Gene expression and biological processes influenced by deletion of Stat3 in pulmonary type II epithelial cells

Background The signal transducer and activator of transcription 3 (STAT3) mediates gene expression in response to numerous growth factors and cytokines, playing an important role in many cellular processes. To better understand the molecular mechanisms by which Stat3 influences gene expression in the lung, the effect of pulmonary epithelial cell specific deletion of Stat3 on genome wide mRNA expression profiling was assessed. Differentially expressed genes were identified from Affymetrix Murine GeneChips analysis and subjected to gene ontology classification, promoter analysis, pathway mapping and literature mining. Results Total of 791 mRNAs were significantly increased and 314 mRNAs were decreased in response to the deletion of Stat3Δ/Δ in the lung. STAT is the most enriched cis-elements in the promoter regions of those differentially expressed genes. Deletion of Stat3 induced genes influencing protein metabolism, transport, chemotaxis and apoptosis and decreased the expression of genes mediating lipid synthesis and metabolism. Expression of Srebf1 and 2, genes encoding key regulators of fatty acid and steroid biosynthesis, was decreased in type II cells from the Stat3Δ/Δ mice, consistent with the observation that lung surfactant phospholipids content was decreased. Stat3 influenced both pro- and anti-apoptotic pathways that determine cell death or survival. Akt, a potential transcriptional target of Stat3, was identified as an important participant in Stat3 mediated pathways including Jak-Stat signaling, apoptosis, Mapk signaling, cholesterol and fatty acid biosynthesis. Conclusion Deletion of Stat3 from type II epithelial cells altered the expression of genes regulating diverse cellular processes, including cell growth, apoptosis and lipid metabolism. Pathway analysis indicates that STAT3 regulates cellular homeostasis through a complex regulatory network that likely enhances alveolar epithelial cell survival and surfactant/lipid synthesis, necessary for the protection of the lung during injury.


Background
The signal transducers and activators of transcription (STATs) were initially identified as a family of latent cytoplasmic transcription factors that are activated by various cytokines, growth factors and other stimuli, and phosphorylated by many protein kinases [1][2][3][4]. In response to various stimuli, STAT family members are phosphorylated by receptor associated kinases, form homo-or heterodimers and are translocated to the cell nucleus where they activate transcription [5]. Recent studies also support the role of unphosphorylated STAT3 that accumulates in nucleus and activates transcription by binding to NFkappaB [6]. STAT3 regulates a variety of biological processes, functioning at both transcriptional and non-transcriptional levels to influence cell growth, survival and metabolism. Its capacity to induce cell transformation and tumorigenesis makes it a potential therapeutic target for various cancers [7,8].
Systemic deletion of Stat3 is embryonic lethal in the mouse, indicating its important role in embryogenesis [9]. Biological roles of STAT3 in various organs and cells have been studied in vitro as well as cell specific deletion in the mouse in vivo. The biological consequences of Stat3 deletion are surprisingly diverse and sometimes contradictory. For example, cardiomyocyte-specific STAT3 deficiency caused cardiac fibrosis and heart dysfunction with advanced age [10]. Hepatic cell specific deletion of Stat3 caused insulin resistance associated with increased expression of gluconeogenic genes [11]. Conditional ablation of Stat3 in respiratory epithelium of the mouse (Stat3 Δ/Δ mice) did not alter lung morphogenesis or function but enhanced susceptibility to hyperoxia and adenoviral induced lung injury whereas overexpression of Stat3C (a constitutive active form of STAT3) in pulmonary epithelium protects against hyperoxic lung injury, suggesting that STAT3 is required for the maintenance of surfactant homeostasis and lung function during injury [12][13][14]. While STAT3 has been proposed as an anti-apoptotic protein through the induction of survival genes such as Bcl2like 1(Bcl-X) and B-cell leukemia/lymphoma 2 (Bcl-2), STAT3 also exerts pro-apoptotic effect through the regulation of insulin-like growth factor binding protein 5 (IGFBP-5) to modulate mammary epithelial apoptosis [4,15]. STAT3 is abundantly and ubiquitously expressed in many tissues and distributed between the cell cytosol and nucleus. A direct effect of non-phosphorylated, cytoplasmic STAT3 on cell motility was reported recently through direct protein-protein interactions [16], indicating a non-transcriptional function of STAT3.
The functions of STAT3 vary in different cellular and physiologic contexts, influencing diverse gene targets by interaction with other proteins and genes. The diversity of STAT3 functions indicate that STAT3 is involved in com-plex genetic networks to maintain cellular homeostasis rather serving a singular role in acute phase responses as initially defined. In the present study, we sought to systemically study the role of STAT3 in pulmonary epithelial cell homeostasis. Using knowledge based gene expression profiling approaches and a conditional system that selectively deleted Stat3 in the respiratory epithelium; we identified a large STAT3-dependent network that influences a wide variety of biological processes in type II alveolar cells in the lung.

