Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Myofibroblasts were discovered in granulation tissues as hybrid cells of fibroblasts and smooth muscle cells (SMC), and ultrastructurally they have both contractile elements observed in SMC and well-developed rough endoplasmic reticula found in fibroblasts (1). Myofibroblasts have been found in a variety of normal tissues and various pathologic situations, including wound-healing, fibromatosis, and a stromal reaction to epithelial tumors (2, 3). Although myofibroblasts are believed to be morphologic intermediates of fibroblasts and SMC, they are not functionally intermediate. Myofibroblasts have been reported to have higher collagen-synthesis activity, especially of type III and I, than do fibroblasts (2). With regard to the wound-healing process, myofibroblasts have been suggested to be involved in tissue contraction, which is necessary for wound closure, as well as in production of the extracellular matrix. These results indicate that myofibroblasts with a high level of metabolic activity are induced in tissue remodeling, a level that is likely not the same as that found in fibroblasts and SMC.

There have been many reports concerning the interaction of myofibroblasts and wound-healing in a variety of tissues (1, 5, 6). In lung tissue, myofibroblasts have been suggested to participate in granulation and fibrogenesis in pulmonary sarcoidosis and lung fibrosis (7, 8). It has also been reported that myofibroblasts are involved in subepithelial fibrosis of the airway in asthma (9), and transforming growth factor (TGF)-1, one of the strong inducers of myofibroblasts, is partly responsible for this fibrosis (13). In addition, because TGF-1 is secreted by most cells in a latent form, some kinds of activatorssuch as thrombospondin (TSP)-1, which has recently been proposed to be a major activator under physiologic conditions (16)might be implicated in the remodeling as well. From these observations, we hypothesized that myofibroblasts play a central role in the subepithelial fibrogenetic remodeling process in the asthmatic airway, and that epithelial shedding might be an important trigger leading to this process. Therefore, the following wound-healing model of epithelium was designed.

To explicate the intricate puzzle of myofibroblast pathogenesis in the airway-wall remodeling process, we investigated the effects of mechanical injury to airway epithelial cells on myofibroblast induction in an in vitro culture system. We have previously reported that coculture of guinea-pig tracheal epithelial cells and fibroblasts on and beneath an amnion membrane facilitates the differentiation of pseudostratified epithelial cells, almost identical to those of in vivo trachea (17, 18). Since then we have mechanically scraped epithelial cells as a model of epithelial shedding in asthmatic response and examined myofibroblast induction during the redifferentiation process of epithelial cells.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of Epithelial Cells and Fibroblasts from Guinea Pig Tracheas

The guinea-pig epithelial cells and fibroblasts were prepared as described previously (17, 18). Female Hartley-strained guinea pigs (Japan SLC Inc., Shizuoka, Japan) were used for the following experiments. The animals were killed by exsanguination under anesthesia with an intraperitoneal injection of 50 mg/kg pentobarbital. Their tracheas were immediately removed, and the tracheal epithelial cells were isolated by mild digestion with 1 mg/ml of pronase (Sigma, St. Louis, MO). The cells were collected by flushing the inside of the tracheal lumen with Dulbecco's modified Eagle's medium/Ham's F-12 (DMEM/F12) containing 5% fetal calf serum (FCS) (GIBCO BRL, Grand Island, NY), washed several times with phosphate-buffered saline (PBS), and resuspended in the culture medium.

After isolation of the epithelial cells, the tracheas were minced to pieces of around 1 to 2 mm³. The minced pieces were vigorously washed in PBS, and the washing procedure was repeated until the supernatant became clear to remove the remaining epithelial cells. The pieces were then cultured in Eagle's minimum essential medium (MEM) containing 10% FCS (GIBCO BRL) on a six-well culture plate (Corning, Inc., Corning, NY) by replacing the medium every 3 d. It was observed that the fibroblasts migrated from each piece and grew to subconfluence in each well during the first 2 wk. The remaining pieces were then removed by gentle pipetting, and the outgrowing fibroblasts from those pieces were harvested from the wells using trypsin ethylenediaminetetraacetic acid. The recovered cells were washed with PBS and then resuspended in MEM-FCS. Non-spindle-shaped cells were mechanically removed by a Pasteur pipette under a microscope. In total, the fibroblasts were passaged five to seven times, and for each experiment they were immunohistochemically stained with an anti -smooth muscle actin (-SMA) monoclonal antibody (mAb) (Sigma) to assess the contamination of SMC (18).

Three-Dimensional Cell Culture with Human Amnion Membrane and Experimental Wound Model

Epithelial cells and fibroblasts were cultured in the human amnion chamber using a modified method according to previous reports (17, 18). Briefly, a human amnion was peeled away from the chorion of a normal-term placenta obtained immediately after delivery, and immersed in 0.25 M NH₄OH. The epithelial layer and debris of each amnion were scraped off, and the membrane was placed within a tissue-holding device with its epithelial side facing upward. The device was composed of two concentric polycarbonate rings, each having an outside diameter of 30 mm and an inside diameter of 14 mm.

