Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Asthma is a complex disease with increasing worldwide incidence and significant morbidity and mortality. The disease is characterized by exaggerated narrowing of the conducting airways of the lung in response to bronchoconstrictor stimuli (airway hyperresponsiveness). Morphologic changes in the airways of patients with asthma include eosinophilic inflammation, mucus metaplasia (1), subepithelial fibrosis (1, 2), and smooth muscle hypertrophy (3). T lymphocytes in the airways of patients with asthma are principally of the T helper (Th)2 type and produce the Th2 cytokines interleukin (IL)-4, IL-5, and IL-13 (4, 5). The genes encoding these cytokines are located on human chromosome region 5q25-31, a region mapped by linkage analysis to asthma in several genomewide screens (6). Many of the phenotypic features of asthma can be reproduced by antigen sensitization and inhalational challenge in mice. Recently, we and others have shown that inhibition of IL-13 blocks antigen-induced effects on the airways and that administration of IL-13 to the airways of mice can itself induce mucus metaplasia, eosinophilic inflammation, and airway hyperresponsiveness (7, 8). These effects of IL-13 do not require the participation of T cells or B cells because they are retained in recombinase-activating gene-1 null mice (7), suggesting that they are largely dependent on the effects of IL-13 on resident airway cells. However, the contributions of various airway cell types to the in vivo effects of IL-13 are largely unknown. IL-13, like the closely related cytokine IL-4, binds to receptors composed of the common IL-4 receptor (IL-4R)  subunit and one of two IL-13 subunits (1 or 2). Current evidence suggests that only receptors containing the IL-13 receptor (IL-13R) 1 subunit are capable of inducing subsequent signals. Ligation of the IL-13R1/IL-4R receptor complex induces tyrosine phosphorylation of a member of the signal transduction and transactivation (STAT) family, STAT6, which is essential for many of the known biologic functions of IL-13 and IL-4. Although other signaling pathways, including insulin receptor substrate (IRS)-1 and IRS-2 (9), can be activated by IL-13, they have not been shown to contribute to IL-13-mediated development of allergic asthma. Mice lacking STAT6 (10) and mice lacking STAT6 activation by IL-13, such as mice treated with an IL-13R antagonist (7, 8), are protected from many of the phenotypic features of allergic asthma, suggesting that this pathway is critical to the contribution of IL-13 to asthma. Once phosphorylated, STAT6 forms a homodimer and translocates to the nucleus, where it directly binds to regulatory DNA sequences and modulates expression of IL-13 responsive genes. Because IL-13 principally affects cell behavior by altering gene expression, we initiated exploration of the molecular mechanisms underlying the contributions of IL-13 to asthma by examining the global effects of IL-13 on gene expression in the three major resident cell types in the airway wall: epithelial cells, smooth muscle cells, and fibroblasts.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture and Reagents

Normal human bronchial epithelial cells (NHBE), bronchial smooth muscle cells (BSMC), and normal human lung fibroblasts (NHLF) were purchased from Clonetics (San Diego, CA) and grown at 37°C in small airways growth medium, smooth muscle growth medium, and fibroblast growth medium, respectively, as recommended. All cells used for each cell type were obtained from a single donor. Rabbit polyclonal anti-STAT6 antibody was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal antiphosphotyrosine antibody PY20 was obtained from Transduction Laboratories (Lexington, KY). Recombinant human IL-13 was obtained from R&D Systems (Minneapolis, MN). Leupeptin and aprotinin were obtained from Calbiochem-Novabiochem Corp. (La Jolla, CA) and phenylmethylsulfonyl fluoride (PMSF) and pepstatin were obtained from Sigma Chemical Co. (St. Louis, MO).

Western Blot Analysis

Cells were grown to confluence in 100-mm tissue culture plates, placed in basal medium overnight, and treated with recombinant IL-13 (100 ng/ml) or phosphate-buffered saline (PBS) for 20 min. Total cell extracts were obtained using radioimmunoprecipitation assay buffer (0.1% sodium dodecyl sulfate, 1% sodium deoxycholic acid, 1% Nonidet P40, 1 mM vanadate, 0.5 mM molybdate) in the presence of 10 µg/ml pepstatin, 10 µg/ml leupeptin, 5 µg/ml aprotinin, and 1 mM PMSF. Protein concentrations were determined by the BCA method (Pierce, Rockford, IL).