Identification of differentially expressed genes in alveolar type II epithelial cells from Stat3 Δ/Δ mice
In previous studies Stat3 mRNA and protein expression were markedly reduced in Type II cells isolated from Stat3 Δ/Δ mice, being less than 10% of control levels [12]. To identify the RNAs influenced by the conditional deletion of Stat3 Δ/Δ , RNAs isolated from alveolar epithelial type II cells of control and Stat3 Δ/Δ mice were compared using Affymetrix murine genome MOE430 gene chips. The complete dataset can be found at Gene Expression Omnibus (GEO) [17]; Accession no. GSE6846. Total of 1105 genes were identified as significantly altered using the criteria described in Method. Among them, 791 mRNAs were increased and 314 mRNAs were decreased in response to the deletion of Stat3 Δ/Δ in the lung (Additional file 1). Changes in mRNA expression of a subset of genes including Malt1, Rnt4, Reg3g, Bcl2l1, Cds2, Cdipt, Fasn, Acox2, Akt2, Gpam, Foxj1, Abca3, Srebf1, Srebf2 and Scap were validated by real-time RT-PCR. Genes cross validated by RT-PCR are listed in Table 1 and indicated by asterisks (*) in Additional file 1.

Functional classification of differentially expressed genes revealed the dysregulation of various biological processes in type II epithelial cells from Stat3
Δ/Δ mice Differentially expressed genes were classified according to Gene Ontology (GO) classification on Biological Process. The Fisher Exact Test was used to calculate the probability of each category that was overrepresented in the selected list using the entire MOE430 mouse genome as reference dataset. Deletion of Stat3 from type II cells significantly induced the genes involved in protein metabolism, protein transport, chemotaxis and apoptosis and decreased the expression of genes in lipid synthesis and metabolism ( Table 2).

Promoter analysis identified putative common regulators of the differentially expressed genes
To identify putative common transcription factors regulating the type II cell responses to Stat3 deletion, promoter region (-2kb to exon1) of differentially expressed genes were searched for overrepresented cis-elements using Mat-Inspector (Genomatix) vertebrate matrix library. In com-pare with the sequence of random chosen gene promoters, the cis-elements significantly enriched in the promoter region of differentially expressed genes were selected based on a binomial probability calculation and their percentage frequency in our selected gene list (Adjusted p Value < 0.001 and frequency > 50 %). STAT, EGRF, AHRR, SP1F, ZF5F, E2FF, HIFF, SREB and AP2F were the most overrepresented cis-elements and may therefore mediate changes in gene expression in cells from Stat3 Δ/Δ mice ( Table 3). The finding that STAT was the Note: Gene Ontology Analysis was preformed using public available web-based tool DAVID [81]. Overrepresented biological processes were selected at the threshold of Fisher Exact Test P-Value ≤ 0.05 and minimum gene counts belonging to an annotation term ≥ 2%. most enriched cis-elements indicated the identification of a sub-group of potential direct transcriptional targets of Stat3 in lung epithelial cells. Sterol regulatory element binding factors (Srebf1 and 2, known as important transcriptional regulators of fatty acid and steroid biosynthesis), were significantly decreased in Stat3 Δ/Δ mice. The SREB binding site was overrepresented in differentially expressed genes indicating that it is a potential regulator of the lipid metabolism pathways altered in Stat3 Δ/Δ mice. Other significantly enriched TFBS include SP1F and HIFF. STAT3 and SP1 function cooperatively to activate the C/ EBP promoter, the SP1 site being required for IL-6 induction and transactivation by STAT3 [18]. HIF1A, SP1, SMAD3 AND SMAD4 can form multifactor complex, regulating VEGF and erythropoietin gene transcription through functional cooperation and association [19][20][21]. Consistent with the promoter analyses and literature findings, the mRNA expression of Klf5 (a member of the SP1 family) and Hif1a, Smad3 and Smad4 was simultaneously increased by Stat3 deletion, indicating the potential transcriptional complex formation among the corresponding transcription factors.

