheroids 3 times in complete medium.
   5. Proceed to steps 2.6-2.17 of the protocol described in the section 2.
   6. Image spheroids via time-lapse imaging every 10 min for 24-72 h (Figure 2), or via longitudinal imaging, daily, for up to 7 days (Figure 3). Use a laser scanning confocal microscope, 10x air objective (0.4 numerical aperture and 3.1 mm working distance), 1024 x 1024 pixels, exposure time 8 µs/pixel, pinhole 90 µm, 6 x 15 µm z-steps and 3 fields of view-each containing one spheroid.
      NOTE: For time-lapse imaging, use minimal laser power and equip the microscope with an environmental chamber with temperature, humidity and gas control. Culture the embedded spheroids at 37 °C for > 8 h or overnight prior to imaging to minimize the time on the microscope.
2. Immunofluorescence staining of spheroids (Duration 2 days)
   NOTE: The procedure described here is adapted and optimized from previously published protocols^8,9. This method can be used after the protocol outlined in the sections 2 and 3.1.
   1. Freshly prepare the following solutions.
      1. Fixing solution: 1x PBS containing 4% PFA (vol/vol) [e.g., add 5 mL of 16% (vol/vol) PFA to 15 mL of 1x PBS].
      2. Fixing and permeabilizing solution: 1x PBS containing 4% PFA (vol/vol) and 0.5% Triton X-100 (vol/vol) [e.g., add 50 µL of Triton X-100 to 10 mL of fixing solution].
      3. Blocking solution: 1x PBS containing 1% FBS (vol/vol) and 1% BSA (wt/vol) [e.g., dissolve 100 mg of BSA in 10 mL of 1x PBS, then add 100 µL of FBS].
      4. Washing solution: 1x PBS containing 0.05% Tween 20 (vol/vol) [e.g., add 25 µL of Tween 20 to 50 mL of 1x PBS].
   2. Remove the culture medium from the SID(s) and wash once with warm 1x PBS.
   3. Add 2 mL of fixing and permeabilizing solution per SID and incubate at room temperature for 5 min.
   4. Remove the fixing and permeabilizing solution and add 2 mL of fixing solution per SID and incubate at room temperature for 20 min.
   5. Wash 3 times with the washing solution.
   6. Add 2 mL of blocking solution per SID and incubate at 4 °C for 24 h with mild shaking.
      NOTE: At this step, samples can be incubated at 4 °C, over the weekend. Please note that a longer blocking time might interfere with the immunofluorescence labeling procedure.
   7. Dilute primary antibodies in blocking solution.
   8. Add 150 µL of blocking solution with primary antibodies/well in a 48-well plate.
   9. Using fine tweezers, carefully detach the plug of collagen I containing a spheroid and transfer into a well. Repeat if multiple spheroids are labeled. Wipe any liquid that might remain on the tweezers when working with different antibodies, to prevent contamination.
   10. Incubate overnight at 4 °C with mild shaking.
   11. Add 300 µL of washing solution to empty and full wells.
   12. Carefully transfer the plug of collagen I containing a spheroid into a “wash” well.
   13. Wash 3 times with washing solution for 2 h, at room temperature and with mild shaking.
       NOTE: Transferring the plugs of collagen I containing a spheroid, instead of aspirating the solutions, can help reducing sample loss and sample damage.
   14. Dilute secondary antibodies, 4’,6-diamidino-2-phenylindole (DAPI) and/or phalloidin in blocking solution.
   15. Add 150 µL of blocking solution with secondary antibodies, DAPI and/or phalloidin per well.
   16. Using tweezers, carefully transfer the collagen plug containing the spheroids into the well. Repeat if multiple spheroids are labeled.
   17. Incubate at room temperature for 1 h with mild shaking.
   18. Repeat steps 3.2.11 and 3.2.12.
   19. Wash 3 times with washing solution for 30 min, at room temperature with mild shaking.
   20. Using a razor blade, cut a 3-hole PDMS insert into three parts so that each part contains a hole.
   21. Place two pieces of PDMS onto a microscope slide.
   22. Add a drop of mounting solution using a P200 tip with the end cut off. Prevent bubble formation.
   23. Carefully transfer a collagen I plug into each hole.
       NOTE: If the spheroid was positioned at the top of the collagen plug, invert the collagen plug so that the spheroid ends up closer to the glass coverslip.
   24. Place a coverslip on top and seal using tape.
   25. Let the sample dry at room temperature for 10 min, protected from light.
   26. Store samples at 4 °C, protected from light until imaging.