To detect tyrosine phosphorylation of STAT6, aliquots of lysates containing equal protein concentrations were precleared with protein G-sepharose, incubated with anti-STAT6 rabbit polyclonal antibody for 3 h at 4°C, and immune complexes were captured by protein G-sepharose for 90 min at 4°C. The beads were washed, and immunoprecipitated proteins were solubilized in 5× reducing Laemmli sample buffer. Samples were separated by SDS-polyacrylamide gel electrophoresis (PAGE) on 10% polyacrylamide gel and transferred to Immobilon membranes for Western blotting with antiphosphotyrosine antibody PY20. Membranes were blocked in 3% bovine serum albumin and incubated with primary antibody for 1 h at room temperature followed by secondary antibody incubation with antimouse horseradish peroxidase (HRP) and developed with enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ).

RNA Extraction

Cells were grown to confluence in growth medium and placed in basal medium overnight before stimulation with recombinant IL-13 (100 ng/ml) or PBS. After 6 h, total RNA was isolated using Trizol Reagent (GIBCO BRL, Grand Island, NY). RNA quantity was determined by optical density measurement at 260 nm and quality was determined by RNAse negative 1% agarose gel.

Preparation of Labeled Cellular RNA

A total of 10 to 15 µg of total RNA was used for double-stranded complementary DNA (cDNA) synthesis. Double-stranded cDNA was generated with a cDNA synthesis kit (Superscript cDNA Synthesis System; Life Technologies, Gaithersburg, MD) using an oligo(dT)₂₄ primer containing a T7 RNA polymerase promoter site at the 3' end (Genset, La Jolla, CA). The cDNA was extracted with phenol/chloroform, ethanol precipitated, and used as a template for in vitro transcription with biotin-labeled nucleotides (BioArray HighYield RNA Transcript Labeling Kit; Enzo Diagnostics, Farmingdale, NY). A total of 12.5 µg of the labeled cellular RNA (cRNA) was fragmented at 94°C for 35 min in fragmentation buffer (40 mM Tris-acetate, pH 8.1/100 mM potassium acetate, 30 mM magnesium acetate), and a hybridization mix was generated by addition of herring sperm DNA (0.1 mg/ml; Sigma), sodium chloride (1 M), Tris-acetate (10 mM), and Tween-20 (0.0001%). A mixture of four control bacterial and phage cRNA (1.5 pM BioB, 5 pM BioC, 25 pM BioD, and 100 pM Cre) was included to serve as an internal control for hybridization efficiency.

Hybridization of Microarrays

Aliquots of each sample (10 µg cRNA in 200 µl hybridization mix) were hybridized to a Genechip HuGene FL array (Affymetrix). After hybridization, each array was washed, stained with streptavidin phycoerythrin (Molecular Probes, Eugene, OR), washed again, hybridized with biotin-labeled antistreptavidin phycoerythrin antibodies (Vector Laboratories, Burlingame, CA), restained with streptavidin phycoerythrin, scanned (GeneArray scanner G2500A; Hewlett Packard, Palo Alto, CA), and washed according to procedures developed by the manufacturer (Affymetrix). Samples obtained from four individual dishes of each cell type were included for each condition examined.

Analysis of Genechip Data

Scanned output files were visually inspected for hybridization artifacts. Arrays lacking significant artifacts were analyzed using Genechip 3.3 software (Affymetrix). Arrays were scaled to an average intensity of 150 per gene and analyzed independently. The expression value for each gene was determined by calculating the average of differences (perfect match intensity minus mismatch intensity) of the probe pairs in use for that gene.

The expression analysis files created by Genechip 3.3 software were then transferred to a database (Microsoft Access; Microsoft, Redmond, WA) linked to Internet genome databases (e.g., National Heart, Lung and Blood Institute, Swiss Prot, and GeneCards) to update gene definitions. Fold changes were determined by dividing the average difference of each sample by the mean of the average differences of the PBS samples. Because meaningfulness of low and negative intensity readings are unclear, a value of 20 was assigned to all measurements lower than 20 to facilitate calculations. Genes that were either increased at least 1.5-fold or decreased at least by 50% from baseline in three or four samples were included in subsequent analyses. For further data analysis and data presentation, we used Spotfire Pro 4.0 (Spotfire, Somerville, MA). For cluster analysis we used Gene Cluster and Treeview programs described by Michael Eisen and colleagues (11).

