Oncogene (2005) 24, 3397–3408 & 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc

Stat3 regulates common to both wound healing and cancer

Daniel J Dauer1,3, Bernadette Ferraro1,3, Lanxi Song1,3, Bin Yu1,3, Linda Mora2,3, Ralf Buettner2,3, Steve Enkemann2,3, Richard Jove2,3 and Eric B Haura*,1,3

1Thoracic Oncology/Experimental Therapeutics, H Lee Moffitt Cancer Center and Research Institute, Tampa, FL 33612, USA; 2Molecular Oncology Programs, H Lee Moffitt Cancer Center and Research Institute, Tampa, FL 33612, USA; 3Department of Interdisciplinary Oncology, University of South Florida College of Medicine, Tampa, FL 33612, USA

Wound healing and cancer are both characterized by cell hallmarks of cancer consist of an analogous program proliferation, remodeling of extracellular matrix, cell that allows for deregulated cell growth, tumor cell invasion and migration, new blood vessel formation, and invasion, angiogenesis, and metastasis (Hanahan and modulation of blood coagulation. The mechanisms that Weinberg, 2000). These observations have been the link wound healing and cancer are poorly understood. We backdrop for the concept that tumors are ‘wounds that report here that Stat3, a common signaling mechanism do not heal’ (Dvorak, 1986). While the relationship involved in oncogenesis and tissue injury, regulates a between inflammation/wound healing and cancer is common set of genes involved in wound healing and widely accepted, the molecular mechanisms that link cancer. Using oligonucleotide arrays and quantita- inflammation and cancer remain unclear (Coussens and tive real-time PCR, we evaluated changes in global gene Werb, 2002). Here, we evaluated the ability of Stat3 to expression resulting from expression of Stat3 in lung coordinate pathways common to both wound healing epithelial cells. We report here previously uncharacterized and cancer. We postulated that signaling pathways genes induced by Stat3 implicated in signaling pathways regulated by bothoncogenic signals and tissue injury common to both wound healing and cancer including cell signals coordinate changes common to invasion and migration, angiogenesis, modulation of bothevents. coagulation, and repression of interferon-inducible genes. Originally identified as a key component linking Consistent with these results, we found increased Stat3 normal cytokine signals to transcriptional events in activity associated with wound healing in chronically cells, Stat3 is now thought to play a major role in tumor inflamed mouse lungs and increased Stat3 activity was formation (Levy and Darnell, 2002; Yu and Jove, 2004). identified at the leading edge of lung tumors invading Stat3 is activated by oncogenic signals suchas Src and adjacent nontumor stroma. These findings provide a EGFR signaling as well as by cytokines and mitogens molecular basis for understanding cancer as a deregula- involved in wound healing such as interleukin-6 and tion of normal wound healing processes. hepatocyte-growth factor (HGF) (Song et al., 2003). A Oncogene (2005) 24, 3397–3408. doi:10.1038/sj.onc.1208469 constitutively activated mutant of Stat3 can promote Published online 28 February 2005 cellular proliferation and survival as well as lead to cellular transformation, providing genetic evidence for Keywords: STAT ; cancer; wound healing; the intrinsic oncogenic potential of Stat3 (Bromberg Stat3; lung cancer et al., 1999). Stat3 regulates a number of pathways important in tumorigenesis including cell cycle progres- sion, apoptosis, tumor angiogenesis, invasion and metastasis, and tumor cell evasion of the immune Introduction system (Bromberg et al., 1999; Grandis et al., 2000; Yu and Jove, 2004). In addition to processes involved in In response to tissue injury, local cytokines and growth tumor formation in epithelial cells, Stat3 plays a crucial factors act on epithelial, mesenchymal, and immune cells role in normal wound healing and response to injury. to orchestrate healing of the wound. Wound healing is Stat3 is activated in lung tissues following either generally characterized by processes involved in cell lipopolysaccharide (LPS) exposure or intrapulmonary proliferation, remodeling of extracellular matrix, cell deposition of IgG immune complexes (Gao et al., 2004; invasion and migration, new blood vessel formation, Severgnini et al., 2004). Selective Stat3 deletion in lung and modulation of blood coagulation. Similarly, the epithelial cells does not compromise postnatal lung function but mice exposed to hyperoxia develop rapid *Correspondence: EB Haura, Thoracic Oncology and Experimental lung injury characterized by alveolar capillary leak and Therapeutics Programs, H Lee Moffitt Cancer Center and Research acute respiratory distress (Hokuto et al., 2004). Similary, Institute, MRC3 East, Room 3056, 12902 Magnolia Drive, Tampa, FL 33612-9497, USA; E-mail: hauraeb@moffitt.usf.edu selective loss of Stat3 in keratinocytes results in impaired Received 10 November 2004; revised 8 December 2004; accepted 20 wound healing and defects in keratinocyte migration December 2004; published online 28 February 2005 (Sano et al., 1999). Stat3, wound healing and cancer DJ Dauer et al 3398 Results repressed by Stat3C. The top 50 genes in each of these two groups are shown in Tables 1 and 2, respectively. Identification of Stat3-regulated genes using Complete lists of the significant upregulated and down- oligonucleotide microarrays regulated genes are included in the Supplementary Data (Tables A–C). Importantly, we identified targets that To investigate whether Stat3 regulates genes involved in have been previously reported as Stat3 targets in the bothwound healing and cancer, we investigated literature. These include upregulated genes such as genome-wide changes in gene expression in lung fibrinogen, lipopolysaccharide-binding (LBP), epithelial cells. We developed an adenoviral vector that SOCS3, c/EBPd, BCL6, JunB, and (Fujitani et al., expresses a mutant of Stat3, designated Ad-Stat3C, 1994; Schumann et al., 1996; Hutt et al., 2000; Reljic which is constitutively active in the absence of tyrosine et al., 2000; Bowman et al., 2001; Duan and Simpson- phosphorylation and can transform rodent fibroblasts Haidaris, 2003; He et al., 2003). These putative Stat3- (Bromberg et al., 1999). Quiescent A549 lung cells were regulated genes were then grouped into distinct known infected witheither Ad-Stat3C or a control adenoviral biological functions of the encoded proteins, including vector expressing green fluorescent protein (Ad-GFP). apoptosis, cell cycle, cell signaling, invasion, and These cells were established from a human bronchio- inflammation. We used Onto-Express to aid in under- loalveolar cell carcinoma and have features resembling standing functional roles for genes identified as differ- et al type II alveolar epithelial cells of the lung (Lieber ., entially expressed (Khatri et al., 2002; Khatri et al., 1976). Preliminary results had identified an optimal 2004). Onto-Express is a tool designed to mine available multiplicity of infection (MOI) of 50 resulting in functional annotation data and identify relevant and equivalent levels of Stat3 DNA binding compared with significant functional biological processes. In addition, IL-6 stimulation and resulting in nearly 90% of the cells eachgene was researched by manual examination of expressing GFP (data not shown). Figure 1 demon- published literature using PubMed. This analysis is strates that Stat3–DNA-binding activity is detected after shown in Table 3. as early as 8 hfollowing infection and increases One of the more interesting findings was the throughout the infection. identification of a large number of genes induced by To control for biological and experimental variation, Stat3 that are involved in cell invasion/migration and five 10 cm dishes of cells were infected with either Ad- remodeling of extracellular matrix. These included genes GFP or Ad-Stat3C (10 individual plates). At 24 hafter in the chemoattractants family such as CCL2 and infection, total RNA was collected and used as the RNA CXCL2, proteases in the cathepsin and source for microarray analysis. Eachdishof cells was families suchas uPA and its uPAR, and genes used for one Affymetrix U133A Human GeneChip. involved in bothcell invasion and blood coagulation Differentially expressed genes between Ad-Stat3C and pathways (such as PAI-1). Several genes encoding Ad-GFP were identified using significance analysis of coagulation proteins were similarly upregulated by Stat3 microarrays (SAM) (Tusher et al., 2001). This is a including fibrinogen, PAI-1 (SERPINE1), and throm- method of identifying statistically significant changes in bomodulin. We also identified genes involved in gene expression taking into account multiple testing and angiogenesis including EPAS1, adrenomedullin, and a family-wise error rate. We accepted all genes identified angiopoietin-like protein 4 (ANGPTL4; also known as by SAM as differentially regulated by at least 1.5-fold. PPARg-angiopoietin-related protein), as being upregu- While this choice was somewhat arbitrary, it was largely lated by Stat3. VEGF, previously identified as a Stat3 based on a report demonstrating that a 50% change in target gene, was identified by SAM as being upregulated gene expression can result in tumorigenesis (Yan et al., by Stat3 but with a fold change of slightly less than 1.5 2002). We identified 200 genes induced by Stat3C that in our analysis (Niu et al., 2002). were considered significant and we identified 150 genes In addition, we found that Stat3C expression repressed a large number of interferon-signaling genes. They include IFIT1, IFIT4, and IFITM1, OAS1, OAS2 and OASL, IFI27 and IFI35, GIP2 and G1P3, TRAIL, MX1 and MX2, and PRKR, and Stat1. Onto-Express calculated the P-value of this occurring by chance as 4.8 Â 10À11, indicating that repression is a true function of Stat3 and not a chance occurrence in the data set. These findings are consistent with the opposing roles of Stat1 and Stat3 in controlling cell proliferation, survival, and immune responses in many contexts (Yu and Jove, 2004).

