. 2014 Oct 31;8(5):440–451. doi:10.4161/cam.36224

An introduction to the wound healing assay using live-cellmicroscopy

James E N Jonkman¹,Judith A Cathcart¹,Feng Xu¹,Miria E Bartolini¹,Jennifer E Amon²,Katarzyna M Stevens²,Pina Colarusso^2,^3,^*

¹Advanced Optical Microscopy Facility;University Health Network; Toronto, ON Canada

²Live Cell Imaging Facility; Snyder Institutefor Chronic Diseases; University of Calgary; Calgary, ABCanada

³Department of Physiology and Pharmacology;University of Calgary; Calgary, AB Canada

Correspondence to: PinaColarusso; Email:gcolarus@ucalgary.ca

Received 2014 Apr 17; Revised 2014 Aug 17; Accepted 2014 Aug 25; Collection date 2014 Sep-Oct.

PMC Copyright notice

PMCID: PMC5154238 PMID:25482647

Abstract

The wound healing assay is used in a range of disciplines to study the coordinatedmovement of a cell population. In this technical review, we describe the workflow of thewound healing assay as monitored by optical microscopy. Although the assay isstraightforward, a lack of standardization in its application makes it difficult tocompare results and reproduce experiments among researchers. We recommend generalguidelines for consistency, including: (1) sample preparation including the creation ofthe gap, (2) microscope equipment requirements, (3) image acquisition, and (4) the use ofimage analysis to measure the gap size and its rate of closure over time. We also describeparameters that are specific to the particular research question, such as seeding densityand matrix coatings. All of these parameters must be carefully controlled within a givenset of experiments in order to achieve accurate and reproducible results.

Keywords: wound healing assay, scratch assay, physical exclusion assay, collective cell migration, sheet migration, live-cell microscopy

Abbreviations

DIC: differential interference contrast
ECM: extracellular matrix
HUVEC: human umbilical vein endothelial cells
NA: numerical aperture

I. Introduction

The wound healing assay is a standardin vitro technique for probingcollective cell migration^1-6 in twodimensions. In this assay, a cell-free area is created in a confluent monolayer by physicalexclusion or by removing the cells from the area through mechanical, thermal or chemicaldamage. The exposure to the cell-free area induces the cells to migrate into the gap. Asequence of representative images from a wound healing assay carried out on a confluentendothelial monolayer is shown inFigure 1. In this example, the monolayer was scratched with apipette tip and the migration into the gap was imaged over several hours using atransmitted-light microscope equipped for live-cell imaging. Note that the cells remain incontact during their directed and coordinated movement into the gap.

Figure 1. — Images from a scratch assay experiment at different time points. Human umbilical veinendothelial cells (HUVEC) were plated on gelatin-coated plastic dishes, wounded with ap20 pipette tip, and then imaged overnight using a microscope equipped with pointvisiting and live-cell apparatus. Scale bar = 120 μm.

The type of collective cell migration probed by the wound healing assay is known as sheetmigration. This migration is exhibited by epithelial and endothelial monolayers that move intwo dimensions while maintaining their intercellular junctions.^1-5 Sheet migration occursin diverse processes such as cancer metastasis,⁷^,⁸^,^9,10 embryonicmorphogenesis,^11,12 andtissue injury.^12,13 Sheetmigration involves a complex interplay among mechanical forces, molecular interactions andbiochemical cascades that are triggered by the exposure of the cellular monolayer to freespace, as when cells are exposed to the gap in the wound healing assay.^14-21 Although a strict definition of sheet migration requires cells tomaintain intercellular junctions, there is some evidence that cells lacking intercellularjunctions such as fibroblasts exhibit some characteristics of sheet migration.²²

The most common information derived from the wound healing assay is the rate of gapclosure, which is a measure of the speed of the collective motion of the cells. In a typicalexperiment, the gap closure rate is measured under different conditions, such as by treatingthe cells with RNA interference (RNAi),^15,16 modulating the extracellular matrix composition,^23-25 or by changing otherenvironmental variables such as substrate stiffness.²⁶ In this way, the underlying mechanisms governing sheet migrationare revealed. The wound healing assay is also easily adapted for medium to high throughputapplications such as small molecule screening²⁷ and drug discovery.²⁸

A researcher new to the wound healing assay is presented with a plethora of choices atevery step of the technique. This technical review provides researchers with an introductionto this technique while emphasizing guidelines for quantitative and reproducible results. Wecover the steps in setting up the wound healing assay, from sample preparation to imageanalysis. Although the wound healing assay is widely applied, it lacks standardization inexperimental approach. In practice, a wide variety of methods are used to create the woundor gap, and then to monitor and quantify the dynamics of the migration into the cell-freearea. Without standardization of experimental variables, it is difficult to make meaningfulcomparisons among different reports in the literature. Furthermore, to achieve a high levelof accuracy and reproducibility, a well-defined procedure for the wound healing assay iscrucial. For example, a gap of consistent size facilitates quantification within a series ofexperiments. To streamline the analysis, the gap should have relatively smooth edges andlittle cellular debris within the gap. Appropriate controls and experimental conditionsshould be recorded in parallel to improve data quality and the assay should have anobjective endpoint. It is also important to consider the influence of the extracellularmatrix and the substrate on the cellular migration and interactions after creating the gapin the monolayer. Developing a standard approach for the wound healing assay promotescomparability between studies and could lead to a useful database containing collectivemigration data for a variety of cell lines.

II. Sample Preparation: Cell Culture

Wound healing assays use cells derived from either cell lines or primary isolations fromblood or tissue. Reproducible cell culture conditions²⁹ are required for stable phenotypes that are the foundation ofsuccessful assays. For most cell lines, culturing protocols can be obtained from centralresource collections such as the American Tissue Type Collection (ATCC;http://www.atcc.org),which include information such as the recommended growth medium, sub-culturing and expecteddoubling time. For cells derived from primary isolations, however, the protocols are oftendeveloped in individual laboratories and are more difficult to standardize. In practicalterms, it is important to keep the conditions as consistent as possible.

