Cell Motility and the Cytoskeleton 37:33–43 (1997) Cell Migration and Proliferation During the In Vitro Wound Repair of the Respiratory Epithelium Jean-Marie Zahm,* Hervé Kaplan, Anne-Laure Hérard, Fabrice Doriot, Denis Pierrot, Pascal Somelette, and Edith Puchelle INSERM U314, IFR53, CHU Maison Blanche, Reims, France The respiratory epithelium is frequently injured by inhaled toxic agents or by micro-organisms. The epithelial wound repair represents a crucial process by which surface respiratory cells maintain the epithelial barrier integrity. The repair process involves both cell migration and proliferation, but as yet, the kinetic of these two mechanisms has not been extensively studied. Using an in vitro model of human respiratory epithelium wound repair, proliferative cell immunofluorescent staining and a computer-assisted technique allowing the tracking of living cells, we studied the cell proliferation and migration during the wound repair process. Respiratory epithelial cells were dissociated from human nasal polyps and cultured on a collagen I matrix. At confluency, a chemical wound was made on the culture. We observed that the cell mitotic activity peaked at 48 h after wounding (23% of the cells) and mainly concerned the cells located 160 to 400 µm from the wound edge. The migration speed was highest (35 to 45 µm/h) for the spreading cells at the wound edge and progressively decreased for the cells more and more distant from the wound edge. The temporal analysis of the cell migration speed during the wound repair showed that it was almost constant during the first 3 days of the repair mechanism and thereafter dropped down until the wound closure was completed (after 4 days). We also observed that over a 1-hour period, the intra-individual and interindividual variation of the cell migration speed was 43% and 37%, respectively. These results demonstrate that cell proliferation and cell migration during respiratory epithelial wound repair are differently expressed with regard to the cell location within the repairing area. Cell Motil. Cytoskeleton 37:33–43, 1997. r 1997 Wiley-Liss, Inc. Key words: airway wound; migrating cells; proliferative cells; spreading cells; imaging analysis; cell tracking INTRODUCTION The respiratory epithelium is frequently injured by inhaled toxic agents or micro-organisms, leading to the alteration of the epithelium barrier integrity. Whatever the source of injury, the response of the respiratory surface epithelium to an acute injury can be characterized by a succession of cellular events varying from the loss of surface epithelial impermeability to the desquamation of cells from the epithelium. The epithelial wound repair represents a crucial process by which the remaining surface respiratory cells restore the epithelial barrier integrity. Numerous models have been used to explore the r 1997 Wiley-Liss, Inc. capacity of the respiratory epithelium to repair following injury. Most of these models have been developed in vivo in different animal species and have allowed the deduction that denudation of the airway epithelial surface leads Contract grant sponsors: Association Française de Lutte Contre la Mucoviscidose (AFLM), Pôle Technologique Régional GBM, INSERM (Contrat Normalisé d’Etude Pilote), Ministère de l’Enseignement Supérieur et de la Recherche. *Correspondence to: Jean-Marie Zahm, INSERM U314, CHU Maison Blanche, 45, rue Cognacq-Jay, 51092 Reims Cedex, France. Received 19 July 1996; accepted 18 November 1996. 34 Zahm et al. to the surrounding epithelial cell migration and proliferation [Gordon and Lane, 1976; Keenan et al., 1982; McDowell et al., 1979; Nikula et al., 1988; Shimizu et al., 1994]. However, the kinetic of these two major processes during the wound repair has not been extensively studied, due to the difficulty introduced by in vivo models. In addition, in all these animal models of injury and repair, the inflammatory stimulus engages cellular factors and serum-derived molecules into the process of repair. To overcome the complex association of factors in the in vivo models, culture models of respiratory epithelial cells allow more clearly the identification of the cellular and molecular alterations associated with the wound repair processes [Zahm et al., 1991, 1993]. At the molecular level, alterations in cytoskeletal protein patterns, binding of growth factors and interactions with extracellular matrix proteins and metalloproteinases through specific cells receptors (integrins), can govern the cell migration in the repairing area. The flattened cells which migrate into the wounded site change their phenotype, becoming poorly differentiated cells expressing keratin 14 and vimentin intermediate filaments [Shimizu et al., 1994; Buisson et al., 1996a]. The cell spreading and migration which occur during wound repair depend on cell-matrix interactions mediated by fibronectin and one of its cellular receptors, the a5b1 integrin [Hérard et al., 1996]. Recent results have also documented the increase and localization of gelatinolytic metalloproteinases associated with the migratory process of epithelial cells during the respiratory epithelium repair [Buisson et al., 1996b]. However, up to now, the specific role of cell migration and cell proliferation in the wound repair process of the respiratory epithelium has not been extensively documented. In the present study, we used