In this protocol, we present the experimental procedures of a cell spreading assay that is based on live-cell microscopy. We provide an open-source computational tool for the unbiased segmentation of fluorescently labeled cells and quantitative analysis of lamellipodia dynamics during cell spreading.
Cell spreading is a dynamic process in which a cell suspended in media attaches to a substrate and flattens itself from a rounded to a thin and spread-out shape. Following the cell-substrate attachment, the cell forms a thin sheet of lamellipodia emanating from the cell body. In the lamellipodia, globular actin (G-actin) monomers polymerize into a dense filamentous actin (F-actin) meshwork that pushes against the plasma membrane, thereby providing the mechanical forces required for the cell to spread. Notably, the molecular players that control the actin polymerization in lamellipodia are essential for many other cellular processes, such as cell migration and endocytosis.
Since spreading cells form continuous lamellipodia that span the entire cell periphery and persistently expand outward, cell spreading assays have become an efficient tool to assess the kinetics of lamellipodial protrusions. Although several technical implementations of the cell spreading assay have been developed, a detailed description of the workflow, which would include both a step-by-step protocol and computational tools for data analysis, is currently lacking. Here, we describe the experimental procedures of the cell spreading assay and present an open-source tool for quantitative and unbiased analysis of cell edge dynamics during spreading. When combined with pharmacological manipulations and/or gene-silencing techniques, this protocol is amenable to a large-scale screen of molecular players regulating lamellipodial protrusions.
Lamellipodial protrusions are prominent cytoskeletal structures formed at the front of a migrating cell. In lamellipodia, polymerization of actin with the aid of the Arp2/3 complex and formins creates a fast-growing branched actin meshwork that pushes against the plasma membrane1,2. The pushing force generated by the actin meshwork physically propels the cell forward1,3,4,5. Depletion of the Arp2/3 complex or disruption of signaling pathways essential for lamellipodial protrusions often impair cell migration6, 7. Although migration of lamellipodia-deficient cells has also been reported8,9, the importance of lamellipodia in cell migration is evident as depletion of this protrusive structure perturbs the cell's ability to move through complex biological microenvironments6,10.
A major hindrance to understanding the regulation of lamellipodia in migrating cells is the natural variability in lamellipodial protrusion kinetics, size, and shape11,12,13,14. Furthermore, recent studies have demonstrated that lamellipodia exhibit complex protrusive behaviors, including fluctuating, periodic, and accelerating protrusions14,15. Compared to the highly variable lamellipodia of migrating cells6,16, lamellipodia formed during cell spreading are more uniform12. Since the protrusive activity of spreading and migrating cells is driven by identical macromolecular assemblies, which include a branched actin network, contractile actomyosin bundles, and integrin-based cell-matrix adhesions17,18, spreading cells have been widely used as a model for investigating the regulation of lamellipodia dynamics.
Cell spreading is a dynamic mechanochemical process where a cell in suspension first adheres to a substrate through integrin-based adhesions17,19,20 and then spreads by extending actin-based protrusions21,22,23. During the spreading phase, lamellipodia emanating from the cell body protrude isotropically and persistently with little to no retraction or stalling12. The most commonly used cell spreading protocols are endpoints assays, where spreading cells are fixed at various times after plating19,24. These assays, although quick and simple, are limited in their diagnostic power to detect changes in the dynamic features of lamellipodia. To determine the molecular mechanisms that control lamellipodia dynamics, the Sheetz group pioneered the use of quantitative analysis of live spreading cells and uncovered many fundamental properties of cell edge protrusions11,12,22. These studies have demonstrated that the live-cell spreading assay is a robust and powerful technique in the toolbox of a cell biology laboratory. Despite that, a detailed protocol and open-source computational tool for a live-cell spreading assay are currently unavailable for the cell biology community. To this end, our protocol outlines the procedures of imaging live spreading cells and provides an automated image analysis tool. To validate this method, we used Arp2/3 inhibition as an experimental treatment and showed that inhibiting the function of the Arp2/3 complex did not arrest cell spreading but caused a significant reduction in cell protrusion speed, as well as the stability of cell edge protrusions, giving rise to jagged cell edges. These data demonstrate that the combination of live-cell imaging and automated image analysis is a useful tool for analyzing cell edge dynamics and identifying molecular components that regulate lamellipodia.
