Insulin receptor substrate-1/2 mediates IL-4-induced migration of human airway epithelial cells

Abstract

Migration of airway epithelial cells (AEC) is an integral component of airway mucosal repair after injury. The inflammatory cytokine IL-4, abundant in chronic inflammatory airways diseases such as asthma, stimulates overproduction of mucins and secretion of chemokines from AEC; these actions enhance persistent airway inflammation. The effect of IL-4 on AEC migration and repair after injury, however, is not known. We examined migration in primary human AEC differentiated in air-liquid interface culture for 3 wk. Wounds were created by mechanical abrasion and followed to closure using digital microscopy. Concurrent treatment with IL-4 up to 10 ng/ml accelerated migration significantly in fully differentiated AEC. As expected, IL-4 treatment induced phosphorylation of the IL-4 receptor-associated protein STAT (signal transducer and activator of transcription)6, a transcription factor known to mediate several IL-4-induced AEC responses. Expressing a dominant negative STAT6 cDNA delivered by lentivirus infection, however, failed to block IL-4-stimulated migration. In contrast, decreasing expression of either insulin receptor substrate (IRS)-1 or IRS-2 using a silencing hairpin RNA blocked IL-4-stimulated AEC migration completely. These data demonstrate that IL-4 can accelerate migration of differentiated AEC after injury. This reparative response does not require STAT6 activation, but rather requires IRS-1 and/or IRS-2.

the airway epithelium is a target of inflammatory and physical insults that occur in several airways diseases including asthma (2, 25). Repair of the airway mucosa after injury requires that epithelial cells near the wound edge then shift their phenotype and migrate into the wound region (12, 13). Once the wound region is covered by new cells, the epithelial cell phenotype shifts again into more differentiated cells (22). Each step in the repair process may be mediated by both constitutive and inflammatory factors.

T lymphocytes are found in asthmatic airways (7) and infiltrate into the airway mucosa (5); fluid recovered by bronchoalveolar lavage from asthmatic airways is enriched in IL-4, IL-5, IL-13, and granulocyte/macrophage colony-stimulating factor (GM-CSF) but not IFNγ (37), indicating the presence of Th2 subclass CD4+ lymphocytes. Activation of Th2 lymphocytes may stimulate the epithelium to secrete these cytokines and chemokines (4, 16) and growth factors such as transforming growth factor-β (TGF-β) (37, 45). Whether repair of the airway epithelium also may be modulated by infiltrating lymphocytes, however, is not clear.

IL-4, a significant inflammatory mediator in asthmatic airways, enhances production of chemokines such as eotaxin (15), IL-8, and RANTES in the BEAS-2B (15) and 9HTEo- (35) airway epithelial cell (AEC) lines and in primary AEC collected from both normal and asthmatic subjects (27). In addition to its ability to stimulate chemokine release, some data suggest that IL-4 can mediate repair functions in epithelium. Treatment of monolayers of the Calu-3 human lung epithelial cell line with either IL-4 or IL-13 decreases barrier function, as assessed by mannitol flux measurements and decreased expression of the tight junction protein ZO-1 (1). One regulatory component of focal adhesion complexes, paxillin, which serves as an anchor point for actin fibers (44), was decreased dramatically after treatment of primary, human AEC with either IL-4 or IL-13; this in turn decreased cell adhesion to matrix (36). Similar changes in barrier function and adhesion are a necessary step in the initial stages of cell migration (29).

IL-4 binds two distinct receptor complexes: the combination of IL-4Rα and the common γ-chain (γc) of the IL-2 receptor (type I receptor), or that of IL-4Rα with the IL-13Rα1 subunit (type II receptor) (26, 39). After activation, the cytoplasmic tails of these subunits associate with tyrosine kinases of the Janus family (JAK1–3 and the related TYK2) (32, 34). IL-4Rα associates with JAK1 (34) or JAK2 (38), γc with JAK3, and IL-13Rα1 with either JAK2 or the related TYK2 (38). Dimerization of the receptor chains enhances JAK activity and leads to phosphorylation of tyrosine residues in the cytoplasmic domain of IL-4Rα. These residues then act as docking sites for signaling molecules containing Src homology 2 (SH2) domains (23), such as STAT6, one of the family of signal transducers and activators of transcription (STAT). After phosphorylation, STAT6 homodimerizes and translocates to the nucleus to act as a transcriptional activator.

