Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The airway epithelium is a target of inflammatory and physical stimuli in obstructive diseases such as asthma and bronchopulmonary dysplasia. The epithelium both provides a physical barrier to the external environment and regulates several key metabolic functions of airways, including fluid and ion transport to the airway lumen, mucociliary clearance, and airway diameter. Repair of a damaged epithelium may be a necessary part of restoring airway function to its normal state. Repair generally involves several steps, including migration and spreading of epithelial cells at the margin of the injury into the damaged region and proliferation of new epithelial cells (1). Each step can be modulated actively by growth factors secreted by constitutive cells within the airway, such as fibroblasts (4), or depressed by mediators secreted by inflammatory cells that have migrated into the airways, such as eosinophils (5, 6). Each of these processes may be mediated by integrins found on epithelial cells (7, 8). Integrins mediate cell adhesion of many cell types, including epithelial cells, to extracellular matrix (ECM) (9, 10). Signal transduction from integrins after binding of matrix protein proceeds via several intermediate pathways (11, 12) to regulate adhesion (13, 14), migration (15), proliferation (16, 17), and apoptosis (18, 19) of epithelial cells.

Dissolution of integrin-mediated adhesions must occur when epithelial cells detach from the underlying basement membrane to migrate to a new site. The matrix composition of the basement membrane may alter this process. Epithelial cells are known to express integrins for collagen, laminin-1, and fibronectin found in basement membranes (20). One recent report demonstrates that one variant of laminin, laminin-2, is expressed in the basement membrane of airways in subjects with chronic asthma but not in nonasthmatic subjects (23). It has been suggested that integrin receptors found on airway epithelial cells, such as ₂₁ and ₆₁, bind laminin-2 less well than laminin-1 (24, 25). It is possible, then, that expression of laminin-2 or other matrix proteins in asthmatic airways decreases epithelial cell migration and repair over the basement membrane.

The purpose of this study was to demonstrate the extent to which interference with integrins would block airway epithelial repair after injury. We also determined the ability of epithelial cells to repair after injury over different ECM proteins. Human airway epithelial cell lines grown in monolayer culture were studied after mechanical injury using time-lapse video microscopy. Monolayer repair was observed in cells grown on one of three matrix proteins: collagen-IV, laminin-1, and laminin-2. Our data demonstrate that human airway epithelial cells are able to use each matrix protein to repair injury in monolayer culture. The ₁-integrin modulates cell repair and detachment in cells grown on each of these matrix proteins, whereas -integrin subunits regulate wound repair only in cells grown on collagen-IV.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials

Insulin, penicillin, streptomycin, hydrocortisone, transferrin, bovine serum albumin, triiodothyronine, Ham's F12 medium, trypsin, epidermal growth factor (EGF), ethylenediaminetetraacetic acid (EDTA), interleukin-1, and type 25 protease from Bacillus polymyxia were obtained from Sigma (St. Louis, MO). Fetal calf serum (FCS) was obtained from Hyclone (Logan, UT) and was heat-denatured prior to use. Monoclonal antibodies (mAbs) against ₁-integrin (CD29) (clone Lia 1/2), ₂-integrin (CD49b) (clone Gi9), ₃-integrin (CD49c) (clone M-Kid2), ₅-integrin (CD49e) (clone SAM1), and ₆-integrin (CD49f) (clone GoH3) were obtained from Immunotech, Inc. (Westbrook, ME). Antimurine immunoglobulin (Ig) G₁ and IgG₂ isotype control mAbs also were obtained from Immunotech. Fluorescein isothiocyanate (FITC)-conjugated goat antimouse immunoglobulin was obtained from Becton-Dickinson (San Jose, CA). Human placental collagen-IV and mouse laminin-1 were obtained from Sigma. Human placental laminin-2 (merosin) was obtained from GIBCO BRL (Grand Island, NY).