an in vitro wounding model of dissociated respiratory epithelial cells cultured from human nasal polyps to evaluate the dynamics of the cell migration and proliferation during wound repair. We also reported in the present work, a novel method, using image analysis, to provide data on the migration kinetics of either individual cells or a cell population during the wound repair process. We observed that the cell mitotic activity peaked at 48 h after wounding. The temporal analysis of the cell migration speed showed that, despite large intra-individual and interindividual variations, the overall migration speed of the cell population at the wound edge remained fairly constant during the first 3 days of repair and decreased progressively up to the wound closure. in RPMI 1640 culture medium (Gibco, Grand Island, NY). The nasal tissue was then dissociated by enzymatic digestion (pronase 0.1%) for 12 h. The dissociated surface epithelium was removed from the tissue by gentle agitation and enzymatic digestion was stopped by adding 10% calf serum (Seromed, Biochrom, Berlin, Germany). The cellular pellet collected after centrifugation at 150 3 g for 10 min was constituted by isolated mucous cells, ciliated or basal cells, small clumps of reaggregated cells and small epithelial sheets. These cells were resuspended in a serum-free medium and were plated on a type I collagen gel matrix prepared from rat tail tendons according to the technique developed in our laboratory and earlier described by Chevillard et al. . The cells were cultured in an humidified atmosphere containing 95% air and 5% CO2 in RPMI 1640 medium supplemented with 1 µg/ml insulin (Sigma, L’Isle D’Abeau, France), 1 µg/ml transferrin (Sigma), 10 ng/ml epidermal growth factor (Serva, Heidelberg, Germany), 0.5 µg/ml hydrocortisone (Sigma), 10 ng/ml retinoic acid (Sigma), 100 U/ml penicillin, and 100 µg/ml streptomycin. The use of type I collagen gel and growth factor-supplemented culture medium is the most suitable condition to achieve differentiated cultures of human nasal epithelial cells [Chevillard et al., 1993]. In Vitro Epithelial Wounding After 3 days in culture, when the cells had reached confluency, the culture medium was removed from the culture dish. A 2-µl drop of NaOH 0.1 M was deposited in the center of the culture dish and immediately diluted in 1 ml of culture medium. The cells in contact with the NaOH drop desquamated from the collagen I matrix, creating a circular wound of about 30 mm2 in area. The wounded culture was then rinsed with 1 ml of culture medium and incubated in air with 5% CO2 at 37°C. Wound Area Determination Every 12 h during the wound repair process, the wounded culture dish was placed on the stage of an inverted microscope (Nikon TMS-F) equipped with a 31 objective and a video CCD camera (Cohu 4700). The video image of the wound was displayed on a video monitor and the wound edge was drawn manually. From this drawing, the wound area was determined and expressed as a function of time. Cell Proliferation MATERIALS AND METHODS Respiratory Epithelial Cell Culture Nasal polyps were obtained from patients undergoing nasal polypectomy and were immediately immersed For cell proliferation quantitation, wounded cultures were cryofixed every 24 h throughout the wound repair process. The central area of the culture (about 200 mm2 in area) in which the wound was performed, was cut Cell Proliferation and Migration in Wound Repair out from the culture dish, embedded in OCT compound, immersed in liquid nitrogen for 5 min and then kept at 280°C. Thick sections (5 µm) perpendicular to the wounded culture were obtained using a Reichert Cryocut, sliced on a glass slide and dehydrated in air. For immunostaining, the sections were immersed in methanol at 220°C for 5 min and rinsed in 0.1 M phosphatebuffered saline (PBS). Each section was then incubated for 1 h with a mouse monoclonal antibody (MIB-1, Immunotech, France) which recognizes the Ki67 nuclear antigen and which was diluted 1/50 in 0.1 M PBS and thereafter incubated with a biotin-conjugate goat antimouse IgM, diluted 1:25 in 0.1 M PBS for 1 h. The cells were then stained with streptavidin-FITC for 30 min. The sections were mounted in glycerol-PBS-1,4 diazabicyclo 2-2 octane and observed under a Zeiss Axiophot microscope equipped with epifluorescence illumination. Successive microscopic fields of 40 µm in length radiating out from the wound edge were observed. In each field, the number of cells being positively stained for the Ki67 was quantified and expressed as a percentage of proliferative cells with regard to the total number of cells. Video Recording of the Cell Migration Immediately after wounding, cell cultures were incubated for 15 min with Hoechst 33258 (Sigma) at 0.1 mg/ml in culture medium, allowing the fluorescent staining of the nucleus in living cells. The culture was then washed twice with culture medium to remove the excess of fluorescent dye. Following that, the wounded culture dish was placed on the stage of a Zeiss IM35 inverted microscope and was enclosed in a small transparent culture chamber (Climabox) with 5% CO2 in air at 37°C. The microscope was equipped with an epifluorescence illumination (Hg lamp) through an excitation filter at 360 nm, and an emission filter at 510 nm, and with a low level SIT camera (Lhesa 4036). A special electronically operated shutter in the excitation