Integrin αvβ8-Mediated Activation of Transforming Growth Factor-β Inhibits Human Airway Epithelial Proliferation in Intact Bronchial Tissue (original) (raw)

Abstract

Transforming growth factor (TGF)-β is a potent multifunctional cytokine that is an essential regulator of epithelial proliferation. Because TGF-β is expressed almost entirely in a latent state in vivo, a major source of regulation of TGF-β function is its activation. A subset of integrins, αvβ8 and αvβ6, which are expressed in the human airway, has recently been shown to activate latent TGF-β in vitro, suggesting a regulatory role for integrins in TGF-β function in vivo. Here we have developed a novel, biologically relevant experimental model of human airway epithelium using intact human bronchial tissue. We have used this model to determine the function of integrin-mediated activation of TGF-β in the airway. In human bronchial fragments cultured in vitro, authentic epithelial-stromal interactions were maintained and integrin and TGF-β expression profiles correlated with profiles found in normal lung. In addition, in this model, we found that either the integrin αvβ8 or TGF-β could inhibit airway epithelial cell proliferation. Furthermore, we found that one mechanism of integrin-αvβ8-dependent inhibition of cell proliferation was through activation of TGF-β because anti-β8 antibody blocked the majority (76%) of active TGF-β released from bronchial fragments. These data provide compelling evidence for a functional role for integrin-mediated activation of TGF-β in control of human airway epithelial proliferation in vivo.

Homeostatic programs tightly regulate the proliferation of airway cell types and disturbances in this regulation can contribute to airway diseases such as lung cancer or asthma. 1 The mechanisms of airway homeostasis are incompletely understood but involve the interaction of cells with the extracellular matrix and with soluble cytokines/growth factors. 2-4 Integrins are the major family of molecules that regulate cell-extracellular matrix interactions whereas TGF-β constitutes a major family of cytokines/growth factors that regulates airway epithelial cell and subepithelial myofibroblast proliferation, pathological hallmarks of lung cancer and asthma, respectively. 2,5

Abundant evidence indicates a role for integrins in epithelial cell proliferation in vitro, whereas there is little known on the role of integrins in cell proliferation in vivo. 6 Integrin effects on cell proliferation have been attributed to integrin-ligand interactions resulting in the initiation of intracellular signaling pathways mediated through integrin cytoplasmic domains. 3,7 However, direct interactions of integrins with TGF-β have recently been described, raising the possibility that integrins might also function as regulators of extracellular signals. 8,9 Such cell surface regulation of cytokine activity could provide a mechanism for autocrine and paracrine regulation of homeostasis.

TGF-β is secreted by almost every cell in the body in a latent form that must be activated for it to bind to its receptors and initiate signaling through the TGF-β-signaling mediators. 10 Latency of TGF-β is maintained through the noncovalent association of the active TGF-β peptide with its propeptide, the latency-associated peptide (LAP). 11 The mechanisms regulating activation of TGF-β are incompletely understood but involve either conformational alterations or proteolysis of the LAP. 8,11-14 Of note, the LAPs of TGF-β1 and -β3 contain integrin recognition motifs, Arg-Gly-Asp (RGD), that serve as high-affinity ligand-binding sites for the airway integrin αvβ8. 8 We have recently shown in lung cancer cell lines that αvβ8 binds to the RGD sequence of the LAP of latent TGF-β and mediates activation of the latent TGF-β complex through a mechanism that is dependent on the transmembrane protease, MT1-MMP. 8

In the airway, integrin-mediated activation of TGF-β is likely to regulate both cell proliferation and extracellular-matrix production. For instance, in two-dimensional culture, αvβ8-mediated activation of TGF-β inhibited cell growth. 15 In lung cancer xenografts, heterologous expression of αvβ8 was associated with an increase in soluble active TGF-β, tumor growth inhibition, and excess extracellular matrix production. 8,15 However, no studies have addressed the role of integrin-mediated activation of TGF-β in cell proliferation and extracellular matrix production in the airway in vivo.

The integrin αvβ8 is expressed in the basal cells of the human airway. 15 Basal cells are the major proliferative compartment of the proximal airway and are the predominant cell type in direct contact with the basement membrane. 16 Autocrine and paracrine effects of TGF-β on basal cells could influence airway epithelial and subepithelial myofibroblast proliferation and basement membrane thickness and, thus, could be important in understanding both the pathogenesis of lung cancer and of airway remodeling in asthma. 1 However, airway basal cells are difficult to study for several reasons. For one, they resist isolation and growth in vitro cultures. For another, in the bronchial tree of rodents, they are essentially absent and their function may be served by other cell types. 17,18 In this regard, rodent models do not accurately reflect the biology of human airways. Therefore, to examine human airway proliferation in a biologically relevant context, a new system was required.