1. Cell Seeding
NOTE: The described cell spreading protocol was performed using mouse embryonic fibroblasts (MEFs) expressing PH-Akt-GFP (a fluorescent marker for PIP3/PI(3,4)P2). This cell line was generated by genomically integrating an expression construct for PH-Akt-GFP (Addgene #21218) by CRISPR-mediated gene editing. However, other fluorescent markers that are expressed transiently or integrated in the genome can also be used in this assay. For optimal image segmentation, we recommend using fluorescent markers that are evenly distributed in the cytoplasm, e.g., cytosolic GFP.
2. Drug Incubation and Cell Recovery
3. Magnetic Chamber Preparation
4. Image Acquisition
5. Analysis of cell area, circularity and protrusion dynamics during cell spreading
6. Quantify cell edge dynamics during cell spreading using kymographs
The above protocol describes the experimental procedures for the live-cell imaging of spreading cells and a computational tool for the quantitative analysis of cell spreading dynamics. The computational tool can be used in a low- or high-throughput format to identify the molecular players regulating the actin polymerization machinery at the cell leading edge.
The schematic representation of the experimental procedures is depicted in Figure 1. The cell spreading assay was performed on immortalized mouse embryo fibroblasts stably expressing the pleckstrin homology (PH) domain of the Akt protein kinase tagged with eGFP25. The cells were detached with trypsin-EDTA and were allowed to recover in suspension for 45 minutes. During the recovery step, cells replenished their integrin receptors on the plasma membrane as indicated by the fast and synchronous attachment of the recovered cells to the fibronectin coated coverslips (Figure 2). Without the recovery period, spreading cells exhibited a broad distribution of cell size indicating a high variability in the onset of cell spreading (Figure 2A and B). Next, cells were plated on a fiducially-marked coverslip and their spreading dynamics were visualized by spinning disk confocal microscopy (schematics shown in Figure 1A – H). Throughout the image acquisition, we considered fields of view that featured cells with a signal-to-noise ratio of 2.5 or above. This was an important consideration as the subsequent image segmentation is sensitive to the cells' fluorescence intensity relative to the background. In our experiments, we acquired images every 6 seconds for 15 minutes (schematics shown in Figure 1I-J). In agreement with previous reports16, imaging at a 6 second frame rate ensured sufficient temporal resolution for capturing the dynamics of individual protrusion and retraction events, while allowing us to acquire several fields of view in parallel. The resulting time-lapse images were analyzed using the custom-build Python software (Figure 3).
An unbiased quantification of cell spreading was performed by using two distinct analytical procedures: (i) Morphodynamic profiling of spreading cells (Figure 3A, C and 4) and (ii) Kymograph analysis of cell edge dynamics (Figure 3B and 5). The analysis of cell spreading by morphodynamic profiling involves the automated detection of the spreading and fiducial cells in the field of view (Figure 3C left), followed by a frame-by-frame image segmentation and detection of the spreading cell boundary. The segmentation is performed by global intensity thresholding the individual frames. The threshold value is calculated as the local minimum between the first and second intensity modes on the image histogram26. Images with a unimodal, right-skewed histogram are segmented by the triangle thresholding algorithm27. Following cell segmentation, the morphodynamic characteristics of the spreading cells (i.e., cell area, aspect ratio, and cell circularity) were computed (Figure 3C right).
Consistent with published results12,28, cell spreading was driven by an isotropic expansion of lamellipodia as indicated by a sigmoidal shape on the representative cell area plot (Figure 3C right, blue curve and S1). The area plot showed that the cell area increased by approximately 3-fold before reaching a plateau (Figure 4A and B). Throughout the process of cell spreading, the cells remained circular (circularity = 0.70 ± 0.076) and displayed veil-like protruding cell edges, which are indicative of lamellipodial protrusions (Figure 4C).
