Cellular processes such as cell migration have traditionally been studied on two-dimensional, stiff plastic surfaces. This report describes a technique for directly visualizing protein localization and analyzing protein dynamics in cells migrating in a more physiologically relevant, three-dimensional matrix.
Traditionally, cell migration has been studied on two-dimensional, stiff plastic surfaces. However, during important biological processes such as wound healing, tissue regeneration, and cancer metastasis, cells must navigate through complex, three-dimensional extracellular tissue. To better understand the mechanisms behind these biological processes, it is important to examine the roles of the proteins responsible for driving cell migration. Here, we outline a protocol to study the mechanisms of cell migration using the epithelial cell line (MDCK), and a three-dimensional, fibrous, self-polymerizing matrix as a model system. This optically clear extracellular matrix is easily amenable to live-cell imaging studies and better mimics the physiological, soft tissue environment. This report demonstrates a technique for directly visualizing protein localization and dynamics, and deformation of the surrounding three-dimensional matrix.
Examination of protein localization and dynamics during cellular processes provides key insight into protein functions. Genetically encoded fluorescent tags provide a unique method for observing protein localization and dynamics. Using this technique, we can analyze the subcellular accumulation of key, force-generating cytoskeletal components in real-time as the cell maneuvers through the matrix. In addition, using multiple fluorescent tags with different wavelengths, we can examine the localization of multiple proteins simultaneously, thus allowing us to test, for example, whether different proteins have similar or divergent roles. Furthermore, the dynamics of fluorescently tagged proteins can be quantified using Fluorescent Recovery After Photobleaching (FRAP) analysis. This measurement assays the protein mobility and how stably bound the proteins are to the cytoskeletal network.
By combining live-cell imaging with the treatment of protein function inhibitors, we can examine in real-time the changes in the distribution of proteins and morphology of migrating cells. Furthermore, we also combine live-cell imaging with the use of fluorescent tracer particles embedded within the matrix to visualize the matrix deformation during cell migration. Thus, we can visualize how a migrating cell distributes force-generating proteins, and where the traction forces are exerted to the surrounding matrix. Through these techniques, we can gain valuable insight into the roles of specific proteins and their contributions to the mechanisms of cell migration.
1. Generation of stable cell line (e.g. MDCK cells)
2. Surface modification of glass bottom dish for optimal collagen binding (Optional)
3. Preparation of 3D collagen gel with tracer particles
4. Procedure for time-lapse image acquisition
5. FRAP procedure and analysis
6. Representative Results
An example of the live-cell imaging of healthy epithelial cells within the matrix is shown in Figure 3. Healthy cells exhibit a smooth, continuous membrane, and distinct nucleus, whereas unhealthy cells often have a disrupted membrane, and an excessive number of vacuoles. In a 3D matrix, single epithelial cells migrate1, and over the course of several days, epithelial cells form three-dimensional, spherical, multi-cellular cysts within the matrix2. The cells are also highly dynamic, and migrate within the cyst (Figure 3). The matrix deformation as a result of the traction forces exerted by migrating cells is analyzed through the displacement of tracer particles embedded in the surrounding matrix (Figure 4). The tracer particle movements are displayed as a maximum projection image of different time points, and each time point is pseudo-colored to indicate the time (Figure 4B). Alternatively, the images of an individual tracer particle are displayed as a montage to show the movement of the tracer particle over time (Figure 4C). All image analysis was done using ImageJ. These qualitative assessments of matrix deformation, and therefore the forces exerted by migrating cells, are useful first approximation of traction force distribution. The quantitative estimation of traction forces exerted by migrating cells in 3D is beyond the scope of this protocol and is described elsewhere3,4.
A typical fluorescence recovery after photobleaching experiment is shown in Figure 5. The region of interest must be photobleached using optimized laser settings so that the fluorescence intensity is visibly diminished compared to background levels (to maximize signal-to-noise ratio), while maintaining healthy and undamaged cells. The optimal setting must be determined empirically as each FRAP setup is different (e.g., FRAP using laser-scanning confocal system).
The exposure and time-interval of post-FRAP image acquisition must be carefully controlled to avoid background photobleaching. The use of lower laser power, short exposure time, and the use of sensitive camera and high efficient optics are essential for high quality FRAP imaging. If the fluorescence bleaching during the image acquisition is significant, this background fading must be quantified and normalized from the observed fluorescence intensity recovery. Due to 3D nature of environment, the z-component of fluorescence intensity recovery may become significant. Therefore, it is important to define a photobleached volume by taking pre and post 3D-stack images of photobleached region. Further analysis and modeling of fluorescence recovery is discussed elsewhere5,6.
