Quantitative Analysis of the Effects of Extracellular Matrix Proteins on Membrane Dynamics Associated with Corneal Epithelial Cell Motility

Abstract

Purpose.: Interaction of corneal epithelial cells with components of the basement membrane regulates cell morphology, adhesion, and migration. The authors examined the effects of basement membrane components on the behavior of human corneal epithelial (HCE) cells, including movement of the cell membrane in relation to the direction of cell migration.

Methods.: HCE cells were plated on fibronectin, laminin, collagen types I and IV, and bovine serum albumin (BSA). Cell adhesion was evaluated by staining with crystal violet. Cell movement was monitored by time-lapse microscopy, and an image processing program was developed for quantitative analysis of movement of the cell membrane in relation to the direction of cell migration. Actin cytoskeleton, focal adhesions, and the K⁺ channel Kv2.1 were detected by fluorescence microscopy.

Results.: Laminin, fibronectin, and collagen types I and IV each promoted cell adhesion compared with that apparent with BSA. Fibronectin and collagen types I and IV, but not laminin, increased cell motility. Laminin had no effect on membrane movement compared with that observed in the presence of BSA. Collagen types I and IV each increased membrane movement but in a manner independent of the direction of cell migration. Fibronectin induced membrane movement in the direction of cell migration. Fibronectin induced the formation of thick bundles of F-actin and focal adhesions at the cell periphery and the localization of Kv2.1 at the leading edge of the cell, whereas laminin and collagen types I and IV elicited the formation of thinner bundles of F-actin and smaller focal adhesions.

Conclusions.: Fibronectin may modulate the directional migration of corneal epithelial cells.

The corneal epithelium is a multilayered tissue that provides a barrier to the external environment and thereby supports homeostasis of the internal environment of the eye. The extracellular matrix (ECM) plays key roles in the development, growth, differentiation, and migration of the corneal epithelium. The basal layer of the corneal epithelium adheres to a basement membrane that separates the epithelium from the interstitial stroma, the ECM of which consists predominantly of collagen type I. The basement membrane is composed of several ECM proteins, including laminin, collagen type IV, collagen type VII, and proteoglycans. ^1–6 Healing of epithelial wounds in the cornea is mediated by the migration, proliferation, and remodeling of the remaining epithelial cells surrounding the wound. ^7–9 Expression of fibronectin is induced at the wound site, and the remaining epithelial cells upregulate integrin chains that bind fibronectin in order for them to cover the area of the defect. ^10–12 We have previously shown that migration of the corneal epithelium as a sheet in organ culture is promoted by the addition of fibronectin. ^13–15 Exogenous fibronectin also stimulates the healing of corneal epithelial wounds in the rabbit in vivo. In contrast to fibronectin, the expression of laminin disappears at the leading edge of epithelial migration during wound healing. ^16–18  

Cell adhesion and migration are dependent on the composition of the ECM, the expression of integrin receptors for ECM proteins, and the affinity of the interaction between these receptors and their ECM ligands. ^19–22 The interaction of cells with a permissive ECM substrate induces the clustering of integrins and the consequent assembly of focal adhesion complexes linked to the actin cytoskeleton. ^23,24 The outside-in signaling triggered by the substrate-integrin interaction promotes cell adhesion or migration. ^20,25,26 Changes in focal adhesion complexes during corneal epithelial adhesion or migration have thus been observed. ^27,28 Cells migrating on a substrate undergo four distinct processes, including cell extension, attachment, contraction, and detachment. Directional migration allows cells to move rapidly between locations, whereas random migration allows them to probe their local environment. ^29,30 Directional migration is associated with the formation of membrane protrusions and membrane ruffling at the leading edge of the cell. ^19,31,32 Moreover, several ion channels and water channels become localized at the leading edge of membrane ruffles in migrating cells and contribute to cell adhesion and migration. ^33,34  

The dynamics of corneal epithelial cell migration have been characterized by time-lapse video microscopy. ^35–37 The distribution of various ECM proteins in the normal or wounded cornea has also been studied by immunofluorescence analysis, ^18,38 and such proteins have been shown to affect cell adhesion structures and the behavior of corneal epithelial cells. ^18,39 With the use of time-lapse microscopy, we have now examined in detail the effects of various ECM proteins, including components of the basement membrane, on the adhesion and motility of corneal epithelial cells. Moreover, we performed quantitative analysis of the effects of these proteins on movement of the cell membrane relative to the direction of cell motility. 