Pathway analysis revealed known and novel functions of STAT3 in the lung
Pathway enrichment test is an unbiased way to answer the question, ''Are the differentially expressed genes enriched in certain pathways?'' To address this issue, we compared the overlap of differentially expressed genes in Stat3 Δ/Δ mice with the known biological pathways in KEGG (Kyoto Encyclopedia of Genes and Genomes) using 1) the mouse genome and 2) a list of genes shown to be least changed in response to the Stat3 deletion in the gene array as background. Results from both analyses were consistent. Jak-Stat Signaling Pathway, Apoptosis, Cytokine-Cytokine Receptor Interaction, Insulin Signaling Pathway, Mapk Signaling Pathway, Focal Adhesion, and Wnt Signaling Pathway were among the most enriched pathways identified from the analysis of differentially expressed genes (Table 4). mRNAs mediating steroid biosynthesis and fatty acid metabolism were mostly decreased in Stat3 Δ/Δ mice. The most overrepresented pathways and the differentially expressed genes associated with those pathways are illustrated in Additional files 2. The finding that known Stat3 functions, including the Jak-Stat Signaling Pathway were identified from microarray analysis of Stat3 Δ/Δ mice type II cells using knowledge integration approaches provides support for the utility of the analysis to detect novel pathways regulated by Stat3.

Stat3 influences protein metabolism in lung type II cells
''Protein metabolism'' was the most enriched biological process (5.83E-10), accounting for 20% of the induced genes caused by the deletion of Stat3. More than 40 genes encoding proteins involved in protein ubiquitination/ ubiquitin cycle were present (Table 5). Of interest, Casitas B-lineage lymphoma b (Cblb) was increased 7.3 fold. CBLB is a member of Cbl ubiquitin ligases (E3) protein family that are tyrosine-phosphorylated in response to a Overrepresented pathways were identified by comparison the overlap of differentially expressed genes and all genes in MOE430 mouse genome (reference) with the known KEGG pathways. A pathway is considered to be over-represented when a probability P-Value ≤ 0.01 and gene frequency (genes in the pathway/total number of differentially expressed genes) ≥ 2%. Note: promoter sequence (-2000 -0 bps) from 886 out of 1105 differentially genes were available for analysis. The P-Value is calculated using binomial distribution probability. 5 random sets of 1000 mouse gene -2kb promoter regions were used to calculate the background frequency. The Q-Value is calculated using single-step Bonferroni adjustment to control for the multiple comparison effect.
wide variety of receptor mediated stimuli, including epidermal growth factor receptors, cytokine receptors such as colony stimulating factor family receptors (increased 2-3 fold in the Stat3 Δ/Δ cells) and immune complex receptors such as Fcgr2b (increased 2.7 fold). CBLB functions as an adaptor protein that regulates many signal-transduction pathways including JAK-STAT signaling pathways through its ability to interact with critical signaling molecules [22]. The expression of genes encoding known CBLB binding partners was significantly increased in type II cells from the Stat3 Δ/Δ mice, including SYK kinase (increased 4 fold), JAK2 (increased 2.1 fold), PKC (increased 2.6 fold), phosphatidylinositol 3-kinase (Pik3r1 increased 1.8 fold, Pik3c2a increased 3 fold), adenylate cyclase-associated protein 1 (Cap1 increased 5 fold) and Crk (increased 2.5 fold). These signaling molecules can in turn modulate various down stream signaling cascades including those mediated by small G-proteins (multiple members of RAS oncogene family were induced 2-3 fold), as well as PI3K/ AKT mediated signaling and Nfat induction (increased 4.1 fold) [23][24][25][26][27]. CBLB physically interacts with another ubiquitin-protein ligase, NEDD4 (increased 3.8 fold), an interaction that reverses CBLB effects by targeting CBLB for ubiquitination and proteasomal degradation [28]. SLIM is a known E3 ubiquitin ligase specifically interacts with activated STAT1 and STAT4, inducing their ubiquitination and degradation [29]. We speculate that CBLB may act in a similar manner to regulate STAT3 signaling. In addition, SUMO1 and SUMO1 specific peptidase 2 were both increased more than 2 fold in Stat3 Δ/Δ mice. SUMO1 can covalently modify many targets including STAT1 and glucocorticoid receptor NR3C1 (increased 5 fold) and regulate protein stability and transcriptional activity [30]. Taken together, deletion of STAT3 activates a number of molecules regulating protein metabolism, stability and routing, indicating the potential role of ubiquitination and sumoylation in cytokine signaling and STAT3 activation that influences cellular adaptation.