The epithelial cells were seeded over the amnion membrane in the upper compartment of the chambers at 1 × 10⁶ cells/chamber. Each chamber was placed in a six-well culture plate containing 5 ml DMEM/F12-FCS and maintained at 37°C in an incubator with 5% CO₂ and 95% air. The epithelial cells were maintained by immersion feeding during the first week of culturing, and were then maintained under air-liquid interface feeding for another 2 wk. Fibroblasts were then seeded on plastic sheets (Wako Pure Chemicals Ltd., Osaka, Japan), which were placed on the bottom of the culture plate at a density of 5 × 10⁵ cells/well. After coculturing the epithelial cells with fibroblasts for 10 d, two scrape-line injuries 1 mm in width were made centrally across the epithelial layer using a Pasteur pipette (Figures 1A and 1B). Each injured site was confirmed by phase-contrast micrography. The denuded epithelial cells were washed out with PBS and continuously cultured with fibroblasts for 2, 4, or 8 d.

View larger version (80K): [in this window] [in a new window]   Figure 1.   (A) The human amnion chamber with well-differentiated epithelial cells cultured in its upper compartment. (B) Two scrape-line injuries with 1 mm in width were made centrally across the epithelial layer using a Pasteur pipette; the denuded epithelial cells were then washed with PBS and continuously cultured with fibroblasts for 2, 4, or 8 d.

Electron Microscopic Observations

The amnion membranes with the cultured epithelial cells were removed from the tissue-holding devices before and on Days 2, 4, and 8 after injury. On Day 4 after injury, the plastic sheet on which the fibroblasts were cultured was also detached from the bottom of the six-well culture plate. Each membrane and the plastic sheet were washed twice with PBS, fixed in 2% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.2), and postfixed in 1% osmium tetroxide for 1 h in sodium phosphate buffer (0.1 M, pH 7.2). They were then dehydrated in a graded series of ethanol (50 to 100%) followed by propylene oxide, and were embedded in Epon 812 (Abbot, North Chicago, IL). Ultrathin sections were cut using an Ultrotome V 2088 (LKB-Produkter AB, Bromma, Sweden), and stained with uranyl acetate and lead citrate. The sections were evaluated with an electron microscope (H-7000; Hitachi Ltd., Tokyo, Japan).

Immunohistochemistry

Before and on Days 2, 4, and 8 after injury, the cultured fibroblasts on the plastic sheets were fixed in 4% paraformaldehyde solution for 10 min at 4°C, washed with distilled water, and dried. Before and on Day 4 after injury, the amnion membranes with epithelial cells were fixed in 4% paraformaldehyde for 10 min at 4°C, embedded in OCT compound (Miles, Elkhart, IN), and frozen in isopentane cooled by dry ice. The amnion membranes with intact epithelial cells were also treated as a control at the same time. The 8-µm sections were prepared on poly-L-lysine-coated slides for immunohistochemistry. Immunohistochemical staining was carried out with mAbs using avidin biotinylated enzyme complex method (Vectastain kit; Vector Laboratories, Burlingame, CA). Specific primary antibodies included anti--SMA at a concentration of 20 µg/ml, anti-TGF-1 (Chemicon International, Inc., Temecula, CA) at 10 µg/ml, and anti-TSP (Immunotech, Marseille, France) at 10 µg/ml. The specificity of the anti- TGF-1 and anti-TSP antibodies was determined by Western blot analysis with the protein of the guinea-pig platelets, which revealed a single band. As a control, nonimmune mouse immunoglobulin (Ig) G for anti--SMA and anti-TGF-1, and nonimmune rat IgG for anti-TSP, were used. Color development was performed with 3,3'-diaminobenzidine tetrahydrochloride (DAB) as a chromogen, and specimens were counterstained with Myer's hematoxylin. -SMA-immunoreactive fibroblasts were counted in a blinded fashion and expressed as the number of cells per 10⁵ µm² of five different experiments. We present these results as the percentage of -SMA-positive cells in the total fibroblasts.

Labeling of Apoptotic Cells

Before and on Day 8 after injury, the cultured fibroblasts on the plastic sheets from different experiments were fixed in 4% paraformaldehyde solution for 10 min at 4°C, washed with distilled water, and dried. A deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling (TUNEL) assay was performed to identify the cells undergoing apoptosis using the In Situ Apoptosis Detection Kit (ApopTag; Intergen Company, Purchase, NY) according to the manufacturer's protocol.

DNA Probes

We obtained the guinea-pig complementary DNA (cDNA) of an -SMA fragment from the total RNA of guinea-pig SMC by reverse transcription polymerase chain reaction (PCR) using a GeneAmp RNA PCR kit (Perkin-Elmer, Foster City, CA) with a set of gene-specific primers, as shown in Table 1. Simultaneously, the guinea-pig cDNAs of TGF-1, TSP-1, procollagen 1(I), and procollagen 1(III) fragments were obtained from the total RNA of guinea-pig fibroblasts, as well as that of integrin 6 fragment from the total RNA of guinea-pig epithelial cells. Each primer was designed on the basis of published cDNA sequences (19). The amplified PCR products were ligated into pCR II-TOPO vector (Invitrogen, Carlsbad, CA) and transformed into One Shot chemically competent cells (Invitrogen). After isolating plasmid DNAs from clones containing the inserts, the plasmids were sequenced by the dideoxynucleotide chain-termination method with an autosequencer (ABI PRISM-310; Perkin-Elmer). The nucleotides from the guinea-pig cDNAs were highly homologous to the published sequences (19), and were submitted to the Genbank database (accession numbers AF169349 for -SMA, AF169347 for TGF-1, AF169345 for TSP-1, AF169346 for procollagen 1[I], AF169348 for procollagen 1[III], and AF169344 for integrin 6).