Typically, wound healing assays are carried out with a thin layer of cells grown on aplastic or glass substrate. For many epithelial and endothelial cells, thin layers of cellsare easily obtained because the cells form confluent monolayers. A good starting point is towork out the seeding density and incubation time required to produce a confluent monolayerof cells. Other important factors include the frequency and volume of media changes as wellas the passage number (if applicable). The wound healing assay should be initiated at thesame time point after the cells become confluent because the results obtained may vary asthe monolayer matures.

The underlying substrate and the associated extracellular matrix (ECM) are also importantconsiderations in the wound healing assay. Some cell types can grow directly on plastic orglass substrates, while others require a coating of extracellular matrix components such asgelatin, collagen, or fibronectin in order to adhere. It is important to recognize that theECM does more than provide an adhesive surface for the cells and can also participate insignalling involved in cell migration.³⁰ In addition, the stiffness of the underlying substrate can alsomodulate the dynamics of the migration into the gap.³¹

The dish or chamber used to culture the cells should be carefully chosen to match theexperimental approach. Multi-well cell culture dishes (6, 12, or 24 wells) are a popularchoice because they are inexpensive and fabricated from plastic treated to encourage celladhesion. The wells in these dishes are also large enough to provide clearance formanipulating the monolayer as when creating a scratch wound or working with an insert. Forhigh-resolution imaging, as discussed below, it is necessary to use dishes with thin plasticor glass bottoms on the order of 160-190 μm (Number 1.5 coverslips). Although almostall cells will grow well on plastic, some cell types will not grow on glass unless it isspecially treated and coated with serum, gelatin or other extracellular components.

During the course of the wound healing assay, the cells will migrate into the gap, and theywill also proliferate. In most wound healing studies, the desired result is to suppressproliferation so that it does not interfere with the measurement of migration. Drugs such asactinomycin C¹³ can be used to arrestmitosis at various stages, but the dosage must be carefully controlled to avoid apoptosis orother toxic effects which may affect cell migration. Serum starving³² is the most common non-pharmaceutical method forminimizing proliferation in wound healing assays, but the degree of serum starving has to beworked out for each cell type under investigation. Primary cells do not tolerate serumstarving as well as established cell lines, and often require a reduced concentration ofserum rather than its complete absence within the medium. A note of caution when using serumstarving: a recent report suggests that serum starving induces many complex andunpredictable responses in different cell lines.³³

III. Sample Preparation: Creating the Gap

A cell-free gap can be created in a cell monolayer either by direct manipulation or byphysical exclusion. These approaches have been characterized as “celldepletion” or “cell exclusion” collective cell migration assays.⁴ Direct manipulation destroys specificregions within the monolayer through mechanical, electrical, chemical or thermal means.Physical exclusion creates cell-free areas by placing barriers within the cell culture platesuch as plastic inserts, liquids or gels. The physical exclusion methods generally causeminimal damage to the remaining cells compared to the direct damage of the monolayer. Herewe limit our discussion to two techniques that rely on equipment that is readily availablein laboratories equipped for cell culture: the scratch (direct manipulation) and insert(physical exclusion) approaches. Interested readers are directed towards a detailed reviewthat discusses their relative merits and applications of the more novel gap-creationmethods.³⁴ Whatever the method used,it is important to note that recent research suggests that gap geometry affects the closurerate regardless of gap surface area uniformity.^35,36

To simulate wounding, the most common approach is to create a gap by scratching a confluentmonolayer with a pipette tip, needle, or other sharp tool. This is known as the scratchassay and is a good choice because it is inexpensive and easy to implement. Usually, apipette tip is used to create a scratch one well at a time. As the scratch is createdmanually, it can be difficult to generate reproducible wounds. It is important to angle thepipette correctly as well as to apply consistent pressure to create a consistent gapwidth.³⁷ In addition, applying toomuch pressure may damage the extracellular matrix, which can affect migration rates.²⁸ While the scratch method is usuallyperformed one well at a time, some groups have scaled up the technique to multi-well platesusing devices with multiple pins.²⁷

Alternatively, the gap can be created by physically excluding the cells using inserts. Suchinserts create a linear or circular gap by adhering to the treated dish bottom andpreventing cell growth in a predefined region. A linear exclusion commercial insert(www.ibidi.com)is shown inFigure 2 (examples ofother insert suppliers include Cell Biolabs, and Platypus Technologies). The wound-healingassay is initiated by removing the insert. The advantages of using inserts include requiringfewer cells when seeding as well as more reproducible gap sizes compared to the scratchmethod. While the gap size created by the insert can be more reproducible, cells clinging tothe insert can be torn out of the monolayer leaving jagged edges. This means that the gapedges are not necessarily better defined compared to a scratch assay. In addition,commercial inserts are more costly than using the scratch method. Inserts can be re-used tomitigate the cost, but over time their adhesion to the bottom of the dish may becompromised.

Figure 2. — Silicone insert for gap creation via cell exclusion in a standard 24-well plate(A). Note the cells are seeded on both sides of the insert as shown in(B). Before the start of the experiment, the insert is removed and thecells are monitored as they move into the gap associated with the thin silicone stripseparating the two wells.

It is a good idea to rinse twice with phosphate buffered saline or a suitable buffer beforethe cell culture medium is replaced and image acquisition is started. Rinsing will removedebris from damaged or dead cells, particularly after mechanical scratching. Mechanicalscratching may also release more growth factors from the damaged cells compared to physicalexclusion methods; replacing the growth medium with fresh medium (with or without particulargrowth factors as needed) after scratching helps to control the factors available to cellsfor migration.

IV. Microscope Configuration and Image Acquisition for the Wound Healing Assay

Fluorescence has become the most important contrast technique for optical microscopy ofbiological specimens. Entire cells, specific organelles within cells, or molecular speciesexpressed in cells can be labeled with fluorescent contrast agents. These labels emit avariety of colours when they are excited with light of an appropriate wavelength, whileunlabeled entities remain dark. By carefully choosing fluorophores, there is littleinterference between fluorophores of different colours and one can see the spatialrelationship between various objects. With all of the advantages of using fluorescence forcellular imaging, it may come as a surprise that fluorescence isnot alwaysthe best contrast technique for wound-healing assays. For wound healing experiments,transmitted-light techniques are usually sufficient to track the area of the gap overtime.²³ Long-term imaging withfluorescence leads to both photobleaching (fading of the fluorescence over time) andphototoxicity (light-induced cell death). Fluorescence labeling itself for living cellsintroduces potentially migration-affecting or even toxic molecules into the cells. For thesereasons, transmitted-light techniques are preferred for monitoring collective cell migrationunless fluorescence imaging is required for visualizing cellular processes over and beyondmigration into the gap.