STAT3 influences lipid homeostasis in lung type II cells
Expression of genes encoding sterol regulatory element binding factor 1 and 2 (Srebf1 and Srebf2), their cleavage activating protein (Scap) and multiple SREBP target genes involved in lipid metabolism were decreased in Stat3 Δ/Δ mice type II cells. As depicted in Figure 1, genes dedicated to the biosynthesis of fatty acid, phospholipid and choles- Genes involved in lipids biosynthesis and clearance pathways were altered in type II cells from Stat3 Δ/Δ mice terol were down regulated, with the exception of 3hydroxy-3-methylglutaryl-Coenzyme A synthase (Hmgcs), which was increased. HMGCS participates in other metabolic pathways, including valine, leucine and isoleucine degradation (KEGG 00280). This analysis indicates that metabolic pathways regulating fatty acid, phospholipid, and cholesterol biosynthesis were coordinately decreased after deletion of Stat3 in type II cells, supporting an important role of STAT3 in regulating lipids biosynthesis in the lung. LDL receptor (Ldlr), mediating cholesterol uptake, and ATP-binding cassette A3 (Abca3), important for phospholipid transport, lamellar body formation and pulmonary surfactant secretion in alveolar type II cells were decreased in the Stat3 Δ/Δ mice cells [36][37][38]. Recent in vitro studies from our group confirmed the direct binding of SREBP1c to the Abca3 promoter (Besnard et.al, submitted for publication). On the other hand, the increased expression of Abca1, a key facilitator of cellular cholesterol and phospholipid export [39,40] and high density lipoprotein binding protein (Hdlbp), which may function in the removal of excess cellular cholesterol, suggests that cholesterol and phospholipid clearance were induced after deletion of Stat3. Adenosine kinase (Adk), was increased 8.2 fold. Cellular role of ADK in lipid metabolism is somewhat controversial [41], Adk deficient mice developed neonatal hepatic steatosis and die within 14 days with fatty liver [42]. Increased expression of lipid export genes and decreased expression of genes mediating lipid biosynthesis likely cause a reduction of total lipids level in the type II cells of Stat3 Δ/Δ mice.
Previous studies demonstrated the susceptibility of Stat3 Δ/ Δ mice to lung injury and death related to surfactant dysfunction [12]. Consistent with our prediction from the present mRNA microarray analysis, the saturated phosphatidycholine (SatPC) content in bronchoalveolar lavage fluid was significantly decreased in Stat3 Δ/Δ mice ( Figure 2). Significantly decreased SatPC synthesis and abnormalities in lamellar body numbers and morphology were also observed in Stat3 Δ/Δ mice (data not shown). Taken together, a number of genes regulating surfactant lipid homeostasis were altered in type II cells isolated from the Stat3 Δ/Δ mice, consistent with biochemical, functional, and morphologic changes in the surfactant system that is exacerbated by oxidant stress [12] or expose to pathogens [13].

