Hierarchical control of interleukin 13 (IL-13) signals in lung fibroblasts by STAT6 and SOX11

Interleukin (IL)-13 is a signature cytokine of type 2 inflammation important for the pathogenesis of various diseases, including allergic diseases. Signal transducer and activator of transcription (STAT) 6 is a critical transcriptional factor for the IL-13 signals; however, it remains unknown how expression of the IL-13–induced genes is differentiated by the transcriptional machineries. In this study, we identified IL-13–induced transcriptional factors in lung fibroblasts using DNA microarrays in which SOX11 was included. Knockdown of SOX11 down-regulated expression of periostin and CCL26, both of which are known to be downstream molecules of IL-13, whereas enforced expression of SOX11 together with IL-13 stimulation enhanced expression of periostin. Moreover, we found that in DNA microarrays combining IL-13 induction and SOX11 knockdown there exist both SOX11-dependent and -independent molecules in IL-13–inducible molecules. In the former, many inflammation-related and fibrosis-related molecules, including periostin and CCL26, are involved. These results suggest that SOX11 acts as a trans-acting transcriptional factor downstream of STAT6 and that in lung fibroblasts the IL-13 signals are hierarchically controlled by STAT6 and SOX11.

Interleukin (IL)-13 is a signature cytokine of type 2 inflammation important for the pathogenesis of various diseases, including allergic diseases. Signal transducer and activator of transcription (STAT) 6 is a critical transcriptional factor for the IL-13 signals; however, it remains unknown how expression of the IL-13-induced genes is differentiated by the transcriptional machineries. In this study, we identified IL-13-induced transcriptional factors in lung fibroblasts using DNA microarrays in which SOX11 was included. Knockdown of SOX11 down-regulated expression of periostin and CCL26, both of which are known to be downstream molecules of IL-13, whereas enforced expression of SOX11 together with IL-13 stimulation enhanced expression of periostin. Moreover, we found that in DNA microarrays combining IL-13 induction and SOX11 knockdown there exist both SOX11-dependent and -independent molecules in IL-13-inducible molecules. In the former, many inflammation-related and fibrosis-related molecules, including periostin and CCL26, are involved. These results suggest that SOX11 acts as a trans-acting transcriptional factor downstream of STAT6 and that in lung fibroblasts the IL-13 signals are hierarchically controlled by STAT6 and SOX11.
IL-13 is a pleiotropic cytokine acting on various kinds of cells by binding its receptor, which is composed of the IL-13 receptor ␣1 chain and the IL-4 receptor ␣ chain on the cell surface of B cells, T cells, mast cells, macrophages, epithelial cells, fibroblasts, smooth muscle cells, and endothelial cells, and exerting its pathophysiological roles (10). It is of note that the actions of IL-13 on fibroblasts are important for accelerating inflammation and generating fibrosis. IL-13 induces expression of various chemokines, recruiting inflammatory cells, including CCL26/eotaxin-3 and profibrotic molecules such as tenascin-C and periostin, and activating fibrosis-related molecules such as transforming growth factor-␤ and several matrix metalloproteins (1,11).
Signal transducer and activator of transcription (STAT) 6 is a critical transcriptional factor for IL-13 signaling (12,13). After it is phosphorylated and activated by Janus kinases following the ligation of IL-13 to its receptor, it recognizes the sequence motif TTCN 3-4 GAA, regulating the transcriptional activities of the IL-13-targeted genes. However, it remains unknown how expression of the IL-13-induced genes is differentiated by the transcriptional machineries.
Periostin is a matricellular protein downstream of the IL-13 signals (14). We found that expression of periostin requires STAT6 in model mice because periostin induction is diminished in STAT6-deficient mice (15) and that periostin is highly expressed in the basement membrane in asthma patients. Crucially, it is positively correlated with poor prognosis of these patients (16,17). We then showed that periostin acts as a proinflammatory molecule by activating NF-B, a transcriptional factor important for inflammation, in keratinocytes or fibroblasts in a paracrine or autocrine manner (15,18). Moreover, because periostin acts as a surrogate marker for IL-13 reflecting type 2 inflammation (14), measurement of serum periostin has been applied to predicting the efficacy of IL-13 antagonists (19,20). However, the precise mechanism of periostin expression by IL-13 has remained elusive.
In the SOX (SRY-related HMG box) family, SOX11 is a member of the SOX C group as well as SOX4 and SOX12 (21). The members of the SOX C group have DNA-binding HMG domains close to the N terminus and transcriptional activation domains in the C-terminal regions. SOX11 is important for organ development in embryos; SOX11-deficient mice immediately die after birth, probably due to heart defects. These mice manifest craniofacial and skeletal malformations; asplenia; and hypoplasia of the lung, stomach, and pancreas, suggesting widespread roles of SOX11 in tissues or organ development (22). Additionally, SOX11 plays a critical role in neural differentiation in both embryos and adults (23)(24)(25). Moreover, it has been reported that SOX11 is expressed in various malignancies, including mantle cell lymphoma, suggesting the oncogenic activity of SOX11 in certain types of malignancies (26). However, there is no report showing that SOX11 exerts immunomodulatory effects.
In this study, we identified IL-13-induced transcriptional factors in lung fibroblasts using DNA microarrays in which SOX11 was included. We then showed that SOX11 acts as a trans-acting transcriptional factor for periostin. Moreover, we found that in DNA microarrays combining IL-13 induction and SOX11 knockdown there exist both SOX11-dependent and -independent molecules in IL-13-inducible molecules. The former involves many inflammation-related and fibrosisrelated molecules. These results suggest that SOX11 acts as a trans-acting transcriptional factor downstream of STAT6 and that in lung fibroblasts the IL-13 signals are hierarchically controlled by STAT6 and SOX11.