Most cellular monolayers are thin and almost invisible under transmitted-light (ie:brightfield) illumination; however, reasonable contrast can be achieved by adding a fewspecial components to the light path.^38,39 For example, phase contrast is available on almost all standardcell culture microscopes and often is available on more advanced imaging systems because itis straightforward to implement and compatible with plastic dishes. Differentialinterference contrast (DIC) is another transmitted-light technique that can provide contrastfor unlabeled cells. DIC, however, requires more sophisticated optical components (prismsand polarizers) and is more expensive and more difficult to set up than phase contrast. Inaddition, DIC typically requires samples to be grown on glass rather than standard plasticdishes. As most wound healing assays monitor the overall gap rather than individual cellularstructures, the cheaper and simpler phase contrast technique is more commonly used.Representative phase contrast images from a wound healing assay, carried out on endothelialcells, are shown inFigure 1.

Transmitted-light microscopes range in complexity from a basic cell-culture microscopeshown inFigure 3A to a motorized,environmentally-controlled, automated microscope with camera as shown inFigure 3B. In the simplest approach,the researcher can score the size of the gaps by inspecting the dishes at regular timepoints (for example, every 2 hours) on a cell culture microscope. If the microscope isequipped with an eyepiece reticule, the researcher can go one step further and attempt tomeasure the gap widths directly. Visual scoring and manual measurements, while adequate forobtaining preliminary data, are tedious, imprecise and prone to user bias. For more accuratemeasurements, one can add a digital camera to the manual microscope and record images of thewound healing over time for subsequent analysis. However, capturing images manually quicklybecomes onerous as wound healing assays can take many hours to complete, and it is difficultto keep track of the same field-of-view along each wound. For these reasons, some degree ofautomation is desirable for monitoring the wound healing assay.

Ideally, the microscope should be equipped with image acquisition software and ascientific-grade digital camera, environmental chamber, motorized stage for multi-positionacquisition, and motorized focus with autofocus capability to minimize focal drift overtime. An automated system relies on software for triggering image acquisition using adigital camera at regular time intervals as the wound recovers, for example every 15 to30 min over a 4- to 24-hour period. The environmental chamber replicates an incubatoron the microscope by controlling temperature, pH and humidity⁴⁰; it allows samples to remain on the microscope ratherthan being repeatedly taken in and out of an incubator, thus avoiding any compromises incell physiology arising from fluctuations in these parameters. Live-cell chambers range fromintegrated commercial systems, costing thousands of dollars, to inexpensive homebuiltdevices that can be as effective as the commercial systems.⁴¹ Temperature control can be achieved by using anenclosure that wraps around the entire microscope, a more compact stage-top incubator, or acombination of both. Furthermore, although most cell types can be maintained in aspecialized atmospheric buffer for short periods, it is advisable to carry out the woundhealing in the cell culture medium. This usually requires the use of CO₂ at5% or 10%, but of course the exact concentration depends on cell type.CO₂ can be delivered to the cells at the desired concentration by using a gasmixer with a 100%-CO₂ cylinder, or directly from a specialty cylinderprepared with a custom gas mixture. It is important to humidify the gas before it enters theincubator to avoid evaporation. With evaporation, cell viability would be compromised byexposing the preparation to osmotic stress. A software package combined with a motorizedstage and focus allows for recording images at pre-specified x, y and z stage coordinatesfor multiple points along each wound. Images of different conditions set up in a multi-welldish can then be acquired during the same experiment. This saves time and ensures that thevarious experiments are carried out under the same conditions.

Tips for optimal imaging

Here we will briefly review several parameters that are important for optimal imaging ofthe wound healing assay. For more detailed descriptions, the reader is directed to severalexcellent references that describe the elements of light microscopy.^38,39

Koehler illumination

For optimal transmitted-light imaging, the microscope should be set up for Koehlerillumination. This frequently-overlooked procedure is necessary to ensure evenillumination across the sample and optimal contrast (note: some cell culture microscopesare preset and do not allow for these adjustments). A full Koehler alignment is describedin the references.^38,⁴² A quick guide to setting upKoehler illumination, which assumes proper bulb alignment, is as follows: (1) Ensure thefield diaphragm and condenser diaphragm (also known as irises or apertures) are fullyopened. (2) Bring the sample into focus and do not touch the focus knobs again until steps3-5 are complete. (3) Close the field diaphragm to its smallest diameter (it will notclose completely). (4) Next adjust the condenser until the edges of the field diaphragmare in focus. The diaphragm image will be overlaid on the focused sample image. (5) Centrethe field diaphragm using the adjustment knobs on the condenser. (6) Open the fielddiaphragm just to the edges of the desired field-of-view to minimize scatter and glare. Inaddition, the optical path should be free of dust, debris or other image clutter that maydegrade the optical performance and complicate subsequent analysis.

Phase contrast

Phase contrast objectives contain a ring for creating contrast based on differences inoptical path length through the sample. Phase objectives are usually denoted as PH1, PH2,etc., with the numbers denoting the size of the ring. To achieve the phase effect, choosea corresponding phase ring in the microscope's condenser, which illuminates thesample with a bright ring of light. Ensure that the bright ring from the condenseroverlays the ring of the objective by using a Bertrand lens or by removing one of theeyepieces and peering into the empty ocular position – most microscopes haveadjustment screws to fine-tune the alignment if necessary. The side of the well containingthe cells can shift the rings so try to avoid creating wounds near the edges of thewell.