4. Image processing to analyze cancer invasion over time

NOTE: The format required for this macro is a single-channel x,y,t image saved as a .tiff file.

1. If multiple z-slices were acquired (i.e. x,y,z,t image), open the image in Fiji and select Image | Stacks | Z Project. It is recommended to use the Max Intensity option. Alternatively, a single z-slice can be used to run the macro. Save the image as a .tiff file.
2. Create a separate “Processing” folder on the Desktop.
3. Select File | Save As | Image Sequence. Use the TIFF format, update the digits number according to the number of frames, check the box to use slice label as file name and select Ok. Select the “Processing” folder created in step 2.
4. Open one image from the “Processing” folder. It should be an x,y image, corresponding to a single timepoint.
5. Select Image | Adjust | Auto Threshold. In the dropdown menu select the method Try all and check the box for white objects on black background.
6. A montage image appears showing the result for each automated thresholding method.
7. Identify the best automated thresholding method, for example RenyIEntropy.
8. To confirm the choice for the automated thresholding method, open any other image from the “Processing” folder and test the chosen thresholding method using Image | Adjust | Auto Threshold.
9. Close all images.
10. Download the SpheroidAreaTime macro (Supplementary File 2).
11. Open the Fiji macro by drag-and-drop.
12. In line 58, update the automated thresholding method as needed.
13. In line 62, update the size range according to the cell dimensions.
14. Select Run.
15. Select the “Processing” folder and type in “Processing” as the Parent folder, then select Ok.
16. Once the run is over, save the table “Summary”. The first column indicates the name of the image; the third column indicates the spheroid area; the sixth, seventh and eight columns specify the parameters for an ellipse fitted onto the spheroid.
17. The “Processing” folder now contains a processed image for each time point, with the extension _SpheroidArea.
    NOTE: If the spheroid center is dim, fill it with white using the selection tools, for all timepoints, and then proceed to step 3. Similarly, the space around the spheroid can be filled with black to “clean” the image. If the image is clean, comment the line 61 to speed up the run.

Results

Due to its biocompatibility, PDMS is widely used for microfabrication of confining wells, stamps and molds, which revolutionized micropatterning and microfluidic devices. In the method described here, it is used to create SIDs, customizable wells that optimize spheroid embedding and imaging procedure. Figure 1 illustrates the major components used in the fabrication of the SIDs. To cast the PDMS mold, a 1-mm thick spacer is 3D printed (Figure 1A,B), placed between the two glass plates, sealed with large clips. Pouring and baking PDMS in the space between the plates forms an 1-mm thick sheet of PDMS. The SID schematic (Figure 1C) indicates the dimensions of the optimal device, however slight variation occurs in the hole distances, due to manual punching of holes using biopsy punches (Figure 1D). Figure 1D,F indicate clean circular cuts with evenly spaced holes within the PDMS insert.

Using the SIDs facilitates efficient embedding, and hence recording of the invasion of cancer cells inside collagen I using time-lapse (Figure 2, Video 1) or longitudinal (Figure 3) imaging. Despite using the same inversion frequencies for all the SIDs during the collagen I polymerization, spheroid positions inside the collagen plug will slightly vary. Therefore, it is important to use an objective with a long working distance (>1 mm). Otherwise, focusing and imaging may be difficult for spheroids positioned close to the top of the collagen plug. In contrast, spheroids positioned close to the bottom of the collagen plug, will have cells which move to the glass surface and migrate on the glass, instead of invading into the collagen I matrix. Videos of such spheroids need to be discarded. The thickness of the collagen plug, here approximately 800 µm, is controlled by the volume dispensed in each hole of the SID and adjusted to invasion distances spheroids exhibit in this protocol. Thickness of the collagen plug can be lowered by dispensing a lower volume of collagen I in each hole of the SID, when using smaller or less invasive spheroids.