Materials and Methods

Materials

A mixture of Dulbecco's modified Eagle's medium and nutrient mixture F-12 (DMEM/F-12) as well as fetal bovine serum, trypsin-EDTA, and gentamicin were obtained from Invitrogen-Gibco (Carlsbad, CA), and crystal violet was from Nacalai Tesque (Kyoto, Japan). Bovine serum albumin (BSA), bovine insulin, cholera toxin, and human recombinant epidermal growth factor were obtained from Sigma-Aldrich (St. Louis, MO). Glass-bottom culture dishes (35 mm) were from Iwaki (Tokyo, Japan). Fibronectin and laminin were obtained from Roche (Basel, Switzerland), and collagen type I and type IV were from Nitta Gelatin (Osaka, Japan). Mouse monoclonal antibodies to phosphotyrosine and rabbit polyclonal antibodies to the voltage-gated K⁺ channel Kv2.1 were obtained from Millipore (Billerica, MA), and Alexa Fluor 488–labeled goat antibodies to mouse or rabbit immunoglobulin G, rhodamine-phalloidin, and TOTO-3 were from Invitrogen (Carlsbad, CA). 

Cells and Cell Culture

Simian virus 40–immortalized human corneal epithelial (HCE) cells ⁴⁰ were obtained from RIKEN Biosource Center (Tsukuba, Japan). They were passaged in supplemented hormonal epithelial medium (SHEM), which comprises DMEM/F-12 supplemented with 15% heat-inactivated fetal bovine serum, bovine insulin (5 μg/mL), cholera toxin (0.1 μg/mL), human recombinant epidermal growth factor (10 ng/mL), and gentamicin (40 μg/mL). For experiments, HCE cells were cultured for 24 hours in unsupplemented DMEM/F-12, isolated by treatment with trypsin-EDTA, suspended in unsupplemented DMEM/F-12, and plated at a density of 2 × 10⁴ cells per well of 96-well plates or of 3 × 10⁴ cells per glass-bottom dish. The culture plates and dishes had been coated with fibronectin (10 μg/mL), collagen type I (10 μg/mL), collagen type IV (10 μg/mL), or laminin (10 μg/mL), each in the presence of 1% BSA or with 1% BSA alone (negative control). 

Cell Adhesion Assay

Cells plated in 96-well plates were incubated for 1 hour, washed twice with Ca²⁺- and Mg²⁺-free phosphate-buffered saline (PBS−), fixed for 15 minutes at 37°C with 3.7% formalin in PBS−, and stained with 1% crystal violet. They were then washed with water and allowed to dry in air, and the remaining attached cells were quantified by measurement of absorbance at 570 nm with the use of a microplate reader. 

Analysis of Cell Motility and Morphology

Cells plated in glass-bottom dishes were cultured for 24 hours and then incubated in a humidified chamber containing 5% CO₂ for 3 hours in DMEM/F-12 supplemented with 25 mM HEPES (pH 7.5). Five to 10 cells per field were monitored during the incubation period with the use of a fluorescence inverted microscope (Axioscope; Zeiss, Oberkochen, Germany). Phase-contrast images were obtained with a charge-coupled device (CCD) camera at 5-minute intervals. Nuclei positions were tracked to quantify cell motility with the use of Move-tr/2D software (Library, Tokyo, Japan). 