Culture of Human Airway Epithelial Cell Lines

The 16HBE14o cells, a generous gift of Dr. Dieter Gruenert, University of California San Francisco, were SV₄₀-transformed human central airway epithelial cells characterized previously (26, 27). Cells were grown on matrix-coated flasks or plates in Eagle's modified essential medium containing 10% FCS, 2 mM L-glutamine, 100 µg/ml streptomycin, and 100 U/ml penicillin G (medium B), and incubated at 37°C in 5% CO₂. Cells were passaged into matrix-coated six-well plates for wound-repair experiments and into either two-well or four-well matrix-coated slides for other experiments. Plates were coated with either 0.5 µg/ml collagen-IV, 10 µg/ml laminin-1, or 10 µg/ml laminin-2. In each case, plates were treated overnight to dry and washed with serum-free medium before use.

Flow Cytometry

Cells grown in T24 flasks on specified matrix protein were collected after incubation with 0.5% trypsin and 0.2% EDTA for 15 min with a rubber spatula, and then washed twice in medium C (medium B minus FCS). Flow cytometric analysis was performed by standard techniques using a murine IgG₁ or IgG₂ isotype mAb as a control, or one of the mAbs for ₁-, ₂-, ₃-, ₅-, or ₆-integrin subunits. For each condition, ~ 500,000 cells were blocked with 5% human serum albumin in fluorescence-activated cell sorter (FACS) buffer for 10 min, and then stained with primary mAb for 60 min, followed by washing and incubation with an FITC-conjugated goat antimouse IgG (Becton-Dickinson) for 30 min. Cells then were washed, fixed in 2% paraformaldehyde, and stored at 4°C until analysis. Cells were analyzed on a Becton-Dickinson FACScan cytometer.

In additional experiments, the expression of integrins before and 24 h after treatment with EGF was assessed. Monolayers grown to confluence on either 0.5 µg/ml collagen-IV, 10 µg/ml laminin-1, or 10 µg/ml laminin-2 (n = 1 each) were placed in medium C overnight. Cells from some monolayers were collected, fixed, and stained for the presence of ₁-, ₂-, ₃-, and ₆-integrin subunits using flow cytometry. Additional monolayers were treated with either 15 ng/ml EGF or sham in medium C for 24 h, after which cells were collected, stained for the presence of integrin subunits, and fixed in the same manner as in the preceding paragraph.

Monolayer Wound Repair Assay

This assay was used to demonstrate the effect of integrin monoclonal antibodies on epithelial cell migration and wound repair. We have previously published details of this method (28, 29). Briefly, 16HBE14o cells were grown until confluent in medium B in six-well plates (~ 6 to 8 × 10⁵ cells/well) and then placed into medium C overnight. The next day, cells were washed and placed in 3 ml serum-free medium C. A small wound was made in the confluent monolayer with a rubber stylet. In preliminary experiments, use of the stylet to remove cells without disturbing the underlying protein matrix was verified by demonstrating an intact matrix immediately after cell removal with the use of an appropriate antimatrix antibody and fluorescence microscopy. In preliminary experiments, staining of freshly made wounds with either an anti-pan-collagen or anti-pan-laminin mAb (Sigma) labeled with FITC demonstrated an intact or near-intact matrix in the wound site. All wounds were viewed immediately after creation by phase-contrast microscopy to look for signs of matrix removal within the wound; wounds with evidence of matrix removal were discarded. Wound closure was measured serially for 24 h starting immediately after wound creation. Microscope images were photographed using a Sony Iris CCD camera (Sony, Inc., Rolling Meadows, IL) on a Nikon Diaphot inverted-stage microscope (Nikon Inc., Morton Grove, IL). Video images were digitized using a Power Macintosh 6100 AV computer (Apple Computer, Inc., Cupertino, CA) and Apple Video Player software (Apple Computer). Analysis of perimeter length and area of the remaining wound in each image was performed using NIH Image software (Wayne Rasband, National Institutes of Health, Bethesda, MD, available on the Internet at http://rsb.info.nih.gov/nih-image).