Involvement of a trans-acting transcriptional mechanism through STAT6 in IL-13-mediated expression in lung fibroblasts
It had already been established that STAT6 plays a critical role in IL-13 signaling (12,13). We first examined whether STAT6 is required for IL-13-mediated induction of the POSTN, CCL26, and SOCS1 genes, known to encode downstream molecules of IL-13 (15,27,28). IL-13 induced transient tyrosine phosphorylation of STAT6 in MRC-5 cells, a human lung fibroblast cell line (Fig. 1A). Knockdown of STAT6 significantly decreased expression of all of POSTN, CCL26, and SOCS1 genes (Fig. 1B). Moreover, secretion of periostin protein in IL-13-stimulated MRC-5 cells was mostly diminished by knockdown of STAT6 or treatment with a STAT6 inhibitor, AS1517499 (Fig. 1, C and D). These results suggest that induction of periostin, CCL26, and SOCS1 by IL-13 in lung fibroblasts requires STAT6.
We next examined the kinetics of periostin expression in MRC-5 cells. In contrast to phosphorylation of STAT6, expression of periostin at both the mRNA and protein levels was prolonged and gradually increased after IL-13 stimulation up to 72 h (Fig. 2, A and B). Moreover, expression of SOCS1 and CCL26 at the mRNA level was also prolonged and sustained up to 72 h. This raised the possibility that in lung fibroblasts IL-13-mediated expression requires a trans-acting transcriptional mechanism through STAT6. To address this, we examined the effects of cycloheximide, an inhibitor against de novo protein synthesis, on IL-13-dependent expression of the POSTN, CCL26, and SOCS1 genes. Cycloheximide significantly decreased the expression of all of these IL-13-inducible genes (Fig. 2C), although there was still induction of these genes by IL-13. These results suggest that in lung fibroblasts a transacting transcriptional mechanism through STAT6 is involved in the IL-13-mediated expression of periostin, CCL26, and SOCS1.