Monocyte Chemotactic Protein-1 and IL-6 Enzyme-Linked Immunosorbent Assay

Monocyte chemotactic protein (MCP)-1 enzyme-linked immunosorbent assay (ELISA) was performed using monoclonal antihuman MCP-1 antibody and biotinylated antihuman MCP-1 antibody as described by the manufacturer (R&D Systems) with the exception of the detection method. For detection, 40 µl of 2 µg/ml alkaline-phosphatase-conjugated streptavidin (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) was added and incubated for 1 h. The plate was washed five times with 0.05% Tween-20 in PBS, and incubated with 100 µl per well 1 mg/ml p-nitrophenyl phosphate (Sigma) for 5 min. The reaction was stopped by adding 30 µl 0.5 M NaOH, and optical densities were measured at 405 nm. Test conditions were established using recombinant human MCP-1 (R&D Systems) as a standard and positive control. MCP-1 concentrations in the samples were obtained by using a standard curve of diluted recombinant human MCP-1 (R&D Systems).

The IL-6 concentration in conditioned media was measured using ELISA (IL-6 optEIA kit, no. 2645 KI; Pharmacia, Piscataway, NJ), as described by the manufacturer's protocol, except for the use of a chemiluminescent developing substrate (Pierce, Pico, Rockville, IL). The 96-well plates (Fluoronunc Module Masisorp Surface; Nalge NUNC, Rochester, NY) were coated overnight with the antihuman IL-6 monoclonal capture antibody (Pharmacia) and blocked with 10% fetal bovine serum-PBS for 1 h at room temperature. Undiluted samples were incubated in the plates for 2 h on a shaker. Biotinylated antihuman IL-6 detection antibody conjugated with HRP-streptavidin (Pharmacia) was incubated in the plates for 1 h on a shaker. Plates were developed using a chemiluminescent substrate (Pierce, Pico) and measured after 15 min using a luminometer (Victor2 1420 Multilabel Counter; Perkin-Elmer, Wallac, Boston, MA). IL-6 concentrations were determined from a standard curve of diluted recombinant human IL-6 (Pharmacia).

TaqMan Amplification

Aliquots of RNA used for oligo array experiments were used for the preparation of cDNA for TaqMan amplification. Gene-specific primers were used in multiplex reverse transcriptase/polymerase chain reaction (RT-PCR) to generate gene-specific cDNAs. The "Primer Express" package from Perkin Elmer (Foster City, CA) was used to design both RT-PCR and TaqMan primers/probes. The sequences for the primers/probes can be found on the following website http://owl.ucsfmedicalcenter.org. TaqMan primer-probe sets were designed to span exon-intron junctions whenever possible. Optimization of multiplex PCR was done as described (12), using KlenTaq DNA polymerase (cDNA Advantage Mix; Clontech, Palo Alto, CA). The reaction was heated at 94°C for 2 min followed by zero to 25 cycles at 94°C for 30 s, 55°C for 15 s, and 70°C for 45 s. The number of PCR cycles was kept as low as possible to enable efficient gene quantification. Gene-specific multiplex RT-PCR was performed several times with different numbers of PCR cycles, and all data generated were highly reproducible in the range of zero to 25 cycles. For each RNA sample, dilution curves were generated to confirm linear dependence of the signal on the concentration of template RNAs. Cycle thresholds were determined for each reaction, normalized to the cycle threshold for the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and expressed as relative copy number by comparison to a standard curve. Mean values were calculated from at least three separate samples for each condition analyzed and expressed as fold difference from mean values for the same gene after treatment with PBS.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

IL-13 Induces STAT6 Phosphorylation in NHBE, BSMC, and NHLF

To determine whether IL-13 is capable of activating the same signaling pathway in NHBE, BSMC, and NHLF, we assayed tyrosine phosphorylation of STAT6 by immunoprecipitation followed by Western blotting with antiphosphotyrosine antibody PY20. As shown in Figure 1A, a band of the appropriate molecular mass to be STAT6 was phosphorylated in response to IL-13 in all three cell types. Similar amounts of STAT6 were present and immunoprecipitated in equal quantities of cell lysates (Figure 1B).

View larger version (52K): [in this window] [in a new window]   Figure 1.   (A) Phosphotyrosine immunoblot of cell lysates of NHLF, NHBE, and BSMC incubated with or without IL-13 for 20 min. Lysates were immunoprecipitated with STAT6 antibody and probed with antiphosphotyrosine antibody. (B) STAT6 immunoblot of the filter used in Figure 1A.