Role of Akt in Stat3 regulated lipid metabolism
Our observations support the view that the decrease in SREBP, at least in part, results in decreased expression of genes regulating lipid biosynthesis and metabolism in type II cells from Stat3 Δ/Δ mice. SREBPs are master regulators of lipid metabolism. The transcriptional targets, and the pathways mediated by SREBP in liver have been well studied [43]. SREBPs are expressed in the developing lung, SREB1c increases in the developing lung concomitantly with the perinatal increase in surfactant and lipid synthesis, surfactant protein and Abca3 expression, genes critical for surfactant function.
However, the role of Stat3 in regulating SREBP and associated lipid metabolism in the lung is largely unknown. In the present analysis, we sought to identify mechanisms by which Stat3 regulates SREBPs and associated lipid biosynthesis pathways in alveolar type II cells in the lung. The regulation of SREBPs occurs at both transcriptional and post-transcriptional levels. The post-transcriptional regulation requires SCAP. Cre-mediated disruption of Scap significantly reduced Srebf1 and 2 levels as well as SREBP target gene expression in liver [44]. There are multiple potential STAT sites on the Scap promoter. Thus, Stat3 may influence lipid biosynthesis through the transcriptional regulation of Scap. At transcriptional level, nuclear hormone receptors (Nr1h2 and Nr1h3) and Pgc1-alpha and beta are known to regulate SREBPs expression in liver. Increased expression of Srebf1 and Fasn was associated with increased hepatic triglyceride content in Stat3 deficient mice [10,45]. In the present study, expression of SREBP and their down stream targets were decreased in type II cells from Stat3 Δ/Δ mice (a finding that contrast Surfactant saturated phosphatidylcholine (SatPC) Figure 2 Surfactant saturated phosphatidylcholine (SatPC) was significantly decreased in bronchoalveolar lavage fluid from the Stat3 Δ/Δ mice, n = 8 per group. Aliquots of BALF were extracted with chloroform/methanol (2:1) and SatPC was isolated with osmium tetroxide followed by measurement of phosphorus as described previously [12]. Statistical differences were analyzed by Student t-test.
with the findings in the liver) without changes in other known regulators such as Nr1h2, Nr1h3, Pgc1-alpha or Pgc1-beta, indicating the likely presence of alternative regulatory mechanisms in lung cells.
We hypothesize that AKT plays an important role in Stat3 regulated SREBP expression and associated lipogenesis in lung based on the following observations: 1) Protein and mRNA levels of Akt were increased in cells constitutively expressing active Stat3 and were reduced after Stat3 depletion [46]. STAT3 binds to Akt1 promoter was confirmed by ChIP assay [47], suggesting that A kt maybe a direct transcriptional target of Stat3. AKT, on the other hand, inhibits Stat3 transcriptional activity and phosphorylation [48]. Decreased Akt gene expression seen after deletion of Stat3 may represent a direct effect of Stat3 deficiency or to a negative regulatory response to STAT3 deficiency, 2) AKT activation induces both Srebf1 and Srebf2 mRNAs and proteins as well as key enzymes in the cholesterol, fatty acid and membrane lipid biosynthesis pathways [49], 3) Multiple lines of evidence suggest that PI3K influences Stat3 activation. STAT3 binds directly to the PI3K regulatory subunits [50][51][52]. The expression of genes encoding for several PI3K subunits were altered after deletion of Stat3 (see Additional file 1), supporting the involvement of PI3K/Akt signaling in Stat3 regulated bioprocesses in lung, and 4) Since AKT physically interacts with FOXA2 and regulates FOXA2-dependent transcriptional activity [53], the effects of AKT may be mediated, in part, via Foxa2. Foxa2 expression was reduced more than 2 fold in cells from the Stat3 Δ/Δ mice. FOXA2 regulates lipid metabolism in both lung and liver [54,55]. Deletion of either Stat3 or Foxa2 resulting in decreased expression of a number of the overlapping genes that play important roles in surfactant homeostasis including Abca3 [55]. Thus, interactions between AKT and FOXA2 represent another potential mechanism by which lipid metabolism is influenced in Stat3 Δ/Δ mice.