The cis-acting transcriptional mechanism of STAT6 for IL-13-induced periostin expression
Treatment with cycloheximide significantly, but not completely, diminished IL-13-mediated expression of periostin as shown in Fig. 2C, which suggested that, through STAT6, there might exist a cis-acting transcriptional mechanism in addition to a trans-acting mechanism. To explore this possibility, we prepared a series of plasmids encoding promoter regions in the Ϫ5114 to ϩ1 bp site relative to the transcription start site in the POSTN gene. We then applied them to transient transfection and luciferase assay using HEK293T cells. Because HEK293T cells lack STAT6, we cotransfected the plasmid encoding each promoter region together with the plasmid encoding human STAT6. These regions contain one TTCN 4 GAA sequence (N 4 GAS motif), a typical STAT6-binding site, and four TTCN 3 GAA sequences (N 3 GAS motifs 1-4), atypical STAT6binding sites (Fig. 3A). When we narrowed down the promoter region from Ϫ835 to Ϫ97 bp, luciferase activity decreased significantly, suggesting that this region contains an IL-13responsive element (Fig. 3B). Because this same region contains N 3 GAS motif 1 (Ϫ134/Ϫ126), we examined whether this motif was required for IL-13 responsiveness by introducing a mutation. Introduction of a mutation into N 3 GAS motif 1 (Ϫ134/Ϫ126), but not into the N 4 GAS motif (Ϫ1462/Ϫ1453), diminished IL-13 responsiveness (Fig. 3C). Electrophoretic mobility shift assay (EMSA) showed that the probe corresponding to this motif specifically binds to STAT6 (Fig. 3D). Moreover, we confirmed that tyrosyl phosphorylation of STAT6 is required for induction of periostin by IL-13 by using the mutated form of STAT6, Y641F, in which the phosphorylated tyrosyl residue of STAT6 is mutated (Fig. 3E). Taken together, these results demonstrate that STAT6 acts as a cis-regulatory mechanism for transcription of periostin induced by IL-13 via the N 3 GAS motif located at Ϫ134 to Ϫ126 bp.

Identifying SOX11 as a candidate for a trans-acting factor for IL-13-induced transcription of periostin in lung fibroblasts
To identify candidates for trans-acting factors for IL-13induced transcription of periostin in lung fibroblasts, we searched for transcriptional factors among the IL-13-inducible molecules in lung fibroblasts using DNA microarray analyses, finding that expressions of four transcription-related factors-SOX11, PIM1, PRRX2, and ID2-were significantly up-regulated in IL-13-stimulated MRC-5 cells (Fig. 4A). The analyses of quantitative RT-PCR (qRT-PCR) showed that expression of

Control of IL-13 signals by SOX11
SOX11, PIM1, and PRRX2, but not ID2, was increased by stimulation of IL-13 in MRC-5 cells (Fig. 4B). Among these three molecules, we focused on SOX11 for two reasons. 1) SOX11 was the only SOX family member induced by IL-13 (Fig. 4C), and 2) the Gene Expression Omnibus (GEO) at NCBI shows in a data set that expression of both POSTN and SOX11 are upregulated in IL-13-treated IMR-90 cells, another human fibroblast cell line (GEO accession number GDS5256; Fig. S1). We confirmed that IL-13 up-regulated expression of SOX11 protein in MRC-5 cells (Fig. 4D). Moreover, knockdown of STAT6 significantly decreased SOX11 expression induced by IL-13 in MRC-5 cells (Fig. 4E). In contrast, IL-13 did not induce expression of SOX11 in mouse primary lung fibroblasts, although it induced expression of periostin and SOCS1 via STAT6, indicating that the expression mechanism of SOX11 is different in humans and mice (data not shown). These results demonstrate that SOX11 is a downstream molecule of the IL-13 signal and indicate that SOX11 may be a candidate for a trans-acting factor for IL-13-induced transcription of periostin in lung fibroblasts.

Regulation of IL-13-induced periostin expression by SOX11 in lung fibroblasts
We next examined the regulatory roles of SOX11 in IL-13induced periostin expression using primary normal human lung fibroblasts (NHLFs). IL-13 induced expression of POSTN, CCL26, and SOCS1, which were significantly down-regulated by either knockdown of STAT6 (Fig. 5A) or treatment with cycloheximide ( Fig. 5B), in these cells as well as in MRC-5 cells. Moreover, knockdown of STAT6 also significantly decreased IL-13-induced SOX11 expression in these cells (Fig. 5C). To clarify the role of SOX11 in transcriptional regulation by IL-13, we examined the effects of knockdown of SOX11 on expression of periostin, CCL26, and SOCS1 induced by IL-13. Knockdown of SOX11 significantly diminished expression of periostin and CCL26 at the mRNA level and of periostin at the protein level,