For proper analysis of the invasion using the Fiji macro, it is critical for cancer cells to be properly labeled over the course of the imaging session. As noted in the step 4.17., the image processing step allows for image correction if the labeling is sub-optimal. While we illustrate longitudinal imaging over the course of 6 days (Figure 3), which requires the stable expression of cytoplasmic and/or nuclear fluorescent protein, similar longitudinal imaging could be performed over a shorter time using labeling with dyes.

Following live imaging, we present some results from the immunolabeling procedure for the epithelial cadherin (E-cadherin), cortactin and F-actin (Figure 4). In these examples, we used the 4T1 and 67NR cell lines. Figure 5 shows step-by-step illustration of the image processing procedure using the Fiji macro to measure the area of the spheroid over time.

Figure 1: Fabrication of the SIDs. (A) Top view schematic representation of the spacer. (B) 3D printed spacer. (C) Top view schematic representation of the SID. A 3-hole PDMS inserts (D) is bound to a glass bottom dish (E), creating the final SID (F). Dimensions are in millimeters. Please click here to view a larger version of this figure.

Figure 2: Live imaging of spheroids. Representative 20 h and 40 h timepoints from Video 1, maximum projection of a 4T1 spheroid. 4T1 cells were labeled using a cytoplasmic dye. Scale bar, 100 µm. Please click here to view a larger version of this figure.

Figure 3: Longitudinal imaging of spheroids. Representative micrographs (maximum projection) of a mixed 4T1/67NR spheroid imaged daily. 4T1 and 67NR cells stably express the cytoplasmic mScarlet and green fluorescent protein (GFP), respectively. Scale bar, 100 µm. Please click here to view a larger version of this figure.

Figure 4: Immunofluorescence imaging of spheroids. Representative micrographs (maximum projection) of a 4T1 spheroid fixed 2 days after embedding in collagen I. E-cadherin (cyan, A), cortactin (yellow, B) and F-actin (magenta, A and B) were labeled. Scale bar, 100 µm. Please click here to view a larger version of this figure.

Figure 5: Image processing analysis. Step-by-step illustration of the image processing procedure using the Fiji macro (A) to measure the area of the spheroid over time (B). Please click here to view a larger version of this figure.

Video 1: Representative video of a 4T1 spheroid imaged every 10 min for 46 h and 10 min. 4T1 cells were labeled using a cytoplasmic dye. Scale bar, 100 µm. Please click here to download this video.

Supplemental file 1: 3D model of the spacer (STL file). Please click here to download this file.

Supplemental file 2: SpheroidAreaTime macro. Please click here to download this file.

Discussion

The 3D printed spacer was designed to create 1-mm thick sheets of PDMS that can then be used to easily create various shapes of PDMS, as required by the experimental applications. Due to the simplicity of its fabrication and the freedom to alter the design, this method of PDMS casting was chosen for the initial design of the SID. If high volume of SIDs is required, production can be made more efficient by creating a 3D-printed mold, which already contains PDMS disks with three equally spaced holes, and reducing the process to one step. This would eliminate the need to punch out each disk, along with the subsequent three holes, and decrease the overall preparation time.

While the 3-hole punch is developed for use with 35 mm glass bottom dishes, other sizes are available which allow for more holes and hence, more spheroids to be imaged in parallel. In addition, custom-size cover glass is also commercially available, which, in combination with 3D printed holders, can allow for high-throughput spheroid assays. With such an approach, limiting factor is the speed of data acquisition- for example, in our multicolor time-lapse confocal imaging, acquiring 3D stack of a single spheroid requires approximately 2.5 minutes. Therefore, acquiring 3D stacks for 3 spheroids in the SID requires approximately 8 minutes. As a result, to maintain our preferred frequency of imaging at 10 minutes per stack, we cannot increase the number of holes in the SIDs.

To record and quantify the invasion of living cancer cells in the 3D spheroid model, bright-field imaging can be used⁵. However, fluorescence microscopy is preferred, as it provides increased contrast, and ease and precision in the image processing. If the generation of a cell line expressing a cytoplasmic and/or nuclear fluorescent protein is not possible, we propose the use of the cytoplasmic dyes. As the retention time of cytoplasmic dyes inside cells is three days in our imaging conditions, cells should be labeled following the 3-day period in hanging drop, and immediately before the embedding. Labeling of cells post-embedding may non-specifically label the collagen, and reduce cell labeling. Imaging for longer than 3 days post-embedding requires the use of a different spheroid seeding protocol¹⁰ or cell lines stably expressing fluorescent protein(s).