Numerical aperture, contrast and resolution

The numerical aperture (NA) of the objective and condenser determines the spatialresolution when using transmitted-light techniques.^38,39 The lateral spatial resolution, r, isgiven byr = 1.22 λ/(NA_objective +NA_condenser), where λ is the centre wavelength of the incidentlight.³⁸ Objectives typically areinscribed with the NA after the magnification, such as 20x/0.75 NA. Often a range ofnumerical apertures is inscribed directly on the condenser as well. For maximum spatialresolution, the condenser diaphragm should be set to match the numerical aperture of theobjective once the microscope has been aligned for Koehler illumination. If the condenseris not labeled with the NA, the reader is referred to detailed protocols in thereferences.^38,42 Forincreased contrast, the condenser diaphragm can be closed, but this reduces the numericalaperture of the condenser, leading to a loss of spatial resolution in the image. Finally,the use of a bandpass filter (usually blue or green) can improve the contrast in bothphase and DIC techniques.⁴² For a10x/0.3 NA objective and a condenser with 0.3 NA, common settings used to image woundhealing, the lateral spatial resolution of the microscope is about 1.1 μm if thewavelength λ illuminating the sample is about 550 nm.

Choice of objective

Choose the highest magnification objective that allows you to visualize both sides of thewound, or preferably create a wound that fits your best objective. For example, a gap of500 μm fits in the field-of-view of a 10x objective using a standard CCD cameraassuming there is no additional magnification after the objective. One can generally get ahigher-quality phase-contrast image from a higher power objective (10x, as opposed to 5xor lower) while maintaining a reasonable field-of-view.

Note that it is important to choose a sample chamber that matches the working distance ofthe objective. The working distance is the distance from the front surface of theobjective lens to where the lens focuses. In general, low NA objectives have longerworking distances than high NA objectives. In practical terms, low NA (<0.5 NA)objectives must be used to image cells grown in standard multi-well dishes (6, 12, 24wells) because these objectives can focus through the thick plastic of the wells. High NAobjectives, in contrast, have shorter working distances and require sample chambers withthin plastic or glass bottoms, usually on the order of 170 μm.

Camera settings

When digital images are acquired, the camera pixels should be calibrated by recording areference image of a stage micrometer.⁴² In this way, the image dimensions can be calibrated to match theactual physical dimensions of the sample. For cameras with rectangular chips, considerrotating the camera to maximize the amount of wound gap that you cover while still seeingboth sides of the gap. Finally, before beginning your experiment, place a blank slide onthe microscope, defocus at least a millimeter, and snap a background image for use inflat-fielding, a procedure used to correct uneven illumination (more details on flat-fieldcorrection in the next section).

V. Image Processing and Analysis of the Wound Healing Assay

1. Analysis approach and experimental endpoint

Once the digital images are recorded, the gap size can be measured as a function of timeusing nearly any image analysis software package (such as ImagePro Premier, Metamorph, orthe open-source Fiji/ImageJ). One might consider tracking the gap width by drawing linesalong the leading edges of each cell front, and then measuring the decrease in the averagedistances of the lines as the wound closes. However, the cells at the scratch edges oftengrow into the gap at different rates, leading to an ill-defined cell front as theexperiment progresses. Manually tracing the leading edges is also time-consuming if asequence of images is acquired over time. Some groups try to measure the exact time of thewound closure by direct observation. Although this method is simple, it can be difficultto pinpoint the wound-closure event because the cells often do not form a perfectmonolayer in an orderly fashion; moreover, this method takes the maximum amount ofobservation time.

A better approach is to apply image analysis to computationally measure and plot the gaparea as a function of time. From the slope of this plot a simple calculation yields thecell migration rate v_migration in μm/hr, or alternatively thetime it takes for the gap to close to 50% of its original area, which we denotet_1/2gap. These two values are readily measured usingquantitative analysis via the simple steps outlined below, removing subjectivity from thegap closure measurement. The method is robust: even if some groups of cells along the gapclose in faster than others, one can still unambiguously determine the gap areaparticularly during this first half of the wound healing. In addition, by plotting the gaparea versus time, the cell migration rate can be readily extrapolated from the data wellbefore the gap has fully closed in, thereby saving costly imaging time. In fact, it isgenerally not necessary to continue imaging beyond the point where the gap isapproximately half closed. While a consistent gap size is required to compare thet_1/2gap measurement, the cell migration ratev_migration is independent of the initial gap size. However, asthe biological stimuli directing collective cell migration are affected by many factorsincluding gap geometry and size, a relatively consistent gap size is anywaysdesirable.

Here we will describe a standard approach for analyzing the wound healing assay as imagedby transmitted-light techniques. We will illustrate the steps by using a sample data setacquired using phase contrast. The general workflow is a starting point that can beadapted and modified for use in a number of different image analysis packages.Alternatively, several analysis programs (Image-Pro Premier by Media Cybernetics,⁴³ and TScratch⁴⁴) include customized wound-healing applications toautomate the rate measurements, using similar steps to the ones we describe. Furthermore,one can out-source the analysis process entirely by uploading your wound-healing datasetsfor processing by commercial vendors such as ibidi.⁴⁵ With the automated analysis programs, the description in Step 5below is useful for calculating the actual cell migration rate in real-world units ofμm/hr.

2. Data handling: image formats and creating a time-lapse data set

Before beginning the analysis, check that the data files are in a format compatible withyour image analysis program. If the image acquisition system is integrated with the imageanalysis software, data formats should be compatible and this step will not be necessary.If the analysis package cannot open the data sets acquired on the microscope, the datashould be exported from the capture software into a standard format such as TIFF. It isimportant to ensure that the original data sets are saved in the custom and/or proprietaryformat as well, or key instrument and acquisition settings may be lost. Avoid exportingthe data using image formats that degrade the image quality through compression (such asmost JPEG formats). Often the image acquisition software will give the option of savingtime series data within a single file, known as a multi-TIFF. Such data series areconvenient for organizing data sets compared to handling the individual images for eachexperiment. If this option is not available, the TIFF images are typically numbered andthe sequence frames can be imported into a series using the image analysis software.

3. Preprocessing: flatfield correction and edge detection

The first step in image analysis is to separate the objects of interest from thebackground; in a wound healing assay, the object is the gap area while the background isthe area containing cells. Often, this is achieved by manipulating the image histogram,which plots the number of pixels at given intensities within one image. By choosing anintensity cut-off, the pixels can be divided into object and background pixels based onwhether their intensities are higher or lower than the cut-off value, respectively. Thistechnique is known as thresholding while the separation of the image into differentpopulations is known as image segmentation.