Control of IL-13 signals by SOX11
whereas expression of SOCS1 did not decrease (Fig. 5, D and F). Knockdown of SOX11 did not change the expression of STAT6, excluding the possibility that down-regulation of periostin and CCL26 was not via down-regulation of STAT6 (Fig. S2). Down-regulated expression of both periostin and CCL26, but not of SOCS1, by knockdown of SOX11 was observed in MRC-5 cells as well (Fig. 5, E and F). Knockdown effects of STAT6 on SOX11 expression was confirmed at the protein level too (Fig. 5G). Knockdown of PIM1 or PRRX2, transcriptional-related factors identified by DNA microarray (Fig.  4A), affected not at all or only slightly the expression of periostin in MRC-5 cells (data not shown). Moreover, overexpression of SOX11 enhanced periostin expression by IL-13 in MRC-5 cells, whereas overexpression of the mutated forms of SOX11 in which the DNA-binding domain or the transactivation domain is deleted did not have such an effect (Fig. 5H). Both forms of SOX11 impaired the induction activity of periostin by IL-13. These results demonstrate that in lung fibroblasts SOX11 acts as a trans-regulatory molecule for expression of periostin and CCL26, but not SOCS1, by IL-13.

Up-regulation of periostin in lung fibroblasts by enforced expression of SOX11
We next examined whether enforced expression of SOX11 reciprocally up-regulates periostin expression in lung fibroblasts. We transfected the plasmid encoding human SOX11 or control plasmid to MRC-5 cells followed by stimulation of IL-13. When we transfected the control plasmid to MRC-5 cells, IL-13 stimulation up-regulated periostin expression as shown previously, which is due to a combination of cis-and trans-regulation by STAT6 and endogenously expressed SOX11 (Fig. 6). When SOX11 was exogenously overexpressed in MRC-5 cells without IL-13 stimulation, expression of periostin was unchanged. However, when SOX11-overexpressing cells were stimulated with IL-13 for 8 h, periostin expression was significantly up-regulated compared with mock-transfected cells. SOX11 was not coimmunoprecipitated with STAT6, excluding the possibility that these two molecules act by forming a heterodimer (data not shown). These results reinforce the role of SOX11 in the trans-regulation mechanism of periostin, indicating that some transcriptional or post-translational modification induced by IL-13 would be required for this mechanism.

Differentiating SOX11-dependent and -independent genes in IL-13-inducible genes in lung fibroblasts
The finding that knockdown of SOX11 down-regulated IL-13-induced expression of periostin and CCL26 but not of SOCS1 suggested that there would be two groups in IL-13induced genes in lung fibroblasts: SOX11-dependent and -independent genes. To differentiate these two groups, we submitted four groups of genes from NHLFs, with or without stimulation

Control of IL-13 signals by SOX11
of IL-13 and with or without SOX11 knockdown, to DNA microarray analysis. Based on the up-regulation by stimulation of IL-13 (longitudinal axis) and the down-regulation by SOX11 knockdown (horizontal axis), we divided the genes into four groups (Fig. 7A): 1) IL-13-inducible and SOX11-dependent (upper left panel), 2) IL-13-inducible and SOX11-independent (upper right panel), 3) IL-13-noninducible and SOX11-dependent (lower left panel), and 4) IL-13-noninducible and SOX11independent (lower right panel). As expected, both POSTN and CCL26 were included in the IL-13-inducible and SOX11dependent genes, whereas SOCS1 was included in the IL-13-inducible and SOX11-independent genes, validating the classification of the genes by IL-13 induction and SOX11 dependence using DNA microarray analysis.
We then applied the IL-13-inducible and SOX11-dependent genes, defined by more than 4-fold up-regulation by stimulation of IL-13 and more than 40% down-regulation by SOX11 knockdown (470 genes), to DAVID analysis to enrich biological functions in this group. We found that this procedure picked up "signal" (a GO term) with a high enrichment score in which both POSTN and CCL26 were included (Fig. 7B). This group contained chemokines (CCL3L3, CCL7, CCL11, and CCL26), a proinflammatory cytokine (IL6), cytokine receptors (IL1RL1, CSF2RA, and FGFR3), immune receptors (CD3E, FCRL2, FCRL3, and HLA-DRB5), cadherin molecules (CDH26, PCDHB9, and PCDHB10), and an extracellular matrix protein (TNXB) as well as POSTN. These results suggest that there exist two groups in IL-13-induced genes in lung fibroblasts, SOX11dependent and -independent genes, and that the POSTN and CCL26 genes are included in the former group.