To validate the cell spreading assay, we inhibited Arp2/3 with 100 µM CK-666 for 1 hour and 45 minutes and assessed the effect of this treatment on the cell spreading dynamics. In agreement with previous reports13, the suppression of Arp2/3 activity did not result in a significant decrease in the cell spreading speed (Figure 4A and B, pink curve). However, cell shape analysis revealed a significant difference in the circularity of control and CK-666 treated cells (Control: 0.70 ± 0.08 vs. CK-666: 0.54 ± 0.09, p < 0. 1 x 10-3) (Figure 4C). While the control cells remained circular until the plateau, Arp2/3-inhibited cells acquired a polygonal shape which was retained throughout the course of spreading (Figure 4A). Together, these experimental results demonstrate that the described cell spreading assay reveals moderate changes in cell morphodynamics caused by perturbations of the actin polymerization machinery.
While morphodynamic profiling is often sufficient to detect gross alterations in cell spreading dynamics, this analysis has limited ability to identify specific cytoskeletal components that regulate the protrusion-retraction cycles of the cell edge, prompting us to implement an unbiased Kymograph Analysis tool (Figure 5 left, dashed lines). The analysis of cell edge speed revealed a moderate but significant decrease in the average protrusion speed of Arp2/3-inhibited cells compared to control (Control: 37.1 ± 12.87 nm/s vs. CK-666: 28.7 ± 13.4 nm/s, p = 0. 9 x 10-3) (Figure 5C). Furthermore, dynamics of the cell edge of control and Arp2/3-inhibited cells were remarkably different (Figure 5A and D). The control cells protruded persistently with little to no retractions during the rapid expansion phase, which lasted about 200 seconds, and exhibited intermittent retractions during the plateau phase (Figure 5A and D). In contrast, the expansion of Arp2/3-inhibited cells was intervened by retraction events, denoted by the red dots on the graphs (Figure 5B and D). Quantification of retraction frequency showed that Arp2/3-inhibited cells retracted 26% more frequently than control cells (Control: 0.18 ± 0.22 s-1 vs. CK-666: 0.24 ± 0.19 s-1, p = 0.03) (Figure 5D). These data demonstrate the high sensitivity of the kymograph analysis in detecting mild alterations in lamellipodial dynamics.
Figure 1: The experimental workflow of a cell spreading assay. (A – H) Schematics of the cell spreading assay. (A) A 22 mm x 22 mm coverslip is coated with fibronectin diluted in PBS to a final concentration of 2.5 µg/mL. (B) A confluent 10 cm dish of PH-Akt-GFP-expressing mouse embryonic fibroblasts (MEFs) is washed with PBS and treated with 0.05% trypsin-EDTA. The trypsin-treated cells are then split into a 15 mL centrifuge tube and a 6 cm tissue culture dish, both containing cell culture media. (C) From the 15 mL centrifuge tube, 500-1000 µL is pipetted onto the fibronectin-coated coverslip. (D) The 6 cm dish and the 35 mm dish with the coverslip containing the sparsely seeded cells are placed in a 37 °C incubator overnight. Once polarized, these cells will provide the frame of focus for the cell spreading acquisition. (E) An hour before image acquisition, the 6 cm dish's media is replaced with phenol-red free DMEM supplemented with HEPES and the drug of interest. After 1 hour, the cells are treated with 0.05% trypsin-EDTA and transferred to a 15 mL centrifuge tube (Tube A) containing the HEPES/drug-supplemented phenol red free DMEM. The cells in Tube A are then further diluted in another 15 mL centrifuge tube (Tube B), which is placed in the incubator for 45 minutes. (F) As the cells recover, the magnetic chamber is prepared from bottom to top: first the bottom plate is placed on a flat surface, then the coverslip with the polarized cells, the silicone gasket, the main body of the chamber, and finally the transparent cover are laid on top. (G) 1 mL of drug-supplemented phenol red free DMEM is pipetted into the magnetic chamber, which is then brought to the microscope stage. A CFI Plan Apo Lambda 60X Oil objective is selected for image acquisition. (H) The transparent cover is removed and 500 µL of Tube B's contents are pipetted into the magnetic chamber. (I) For image acquisition, appropriate fields of view will contain green "halos", which are suspended cells that have not yet attached to the coverslip. (J) The cells are imaged for 15 minutes. Please click here to view a larger version of this figure.