Figure 1 Expression of Stat3C using adenoviral vectors. Quies- cent A549 cells were infected with50 MOI of eitherAd-Stat3C or Validation of microarray results using real-time Ad-GFP for various times, and nuclear extracts assayed for DNA- quantitative PCR analysis binding activity using STAT EMSA. Ad-Stat3C results in time- dependent increase in Stat3 DNA binding, while Ad-GFP has no To validate the oligonucleotide array data, we per- effect on Stat3 DNA-binding activity formed real-time quantitative PCR (QPCR) on a select

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3399 Table 1 Highest overexpressed genes (50) Symbol Gene name Accession Fold change

STAT3 Signal transducer and activator of transcription 3 (acute-phase response factor) BC000627 14.02925 FGG Fibrinogen, gamma polypeptide NM_000509 6.31786 FGB Fibrinogen, B beta polypeptide BG545288 5.86271 FGA Fibrinogen, A alpha polypeptide NM_000508 5.4178 AKAP2 A kinase (PRKA) anchor protein 2 NM_007203 5.03752 LBP Lipopolysaccharide-binding protein NM_004139 4.54253 SLC2A3 Solute carrier family 2 (facilitated glucose transporter), member 3 NM_006931 4.24708 IGFBP1 Insulin-like growthfactor-binding protein 1 NM_000596 4.11986 CCL2 Chemokine (C-C motif) ligand 2 S69738 4.11116 SOCS3 Suppressor of cytokine signaling 3 BG035761 3.92949 ABCA1 ATP-binding cassette, subfamily A (ABC1), member 1 NM_005502 3.88478 SLC2A14 Solute carrier family 2 (facilitated glucose transporter), member 14 AA778684 3.65108 NPC1 Niemann–Pick disease, type C1 NM_000271 3.42023 ZFP36 Zinc-finger protein 36, C3H type, homolog (mouse) NM_003407 3.41924 UGCG UDP-glucose ceramide glucosyltransferase NM_003358 3.40843 EFNB2 Ephrin-B2 BF001670 3.2662 SOD2 Superoxide dismutase 2, mitochondrial X15132 3.07171 CEBPD CCAAT/enhancer-binding protein (C/EBP), delta NM_005195 3.01762 PFKFB3 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 NM_004566 2.99052 ATF3 Activating 3 NM_001674 2.8937 MAFF v-Maf musculoaponeurotic fibrosarcoma oncogene homolog F (avian) AL021977 2.87145 LDLR Low-density lipoprotein receptor (familial hypercholesterolemia) NM_000527 2.82377 AGT Angiotensinogen (serine (or cysteine) proteinase inhibitor, clade A (alpha-1 NM_000029 2.82277 antiproteinase, antitrypsin), member 8) AKAP12 A kinase (PRKA) anchor protein (gravin) 12 AB003476 2.74377 PPAP2B Phosphatidic acid phosphatase type 2B AB000889 2.73372 ADM Adrenomedullin NM_001124 2.63547 KIAA1237 KIAA1237 protein AI148659 2.62669 EFNA1 Ephrin-A1 NM_004428 2.59892 GFPT2 Glutamine-fructose-6-phosphate transaminase 2 NM_005110 2.59264 ICAM1 Intercellular adhesion molecule 1 (CD54), human rhinovirus receptor NM_000201 2.50961 BCL6 B-cell CLL/lymphoma 6 (zinc-finger protein 51) NM_001706 2.50233 C8FW Phosphoprotein regulated by mitogenic pathways NM_025195 2.49763 FEM1C Fem-1 homolog c (Caenorhabditis elegans) AI862658 2.49356 SAMD4 Sterile alpha motif domain containing 4 AB028976 2.31359 SERPINB3 Serine (or cysteine) proteinase inhibitor, clade B (), member 3 BC005224 2.26622 SFN Stratifin X57348 2.26209 TEAD4 TEA domain family member 4 U63824 2.20177 JUN v-jun sarcoma virus 17 oncogene homolog (avian) NM_002228 2.17961 C3 Complement component 3 NM_000064 2.15312 DUSP5 Dual specificity phosphatase 5 U16996 2.11909 FOSL1 FOS-like antigen 1 BG251266 2.10867 BCL3 B-cell CLL/lymphoma 3 NM_005178 2.10195 FLNB Filamin B, beta (actin-binding protein 278) AV712733 2.09461 TIEG2 TGFB inducible early growthresponse 2 AA149594 2.083 CTSB Cathepsin B NM_001908 2.0828 THBS1 Thrombospondin 1 NM_003246 2.0796 KIAA0146 KIAA0146 protein AV655640 2.06906 SLC4A7 Solute carrier family 4, sodium bicarbonate cotransporter, member 7 AF053755 2.06847 STAT2 Signal transducer and activator of transcription 2, 113 kDa S81491 2.06655 TIP47 Cargo selection protein (mannose 6 phosphate receptor-binding protein) NM_005817 2.0665