In this study, we have developed a novel, biologically relevant human airway model and used it to determine that αvβ8-mediated activation of TGF-β is a major homeostatic mechanism during human airway epithelial repopulation. We demonstrate that αvβ8 is the predominant TGF-β-activating integrin expressed in the bronchial fragment model. Treatment of bronchial fragments with either neutralizing anti-αvβ8 or anti-TGF-β antibodies (Abs) increases airway epithelial proliferation. Finally, we have determined that one mechanism of αvβ8-mediated regulation of proliferation involves the activation of TGF-β because treatment with neutralizing anti-β8 Abs significantly decreases TGF-β activation.

Materials and Methods

Tissues

Fresh lung tissue was obtained in the operating room from patients undergoing pneumonectomies or lobectomies for primary or metastatic lung tumors. Exclusion criteria were active lung infection, present smoking history, and neoadjuvant radiation therapy. Superficial endobronchial tissue from tumor-free, first to fifth generation bronchi was harvested using biopsy cup forceps. The clinical characteristics of each subject is given in Table 1 ▶ . Human tissues were obtained from consenting patients using a protocol approved by the Committee on Human Research at the University of California, San Francisco.

Table 1.

Characteristics of Subjects Used in the Study

Subject no. Sex Age Diagnosis Smoker Neoadjuvant therapy
1 M 55 NSCLC* Former None
2 M 56 NSCLC Former Chemo
3 F 71 Carcinoid Former None
4 F 69 NSCLC Former None
5 M 66 NSCLC Former None
6 M 61 Metastasis Former None
7 F 42 NSCLC Former Chemo
8 M 73 Lymphoma Former None
9 M 63 NSCLC Former None
10 M 85 NSCLC Former None

Bronchial Fragment Culture

The biopsies were immediately put into ice-cold bronchial epithelial growth medium (Clonetics, San Diego, CA). Within 4 hours, under a dissecting microscope, the biopsies were cut into fragments ∼0.5 mm in diameter. Each fragment was then placed in an individual well of an agar-coated 24-well plate and incubated overnight at 37°C in a humidified incubator in 7% CO2 using the liquid overlay culture technique as described. 19

Bronchial Fragment Studies

After recovering overnight, fragments with beating cilia, as evidence of a viable epithelium, were randomly transferred to individual wells on a 96-well agar-coated plate. Each well contained 50 μl of fresh bronchial epithelial growth medium with either neutralizing or control monoclonal Abs. The Abs were either 100 μg/ml anti-β8, clone 37E1; 8 10 μg/ml anti-TGF-β, clone 1D11 that blocks TGF-β1-3 (R & D Systems, Minneapolis, MN); or 100 μg/ml of the isotype-matched control Ab (HLA-A, B, C, clone W6/32; America Type Culture Collection, Rockville, MD). The fragments were transferred to new agar-coated plates with fresh medium and Ab after 1 day and thereafter every other day. At various time points, fragments were fixed for 30 minutes in 10% formalin and embedded in paraffin.

TGF-β Bioassay

The TGF-β bioassay was performed as previously described, 8 with the following modifications. Ciliated airway fragments at day 1 of culture were randomized into groups and were treated with the neutralizing anti-β8, anti-pan-TGF-β, or control Abs as above. After culture overnight, fragment groups (four per group) were co-cultured for 16 to 20 hours with the TGF-β reporter cell line, TMLC 20 in individual wells of a 96-well plate containing neutralizing Abs at the above concentrations. The samples were analyzed as previously described. 8 Relative luciferase units are defined as arbitrary luminometer units of experimental minus the TMLC background values.

Immunohistochemistry

Deparaffinized 4-μm sections of bronchial fragments were processed for immunohistochemistry using the primary Abs polyclonal anti-β8, 15 monoclonal anti-keratin 34βE12 (DAKO Corporation, Carpinteria, CA), monoclonal Ki-67 clone MIB 1 (Immunotech, Westbrook, ME), polyclonal anti-TGF-β1 and anti-TGF-β3 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), followed by the avidin-biotin-peroxidase detection systems LSAB2 or Envision+ (DAKO). Hematoxylin was used as a counterstain. The anti-keratin Ab 34βE12 recognizes the high molecular weight (HMW) cytokeratin types 1, 5, 10, and 14. 16