Another pathway that mediates IL-4 signaling is via either insulin receptor substrate (IRS)-1 or IRS-2. These ∼170-kDa proteins have significant homology and are phosphorylated at a substantial number of serine and tyrosine residues in response to insulin (41), and both contain multiple functional domains for protein-protein interactions and intracellular signal transduction. The importance of IRS-1 to IL-4-mediated signaling was first demonstrated in the 32D myeloid cell line, which lacks endogenous IRS-1; treatment with IL-4 after reconstituting IRS-1 elicited proliferation (46, 47). IL-4Rα contains a motif in the I4R region that is highly homologous to sequences within insulin and insulin growth factor-1 receptors known to bind IRS-1 (21, 50) and IRS-2 (47). Mutating the Y497 residue on the IL-4Rα subunit prevents IL-4-stimulated phosphorylation of IRS-1 (21); whereas the same mutation does not block activation of IRS-2, mutation of P488 on IL-4Rα greatly diminishes tyrosine phosphorylation (46). Among the downstream effectors of IRS-1 and -2 is the p85 subunit of phosphatidylinositol-3 kinase (PI3K) and the adapter protein Grb-2, which associates with SOS, an activator of Ras (49). GTP-bound Ras activates the Ser/Thr kinase Raf, which in turn activates several pathways in the stress-activated protein kinase (SAPK) pathway, including p38 MAPK via the PAK family and MKK3/6 (11), Erk 1/2 (p42/p44 MAPK) (9, 30), and c-Jun kinase (JNK) (32). Each of these pathways regulates migration of AEC (51), suggesting that IL-4-stimulated migration could be regulated via these pathways.

IL-4 can induce production of cell components necessary for migration, such as the tissue inhibitor of metalloproteinase-2 in dermal fibroblasts (18). However, whether IL-4 can modulate AEC spreading or migration has not been examined, and the specific pathways by which IL-4 stimulates migration are not clear. To examine this issue, we evaluated whether IL-4 could elicit migration of airway epithelial cells using two-dimensional repair models of primary, differentiated cells. Our data demonstrate that IL-4 is a potent motogen for airway epithelial cells via a requirement for IRS-1/2 but not STAT6. These results suggest that IL-4 may have positive effects on airway epithelial cell repair after injury.

MATERIALS AND METHODS

Materials.

Medium A consisted of BEBM (CC-3171) and a SingleQuot kit (CC-4171), both purchased from Lonza (Walkersville, MD). Medium B consisted of medium A plus 1.25 μg/ml amphotericin, 100 μg/ml ceftazidime, 80 μg/ml tobramycin, 100 μg/ml vancomycin, 100 μ/ml penicillin, 100 μg/ml streptomycin, 100 μ/ml nystatin, and 50 μg/ml gentamicin. DMEM consisted of DMEM (Cellgro 10-017 CM) supplemented with 10% FCS, 100 g/ml streptomycin, and 100 U/ml penicillin G. Air-liquid interface (ALI) medium consisted of 1:1 BEBM:DMEM. Every 500 ml were supplemented with a SingleQuot kit, 130 μg/ml bovine pituitary extract, 25 ng/ml EGF, 50 nM retinoic acid, 0.5 mg/ml low-endotoxin BSA, and 20 U/ml nystatin. Recombinant IL-4 was purchased from R&D Systems (Minneapolis, MN). Antibodies directed against total and phosphorylated STAT6 were purchased from Cell Signaling (Danvers, MA). Antibodies directed against IRS-1 and IRS-2 were purchased from Upstate (Billerica, MA). Transfection reagent (TransIT-LT1) for the production of lentivirus (LV) was purchased from Mirus, and pCMV6-XL-5-STAT6 was purchased from OriGene (Rockville, MD). All other reagents were obtained from Sigma (St. Louis, MO).

Cell culture.

The use of primary human airway epithelial cells was approved by the University of Chicago Institutional Review Board. Primary cells were obtained from two sources. Cells were purchased from Lonza. These were grown per instructions on collagen IV-coated containers in medium A in 5% CO₂ atmosphere at 37°C.

Epithelial cells also were collected from lungs provided by the Regional Organ Bank of Illinois (Elmhurst, IL). These lungs had been collected for use in lung transplantation but were later discarded. Information provided by the organ bank declared that the subjects donating the lungs did not have lung disease (9 lungs) or had chronic asthma (4 lungs). For the latter set of lungs, no clinical information as to disease severity, previous treatment, or recent exacerbation could be obtained. Mucosal membranes from central airways were dissected and incubated in 1% protease at 37°C for 2 h. Epithelial cells were then removed by pipetting gently up and down several times and then repeating this operation four to five times with fresh medium A. All the cell suspensions were pooled together and centrifuged at ∼460 g for 3 min at 4°C. The cells were grown on collagen IV-coated containers in medium B for 1–3 days in 5% CO₂ atmosphere at 37°C. Then, the strength of the antibiotics and antimycotics in medium B was reduced progressively every day to 80, 50, 25, and 0%.