Discussion
In this study, we showed that in lung fibroblasts there exist both cis-and trans-acting transcriptional mechanisms through STAT6 in IL-13-mediated expression; in the former, STAT6 activated by IL-13 directly binds to the promoter/enhancer region on the targeted genes, whereas in the latter, STAT6 activated by IL-13 induces expression of SOX11, which then induces some IL-13-targeted genes (Fig. 8). This in turn indicates that there exist SOX11-dependent and -independent genes in the IL-13-inducible genes downstream of STAT6. These results have highlighted the importance of SOX11 in the pathogenesis of IL-13-related diseases such as allergic diseases.
The SOX family is composed of nine groups (A-H with the B group including B1 and B2) among which the SOX C group contains three members, SOX4, SOX11, and SOX12, having a common genetic structure (21). The members of the SOX family mainly function in tissue or organ development. We found that SOX11 is the only SOX member induced by IL-13 (Fig. 4C) and then that SOX11 plays a role in inducing certain genes downstream of the IL-13/STAT6 signal. To our knowledge, transcriptional regulation downstream of IL-13/STAT6 is the first example of the immune-regulatory effects of the SOX family. This suggests the unique characteristic of SOX11 in the SOX family and, furthermore, that the functions of SOX11 are broader than what are currently known. Thus far, we do not know in detail how SOX11 induces periostin expression. The requirement of IL-13 stimulation in periostin expression by enforced SOX11 expression (Fig. 6) indicates that enforced SOX11 expression alone is not enough for periostin expression

Control of IL-13 signals by SOX11
induces GATA-binding protein 3, which in turn induces type 2 cytokines IL-4, IL-5, and IL-13 (33). However, to our knowledge, no trans-activating transcriptional system downstream of STAT6 via a transcriptional factor other than GATA-binding protein 3, in cells other than T H 2 cells, or targeting genes other than type 2 cytokines is known. The present finding has added a novel transcriptional mechanism to the signal transduction pathway of IL-13. It has been established that IL-13 is a pleiotropic cytokine exerting various actions in many kinds of cells, including B cells, T cells, mast cells, macrophages, epithelial cells, fibroblasts, smooth muscle cells, and endothelial cells (10). In the present study, we showed that SOX11 acts as a trans-acting transcriptional factor downstream of the IL-13/STAT6 signals in lung fibroblasts. Moreover, we propose that the IL-13/STAT6/ SOX11 pathway in lung fibroblasts is important for both inflammation and fibrosis by inducing inflammation-or fibrosis-related molecules, including periostin and CCL26. However, it remains undetermined whether the IL-13/STAT6/ SOX11 pathway is adopted in other cells. It is known that SOX11 is expressed in various cells, including neural cells, kidney cells, and osteoblasts (21,34), and that epithelial-mesenchymal transition up-regulates SOX11 expression, suggesting that SOX11 expression in fibroblasts would be relatively abundant (35). As far as we could determine from GEO Profiles searches, SOX11 was not listed as an IL-13-inducible gene in bronchial or esophageal epithelial cells. Overall, the IL-13/ STAT6/SOX11 pathway may not be common in different cell lineages but more cell lineage-specific, at least in fibroblasts.
We identified the IL-13-inducible genes as either SOX11-dependent or -independent (Fig. 7A). The upper left and upper right panels corresponding to the IL-13-inducible/SOX11dependent and IL-13-inducible/SOX11-independent genes contained ϳ7700 and ϳ6700 dots, respectively, suggesting that roughly more than half of the IL-13-inducible genes are SOX11-dependent. To establish the role of SOX11 downstream of the IL-13/STAT6 signals in lung fibroblasts, we await the comparison of the phenotypes of mice deficient in STAT6 or SOX11 specifically in fibroblasts; it has been shown that global deficiency of SOX11 is lethal in mice (22). However, in the present study, we show that, based on microarray analysis, the IL-13/STAT6/SOX11 pathway is important for induction of genes related to inflammation and/or fibrosis, including chemokines, proinflammatory cytokines, cytokine receptors, immune receptors, cadherin molecules, and extracellular matrix proteins (Fig. 7, B and C). Consistent with this assumption, we and others have reported that deficiency in either IL-13 or periostin, a downstream molecule of the IL-13 signal, shows comparable protection of pulmonary fibrosis by bleomycin (36,37). Thus, the discovery of involvement of SOX11 in the IL-13/ STAT6 signals would clarify the pathophysiological roles of the IL-13/STAT6 signals. Alternatively, various kinds of agents targeting IL-13 or its receptor to inhibit STAT6 activation could target asthma and pulmonary fibrosis (Clinical Trial NCT01872689) (20,38). Our present finding may give us a clue to finding another option to develop an inhibitor for IL-13/ STAT6 signals.