Determination of Cell Proliferation

Three slides with random sections (4 μm) from each treatment group were immunolabeled with the proliferation marker anti-Ki-67, as above. Before counting, the slides were blinded by an independent investigator. The epithelial cells occupied the most superficial cell layers covering the fragments and were easily identified by their epithelioid morphology. The total number of epithelial cells covering the surface of each fragment and the number of epithelial cells with Ki-67-positive nuclei were counted manually under a standard light microscope using a ×40 objective. After unblinding of the slides, the sum of cells from each fragment was calculated and the percentage of Ki-67-positive epithelial cells determined, henceforth referred to as the Ki-67 labeling index. Alternatively, to determine the concentration of Ki-67-positive cells in either the original intact epithelium, the transition zone, or the outgrowing epithelium, the number of Ki-67-positive epithelial cells found per ×40 field (n = 15) in each epithelial zone was recorded and expressed as labeled cells/high-power field (HPF).

Differential Gene Expression Analysis Using Two-Step Reverse Transcriptase-Polymerase Chain Reaction (PCR)

The method used has been described in detail elsewhere. 21 Briefly, eight fragments cultured for 3 days were snap-frozen separately in liquid nitrogen and individually sonicated for 1 minute in 0.7 ml of fresh RLT buffer (Qiagen, Valencia, CA) with 10 μl/ml of β-mercaptoethanol (Sigma, St. Louis, MO) and 20 μg/ml of linear acrylamide (Ambion Inc., Austin, TX). Total cellular RNA was isolated using a Qiagen RNeasy mini kit as specified by the manufacturer. After DNase treatment (RQ1 RNase-free DNase I; Promega, Madison, WI), RNA was further purified on a second RNAeasy mini column and stored frozen at −70°C. Total RNA was reverse-transcribed using the SuperScript first-strand synthesis system (Invitrogen, Carlsbad, CA). To control for genomic DNA contamination, RT+ and reverse transcriptase reactions were performed. PCR amplification of resulting cDNA was performed using Advantage 2 polymerase (Clontech, Palo Alto, CA) and a mixture of gene-specific primers for the integrin subunits β1, β6, and β8, and the housekeeping gene GAPDH. Primer sequences are available at http://astmagenomics.ucsf.edu. Commercial RNA (human total RNA panel 1, Clontech) was used as a reference pool for transcript quantification. Approximately 0.1 μl of the PCR product was used in a real time PCR reaction run in 10 μl of TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA) as described. 21 Plasmid template controls for the integrin subunits β1 (a gift from Y. Takata, Scripps Research Institute, La Jolla, Ca), β5 (American Type Culture Collection, Rockville, MD), and β8 22 were used as standards for calibration curves. Integrin log copy numbers were plotted against the raw cycle threshold (Ct) values and the resulting calibration curves used to determine the relative transcript copy numbers. For determination of differential gene expression levels, raw Ct values were normalized to Ct values for the housekeeping gene GAPDH.

Statistics

Student’s _t_-test was used for comparison of two data sets; analysis of variance (analysis of variance for parametric and the Mann-Whitney test for nonparametric data sets) or linear regression for more than two data sets. Tukey’s or Dunn’s test were used for parametric and nonparametric data, respectively, to find where the differences lay. Significance was defined as P < 0.05. Statistical software was Prism v.3, and Instat v.3 (GraphPad Software, Inc., San Diego, CA).

Results

Integrin αvβ8 Is the Predominant TGF-β-Activating Integrin Present in Airway Epithelium in Human Bronchial Fragments

We used real-time PCR to assess the expression of β8 transcript in bronchial fragments 21 (Figure 1, a and b) ▶ . The β1 integrin subunit was used as a comparison because the β1 integrins are known to be highly expressed in the airway. 23,24 We expected the transcript levels of β8 to be far less than β1 in bronchial fragments because β8 expression in the airway is confined to the basal cell layer of the epithelium and β1 integrins (ie, α2β1, α3β1, α5β1, and α9β1) are highly expressed throughout the epithelium and in the stroma. 15,23,24 Surprisingly, we found that the β8 mRNA was expressed at only slightly lower levels (0.7-fold difference) than the β1 integrin subunit (Figure 1, a and b) ▶ . Therefore, we conclude that the β8 transcript is highly expressed by the airway epithelium in bronchial fragments.

Figure 1.

Figure 1.

The β8 integrin subunit mRNA is highly abundant in airway fragments in culture. a: The relative abundance of the β8, β6, and β1 integrin subunit transcripts was compared using nested primers to amplify preamplified oligo-dt-primed cDNA from airway fragments after 3 days in culture (n = 8). Shown are the relative transcript numbers for each integrin subunit from each fragment. Bars indicate the median value. Each circle, square, or diamond represents a single fragment. b: The same data shown in a is depicted in bar graph form to show SE. *, P < 0.001.