In each experiment, one well was used as a negative control (no EGF or mAb) and one well was treated with 15 ng/ml EGF, previously demonstrated to be a potent accelerant for epithelial monolayer wound closure (29). Other monolayers were treated immediately after wound creation with 15 ng/ml EGF plus one of the mAbs for ₁-, ₂-, ₃-, or ₆-integrin (0.1 to 10 µg/ml, one concentration per well). Repair was followed for 24 h by time-lapse video microscopy. Each set of experiments was performed on each of the three ECM proteins used: collagen-IV, laminin-1, or laminin-2.

In separate control experiments, monolayers grown on each of these matrix proteins were treated immediately after wound creation with 3 µg/ml of either ₁-, ₂-, ₃-, or ₆-integrin mAb, without EGF, and were followed for 24 h by time-lapse video microscopy (n = 4 wells for each mAb and each matrix protein). A positive control, 15 ng/ml EGF alone, and a negative control, with neither EGF nor mAb, were done at the same time to ensure valid experiments.

Epithelial Cell Apoptosis In Situ after Wound Creation

These experiments tested whether delay in wound closure was due to cell apoptosis at the leading edge of the wound. Epithelial cells on collagen-IV, laminin-1, or laminin-2 matrix were grown to confluence in four-chamber glass slides (n = 3 for each matrix). Cells were incubated in serum-free medium C overnight prior to starting experiments. An approximately 1.0-mm² wound was created in each chamber, after which chambers were treated with one of the following conditions: 10 µg/ml anti-₁-integrin mAb alone, 10 µg/ml anti-₁-integrin mAb and 15 ng/ml EGF, 10 µg/ml of IgG isotype control mAb, or 10 µg/ml IgG isotype control mAb with 15 ng/ml EGF. Slides were incubated for 24 h, fixed for 48 h with 4% paraformaldehyde at room temperature, and then air-dried. Apoptotic cells were demonstrated by labeling free 3'-hydroxyl groups of DNA using a terminal deoxynucleotidyl transferase- mediated deoxyuridine triphosphate biotin nick end-labeling (TUNEL) TACS II fluorescent assay kit (Trevigen, Inc., Gaithersburg, MD). Kit instructions were followed. Slides were counterstained with 1 mM 4'6-diamidino-2-phenyindole (DAPI) for 5 min and then washed five times in distilled, deionized water. Slides were stored in the dark until they were viewed under fluorescent microscopy. Positive cells around the wound edge and in the interior of the chamber away from the wound were counted immediately.

Epithelial Cell Proliferation In Situ after Wound Creation

These experiments tested whether delay in wound closure was due to cell proliferation at the leading edge of the wound. Epithelial cells on collagen-IV, laminin-1, or laminin-2 matrix were grown to confluence in four-chamber glass slides (n = 3 for each matrix). Cells were incubated in serum-free medium C overnight prior to starting experiments. An approximately 1.0-mm² wound was created in each chamber, after which slides were treated with 15 ng/ ml EGF, or sham, and incubated for 18 h. Phase-contrast images were collected at 0 and 24 h after wound creation. Cells were fixed for 1 h in 4% paraformaldehyde at room temperature, and then air-dried. Cells were then treated with 1 mM DAPI and 1.0 µg/ml rat antihuman cyclin-B₁ (PharMingen Inc., San Diego, CA), a marker for progression to G2 and S phase in the cell cycle (30, 31), for 10 min at 4°C. A secondary FITC-tagged goat antirat IgG (1:128) was then added for 150 min, after which slides were rinsed. Slides were stored in the dark until they were viewed under fluorescent microscopy. The number of proliferating (cyclin-B-positive) nuclei at the wound edge was counted as a percentage of total nuclei (DAPI-positive).

Data Analysis

Cell apoptosis (TUNEL-positive) is expressed as percent of cells counted at the wound edge. Wound closure is expressed either as area (µm²) or as percent of control area. In previous video microscopy experiments using cell monolayers, intraobserver variability was < 2% and interobserver variability was < 4% for all measurements (28, 29). The time required for 50% closure of a wound was calculated by linear interpolation between the two closest data points and expressed in minutes. When this exceeded 24 h, the maximum time (24 h) was used.