IL-13 Induces Different Patterns of Gene Expression in Primary Airway Cells

To determine whether IL-13 had similar effects on gene expression in each cell type and to identify the major genes induced and inhibited, we evaluated the global pattern of gene expression after 6 h of stimulation with IL-13 with Affymetrix genechips containing probe sets to detect 6,500 human genes. Cluster analysis viewed using the Treeview program demonstrated different patterns of gene expression in each cell type tested (Figure 2). In NHBE, NHLF, and BSMC, 80, 79, and 204 genes, respectively, were induced by at least twofold in at least three of the four samples. Of these genes, only two were common between airway epithelial cells and airway smooth muscle (TRAIL, lymphoma proprotein convertase), none were common between airway epithelial cells and fibroblasts, three were common between airway smooth muscle cells and fibroblasts (vitamin D receptor, placental bikunin, SS-A/Ro ribonucleoprotein autoantigen), and none were induced in all three cell types. Another outstanding feature of this cluster analysis was the large number of genes that were affected by IL-13 in airway smooth muscle cells, a likely target for the induction of airway hyperresponsiveness.

View larger version (63K): [in this window] [in a new window]   Figure 2.   Cluster analysis of expression of ~ 1,200 genes with a gene expression ratio of > 1.5 or < 0.5 (in at least n-1 samples) to baseline after a 6-h incubation with IL-13. Each row represents one gene and each column represents one IL-13-stimulated cell sample. For each sample, expression values after IL-13 treatment are compared with the average gene expression of unstimulated (PBS control) samples. Green represents transcript levels below baseline, black represents transcript levels equal to baseline, and red represents transcript levels above baseline. Lines to the right of the cluster analysis represent clusters of genes induced by IL-13 in individual cell types and lines to the left represent clusters of genes inhibited by IL-13. (A) Genes induced in airway epithelial cells. (B) Genes induced in NHLF. (C) Genes inhibited in BSMC. (D) Genes induced in BSMC.

Because of the high cost of this approach, we limited this experiment to analysis of a single time point. At this time point, changes in gene expression reflect both genes that are directly modulated by IL-13-induced signals (e.g., through STAT6) and genes that are modulated indirectly through IL-13-induced effects on genes whose products can themselves modulate subsequent gene expression (e.g., transcription factors). Further evidence for the divergence of effects of IL-13 in each of the cell types examined comes from analysis of the specific transcription factors induced by IL-13. As shown in Table 1, IL-13 induced at least a twofold increase in several transcription factors in each cell type, but there was no overlap between cell types.

Expression data were sorted for genes with > 1.5-fold induction over the baseline in at least three of four treated dishes, and cell-specific lists of genes induced or inhibited by IL-13 are presented in Tables 2 to 5. Induced genes were further evaluated by searching existing databases to assign as many as possible to functional groups.

IL-13 Induces Expression of Several Signaling Effectors, Signaling Receptors, Contractile Proteins, and Ion Channels in BSMC

Among the most dramatic effects of IL-13 on BSMC was induction of a number of signaling molecules and signaling receptors that could prime these cells for enhanced responses to a wide array of other stimuli. Notable among these were several components of mitogen-activated protein (MAP) kinase signaling pathways (e.g., jnk1b2, jnk2, Mek5c, and MAPkAPK2), phospholipase A2 and diacylglycerol kinase . Enhanced activity of these pathways could contribute to airway smooth muscle cell proliferation and/ or hypertrophy, and the increase in airway smooth muscle mass that characterizes asthma. Among the most highly induced were signaling molecules of the Src family, fgr and the CXC chemokine receptor CXCR2, which are of particular interest because neither of these proteins has been previously identified on smooth muscle cells. IL-13 also increased expression of the signaling IL-13R1 subunit, an effect that could result in a positive feedback loop that would amplify and prolong responses to IL-13 in these cells.

IL-13 also induced expression of several contractile proteins (sarcolipin, dystroglycan-associated protein, smooth muscle myosin heavy chain, and cardiac  myosin heavy chain) and ion channels (e.g., KCNQ2, KVLQT1 and CLCL3), effects that could alter the contractile response of these cells and thereby contribute to airway hyperresponsiveness. Finally, IL-13 induced expression of secreted factors in airway smooth muscle cells, such as the basic fibroblast growth factor (bFGF) and the IL-6 family cytokine leukemia inhibitory factor (LIF), which could contribute to the asthma phenotype through autocrine effects or paracrine effects on other airway cells.