Cells at 70% confluence were replated onto collagen IV-coated 12-mm Transwell membranes in submersion culture for 5–7 days in ALI medium to 80% confluence. Cells were then changed to ALI culture conditions for an additional 21 days, changing the medium every other day on the basal side only (24). By 21 days, cells in ALI culture displayed characteristics of differentiated epithelium, and numerous ciliated cells could be identified on phase-contrast microscopy.

The cell line 16HBE14o^- was obtained from Dieter Gruenert (California Pacific Medical Center, San Francisco, CA). These were grown in DMEM and were used before passage 30.

Wound repair assay.

We have previously published details of this method (40, 51). Briefly, confluent monolayers were washed twice and placed in serum-free or defined medium appropriate for the cells being studied. Mediators or control diluent (DMSO) were added as appropriate. In all experiments, the concentration of DMSO was <0.1% in culture wells. Linear wounds of 0.7- to 1.3-mm width were made with a rubber stylette. 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 significant matrix removal were discarded. Wound closure was measured serially for 24 h starting immediately after wound creation. Wounds were photographed edge to edge, and images were assembled using Photoshop (Adobe). Measurements were made using ImageJ software (Wayne Rasband, National Institutes of Health, Bethesda, MD). A 200 × 200 pixel grid was overlaid onto the panel. Approximately 15–20 perpendicular lines were drawn starting at the top edge of the wound where a grid intersected, and ending at the bottom edge; the mean length represented average closure. Values were normalized to time 0 values. The coefficient of variation was <20% in each panel. Intra-operator variance of measurement was <1%, and inter-operator variance was <3%, for wounds of 1-mm width.

Western blot analysis.

Confluent monolayers were wounded using a cell rake adapted from a metal lice comb (51). Monolayers were pretreated with inhibitor or sham 15 min before wounding and afterwards incubated at 37°C for 5 min to 6 h. Immunoblotting was performed as described (51). Membranes were rehybridized with an antibody for actin when appropriate to control for differences in protein loading. In some experiments, densitometry was done using ImageJ software.

Extraction of nuclear proteins.

Nuclear proteins were collected at the conclusion of experiments using a Nuclear Extract kit (Active Motif, Carlsbad, CA). Kit directions were followed. Nuclear protein lysates were then resolved for the presence of STAT6 by Western blot.

Generation of a dominant negative STAT6 cDNA.

A truncated STAT6 cDNA was generated starting with pCMV6-XL-5-STAT6 (acc. no. NM 003153, OriGene). To generate a truncated STAT6, PCR was done using forward primer 5′-GGGAGGTGCTAGCAGGGCCAGCCT-3′ corresponding to nt 143–166 and containing an NheI site (underlined), and reverse primer 5′-TAGTGTACACCTAACCCCTGCCATCCTT-3′ corresponding to nt 2177–2204 with the necessary changes to get at stop codon (TAG) at amino acid 641 and a BsrGI site (underlined) for cloning. The PCR fragment was purified and subcloned into pGEM-T Easy Vector (Promega), and the ligation reaction was used to transform DH-5α cells (Invitrogen), from which colonies were selected. The resulting cDNA, 640 amino acids in length, was then ligated into LV.

Generation of silencing hairpin RNA for IRS-1 and IRS-2.

Silencing hairpin RNA (shRNA) were generated based on a previously published target sequence: for IRS-1, 5′-GTCAGTCTGTCGTCCAGTA-3′ (8) or a nonspecific (NS) shRNA, 5′-ACAAGACCTAAGTGCACTG-3′ (Karim Bouzakri, Karolinska Institutet, Stockholm, Sweden, personal communication); for IRS-2, 5′-CAACAACAACAACCACAGC-3′ (IRS-2A) (3), 5′-GGTGACGCTGCAGCTTATGA-3′ (IRS-2B) (43), or the same NS shRNA used for IRS-1 (K. Bouzakri, personal communication).

Transduction and expression of either shRNA or cDNA using lentiviral infection.

Biohazard Hazard Safety Level 2 containment practices were strictly observed at all times for the generation of LV. Constructs containing the shRNA of interest were ligated into the lentiviral vector pLentilox (pLL) 3.7 using appropriate restriction enzymes, with confirmation of orientation by sequencing. The lentiviral vector has a GFP encoded downstream of the shRNA driven by a separate pCMV promoter; demonstration of GFP expression indicates stable transduction. Constructs containing the cDNA of interest were also ligated into pLL3.7 using appropriate restriction enzymes; in this process, the GFP was removed, and so a separate pLL3.7 expressing GFP only was generated for use.