Two pre-processing steps can make the image segmentation much more successful. First, itis helpful to correct the background in the image. Even if the microscope has beenproperly adjusted for Koehler illumination, it is common to have intensity variations of10% or more across the camera's field-of-view.Figure 4 shows images from a typical researchmicroscope before (A) and after correction (B). The background intensity variation issubtle in the uncorrected image, but is evident when a line profile drawn across the imageis examined inFigure 4C (black,upper line). Such background intensity variations can make image segmentation difficulteven after edge detection and smoothing. Fortunately, the intensity variations can bereadily removed by a flat-field correction. This correction involves dividing the originalimage by a background image, and then scaling the resulting image back to a reasonableintensity. The most rigorous approach is to acquire a background image of a blank chamberacquired during the wound healing experiment as mentioned earlier in the tips for optimalimaging. Alternatively, software routines such as a “rolling ball” or a“flat-field” filter can be applied to the image to calculate an estimate ofthe background. Both methods are usually implemented simply as a “flat-fieldcorrection” menu item in 2D image analysis software.Figure 4B shows the results of flat-fieldcorrection of the image shown in (A) using a rolling ball filter. The corrected lineprofile shown inFigure 4C (red,lower) shows little variation in intensity across the field-of-view compared to theuncorrected line profile (black, upper).

Figure 4. — The illumination across the field-of-view is corrected by flat-fielding. Linesalong the same image coordinates are drawn on original image (A) andthe corrected image (B). In (C), line profiles before(black) and after (red) flat-fielding show pixel intensities along the correspondinglines shown in (A) and (B). The black line profile drawnon the original image (A) shows uneven intensities when the pixelintensities across the line are plotted inC (also shown in black).Note that the slope of the red line from the corrected image is near 0 once flatfielding has been applied, indicating a stable average intensity of the pixelintensities along this line. Cells were derived from a human bronchial epithelialcell line (BEAS-2B cells).

Second, with transmitted-light (including phase contrast) images, it is helpful toincrease the contrast between the gap and the area that contains cells. Inspection of thewound healing image inFigure 5Areveals that, although the gap and cells have similar average intensity values intransmitted-light images, they differ in the local variation in pixel intensities. Thatis, the gap area is smooth and shows little variation in intensity, while abrupt changesin pixel intensities can be seen within the cell monolayer and near the gap-cellinterfaces. Processing routines, such as edge detection or variance filters, can be usedto create images that highlight the magnitude of changes in the pixel intensities amongneighbouring pixels. Where there are small changes in intensity among neighbouring pixels,the pixels are assigned low values (closer to black) and where there are largerfluctuations the pixels are assigned high values (closer to white). An example of an imageprocessed by a Sobel filter, a type of edge filter, is shown inFigure 5B. The edge detection also accentuatesnoise, but a Gaussian smoothing filter can help to reduce the noise (Figure 5C).

Figure 5. — The images show the workflow involved in a standard wound healing analysis protocolafter the flat-fielding step shown in Figure 4. Data were processed using Image-ProPremier (Rockville, MD). The images are derived from a time series of a sheetmigration assay carried out with BEAS cells. The first image in a time series isshown (A) The original greyscale image (A) ispre-processed to make the pixel separation into the gap or cells easier. First, astandard edge detection step (Sobel filter) is applied to (B), followedby smoothing (C). (D) is the histogram of image(C). Note that a single intensity value can separate the gap from thecells (solid line showing the threshold value is shown in histogram).(E) shows the image that is created by using a single thresholdintensity value to segment the image into two populations: the gap versus cells. Inthis way, the number of pixels in each population can be calculated and give therelative areas covered by each population. The small islands of red pixels betweenthe cells can be excluded from the analysis based on size as shown in(F). Here the detected gap pixels are shown in blue compared to theexcluded pixels shown in red.

4. Image segmentation

Having prepared and pre-processed the time-lapse image sets, we are now ready to perform“image segmentation,” which is carried out by selecting the gap area using anintensity threshold. The threshold is an intensity cut-off that separates the features ofinterest (gap) from the background (cells). Most analysis software packages provide a toolfor interactively choosing the threshold, allowing one to subjectively slide the thresholdvalue higher and lower until the desired features are selected. A histogram of theintensity values in the image can help to guide the intensity threshold choice.Figure 5D shows the histogram of theresulting image inFigure 5Cafter edge detection and smoothing. As indicated on the histogram, a single intensityvalue is chosen to threshold the image into two populations of pixels, corresponding tothe gap and the cells, respectively. The detected gap pixels are shown overlaid on theimage inFigure 5E. There arealso a number of automatic thresholding routines such as Otsu or maximum entropy that maywork, thereby providing an objective way of setting the threshold. More sophisticatedsegmentation routines, such as Image Pro's Smart Segmentation algorithm,⁴³ use texture analysis or patternrecognition to define the features of interest more precisely, which can help to pick outthe gap area from the cells more accurately.

After selecting the optimal threshold, the software can split the image up into objects,which are groups of contiguous pixels that all pass the threshold criterion, in this case,a given pixel intensity. In ImageJ, this is accomplished using the “AnalyzeParticles” feature, whereas Image Pro Premier refers to this as “CountObjects.” The largest object is generally the gap area, but there may still be somesmaller regions in the cell area that have the same intensity as the gap area, and aretherefore selected along with the gap. In most analysis software there is a method forexcluding these unwanted regions by setting a minimum size limit for the objects, theresult of which is shown inFigure 5F. As the gap closes, a simple size criterion may nolonger be sufficient for unambiguously selecting the gap area: stopping the analysis whenthe gap reaches 50% closure avoids this problem as discussed in the nextsection.