For analysis of the direction of cell movement and changes in cell morphology, images were collected with a CCD camera at 6-second intervals for 15 minutes. We developed an image processing procedure for automatic detection of the cell outline in phase-contrast images. Image processing included six steps: edge detection with the use of a Sobel filter, ⁴¹ prebinarization with the use of a histogram, ⁴² noise removal by border-following, ⁴³ dilation, ⁴⁴ extraction of the cell outline by-border following, ⁴⁵ and paint processing with the use of a seed-filling algorithm. ⁴⁶  

The boundary of a cell was thus first detected approximately with a Sobel filter, which allows the gradient of pixel brightness at a given point to be calculated (step 1). The image had 256 pixel gray values after application of the Sobel filter. A large amount of information is not needed to distinguish the area of the cell from the background, and it is advantageous to reduce the amount of information for detection of a cell outline. Cell area and background were classified as white and black, respectively. Prebinarization was applied to the images with a defined threshold value based on a histogram (step 2), yielding images with only two pixel colors (white and black). A substantial level of noise in the images hindered extraction of the cell outline. The noise level was therefore reduced by border-following, which removes noise smaller than a defined level (step 3). Given that parts of the cell outline were broken locally after prebinarization and border-following, we applied dilation, in which black pixels were set to white in the neighborhood of a white pixel (step 4). The cell outline was then extracted again by border-following (step 5), after which the interior of the cell was painted with the use of a seed-filling algorithm (step 6). Quantification of the difference between two processed images separated by a defined time interval allowed determination of the length of membrane movement at a given point, P, at the cell edge and with an angle Φ relative to the direction of migration and the centroid (O) of the cell (see Fig. 3). 

Immunofluorescence Microscopy

HCE cells plated in glass-bottom dishes were incubated for 60 minutes and then fixed for 15 minutes at 37°C with 3.7% formalin, washed with PBS−, and incubated for 1 hour at room temperature with 1% BSA in PBS−. The cells were then incubated for 1 hour with antibodies to phosphotyrosine or to Kv2.1 [1:200 dilution in PBS− containing 1% BSA, washed with PBS−, and incubated for 1 hour with Alexa Fluor 488–conjugated goat secondary antibodies (1:1000 dilution), rhodamine-phalloidin (1:200 dilution), or TOTO-3 (1:1000 dilution) in PBS− containing 1% BSA. They were then examined with a laser confocal microscope (LSM5; Zeiss). 

Statistical Analysis

Quantitative data are presented as mean ± SE. Differences were analyzed with Dunnett's test. P <0.05 was considered statistically significant. 

Results

We first examined the effects of laminin, fibronectin, and collagen type IV as basement membrane components, of collagen type I as a positive control, and of BSA as a negative control on the adhesion of cultured HCE cells. Culture plates were coated with each ECM protein in the presence of the same concentration of BSA as that used for the negative control. The cells attached to laminin, fibronectin, collagen type IV, and collagen I to a significantly greater extent than they did to BSA alone (Fig. 1). Cell attachment to collagen type IV or collagen type I was more pronounced than that to fibronectin or laminin. 

Figure 1.

 

Effects of ECM proteins on adhesion of HCE cells. Cells were incubated for 1 hour at 37°C in 96-well plates that had been coated with laminin, fibronectin, collagen type I, or collagen type IV, each in the presence of BSA, or with BSA alone. They were then washed and fixed, and the number of attached cells was evaluated by staining with crystal violet and measurement of absorbance at 570 nm. Data are mean ± SE from three independent experiments. *P < 0.05 for the indicated comparisons (Dunnett's test).

The effects of the various ECM proteins on the motility of HCE cells were examined by time-lapse microscopy. Analysis of the position of cell nuclei during incubation of the cells for 3 hours revealed that the distance moved by nuclei was significantly greater on dishes coated with fibronectin, collagen type IV, or collagen type I than on those coated with BSA alone (Fig. 2). Cell motility on laminin did not differ significantly from that on BSA alone. 

Figure 2.

 

Effects of ECM proteins on motility of HCE cells. Cells were incubated for 3 hours in glass-bottom dishes that had been coated with laminin, fibronectin, collagen type I, or collagen type IV, each in the presence of BSA, or with BSA alone, during which time they were monitored by time-lapse microscopy. Images were analyzed for determination of cell motility (distance traveled). Data are mean ± SE from three independent experiments. *P < 0.05 (Dunnett's test).