Lentiviral vectors were cotransfected with psPAX2 (generous gift of Didier Trono, Swiss Institute of Technology, Lausanne, Switzerland) and envelope protein vector pHCMV-G into a HEK-293T cell line. Cells were transfected in 11 ml of DMEM containing 20 μg of total DNA and 40 μl of Mirus transfection reagent. Cells were incubated at 33°C in 5% CO₂ for 48 h, after which supernatants containing virus were passed through 0.45-μm filters and then centrifuged at 27,000 g for 6 h. The LV pellet was resuspended in 4–6 ml of DMEM, aliquoted, and frozen until use. Expression and toxicity of all LV supernatants were determined by serial dilution using the 16HBE14o- cell line; the dilution combining highest expression of GFP as determined by fluorescence microscopy and least toxicity as measured by cell lifting was chosen for infection of primary cells.

Primary cells were infected at ∼50% confluence while in submersion culture in medium A. For experiments to insert an shRNA, LV containing vectors for the shRNA of interest, a NS shRNA, or an empty vector (LV pLL3.7-null) were used. For experiments to insert a cDNA, LV containing vectors for the cDNA of interest was used in a 5:1 ratio with pLL3.7-GFP containing a cDNA for GFP only to monitor infection efficiency. Controls were done using pLL3.7-GFP only. In both sets of experiments, cells were infected with an appropriate dilution of LV suspended in 5 ml of medium A and 8 μg/ml polybrene for 5 h at 33°C. Cells were then washed and incubated in fresh medium; at ∼70% confluence, cells were transferred to Transwell membranes and grown in ALI medium. Cultures in ALI were examined daily for GFP expression; cultures that failed to demonstrate appropriate expression in ≥80% of cells were discarded. Additional cells in each experiment were sham-infected with LV; these cells were otherwise treated the same as infected cultures and examined side-by-side with those cultures as an additional control. In these sham-infected cells, no GFP expression could be demonstrated.

Data analysis.

Wound repair data are referenced to time 0 wound width for each wound and are expressed as means ± SE. To avoid confounding problems with multiple analyses along the time-response curve, differences were analyzed at 24 h. Real-time RT-PCR data are expressed as comparative cycle time ratios using 18S rRNA as an internal standard. Differences were examined by analysis of variance; when significant differences were found, post hoc analysis was done using Fisher's protected least significant differences test. Differences were considered significant when P < 0.05.

RESULTS

IL-4 elicits migration of primary, differentiated AEC in ALI culture.

AEC grown in ALI culture for 3 wk were treated with 0.1–10 ng/ml IL-4, or sham diluent, immediately after wounding, and repair followed using digital photomicroscopy for 24 h. In these experiments, starting width for all wounds was 0.86 ± 0.03 mm (n = 33 wounds). Treatment with IL-4 accelerated migration significantly over 24 h (Fig. 1A). In AEC obtained from four lungs of donors with chronic asthma, cells grown in ALI culture were treated immediately after wounding with 1 or 10 ng/ml IL-4, or sham diluent, immediately after wounding, and followed for 24 h. As with cells collected from normal lungs, treatment with IL-4 accelerated migration significantly over 24 h (Fig. 1B). Control wounds in asthmatic cells did not close more quickly than control wounds in normal cells.

Fig. 1.

Migration of airway epithelial cells (AEC) in response to IL-4. Cells were injured using a rubber stylette, and migration was followed for 24 h by digital photomicroscopy. A: migration of primary AEC grown under air-liquid interface (ALI) culture conditions for 3 wk. N = 6–10 experiments in each group collected from 4 normal donors. Statistical significance of difference was analyzed by 1-way ANOVA (*P = 0.01 vs. control). Data are expressed as means ± SE. B: migration of differentiated AEC collected from asthmatic lungs in response to IL-4. Cells were grown under ALI culture conditions for 3 wk and then injured using a rubber stylette. Migration was followed for 24 h by digital photomicroscopy. N = 14–16 experiments in each group collected from 4 asthmatic donors. Statistical significance of difference was analyzed by 1-way ANOVA (*P = 0.02 vs. control at 24 h). Data are expressed as means ± SE.

Role of STAT6 following injury and treatment with IL-4.

We first examined whether STAT6 was phosphorylated in response to treatment with IL-4 after injury in differentiated, primary AEC grown in ALI conditions for 3 wk. Protein lysates were collected before or up to 2 h after injury and resolved by Western blot. Treatment with 10 ng/ml IL-4 elicited prompt phosphorylation of STAT6 (Fig. 2A); injury alone did not elicit discernible STAT6 phosphorylation. Treatment with 10 ng/ml IL-13 also elicited STAT6 phosphorylation, whereas treatment with either 10 ng/ml IL-1β or 20 ng/ml TNFα did not, as expected (Fig. 2B).

Fig. 2.

A: phosphorylation of STAT6 in AEC after injury and treatment with IL-4. Cells grown in ALI culture for 3 wk were injured and treated immediately with 10 ng/ml IL-4 or control vehicle only. Cell lysates collected over 2 h were resolved by Western blot for total and phospho-STAT6. Image is representative of 3 experiments. B: phosphorylation of STAT6 in AEC grown in ALI culture for 3 wk after injury and treatment with 10 ng/ml IL-4, 10 ng/ml IL-13, 10 ng/ml IL-1β, or 20 ng/ml TNFα. Cell lysates 30 min after injury and treatment were resolved by Western blot for total and phospho-STAT6. Image is representative of 3 experiments.