The top 50 genes ranked by fold change identified by SAM as being upregulated by Stat3C

number of novel genes identified as being differentially Chronic inflammation of the mouse lung results in Stat3 regulated. Table 4 shows the QPCR results for 47 activation selected genes that are identified by SAM as being either upregulated or repressed by Stat3 at levels greater or Our microarray data demonstrate that Stat3 activation equal to 1.5-fold. There is a good agreement between results in activation of a genetic program involved in SAM and QPCR withregard to genes identified as being wound healing characterized by upregulation of genes differentially expressed in cells overexpressing Stat3C. involved in cell invasion, chemotaxis, angiogenesis, and Notable exceptions were SERPINA3 and SERPINB4, blood coagulation. To determine if Stat3 activation where QPCR identified extremely large fold changes occurs as part of the normal wound healing process in compared to microarray analysis. the lung, we used a previously validated model of mouse

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3400 Table 2 Highest repressed genes (50) Symbol Gene name Accession Fold change

IFIT1 Interferon-induced protein withtetratricopeptide repeats 1 NM_001548 0.09117 MX1 Myxovirus (influenza virus) resistance 1, interferon-inducible protein p78 (mouse) NM_002462 0.18045 MDA5 Melanoma differentiation associated protein-5 NM_022168 0.2016 HSPA6 Heat shock 70 kDa protein 6 (HSP70B0) NM_002155 0.20819 RIG-I DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide NM_014314 0.21216 TRIM22 Tripartite motif-containing 22 AA083478 0.22661 G1P2 Interferon, alpha-inducible protein (clone IFI-15K) NM_005101 0.22809 OAS1 20,50-oligoadenylate synthetase 1, 40/46 kDa NM_002534 0.26107 IFI44 Interferon-induced protein 44 NM_006417 0.2746 IFI27 Interferon, alpha-inducible protein 27 NM_005532 0.28058 OASL 20-50-oligoadenylate synthetase-like NM_003733 0.28894 IFIT4 Interferon-induced protein withtetratricopeptide repeats 4 NM_001549 0.30546 OAS2 20-50-oligoadenylate synthetase 2, 69/71 kDa NM_016817 0.33235 FLJ20035 Hypothetical protein FLJ20035 NM_017631 0.33684 CEB1 Cyclin-E-binding protein 1 NM_016323 0.34414 USP18 Ubiquitin specific protease 18 NM_017414 0.34625 TRIM14 Tripartite motif-containing 14 NM_014788 0.37021 IFITM1 Interferon induced transmembrane protein 1 (9–27) AA749101 0.39253 DNAJB4 DnaJ (Hsp40) homolog, subfamily B, member 4 BG252490 0.42397 SP110 SP110 nuclear body protein AA969194 0.42548 RI58 Retinoic acid- and interferon-inducible protein (58kDa) N47725 0.42724 IRF7 Interferon regulatory factor 7 NM_004030 0.43228 OAS3 20-50-oligoadenylate synthetase 3, 100 kDa NM_006187 0.45127 TNFSF10 Tumor necrosis factor (ligand) superfamily, member 10 NM_003810 0.46383 STAT1 Signal transducer and activator of transcription 1, 91 kDa BC002704 0.46874 FLJ38348 Hypothetical protein FLJ38348 AV755522 0.48173 TRIM5 Tripartite motif-containing 5 AF220028 0.49974 LAP3 Leucine aminopeptidase 3 NM_015907 0.50196 IFI35 Interferon-induced protein 35 BC001356 0.50389 FLJ22693 Hypothetical protein FLJ22693 NM_022750 0.50852 HSPA1A Heat shock 70 kDa protein 1A NM_005345 0.52195 FLJ20073 Hypothetical protein FLJ20073 NM_017654 0.52783 AQP3 Aquaporin 3 N74607 0.53447 FLJ20637 Hypothetical protein FLJ20637 NM_017912 0.54366 PGRMC2 membrane component 2 NM_006320 0.5443 HSPA1B Heat shock 70 kDa protein 1B NM_005346 0.55049 PDK4 Pyruvate dehydrogenase kinase, isoenzyme 4 NM_002612 0.55581 HIS1 HMBA-inducible AW193511 0.55584 BCL2L13 BCL2-like 13 (apoptosis facilitator) NM_015367 0.5583 PCTAIRE2BP Tudor repeat associator withPCTAIRE 2 AW129593 0.55974 LAMP3 Lysosomal-associated membrane protein 3 NM_014398 0.5611 HPGD Hydroxyprostaglandin dehydrogenase 15-(NAD) J05594 0.56132 KIAA0342 KIAA0342 gene product AA035414 0.56794 C14orf2 14 open reading frame 2 NM_004894 0.57381 MAC30 Hypothetical protein MAC30 BF038366 0.58161 GRB14 Growthfactor receptor-bound protein 14 NM_004490 0.58167 PRPF4B PRP4 pre-mRNA processing factor 4 homolog B (yeast) Z25435 0.584 SIAT8D Sialyltransferase 8D (alpha-2, 8-polysialyltransferase) NM_005668 0.58629 TLR3 Toll-like receptor 3 NM_003265 0.5867 ID3 Inhibitor of DNA-binding 3, dominant negative helix–loop–helix protein NM_002167 0.58841

The top 50 genes ranked by fold change identified by SAM as being downregulated by Stat3C

lung chronic inflammation caused by the noncarcino- withthincytoplasmic extensions, and plump type II genic food additive butylated hydroxytoluene (BHT) pneumocytes that bulge into the lumens of the alveolar (Bauer et al., 2001a). Administration of BHT causes sacs. A few pulmonary macrophages are present in the alveolar type I cell necrosis followed by type II cell interalveolar septa or alveolar sacs. No noticeable hyperplasia as a compensatory mechanism. This is changes were identified in tissues during weeks 1 and 2 followed by a process of wound healing characterized after treatment, but by 3–4 weeks after treatment, lungs by cellular proliferation, angiogenesis, and inflamma- have moderate to marked capillary engorgement, inter- tory cell infiltrate into the alveoli. Figure 2a shows the stitial and alveolar edema, and alveolar hemorrhage. morphological change of chronic lung injury by BHT. There is a moderate to marked increase in the cellularity Lungs of control mice have terminal bronchioles lined of the thickened alveolar septa, comprised of hyper- by numerous simple columnar, dome-shaped epithelial trophic and hyperplastic cuboidal type II pneumocytes cells (Clara cells) withabundant apical cytoplasm. with round hyperchromatic nuclei, and a moderate Alveoli are composed of flattened type I pneumocytes number of pulmonary macrophages in alveolar septa

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3401 Table 3 Stat3 target genes grouped into functional classes Gene Accession Fold change Gene Accession Fold change

Cell cycle CEBPD NM_005195 3.01762 MYC NM_002467 1.72323 SFN BC000329 2.26209 CDKN1B BC001971 0.63824