The unique combination of live-cell imaging techniques: GFP to follow the dynamics of cytoskeletal proteins, red fluorescence tracer particles to monitor the matrix deformation, and the addition of inhibitors, can be used simultaneously for the analysis of protein dynamics, traction force and molecular pathways.
Figure 1. Preparation of collagen gel. A) Polymerized collagen matrix on a glass-bottom dish. The pink color of the gel is due to embedded fluorescent particles. B) Procedure to treat the glass bottom dishes to crosslink the collagen gel to the glass surface. First, the glass bottom dish is treated with 3-Aminopropyltrimethoxysilane solution, then glutaraldehyde solution that crosslinks the collagen matrix to the glass.
Figure 2. Schematics of confocal/FRAP microscope setup. The confocal microscope is based on a Zeiss AxioObserver with a CoolSnap HQ II CCD camera and completely automated by Slidebook software (Intelligent Imaging Innovations). The confocal unit is custom designed and based on Yokogawa spinning disk unit CSU10 and two solid-state lasers (488 nm with 50 mW and 561 nm with 40 mW) with Acousto-Optical Tunable Filter (AOTF) to allow milliseconds switching between two lasers. The emission filters are 525/50 nm and 620/60 nm (#118661 and #118085, Chroma Technology) and dichroic mirror is 488-568 BrightLine Dual band (Semrock). The objectives are a long working distance 40x C-Apochromat objective with NA 1.1 and working distance of 0.62 mm, and a 63x Plan-Apochromat objective with NA 1.4 and a working distance of 0.19 mm. The microscope also includes a FRAP photoablation system that consists of a fiber optically pumped dye laser, a computer controlled beam position and intensity, and a diffraction limited spot size. Furthermore, the microscope is equipped with an x-y motorized stage that includes 0.1 micron linear encoders on each axis. During the time-lapse imaging, the environmental temperature is maintained by a custom designed microscope chamber and a heater with a feedback temperature control. To isolate any noise and vibration, the entire microscope system is on a vibration-free table.
Figure 3. Live cell imaging of epithelial cells expressing GFP-actin. These cells formed a cyst after 4 days of culture in a 3D collagen matrix. Some cells move along the surface of cyst (yellow arrowhead), while others migrate within the interior of cyst (red arrowhead). Scale bar 10 μm, time in hours.
Figure 4. The effect of Rho-kinase inhibition on traction force. A) DIC image of a migrating MDCK cell expressing GFP tagged nuclear marker. The image was taken immediately before the treatment of Rho-kinase inhibitor Y-27632. Scale bar 10 μm. B) Particle displacement resulting from the addition of Y-27632. The particle positions at various time points (0 – 52 min) were pseudo-colored according to the intensity scale, then projected onto a single image. The white region is the GFP positive nucleus in the migrating cells. Scale bar 10 μm. Time in minutes. C) The particle movement at the trailing edge of the cell (see arrow in B). Asterisk denotes the last frame captured before Y-27632 addition. The tracer particle moved toward and away from the trailing edge of the migrating cell before and after the addition of Y-27632, respectively. Scale bar 1 μm.
Figure 5. FRAP analysis of GFP-actin expressing cell migrating in a three-dimensional matrix. A) Timelapse images of the GFP-actin expressing cell of before and after the photobleaching. A small region of the GFP-actin accumulated at the cell rear was photobleached (red arrowhead) at timepoint 0. Time in seconds, scale bar 5 μm. B) The average fluorescence intensity of the photobleached region (solid line), and the exponential fit of the fluorescence recovery (circles) plotted over time. The fluorescence intensity was normalized to the initial value before photobleaching.
Here we describe a method for using live-cell imaging to study the mechanisms of cell migration in a three-dimensional matrix. The success of this technique depends on obtaining “good” clones stably expressing GFP-tagged proteins. The low level of GFP proteins will require an excess excitation exposure that compromises cell health, while too high GFP level will have undesirable side effects on the cell. Thus, the vector choice to transfect genes into cells is important (e.g., the promoter, fluorescence tag, etc). There are many fluorescent tags other than GFP, including red fluorescent protein (RFP)7,8. These protein tags require specific excitation light wavelength, and therefore different sets of fluorescence filters. In addition, different antibiotic resistant genes (e.g., G418, hygromycin, puromycin etc) are also available and can be used to generate stable clones.