IL-13 Induces Expression of Different Secreted Proteins, Signaling Molecules, and Receptors in NHLF

As noted in Table 2, distinct members of some of the same functional classes of genes were induced by IL-13 in NHLF. Most noteworthy was induction of secreted cytokines and growth factors. These include MCP-1, a chemokine that has been shown to induce collagen production from fibroblasts and contribute to models of asthma and tissue fibrosis (15), and the growth factor IL-6, whose overexpression has been shown to induce subepithelial fibrosis in transgenic mice (personal communication, Dr. Jack Elias, Yale University). To confirm that these effects of IL-13 might be biologically significant, we measured the secretion of MCP-1 and IL-6 by IL-13-stimulated NHLF with an ELISA (see Figure 3). As shown in Figure 4, IL-13 induced a more than twofold increase in secretion of each of these gene products. IL-13 also increased expression of the transcript for potent angiogenic factor angiopoietin 1, which could contribute to the angiogenesis seen in asthmatic airways, and a platelet-derived growth factor isoform that would be expected to induce proliferation of smooth muscle cells and fibroblasts.

View larger version (10K): [in this window] [in a new window]   Figure 3.   Results of ELISA from conditioned media of IL-13-stimulated lung fibroblasts for MCP-1 (A) or IL-6 (B). The cells were treated with PBS (control) or IL-13 at 100 ng/ml for 24 or 48 h. Error bars represent standard error of the mean for values from three dishes analyzed at each time point.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Comparison of fold induction in selected genes in lung fibroblasts as measured by the TaqMan method or the oligonucleotide array method.

To confirm that genechip analysis was accurately measuring RNA abundance, we used the same input RNA used for genechip analysis from treated and untreated fibroblasts to determine RNA abundance with a different technique, TaqMan quantitative PCR (Figure 4). We chose three genes for this analysis (IL-6, MCP-1, and the vitamin D receptor) that were induced more than twofold by IL-13, two genes (the chloride channel, CLCN3, and serum response factor) that were induced between 1.5- and twofold, and nine genes that were expressed but unaffected by IL-13. Although there were quantitative differences in the determined fold induction, especially for the three most highly induced genes, both methods demonstrated increased expression of the same five genes and essentially no change in expression of the other nine genes.

IL-13 Induces Expression of Secreted Extracellular Matrix Proteins, Proteases, and Protease Inhibitors in NHBE

The most prominent groups of genes induced by IL-13 on NHBE were genes whose products are involved in production and turnover of the extracellular matrix (Table 2A). IL-13 also increased expression of a small number of signaling receptors and effectors in these cells, but this effect was not as prominent as that seen in airway smooth muscle cells or fibroblasts. As with induction of angiopoietin-1 in fibroblasts, induction of the gene for the angiogenic agonist placental growth factor could contribute to airway mucosal angiogenesis in asthma. In each cell type, IL-13 also inhibited expression of a large number of genes (Tables 3B, 4B, and 5B). Analysis of the functional significance of these genes is also likely to be informative and is getting under way.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The results of this study demonstrate that a single cytokine, IL-13, using a common signaling pathway that activates a common transcription factor (STAT6) induces different patterns of gene expression in three different primary cell types, with virtually no overlap in the genes that is induced or inhibited in each cell type. The resulting gene expression profile suggests a coordinate and distinct contribution to asthma pathogenesis by each of the cell types examined.

As noted previously, IL-13 has been identified as a critical mediator of allergic airway responses in mice and has been suggested to be important in human asthma. In mice, IL-13 has been shown to cause airway hyperresponsiveness, mucus metaplasia, eosinophilic inflammation, and subepithelial fibrosis (7, 8, 18), which are all features of the human disease (2, 3, 19). The molecular mechanisms underlying these responses are largely unknown. We thus sought to use information gained by global analysis of gene expression to generate a hypothesis about the cellular targets for IL-13 in the airways and the molecular pathways by which IL-13 might affect the function of these target cells. In that regard, the numerous genes that were affected by IL-13 in airway smooth muscle cells was especially informative and suggests that smooth muscle is likely an important target for IL-13 in the airways. Induction of signaling proteins, including several components of MAP kinase cascades, and of contractile proteins and ion channels, suggests that IL-13 could coordinately enhance the contractile (and perhaps, proliferative) response of airway smooth muscle. Among the contractile proteins induced was dystrophin-associated glycoprotein (DAG), a glycoprotein known to be decreased in disorders associated with muscle weakness, such as Duchenne's muscular dystrophy, with a suspected role in muscle anchoring (20). More recently, DAG has also been found to be increased at sites of muscle injury and muscle regeneration, and to contribute to regeneration of tensile strength of the muscle (21). Another induced gene, sarcolipin, is a modulator of Ca²⁺-adenosine triphosphatase (SERCA) type pumps, which are known to have a role in regulating the contraction of fast-twitch skeletal muscle (22).