Comparisons between multiple groups were made by analysis of variance; when significant differences were found, further comparisons were made by Fisher's least significant difference test. Comparisons between two groups were made by paired Student's t test. Bonferroni's correction was made as appropriate for multiple comparisons. Differences were considered significant when P < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

The 16HBE14o cell line attached and grew well on each of the ECM proteins used in this study. No differences were observed in initial attachment, time to confluence, or morphology of the cells, as judged by light microscopy based on matrix protein composition.

Integrin Subunit Expression on 16HBE14o Cells by Flow Cytometry

The ₁-, ₂-, ₃-, and ₅-integrin subunits were expressed on the cell surface of 16HBE14o cells as demonstrated by flow cytometry (Figure 1). Expression of the ₅-integrin subunit was above that for control IgG but substantially less so than for ₁-, ₂-, and ₃-integrin subunits. Expression of the ₆-integrin subunit was demonstrated at a lower mean fluorescence, which was not different from that for the IgG control. Expression of each integrin subunit was equivalent on 16HBE14o cells grown on collagen-IV, laminin-1, and laminin-2 (data not shown). Subsequent wound repair experiments focused on the interactions of 16HBE14o cell monolayers with each matrix protein via ₂₁, ₃₁, or (as a control) ₆₁ integrin receptors during wound repair.

View larger version (24K): [in this window] [in a new window]   Figure 1.   Demonstration of 1-, 2-, 3-, and 6-integrin subunit expression on the cell surface of 16HBE14o cells by flow cytometry. Expression of 6-integrin is less in 16HBE14o cells compared with the other subunits. Expression of other integrin subunits is equivalent. Representative flow cytometry histograms taken from three experiments.

There was no difference in integrin subunit expression as assessed by flow cytometry after treatment with 15 ng/ ml EGF for 24 h compared with untreated cells (data not shown).

Repair of 16HBE14o Monolayers after Injury

The 16HBE14o monolayers were grown on either collagen-IV, laminin-1, or laminin-2 matrix, and monolayer wounds were created without difficulty in each monolayer in each experiment. Initial wound area was significantly larger for monolayers grown on collagen-IV compared with either laminin matrix (Table 1). Control wounds closed modestly over 24 h, whereas wounds in monolayers treated with 15 ng/ml EGF closed substantially in the same time period (Figure 2). The 50% closure time was similar for monolayers grown on laminin-1, laminin-2, and collagen-IV despite the difference in initial area (Table 2). Treatment with an isotype IgG₁ control mAb had no effect on wound closure in separate EGF-treated monolayers (n = 24 monolayers, data not shown).

View larger version (21K): [in this window] [in a new window]   Figure 2.   Wound closure in monolayers grown on collagen-IV, laminin-1, or laminin-2. (A) Closure of untreated (control) wounds. (B) Closure of wounds in monolayers treated with 15 ng/ml EGF. There were no significant differences in wound closure in either condition based on underlying matrix protein.

For each matrix protein, treatment with the anti-₁- integrin mAb slowed wound closure. Inhibition of wound closure was both time- and concentration-dependent but was not complete, in that some wound closure still occurred even in the presence of 10 µg/ml of antibody (Figure 3). After 24 h, the remaining wound area in monolayers grown on collagen-IV matrix and treated with EGF + 10 µg/ml Lia 1/2 was 48 ± 16% of time 0, versus 13 ± 7% for EGF alone (P = 0.05). The remaining wound area in monolayers grown on laminin-1 matrix and treated with EGF + 10 µg/ml Lia 1/2 was 84 ± 19% of time 0 (n = 5), versus 13 ± 6% for EGF alone (n = 5, P < 0.002). The remaining wound area in monolayers grown on laminin-2 matrix and treated with EGF + 10 µg/ml Lia 1/2 was 50 ± 15% of time 0 (n = 5), versus 16 ± 9% for EGF alone (n = 5, P = 0.05). The 50% closure time was greater in wounds treated with both EGF and the anti-₁-integrin mAb compared with EGF alone for both collagen-IV and laminin-1, but not for laminin-2 (Table 3).