We also compared the expression of the β8 subunit relative to the β6 integrin subunit transcript because β6 is the only other integrin β subunit known to participate in the activation of TGF-β. 8,9,13,25 Like β8, the expression of the β6 integrin subunit in the airway is confined to the epithelium. 26 The β8 transcript was expressed at higher levels in each of eight fragments than the β6 transcript (4.5-fold difference, P < 0.001) (Figure 1, a and b) ▶ supporting previous reports that β6 is not highly expressed in noninflamed airway epithelium. 26-28 Because β8 and β6 mRNA in tissues correlates with the presence of αvβ8 and αvβ6 protein, 27,29,30 these data suggest that the integrin αvβ8 is more highly expressed than the integrin αvβ6 in bronchial fragments. We therefore focused our subsequent investigations on determining the function of αvβ8 in bronchial fragments.

The real-time PCR methodology was validated for determining relative integrin subunit expression levels by demonstrating that preamplification of small amounts of control RNA (10 ng) did not change the relative representations of the β8, β6, or β1 integrin subunit and by showing essentially no differences in amplification efficiency between the same subunits throughout a large range (108) of input template concentration (data not shown).

The TGF-β-Activating Integrin αvβ8 Is Expressed by Airway Basal Cells during Airway Epithelial Repopulation

After harvest, bronchial fragments were partially covered by an intact normal airway epithelium (Figure 2, a and b) ▶ . From this intact epithelium, cells radiated outwards to cover the whole surface of the fragments by 7 days of culture (Figure 2c) ▶ creating a new squamoid epithelium (Figure 2b) ▶ . This new epithelium gradually became stratified with no obvious sign of goblet or ciliated cell differentiation during the 1-week culture period used in this study.

Figure 2.

Figure 2.

The airway integrin subunit β8 is expressed in the proliferative and nonproliferative basal cell compartment in the bronchial fragment model. a–c: H&E staining. a: Normal human bronchus. b: Bronchial fragment cultured for 1 day. The fragment is only partially covered by an epithelium. Outgrowing epithelial cells (arrows) migrate from the intact epithelium at a transition zone (arrowhead) to cover the exposed subepithelial stroma. c: Bronchial fragment cultured for 7 days. The exposed stroma (area between arrowheads) has been completely covered with outgrowing epithelium (arrow). d–f: Immunostaining using polyclonal anti-β8 Ab. d: The integrin subunit β8 is mostly restricted to basally located cells in normal bronchial epithelium (arrows). e: After 1 day of culture, the integrin subunit β8 is expressed by outgrowing epithelial cells (arrow), in all cells in the tapering transition zone (arrowhead), and in the basal cells of the original epithelium. f: After 7 days of culture, membrane staining for β8 in all layers of the outgrowing epithelium is seen (arrows). g–i: Immunostaining using the anti-HMW keratin Ab 34βE12. g: In normal bronchial epithelium, 34βE12 stains basal cells (arrows). h: After 1 day of culture, 34βE12 staining is seen in the outgrowing epithelium (arrow) and in the basal compartment of the intact ciliated epithelium (arrowhead). i: After 1 week of culture, only basally located cells in the outgrowing epithelium stain with 34βE12. j–l: Immunolabeling using an anti-Ki-67 Ab that labels proliferating cells. j: Ki-67-positive cells are rare in normal adult airway. A single immunopositive stromal cell (arrow) is shown. k: After 2 days of culture a few proliferating cells (arrow) could be found in the transition zone (arrowhead). l: After 1 week of culture, numerous immunopositive cells are seen in the transition zone (arrowhead) and in the outgrowing epithelium (arrow). Shown are representative fields (a–l). Scale bar, 50 μm.

We confirmed that the integrin subunit β8 was expressed on the cell membrane and in the cytoplasm of basally located cells in the normal conducting airways, as earlier reported (Figure 2d) ▶ . 15 The same staining pattern was observed in the intact epithelium on the fragments during the culture period (Figure 2e) ▶ . In the transition zone between the original epithelium and the new outgrowing epithelium, β8 immunolabeling was found throughout the thickness of the epithelium but was more intense in the basally located cells (Figure 2e) ▶ . After 7 days of culture, membrane staining for β8 was seen in all layers of the newly outgrown stratified epithelium without a preference for basally located cells (Figure 2f) ▶ . An anti-β6 Ab that stains formalin-fixed paraffin-embedded tissue was not available to make direct comparisons of β6 and β8 immunostaining.