light path was developed in order to automatically illuminate the culture at short periods of time, and to simultaneously record the fluorescent images. This device, based on timing semiconductor circuits, allowed the definition of the recording time of each video sequence (from 1 to 99 s) and the time interval between two successive sequences (from 1 to 99 min). Experimental Procedure Video recordings of the fluorescent-stained cell nucleus were performed following three sets of experiments during the wound repair process. In the first set of experiments, we recorded for 2 s every 12 h the cell nucleus at the wound edge, throughout the wound repair process. In the second set of experiments, at day 2 of the wound repair, we recorded successive microscopic fields which were radiating out from the wound edge. In the third set of experiments, at day 2 of the wound repair, we 35 recorded for 2 s every 5 min, the fluorescent nucleus of cells in the same microscopic field at the wound edge. From these video recordings, we used an image analysis technique that we developed to analyze the migration of individual cells forming continuous sheets. In one experiment, using phase-contrast illumination, we continuously recorded the cells at the wound edge for a 10-min period in order to visualize the cell spreading. Cell Tracking and Cell Migration Measurement The images were digitized as a 766 3 574 pixels and 24-bit array from the video recordings, using a Sparc-Classic workstation equipped with a XVideo card (Parallax Graphics, Santa Clara, CA). Before treatment, the images were reduced to 512 3 512 pixels and 8-bit grey levels. We developed a software in C language using the Xlibrary (version 1.1 of the XWindow system) and a tool kit available in XView (Sun Microsystems, Mountain View, CA). The software is user-interfaced and has three main functions: (1) the detection of the cell nucleus, (2) the computation of the trajectories of the nucleus, and (3) the analysis of the nucleus trajectories. The detection of the cell nucleus can be done manually by using the computer mouse or automatically by using image segmentation methods based on thresholding. The tracking of several nuclei together is done by using trajectory-based methods [Aggarwal and Nandhakumar, 1988]. The general scheme of these algorithms is to minimize the mean cost of all the computed trajectories, using one or several parameters such as the regularity of cell motion, the low deviation of the cell trajectory or the nearness of the successive positions of the cells. For the main part, we used nearness and low deviation of the trajectories as parameters of minimization for several successive images and nearness parameters for only two successive images. After trajectory computation, the software draws all the trajectories of the selected nucleus. From this drawing, each nucleus trajectory can be selected individually and the migration speed is calculated from one nucleus position to the other. The mean migration speed for all the nucleus trajectories is also calculated. The software is able to track a cell population of about 100 cells together. The computation time for 100 cells in eight images is about 2 s using an Unix Workstation (Sun Microsystems). Statistical Analysis All the data are expressed as mean 6 S.D. (standard deviation of the mean). The unpaired Student’s t test was used to determine significance (defined as p , 0.05). Curve fitting was performed on a Macintosh Quadra 650 using CA-Cricket Graph III (Computer Associates, New York, NY). 36 Zahm et al. Fig. 1. Set of low magnification (31 objective) micrographs showing the in vitro wound repair process of the respiratory epithelium. The first micrograph (t0) was taken immediately after wounding. The following micrographs were taken every 30 h during the wound repair. The progressive closure of the wound is observed. Bar 5 1 mm. RESULTS Wound Repair Process Human surface respiratory epithelial cells cultured on a type I collagen reached confluency within 2 to 3 days. The contact of a drop of NaOH with a confluent respiratory epithelial cell culture induced a localized injury characterized by the cell lysis and desquamation and the denudation of the collagen matrix on which the cells were growing. The chemically induced wounds in the respiratory epithelial cell culture consisted of circular holes. Microphotographs taken using an inverted microscope at low magnification (31 objective), immediately after wounding and thereafter every 30 h during wound repair, show the progressive closure of the wounded surface (Fig. 1). The early reepithelialization pattern displayed by the respiratory cells immediately after injury, appeared to involve the spreading of the cells at the wound edge and the movement of cell sheets toward the wound [Zahm et al., 1991]. The migration process is uniformly distributed around the wound surface as suggested by the concentricity of the successive wound surfaces. A regular feature which progressively appeared during the wound repair is the puckering of cell sheets in the repaired area (Fig. 1, t90). The observation of semithin sections of wounded cultures revealed that the puckers radiating around the wound are made of multilayered cells. The mean wound surface plotted versus time decreased sharply up to the 50th hour of repair (Fig. 2). A third degree polynomial fit through the