RNA Isolation and Northern Blot Analysis

An RNeasy total RNA kit (Qiagen, Hilden, Germany) was used to extract the total RNA from fibroblasts and epithelial cells from three different experiments. A 4-µg sample of total RNA from each cell was electrophoresed on a 4% formaldehyde/1% agarose gel and transferred to a Hybond N+ nylon membrane filter (Amersham, Little Chalfont, Bucks, UK). Hybridization was performed with [³²P]deoxycytidine triphosphate-labeled -SMA cDNA, TGF-1 cDNA, TSP-1 cDNA, procollagen  1(I) cDNA, and procollagen 1(III) cDNA at 68°C for 60 min in ExpressHyb hybridization solution (Clontech, Palo Alto, CA). The blots were washed with 2× 0.15 M NaCl/0.015 M sodium citrate (SSC)/ 0.05% sodium dodecyl sulfate (SDS) at room temperature for 30 min, followed by 0.1× SSC/0.1% SDS at 50°C for 30 min. The membrane was exposed to a BAS-SR imaging plate (Fuji Film, Tokyo, Japan), and analyzed by a BAS5000 imaging analyzer (Fuji Film). The messenger RNA (mRNA) from each epithelial cell and fibroblast was quantitated by densitometric analysis. Values were normalized on the basis of hybridized -actin mRNA, and are expressed as percentages of the preinjury values.

TGF-1 Assay in Culture Medium

The TGF-1 levels were measured for each coculture supernatant from injured epithelial cells and fibroblasts before and on Days 2, 4, and 8 of injury, and simultaneously from intact cocultured cells. The culture medium was exchanged for fresh DMEM/ F12-FCS 24 h before collection. Sample series from each experiment were continuously collected from three different experiments. Those supernatants were centrifuged at 10,000 × g at 4°C for 10 min, and then the TGF-1 immunoreactivities in the samples were analyzed by enzyme-linked immunosorbent assay (TGF-1 ELISA kit; R&D Systems, Inc., Minneapolis, MN). In this assay the minimal detection limit for TGF-1 was 7 pg/ml, and no significant cross-reactivities with TGF-2 and TGF-3 were observed.

A direct bioassay for TGF- activity was also carried out. Growth inhibition of the TGF--sensitive mink lung cell line CCL-64 (American Type Culture Collection, Rockville, MD) was used to quantify the TGF- concentrations in the supernatants (25). Serial dilutions of the supernatants were prepared on 96-well microplates (Corning) to a volume of 100 µl. CCL-64 cells were seeded at 3 × 10⁴ cells/well and cultured for 3 d in 200 µl RPMI 1640 (Sigma) with 10% FCS. Subsequently, the inhibition of cell growth was determined by a tetrazorium-based assay. The supernatants were carefully removed, and 100 µl of a 1% solution of 3,(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) (Sigma) in the medium with 10% FCS was added to each well. After a 1-h incubation at 37°C in 5% CO₂/air, the medium was removed and the MTT-formazan crystals were dissolved in 100 µl dimethylsulfoxide (Wako). Plates were placed on a shaker for 5 min, and the absorbance was measured with a microplate reader (Bio-Rad Laboratories Ltd., Richmond, VA) at a wave length of 540 nm. Samples were tested in triplicate, and the TGF- concentrations were determined by the growth inhibition caused by the sample, compared with a standard curve obtained by 1,000-16 pg/ml recombinant TGF- (Biomedical Technologies, Inc., Stoughton, MA).

Inhibitory Study Against TGF-1 and TSP-1

In order to elucidate whether TGF-1 and TSP-1 are involved in injury-related myofibroblast induction, we examined the transformation of fibroblasts into myofibroblasts using two neutralizing antibodies against TGF-1 or TSP (Immunotech), or LSKL, a peptide that blocks the TSP-1-induced activation of TGF- (26) in three different epithelial injury models. Each antibody and the blocking peptide, at final concentrations of 1 mg/ml, 10 mg/ml, and 80 nM, were simultaneously added to the culture medium to inflict the mechanical epithelial injury. The culture medium and the neutralizing antibodies or blocking peptide were replaced after 36 h. As a control, nonimmune mouse IgG for anti-TGF-1 and nonimmune rat IgG for anti-TSP were used. On Day 4 after injury, fibroblasts with and without the blocking reagents were fixed in 4% paraformaldehyde and subjected to the immunohistochemical study for -SMA.

Aprotinin (Sigma), an inhibitor of plasmin, was also added to examine the participation of plasmin in TGF-1-mediated myofibroblast induction in three different experiments at a concentration of 1,000 U/ml. The transformation of fibroblasts into myofibroblasts was estimated as well.

Statistical Analysis

Data are presented as means ± standard error of the mean (SEM). Statistics were obtained using Statview 5.0 (Abacus Concepts, Inc., Cary, NC). The data from the time-course experiments were analyzed by Dunnett's test as a multiple comparison test, and the data between the two groups were evaluated by an unpaired Student's t test. For all comparisons, statistical significance was accepted as P < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Histologic Analysis of Re-epithelialization

Electron microscopic observations revealed that epithelial cells on an amnion membrane cocultured with fibroblasts show pseudostratified columnar epithelium with ciliated cells and goblet cells (Figure 2A), almost identical to that of the in vivo airway as described previously (18), before injury. With a Pasteur pipette, mechanical scrapes were made across the epithelial layer, and epithelial cells alone were removed from an injured site, whereas the amnion membrane remained undamaged (Figures 1A and 1B). At 1 d after injury (Day 2), the epithelium-denuded part of the amnion membrane was thoroughly covered with flat nonciliated cells (Figure 2B). Redifferentiation of epithelial cells proceeded from the proximal to the distal site of the injury. The epithelial cells were entirely changed to a cuboidal shape with ciliated cells on Day 4 after injury (Figure 2C), and returned to a pseudostratified and columnar epithelium with ciliated cells and goblet cells on Day 8 (Figure 2D).