Stat3 influences expression of genes mediating apoptosis and cell survival
Many genes modulating apoptosis/cell survival were altered in response to the deletion of Stat3 from type II cells, including multiple Bcl-2 family members (Bcl2, Bcl2l1, Mcl1, Bcl2l11 and Bad), caspase 3 (Casp3), FADDlike apoptosis regulator (Cflar, also known as Flip), mucosa associated lymphoid tissue lymphoma translocation gene 1 (Malt1), prostaglandin-endoperoxide synthase 2 (Ptgs2) and nuclear receptor subfamily 3, group C, member 1 (glucocorticoid receptor, Nr3c1). In depth literature mining identified more apoptosis related genes than did analysis by Gene Ontology annotation. As depicted in Figure 3, our study indicates that STAT3 regulates apoptosis in a complex manner via processes that occur in multiple intracellular locations. Thus, STAT3 appears to serve as key regulator of apoptosis in alveolar type II cells. Among those apoptosis related genes, Malt1, Ptgs2 and Nr3c1 were strongly induced (> 5 fold in compare with control). MALT1 interacts with BCL10 (increased 2 fold in Stat3 Δ/Δ mice). The formation of this complex is essential for NF-kappaB activation that, in turn, play a role in cell survival. IKKB phosphorylates BCL10 in its MALT1 interaction domain, causing BCL10 and MALT1 to disassociate, resulting in attenuation of NFKB signaling and cytokine production [56,57]. PTGS2 also known as COX-2, a key enzyme in prostaglandin biosynthesis, that is highly expressed in alveolar type II cells. The expression of Ptgs2 is increased in epithelial tumors, including non-small cell lung and prostate cancers via activation of the IL-6/ GP130/STAT3 signaling pathway [58]. This pathway could contribute to tumor formation by promotion of tumor cell resistance to apoptosis via inhibitor of apoptosis (IAP)-dependent mechanism [58,59]. Consistent with these observations, Ptgs2, Il6st and two of the IAP family members (Birc4 and Birc6) were correspondingly induced in Stat3 Δ/Δ cells (5.7, 1.7, 3.2 and 2.3 fold respectively). Nr3c1 (increased 5.1 fold) encodes a receptor for glucocorticoids that can act as both a transcription factor and as a regulator of other transcription factors. STAT3 and NR3C1 physically interact to mediate effect of glucocorticoid on the IL-6-mediated inflammatory response [60][61][62]. NR3C1 also interacts with stress-responsive transcription factors (Hif1a, increased 4.7 fold), mitogen activated protein kinase 8 (Mapk8 or Jnk, increased 4.7 fold) and tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, epsilon polypeptide (Ywhal, increased 3.3 fold), a 14-3-3 family of proteins implicated in the pathogenesis of small cell lung cancer [63,64].
STAT3 is likely to regulate apoptosis by multiple mechanisms including gene transcription. Bcl-xL is the direct transcription target of STAT3 [65]. STAT3 can serve as an anti-apoptotic factor by transcriptional up-regulating the expression of Bcl-xL [66]. The decrease of Bcl-xL may represent a direct response to Stat3 deletion. The fact that expression of Bcl-xL blocked the apoptotic effects of the adenovirus in lung injury suggested that Bcl-xL may mediate the role of STAT3 in the regulation and survival of the respiratory epithelium [13]. The Pi3k-Akt pathway represents a second mechanism by which STAT3 influences cytoprotection. Pi3k-Akt signaling mediates a wide range of down stream targets to regulate apoptosis [67]. For example, AKT phosphorylates multiple Bcl-2 family members, including BAD and Bcl-xL [68], inhibits caspase 3 activation [66] and blocks cytochrome C release from mitochondria [69]. Another mechanism by which STAT3 modulates apoptosis is through protein-protein interactions. Bcl-xL is a direct transcriptional target of STAT3; Bcl-xL interacts with VDAC1 to regulate the outer mitochondrial membrane channel induce apoptosis [70,71].
The expression of Casp3 as well as its interaction partners was induced in Stat3 Δ/Δ mice. NR3C1 has both pro-and anti-apoptotic effects. NR3C1 physically interact with STAT3, HIF1A, MAPK8, YWHAL [63,64,76]; these stressresponsive transcription factors and signaling molecules were largely induced in the present array from Stat3 Δ/Δ type II alveolar epithelial cells. The close transcriptional communication and physical interactions among these transcriptional regulators likely play a critical role in regulating the balance of apoptosis and cell survival. In the present study, effects of STAT3 deletion were assessed in type II epithelial cells purified from the adult mouse lung. mRNA was isolated immediately after isolation to avoid cell culture dependent alteration in gene expression. It is possible that the cells have undergone added cellular stress during protease treatment, isolation and purification, which in turn may influence the expression of genes.
Our results support the view that STAT3 regulates the balance between a subset of pro-and anti-apoptotic genes, determining the cell death or survival through multiple mechanisms. Consistent with the present microarray prediction, cleaved caspase-3 and TUNEL positive cells were Genes involved in apoptotic pathways were altered in type II cells isolated from Stat3 Δ/Δ mice Figure 3 Genes involved in apoptotic pathways were altered in type II cells isolated from Stat3 Δ/Δ mice. Genes associated with apoptosis were identified via literature mining using Pathway Studio 4.0 (Ariadne Genomics, Inc.). STAT3 regulates apoptosis positively and negatively, via processes occurring in multiple intracellular compartments. Genes up-regulated in Stat3 Δ/Δ mice are framed in red. Genes down-regulated in Stat3 Δ/Δ mice are framed in green. Each dashed line demarcated with a larger dot linking "apoptosis" and each gene node indicates a regulatory relationship based on the literature references. Regulatory relationships are denoted by line colors (green, positive regulation; red, negative regulation; and gray, regulation direction is unknown).
significantly increased in Stat3 Δ/Δ mice following adenoviral infection and the apoptosis can be blocked by expression of Bcl-xL [13].