experimental data gave the following curve equation: Wa(t) 5 27.99 2 0.37t 2 1.28 3 1023t2 2 2.14 3 1025t3 Fig. 2. Temporal evolution of the wound surface during the wound repair process of the respiratory epithelial cells, whose micrographs are shown in Figure 1. Each point represents the mean 6 S.D. of three experiments. The curve corresponds to a third-degree polynomial fit through the experimental data. The decrease in wound area (0.37 mm2 per hour) appears linear up to 60 h and then rapidly flattens out up to wound closure. where Wa is the wound area and t the time of repair. As shown by the equation of the fit curve, the first-order coefficient corresponded to the linear part of the fit curve. The decrease in wound area appeared therefore linear up to 50 h of repair due to the fact that the second- and third-order coefficients of the fit equation interfered only about 10% in the rate of decrease. However, on and after the 60th hour of repair, the third-order coefficient acts as a factor flattening out the curve and alters the rate of decrease by 16%, 39% and 77% at 60, 80 and 100 h of repair, respectively. This observation indicates that the repair process becomes slower when most of the injured surface was recovered by the repairing cells. Cell Proliferation In order to assess whether the in vitro model of respiratory epithelium wound repair involved cell proliferation, wounded cultures at different days of the repair process were immunofluorescently stained for the nuclear antigen Ki67 which is a marker of cycling cells. The micrograph in Figure 3a shows a number of Ki67positive cells in the culture 2 h after wounding (day 0). As shown in Figure 3b, the number of proliferative cells in the repairing area increased dramatically at 48 h after wounding (day 2). When the wound closure was completed (day 4), the number of proliferative cells fell down to a value similar to that obtained at day 0 (data not Cell Proliferation and Migration in Wound Repair 37 Fig. 3. Proliferative activity of respiratory epithelial cells during the in vitro wound repair. The proliferative cells were detected by immunofluorescence of the Ki67 antigen. a: A few Ki67-positive cells are observed at day 0 of wound repair. b: Forty-eight hours after wounding, numerous Ki67-positive cells are present in the repairing area. c: Percentage of proliferative cells at day 0, day 2 and day 4 of the wound repair process. A significant increase in the percentage of proliferative cells was observed at day 2 ( p , 0.05). At wound closure (day 4), the percentage of proliferative cells was not significantly different as compared with day 0. Each bar represents the mean percentage 6 S.D. of proliferative cells quantitated on at least 12 different repairing areas. d: Spatial evolution, at day 2, of the percentage of proliferative cells in the repairing area. A significant increase ( p , 0.05) in the percentage of proliferative cells was observed at a distance of 80 to 320 µm from the wound edge. Each value represents the mean 6 S.D. of three measurements. shown). Figure 3c illustrates the variations in the proliferative activity of the cells in the repairing area. The percentage of proliferative cells was 7.1 6 7.0% at day 0, increased significantly ( p , 0.05) up to 23.0 6 25.6% at day 2 and decreased to 2.8 6 3.0% at day 4 when the wound closure occurred. In order to assess where the proliferative cells were located, we quantified their number in areas more and more distant from the wound edge. The variation at day 2 in the percentage of proliferative cells in relation to the distance from the wound edge is represented in Figure 3d. A slight increase in the number of proliferative cells was observed at a short distance from the wound edge. The cells located in the repaired area at a distance of 80 to 320 µm from the wound edge were characterized by a significant increase ( p , 0.05) in their proliferative activity. The decrease in proliferation between 160 µm and 240 µm was not significant compared to 160 µm, but the increase in proliferation at 320 µm was significantly higher ( p , 0.05) as compared to the proliferation at 250 µm. On the contrary, the proliferation of cells located at a distance higher than 400 µm from the wound edge dramatically decreased to only 1–2%. These results demonstrate that the proliferative activity is maximal at day 2 of the repair process and mainly concerns the cells located within an area behind the first cell rows of the wound border. Cell Spreading To analyze the dynamics of cell motility at the wound edge of wounded respiratory epithelial cell cultures, video recordings of the wounded cultures were made under phase-contrast illumination. From 60 images recorded every min for 1 h, a movie was produced, allowing the visualization of lamellipodia dynamics of the cells located at the wound edge. Four pictures 38 Zahm et al. Fig. 4. Series of phase-contrast images, taken at 3-min intervals, of respiratory epithelial cells migrating and extending lamellipodia at the wound edge. The leading edge of the cell lamellipodia has been visualized by a white line. The cells located at the wound edge are characterized by a functional polarity (monopolar) with a continuously changing lamellipodia shape. Bar 5 10 µm. extracted from the movie every 3 min are presented in Figure 4. In order to better visualize the front of the cell lamellipodia, its