View larger version (25K): [in this window] [in a new window]   Figure 3.   Wound closure in monolayers grown on collagen-IV, laminin-1, or laminin-2: effect of the 1-integrin mAb Lia 1/2. (A) Collagen-IV experiments; (B) laminin-1 experiments; (C ) laminin-2 experiments. In each set of experiments, negative control wells were untreated and positive control wells were treated with 15 ng/ml EGF alone. Wound closure was blocked substantially, but not completely, by treatment with the anti- 1-integrin mAb in each set of experiments. n = 5 to 7 experiments on collagen-IV, 4 to 6 experiments on laminin-1, and 5 experiments on laminin-2. *P = 0.05, and §P < 0.002 versus EGF alone.

Treatment with the anti-₃-integrin mAb slowed wound closure on 16HBE14o cells grown on either collagen-IV or laminin-1. Inhibition of wound closure was both time- and concentration-dependent but was not complete even with addition of high concentrations of antibody (Figure 4). After 24 h, the remaining wound area in monolayers grown on collagen-IV matrix and treated with EGF + 10 µg/ml M-Kid2 was 49 ± 9% of time 0 (n = 5), versus 3 ± 3% for EGF alone (n = 5, P < 0.03). The remaining wound area in monolayers grown on laminin-1 matrix and treated with EGF + 10 µg/ml M-Kid2 was 58 ± 4% of time 0 (n = 5), versus 8 ± 6% for EGF alone (n = 5, P < 0.0001). The 50% closure time was greater for wounds treated with both EGF and the anti-₃-integrin mAb versus EGF alone when grown on either collagen-IV or laminin-1 (Table 4). However, treatment with the anti- ₃-integrin mAb slowed wound closure only modestly in monolayers grown on laminin-2 (Figure 4). The remaining wound area in monolayers grown on laminin-2 matrix and treated with EGF + 10 µg/ml M-Kid2 was 15 ± 5% of time 0 (n = 5), versus 0 ± 0% for EGF alone (n = 5, P = 0.05). The 50% closure time was not different in monolayers grown on laminin-2 and treated with both anti-₃-integrin mAb and EGF compared with EGF alone (Table 3).

View larger version (25K): [in this window] [in a new window]   Figure 4.   Wound closure in monolayers grown on collagen-IV, laminin-1, or laminin-2: effect of the 3-integrin mAb M-Kid2. (A) Collagen-IV experiments; (B) laminin-1 experiments; (C ) laminin-2 experiments. In each set of experiments, negative control wells were untreated and positive control wells were treated with 15 ng/ml EGF alone. Wound closure was blocked substantially, but not completely, by treatment with the anti-3-integrin mAb in each set of experiments. n = 5 experiments on each matrix. *P = 0.05, §P = 0.03, and P < 0.0001 versus EGF alone.

Treatment with the anti-₂-integrin mAb moderately slowed wound closure of 16HBE14o cells grown on collagen-IV (Figure 5). After 24 h, the remaining wound area in monolayers grown on collagen-IV matrix and treated with EGF + 10 µg/ml Gi9 was 25 ± 9% of time 0 (n = 5), versus 11 ± 8% for EGF alone (n = 5, P < 0.03). However, when cells were grown on either laminin-1 or laminin-2, treatment with this mAb did not slow wound closure (Figure 5). The remaining wound area after 24 h in monolayers grown on laminin-1 matrix and treated with EGF + 10 µg/ml Gi9 was 14 ± 9% of time 0 (n = 4), versus 5 ± 3% for EGF alone (n = 5, P = NS). Wounds created in monolayers grown on laminin-2 matrix and treated with EGF closed completely after 24 h, regardless of mAb concentration (Figure 5). The 50% closure time reflected these differences (Table 3).

View larger version (26K): [in this window] [in a new window]   Figure 5.   Wound closure in monolayers grown on collagen-IV, laminin-1, or laminin-2: effect of the 2-integrin mAb Gi9. (A) Collagen-IV experiments; (B) laminin-1 experiments; (C ) laminin-2 experiments. In each set of experiments, negative control wells were untreated and positive control wells were treated with 15 ng/ml EGF alone. Wound closure was blocked partially by treatment with the anti-2-integrin mAb in cells grown on collagen-IV, but not in cells grown on either laminin-1 or laminin-2. n = 5 to 8 experiments on collagen-IV, 4 or 5 experiments on laminin-1, and 5 experiments on laminin-2. *P = 0.03 versus EGF alone.