IL-4 signaling via STAT6 may be opposed by other cytokines: for example, pretreatment of human AEC with IFNγ blocks IL-4Rα/IL-13Rα1-mediated phosphorylation of STAT6 and increases mRNA expression of both SOCS-1 and SOCS-3, thereby reducing IL-4-stimulated production of eotaxin-3 (16). However, one recent study (52) demonstrated an opposite result: prolonged pretreatment with IFNγ augmented IL-4-stimulated production of eotaxin-3 in BEAS-2B cells by upregulating expression of IL-4Rα. We asked whether concurrent treatment with IFNγ would block IL-4 accelerated migration and whether this would be mediated via changes in STAT6 phosphorylation. In these experiments, differentiated, primary AEC grown in ALI conditions for 3 wk were treated with sham diluent, 10 ng/ml IL-4, 100 ng/ml IFNγ, or both, immediately after injury. Both cytokines accelerated migration, but the combination was no more effective than either cytokine alone (Fig. 3A). We then examined whether concurrent treatment with IFNγ altered IL-4-stimulated phosphorylation of STAT6. Cells were treated with 10 ng/ml IL-4 alone or in combination with 1–100 ng/ml IFNγ for 30 min, after which protein lysates were collected and resolved to examine STAT6 by Western blot. Concurrent treatment with IFNγ did not decrease IL-4-induced STAT6 phosphorylation significantly (Fig. 3B). These data strongly suggested that IFNγ neither modulated STAT6 phosphorylation nor migration induced by IL-4.

Fig. 3.

A: migration of differentiated AEC in response to IL-4: effect of IFNγ. Cells were grown under ALI culture conditions for 3 wk and injured using a rubber stylette, and migration was followed for 24 h by digital photomicroscopy. N = 5–8 experiments in each group collected from 2 normal donors. Statistical significance of difference was analyzed by 1-way ANOVA (*P = 0.05; †P = 0.02 vs. control). Data are expressed as means ± SE. B: STAT6 phosphorylation after treatment with 10 ng/ml IL-4 alone or combined with 1–100 ng/ml IFNγ. Cells in ALI culture × 3 wk were treated as noted for 30 min. Image shown is representative of 2 experiments.

If IL-4-stimulated migration requires phosphorylation and activation of STAT6, then inhibition of STAT6 should block migration. To test this, we expressed a cDNA encoding a dominant negative, truncated STAT6 using a lentiviral vector. The STAT6-tr generated at 640 amino acids in length was missing a critical tyrosine at amino acid 641 (31, 33, 48), important to activation and dimerization, and a region in the COOH terminus important to transcriptional activation (28). Lentiviral infection used as expression in a high proportion of cells was necessary to ensure an effect in AEC migration. GFP expression from the LV construct was demonstrated as early as 48 h after infection in submersion culture and was expressed in ≥80% of all cells 3 wk after conversion to ALI culture (Fig. 4A). In primary AEC infected with LV to express STAT6-tr cDNA while in submersion culture and subsequently grown in ALI conditions for 3 wk, significant expression of STAT6-tr was demonstrated by Western blot (Fig. 4B) compared with cells infected with an empty control LV (EV control) or sham-infected cells. In the cells infected to express STAT6-tr, treatment with 10 ng/ml IL-4 for 6 h did not elicit an increased abundance of native STAT6 in nuclear protein extracts, whereas similar treatment in either sham-infected cells or EV control cells did increase the abundance of STAT6 in nuclear protein extracts (Fig. 4C). This suggested that the STAT6-tr interfered with STAT6 signaling as expected.

Fig. 4.