To examine the relation between changes in cell morphology and the direction of cell migration, we obtained time-lapse images during a 30-minute period of HCE cells cultured in glass-bottom dishes coated with the various ECM proteins. We also developed an image processing system for detection of the cell outline (Fig. 3). Analysis of cells cultured on BSA alone or on laminin revealed only small changes in the cell outline during cell motility (Figs. 4A, 5A). The relative value for the morphologic velocity of the membrane of cells on BSA alone or on laminin compared with the position at time 0 ranged from −0.04 to 0.04 μm/s (Figs. 4B, 5B), whereas the absolute values were <0.012 μm/s and did not correspond to the direction of cell migration (Figs. 4C, 5C). In contrast, fibronectin induced an apparent marked change in cell outline during cell motility (Fig. 6A). The relative value for the morphologic velocity of the cell membrane on fibronectin ranged from −0.010 to 0.08 μm/s (Fig. 6B) and was maximal in the direction of cell migration. The corresponding absolute values were 0.023 ± 0.003 μm/s and were again maximal in the direction of cell movement (Fig. 6C). The outline of cells cultured on collagen type IV was also altered during cell motility (Fig. 7A). The relative value for the morphologic velocity of the cell membrane on collagen type IV ranged from −0.06 to 0.04 μm/s, and the corresponding absolute values were 0.019 ± 0.003 μm/s (Figs. 7B, 7C). The membrane morphologic velocity, However, did not coincide with the direction of cell movement for cells cultured on collagen type IV. Finally, collagen type I also induced membrane movement in HCE cells (Fig. 8A). The relative value for the morphologic velocity of the cell membrane ranged from −0.08 to 0.012 μm/s, with the corresponding absolute values 0.025 ± 0.001 μm/s (Figs. 8B, 8C). The membrane morphologic velocity for cells cultured on collagen type I was also independent of the direction of cell migration. The mean absolute value of membrane morphologic velocity for cells on fibronectin did not differ significantly from that for cells on collagen type IV or type I. We also calculated the ratio of the mean directional value of membrane morphologic velocity to the mean vertical value. The ratio for fibronectin (0.46 ± 0.12) was significantly (P < 0.05) smaller than that for collagen type IV (0.86 ± 0.08) or collagen type I (1.1 ± 0.06). These results thus showed that fibronectin promotes membrane protrusion at the leading edge of migrating HCE cells, whereas collagen types I and IV promote random membrane protrusion independent of the direction of cell movement. 

Figure 3.

 

Image processing system. Images of HCE cells cultured in glass-bottom dishes coated with ECM proteins were acquired by time-lapse microscopy and processed for analysis of changes in membrane morphology in relation to the direction of migration. O, centroid of the cell; P, point at the cell periphery; Φ, angle of the line connecting O and P relative to the direction of migration (arrow).

Figure 4.

 

Effect of BSA on the membrane morphology and direction of motility of HCE cells. Cells plated on a glass-bottom dish coated with BSA alone were incubated for 3 hours and subjected to time-lapse microscopy. Images were collected at 6-second intervals during the final 15 minutes of incubation and were processed as shown in Figure 3. (A) Representative images obtained at 0, 5, 10, and 15 minutes. Quantitation of (B) relative and (C) absolute values of membrane morphologic velocity averaged for a total of five cells. Red circle: position of the membrane at time 0. Arrow: direction of cell movement. Individual cells were analyzed in three independent experiments. Relative morphologic velocity and absolute morphologic velocity are shown in micrometers per second.

Figure 5.

 

Effect of laminin on the membrane morphology and direction of motility of HCE cells. Cells plated on dishes coated with laminin plus BSA were analyzed as in Figure 4. Relative morphologic velocity and absolute morphologic velocity are shown in micrometers per second.

Figure 6.

 

Effect of fibronectin on the membrane morphology and direction of motility of HCE cells. Cells plated on dishes coated with fibronectin plus BSA were analyzed as in Figure 4. Relative morphologic velocity and absolute morphologic velocity are shown in micrometers per second.

Figure 7.

 

Effect of collagen type IV on the membrane morphology and direction of motility of HCE cells. Cells plated on dishes coated with collagen type IV plus BSA were analyzed as in Figure 4. Relative morphologic velocity and absolute morphologic velocity are shown in micrometers per second.