Treatment with the anti-₆-integrin mAb did not slow wound closure of 16HBE14o cells grown on either collagen-IV, laminin-1, or laminin-2 matrix (Figure 6). The 50% closure time was also similar in each group regardless of mAb treatment (Table 3).

View larger version (26K): [in this window] [in a new window]   Figure 6.   Wound closure in monolayers grown on collagen-IV, laminin-1, or laminin-2: effect of the 6-integrin mAb GoH3. (A) Collagen-IV experiments; (B) laminin-1 experiments; (C ) laminin-2 experiments. In each set of experiments, negative control wells were untreated and positive control wells were treated with 15 ng/ml EGF alone. Wound closure was blocked partially by treatment with the anti-6-integrin mAb in cells grown on collagen-IV, but not in cells grown on either laminin-1 or laminin-2. n = 3 experiments on collagen-IV, 7 experiments on laminin-1, and 6 experiments on laminin-2.

Treatment of monolayers with an integrin mAb immediately after wound creation, in the absence of EGF, did not change wound closure appreciably compared with control monolayers, regardless of underlying protein matrix (Figure 7). The 50% closure time was similar in each group regardless of mAb treatment (Table 4). In each experiment, the monolayers treated with EGF demonstrated appropriate acceleration in wound closure.

View larger version (25K): [in this window] [in a new window]   Figure 7.   Wound closure in monolayers grown on collagen-IV, laminin-1, or laminin-2: response to integrin antibodies in the absence of EGF. (A) Collagen-IV experiments; (B) laminin-1 experiments; (C ) laminin-2 experiments. In each set of experiments, negative control wells were untreated and positive control wells were treated with 15 ng/ml EGF alone. Each of the integrin antibodies was added at a final concentration of 10 µg/ ml. Wound closure in the presence of each antibody was not different from that in control monolayers. n = 4 experiments on each matrix.

Apoptosis of 16HBE14o Cells at the Wound Edge

Examination of wound edges 24 h after creation in monolayers treated with the anti-₁-integrin mAb demonstrated few TUNEL-positive cells at the wound edge (Figure 8). Similar results were found in monolayers grown on either collagen-IV, laminin-1, or laminin-2. The ratio of TUNEL-positive nuclei to all nuclei at the wound edge was < 1% in all monolayers.

View larger version (65K): [in this window] [in a new window]   Figure 8.   Apoptosis in an airway epithelial monolayer wound after 24 h: representative photomicrographs. (A) Fluorescent TUNEL staining for DNA nicking in cell nuclei; (B) DAPI counterstain of the same monolayer. Few apoptotic cells are seen at the edge of the monolayer wound. Original magnification: ×100 in both images.

Proliferation of 16HBE14o Cells at the Wound Edge

Examination of wound edges 18 h after creation in monolayers treated with EGF for cyclin-B₁ labeling demonstrated that the ratio of proliferating (cyclin-B₁-positive) cells to all cells at the wound edge was < 3% (n = 3) in all monolayers (Figure 9).

View larger version (69K): [in this window] [in a new window]   Figure 9.   Proliferation at an airway epithelial monolayer wound edge after 18 h: representative photomicrographs. (A) Fluorescent cyclin-B1 labeling in cell nuclei; (B) DAPI counterstain of the same monolayer. Few proliferating cells are seen at the edge of the monolayer wound. Original magnification: ×100 in both images.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The purpose of this study was to demonstrate the extent to which integrin receptors mediate airway epithelial repair after injury, and to determine the ability of airway epithelial cells to repair after injury over different ECM proteins. A human airway epithelial cell line grown in monolayer culture on either collagen-IV, laminin-1, or laminin-2 matrix protein was studied after mechanical injury using time-lapse video microscopy.