Apoptosis IGFBP1 NM_000596 4.11986 BCL2A1 NM_004049 1.75704 NFIL3 NM_005384 1.95134 CASP4 U25804 1.59452 ATF3 NM_001674 2.89370 STK17A AW194730 1.53870 TRAF4 NM_004295 1.56184 PDCD2 BF446180 0.58867 GADD45B NM_015675 2.01087

Signaling BCL3 NM_005178 2.10195 TIEG2 AA149594 2.08300 SOCS3 BG035761 3.92949 RAB27A U38654 1.62739 JUNB NM_002229 1.92353 JUN NM_002228 2.17961 DUSP5 U16996 2.11909 DUSP1 NM_004417 1.82851 FOSL1 BG251266 2.10867 LYN NM_002350 1.52019 FOSL2 N36408 1.90471 EGFR AW157070 1.51298 IL6R NM_000565 1.71201 JAK2 AF001362 1.51837 TNFRSF12A NM_016639 1.67420 IRF1 NM_002198 1.61695 IL1R1 NM_000877 1.82161

Angiogenesis ADM NM_001124 2.63547 EPAS1 AF052094 1.56013 ANGPTL4 NM_016109 1.99292

Coagulation FGA NM_000508 5.41780 PLAUR U08839 1.82458 FGB BG545288 5.86271 SERPINE1 NM_000602 1.81434 FGG NM_000509 6.31786 THBS1 NM_003246 2.07960 PLAU K03226 1.71778 TFPI2 L27624 1.57622 THBD NM_000361 1.95048

Lipids ABCA1 NM_005502 3.88478 NPC1 NM_000271 3.42023 LDLR NM_000527 2.82377 VLDLR L22431 1.71423 UGCG NM_003358 3.40843 HM74 NM_006018 1.60217

Invasion FGA NM_000508 5.41780 SERPINB3 BC005224 2.26622 FGB BG545288 5.86271 COL5A1 AI983428 1.65357 FGG NM_000509 6.31786 SERPINB4 U19557 1.91011 CTSB NM_001908 2.08280 PLAUR U08839 1.82458 SERPINE2 AL541302 1.64229 CXCL2 M57731 1.67439 CTSL NM_001912 1.79108 THBD NM_000361 1.95048 CCL2 S69738 4.11116 ANGPTL4 NM_016109 1.99292 RAB27A U38654 1.62739 SERPINB1 NM_030666 1.75773 NDRG1 NM_006096 2.05333 CEACAM1 X16354 1.66495 PLAU K03226 1.71778 SERPINE1 NM_000602 1.81434 ICAM1 NM_000201 2.50961 THBS1 NM_003246 2.07960 PLEKHC1 Z24725 1.75166 SERPINB9 BC002538 1.52483 FLRT3 NM_013281 1.75891 TFPI2 L27624 1.57622 SERPINA3 NM_001085 1.82817

Inflammation/immunity SOD2 X15132 3.07171 EFNA1 NM_004428 2.59892 BCL3 NM_005178 2.10195 ICAM1 NM_000201 2.50961 TRAF4 NM_004295 1.56184 IL4R NM_000418 1.91542 GG2-1 BC005352 1.59366 IL6R NM_000565 1.71201 CXCL2 M57731 1.67439 IL1R1 NM_000877 1.82161 EFNB2 BF001670 3.26620 C1S M18767 1.62531 LBP NM_004139 4.54253 C1R AL573058 1.64849 CCL2 S69738 4.11116 IRF1 NM_002198 1.61695 C3 NM_000064 2.15312

Transcription factors BCL6 NM_001706 2.50233 FOSL2 N36408 1.90471 CEBPD NM_005195 3.01762 MYC NM_002467 1.72323 BCL3 NM_005178 2.10195 PLAGL1 BG547855 1.79162 MAFF AL021977 2.87145 CITED2 AF109161 1.77378 NFIL3 NM_005384 1.95134 JUN NM_002228 2.17961 ATF3 NM_001674 2.89370 GATA6 D87811 1.65210 JUNB NM_002229 1.92353 ATBF1 NM_006885 1.57761 FOSL1 BG251266 2.10867 MAFK NM_002360 1.69499 ETS2 NM_005239 1.94842

Interferon signaling IFIT1 NM_001548 0.09117 SP110 AA969194 0.42548 MX1 NM_002462 0.18045 RI58 N47725 0.42724 MDA5 NM_022168 0.20160 IRF7 NM_004030 0.43228

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3402 Table 3 (continued ) Gene Accession Fold change Gene Accession Fold change

TRIM22 AA083478 0.22661 OAS3 NM_006187 0.45127 G1P2 NM_005101 0.22809 TNFSF10 NM_003810 0.46383 OAS1 NM_002534 0.26107 STAT1 BC002704 0.46874 IFI44 NM_006417 0.27460 TRIM5 AF220028 0.49974 IFI27 NM_005532 0.28058 IFI35 BC001356 0.50389 OASL NM_003733 0.28894 PRKR NM_002759 0.58871 IFIT4 NM_001549 0.30546 G1P3 NM_022873 0.60681 OAS2 NM_016817 0.33235 MX2 NM_002463 0.62577 CEB1 NM_016323 0.34414 PLSCR1 NM_021105 0.64449 TRIM14 NM_014788 0.37021 HLA-DMA X76775 0.64550 IFITM1 AA749101 0.39253 ISG20 NM_002201 0.65778

Some genes are grouped into multiple functional classes based on diverse functions of the gene product