To characterize the basally located cell type that stained with the β8 Ab, we used an anti-HMW cytokeratin Ab that recognizes basal cells of the normal human bronchial epithelium (Figure 2, g to i) ▶ . 16 HMW cytokeratin staining was found in the cytoplasm of the basally located cells in the intact epithelium confirming, that in the normal human airway, the cells that express β8 are basal cells. The epithelial cells that had migrated out to cover the exposed stroma of the cut surfaces of the fragments were also positive for HMW keratin. Therefore, these cells were also basal cells. This is consistent with other studies reporting that migrating cells initially filling a mechanical wound space are derived from the basal cell population. 31 After 7 days of culture, when the migrating cells had begun to differentiate to form a multilayered new epithelium, only the basally located cells were immunolabeled with the anti-HMW cytokeratin Ab (Figure 2i) ▶ , whereas all layers were immunolabeled with the anti-β8 Ab (Figure 2f) ▶ . This indicates that β8 is expressed in nonbasal cell types during early airway differentiation events whereas in the normal airway it is confined to basal cells.

In the normal human airway there is a low turnover, with fewer than 1% of cells labeled with Abs to the Ki-67 antigen (Figure 2j) ▶ , a nuclear protein expressed in late G1 through M phase. 16,32 The predominant proliferating airway epithelial cell type is the basal cell, suggesting that it functions as a progenitor cell. 16 In cultured fragments, Ki-67-labeled airway basal cells could be easily detected (Figure 2, k and l) ▶ . During the first 2 days of culture, a significantly higher concentration of Ki-67-positive cells (2.5 ± 0.4 cells/HPF, P = 0.02) were located in the immediate vicinity of the transition zone between the original intact epithelium and the outgrowing epithelial cells than in the original intact epithelium (1.3 ± 0.3 cells/HPF) (Figure 2k) ▶ . At this time point, the outgrowing epithelium had little Ki-67 staining (0.6 cells/HPF), indicating that cell migration accounted for the initial outgrowth, not proliferation. By 1 week of culture, a significantly higher concentration of Ki-67-positive cells was found in both the transition zone (7.6 ± 1.5 cells/HPF, P = 0.002) and outgrowing epithelium (3.3 ± 0.8 cells/HPF, P = 0.005) (Figure 2l) ▶ consistent with the repair sequence found in vivo involving flattening of cells, migration, and then proliferation. 33

Bronchial Fragments as a Model to Assess Airway Epithelial Proliferation in a Complex Biological Environment

To examine airway proliferation in intact bronchial fragments, we quantified Ki-67 immunostaining using manual counting. We found that the Ki-67-labeling indices of untreated fragments, at each time point, varied considerably (Figure 3a ▶ , open circles) presumably because individual fragments, although obtained from a single patient, represented a heterogeneous population varying in size and in the amount of intact original epithelium and stroma. Despite the heterogeneity of the fragments, statistically significant differences could be seen in the proliferation indices between the populations of fragments at different time points (Figure 3, a and b ▶ ; open circles, open bars). Thus, significant increases in proliferation were seen at days 4 and 6 of culture (day 2 versus day 4, P < 0.001; day 2 versus day 6, P < 0.01) (Figure 3, a and b ▶ ; open circles, open bars).

Figure 3.

Figure 3.

The integrin αvβ8 inhibits airway epithelial proliferation in bronchial fragments. Airway fragments were cultured in individual wells for 2, 4, and 6 days in the presence or absence of a neutralizing anti-β8 Ab. Cell proliferation was assessed using Ki-67 immunostaining as a surrogate proliferation marker. Ki-67-labeling index was defined as the percentage of Ki-67-positive airway epithelial cells present in three random sections of each fragment. Each circle represents one fragment. a: Time course demonstrating that proliferation is significantly increased at days 4 and 6 of culture. The maximal increase is seen at day 4. The number of proliferating cells is significantly higher in anti-β8-treated (closed circles) relative to nontreated fragments (open circles) at day 4 of culture. The bar indicates the mean value. Each circle represents a single fragment. b: The experiment shown in a is shown in bar graph form to show SE. c: The experiment shown at day 4 in a is compared to two sequential experiments, each performed with fragments obtained from a different patient, using anti-β8 Abs (closed circles) compared to an isotype-matched control Ab (open circles). d: The experiment shown in c is shown in bar graph form to show SE. All three experiments show significant differences in cell proliferation between anti-β8 and control treated groups. *, P < 0.05; **, P < 0.001.