View larger version (85K): [in this window] [in a new window]   Figure 2.   Transmission electron micrographs show the characterization of guinea-pig epithelial changes before (A) and on Days 2 (B), 4 (C), and 8 (D) after injury. The culture of epithelial cells on an amnion membrane was interrupted by mechanical scrape injuries, as shown in Figure 1B. On Day 2 after injury, the deficient part was covered with a confluent monolayer of flat, nonciliated epithelial cells (B). The epithelial cells were cuboidal with ciliated cells on Day 4 (C), and then took on a pseudostratified appearance with ciliated cells and goblet cells on Day 8 (D). Bar = 5 µm.

Myofibroblast Induction after Epithelial Shedding

After the fifth passage, the purity of fibroblasts was morphologically and immunohistologically determined. Fibroblasts showed a typical peculiar spindle shape and formed a confluent monolayer with no piling up of cells. No endothelial and epithelial cells were observed. In addition, these spindle-shaped cells were not stained with mAb to -SMA and keratin (data not shown).

To evaluate the existence of myofibroblasts, immunohistochemical study using -SMA mAb was performed. The experiment revealed that -SMA-immunoreactive fibroblasts increased on Day 4 after injury (71.5 ± 2.8% of total cell count) compared with those before injury (4.5 ± 1.1% of total cell count), and decreased on Day 8 (18.8 ± 1.2% of total cell count) (Figure 3). To determine whether the expression of -SMA mRNA was correlated with that of -SMA protein, a Northern blot analysis was performed. The size of the guinea-pig -SMA mRNA (1.7 kb) was found to be consistent with that of human -SMA mRNA (19). Concurrently, -actin was recognized by the same guinea-pig -SMA cDNA, whose size was consistent with that of human -actin (2.1 kb) (27). The densitometric quantitation of -SMA mRNA was normalized on the basis of hybridized -actin mRNA, and is expressed as a percentage of the preinjury values to be 537± 357, 5,337 ± 1,616, and 677 ± 464% in fibroblasts on Days 2, 4, and 8 after injury, respectively. A very small amount of -SMA mRNA was constitutively expressed in guinea-pig fibroblasts before injury; however, its expression was significantly increased on Day 4 after epithelial injury (P < 0.05) and decreased on Day 8 (Figure 4).

View larger version (198K): [in this window] [in a new window]   View larger version (10K): [in this window] [in a new window]   Figure 3.   (A-D) Immunohistochemistry for -SMA in fibroblasts before (A) and on Days 2 (B), 4 (C), and 8 (D) after epithelial injury. Color development was carried out with DAB as a chromogen, and specimens were counterstained with Myer's hematoxylin. -SMA-positive cells increased significantly on Day 4. Bar = 10 µm. (E) Dynamics of -SMA expression in guinea-pig fibroblasts after epithelial injury. The fibroblasts were counted in a blinded fashion and are expressed as percentages of the total cell count of five different experiments. Each result is expressed as the mean ± SEM. * P < 0.05 compared with the number of fibroblasts before injury. The number of -SMA-immunoreactive fibroblasts was significantly increased on Day 4 after injury and decreased on Day 8.

View larger version (28K): [in this window] [in a new window]   Figure 4.   Northern blot analysis demonstrating the time course of -SMA mRNA expression in fibroblasts before and on Days 2, 4, and 8 after epithelial injury. Each lane contains the total RNA (4 µg) isolated from fibroblasts, RNA that was electrophoresed onto a nylon membrane and hybridized with 32P-labeled -SMA cDNA. The -SMA cDNA probe recognized not only -SMA but also -actin, which was used as an internal standard. The mRNA expression of -SMA was transiently increased on Day 4 after epithelial injury.

By electron microscopic observation, myofibroblasts were morphologically confirmed. Fibroblasts on Day 4 after epithelial injury had features of myofibroblasts such as bundles of microfilaments with dense bodies running parallel to the long axis of the cell, notched nuclei, and well-developed rough endoplasmic reticula. In the extracellular space, there were collagen fibers with finer fibrillar materials without periodicity (Figure 5).

View larger version (134K): [in this window] [in a new window]   Figure 5.   Transmission electron micrograph of fibroblasts on Day 4 after injury. The fibroblasts on Day 4 had features of myofibroblasts such as bundles of microfilaments with dense bodies running parallel to the long axis of the cell (arrowheads), notched nuclei, and well-developed rough endoplasmic reticula (arrow). In the extracellular space there were collagen fibers with finer fibrillar materials without periodicity (asterisks). Bar = 10 µm.

Investigating the termination of the expression of myofibroblasts on Day 8, we evaluated the presence of apoptosis in our wound-healing model. TUNEL staining was performed against fibroblasts before and on Day 8 after epithelial injury. The number of apoptotic cells was increased on Day 8 (39.2 ± 7.2%), compared with that before injury (1.9 ± 0.5%).