For live-cell imaging, transiently transfected cells can also be used. Although this minimizes the time and effort required to develop a stable cell line, transiently transfected cells are generally unhealthy, and therefore less reliable for use in experiments. In addition, the population of transiently transfected cells is a heterogeneous mixture of expressing and non-expressing cells, and the expressing cells will lose fluorescent protein expression over time. In comparison, stably expressing cells are typically cloned from a single colony of cells and protein expression should be stable and homogeneous, although the expression level will be lower than in transiently expressing cells. Since the colony of stable cells is likely to have originated from a single cell, it is important to pick colonies that have morphology and behavior similar to normal, untransfected cells, to avoid phenotypes that are specific only to that clone (unless the transfected plasmid is designed to perturb cell morphology). Multiple colonies should be used for experiments to avoid phenotypes due to clonal variation. Alternatively, viral transfection or electroporation may be used to introduce genes of interest. This viral transfection is particularly useful for difficult to transfect cells or primary cells (see the detailed protocols of other transfection techniques described elsewhere in this journal).
The trickiest part of this procedure is subcloning. Stable clones can be picked using the cotton swab method described here, or alternatively, using a cloning ring. Another method for selecting GFP-positive cells is using FACS. FACS selects cells based solely on the magnitude of fluorescence intensity. In comparison, using the cloning method described here, the user can visually identify not only the level of GFP expression and but also its cellular localization using a microscope. Also, cloning using FACS typically results in a more heterogeneous population of cells. To overcome this issue, FACS-sorted cells can be plated directly into 96 well dishes, so that each well contains only one cell. Colonies originating from single cells should be a homogeneous population.
The method used to attach the collagen matrix to the glass bottom dish (Figure 1) is critical for live-cell imaging and culturing cells in a gel for an extended period of time. The initial step for attaching the matrix to the glass is the formation of a silane monolayer. Failure of silane to deposit into a monolayer will often result in collagen matrix detachment from the glass. This is because the cells embedded within the matrix exert contractile traction forces, and cause the matrix to shrink, thus increasing the likelihood of matrix detachment from the dish. Therefore, silane/glutaraldehyde treatment of the dish is critical for maintaining gel attachment to the dish. Note that collagen assembly conditions (concentration, time, and pH) may need to be optimized for different collagen type and batch. In addition, the protocol can be used with different types of collagen or other 3D gels such as Matrigel or fibroblast derived matrix9.
Although many cells do not show adverse effects in response to HEPES buffer, some cells are more finicky. To maintain cell health during time-lapse acquisition, a CO2 incubator can be added to control the pH within the microscope enclosure. In addition, laser exposure to the cells should be minimized. If the fluorescence emitted by the cells is too dim, the fluorescence can be collected more effectively using a more sensitive camera, or a better objective, or a more efficient filter set, etc. Cell morphology within the microscope enclosure should be compared with its morphology in the CO2 incubator to make sure that cells are healthy and behaving normally. Since this method describes single cell based assays, some cell-to-cell variation is expected. Therefore, repeating experiments and data quantification is essential.
The authors have nothing to disclose.
We thank Dr. Grant Sumida for critical reading of the manuscript. This work was supported by a Beckman Young Investigator Award (SY), a Hellman Family New Faculty Award (SY), a NIH EUREKA, the University of California Cancer Research Coordinating Committee.
Name of the reagent | Company | Catalogue number | Comments |
Collagen, bovine, Type I | BD Biosciences | 354231 | Stock is about 3 mg/ml |
3-aminopropyltrimethoxysilane | Sigma Aldrich | 281778 | Dilute in water |
glutaraldehyde | Sigma Aldrich | 340855 | Dilute in PBS |
1M Hepes | Invitrogen | 15630-080 | |
Fluospheres polystyrene microspheres 1 µm, red fluorescence (580/605) | Invitrogen | F13083 | |
Geneticin (G418) | Invitrogen | 11811-031 | |
Culture media components: | |||
DMEM | Invitrogen | 31600-034 | |
Fetal Bovine Serum | Atlanta Biologicals | S115500 | |
Penicillin/Streptomycin | Invitrogen | 15140-122 | |
Kanamycin | Invitrogen | 15160-054 |