border was underlined by a white drawing. The lamellipodia formation was a characteristic feature of the cells at the front row of the wound edge. The lamellipodia remained strictly localized at the ‘‘head’’ or front of migrating cells. We can easily observe that the shape of the cell edge is continuously modified during the spreading process, but the cell functional polarity (monopolar) remained constant throughout the repair process, leading to the unidirectionality of cell movement. Cell Migration To track the migratory cells repairing the wound, we stained the cell nuclei with Hoechst fluorescent dye. The fluorescent images obtained under these conditions are less complex than phase-contrast images (see Fig. 4) and therefore are more easily analyzed with the image analysis software specifically developed for the cell velocity measurement (Fig. 5). The micrographs in Figure 5a show cell nuclei stained with the Hoechst dye at the edge of a wound at day 2 of the wound repair. Figure 5b represents the same nuclei recorded 60 min later. The trajectories determined from the field analyzed every 15 min over the 60-min period of observation were fairly straight (Fig. 5c and d). As shown in Figure 5d, the software that we developed allowed us to calculate the angular deviation (a1, . . . , a4) from the horizontal of the successive trajectories of the nuclei. The mean deviation angle for all the selected nuclei was 0.30 6 0.11 rd, which confirmed the uniform directionality of the cell migration. These trajectories represent a typical behavior of the radian migrating cells which actively participated to the wound repair process. Cell Proliferation and Migration in Wound Repair 39 Fig. 5. Quantitative evaluation of cell migration velocity. a: Fluorescent staining with a DNA fluorescent dye (Hoechst 33258) of the nucleus of respiratory epithelial cells at the edge of a wounded culture at day 2 of the wound repair process. b: The same cells recorded 1 h later. c: Trajectories of cell nuclei recorded over a 60-min period. d: Detailed trajectories of a particular nucleus analyzed every 15 min for 1 h. The angle a represents the deviation with the horizontal of the nucleus trajectory. e: Temporal evolution of the migration velocity of cells at the wound edge throughout the wound repair process. A significant decrease in the migration speed was observed at 72 h. Each bar represents the mean 6 S.D. of 15 measurements. f: Temporal evolution for 1 h of the migration speed of cells at the wound edge. The same 15 cells were recorded every 5 min for 1 h. Huge variations were observed from one cell to the other. Each bar represents the mean 6 S.D. of the migration speed for the 15 cells. The migration speed of the cells at the wound edge during the wound repair process is represented in Figure 5e. We measured an almost constant speed (35 to 40 µm/h) for up to 60 h of repair, followed by a significant decrease from hour 72 until the wound closure. When compared with the wound area decrease reported in Figure 2, the time of 60 h may be significant, because it roughly corresponded to the time up to where the decrease in wound area was roughly constant. Conse- quently, the decrease in the cell migration velocity induced an alteration in the wound area decrease rate. Figure 5f represents the mean migration speed of a cell population measured every 5 min for 1 h at the wound edge at day 2. The high standard deviation of the migration speed calculated every 5 min for the cell population indicated that huge variations (interindividual variation 5 37%) in cell migration speed were observed for different cells during the same time period. In 40 Zahm et al. Fig. 6. Spatial evolution of the migration speed of cells at day 2 of the wound repair process. The cell migration speed progressively and significantly ( p , 0.05) decreased with the increase in the distance from the wound edge. Each bar represents the mean 6 S.D. of 15 measurements. addition, the mean coefficient of variation of the migration speed for the same cell measured every 5 min for a 1-h period was 43% (5intra-individual variation), indicating that the migration speed of one given cell was not constant over a 1-h period. The mean migration speed of cells measured in successive areas progressively more distant from the wound edge is represented in Figure 6. We observed that the cell migration speed continuously and significantly ( p , 0.05) decreased with the increase in the distance from the wound edge. The epithelial cells at the wound edge moved faster than the cells which were distant from the wound edge. At a distance higher than 1.6 mm from the wound edge, the cell migration was no longer detectable. DISCUSSION The experiments described in the present work show that the respiratory epithelium is able to repair wounds through a combination of cell migration and cell proliferation. Cell proliferation was quantified through the Ki67 antigen which is a marker of cycling cells, i.e., cells in G1, S, G2 and M phases. In our in vitro model of respiratory epithelial wound repair, cell proliferation took place preferentially in the repairing area, with a mitotic activity which peaked at 48 h. When the wound closure was completed, the mitotic activity subsided. These results are consistent with the recent results of Shimizu et al.  who demonstrated a similar pattern of mitotic activity (24% of