TGF-1 Expression in the Wound-Healing Model

We examined the expression of TGF-1 in our wound-healing model. Northern blot analysis demonstrated the TGF-1 mRNA levels in epithelial cells and fibroblasts before and on Days 2, 4, and 8 after injury. The size of the guinea-pig TGF-1 mRNA (2.5 kb) was found to be consistent with that of human TGF-1 mRNA (20). The densitometric quantitation of TGF-1 mRNA was normalized on the basis of hybridized -actin mRNA, and is expressed as a percentage of the preinjury values to be 110 ± 23, 90 ± 1, and 107 ± 9% in fibroblasts; and 117 ± 12, 97 ± 18, and 110 ± 10% in epithelial cells on Days 2, 4, and 8 after injury, respectively. TGF-1 mRNA was continuously expressed in those cells; however, there was no significant change after the injury (Figures 6A and 6B).

View larger version (39K): [in this window] [in a new window]   Figure 6.   Northern blot analysis demonstrating the time course of TGF-1 mRNA expression in epithelial cells (A) and in fibroblasts (B) before and on Days 2, 4, and 8 after epithelial injury. Each lane contains the total RNA (4 µg) isolated from epithelial cells and fibroblasts, RNA that was electrophoresed onto a nylon membrane and hybridized with 32P-labeled TGF-1 cDNA. The levels of -actin mRNAs are shown as loading controls. The mRNA of TGF-1 was continuously expressed in epithelial cells and fibroblasts without any significant change.

We also examined the immunohistochemical localization of TGF-1 of the amnion membrane with injured epithelial cells and fibroblasts during the wound-healing process. The amnion membrane was strongly immunoreactive to TGF-1, compared with epithelial cells, before the experimental injury. However, on Day 4 after injury, at the same time the myofibroblasts were induced, the immunoreactivity to TGF-1 in the amnion membrane disappeared, whereas strong immunoreactivity remained in the control membrane with noninjured epithelial cells (Figure 7). In contrast, TGF-1-immunoreactive fibroblasts were found to be increased on Day 4 after injury (Figure 8).

View larger version (147K): [in this window] [in a new window]   Figure 7.   Immunohistochemistry for TGF-1 in amnion membranes after epithelial injury (A and B) and that in the control without injury (C and D). The amnion membranes with epithelial cells were fixed on Days 1 (A and C) and 4 (B and D). Color development was carried out with DAB as a chromogen, and specimens were counterstained with Myer's hematoxylin. The amnion membranes were strongly stained with TGF-1 mAb from the beginning of the experiment. On Day 4 after injury, the immunoreactivity of the membrane (B) was remarkably lessened compared with that of the noninjured control (D). Bar = 10 µm.

View larger version (142K): [in this window] [in a new window]   Figure 8.   Immunohistochemistry for TGF-1 in fibroblasts before (A) and on Day 4 (B) after epithelial injury. Color development was carried out with DAB as a chromogen, and specimens were counterstained with Myer's hematoxylin. The number of TGF-1-positive cells increased significantly on Day 4. Bar = 10 µm.

In addition, we measured the levels of immunoreactive TGF-1 in the culture medium obtained before and on Days 2, 4, and 8 after injury to the epithelial cells. All the samples were activated by acidic pH before measurement because active TGF-1 was not detected in any of the supernatants. Fresh conditioned medium originally contained 408 ± 51 pg/ml of TGF-1 due to FCS, of which the subtraction from the measured level revealed the amount of TGF-1 newly released from the coculture system (within 24 h). TGF-1 release in the culture medium before and on Days 2, 4, and 8 was 579 ± 200, 936 ± 284, 1,188 ± 310, and 968 ± 410 pg/ml in the injured group; whereas it was 432 ± 137, 600 ± 200, 726 ± 186, and 611 ± 330 pg/ml in the noninjured group. The culture medium generated increased quantities of TGF-1, which were maximal by Day 4 after injury; however, there was no statistical significance (Figure 9). With the bioassay for TGF- activity, the concentrations of biologically active TGF- in the supernatants during the wound-healing process were undetectable when determined by the inhibitory response of CCL-64 cells using an MTT assay.

View larger version (22K): [in this window] [in a new window]   Figure 9.   The amount of latent TGF-1 released in the culture supernatants from injured epithelial cells and fibroblasts, as well as from intact cells, before and on Days 2, 4, and 8 (n = 3). TGF-1 immunoreactivity in the samples was analyzed by ELISA. All samples were activated by acidic pH before measurement. The culture medium from the injured group generated increased quantities of TGF-1, which were maximal by Day 4. However, there was no statistical significance.

Expression of TSP Protein and mRNA in Fibroblasts

The measurement of TGF-1 demonstrated increased levels in the supernatants from Day 4 after injury; however, the increment was not significant. Moreover, the detectable TGF-1 was entirely in a latent form. These data suggest the contribution of some kind of TGF-1 activator, whose expression might be increased during this wound-healing process. TSP-1 is one possible candidate, as has been recently described (16). To assess TSP production in the fibroblasts during this experiment, an immunohistochemical study using anti-TSP antibody was performed, which revealed that the levels of TSP-immunoreactive fibroblasts increased on Day 4 after injury (Figure 10). Total mRNA was extracted from the fibroblasts before and on Days 2, 4, and 8 after injury and was provided for the Northern blot analysis. The guinea-pig TSP-1 message (6.0 kb) was found to be identical in size to the published human TSP-1 mRNA (21). The densitometric quantitation of TSP-1 mRNA was normalized on the basis of hybridized -actin mRNA, and is expressed as a percentage of the preinjury values to be 217 ± 3, 617 ± 147, and 83 ± 15% in fibroblasts on Days 2, 4, and 8 after injury, respectively. The expression of TSP-1 mRNA was slightly detectable before and on Day 2 after injury, significantly increased on Day 4 (P < 0.05), and returned to control values on Day 8 (Figure 11), with this time course correlating with the induction of myofibroblasts.