Blocking IL-4-accelerated migration of differentiated AEC by expression of a truncated, dominant negative STAT6 cDNA. A: expression of GFP as a marker for successful infection with lentivirus (LV) containing cDNA. Undifferentiated AEC were infected with either LV pLL3.7-STAT6-tr + LV pLL3.7-GFP control in 5:1 ratio (STAT6-tr), LV pLL3.7-GFP control alone (EV control), or were sham-infected. GFP fluorescence was examined 3 wk after conversion to ALI culture. Bar, 20 μm. B: abundance of STAT6 and STAT6-tr protein after infection as in A. Protein lysates collected from cells in ALI culture × 3 wk were resolved for total STAT6 by Western blot. Lane loading was confirmed by rehybridization with an antibody directed against actin. The antibody used recognizes both STAT6 and STAT6-tr. C: abundance of STAT6 and STAT6-tr protein in cell nuclei after infection as in A. Cells in ALI culture × 3 wk after infection were treated with 10 ng/ml IL-4 or control diluent for 6 h, after which nuclear proteins were collected and resolved for total STAT6 by Western blot. Increased abundance of STAT6 is seen in nuclear protein extracts after IL-4 treatment in both sham-infected and EV control cells but not in cells infected with pLL3.7-STAT6-tr. D: migration of AEC after infection with LV pLL3.7-STAT6-tr-cDNA. Cells were grown in ALI culture × 3 wk after infection and then injured using a rubber stylette. Cells were then treated with 10 ng/ml IL-4 or sham diluent, and migration was followed for 24 h by digital photomicroscopy. N = 8 experiments in each group. Statistical significance of difference was analyzed by 2-tailed t-test (*P = 0.01 vs. control at 24 h). Data are expressed as means ± SE. E: migration of AEC after infection with LV pLL3.7-GFP control (EV control). Cells were grown in ALI culture × 3 wk after infection and then injured using a rubber stylette. Cells were then treated with 10 ng/ml IL-4 or sham diluent, and migration was followed for 24 h by digital photomicroscopy. N = 7–8 experiments in each group. Statistical significance of difference was analyzed by 2-tailed t-test (*P = 0.01 vs. control at 24 h). Data are expressed as means ± SE.

Subsequent treatment with 10 ng/ml IL-4 immediately after injury, however, accelerated cell migration over 24 h to a similar extent in both sets of cells. Treatment with IL-4 accelerated migration significantly in cells infected to express STAT6-tr over 24 h (Fig. 4D). Likewise, IL-4 accelerated migration in EV control cells (Fig. 4E). There was no significant difference in closure rate between the controls for each set of experiments. IL-4-stimulated migration was also similar in sham-infected cells as for EV control cells (data not shown).

Role of IRS-1/IRS-2 following injury and treatment with IL-4.

We first examined IRS-1 and IRS-2 abundance in primary AEC grown in ALI conditions for 3 wk. Both IRS-1 and IRS-2 were demonstrated in protein lysates collected from AEC in ALI culture × 3 wk using appropriate antibodies (data not shown). Both IRS-1 and IRS-2 have multiple serine and tyrosine phosphorylation sites, some of which are inhibitory and some stimulatory (41, 42); we judged it impractical to review each phosphorylation site, and combinations thereof, to ascertain IRS-1 and IRS-2 activation. Therefore we decided to test directly the role of IRS-1 and IRS-2 in AEC migration. We first examined the role of IRS-1 by expressing a shRNA targeting IRS-1 using a LV vector. Expression of the LV vector could be demonstrated within 48 h after infection and persisted through 3 wk of ALI culture (Fig. 5A). Expression of IRS-1 protein was reduced by ∼90% in these 3-wk cells compared with cells infected with NS shRNA based on analysis of a Western blot normalized for loading (Fig. 5B). Expression of IRS-2 in the same cells was not affected by silencing IRS-1 (Fig. 5C).

Fig. 5.

Blocking IL-4-accelerated migration of differentiated AEC by expression of an shRNA directed against insulin receptor substrate (IRS)-1. A: expression of GFP as a marker for successful infection with LV containing shRNA. Undifferentiated AEC were infected with either LV pLL3.7-shRNA-IRS-1 or LV pLL3.7-shRNA-NS control, or were sham-infected. GFP fluorescence was examined 3 wk after conversion to ALI culture. Bar, 20 μm. B: knockdown of IRS-1 protein after expression of an shRNA directed against IRS-1. Protein lysates collected from cells in ALI culture × 3 wk were resolved for total IRS-1 by Western blot. Lane loading was confirmed by rehybridization with an antibody directed against actin. C: abundance of IRS-2 protein in protein lysates collected from same groups as in B, as resolved by Western blot. D: migration of AEC infected with LV pLL3.7-shRNA-IRS-1. Cells were grown in ALI culture × 3 wk after infection and then injured using a rubber stylette. Cells were treated with 10 ng/ml IL-4 or sham diluent, and migration was followed for 24 h by digital photomicroscopy. N = 10–11 experiments in each group. Data are expressed as means ± SE. E: migration of AEC infected with LV pLL3.7-shRNA-NS control. Cells were grown in ALI culture × 3 wk after infection and then injured using a rubber stylette. Cells were then treated with 10 ng/ml IL-4 or sham diluent, and migration was followed for 24 h by digital photomicroscopy. N = 7 experiments in each group. Statistical significance of difference was analyzed by 2-tailed t-test (*P = 0.02 vs. control at 24 h). Data are expressed as means ± SE.