proliferative cells at 48 h after wounding) in the in vivo wound repair process of the rat tracheal epithelium. The major event in the wound repair process is the cell spreading and migration which allows the epithelium recovery of the denuded collagen matrix. Time-lapse video recordings associated with computer methods allow the continuous monitoring of the process of cell spreading and migration. As seen in Figure 4, under phase-contrast microscopy observation, it is difficult to individualize the cells and to monitor their movement in a confluent cell culture. To overcome this problem, we stained the cell nucleus with a DNA dye (Hoechst 33258) which is excitable by UV illumination. Taking into account what we have previously demonstrated, that the duration of UV illumination can induce dramatic changes in the physiological state of the cells [Zahm et al., 1994], we used an electronic timer which allowed us to reduce the UV illumination time to only a few seconds for each video recording. To track the moving cells, we used trajectory-based methods [Aggarwal and Nandhakumar, 1988] which commend several assumptions. Firstly, the nuclei, which are used as markers of the cells, are considered as a rigid body and are indistinguishable from each other. Secondly, each nucleus is tracked using only one representative point: the center of the nucleus. While the cells are tracked using the center of the nucleus as a single marker, only translation movements of the cells are observable. Other movements, such as cell rotation, for example, are not detectable. The trajectory-based method raises the problem of the occlusions which concern the objects visible in an image and no longer visible in the following image of the sequence. We also have to consider the case of objects which are not present in an image and become visible in the following images. Some algorithms dealing with the occlusions have been described [Fletcher et al., 1991; Rangarajan and Shah, 1991; Sethi and Jain, 1987], but these algorithms generally only take into account the missing objects in successive image sequences. In the images from our wound repair model, the number of nucleus in an image is continuously increasing throughout the set of images. To overcome this problem, we used an algorithm which considered that the number of tracked nuclei was constant in all of the successive images. With that purpose, the nuclei were selected manually in the set of images. This is done easily and quickly using the computer mouse and is facilitated by the cell nucleus staining which made the nuclei clearly visible in the images. Two sets of methods for establishing the correspondence between the selected points are available. The first methods are based on the regularity of the migration of Cell Proliferation and Migration in Wound Repair the objects [Sethi and Jain, 1987] and are called ‘‘smoothing methods.’’ To determine the trajectories of the nuclei, we have to maximize the smoothest set of trajectories. Smoothing can be controlled by several factors such as nearness, speed of motion or deviation of the trajectories. The second methods are predicting methods [Fletcher et al., 1991], by which the estimation of the position of the objects in the image is computed using information from the partial trajectory of the objects. The object which is nearest to the calculated position is chosen and the predicting factors corrected according to the knowledge of this new position. The process must be initiated using the optical flow method [Uras et al., 1988] or a smoothing method, on at least three images. Several techniques have been described for analyzing cell motility, but most of these techniques are applied to isolated migrating cells. Tatsuka et al.  described a method for quantitative measurement of postconfluent cell populations, but the latter authors measured a motility index of a cell population and not the locomotion speed of individual cells. The main advantage of the computerassociated technique that we have developed is that we can track a population of confluent cells over a long period of time, and measure their individual migration speed, allowing the study of the factors controlling and coordinating the locomotory machinery. The ability of cells to move is crucial for many biological processes, such as embryogenesis and wound healing. In this current study, we analyzed the cell migration during the wound repair process of the respiratory epithelium. The cells at the wound edge are characterized by a mean migration speed ranging from 35 to 45 µm/h during the first 60-h period of the wound repair process. We have previously shown [Zahm et al., 1992] that in a smaller mechanically made wound (0.03 mm2 in surface), the cell migration speed reached 26 µm/h. But in this latter wound model, the wound closure occurred within 6 to 8 hours. The decrease in the migration speed measured at the end of the wound repair process is most likely due to contact inhibition effects. Recently, Lee et al. , by analyzing the measurements of cell path, in conjunction with local cell velocity and the distance to the nearest neighbor of a cell, have demonstrated, in detail, the behavior of a cell as it migrates and interacts with other cells. These authors showed that as two cells approach each other, their speed decreases, whereas when a cell speeds away from its neighbor, its speed