The Integrin αvβ8 Inhibits Epithelial Cell Proliferation during Epithelial Repopulation in Bronchial Fragments

To determine whether αvβ8 affects airway proliferation in complex biological environments, we treated bronchial fragments with a neutralizing anti-β8 Ab (Figure 3, a and b ▶ ; closed circles, filled bars). After 2 days in culture, we found a slight increase in the percentage of labeled airway epithelial cells, when compared to the normal airway in vivo (<1%), but no difference between the Ab treated (4.2%) and control (media alone) groups (5.3%) (Figure 3, a and b) ▶ . In contrast, after 4 days of culture, we found a statistically significant increase in the number of proliferating cells in the anti-β8-treated group (38.5%) compared to the control group (28.0%, P < 0.05) (Figure 3, a and b) ▶ . After 6 days in culture, the rates of proliferation of both groups declined and no significant differences were seen between the groups (23% and 21%). Based on these results, subsequent proliferation studies were done only at day 4 of culture.

To confirm the negative effect of αvβ8 on cell proliferation at day 4 of culture, we tested fragments obtained from two additional patients with neutralizing anti-β8 and isotype-matched control Abs (Figure 3, c and d) ▶ . Again, a significant increase in mean epithelial cell proliferation was seen in the anti-β8 treated groups. No differences were seen between control groups treated with control Abs or medium alone indicating that the control Ab had no effect on proliferation (Figure 3, c and d) ▶ .

TGF-β Inhibits Airway Epithelial Cell Proliferation in Bronchial Fragments

As a first step in determining if the mechanism of αvβ8-mediated growth inhibition in intact bronchial fragments could be mediated through TGF-β, we assessed the presence of the TGF-β isoforms in bronchial fragments using immunolocalization and quantitative PCR strategies. Using Abs to the two isoforms of TGF-β that contain the RGD binding motif, TGF-β1 and TGF-β3, we determined that both TGF-β1 and -β3 were expressed in airway fragments (Figure 4a) ▶ . The TGF-β1 Ab lightly stained the intact and newly formed epithelium, scattered submucosal fibroblasts, macrophages, and endothelium (Figure 4a ▶ , top). The TGF-β3 Ab heavily stained the intact and newly formed epithelium, the submucosal fibroblasts, macrophages, and blood vessels (Figure 4a ▶ , bottom). In the bronchial epithelium of the airways of normal lung samples, the TGF-β1 and -β3 Abs stained in an identical pattern as seen in the intact epithelium of the bronchial fragments (data not shown). In bronchial fragments, the TGF-β3 transcript number, as assessed by real-time PCR, was expressed ∼200-fold in excess of the TGF-β1 transcript (Figure 4b) ▶ . However, the TGF-β2 transcript was the most highly expressed, ∼700-fold in excess of the TGF-β1 transcript (Figure 4b) ▶ .

Figure 4.

Figure 4.

TGF-β-signaling pathways are involved in the inhibition of airway epithelial cell proliferation in the airway fragment model. a: Immunostaining of TGF-β1 and TGF-β3 of bronchial fragments at day 1 of culture. TGF-β3 staining (bottom) is intense compared to TGF-β1 (top). The area between the arrows is the new epithelium. v indicates a vessel. b: Real-time PCR of bronchial fragments for TGF-β1, -β2, and -β3. Shown is the approximate transcript copy number. Bar indicates mean values. Each square or diamond represents a single fragment, n = 8. All fragments were obtained from the same patient. c: Neutralizing anti-TGF-β Abs significantly increase cell proliferation in three sequential experiments. Each experiment represents fragments obtained from a different patient. Each circle represents a single fragment. The bar indicates the mean value. *, P < 0.001. d: Antibodies to β8 block release of active TGF-β by bronchial fragments. Bronchial fragments were treated with or without anti-β8 or anti-TGF-β and co-cultured overnight with mink lung epithelial cells stably expressing firefly luciferase cDNA under the transcriptional control of the TGF-β-responsive, plasminogen-activator inhibitor promoter. 20 Each data point represents four fragments. Data are pooled from fragments obtained from two patients. Shown are relative luminescence units. *, No treatment versus anti-β8 or anti-TGF-β, P < 0.001. Scale bar, 50 μm.

We next determined that active TGF-β played a role in regulating airway epithelial cell proliferation in bronchial fragments. In three consecutive experiments, we found a significant increase in Ki-67 labeling in fragments cultured with anti-TGF-β compared with the control Ab (Figure 4c) ▶ . Although, in two of the three experiments, treatment with the anti-TGF-β Ab increased cell proliferation to a greater extent than treatment with the anti-β8 Ab, the overall effects were similar (Table 2) ▶ .

Table 2.