Figure 2: The effect of recovery time on cell spreading. Cells maintained in suspension for the indicated time (cell recovery step in the protocol) were plated on fibronectin-coated coverslips for 15 minutes, fixed with 4% paraformaldehyde and imaged by phase contrast microscopy. (A) Top panels: phase contrast images acquired with a 20X objective. Bottom panels: watershed-segmented cell masks with the cell areas color-coded. (B) Quantification of cell area. Please click here to view a larger version of this figure.
Figure 3: Graphical User Interface (GUI) and the working principles of image processing and analysis software. (A) The GUI of the "Cell spread area" tab. Refer to Step 5.3 for instructions. (B) The GUI of the "Kymograph generator & analysis" tab. Refer to Step 6 for instructions. (C) The image processing pipeline of the software. The software first identifies spreading cells (labeled by a green bounding box) in the whole field of view. The spreading cells are identified based on their intensity value, circularity, and aspect ratio. The identified spreading cells are then segmented frame-by-frame using global intensity thresholding. Each binary mask is processed by median filtering and binary hole filling followed by morphological closing to smoothen the cell edge. The red outline corresponds to the segmented cell boundary. The cell's area, aspect ratio, and circularity are extracted from the binary cell map. The graph shows the area and circularity of a representative cell over time. Upon cell seeding, cells do not start spreading immediately, giving rise to the lag phase seen on the graph. Following the lag phase, cells spread rapidly during the rapid expansion phase and eventually reach a plateau phase. Please click here to view a larger version of this figure.
Figure 4: Representative results of cell spread area analysis upon Arp2/3 inhibition. (A) Representative images of PH-Akt-GFP-expressing MEFs spreading on a fibronectin-coated coverslip over the course of 3 minutes. The red line indicates the cell boundary extracted by the cell segmentation algorithm. Top panels: 0.1% DMSO-treated cells. Bottom panels: 100 µM CK-666 (Arp2/3 inhibitor)-treated cells. (B) A graph showing cell spread area over time. The cell spread area was quantified as fold changes relative to the average cell spread area of control cells. Blue and pink lines represent control and Arp2/3-inhibited cells, respectively. The shaded regions indicate the upper and lower standard deviation of the cell spread area. (C) A bar plot with individual data points showing the average cell circularity of control and Arp2/3-inhibited cells. All error bars represent standard deviation. *, p < 0.05, **, p < 0.01, ***, p < 0.001, n.s. (not significant, p > 0.05) as detected by parametric student t-tests. Please click here to view a larger version of this figure.
Figure 5: Representative results of kymograph analysis of spreading cells upon Arp2/3 inhibition. (A – B) Representative images and kymographs extracted from 0.1% DMSO-treated (control) cells and 100 µM CK-666-treated (Arp2/3 inhibitor) cells. Left panels: inverted grayscale images of a control and Arp2/3-inhibited cell. The dashed lines correspond to the pre-defined lines where kymographs were extracted. Right panels: kymographs are extracted along the dashed lines shown on the grayscale images. The plot boundaries are color-coded to match the dashed lines on the grayscale images. The dashed line in each plot indicates the slope of the curve from which the average protrusion speed was calculated. To pinpoint the plateau phase, a logistic growth curve was fitted to the data points and the plateau was derived from the parameter, c (Supplementary Figure 1). The red dots denote the retraction events. (C) A bar plot with individual data points showing the average protrusion speeds of the control and Arp2/3-inhibited cells. All error bars represent standard deviation. (D) A bar plot with individual data points showing the frequency of retraction events of the control and Arp2/3-inhibited cells. (C – D) *, p < 0.05, **, p < 0.01, ***, p < 0.001, n.s. (not significant, p > 0.05) as detected by non-parametric Mann-Whitney tests. Please click here to view a larger version of this figure.