Control of IL-13 signals by SOX11
purchased from Lonza and maintained as the vendor recommends. These cells were stimulated with 50 ng/ml IL-13 (Pep-roTech), and then the supernatants were applied to ELISA. For knockdown experiments, MRC-5 cells and NHLFs were cultured with small interfering RNA (siRNA) for 24 h followed by stimulation with 50 ng/ml IL-13 for 24 h. Then RNA extracts and the supernatants were applied to qRT-PCR or ELISA, respectively. For some experiments, MRC-5 cells and NHLFs were pretreated with the indicated concentration of a STAT6selective inhibitor, AS1517499 (AXON Medchem), or 10 g/ml cycloheximide for 30 min and then treated with 50 ng/ml IL-13 for 24 h. For overexpression experiments, MRC-5 cells were transfected with plasmid for 24 h followed by stimulation with 5 ng/ml IL-13 for 8 h. Then RNA extracts were applied to qRT-PCR.

Knockdown of mRNA by siRNA
siRNA oligonucleotides were purchased from Dharmacon/ GE Healthcare. Cells were transfected with ON-TARGETplus siRNA for STAT6, SOX11, PIM1, PRRX2, or control at the indicated concentrations and for the indicated times in the presence of RNAiMAX reagent (Thermo Fisher Scientific). The nucleotide sequences are depicted in Table S1. Silencing of target genes was confirmed by qRT-PCR.

ELISA for periostin production
ELISA for periostin was performed using two kinds of antiperiostin Abs, SS18A and SS17B (Shino-Test) as described previously (39).

qRT-PCR
qRT-PCR was performed as described previously (18). Primers for qRT-PCR are described in Table S2.

Expression vectors and reporter gene constructs
We subcloned the proximal 5Ј region of the POSTN gene (Ϫ5114/ϩ1 bp relative to the transcription start site) into the XhoI and NcoI sites of the firefly luciferase reporter vector pGL4.10 (Promega). This construct was used as a template to generate a series of deletion plasmid constructs by inverse PCR. The mutation in the STAT6-binding site of the promoter region was introduced by PCR-based mutagenesis (40). We generated two mutation constructs in the N 4 GAS motif (Ϫ1462/Ϫ1453, TTCTTTTGAA to TATTTTTGAA), and in the N 3 GAS motif (Ϫ134/Ϫ126, TTCCTGGAA to TATCTG-GAA). The plasmid pME18S containing human STAT6 cDNA was prepared as described previously (41). STAT6 Y641F and the mutated form of SOX11 deleting the DNA-binding domain (⌬HMG) were generated using inverse PCR with primers designed to alter tyrosine to phenylalanine (TAT to TTT) or to delete HMG box domain, respectively. The mutated form of SOX11 deleting the transactivation domain (⌬TAD) was constructed by PCR with primers to delete the C-terminal 34 amino acids. Primers for introducing mutant constructs are as follows: 5Ј-TTTGTCCCAGCTACCATCAA-GATG-3Ј for STAT6 Y641F forward, 5Ј-ACCCCTGCCATC-CTTACCCATCT-3Ј for STAT6 Y641F reverse, 5Ј-AAGCCC-AAAATGGACCCCTCG-3Ј for SOX11 ⌬HMG forward, 5Ј-CGACGCCGTCTTGCACCAGTCTG-3Ј for SOX11 ⌬HMG reverse, 5Ј-ACTCGAGATCATGGTGCAGCAGGCGGA-3Ј for SOX11 ⌬TAD forward, and 5Ј-AATACTAGTTCACAGG-CTGCCCTCGCTGAACG-3Ј for SOX11 ⌬TAD reverse.