increases quickly. This behavior of neighboring cells could also explain the huge variations in cell locomotion that we observed when analyzing the local movement of a cell population at the wound edge for a short period of time (Fig. 5). In analyzing the migration of cell populations located at an increasing distance from the wound edge, 41 we observed a negative relationship between the cell migration speed and the distance from the wound edge: the cells located far from the wound edge moved more slowly than the cells located at the wound edge. As shown in Figure 3, the maximum increase in cell proliferation occurred in the intermediate area of the repaired wound. Lee et al.  have shown that an increase in cell density induced by cell proliferation is immediately accompanied by the decrease in the average speed of migration of a cell population. In addition, several years ago, Abercrombie and Heaysman  showed a significant inverse relationship between the speed of movement of a cell and the number of other cells with which it was in contact during the observed movement. The decrease in the cell migration that we observed in the intermediate area of the repairing cultures could therefore be related to an increase in the proliferation rate of these cells. The decrease in cell migration at the end of the wound repair process could be related to a contact inhibition of movement occurring when junction has been established between opposing sheets of cells, as earlier described by Abercrombie and Heaysman . Leading cells, migrating toward the wound, are polarized in that they extend lamellipodia only along the free edge of the wound. This phenomenon has been described as ‘‘contact-stimulated migration’’ by Thomas and Yamada  and could give direction to the migratory path. The movement of cells in a given direction has been related to the fact that they do not usually move in a direction that will take them over the surface of another cell, but they are free to move in any other direction. In a wounded area, cells will consequently be released from contact inhibition of locomotion and move toward the denuded area [Abercrombie and Heaysman, 1970]. An important feature to consider is the highly dynamic formation of lamellipodia at the leading edge of the cells located at the front of the wound. Protrusion of the cell membrane is coupled to polymerization of actin filaments at the leading edge. Different mechanisms for generating protrusive force have been proposed [Mitchison and Cramer, 1996]. Either motor proteins could drive protrusion, or actin polymerization itself could produce force. The protrusion mechanism is followed by cell adhesion and traction. The leading cells at the wound edge seem to act as a tractor for the cells located behind them. Cell-cell interactions and probably cell-matrix interaction alterations govern the ability of the cell population to migrate. These mechanisms are dependent on the extracellular matrix which forms the support for the cells [Roman and McDonald, 1991]. During the wound repair, cell surface adhesion molecules play a likely key role by modulating the cell contact with the extracellular matrix or with neighboring cells. Among these cell adhesion molecules, the integrins are known to 42 Zahm et al. be involved in the regulation of cell migration and proliferation [Albelda, 1991]. We have recently shown that specific cellular receptors for extracellular matrix are involved in cell migration during airway epithelium repair [Herard et al., 1996]. The a5b1 integrin receptor and its ligand, fibronectin, are upregulated in airway migrating cells and govern the wound repair process. A characteristic feature of the migratory process observed in the present in vitro wound model is the convergence of the migratory path of the cells located around the wound. The signals governing the uniform directionality of cell migration during wound repair are not well characterized. The guidance of motility could be related to chemotaxis, haptotaxis and contact guidance. Recently, Nishimura et al.  demonstrated that keratinocytes are able to migrate towards the negative pole in direct current electric fields that are of the same magnitude as measured in vivo near wounds in mammalian skin. This electrically directed movement of cells suggests that galvanotaxis may be one of the physiological mechanisms involved in wound healing. CONCLUSIONS Our results clearly show that in response to wounding, respiratory epithelial cells spread, migrate and proliferate, leading to the reestablishment of the barrier function of the epithelium. The tracking of cells at the wound edge associated with video microscopy techniques points out the coordinated and directional migration path of the cells. The cells located at the wound edge are characterized by the highest migration velocity and the lowest proliferative activity. The nature of the signals involved in propagating cell replication in the cells located behind the wound edge and how it might be related to cell migration is unknown. Propagation of information is probably an important mechanism to organize coordinated activity within a repairing area, and the study of this propagation will be the further step in the understanding of wound repair processes in respiratory epithelial cells. ACKNOWLEDGMENTS The authors are indebted to Ms. R. Smith for reviewing the paper and thank