Conclusion
Our previous studies demonstrated that Stat3 plays critical role in cyto-protection during lung injury [12,13]. Present data support the role of Stat3 in enhancing epithelial cell survival and surfactant lipid synthesis that contribute to the maintenance of lung function. Deletion of Stat3 from type II alveolar epithelial cells induced the expression of the genes regulating protein metabolism, protein transport, chemotaxis and apoptosis while decreasing the expression of genes regulating lipid synthesis and metabolism. Critical to pulmonary function during injury, Stat3 influences the expression of genes regulating surfactant lipid synthesis and surfactant homeostasis including Abca3. As illustrated in Figure 4, the present study identi-fied a complex regulatory network by which Stat3 regulates gene expression in type II alveolar cells that is required for cellular homeostasis following injury. STAT3 likely interacts with AKT/FOXA2 in the regulation a number of biological processes in alveolar type II cells, including cell survival/apoptosis, cholesterol and fatty acid biosynthesis required for surfactant homeostasis and lung function.

Transgenic Mice
SP-C-rtTA/(tetO)7CMV-Cre/Stat3 flx/flx triple-transgenic mice were generated as described previously [12]. Stat3flx/ flx mice were a kind gift of Dr. Takeda (Hyogo College of Medicine, Japan) [8]. In the presence of doxycycline, exon 21 of the Stat3 gene is permanently deleted from respiratory epithelial cells prior to birth (Stat3 Δ/Δ mice) [12].  [12,13]. Present data support the role of Stat3 in enhancing epithelial cell survival and surfactant lipid synthesis that contribute to the maintenance of lung function. For simplicity, only representative genes were included. mates (double transgenic,SP-C-rtTA, or tetO7CMV-Cre and Stat3 flx/flx ) were used for the experiments. Doxycycline was administered to the dams in the food at a concentration of 625 mg/kg (Harlan Teklad, Madison, WI) from embryonic day 0 (E0) to postnatal day 25 (P25), resulting in extensive deletion of Stat3 in respiratory epithelial cells [12]. As previously described, deletion of Stat3 did not alter lung size, morphology or survival under non-stressed condition [13].