View larger version (138K): [in this window] [in a new window]   Figure 10.   Immunohistochemistry for TSP in fibroblasts before (A) and on Day 4 (B) after epithelial injury. Color development was carried out with DAB as a chromogen, and specimens were counterstained with Myer's hematoxylin. The number of TSP-positive cells increased significantly on Day 4. Bar = 10 µm.

View larger version (47K): [in this window] [in a new window]   Figure 11.   Northern blot analysis demonstrating the time course of TSP-1 mRNA expression in fibroblasts before and on Days 2, 4, and 8 after epithelial injury. Each lane contains the total RNA (4 µg) isolated from fibroblasts, RNA that was electrophoresed onto a nylon membrane and hybridized with 32P-labeled TSP-1 cDNA. Levels of -actin mRNAs are shown as loading controls. The mRNA expression of TSP-1 was transiently increased on Day 4 after epithelial injury.

Integrin v6 Expression in the Wound-Healing Model

We used Northern blot analysis to estimate the integrin v6 mRNA levels in epithelial cells and fibroblasts before and on Days 2, 4, and 8 after epithelial injury. In our injury model, integrin v6 mRNA was found to be continuously expressed in epithelial cells and to demonstrate no significant change after the injury, whereas it was absent in fibroblasts (data not shown).

Collagen Synthesis in Fibroblasts after Injury

We then determined by Northern blot analysis whether the expression of procollagen 1(I) and procollagen 1(III) mRNA was correlated with the time course of the expression of -SMA mRNA after epithelial injury. Total mRNA was extracted from fibroblasts before and on Days 2, 4, and 8 after injury. The guinea-pig procollagen 1(I) transcript was estimated to be 5.8 and 4.8 kb, and the guinea-pig procollagen 1(III) RNA to be 4.9 kb, which is consistent with published mRNA values (22, 23). The densitometric quantitation of each mRNA was normalized on the basis of hybridized -actin mRNA, and the values are expressed as percentages of each preinjury value. The values of procollagen 1(I) mRNA were 237 ± 43, 897 ± 129, and 90 ± 15% in fibroblasts on Days 2, 4, and 8 after injury, respectively; and of procollagen 1(III) mRNA on Days 2, 4, and 8 were 243 ± 18, 1,360 ± 155, and 147 ± 37%, respectively. The expression of procollagen 1(I) and procollagen 1(III) mRNA was significantly increased on Day 4 after epithelial injury (P < 0.05) and decreased on Day 8 (Figure 12).

View larger version (42K): [in this window] [in a new window]   Figure 12.   Northern blot analysis demonstrating the time course of procollagen 1(I) and procollagen 1(III) mRNA expression in fibroblasts before and on Days 2, 4, and 8 after epithelial injury. Each lane contains the total RNA (4 µg) isolated from fibroblasts, RNA that was electrophoresed onto a nylon membrane and hybridized with 32P-labeled procollagen 1(I) and procollagen 1(III) cDNA. The levels of -actin mRNAs are shown as loading controls. The levels of procollagen 1(I) and procollagen 1(III) were transiently increased on Day 4 after epithelial injury.

Inhibitory Study against TGF-1 and TSP-1

To confirm the participation of TGF-1 or TSP-1 in the induction of myofibroblasts, we respectively added either neutralizing antibodies against TGF-1 or TSP, or LSKL, to the conditioned medium after epithelial injury, and the number of cells immunoreactive for -SMA was determined on Day 4. -SMA-immunoreactive fibroblasts with the blocking reagents significantly decreased (3.2 ± 1.1% of total cell count with anti-TGF-1 antibody; 2.8 ± 0.4% of total cell count with anti-TSP antibody; and 3.8 ± 1.3% of total cell count with LSKL) compared with those without blocking reagents (73.6 ± 4.7% of total cell count) (P < 0.05) (Figure 13). Nevertheless, aprotinin had no inhibitory effect on the expression of -SMA-immunoreactive fibroblasts (68.9 ± 1.2% of total cell count).

View larger version (11K): [in this window] [in a new window]   Figure 13.   Immunoreactivity of fibroblasts for -SMA with or without blocking reagents, including neutralizing antibodies against TGF-1, TSP1, LSKL, or aprotinin, on Day 4 after epithelial injury. The fibroblasts were counted in a blinded fashion and are expressed as percentages of the total cell count of three different experiments. Each result is expressed as the mean ± SEM. * P < 0.05 compared with the number of fibroblasts without blocking reagents. The immunoreactivity to -SMA was significantly decreased by adding either neutralizing antibody or blocking peptide to the conditioned medium after injury. Aprotinin was found to have no inhibitory effect on the expression of -SMA-immunoreactive fibroblasts.