Silencing IRS-1 also abrogated the migration response to IL-4 in AEC grown in ALI culture conditions. Treatment with 10 ng/ml IL-4 did not accelerate migration significantly in cells infected with LV pLL3.7-shRNA-IRS-1 after 24 h (Fig. 5D). In contrast, IL-4 treatment did accelerate wound closure in cells infected with LV pLL3.7-shRNA-NS control (Fig. 5E). IL-4-stimulated migration was similar in sham-infected cells as for cells infected with LV pLL3.7-shRNA-NS control (data not shown). These data demonstrated clearly that IRS-1 mediated migration stimulated by IL-4 in differentiated AEC.

We also examined the role of IRS-2 in AEC migration by expressing shRNA targeting IRS-2 using a LV vector. Expression of GFP from the LV vector could be demonstrated within 48 h after infection and persisted through 3 wk of ALI culture (Fig. 6A). Expression of IRS-2 protein in these 3-wk cells was reduced by ∼80% compared with cells infected with NS shRNA based on analysis of a Western blot normalized for loading (Fig. 6B). Expression of IRS-1 in the same cells was affected minimally by silencing IRS-2 (Fig. 6C).

Fig. 6.

Blocking IL-4-accelerated migration of differentiated AEC by expression of an shRNA directed against IRS-2. A: expression of GFP as a marker for successful infection with LV containing shRNA. Undifferentiated AEC were infected with either LV pLL3.7-shRNA-IRS-2 or LV pLL3.7-shRNA-NS control, or were sham-infected. GFP fluorescence was examined 3 wk after conversion to ALI culture. Bar, 20 μm. B: knockdown of IRS-2 protein after expression of an shRNA directed against IRS-2. Protein lysates collected from cells in ALI culture × 3 wk were resolved for total IRS-2 by Western blot. Lane loading was confirmed by rehybridization with an antibody directed against actin. C: abundance of IRS-1 protein in protein lysates collected from same groups as in B, as resolved by Western blot. D: migration of AEC infected with LV pLL3.7-shRNA-IRS-2B. Cells were grown in ALI culture × 3 wk after infection and then injured using a rubber stylette. Cells were treated with 10 ng/ml IL-4 or sham diluent, and migration was followed for 24 h by digital photomicroscopy. N = 8 experiments in each group. Data are expressed as means ± SE. E: migration of AEC infected with LV pLL3.7-shRNA-NS control. Cells were grown in ALI culture × 3 wk after infection and then injured using a rubber stylette. Cells then were treated with 10 ng/ml IL-4 or sham diluent, and migration was followed for 24 h by digital photomicroscopy. N = 6 experiments in each group. Statistical significance of difference was analyzed by 2-tailed t-test (*P = 0.005 vs. control at 24 h). Data are expressed as means ± SE.

Silencing IRS-2 also abrogated the migration response to IL-4 in AEC grown in ALI culture conditions. Two different shRNA targeting IRS-2, designated IRS-2A and IRS-2B, were used, and treatment with 10 ng/ml IL-4 did not accelerate migration significantly in cells infected with the IRS-2B shRNA (Fig. 6D). Similar results were obtained in cells infected with the IRS-2A shRNA (data not shown). With the IRS-2B shRNA infection, baseline migration (not IL-4-treated) was moderately slowed compared with infection with the LV pLL3.7-shRNA-NS control (Fig. 6E), whereas IL-4 treatment of cells infected with IRS-2B or IRS-2A shRNA did not stimulate migration. In the cells infected with the NS control, IL-4 treatment accelerated wound closure (Fig. 6E). IL-4-stimulated migration was similar in sham-infected cells as for cells infected with LV pLL3.7-shRNA-NS control (data not shown). These data demonstrated clearly that IRS-2, like IRS-1, mediated IL-4-stimulated migration in differentiated AEC.

DISCUSSION

Repair of the airway mucosa after injury requires the integration of multiple external signals from growth factors, cytokines, and chemokines, as well as physical stimuli. These integrated signals prompt migration of cells into the sites of injury and subsequent phenotype shifting of cells into required subtypes. Epithelial injury is a frequent finding in asthma, even when mild (2, 25), and correlates with airway hyperresponsiveness (20). The inability to repair epithelial defects may be due to continued inflammation alone or to its combination with specific impairments in migration and differentiation. While the effects of multiple growth factors such as epidermal growth factor, keratinocyte growth factor, and insulin growth factors as stimulants to AEC proliferation and migration are established, there has been little consideration as to whether proinflammatory cytokines and mediators affect epithelial repair after injury.