Transient transfection and luciferase assay
Transfection was performed using HEK293T cells with FuGENE 6 reagent (Promega). Both the plasmids encoding each POSTN promoter and human STAT6 cDNA were cotransfected into the HEK293T cells. The pRL-TK plasmid (Promega) was also cotransfected as an internal control to normalize transfection by Renilla luciferase activity.

Overexpression of SOX11 protein
MRC-5 cells were transfected with 2.5 g of SOX11 expression plasmid or control in the presence of 100 l of nucleofector solution specifically formulated for transfection of fibroblasts (Lonza) using the Amaxa Nucleofection system (Lonza). Immediately after pulsing, the cells were transferred to 12-well plates and cultured for 24 h followed by stimulation with 5 ng/ml IL-13 for 8 h. Then RNA extracts were applied to qRT-PCR.

EMSA
Nuclear extracts were prepared from HEK293T cells transiently transfected with pME18S encoding human STAT6 cDNA. The primer sequences in this assay were as follows: biotin-5Ј-CGATGCTTCCTGGAAAGAGTTC-3Ј for the wildtype (WT) POSTN promoter and biotin-5Ј-CGATGCTATCT-GGAAAGAGTTC-3Ј for the mutant probe. For competitive inhibition, a 100-fold molar excess of nonbiotinylated WT probe was added to the reaction mixture 15 min before adding the WT probe. For the supershift assay, nuclear extracts were preincubated with anti-STAT6 Ab or control IgG for 15 min. The samples were separated on a 5% nondenaturing polyacrylamide gel containing 0.4ϫ Tris-buffered saline at 4°C. Samples were transferred to nylon membranes (Hybond-N, GE Healthcare) for 50 min at 380 mA with stirring. Membranes were cross-linked at 120 mJ/cm 2 for 1 min with a UV Stratalinker 2400 cross-linker (Agilent Technologies). Immunodetection was performed with a Chemiluminescent Nucleic Acid Detection Module kit (Thermo Fisher Scientific).

DNA microarray analysis
MRC-5 cells were stimulated with or without 50 ng/ml IL-13 for 24 h. NHLFs were cultured with siRNA for 24 h followed by stimulation with 50 ng/ml IL-13 for 24 h. One replicate array for each experimental condition was performed. Total RNA with

Control of IL-13 signals by SOX11
an RNA integrity number greater than 9.0 was applied to an Agilent Expression Array (SurePrint G3 Human GE8x60K v2 Microarray for MRC-5 and v3 Microarray for NHLFs, Takara Bio). The signal intensity values themselves or their relative ratios were presented on a heat map and subjected to Multi-Experiment Viewer (MeV) v4.9 software (Dana-Farber Cancer Institute). The identified genes are registered in the NCBI Gene Expression Omnibus (GEO) (for MRC-5 cells, GEO accession number GSE104936; for NHLFs, GEO accession number GSE104937). For gene ontology analysis, the Database for Annotation, Visualization, and Integrated Discovery (DAVID) tool (National Cancer Institute) was used. This database includes the Gene Ontology Database (http://geneontology.org/). 3

Statistical analysis
Data are presented as mean Ϯ S.D. Statistical analyses were performed using Prism 5.0 software (GraphPad Software). The significance of differences was assessed using an unpaired or paired Student's t test. Values of p Ͻ 0.05 were considered statistically significant.