The large number of signaling receptors and effectors induced by IL-13 in these cells also suggests that bronchial smooth muscle could be "primed" by IL-13 to enhance responsiveness to a variety of other contractile or proliferative signals. One of the most prominently induced genes in this category was fgr, a Src family tyrosine kinase previously thought to be restricted to mature cells of myeloid lineage. Fgr has been shown to contribute to cell motility in eosinophils (23), neutrophils (24), and macrophages (25), and to adhesion-dependent degranulation of neutrophils (26), and is thought to play an important role in cytoskeletal reorganization in these cells. Fgr has not been previously described in smooth muscle cells. By analogy to roles played by other Src isoforms, fgr is a candidate that may contribute to smooth muscle cell proliferation in response to IL-13. In addition to effects on individual smooth muscle cells, IL-13 increased the expression of genes, encoding a number of secreted proteins in BSMC. Some of these, such as bFGF and LIF, could affect other cells, such as fibroblasts, in the airway wall and thereby contribute to the subepithelial fibrosis induced by IL-13.

Given the well-established role of IL-13 in inducing fibrosis in the airways and in models of parasitic disease in other organs, we expected to identify dramatic induction of extracellular matrix components such as collagen isoforms in NHLF. Surprisingly, although IL-13 did increase expression of several matrix proteins, this effect was principally seen in NHBE, not in fibroblasts. However, two of the most highly induced genes in fibroblasts encoded the secreted proteins IL-6 and MCP-1, and we were able to confirm that IL-13 increased the secretion of each of these proteins by NHLF. These findings are noteworthy because both IL-6 and MCP-1 have been implicated in models of tissue fibrosis. Overexpression of IL-6 and the related cytokine IL-11 under the control of the airway epithelial CC10 promoter causes subepithelial fibrosis in mice (27). MCP-1 has been shown to induce collagen production from fibroblasts, an effect that appears to involve induction of the potent profibrotic cytokine transforming growth factor (TGF)-. Recently, studies in mice lacking CCR2, a principal receptor for MCP-1, have also suggested a role for this chemokine in induction of airway hyperresponsiveness in a model of allergic asthma (28).

Several of the genes whose transcriptions were affected by IL-13 were somewhat unexpected. T-cell receptor (TCR) and Myf4 were both decreased in NHBE, and cardiac  myosin heavy chain was induced in BSMC by IL-13 stimulation. These genes were previously thought to be restricted to other cell types. At this time, the significance of the presence of these genes in the cells studied is unclear. Whether the presence of these transcripts translate to the presence of meaningful protein levels has not yet been determined and will require further experiments.

Clearly, analysis of global patterns of gene expression, as described here, is a powerful approach to identify transcriptional programs that could contribute to disease pathogenesis. However, this approach can only identify promising candidates and suggest hypotheses that must ultimately be tested in in vivo models. There are also important limitations to our experimental approach that are worth noting. In the current study, we used primary cultures of airway cells. Although this approach has the advantage of allowing us to ascribe any changes identified to events that result only from the effects of IL-13 on that cell type, it is likely that transcriptional responses of cells in culture differ considerably from the responses of the same cell in vivo. Because we have shown in this report that cellular differentiation can have enormous effects on these responses, it will ultimately be essential to confirm that genes of interest are also expressed in the relevant cell type in vivo in response to IL-13. Another important limitation to RNA-based analysis of gene expression as described here is the fact that expression of many proteins is regulated at steps downstream of messenger RNA synthesis. We have begun to address this issue by confirming that at least some of the most interesting changes in gene expression we describe (e.g., induction of IL-6 and MCP-1 in NHLF) are also associated with similar changes in expression of the encoded proteins.

In this set of experiments, we analyzed the effects of a single cytokine (IL-13) on three functionally diverse but anatomically adjacent cell types. Despite initiation of an identical signaling pathway (STAT6), IL-13 induced highly distinct transcriptional programs in each of the three cell types. Furthermore, in each cell type, IL-13 induced genes that encode for proteins that may play a significant role in the pathogenesis of chronic asthma, suggesting that the asthma phenotype is likely the result of coordinated effects of IL-13 on the three airway cell types studied. This hypothesis will require specific confirmation and testing in in vivo models. However, the results from our study should facilitate the design and completion of such studies.