Important note about image acquisition parameters and segmentation

It is advisable to carry out segmentation on a few sample data sets run on differentdays before launching into a full assay. In this way, experimental parameters can beoptimized for robust image segmentation. Sometimes the image acquisition settings orother experimental factors can impede the segmentation and it is important to recognizeand minimize these factors early on. For example, debris can make it difficult tosegment the image into the cells and the gap. When there is debris within the gap, theimage segmentation sometimes will detect a gap area containing multiple holes ratherthan one contiguous section. If left uncorrected, the pixels in the binary imagecorresponding to debris may be excluded from the gap area pixel determination, leadingto a systematic underestimation of the gap area. Of course, it is better to avoid areaswith debris during the acquisition but if debris within the area is unavoidable, thesettings in the image analysis program can often be set to ignore holes within acontiguous area such as the gap (as for example choosing “include holes” inthe Analyze>>Particles command in Fiji/ImageJ).

5. Calculating the wound-healing rate

Having measured the gap area for each frame in the wound healing experiment, we can nowplot gap area as a function of time as shown to derive the cell migration ratev_migration and also the t_1/2gap value. A straight-line fit is areasonable approximation and can be readily accomplished using, for example, Excel'sLinear Trendline feature. The general equation for a line is given by y = mx + b,where m is the slope of the line and b is the y-intercept, which in our case is the gaparea at the start of the experiment. Setting y = b/2 (ie: the point at which the gap ishalf the original area) and solving for x, we find that the t_1/2gap can beexpressed as:

t_{1 / 2 g a p} = \frac{I n i t i a l G a p A r e a}{2 \times | s l o p e |}

(1)

The cell sheet migration rate, v_migration, isthe average velocity at which the cells collectively move into the gap. The slope is equalto dA/dt, where the area A is the width of the gap (w) times the lengthof the gap (l). Assuming that the gap is much longer than thefield-of-view so that cells do not migrate in from the edges, then the length is constant,sodA/dt =l × dw/dt. Also, the width closes in attwice the rate of the cell migration, sodw/dt = 2 ×v_migration. This gives the cell migration rateas:

v_{m i g r a t i o n} = \frac{| s l o p e |}{2 \times l}

(2)

If the graph is plotted with area in μm² and time in hours, thenv_migration conveniently can be expressed in units of μm/hour.Figure 6 illustrates four timepoints (A-D) from a scratch wound healing assay carried out using HUVEC cells with thearea of the gap plotted over time inFigure 6E. Note that the graph of area over time is linearfor the early time points; however, at later time points where the gap is nearly closed,the image analysis routine fails to detect and measure accurate gap areas as shown inFigure 6E. For this reason wedid not fit the curve right to the gap closure. In fact, as the trendline is approximatelylinear, it is sufficient to end the experiment and fit the data up to the half-closuretime. For the experiment shown inFigure 6, the calculated t_1/2gap is3.27 hours and the cell sheet migration rate is 8.35 μm/h. While thet_1/2gap measurement can only be used to compare wound healing assayexperiments for which the initial gap is the exact same width, the cell migration rate ismore broadly comparable across datasets with varying wound areas, and is probably the bestmetric for quantifying the wound healing assay.

Figure 7 illustrates how thisquantitative approach can be applied to compare sheet migration rates between twodifferent cell lines. The gap area was quantified for each frame in each sequence usingImage Pro Premier's Wound Healing application module. Upon selection of anappropriate autothreshold level and spatial calibration, the module automatically runsthrough the sequence and outputs the gap area as a function of time. The data wereexported to Microsoft Excel, plotted as an XY scatter plot, and the slope associated witheach individual data point was obtained by fitting the data to a linear model. Themigration rate was determined from the slope of the line and the length of the gap byusingEquation 2. As shown in 7B, theaverage sheet migration rate for BEAS cells is a little more than double that of MCF7cells.

VI. Conclusions and Recommendations

Ideally, the wound healing assay would be carried out on a microscope that combinesautomated image capture, point visiting and incubation, so that multi-position measurementscan be carried out over time under the same conditions. These systems are particularlyuseful for experiments that require numerous samples, such as screening with drug or RNAilibraries. The acquisition of image sets at identical positions considerably simplifies theanalysis of gap closure, as the wound is sampled at multiple points and/or severalreplicates can be acquired in series.

Although we recommend using automated microscopes equipped for live-cell imaging,wound-healing assays of course can be accomplished using a wide variety of techniques andinstrumentation. The key message of this technical review is to design the experiment toenhance the reproducibility of the results. If the wound healing assay is carried out in astandard way, it enables the comparison between researchers and ensures reproducibility.Working according to a standard method can also help researchers achieve more accurateresults because they will not overlook parameters that affect their own experiment. We endthis technical review by summarizing the experimental guidelines for reproducibility of thewound-healing assay.

Cell culture

Choose a cell type that allows for reproducible cell culture conditions ifpossible.
For growth of cells, use the same volume of cell type-specific media and alwaysgrow cells in identical chambers. Include cell type specific requirements toestablish an ECM.
Always seed cells at the same density and start the wound healing assay at the samedegree of confluence.
Grow cells under the conditions specified for the chosen cell type. Ensure that theCO2 concentration is correct for the cell type.
Plate cells in a plastic-bottomed 24-well dish. The plastic bottom helps cells toadhere, which is vital for migration. The 24 wells are suitable for plating controland experimental conditions in parallel to ensure experiments are all carried out inthe same environment. The size of a 24-well dish is amenable to both scratching andcell culture inserts.

Gap creation

Choose a wounding approach that allows for small and reproducible wounds, therebyallowing for microscopy at higher resolution and better accuracy when measuringwound size and healing.
If a manual wound must be made, use consistent pressure and pipette tip angle tokeep the wound consistent.
If using expensive cells or reagents, practice making the wounds on inexpensivecell lines first.
Rinse the wound to remove debris, and replace with fresh medium and growth factorsas required.
Use an insert to make the gap a consistent width, and/or for small volumes of cellsor reagents.
Aim for a gap size of 0.5 mm, which allows for observation at 4x, 5x or 10xmagnification, and reasonable gap closure rates.

Recording wound healing with a microscope

Choose a microscope with environmental control, automatic acquisition over time andautomated stage controls for regular sampling of wound healing at multiplepositions. While motorized microscopes with built-in incubators may not be availablein individual laboratories, they are likely to be available in microscopy corefacilities.
Phase contrast imaging (adjusted for Koehler illumination) is generally moreappropriate than fluorescence for wound healing.
Record wound healing images until the gap is approximately half closed. In mostcases, there is no need to continue until full closure.
Use a 4x, 5x or 10x objective lens. Maximize the wound area in thefield-of-view.