Summary of Data from Proliferation Studies

Experiment* Fragments (n) Epithelial cells,† mean (SD) Ki-67 labeling index,‡ % (SD) P
Anti-β8 Control Anti-β8 Control Anti-β8 Control
1 20 20 1638 (844) 1916 (1104) 38.5 (15.4) 28.3 (15.1) <0.05
2 23 17 973 (483) 1260 (640) 43.3 (13.3) 32.0 (15.4) <0.05
3 20 17 905 (724) 660 (470) 40.9 (10.9) 19.2 (13.9) <0.001
Anti-TGF-β Control Anti-TGF-β Control Anti-TGF-β Control P
1 17 16 1594 (575) 1715 (1077) 30.1 (13.1) 10.1 (7.1) <0.001
2 18 16 647 (318) 461 (397) 57.6 (9.6) 23.0 (10.6) <0.001
3 18 18 917 (591) 682 (436) 54.7 (11.3) 14.8 (8.8) <0.001

To determine the mechanism of αvβ8-dependent inhibition of airway proliferation, we sought to determine whether αvβ8 and TGF-β might be acting through common pathways. We used bronchial fragments co-cultured with a TGF-β reporter cell line. In this co-culture system, bronchial fragments remained floating above and did not contact or adhere to the reporter cell monolayer. Using this no-contact model, we confirmed that active TGF-β was indeed produced and released by bronchial fragments in culture and that the majority of the active TGF-β produced was αvβ8-dependent (Figure 4d) ▶ . Thus, treatment with β8 Ab inhibited 76% of the TGF-β activation in the co-culture system (Figure 4d) ▶ . Taken together these data strongly suggest that the αvβ8-dependent growth inhibition of airway epithelium in bronchial fragments is because of the αvβ8-dependent release of active TGF-β.

Discussion

The major finding of this study is that integrin-mediated activation of TGF-β inhibits airway epithelial cell proliferation in a model of human airway epithelial repopulation. This finding provides the first direct evidence that an integrin in the human airway can promote homeostasis through the regulation of epithelial cell growth. Furthermore, we have evidence that this regulation is achieved through a novel mechanism, the integrin αvβ8-dependent activation of TGF-β, a cytokine that is critical for airway homeostasis. 2

We established the human airway fragment system to model the human airway in vivo. In this model, biologically relevant airway cellular and stromal components are maintained. Indeed, in bronchial fragments, integrin and TGF-β expression profiles correlate well with the immunohistochemical expression profiles seen in the normal airway in vivo. 15,26,27 Thus, this system overcomes the problems of two-dimensional airway epithelial culture systems where authentic stromal interactions, normal airway epithelial differentiation and normal patterns of integrin expression are lost. In addition, this systembypasses questions of relevance when making comparisons between human and rodent species, which differ significantly in cell-type distribution and airway biology. Furthermore, the human bronchial fragment system is easily manipulated using neutralizing antibodies, suggesting a general application to investigation of human airway biology.

In the bronchial fragment model, the β8 subunit transcript was expressed at levels approaching that of the β1 subunit, a highly promiscuous and abundant subunit in the human airway. 24 Robust αvβ8 immunostaining can be demonstrated in the normal airway epithelium in vivo whereas, in airway epithelial cells in culture, only low levels of β8 can be detected. 15 This suggests that αvβ8 expression is rapidly lost in two-dimensional culture, probably because of loss of the basal cell phenotype. The mRNA level for the β6 integrin subunit and TGF-β1 was low relative to other integrin subunits or TGF-β isoforms. This may be because of the absence of a systemic inflammatory response in bronchial fragments because inflammation may increase both β6 and TGF-β1 message levels. 5,26-28 Therefore, based on integrin and cytokine expression profiles, the bronchial fragment model may be a closer representation of the normal noninflamed airway than the inflamed airway.

The TGF-β1 isoform has been extensively studied in the airway where it is thought to play a role in airway remodeling, airway epithelial differentiation and growth regulation, and tumor suppression. 1,34 In the normal mouse airway, TGF-β1 may be the predominant TGF-β isoform expressed. 35 However, in the human airway there is little quantitative data investigating the relative expression of the various TGF-β isoforms. Interestingly, we found that the TGF-β2 and -β3 isoform transcripts were expressed at levels greater than 100-fold higher than the TGF-β1 isoform in bronchial fragments. Transcript profiling correlated with the relative intensities of immunostaining of the TGF-β1 and -β3 isoforms suggesting that the expression of these isoforms is transcriptionally regulated in bronchial fragments. The TGF-β3 isoform was co-expressed with β8 in the airway parabasal cells, whereas the TGF-β1 isoform was barely detected in the airway epithelium. These data demonstrate that TGF-β2 and TGF-β3 are more highly expressed than TGF-β1 in bronchial fragments. Because the LAP of latent TGF-β2, unlike the LAP of latent TGF-β1 and -β3, does not possess an RGD sequence, latent TGF-β2 is highly unlikely to interact with αv-integrins. 8,9,13 Thus, in this system, the integrin αvβ8 may interact primarily with latent TGF-β3. In support of this, we have used peptides modeled on the RGD binding sites of latent TGF-β1 and TGF-β3 and found that both peptides could effectively inhibit αvβ8-mediated latent TGF-β activation in cultured cell lines (data not shown).