Supplemental Figure 1: A representative kymograph and the curve fitting result. A kymograph extracted from a spreading cell. The blue dashed line corresponds to the distance of the cell edge relative to the first frame. The red solid line corresponds to the fitted curve. To increase the confidence of the curve fitting, the raw data points were first smoothed by a Savitzky-Golay filter. After the curve-fitting, the parameter c, from the logistic equation, was used to identify the plateau point. The raw data point closest to c is designated as the plateau point. Please click here to download this File.
Supplemental Files. Please click here to download this File.
The described cell spreading assay allows for the continuous tracking of morphological changes (e.g., cell size and shape) and cell edge movements (i.e., protrusion speed and retraction frequency), which are features missing in most cell spreading protocols19,24. While commonly used end-point cell spreading assays allow for the determination of cell spreading speed, these assays fail to resolve the temporal dynamics of cell edge movements. The lack of temporal information limits the ability to detect and quantify changes in lamellipodial protrusion-retraction cycles.
Our image processing and analysis software performs a streamlined analysis of the spreading cells, from cell segmentation to data quantification. Manual image analyses of cell spreading usually involve a biased selection of a threshold value or applying an automated segmentation algorithm, which is not suited for high-throughput experiments where many images need to be analyzed. Our software is, therefore, designed to detect and segment spreading cells in an automated fashion, in addition to quantifying protrusion dynamics and morphological descriptors. Together, these features make the described protocol amenable for large-throughput screenings of signaling pathways and molecular players that regulate lamellipodia.
To ensure that the analysis of spreading cells is robust and accurate, a few critical steps in the protocol must be performed with extra caution. The first step of the cell spreading assay involves plating fluorescently-labeled cells at a very low density on a fibronectin-coated coverslip the day before imaging (Figure 1A-D). These polarized cells enable precise focusing of the spreading cells' protruding edges during the image acquisition. If the density of polarized cells is too high, spreading cells have a high chance of landing on or overlapping with the polarized cells, which may lead to a cell segmentation failure. The recovery period post-trypsinization is another critical step in this protocol. Cells treated with the drug of interest, e.g., DMSO or CK-666, are detached from the cell culture dish by trypsin-EDTA (Step 2), followed by a recovery period of 45 mins (Figure 1E). This recovery step allows cells to recover from the proteolytic cleavage of cell surface proteins by trypsin19,24 and synchronizes the onset of cell spreading (Figure 2). If the recovery step is omitted, the cell-to-cell variability in cell spread area increases, reducing the consistency of the biological phenotype.
During the image acquisition, any sample drift inevitably decreases the quality and precision of cell spreading analysis, especially the kymograph analysis. To minimize sample drift, a few measures should be taken. First, the user should optimize the stage movement during the image acquisition. The optimization includes minimizing stage travel between fields of view and reducing the speed of stage movement. Second, it is essential to tightly secure the sample on the microscope stage. If these suggested measures do not eliminate sample drift, post-acquisition processing should be considered. Among many proprietary and open-source computational tools, we recommend using the "Descriptor-based registration" Fiji plugin to correct image shifts and align the movies of spreading cells (instructions can be found at: https://imagej.net/Descriptor-based_registration_(2d/3d)).
It should be noted that the quantitative analysis of cell areas and edge dynamics performed by the software depends heavily on the accuracy of cell segmentation. To ensure precise segmentation, we recommend visualizing cell spreading using a confocal imaging system, preferably a spinning disk confocal microscope which offers high resolution, low photobleaching/phototoxicity, and high signal-to-noise ratio. A confocal microscope efficiently removes out-of-focus fluorescence emitted by the spreading cells, which would otherwise decrease the image segmentation accuracy and cell boundary tracing. If a widefield microscope is used for image acquisition, additional post-acquisitional processing, e.g., image deconvolution, may be required to remove out-of-focus fluorescence and improve the accuracy of cell segmentation. Therefore, the choice of the imaging system should be considered.