Analysis

Develop a protocol for measurement in the early stages of the project. Ensure theprotocol is robust before complete recording of data sets.
Software is available for automated gap area measurements.
For manual gap measurements, use flat-field correction and edge detection toprepare the images for analysis. Set a threshold and use size criteria to select thegap area.
Plot gap area versus time, and fit a line to the data to determinet_1/2gap, the time it takes for the gap to close to half the originalarea, and v_migration, the cell migration rate.

Experimental details for the data sets illustrated in this technical review

Two separate microscope systems equipped with stage-top incubators, temperature,CO_2, and humidity were used to generate illustrative data sets for thistechnical review. The data sets shown inFigures 1,5 and6 were derived from HUVEC grown on plastic 24-welldishes to one day past confluence and then scratched with a pipette using a p20 tip. Thedish containing cells was allowed to settle on the microscope for one hour beforeinitiating image acquisition. The same points in multiple wells were imaged every15 min with a 10x phase objective using an Olympus IX71 microscope (Olympus Canada,Richmond Hill, Ontario) equipped with an automated stage and transmitted white-lightshutter (Prior Scientific, Rockland, MA). The software control of the image acquisitionwas through Volocity (Perkin Elmer, Waltham, MA). The data shown inFigures 4 and7 wereacquired from BEAS and MCF-7 cells seeded in ibidi inserts (Minitube Canada, Ingersoll,Ontario) in a plastic-bottomed 24-well dish that were incubated 12 hours after aconfluent monolayer was formed. The inserts were then carefully removed by peeling themback from one corner as per the manufacturer's instructions, and fresh culture mediumwas added to fill each well to about 25% of the full volume. The dish was placed ona motorized AxioObserver microscope (Carl Zeiss Canada, North York, Ontario) equipped withan incubator, hardware autofocus and Zen software (Carl Zeiss Canada, North York, Ontario)to automate image acquisition for multiple locations across the dish. Images were acquiredevery 20 min for approximately 24 hours with a 10x phase objective.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors would like to thank Dr. Kamala Patel of the University of Calgary for valuablediscussions and for experimental support. We thank Dr. Hong Zhang for preparing humanumbilical vein endothelial cells for the illustrative examples. We also thank ChelseaDoktorchik and Dr. Simon Hirota of the University of Calgary for testing and critiquingstandard protocols for image acquisition and analysis developed by the authors.

Funding

The Live Cell Imaging authors would also acknowledge the Snyder Institute for ChronicDiseases for financial support.