TGF-β2 was the most highly expressed TGF-β isoform transcript expressed in bronchial fragments. The significance of this finding is unclear because very little of the total TGF-β activation in bronchial fragments could be attributed to TGF-β2. This could be an artifact of the bronchial fragment system because latent-TGF-β2 is almost certainly activated in the bronchus in vivo, at least during development. 15 Thus, in bronchial fragments TGF-β2 is likely to be sequestered in a latent state and mechanisms that have been defined for latent-TGF-β2 activation in other models, do not play a major role in the bronchial fragment system. 36

Multiple members of the αv-subfamily of integrins can act as cell surface receptors for the RGD sequence found in latent-TGF-β. 8,9,13 However, only two of the αv-integrins, αvβ8 and αvβ6, have been shown to mediate activation of latent TGF-β. The selective ability among αv-integrins to activate latent TGF-β is not understood but appears to involve αv-integrin-specific differences. Thus, there is evidence for two distinct αv-integrin-specific mechanisms of activation of latent TGF-β, one used exclusively by αvβ8 and one by αvβ6. 8,13 These two mechanisms may have evolved to serve separate functions because αvβ8 appears to be expressed most highly in the normal airway, suggesting a role in epithelial homeostasis, and αvβ6 is expressed most highly in inflamed airways, suggesting a role in epithelial repair. 15,26-28

Integrin αvβ8 has been shown to mediate activation of latent TGF-β through a proteolytic mechanism involving the transmembrane-type metalloprotease, MT1-MMP. 8 In this mechanism, the latent domain of TGF-β is cleaved and active TGF-β is released from the cell surface into the cell culture supernatant. Indeed, consistent with a proteolytic release, we were able to show that active TGF-β was released by bronchial fragments into the culture supernatant and that this release was primarily αvβ8-dependent. Furthermore, we found that in bronchial fragments MT1-MMP transcripts were highly expressed because the levels approached or exceeded those of the abundant MMP-2 and MMP-9 (data not shown).

We demonstrated that αvβ8-dependent activation of latent TGF-β accounted for the majority of active TGF-β released by bronchial fragments. Thus, treatment of bronchial fragments with a neutralizing anti-β8 Ab blocked 76% of the total active TGF-β released. The small amount of active TGF-β released that was not blocked by β8 Abs may be because of lower efficacy of the anti-β8 Ab when compared to the pan-TGF-β Ab. Alternatively, there may be other mechanisms of latent TGF-β activation in bronchial fragments. 8 For instance, we have not ruled out a contribution of the integrin αvβ6 to latent TGF-β activation, although αvβ6 is not highly expressed in either bronchial fragments or normal airway. 26-28 Finally, there may be some nonintegrin-mediated activation of the latent TGF-β2 isoform by bronchial fragments in culture. Interestingly, latent TGF-β2 can also be activated through a metalloproteolytic mechanism. 36

To study integrin function in the airway in intact human tissues, we developed a human model of airway proliferation that is faithful to the epithelial-stromal interactions found in vivo. The bronchial fragment model is technically easy to establish and to manipulate and the use of intact human tissues offers the opportunity to study airway cell behavior in a complex environment allowing direct extrapolation to human biology and pathology. In summary, we have demonstrated that the integrin αvβ8 regulates airway epithelial cell proliferation through the activation of TGF-β using a human model of the airway. This model provides quantitative data in a system that closely approximates the human airway in vivo.

Acknowledgments

We thank Vivian Weinberg (USCF Biostatistics) and Ida Welle (UC Berkeley Department of Epidemiology) for help with statistical analysis, Dean Sheppard and Peter Ten Dijke for reading of the manuscript, Amha Atakilit for reagents, and Walter Finkbeiner for helpful suggestions.

Footnotes

Address reprint requests to Stephen Nishimura, M.D., Department of Pathology, Bldg. 3, Rm. 207, San Francisco General Hospital, 1001 Potrero Ave., San Francisco, CA 94110. E-mail: cdog@itsa.ucsf.edu.

Supported by grants from the National Institutes of Health (grants HL70622 and HL63993), the American Heart Association, the Hellman Family, the University of California at San Francisco academic senate (to S. L. N.), the Norwegian Cancer Society, and The Unger-Vetlesen Medical Fund (to L. F.).

References