RNA Extraction
Alveolar type II cells were isolated from 8 weeks old, sex and age matched littermate control and Stat3 Δ/Δ mice using collagenase and differential plating as described by Rice et al. [77]. Type II cells from 3 mice were pooled to obtain one cell pellet. Three independent pools were generated from control and Stat3 Δ/Δ mice separately for purification of RNA and microarray hybridization. Type II cells were homogenized with TRIzol reagent (Invitrogen, Carlsbad, CA). RNA concentration was measured by spectrophotometer and normalized prior to cDNA synthesis. These cell isolates consist of more than 90% alveolar type II cells with residual alveolar macrophages as the major contaminating cell. Purity was assessed by modified Papanicolaou stain. Purity and number of type II cells isolated from Stat3 Δ/Δ mice were not different from controls.

RNA Microarray Analysis
mRNA was extracted from three independent pools of isolated type II cells from adult Stat3 Δ/Δ and control mice. The cRNA was then hybridized to the murine genome MOE430 (consists of ≈ 45000 gene entries) chips (Affymetrix, Santa Clara, CA) according to the manufacturer's protocol. The RNA quality and quantity assessment, probe preparation, labeling, hybridization and image scan were carried out in the CCHMC Affymetrix Core using standard procedure. RNA quality and quantity were analyzed by spectrophotometer. The A260/A280 ratio was used to determine RNA purity with the acceptable region of 1.9-2.1. Affymetrix Microarray Suite 5.0 was used to scan and quantitate the gene chips under default scan settings. Normalization was performed using the Robust Multichip Average model [78,79]. Data were further analyzed using affylmGUI from R/Bioconductor package [80]. Differentially expressed genes were selected with the threshold of T-Test P-value ≤ 0.05, False Discovery Rate (FDR) ≤ 10% and fold change ≥ 1.5. We prioritized the mRNAs whose abundance consistently changed in multiple probe sets by selecting them without the FDR consideration. Unknown cDNA clones/ESTs and duplicated gene entries were removed from further functional analysis.

Gene Ontology Analysis
Gene Ontology Analysis was performed using public available web-based tool David (database for annotation, visualization, and integrated discovery) [81]. Overrepresented biological processes were selected at the threshold of Fisher Exact Test P-Value ≤ 0.05 and minimum gene counts belonging to an annotation term ≥ 2%.

Promoter Analysis: Transcription Factor Binding Sites Search
Promoter sequences of all differentially expressed genes (-2000 bp upstream from the tentative TSS) were retrieved from our database (originally downloaded from UCSC Genome browser) and searched for over-represented TFBS (Transcription Factor Binding Site) in these sequences using MatInspector (Genomatix) using the complete Vertebrate Matrix Library 6.2. The P value is calculated using binomial distribution probability by comparing the matrix match of the promoter regions of differentially expressed genes with the promoters from random mouse genes sets (-2kb for all the promoters). The single-step Bonferroni adjustment is used to control for the multiple comparison effect (i.e., multiplication of p-value by the number of TFBS in Genomatix Vertebrate Matrix Library).

Pathway Enrichment Test
Overrepresented pathways were identified by comparison the overlap of differentially expressed genes and all genes in MOE430 mouse genome (reference) with the known KEGG pathways. A Fisher's exact test for 2·2 contingency table was used to calculate the statistical significance. A pathway is considered to be over-represented when a probability P value = 0.01 and gene frequency (genes in the pathway/total number of differentially expressed genes) ≥ 2%.

Literature Mining
Potential protein-protein or protein-DNA interactions were identified using Pathway Studio (Ariadne Genomics) that contains MedScan, an automated text-mining tool to search the entire PubMed and other public sources (The current version database contains more than 15 million Medline abstracts/full text). Gene expression profiling results was imported into the Studio and used to interpret pathways, gene regulation networks, and protein interaction maps.

Validation of mRNAs
Real-time RT-PCR was used to cross validate changes in a subset of genes from microarray selection. mRNA was extracted from alveolar type II epithelial cells isolated from Stat3 Δ/Δ and control mice using RNeasy Protect mini kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. RNA concentration was measured by spectrophotometer. cDNA was made with SuperScript