References

1.Ilina O, Friedl P.Mechanismsof collective cell migration at a glance. J Cell Sci2009;122:3203-8. [DOI] [PubMed] [Google Scholar]
2.Friedl P, Gilmour D.Collectivecell migration in morphogenesis, regeneration and cancer. NatRev Mol Cell Biol2009;10:445-57. [DOI] [PubMed] [Google Scholar]
3.Friedl P, Wolf K.Plasticityof cell migration: a multiscale tuning model. J CellBiol2010;188:11-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Vedula SR, Ravasio A, Lim CT, Ladoux B.Collectivecell migration: a mechanistic perspective. Physiology(Bethesda)2013;28:370-9. [DOI] [PubMed] [Google Scholar]
5.Rorth P.Collective cell migration. Annu Rev Cell DevBiol2009;25:407-29. [DOI] [PubMed] [Google Scholar]
6.Rorth P.Fellow travellers: emergent properties of collective cellmigration. EMBO Rep2012;13:984-91. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Pouliot N PH, Burrows A.InvestigatingMetastasis Using In Vitro Platforms. Madame Curie Bioscience Database.Austin: (TX: ):Landes Bioscience, 2000. [Google Scholar]
8.Lecomte N, Njardarson JT, Nagorny P, Yang G, Downey R, Ouerfelli O, Moore MA, Danishefsky SJ.Emergenceof potent inhibitors of metastasis in lung cancer via syntheses based onmigrastatin. Proc Natl Acad Sci U S A2011;108:15074-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Fujisawa T, Rubin B, Suzuki A, Patel PS, Gahl WA, Joshi BH, Puri RK.Cysteaminesuppresses invasion, metastasis and prolongs survival by inhibiting matrixmetalloproteinases in a mouse model of human pancreatic cancer.PLoS One2012; 7:e34437. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Hai J, Zhu CQ, Bandarchi B, Wang YH, Navab R, Shepherd FA, Jurisica I, Tsao MS.L1 celladhesion molecule promotes tumorigenicity and metastatic potential in non-small celllung cancer. Clin Cancer Res2012;18:1914-24. [DOI] [PubMed] [Google Scholar]
11.Weijer CJ.Collectivecell migration in development. J Cell Sci2009;122:3215-23. [DOI] [PubMed] [Google Scholar]
12.Martin P, Parkhurst SM.Parallelsbetween tissue repair and embryo morphogenesis.Development2004;131:3021-34. [DOI] [PubMed] [Google Scholar]
13.Reinhart-King CA.EndothelialCell Adhesion and Migration. Methods of Enzymology:Elsevier, Inc.,2008:45-61. [DOI] [PubMed] [Google Scholar]
14.Kim JH, Serra-Picamal X, Tambe DT, Zhou EH, Park CY, Sadati M, Park J-A, Krishnan R, Gweon B, Millet E, et al. Propulsion and navigation within the advancing monolayersheet. Nat Mater2013;12:856-63. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Vitorino P, Meyer T.Modularcontrol of endothelial sheet migration. Genes Dev2008;22:3268-81. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Simpson KJ, Selfors LM, Bui J, Reynolds A, Leake D, Khvorova A, Brugge JS.Identification of genes that regulate epithelial cell migration using ansiRNA screening approach. Nat Cell Biol2008;10:1027-38. [DOI] [PubMed] [Google Scholar]
17.Chapnick DA, Liu X.Leader cellpositioning drives wound-directed collective migration in TGFbeta-stimulated epithelialsheets. Mol Biol Cell2014;25:1586-93. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Tsai FC, Seki A, Yang HW, Hayer A, Carrasco S, Malmersjo S, Meyer T.A polarizedCa2+, diacylglycerol and STIM1 signalling system regulates directed cellmigration. Nat Cell Biol2014;16:133-44. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Trepat X, Fredberg JJ.Plithotaxisand emergent dynamics in collective cellular migration. TrendsCell Biol2011;21:638-46. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Reffay M, Parrini MC, Cochet-Escartin O, Ladoux B, Buguin A, Coscoy S, Amblard F, Camonis J, Silberzan P.Interplay ofRhoA and mechanical forces in collective cell migration driven by leadercells. Nat Cell Biol2014;16:217-23. [DOI] [PubMed] [Google Scholar]
21.Khalil AA, Friedl P.Determinantsof leader cells in collective cell migration. IntegrativeBiology2010;2:568-74. [DOI] [PubMed] [Google Scholar]
22.Bindschadler M, McGrath JL.Sheetmigration by wounded monolayers as an emergent property of single-celldynamics. J Cell Sci2007;120:876-84. [DOI] [PubMed] [Google Scholar]
23.De Rossi G, Whiteford JR.A novelrole for syndecan-3 in angiogenesis. F1000Res2013; 2:270. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Fong E, Tzlil S, Tirrell DA.Boundarycrossing in epithelial wound healing. Proc Natl Acad Sci U SA2010;107:19302-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Shimizu Y, Boehm H, Yamaguchi K, Spatz JP, Nakanishi J.Aphotoactivatable nanopatterned substrate for analyzing collective cell migration withprecisely tuned cell-extracellular matrix ligand interactions.PLoS One2014; 9:e91875. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Wang Y, Wang G, Luo X, Qiu J, Tang C.Substratestiffness regulates the proliferation, migration, and differentiation of epidermalcells. Burns : journal of the International Society for BurnInjuries2012;38:414-20. [DOI] [PubMed] [Google Scholar]
27.Yarrow JC, Totsukawa G, Charras GT, Mitchison TJ.Screeningfor cell migration inhibitors via automated microscopy reveals a Rho-kinaseinhibitor. Chem Biol2005;12:385-95. [DOI] [PubMed] [Google Scholar]
28.Hulkower K. I., Herber R.L.Cell Migration and Invasion Assays as Tools for DrugDiscovery. Pharmaceutics2011;3:107-24. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Philippeos C, Hughes RD, Dhawan A, Mitry RR.Introduction to cell culture. Methods MolBiol2012;806:1-13. [DOI] [PubMed] [Google Scholar]
30.Clause KC, Barker TH.Extracellular matrix signaling in morphogenesis andrepair. Curr Opin Biotechnol2013;24:830-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Mason BNJC, Joseph P., and Reinhart-King,Cynthia A.MatrixStiffness: A Regulator of Cellular Behavior and Tissue Formation. In:Bhatia SK, ed. EngineeringBiomaterials for Regenerative Medicine: Novel Technologies for ClinicalApplications: Springer Science+BusinessMedia,2012:19-37. [Google Scholar]
32.Davis PK, Ho A, Dowdy SF.Biologicalmethods for cell-cycle synchronization of mammalian cells.Biotechniques2001; 30:1322-6, 8,30-1. [DOI] [PubMed] [Google Scholar]
33.Pirkmajer S, Chibalin AV.Serumstarvation: caveat emptor. Am J Physiol Cell Physiol2011;301:C272-9. [DOI] [PubMed] [Google Scholar]
34.Ashby WJ, Zijlstra A.Establishedand novel methods of interrogating two-dimensional cell migration.Integr Biol (Camb)2012;4:1338-50. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Arciero JC, Mi Q, Branca M, Hackam D, Swigon D.Using acontinuum model to predict closure time of gaps in intestinal epithelial celllayers. Wound Repair Regen2013;21:256-65. [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Vedula SR, Leong MC, Lai TL, Hersen P, Kabla AJ, Lim CT, Ladoux B.Emergingmodes of collective cell migration induced by geometrical constraints.Proc Natl Acad Sci U S A2012;109:12974-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Straatman K.WoundHealing Assay. University of Leicester, College of Medicine, Biological Sciences andPsychology, Advanced Imaging Facilities,2008. [Google Scholar]
38.Inoué S, Spring KR.Video Microscopy :the Fundamentals. New York:Plenum Press, 1997. [Google Scholar]
39.Murphy DB, Davidson MW.Fundamentals ofLight Microscopy and Electronic Imaging. Hoboken,N.J.: Wiley-Blackwell,2013. [Google Scholar]
40.Goldman RD, Spector DL.Live cell imaging: a laboratory manual. Cold Spring Harbor, N.Y.:Cold Spring Harbor Laboratory Press,2005. [Google Scholar]
41.Kulesa PM, Kasemeier-Kulesa JC.Construction of a Heated Incubation Chamber around a Microscope Stage forTime-Lapse Imaging. CSH Protoc2007; 2007:pdb prot4792. [DOI] [PubMed] [Google Scholar]
42.Salmon ED, Canman JC.Properalignment and adjustment of the light microscope. Curr ProtocImmunol2002; Chapter 21:Unit 21.1. [DOI] [PubMed] [Google Scholar]
43.http://www.mediacy.com. [Google Scholar]
44.Geback T, Schulz MM, Koumoutsakos P, Detmar M.TScratch: anovel and simple software tool for automated analysis of monolayer wound healingassays. Biotechniques2009;46:265-74. [DOI] [PubMed] [Google Scholar]
45.http://www.ibidi.com. [Google Scholar]
46.Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, et al. Fiji: an open-source platform for biological-imageanalysis. Nat Methods2012;9:676-82. [DOI] [PMC free article] [PubMed] [Google Scholar]

Movatterモバイル変換

PERMALINK

An introduction to the wound healing assay using live-cellmicroscopy

James E N Jonkman

Judith A Cathcart

Feng Xu

Miria E Bartolini

Jennifer E Amon

Katarzyna M Stevens

Pina Colarusso