and sacs. Figure 2b shows corresponding changes in Stat3 in pathways that regulate cellular proliferation, apop- activation from identical tissues. Stat3 activity is in- tosis, tumor angiogenesis, tumor invasion/metastasis, creased 24 hfollowing exposure to BHT and thenreturns and evasion of an immune response (Yu and Jove, to baseline levels by week 2. By weeks 3–4, there is again 2004). Tissue-specific Stat3 gene deletion mouse models an increase in Stat3 activity that corresponds to the time have also defined an important role of Stat3 in tissue of maximal wounding of the lung epithelium. These data injury and wound healing in epithelial cells of the lung demonstrate that Stat3 is activated during the normal and skin (Sano et al., 1999; Gao et al., 2004; Hokuto lung wound healing process following chronic injury. et al., 2004; Severgnini et al., 2004). In lung epithelial cells, Stat3 is activated in acute lung injury and loss Stat3 activation at the invading edge of lung tumors of Stat3 compromises survival following hyperoxia (Gao et al., 2004; Hokuto et al., 2004; Severgnini et al., We identified a number of genes upregulated by Stat3 2004). Our results identify previously uncharacte- involved in tumor cell invasion including uPA, uPAR, rized genes upregulated by Stat3 that are implicated in and cathepsins B and L. This suggests that Stat3 cell invasion and metastasis as well as pathways activation in tumors may be involved in tumor cell involved in angiogenesis and blood coagulation. We invasion and migration. Based on these results, we also identified a number of interferon-induced genes postulated that Stat3 activation could be important at downregulated by Stat3. Most importantly, our results the leading edge of tumors invading adjacent stromal provide a molecular view that links both wound healing tissues. We examined early stage non-small-cell lung pathways and oncogenic pathways to Stat3. These cancers to determine the localization of Stat3 activity by findings provide a molecular basis for the seminal immunohistochemistry using an antibody that recog- observation of similarities between wound healing and nizes tyrosine-phosphorylated Stat3 (pStat3) (Mora tumor development in that tumors are ‘wounds that do et al., 2002). As shown in Figure 3, we identified cases not heal’ (Dvorak, 1986). Both wound repair and cancer where the bulk of the tumor contained little or no pStat3 are characterized by cell proliferation, remodeling of activity while the edge of the tumor adjacent to stromal extracellular matrix, cell invasion and migration, new tissues contained abundant cytoplasmic and nuclear blood vessel formation, and modulation of blood pStat3 staining. Other tumors demonstrated pStat3 coagulation. Activation of Stat3 by either oncogenic activity throughout the tumor (data not shown). We signals or wound healing signals activates a common were unable to demonstrate activation of ERK in these gene program that affects processes common to both tumors using antibodies that recognize phosphorylated events (Figure 4). However, a crucial difference between ERK proteins, suggesting that the activation of Stat3 is normal wound healing, which involves transient Stat3 not a generalized finding of activated signaling (data not activation, and cancer is that Stat3 is persistently shown). These results suggest that Stat3 activity may not activated in cancer (Yu and Jove, 2004). Thus, be homogenous throughout lung tumors but rather can continuous expression of wound healing genes by be concentrated in tumor cells adjacent to nontumor persistently activated Stat3 in tumor cells may be stromal tissues. These results and the identification of essential for malignant progression and may be pre- Stat3-dependent genes (suchas uPA and uPAR) dictive of outcome. A gene expression program that involved in cell invasion/migration suggest a role for reflects wound healing was found to be expressed in Stat3 in tumor cell invasion. human tumors and was predictive of poor outcome in multiple tumor types (Chang et al., 2004). Given recent results demonstrating a critical role of Stat3 in the Discussion development of skin cancer, our results may also suggest additional mechanisms of tumor promotion by Stat3 Considerable evidence has demonstrated a critical including tissue invasion and angiogenesis (Chan et al., role for Stat3 in the regulation of genes participating 2004).

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3403 Table 4 Stat3 target genes identified by QPCR Microarray Fold change QPCR

Mean s.d. C.I. (95%)

Cell cycle CEBPD 3.02 3.57 1.17 [2.24, 4.90] SFN 1.96 1.72 0.54 [1.11, 2.33]

Apoptosis IGFBP1 4.12 9.35 2.20 [6.31, 12.40] ATF3 2.89 3.29 0.36 [2.79, 3.79] BCL2A1 1.76 1.28 0.60 [0.46, 2.11] PDCD2 0.59 0.44 0.01 [0.43, 0.45]

Signaling BCL3 2.10 1.33 0.05 [1.28, 1.39] IL6R 1.71 5.05 0.04 [5.00, 5.10] EGFR 1.51 1.90 0.60 [1.08, 2.73]

Angiogenesis ADM 2.64 3.70 0.29 [3.37, 4.03] ANGPTL4 1.99 4.03 0.09 [3.90, 4.15] EPAS1 1.56 1.43 0.32 [1.08, 1.79]

Coagulation PLAU 1.65 2.49 0.71 [1.69, 3.29] PLAUR 1.82 1.63 0.09 [1.51, 1.75] THBD 1.95 1.72 0.41 [1.15, 2.29] SERPINE1 1.81 2.01 0.28 [1.69, 2.32] THBS1 2.08 2.29 0.22 [2.04, 2.54] TFPI2 1.56 3.58 1.44 [1.58, 5.57]

Lipids ABCA1 3.88 5.42 0.25 [5.08, 5.76] LDLR 2.82 3.52 0.96 [2.43, 4.61] NPC1 3.42 3.48 0.73 [2.65, 4.30] VLDLR 1.71 1.51 0.42 [1.03, 1.99]

Invasion CTSB 2.06 2.82 0.24 [2.55, 3.09] SERPINE2 1.64 1.63 0.29 [1.29, 1.96] CTSL 1.79 2.14 0.29 [1.81, 2.46] CCL2 4.11 2.95 0.67 [2.03, 3.87] NDRG1 2.05 2.80 0.56 [2.16, 3.44] SERPINA3 1.83 4314.67 3554.13 [292.87, 8336.47] SERPINB3 2.14 4.29 4.80 [À1.14, 9.72] SERPINB4 1.91 753.85 298.67 [415.88, 1091.83] CXCL2 1.67 2.08 0.34 [1.70, 2.46] CEACAM1 1.66 1.16 0.39 [0.71, 1.60]

Inflammation/immunity SOD2 3.03 4.38 0.87 [3.40, 5.37] EFNB2 3.27 7.26 2.18 [4.79, 9.72] EFNA1 2.60 4.82 0.13 [4.68, 4.97] IRF1 1.62 1.99 0.59 [1.17, 2.82] ICAM1 2.51 24.29 6.65 [15.08, 33.50]

Transcription factors BCL6 2.50 2.72 0.54 [2.11, 3.32]

Interferon signaling OAS1 0.26 0.25 0.18 [0.05, 0.45] OAS2 0.33 0.04 0.03 [0.01, 0.08] OAS3 0.45 0.29 0.04 [0.24, 0.33] OASL 0.29 0.14 0.06 [0.07, 0.21] IRF7 0.43 0.37 0.06 [0.30, 0.43] TNFSF10 0.46 0.05 0.01 [0.03, 0.07] STAT1 0.47 0.46 0.17 [0.26, 0.66] PRKR 0.59 0.36 0.07 [0.28, 0.44] PLSCR1 0.64 0.51 0.10 [0.39, 0.62]

QPCR was performed as described in Materials and methods from Ad-GFP- and Ad-Stat3C-infected A549 cells. Fold changes identified by both microarray and QPCR are shown. s.d. and 95% confidence intervals (C.I.) of the fold change for the three QPCR replicates are shown

Based on recent studies demonstrating enhanced et al., 2003; Vultur et al., 2004). Stat3 may also act as a Stat3 activation withincreased cell–cell contact or in- sensor of wounding and similarly regulate genetic creased confluence, we hypothesize that Stat3 may serve programs necessary for epithelial cell migration required as a sensor of tumor cell contact and upregulate genes for proper wound healing. The serine protease uPA and necessary for cell invasion and migration (Steinman its receptor uPAR are important for tumor cell invasion

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3404

Figure 3 Stat3 activation at the leading edge of lung cancer. Non- small-cell lung cancers taken from surgically resected specimens from two distinct cases were assessed for Stat3 activity using pStat3 immunohistochemistry. Top, H&E; bottom, pStat3