Within the described software, two image segmentation algorithms were implemented and optimized to reliably detect cells labeled with dim to moderately bright fluorescent proteins, such as cytosolic green and red fluorescent proteins (GFP and RFP)26,27. However, these segmentation algorithms have a limited dynamic range and are not suitable for detecting cells labeled with extremely bright fluorescent proteins or dyes. In our hands, these algorithms tend to undersegment extremely bright cells due to the skewness of the image histogram towards high intensity pixels. For bright samples, the intensities of the images can be controlled by adjusting the exposure time or the output power of the excitation laser.
With these considerations in mind, this live-cell spreading protocol is a robust and powerful tool to study the dynamics of lamellipodia. The automated image analysis platform is suited to many biological investigations, e.g., high-content screening of molecular/signaling factors that regulate lamellipodial protrusions.
The authors have nothing to disclose.
This work was supported by the Connaught Fund New Investigator Award to S.P., Canada Foundation for Innovation, NSERC Discovery Grant Program (grants RGPIN-2015-05114 and RGPIN-2020-05881), University of Manchester and University of Toronto Joint Research Fund, and University of Toronto XSeed Program.
0.05% Trypsin (0.05%), 0.53 mM EDTA | Wisent Bioproducts | 325-042-CL | |
10.0 cm Petri Dish, Polystyrene, TC Treated, Vented | Starstedt | 83.3902 | |
15 mL High Clarity PP Centrifuge Tube, Conical Bottom, with Dome Seal Screw Cap, Sterile | Falcon | 352097 | |
1-Well Chamlide CMS for 22 mm x 22 mm Coverslip | Quorum Technologies | CM-S22-1 | |
35 mm TC-treated Easy-Grip Style Cell Culture Dish | Falcon | 353001 | |
50 mL Centrifuge Tube, Transparent, Plug Seal | Nest | 602002 | |
6.0 cm Cell Culture Dishes Treated for Increased Cell Attachment, Sterile | VWR | 10861-658 | |
Arp2/3 Complex Inhibitor I, CK-666 | Millipore Sigma | 182515 | |
Camera, Prime 95B-25MM | Photometrics | ||
Dimethyl Sulfoxide, Sterile | BioShop | DMS666 | |
DMEM, 1x, 4.5 g/L Glucose, with L-Glutamine, Sodium Pyruvate and Phenol Red | Wisent Bioproducts | 319-005 CL | |
DMEM/F-12, HEPES, No Phenol Red | Gibco | 11039021 | |
D-PBS, 1X | Wisent Bioproducts | 311-425 CL | |
Fetal Bovine Serum | Wisent Bioproducts | 080-110 | |
Fiji Software | ImageJ | ||
HEPES (1 M) | Gibco | 15630080 | |
Human Plasma Fibronectin Purified Protein 1 mg | Millipore Sigma | FC010 | |
Immersion Oil | Cargille | 16241 | |
L-Glutamine Solution (200 mM) | Wisent Bioproducts | 609-065-EL | |
MEM Non-Essential Amino Acids Solution (100X) | Gibco | 11140050 | |
Micro Cover Glasses, Square, No. 11/2 22 x 22 mm | VWR | CA48366-227-1 | |
Microscope Body, Eclipse Ti2-E | Nikon | ||
Objective, CFI Plan Apo Lambda 60X Oil | Nikon | MRD01605 | |
Penicillin-Streptomycin | Sigma | P4333 | |
Spinning Disk, Crest Light V2 | CrestOptics | ||
Spyder | Anaconda | ||
Stage top incubator | Tokai Hit | ||
Statistics Software, Prism | GraphPad | ||
Tweezers, Style 2 | Electron Microscopy Sciences | 78326-42 |