We successfully formed and imaged spheroids containing anywhere from 60 to 5,000 cells/drop. Small spheroids are ideal for recording invasion over multiple days, as their entire invasion area can easily fit into a single field-of-view of higher magnification (20x-30x) objectives. In addition, they can easily be labeled throughout with cytoplasmic or nuclear dyes. Finally, due to the reduced scattering, each cell in the spheroid can be visualized and segmented. However, small spheroids are barely visible with naked eyes and may require additional labeling with tissue markers. In contrast, larger spheroids are easier to handle, but more susceptible to sinking to the bottom of the dish, due to their weight. Moreover, cells in the spheroid center are not always labeled when using dyes, or visible using confocal microscopy, which has penetration depth of approximately 100 micrometers. To visualize all cancer cells throughout the large spheroids using time-lapse imaging, multiphoton microscopy can be used, offering an extra advantage of collagen fibers visualization by second harmonic generation (SHG) without the need for labeling. Also, imaging can be done with light-sheet microscopy^8,11,12, providing reduced image acquisition time and hence allowing for high-throughput spheroid assays, but also requiring more data storage and advanced image processing. If time-lapse videos are not required, our labeling procedure for fixed 3D spheroids can be further combined with optical clearing^8,10. In addition, cryosectioning of the embedded spheroids can eliminate issues with penetration of dye or antibody during labeling, as well as penetration of light during imaging. In our hands, however, successful cryosectioning was limited to early stages of invasion and non-invasive cells, due to the technical challenges in preservation of long and fragile invasive strands.

Our protocol is also compatible with the use of nuclear dyes, to label cancer cells inside the spheroids, enabling single cell tracking of time-lapse data. The Fiji plugin TrackMate¹³ can be used to automate cell tracking and extract motility parameters of single cells, such as velocity, instantaneous speed and persistence.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
1 N NaOH	Honeywell Fluka	60-014-44
10X Dulbecco’s phosphate-buffered saline (PBS)	Gibco	SH30028.LS
16% paraformaldehyde (PFA)	Alfa Aesar	43368-9M
1X Dulbecco’s phosphate-buffered saline (PBS)	Gibco	20012027
4’,6-diamidino-2-phenylindole (DAPI)	Invitrogen	D1306
48-well plate	Falcon	T1048
Alexa Fluor 647 phalloidin	Life Technologies	A20006
Anti cortactin antibody	Abcam	ab33333	1 to 200 dilution
Anti E-cadherin antibody	Invitrogen	13-1900	1 to 100 dilution
Bovine atelocollagen I solution (Nutragen)	Advanced Biomatrix	501050ML
Bovine serum albumin (BSA)	Sigma Aldrich	A4503-50G
CellTracker Red CMTPX Dye	Invitrogen	C34552
Conical tubes	Falcon	352095
Coverslips	FisherBrand	12-548-5E
Disposable container	Staples		Plastic cups
Disposable transfer pipette	Thermo Scientific	202
DMEM	Fisher Scientific	11965118
Double-faced tape	Scotch
Ethanol	Sigma Aldrich	E7023-500ML
Fetal bovine serum (FBS)	Bio-Techne	S11550
Fluoromount-G	eBioscience	00-4958-02
Glutaraldehyde	Sigma Aldrich	G5882-100mL
Hoescht nuclear stain	Thermo Fischer	62249
Isopropanol	Thermo Fischer	S25371A
MatTek dish (glass bottom dish)	MatTek Corporation	P35G-1.5-14-C
Methyl cellulose	Sigma Aldrich	M6385-100G
MilliQ water
Penicilin/streptomycin solution	Thermo Fischer	15140122
Petri dish	Corning	353003
Pipet tips	Fisherbrand	02-707
Pipets	Gilson	F167300
Poly-L-Lysine	Sigma	P8920
Primary antibodies, user specific
Rat Tail Collagen I	Corning	47747-218
Razor Blade	Personna	74-0001
Secondary antibodies, user specific
Slides	Globe Scientific	1354W-72
Sylgard 184 Silicone	Dow Corning	4019862
Tape	Scotch
Triton X100	Sigma Aldrich	10789704001
Tween 20	Sigma Aldrich	655204-100ML