Figure 8.

 

Effect of collagen type I on membrane morphology and direction of motility of HCE cells. Cells plated on dishes coated with collagen type I plus BSA were analyzed as in Figure 4. Relative morphologic velocity and absolute morphologic velocity are shown in micrometers per second.

We next investigated the effects of ECM proteins on the actin cytoskeleton and focal adhesions by fluorescence microscopy. Cells plated on fibronectin appeared more flattened and spread out than did those plated on BSA alone, laminin, collagen type I, or collagen type IV. Cells cultured on fibronectin also manifested a thick rim of F-actin staining at the cell periphery (Fig. 9). Costaining with antibodies to phosphotyrosine revealed numerous large dotlike structures, presumably corresponding to focal adhesions, associated with the bundles of F-actin at the cell periphery. Cells plated on laminin or collagen type I or IV exhibited a thin rim of F-actin at the cell periphery and numerous small dotlike structures likely corresponding to focal adhesions. In contrast, cells plated on BSA exhibited only a thin rim of F-actin staining (thinner than that for cells plated on laminin or collagen type I or IV) and only a few small dotlike structures positive for phosphotyrosine at the cell periphery. 

Figure 9.

 

Effects of ECM proteins on the actin cytoskeleton and focal adhesions in HCE cells. Cells plated on glass-bottom dishes coated with laminin (LN), fibronectin (FN), Col type I, or collagen type IV, each in the presence of BSA, or with BSA alone were subjected to fluorescence microscopy with antibodies to phosphotyrosine for detection of focal adhesions (green), with rhodamine-phalloidin for detection of F-actin (red) and with TOTO-3 for detection of nuclei (blue). Data are representative of three independent experiments. Scale bar, 10 μm.

Finally, we examined the effects of ECM proteins on the distribution of the voltage-gated K⁺ channel Kv2.1 by immunofluorescence microscopy. Cells cultured on fibronectin manifested preferential localization of Kv2.1 at the leading edge of the cell periphery (Fig. 10). In contrast, cells cultured on laminin or collagen type I or IV did not exhibit such a preferential localization of Kv2.1. 

Figure 10.

 

Effects of ECM proteins on the distribution of Kv2.1 in HCE cells. Cells plated on glass-bottom dishes coated with laminin, fibronectin, collagen type I, or collagen type IV, each in the presence of BSA, or with BSA alone were subjected to immunofluorescence microscopy with antibodies to Kv2.1. Data are representative of three independent experiments. Scale bar, 10 μm.

Discussion

We have shown that laminin, fibronectin, collagen type IV, and collagen type I promote corneal epithelial cell adhesion. Fibronectin, collagen type IV, and collagen type I also increased the motility of corneal epithelial cells, whereas laminin did not have an effect on cell motility greater than that of BSA alone. Moreover, we have developed an image processing procedure for quantitative analysis of changes in the morphology of the cell membrane relative to the direction of cell movement as observed by time-lapse microscopy. This procedure revealed that laminin did not affect membrane morphology to an extent greater than did BSA alone. Collagen types IV and I each induced changes in membrane morphology, but the induced membrane protrusions did not coincide with the direction of cell movement. In contrast, fibronectin induced membrane protrusion in the direction of cell movement. Moreover, fibronectin induced the accumulation of a thick rim of F-actin and the formation of large focal adhesions at the periphery of corneal epithelial cells. Collagen type IV, collagen type I, and laminin induced the formation of a thin rim of F-actin at the cell periphery and the formation of small focal adhesions. These observations thus indicate that ECM proteins differentially affect the adhesion, motility, and morphology of epithelial cells. Various ECM proteins are expressed in spatially and temporally distinct patterns during corneal epithelial wound healing, ^18,38,47 with the remaining corneal epithelial cells moving to cover the wound area in both slow and rapid phases. ^36,48 Corneal epithelial cells may thus respond to different substrates with different migration behaviors during wound healing. Cell migration is also affected and modulated by ECM concentration. ⁴⁹ In the present study, we analyzed the behavior of HCE cells on various ECM proteins each at a concentration of 10 μg/mL. The effects of different concentrations of laminin, fibronectin, and collagen types IV and I on HCE cell behavior remain to be determined. 