Figure 2 Stat3 activation following chronic inflammation of the mouse lung. (a) Formalin-fixed tissues from control and BHT- treated mice were examined for morphological changes following Figure 4 Common events regulated by Stat3 in cancer and wound weekly administration of BHT. Representative H&E sections of healing. A model of Stat3-dependent pathways common to both tissues from control mice and mice following 4 weeks of BHT normal wound healing and oncogenesis exposure are shown. (b) Stat3 DNA-binding activity was assayed using STAT EMSA. Results are shown for two animals for each individual time point. C, Control mice; Wk1–4, mice exposed to weekly BHT injections; D6 and D10, mice killed 6 and 10 days, lization of skin wounds by keratinocytes (Romer et al., respectively, following last BHT injection; IL-6, IL-6 exposed A549 1994). Similarly, cathepsin family members are pro- nuclear extract for positive control; SS, C-20 Â Stat3 Supershift. (c) teases implicated in inflammation, tumor invasion, and Graphical representation of Stat3 intensity following analysis using metastasis through regulation of proteolysis, protein ImageQuant. Mean values of Stat3 intensities for the two control mice were set to a value of 1. Stat3 intensity from individual processing, and matrix degradation (Rao, 2003). samples represents fold change compared to control mean. Results Upregulation of fibrinogen by Stat3 in lung epithelia are again shown for two animals for each individual time point is of interest since fibrinogen expression in the lung has been linked to sustained adhesion and/or survival of circulating tumor cells (Palumbo et al., 2000; Palumbo (Rao, 2003). Binding of uPA to uPAR on the cell et al., 2002). These results suggest that upregulation of surface directs proteolytic activity resulting in extra- Stat3 in nonmalignant stromal or epithelial cells within cellular matrix degradation. Eliminating uPAR function an organ may allow for enhanced tumor cell growth and in lung carcinoma cells results in reduced tumor cell metastasis through modulation of factors controlling invasion and metastasis (Lakka et al., 2001). There is cell adhesion, migration, angiogenesis, and chemotaxis. also evidence for uPAR activation during re-epithelia- Inflammatory conditions that upregulate Stat3 activity

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3405 in lung epithelial cells could conceivably favor the tissue dependent genes essential for wound healing are also milieu conducive for invasion and metastasis of circulat- central to cancer formation and progression. ing cancer cells. Confirmation of this hypothesis would In summary, we found that Stat3 serves to connect suggest that targeting normal stromal cell Stat3 signal- seemingly diverse pathways to regulate inflammation in ing could negate nontumor cell-specific events that aid in the context of oncogenesis. Since Stat3 is upregulated by tumor cell metastasis. bothtissue injury and oncogenic signals, our results link In addition to the genes induced by Stat3, we also common signaling pathways to genetic programs used in identified a large group of interferon-dependent genes bothwound response and cancer formation. Given the repressed by Stat3 (Table 2). In the context of normal link between inflammation and cancer, the regulation of wound healing, Stat3 activation and downregulation of suchcommon pathwaysby Stat3 may be clinically an IFNg-like response may be important in reducing relevant for bothchemoprevention and treatment of tissue damage resulting from proinflammatory cytokines established tumors. Of fundamental importance, our elicited by invading immune cells. This finding is findings provide a molecular basis for understanding consistent with previous studies showing that cells cancer as a deregulation of normal wound healing exposed to IL-6 require Stat3 to downregulate Stat1 processes. activity and prevent an IFNg-like response (Costa- Pereira et al., 2002). Selective loss of Stat3 in endothelial cells results in expanded inflammatory response and Materials and methods tissue damage in response to a systemic inflammatory response (Kano et al., 2003). These conditional Stat3- Adenoviral vectors null animals demonstrated increased levels and pro- Ad-Stat3C was constructed by subcloning Stat3C into the longed periods of proinflammatory cytokine produc- AdTrack-CMV plasmid used to construct recombinant ade- tion. Increasing evidence also demonstrates that tumor noviruses (He et al., 1998). This plasmid also coexpresses GFP cells become insensitive to the tumor suppressor actions through a cytomegalovirus (CMV) promoter. Viral stocks of IFNg through both genetic and epigenetic mechan- were created, purified and titered as previously described (Song isms (Saunders et al., 1994; Kaplan et al., 1998; Shou et al., 2003). Cell were infected withvarying concentrations of et al., 2002; Kulaeva et al., 2003). Interferons have the adenovirus, detailed in eachexperiment, placed directly into ability to negatively regulate cell invasion and metas- the medium of cells and incubated for the indicated times. tasis, cell cycle entry, and induce apoptosis (Bromberg Viral infection was confirmed by visually observing GFP et al., 1996; Chawla-Sarkar et al., 2003). Our studies as expression in infected cells. well as those discussed here suggest that Stat3 activity may contribute to disruption of IFNg-mediated tumor Cell lines and cell culture suppression. Additionally, the genes identified as being A549 cell lines were obtained from ATCC. Cells were grown in downregulated by Stat3 expand the list of proinflam- RPMI-1640 medium supplemented with2 m ML-glutamine matory cytokines previously found to inhibit dendritic (Gibco) supplemented with10% fetal calf serum (FCS) cell functional maturation (Wang et al., 2004). There- (Hyclone). A549 lung cells were seeded overnight on 10 cm fore, suppression of interferon-dependent gene expres- plates in 10% FCS and the next day the media was removed sion by Stat3 may be a required event for bothproper and replaced withcomplete media containing 0.25% FCS. wound healing and escape of tumor cells from the tumor After 48 hof serum deprivation, thecells were infected with either Ad-GFP or Ad-Stat3C at an MOI of 50. Cells were suppressor functions of interferons. counted using a hemocytometer and infected with adenoviral Finally, additional Stat3-regulated genes are involved vectors in Optimem media (Gibco) for the times indicated. To in the duality of wound healing and oncogenesis. control for effects of adenoviral infection on gene expression, MnSOD expression in the lungs has been associated we compared Ad-Stat3C-expressing cells to cells infected with withreduced pulmonary radiation sensitivity, decreased control adenoviral vector termed Ad-GFP. alveolitis, and improved animal survival following whole lung irradiation (Epperly et al., 2000). It is also Nuclear extract preparation and electrophoretic mobility shift a survival factor for cancer cells and can increase the assays (EMSA) resistance of cancer cells to chemotherapy (Hur et al., STAT DNA-binding assays were performed as described 2003). Inhibition of CXC chemokine ligands/CXCR2 previously (Song et al., 2003). Protein–DNA complexes were interaction using CXCR2 gene knockout mice reduced resolved by nondenaturing polyacrylamide gel electrophoresis neutrophil sequestration and lung injury as well as (PAGE) and detected by autoradiography. enhanced animal survival following hyperoxia com- pared withwild-type mice (Sue et al., 2004). CXC Oligonucleotide microarray analysis chemokines acting through the CXCR2 receptor also enhance tumor cell invasion, migration, and metastasis We followed MIAME (Minimum Information About a (Reiland et al., 1999). CCL2, also known as macrophage Microarray Experiment) guidelines for the presentation of our results (Brazma et al., 2001). To control for biological and chemotactic protein (MCP-1), has a critical role in experimental variation, five 10 cm dishes of cells were infected alveolar epithelial cell wound healing and has also been witheitherAd-GFP or Ad-Stat3C (10 individual plates). At shown to be important in lymphoma cell invasion 24 hafter infection, total RNA was collected and used as the (Wakabayashi et al., 1995; Christensen et al., 2004). RNA source for microarray analysis. RNA from each These results further support the concept that Stat3- individual plate was processed for analysis on its own