To determine whether TGF-1 is involved in the regulation of collagen synthesis and TSP-1 expression, we performed a Northern blot analysis in the presence of the neutralizing antibody to TGF-1. The densitometric quantitation of each mRNA was normalized on the basis of hybridized -actin mRNA, and the values are expressed as percentages of each preinjury value. In fibroblasts on Days 2, 4, and 8 after injury with anti-TGF-1 antibody, the respective values of procollagen 1(I) mRNA were 103 ± 8.8, 110 ± 20.8, and 100 ± 11.5%; those of procollagen 1(III) mRNA were 100 ± 15.3, 107 ± 6.7, and 107 ± 3.3%; and those of TSP-1 mRNA were 103 ± 8.8, 127 ± 14.5, and 120 ± 15.3%. Compared with the injury model without neutralization (Figures 11 and 12), the expression of procollagen 1(I), procollagen 1(III) and TSP-1 mRNA was not upregulated under TGF-1-blocking conditions (Figure 14).

View larger version (48K): [in this window] [in a new window]   Figure 14.   Northern blot analysis demonstrating the time course of procollagen 1(I), procollagen 1(III), and TSP-1 mRNA expression in fibroblasts with neutralizing antibody against TGF-1 before and on Days 2, 4, and 8 after epithelial injury. Each lane contains the total RNA (4 µg) isolated from fibroblasts, RNA that was electrophoresed onto a nylon membrane and hybridized with 32P-labeled procollagen 1(I), procollagen 1(III), and TSP-1 cDNA. The levels of -actin mRNAs are shown as loading controls. The expression of procollagen 1(I), procollagen 1(III), and TSP-1 mRNA was not upregulated under TGF-1-blocking conditions.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

By the present coculturing system for airway epithelial cells and fibroblasts, we demonstrated myofibroblast induction by means of the mechanical injury of airway epithelial cells. The appearance of myofibroblasts was synchronized with the epithelial redifferentiation process and increased collagen synthesis in fibroblasts, suggesting that even mechanical injury to epithelium causes tissue remodeling. This model could provide the basis for demonstrating of the underlying role of epithelial-mesenchymal interactions during the airway remodeling process.

Myofibroblast induction was defined by the expression of -SMA using Northern blot analysis and immunohistochemistry. Ultrastructural observation demonstrated the development of cytoplasmic bundles of actin microfilaments, enlarged endoplasmic reticula, and notched nuclei in those cells, all of which are the properties of myofibroblasts. The immunoreactivity of -SMA was absent before epithelial injury, although its mRNA was faintly observed. On Day 4 after injury, -SMA immunoreactivity and its mRNA were transiently upregulated. This phenotype change to that of myofibroblasts was triggered by epithelial shedding and redifferentiation, suggesting a cell-cell interaction between epithelial cells and fibroblasts through humoral factors.

Previous studies have shown that apoptosis mediates the disappearance of myofibroblasts during wound healing (5, 28). TUNEL staining revealed that the number of apoptotic cells was increased on Day 8 compared with Day 1, which also confirmed the presence of an apoptotic mechanism to terminate myofibroblast expression in our wound-healing model.

It is known that various cytokines are implicated in myofibroblast induction, one of which, TGF-, has been extensively investigated (2, 29). Zhang and colleagues (30) have recently reported that TGF-2 levels increased in culture supernatants after epithelial damage. There are only a limited number of differences in the effects of different TGF- isoforms; however, the specific function of each isoform is not fully understood. Although it was uncertain whether the anti-TGF-1 antibody used had cross-reactivity to TGF-2 or TGF-3, several studies have demonstrated high basic levels of TGF-1 expression in asthmatics (14, 15). Thus, we focused on TGF-1 in the present study. We found that neutralization against TGF-1 significantly suppresses the appearance of myofibroblasts, suggesting that TGF-1 plays an important role in the induction of myofibroblasts in our model. We further investigated the source and the transition of TGF-1 by immunohistochemistry and Northern blot analysis. It was found that the immunoreactivity of TGF-1 is primarily localized in the amnion membrane and is totally reduced after epithelial injury, and that mRNA of TGF-1 in epithelial cells and fibroblasts is not influenced by the injury. Further, TGF-1 immunoreactivity transiently appeared on the surface of fibroblasts on Day 4 after injury, which revealed that the cellular localization of TGF-1-protein expression was increased, whereas no active forms of TGF-1 were detected in any culture medium. These findings suggest that both newly synthesized TGF-1 and that stored in the amnion membrane might be supplied for myofibroblast induction in its latent form, and that its activation might occur at the cell surface of fibroblasts. In vivo, TGF-1 is released from a wide variety of cells and is largely stored in the extracellular matrix in a latent form (29). Latent TGF-1 is composed of mature TGF-, latency-associated protein (LAP), and latent TGF- binding protein (LTBP), which is a component of the extracellular matrix microfibrils and binds the latent TGF- to the extracellular matrix. An amnion membrane consists of types IV and V collagen and laminin, which are known to be structural components of basement membrane, and interfaces with an avascular collagenous stroma composed of interstitial types I and II collagen (17, 31). Therefore, the use of an amnion membrane such as the basement membrane and extracellular matrix in the present work can mimic physiologic conditions, and the accumulation of TGF-1 in the amnion membrane could be similar to that in vivo. The TGF- signaling is initiated by the proteolytic cleavage of LTBP, which causes the release of the latent TGF- from the extracellular matrix (29). Proteinase-mediated release might be susceptible to the action of various proteolytic enzymes. Among these, we have already confirmed that adding trypsin to the amnion would indeed lead to the loss of TGF- immunoreactivity (data not shown). Inasmuch as bronchial epithelium is known to produce proteinases such as matrix metalloproteinase (32) and trypsin-like protease (33), it is possible that these proteases might participate in the remodeling process.