Dr. Hannion (Cinique Courlancy, Reims) and Dr. Klossek (Hôpital Jean Bernard, Poitiers) for providing the nasal polyps. This work was supported by grants from Association Française de Lutte contre la Mucoviscidose (AFLM), Pôle Technologique Régional GBM, INSERM (Contrat Normalisé d’Etude Pilote), and Ministère de l’Enseignement Supérieur et de la Recherche. REFERENCES Abercrombie, M., and Heaysman, J.E.M. (1953): Observations of the social behaviour of cells in tissue culture. I. Speed of movement of chick heart fibroblasts in relation to their mutual contacts. Exp. Cell. Res. 5:111–131. Abercrombie, M., and Heaysman, J.E.M. (1954): Observations of the social behaviour of cells in tissue culture. II. Monolayering of fibroblasts. Exp. Cell. Res. 6:293–306. Abercrombie, M., and Heaysman, J.E.M. (1970): Contact inhibition in tissue culture. In Vitro 6:128–142. Albelda, S.M. (1991): Endothelial and epithelial cell adhesion molecules. Am. J. Respir. Cell Mol. Biol. 4:195–203. Aggarwal, J.K., and Nandhakumar, N. (1988): On the computation of motion from sequences of images. A review. Proc. IEEE 76:917–935. Buisson, A.C., Gilles, C., Polette, M., Zahm, J.M., Birembaut, P., and Tournier, J.M. (1996a): Wound repair-induced expression of stromelysins is associated with the acquisition of a mesenchymal phenotype in human respiratory epithelial cells. Lab. Invest. 74:658–669. Buisson, A.C., Zahm, J.M., Polette, M., Pierrot, D., Bellon, G., Puchelle, E., Birembaut, P., and Tournier, J.M. (1996b): Gelatinase B is involved in the in vitro wound repair of human respiratory epithelium. J. Cell. Physiol. 166:413–426. Chevillard, M., Hinnrasky, J., Pierrot, D., Zahm, J.M., Klossek, J.M., and Puchelle, E. (1993): Differentiation of human surface upper airway epithelial cells in primary culture on a floating collagen gel. Epith. Cell Biol. 2:17–25. Fletcher, M.J., Warwick, K., and Mitchell, R.J. (1991): The application of a hybrid tracking algorithm to motion analysis. IEEE 5:84–89. Gordon, R.E., and Lane, B.P. (1976): Regeneration of rat tracheal epithelium after mechanical injury. II. Restoration of surface integrity during the early hours after injury. Am. Rev. Respir. Dis. 113:799–807. Herard, A.L., Pierrot, D., Hinnrasky, J., Kaplan, H., Sheppard, D., Puchelle, E., and Zahm, J.M. (1996): Fibronectin and its a5b1 integrin receptor are involved in the wound repair process of the airway epithelium. Am. J. Physiol. 271 (Lung Cell. Mol. Physiol. 15):L726–L733. Keenan, K.P., Combs, J.W., and McDowell, E.M. (1982): Regeneration of hamster tracheal epithelium after mechanical injury. I. Focal lesions: quantitative morphologic study of cell proliferation. Virchows Arch. 41:193–214. Lee, Y., McIntire, L.V., and Zygourakis, K. (1994): Analysis of endothelial cell locomotion: Differential effects of motility and contact inhibition. Biotechnol. Bioeng. 43:622–634. McDowell, E.M., Becci, P.J., Schurch, W., and Trump, B.F. (1979): The respiratory epithelium. VII. Epidermoid metaplasia of hamster tracheal epithelium during regeneration following mechanical injury. J. Natl. Cancer Inst. 62:995–1008. Mitchison, T.J., and Cramer, L.P. (1996): Actin-based cell motility and cell locomotion. Cell 84:371–379. Nikula, K.J., Wilson, D.W., Giri, S.N., Plopper, C.G., and Dungworth, D.L. (1988): The response of the rat tracheal epithelium to ozone exposure: Injury, adaptation and repair. Am. J. Pathol. 131:373–384. Nishimura, K.Y., Isseroff, R.R., and Nuccitelli, R. (1996): Human keratinocytes migrate to the negative pole in direct current electric fields comparable to those measured in mammalian wounds. J. Cell Sci. 109:199–207. Rangarajan, K., and Shah, M. (1991): Establishing motion correspondance. CVGIP: Image understanding 54:56–73. Cell Proliferation and Migration in Wound Repair Roman, J.R., and MacDonald, J.A. (1991): Cellular processes in lung repair. Chest 100:245–248. Sethi, I.K., and Jain, R. (1987): Finding trajectories of feature points in a monocular image sequence. IEEE Transactions 9:56–73. Shimizu, T., Nishihara, M., Kawaguchi, S., and Sakakura, Y. (1994): Expression of phenotypic markers during regeneration of rat tracheal epithelium following mechanical injury. Am. J. Respir. Cell Mol. Biol. 11:85–94. Tatsuka, M., Jinno, S., and Kakunaga, T. (1989): Quantitative measurement of cell motility associated with transformed phenotype. Jpn. J. Cancer Res. 80:408–412. Thomas, L.A., and Yamada, K.M. (1992): Contact stimulation of cell migration. J. Cell Sci. 103:1211–1214. Uras, S., Girasi, F., Verri, A., and Toore, V. (1988): A computational approach to motion perception. Biol. Cybern. 60:79–87. 43 Zahm, J.M., Chevillard, M., and Puchelle, E. (1991): Wound repair of human respiratory epithelium. Am. J. Respir. Cell Mol. Biol. 5:242–248. Zahm, J.M., Pierrot, D., Chevillard, M., and Puchelle, E. (1992): Dynamics of cell movement during the wound repair of human surface respiratory epithelium. Biorheology 29:459–465. Zahm, J.M., Pierrot, D., and Puchelle, E. (1993): Epidermal growth factor promotes wound repair of human respiratory epithelium. Wound Rep. Reg. 1:175–180. Zahm, J.M., Pierrot, D., and Puchelle, E. (1994): Photodamage induced by UV illumination on respiratory epithelial cells. In Jouffrey B., and Colliex C. (eds.): ‘‘Proceedings of the 13th International Congress on Electron Microscopy.’’ Les Ulis: Les Editions de Physique, Vol. 3A, pp. 89–90.