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3406 individual gene array (five arrays for Ad-GFP, five arrays for Quantitative RT-PCR analysis Ad-Stat3C). Total RNA was isolated from cells withTrizol RNA isolated from A549 cells infected witheitherAd-GFP or reagent and then purified using Qiagen RNA prep kits Ad-Stat3C was pooled to create one sample representing Ad- according to the supplier’s instructions. Total RNA (5 mm) GFP-infected cells and one sample representing Ad-Stat3C- served as the mRNA source for microarray analysis. Poly(A) infected cells. The RNA pools were run on RNA 6000 Nano RNA was converted to cDNA, amplified, and labeled with Chips (Agilent, Palo Alto, CA, USA) and analysed on the biotin following the established procedures (Van Gelder et al., Agilent 2100 Bioanalyzer (Agilent, Palo Alto, CA, USA) to 1990). The arrays used were the Affymetrix confirm their integrity and concentration. Reverse transcrip- U133A GeneChips. These chips contain 22 214 probe sets that tion reactions were carried out using Sensiscript RT and oligo- target known and suspected genes as well as a number of dT (Qiagen, Valencia, CA, USA) witha RNA concentration suspected splice variants. Eachgene is represented by a series of 15 ng/ml. Quantitative Real-Time PCR reactions contained of oligonucleotides that are identical in sequence to the gene, 30 ng of cDNA per port (or per 50 reactions). All reactions and oligonucleotides that contain a homomeric (base transver- were run in triplicate on TaqMan Custom Low Density Array- sion) mismatch at the central base position of the oligomer is Micro Fluidic Cards (Applied Biosystems, Foster City, CA, used for measuring cross hybridization. These arrays offer a USA). Product numbers for eachgene assayed are available in comprehensive survey of human genes as registered in Supplemental Information. Fluorescent emission was recorded GenBank, UniGene (build 133), and the Institute for Genomic in real-time (ABI prism 7900HT; Applied Biosystems, Foster Research(TIGR) databases. Hybridization, staining, and City, CA, USA). Gene expression profiling was completed scanning of the chips followed the prescribed procedure using the Comparative Ct method of relative quantification outlined in the Affymetrix technical manual as previously (User Bulletin #2 – ABI PRISM 7700 Sequence Detection described (Warrington et al., 2000). Hybridized chips were System, Applied Biosystems, Foster City, CA, USA). For each scanned, inspected for hybridization artefacts, and then gene, relative RNA quantities were normalized to the analysed using the GeneChip Operating Software (GCOS) endogenous control, 18s rRNA. EachStat3C replicate was version 1.1, which produces probe-level data in a series of CEL normalized to eachGFP pool and theaverage relative quantity files, one for eachGeneChip. (RQ) is reported for each gene. The mean fold changes were Affymetrix probe-level data in CEL files for five control calculated along withs.d. and 95% confidence intervals of the (Ad-GFP) chips and five experimental (Ad-Stat3C) chips were three replicates. normalized using the affy package in the R statistical language and environment version 1.9.0 (www.bioconductor.org). Probe-level data were stored in an R working directory and Animal studies normalized by the Robust Multichip Average (RMA) method, available as part of the affy package. This consists of a We used an established mouse model of chronic lung background adjustment, quantile normalization, and the inflammation using Balb/c mice chronically exposed to four median-polishsummary method,producing a single normal- weekly BHT injections (Bauer et al., 2001b). The use of mice ized expression set for all 10 chips (Bolstad et al., 2003; Irizarry was approved by the University of South Florida Institutional et al., 2003a, b). Animal Care and Use Committee. Male Balb/C mice 6–8 The expression sets were then exported to a Microsoft Excel weeks of age were purchased from Charles River Laboratory. spreedsheet, and formatted for analysis by the SAM add-in Balb/C mice were injected withan initial BHT dose of 150 mg/ tool for Excel (Tusher et al., 2001). Differentially expressed kg followed by three weekly 200 mg/kg i.p. injections; control genes between Ad-Stat3C and Ad-GFP were identified using mice received corn oil injection. Two mice were euthanatized SAM. This method accepts normalized expression sets and by CO2 inhalation 24 h after every injection, as well as 6 and 10 identifies statistically significant changes in gene expression by days after the final injection. Whole lungs were washed with assigning eachgene a score based on its changein expression ice-cold PBS via injection into the right ventricle to remove relative to the standard deviation of repeated measurements. residual blood and snap frozen in liquid nitrogen. A separate SAM uses permutations of empirical measurements to sample of lung tissue was snap fixed in formalin and mounted estimate the false discovery rate (FDR) for the called list in in paraffin for sectioning. the form of a 90% confidence interval. The options selected for the SAM analysis were: (1) Response type: two-class, unpaired data (class 1 – GFP – Human tumor studies five chips, class 2 – STAT3C – five chips); (2) Data logged: Tumor tissues were obtained from patients undergoing logged (base 2); (3) Weblink Option: Accession number; (4) surgical resection of early stage non-small-cell lung cancer as Number of permutations: 200; (5) Imputation engine: N/A – part of a clinical protocol approved by the University of South no missing data in our experiment; (6) Random number seed: Florida Institutional Review Board. The tumor specimen was generate random number seed. This produces a list of both fixed in formalin and processed for routine histological upregulated and downregulated significant genes along with examination and immunohistochemical analysis. Immunos- options to specify fold-change constraints. The false discovery taining for phospho-Stat3 was performed using a rabbit anti- rate is presented as a 90% confidence interval (as opposed to human polyclonal antibody (Phospho-Tyr705-Stat3; Cell the median), and we accepted a rate of 1/100 as the upper limit Signaling, Beverly, MA, USA) as previously described; as of this interval, indicating a false discovery rate of at most 1%. negative controls, rabbit immunoglobulins were used to We used Onto-Express to aid in understanding functional replace primary antibody (Mora et al., 2002). roles for genes identified as differentially expressed (Khatri et al., 2002; Khatri et al., 2004). Onto-Express is a tool designed to mine available functional annotation data and Acknowledgements identify relevant and significant functional biological pro- We thank Dr Robert Engleman for assistance with patholo- cesses. Gene lists were submitted to Onto-Express after being gical examination of the mouse tissues and Dr Steven Eschrich converted to TXT files. In addition, eachgene was researched for review of the microarray analysis. We thank Dr Jacqueline by manual examination of published literature using Pubmed. Bromberg for providing Stat3C. We thank Gerold Bepler for

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3407 critical review of the manuscript, Zhao Chen for statistical & Research Institute. This work was partially funded by the advice, and Rebecca Alexander for administrative assistance. Chiles Endowment Biomedical Research Program of the This work has been supported in part by the Molecular Florida Department of Health(EBH), by a Department of Imaging Core, Molecular Biology and Sequencing Core, and Defense National Functional Genomics Grant (RJ), and by the DNA Microarray Core at the H Lee Moffitt Cancer Center the H Lee Moffitt Cancer Center.