Our data demonstrate that airway epithelial cells adhere to and repair over any of the tested ECM proteins with equal efficacy. Both collagen-IV and laminin-1 are present in the normal basement membrane (20), whereas laminin-2 is not seen in the basement membrane of normal subjects but is expressed in some subjects with asthma (23). It has been suggested that integrin molecules such as ₂₁ and ₆₁ may bind laminin-2 less well (24, 25). By implication, cells expressing this integrin may be less able to adhere to, or migrate over, this laminin isoform than are other matrix proteins. However, our study demonstrates that 16HBE14o cells repair equally well after injury when grown on laminin-2 compared with either laminin-1 or collagen-IV (Figure 2 and Table 3). Thus, the expression of a different laminin isoform in asthmatic airway basement membrane may not impede repair and reconstitution of the epithelium after injury.

Our data also demonstrate that the very late antigen integrins present on airway epithelial cells modulate repair to some extent, but that blockade with an anti-integrin subunit antibody is not complete regardless of the matrix protein support (Figures 3-6). This was especially true when cells were grown on either laminin-1 or laminin-2, as only ₁ subunit blockade substantially attenuated repair for monolayers grown on these matrix proteins (Figure 3 and Table 4). Furthermore, addition of any of the integrin subunit antibodies to control monolayers did not slow wound closure (Figure 7 and Table 4). These data suggest that adhesive mechanisms other than integrins regulate repair after injury, particularly when cells are grown on laminin.

Cell migration and spreading in the absence of EGF were generally < 20% (Figure 2), and addition of integrin subunit mAb to monolayer wounds in the absence of EGF did not further slow wound closure (Figure 7). One potential explanation is that treatment with EGF elicits a change in integrin expression that then, alone or coupled with another signal, initiates migration and spreading. However, integrin subunit expression was not changed by incubation for 24 h with EGF (Table 1). Alternatively, binding of the EGF receptor and initiation of the resulting tyrosine kinase cascade activate processes that lead to migration and spreading even without any change in integrin receptor expression. These pathways were not examined in our study.

The ₁ integrins may regulate cell spreading and migration by processes other than interaction with ECM. Several of the -integrins, such as ₂₁ and ₃₁, can be localized on some epithelial cell types not only at their junctions with matrix but along points of contact with other cells (32, 33). Integrins at sites of cell-cell interaction as well as at hemidesmosomes may be available for recruitment as epithelial cell migration proceeds (33). Such cell-cell interactions also may mediate functions such as branching cell morphogenesis in mammary epithelial cells (34). Blockade of -integrin subunits located at the points of cell-cell contacts in keratinocytes inhibits cell aggregation, whereas blockade of other -integrin subunits located at the point of cell adhesion to matrix does not (35, 36). Inhibition of airway epithelial cell spreading and migration in the present study with the use of antibodies to - and -integrin subunits, then, may be due to blockade of not only cell-matrix but also cell-cell interactions, as either or both sites may be available during repair. Conversely, the failure of some of the subunit antibodies to inhibit cell spreading and migration suggests that these integrin subunits did not mediate either type of interaction during wound repair.

Cell apoptosis was not observed in monolayers grown on each of the matrix proteins. Wound edges did not have an appreciable number of apoptotic cell nuclei (Figure 8). These data suggest that inhibition of wound closure by anti-integrin antibodies was not the result of the induction of apoptosis, as may result when integrin-matrix binding is prevented (16).

Cell proliferation also was not observed at the wound edge in monolayers grown on each matrix protein (Figure 9). These data suggest that wound closure is not a result of proliferation and ingress of daughter cells into the wound. In a 1-mm²-area wound, the number of cells removed is 3,200 to 12,700 (the average diameter of individual 16HBE14o cells measures between 10 and 20 µm, and resulting area 80 to 320 µm², in confluent monolayers as measured by digital microscopy). In a circular wound, between 200 and 400 cells are at the margin at time 0 based on the average diameter. If the mitotic index near the wound edge is approximately 3% in the first 18 h, only 6 to 20 cells at the margin can proliferate and move into the wound area. Given the range of initial wound area, this accounts for 0.2 to 3.1% of wound closure. Even if cells away from the margin with a similar or higher mitotic index also contribute to wound closure (and these can do so only if they then migrate into the wound after proliferating), the absolute contribution of cell proliferation during the first 18 to 24 h of wound closure in this model is small when compared with average wound closure of 70 to 90% during the first 24 h in monolayers treated with EGF (see Figure 2).