Recently, TSP-1 has been reported to be a major activator for latent TGF- in vivo, and we investigated the effects of epithelial injury on the dynamics of TSP-1 in fibroblasts. Northern blot analysis in the present study showed that TSP-1 mRNA is already expressed in fibroblasts before epithelial shedding and was found to be upregulated on Day 4 after injury. TSP-immunoreactive fibroblasts were also found to be increased, in synchronization with TGF-1 expression. Therefore, we assessed the effects of TSP-1 on myofibroblast induction by blocking TSP-1 using a specific neutralizing antibody and LSKL. LSKL is a sequence of LAP, and TSP-1 interacts with latent TGF-1 through the LSKL peptide (26). To block the TSP-1 activation of latent TGF-1, LSKL was used at a 1,000-fold molar excess to the latent TGF-1 level in the conditioned medium. It was found that anti-TSP neutralizing antibody and the blocking peptide had an inhibitory effect on injury-induced myofibroblast formation. Moreover, TSP-1 upregulation was suppressed by blocking TGF-1. Thus, TSP-1 and TGF-1 were closely related to each other in the sense that TSP-1 would activate TGF-1 and would also be upregulated by TGF-1.

We tested for evidence of latent TGF-1 activation by two other mechanisms. Plasmin, a serine proteinase, is known to be an important activator in the coculture of bovine endothelial cells and SMC (34). To estimate the participation of plasmin, we investigated the effects of epithelial shedding on myofibroblast induction in the presence or absence of aprotinin, an inhibitor of plasmin. Myofibroblast induction was not suppressed by aprotinin, which revealed that plasmin did not play a major role in TGF-1- mediated myofibroblast induction in our in vitro wound-healing model. Integrin v6 has also been reported to be a candidate for activating latent TGF-1 (35). However, in our injury model, integrin v6 mRNA was found to be continuously expressed in epithelial cells and to be absent in fibroblasts. It seems therefore that the integrin v6 in epithelial cells may not account for the activation of TGF-1 in our model.

Although both fibroblasts and myofibroblasts have collagen-producing activity, the activity has been reported to be higher in myofibroblasts than in fibroblasts (2). Thus, several investigators have described a correlation between the appearance of myofibroblasts and subepithelial fibrosis in the airways (9). To elucidate the role of induced myofibroblasts in collagen synthesis in our study, we examined the mRNA levels of procollagen I and III in those cells. We found that those mRNAs were upregulated on Day 4 after epithelial injury, which was synchronized with the dynamics of myofibroblast differentiation. Further, we also found that ultrastructurally there was much precipitation of collagen fibrils around myofibroblasts when compared with fibroblasts. These findings suggest that induced myofibroblasts have more potent collagen-producing activity than fibroblasts. In addition, blocking TGF-1 during the wound-healing process resulted in no upregulation of procollagen mRNA, which confirmed that TGF-1 is strongly involved in collagen synthesis.

Our present finding that mechanical injury induces myofibroblasts may partly explain subepithelial fibrosis in asthma. It has been reported that eosinophil migration into the airway results in the expression of TGF- in asthma, whereas there has been no support for such data in another study (5). In our study, both epithelial cells and fibroblasts were observed to have TGF-1-producing activity. Because most cells have been reported to have the ability to produce TGF-, the key process in subepithelial fibrosis may be an activation of TGF-. We have demonstrated that in injury-induced myofibroblast formation, TGF-1 is activated on the cell surface of fibroblasts through TSP-1. That is, myofibroblast-related fibrogenesis would require the upregulation of such activators as TSP-1 on fibroblasts rather than an upregulation of TGF-1 synthesis.

In addition, we have also demonstrated that myofibroblast induction is synchronized with a disappearance of TGF-1 from the amnion membrane. Considering the present finding that mRNA does not change after epithelial injury, TGF-1 stored in the extracellular matrix may be used in the urgent repair of tissue. This hypothesis would also be supported by the preliminary observation that immunoreactivity to TGF-1 beneath the basement membrane of the guinea-pig airway appears to disappear after mechanical injury (data not shown).

In conclusion, we believe that airway myofibroblasts arise from fibroblasts upon stimulation of TGF-1 and TSP-1, which leads to active fibrogenesis, and that the damaged airway epithelium might play a crucial role in the genesis of these phenomena (see Figure 15). These results suggest that myofibroblasts are responsible for the airway remodeling process through epithelial-mesenchymal interactions. Our wound-healing model could become a new tool for the study of local airway remodeling.

View larger version (36K): [in this window] [in a new window]   Figure 15.   A summarized drawing of the sequence of events leading to the TGF-1-dependent induction of myofibroblasts. The epithelial shedding might initiate the TGF-1 signaling by the proteolytic cleavage of LTBP, which causes the release of the latent TGF-1 from the amnion. TSP-1 is also upregulated by the epithelial shedding or increased TGF-1, and TSP-1 then induces a change in the conformation of the latent TGF-1 (26), allowing the activation of TGF-1 on the surface of the fibroblast. The signal is transduced to the interior of the cell and then induces the myofibroblast. Concurrently, collagen synthesis is upregulated.