Epithelial cells stimulated with IL-4 express several cytokines and chemokines that trigger and maintain airway inflammation. IL-4 also elicits changes in epithelial structure, including downregulation of tight junction proteins and diminished barrier capacity (1). However, its role as a repair factor for AEC after injury, whether stimulatory or inhibitory, has not been demonstrated previously. We show that IL-4 accelerates migration of differentiated AEC in ALI culture. To our knowledge, our data represent the first demonstration of the promigratory (“motogenic”) effect of IL-4 in airway epithelium. We further demonstrate that while IL-4 elicits prompt phosphorylation of STAT6, STAT6 is not a required mediator of IL-4-stimulated AEC migration. We also show the presence of and the requirement for both IRS-1 and IRS-2 in IL-4-stimulated AEC migration. These data, together, are the first demonstration, to our knowledge, of a role for IRS-1 and IRS-2 in cell migration.

IL-4 activation of STAT6 leads to upregulation of several cytokines and chemokines (15, 16, 35). In our study, stimulation of the IL-4 receptor activated STAT6 maximally in differentiated AEC within 30–60 min, a time range similar to that reported in other studies of airway epithelium (16). Blocking STAT6 by the expression of a truncated STAT6 did not significantly alter IL-4-stimulated cell migration. Furthermore, IL-4 signaling via STAT6 may be opposed by other cytokines. For example, pretreatment of human AEC with IFNγ blocks IL-4Rα/IL-13Rα1-mediated phosphorylation of STAT6 and increases mRNA expression of both SOCS-1 and SOCS-3, thereby reducing IL-4-stimulated production of eotaxin-3 (16). However, one recent study (52) demonstrated an opposite result: prolonged pretreatment with IFNγ augmented IL-4-stimulated production of eotaxin-3 in BEAS-2B cells by upregulating expression of IL-4Rα. In our study, treatment with IFNγ enhanced migration but did not alter STAT6 phosphorylation in differentiated AEC. Furthermore, expression of a truncated STAT6 did not block IL-4-stimulated migration. Our data suggest a separation of responses to IL-4, such that signaling of repair and migration, in contrast to other functions, may not require phosphorylation of STAT6.

Silencing either IRS-1 or IRS-2 blocked IL-4-stimulated migration completely. Downstream activation from either IRS-1 or IRS-2 leads to activation of PI3K to Ras and subsequently to activation of SAPKs (10, 30, 32). These then can initiate migration. Silencing IRS-2 vs. IRS-1 may have differential effects in some models. For example, silencing IRS-2 in L6 myotubes leads to greater reduction in Erk and p38 MAPK than seen with silencing IRS-1, whereas silencing IRS-1 led to more marked reductions in glucose uptake, GLUT4 translocation, and insulin-induced actin remodeling (17). Differences in IRS-1- and IRS-2-mediated effects also have been noted in primary skeletal muscle myoblasts (3). In our model, silencing either IRS protein led to equal and complete abolishment of IL-4-stimulated migration, and it is not clear whether each activates PI3K or whether one activates an alternative pathway. This will require clarification in future studies.

The overall role of Th2-associated cytokines in inflammation is clearly context-specific: the effector mechanisms by which Th2-associated cytokines defend the host appropriately (e.g., against infection) are the same that cause inappropriate inflammation when induced inappropriately or in an exaggerated fashion. Stimulation of AEC by IL-4 elicits production of several cytokines and chemokines (15, 27, 35) that contribute to airway inflammation in asthma. Our data suggest that IL-4 may have similar beneficial effects as well in airway epithelium. The balance of this effect and that of other cytokines in airways may then determine whether the net effect of a given cytokine, at a given moment and stage of disease, is reparative, inflammatory, or both.

One potential limitation of this study is the use of lentiviral infection to introduce shRNA to cells in submersion culture, with subsequent generation of differentiated cells over time. While expression of the shRNA for either IRS-1 or IRS-2 was likely constant throughout the 3 wk of ALI culture as demonstrated by reduction in the protein of interest by Western blot at the end of that time, prolonged knockdown of a signaling mediator may have deleterious effects on cell morphology. We detected no changes in cell morphology by light microscopy after introduction of shRNA to knockdown either IRS-1 or IRS-2. Baseline migration without addition of IL-4 in cells expressing the shRNA was not significantly different than cells expressing a nonsense shRNA within each series of experiments. Another potential limitation is that shRNA technology may activate the interferon pathway (6) or have significant off-target effects (19). shRNA delivered via LV may also have some cytotoxic effects (14). We minimized the potential for confounding issues in these experiments through the use of NS, control shRNA in each infection experiment.

In summary, we demonstrate that IL-4 stimulates early steps in migration of differentiated, normal, and asthmatic AEC after mechanical injury. IL-4-stimulated migration requires signaling from both IRS-1 and IRS-2 but not from STAT6. Our data suggest that IL-4 has both proinflammatory and proreparative roles in airway epithelium and may stimulate repair through pathways, such as IRS-1 and IRS-2, that have not previously been appreciated as modulators of cell migration.

GRANTS

This work was supported by National Institutes of Health Grants AI-056352, HL-080417, and HL-073132 and the State of North Carolina.