Reconstitution of a functional epithelial barrier requires an ordered sequence of events, including cell spreading and migration, adhesion, production of provisional matrix, proliferation, and differentiation into needed epithelial cell subtypes. Transmembrane integrins participate in several of these steps. The integrin heterodimer mediates cell adhesion to several different matrix proteins, and the specific combination of - and -subunits determines which of the several matrix proteins serve as a ligand (for review, see 21). Human airway epithelial cells express several -integrins that bind to collagen and laminin, including ₂, ₃, and ₆, each of which is complexed to the ₁-subunit (7, 8). Redistribution of the ₆-integrin subunit to lateral cell borders of epithelial cells has been demonstrated during epithelial repair in human airway xenographs transplanted into SCID mice (22), suggesting an active role for this subunit in repair. In the same study, ₂- and ₃-subunits were demonstrated to be present during all phases of epithelial repair (22). Integrins also modulate migration and adhesion in other epithelial cell types. For example, chick embryo retinal pigment cells require a functioning ₁-integrin subunit for migration (37). Both rat corneal epithelial cells and human keratinocytes migrate over fibronectin matrix using the ₅₁ integrin receptor (38, 39). Alveolar epithelial cells in culture migrate over type I collagen in a process mediated by ₂₁ integrin (40). Thus, integrins participate in the migration and repair in a variety of epithelial cell types. Our data demonstrate that the ₁-, ₂-, ₃-, and ₆-integrin subunits modulate repair in monolayer cultures on collagen-IV after acute mechanical injury. However, integrin subunit function in repair after injury in epithelial cells grown on either laminin-1 or laminin-2 is less certain. Treatment with a ₁-integrin subunit mAb attenuated repair (Figure 3), but treatment with -integrin subunits had little to no effect (Figures 4-6). These experiments suggest that alternative matrix receptors may participate in the repair process. The specific identity of such receptors was not identified in this study.

There are limitations to the present study. We chose to study cell migration using video microscopy, which is complementary to traditional studies of chemotaxis during early wound closure. This method, however, has limitations. Wound closure has been shown to involve both migration and cell spreading, and distinguishing between the two on a digitized image may be difficult. A second limitation relates to the simple matrices upon which our cells were grown, different from the more complex basement membrane in which a variety of matrix proteins, such as fibronectin, laminin, type IV collagen, proteoglycans, entactin and nidogen, are found (20). Interactions with multiple matrix proteins may have a different modulating effect on wound repair when compared with a single matrix protein. Third, although the video microscopy method takes into account more factors that are involved with wound repair in vivo, it is still an in vitro approximation. Cells present within (e.g., sensory nerves) or immediately beneath (e.g., fibroblasts) the epithelial layer may modulate the repair process to a damaged epithelium. Such effects are not present in our in vitro assay. Finally, airway epithelial cell lines in culture may not represent the same phenotype as in a normal trachea. Cultured epithelial cells are known to "dedifferentiate" rapidly, so that after several days in culture, few ciliated or secretory cells are present (41, 42). Although some morphologic features of the cells used in our study are not the same as the pseudostratified columnar epithelium seen in vivo (43), the monolayers are uniform and are similar to those described by other investigators (44). Nevertheless, changes in cell transformation and phenotype in vivo as a response to injury may substantially influence the initial reparative response and the use of integrins in cell spreading and migration.

In summary, we demonstrate that integrins modulate the repair of human airway epithelial cells grown on ECM after mechanical injury. Airway epithelial cells migrate and repair over laminin-2 equally well as on laminin-1 and collagen-IV. Attenuation of repair is matrix-specific, integrin-specific, and incomplete, suggesting that other mechanisms also have roles in epithelial repair.