Cells migrate randomly in response to their local environment, and the speed of such random migration is relatively high. ^30,50,51 However, directional cell migration is required under certain conditions, such as during development, inflammation, and tissue repair. Cell polarity contributes to effective cell translocation, with transmission of driving forces at the leading lamella achieved by substrate adhesion. ^52,53 We have now shown that fibronectin induced membrane protrusion at the leading edge and retraction at the rear of HCE cells during migration. These observations suggest that corneal epithelial cells are polarized during migration on fibronectin, with strengthening of adhesions at the front of the cell and dissolution of those at the rear. Under normal conditions, fibronectin is not expressed at a substantial level in the corneal stroma. ^47,54 However, fibronectin is transiently expressed at wound sites in the corneal stroma, and exogenous fibronectin promotes corneal epithelial adhesion and migration and consequent wound healing in vivo. ^47,54 Fibronectin has previously been shown to play a role in the polarization of HCE cells by inducing the formation of membrane ruffles at the leading edge during cell movement. ²⁷ We found that fibronectin also promotes the formation of focal adhesions during HCE cell migration. Ion channels and water channels are also activated during cell migration and localize to the leading edge of membrane ruffles. ^33,34,55 The voltage-gated K⁺ channel Kv2.1 interacts with focal adhesion kinase at focal adhesions during cell migration. ⁵⁶ We have now shown that Kv2.1 was preferentially localized at the leading edge of the cell periphery in HCE cells cultured on fibronectin. However, immunofluorescence analysis revealed that the colocalization of Kv2.1 with focal adhesion kinase was detected in the cytoplasm of HCE cells but not at the focal adhesion (data not shown). This discrepancy among these results may reflect a difference of cell type or species. In contrast to the effect of fibronectin, collagen types I and IV each induced the formation of membrane protrusions in HCE cells in the present study, but these protrusions were not aligned with the direction of cell movement. Together, our results suggest that physical cues from fibronectin may control directional cell motility and that fibronectin may play a key role in promoting the directional migration of corneal epithelial cells during wound healing. 

The interaction between cells and the ECM plays a central role in the regulation of cell behavior. The binding of integrins to their ECM protein ligands elicits intracellular signaling leading to the tyrosine phosphorylation of focal adhesion components, the formation of focal adhesion complexes, and the rearrangement of the actin cytoskeleton. All the ECM proteins examined in the present study promoted HCE cell adhesion. However, laminin did not increase cell motility or induce the formation of obvious membrane protrusions in spite of its promotion of cell adhesion. Laminin is a component of the basement membrane of the corneal epithelium under normal conditions. ^17,57 Damage to the corneal epithelium, however, results in the disappearance of laminin from the basement membrane. ¹² Laminin expression in the basement membrane is restored after the epithelial defect has healed. Laminin also plays an important role in cell differentiation in vitro. ⁵⁸ These observations suggest that laminin does not regulate cell migration in vivo but rather may contribute to the maturation and maintenance of corneal epithelial cells. 

Image analysis programs for time-lapse microscopy have been used to analyze the behavior of cells during migration. ^27,59–61 They measure several parameters, including the distance of cell movement, cell speed, and cell surface morphology. We have now developed a new image processing program for the measurement of cell motility and the cell outline. We have quantitatively analyzed time-lapse images with this new program for changes in the morphology of the cell membrane relative to the direction of cell movement in HCE cells plated on various ECM proteins. Our program should prove useful for analyzing the effects of the cell microenvironment, including ECM, growth factors, and cytokines, on the dynamics of cell motility in vitro. Further characterization of such effects may provide a better understanding of corneal epithelial wound healing and a basis for the development of new treatments for corneal epithelial wounds. 

Footnotes

 Supported in part by Grant 21791687 from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Footnotes

 Disclosure: K. Kimura, None; S. Kawano, None; T. Mori, None; J. Inoue, None; H. Hadachi, None; T. Saito, None; T. Nishida, None

The authors thank Yasumiko Akamatsu for technical assistance.