References

Bauer AK, Dwyer-Nield LD, Hankin JA, Murphy RC and He TC, Zhou S, da Costa LT, Yu J, Kinzler KW Malkinson AM. (2001a). Toxicology, 169, 1–15. and Vogelstein B. (1998). Proc. Natl. Acad. Sci. USA, 95, Bauer AK, Dwyer-Nield LD, Keil K, Koski K and Malkinson 2509–2514. AM. (2001b). Exp. Lung Res., 27, 197–216. Hokuto I, Ikegami M, Yoshida M, Takeda K, Akira S, Perl Bolstad BM, Irizarry RA, Astrand M and Speed TP. (2003). AK, Hull WM, Wert SE and Whitsett JA. (2004). J. Clin. Bioinformatics, 19, 185–193. Invest., 113, 28–37. Bowman T, Broome MA, Sinibaldi D, Wharton W, Hur GC, Cho SJ, Kim CH, Kim MK, Bae SI, Nam SY, Park Pledger WJ, Sedivy JM, Irby R, Yeatman T, Courtneidge JW, Kim WH and Lee BL. (2003). Clin. Cancer Res., 9, SA and Jove R. (2001). Proc. Natl. Acad. Sci. USA, 98, 5768–5775. 7319–7324. Hutt JA, O’Rourke JP and DeWille J. (2000). J. Biol. Chem., Brazma A, Hingamp P, QuackenbushJ, SherlockG, Spellman 275, 29123–29131. P, Stoeckert C, AachJ, Ansorge W, Ball CA, Causton HC, Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B and Gaasterland T, Glenisson P, Holstege FC, Kim IF, Speed TP. (2003a). Nucleic. Acids Res., 31, e15. Markowitz V, Matese JC, Parkinson H, Robinson A, Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Sarkans U, Schulze-Kremer S, Stewart J, Taylor R, Vilo J Antonellis KJ, Scherf U and Speed TP. (2003b). Biostatistics, and Vingron M. (2001). Nat. Genet., 29, 365–371. 4, 249–264. Bromberg JF, HorvathCM, Wen Z, SchreiberRD Kano A, Wolfgang MJ, Gao Q, Jacoby J, Chai GX, Hansen and Darnell Jr JE. (1996). Proc. Natl. Acad. Sci. USA, 93, W, Iwamoto Y, Pober JS, Flavell RA and Fu XY. (2003). 7673–7678. J. Exp. Med., 198, 1517–1525. Bromberg JF, Wrzeszczynska MH, Devgan G, Zhao Y, Pestell Kaplan DH, Shankaran V, Dighe AS, Stockert E, Aguet M, RG, Albanese C and Darnell JE. (1999). Cell, 98, 295–303. Old LJ and Schreiber RD. (1998). Proc. Natl. Acad. Sci. Chan KS, Sano S, Kiguchi K, Anders J, Komazawa N, USA, 95, 7556–7561. Takeda J and DiGiovanni J. (2004). J. Clin. Invest., 114, Khatri P, Bhavsar P, Bawa G and Draghici S. (2004). Nucleic. 720–728. Acids Res., 32, W449–W456. Chang HY, Sneddon JB, Alizadeh AA, Sood R, West RB, Khatri P, Draghici S, Ostermeier GC and Krawetz SA. (2002). Montgomery K, Chi JT, Rijn Mv M, Botstein D and Brown Genomics, 79, 266–270. PO. (2004). PloS. Biol., 2, E7. Kulaeva OI, Draghici S, Tang L, Kraniak JM, Land SJ and Chawla-Sarkar M, Lindner DJ, Liu YF, Williams BR, Sen Tainsky MA. (2003). Oncogene, 22, 4118–4127. GC, Silverman RH and Borden EC. (2003). Apoptosis, 8, Lakka SS, Rajagopal R, Rajan MK, Mohan PM, Adachi Y, 237–249. DinhDH, Olivero WC, Gujrati M, Ali-Osman F, RothJA, Christensen PJ, Du M, Moore B, Morris S, Toews GB and Yung WK, Kyritsis AP and Rao JS. (2001). Clin. Cancer Paine III R. (2004). Am. J. Physiol. Lung Cell Mol. Physiol., Res., 7, 1087–1093. 286, L68–L72. Levy DE and Darnell Jr JE. (2002). Nat. Rev. Mol. Cell. Biol., Costa-Pereira AP, Tininini S, Strobl B, Alonzi T, Schlaak JF, 3, 651–662. Is’harc H, Gesualdo I, Newman SJ, Kerr IM and Poli V. Lieber M, SmithB, Szakal A, Nelson-Rees W and Todaro G. (2002). Proc. Natl. Acad. Sci. USA, 99, 8043–8047. (1976). Int. J. Cancer, 17, 62–70. Coussens LM and Werb Z. (2002). Nature, 420, 860–867. Mora LB, Buettner R, Seigne J, Diaz J, Ahmad N, Garcia R, Duan HO and Simpson-Haidaris PJ. (2003). J. Biol. Chem., Bowman T, Falcone R, FaircloughR, Cantor A, 278, 41270–41281. Muro-Cacho C, Livingston S, Karras J, Pow-Sang J and Dvorak HF. (1986). N. Engl. J. Med., 315, 1650–1659. Jove R. (2002). Cancer Res., 62, 6659–6666. Epperly MW, Epstein CJ, Travis EL and Greenberger JS. Niu G, Wright KL, Huang M, Song L, Haura E, Turkson J, (2000). Radiat. Res., 154, 365–374. Zhang S, Wang T, Sinibaldi D, Coppola D, Heller R, Ellis Fujitani Y, Nakajima K, Kojima H, Nakae K, Takeda T and LM, Karras J, Bromberg J, Pardoll D, Jove R and Yu H. Hirano T. (1994). Biochem. Biophys. Res. Commun., 202, (2002). Oncogene, 21, 2000–2008. 1181–1187. Palumbo JS, Kombrinck KW, Drew AF, Grimes TS, Gao H, Guo RF, Speyer CL, Reuben J, Neff TA, Hoesel LM, Kiser JH, Degen JL and Bugge TH. (2000). Blood, 96, Riedemann NC, McClintock SD, Sarma JV, Van Rooijen 3302–3309. N, Zetoune FS and Ward PA. (2004). J. Immunol., 172, Palumbo JS, Potter JM, Kaplan LS, Talmage K, Jackson DG 7703–7712. and Degen JL. (2002). Cancer Res., 62, 6966–6972. Grandis JR, Drenning SD, Zeng Q, Watkins SC, Melhem MF, Rao JS. (2003). Nat. Rev. Cancer, 3, 489–501. Endo S, Johnson DE, Huang L, He Y and Kim JD. (2000). Reiland J, Furcht LT and McCarthy JB. (1999). Prostate, 41, Proc. Natl. Acad. Sci. USA, 97, 4227–4232. 78–88. Hanahan D and Weinberg RA. (2000). Cell, 100, 57–70. Reljic R, Wagner SD, Peakman LJ and Fearon DT. (2000). He B, You L, Uematsu K, Matsangou M, Xu Z, He M, J. Exp. Med., 192, 1841–1848. McCormick F and Jablons DM. (2003). Biochem. Biophys. Romer J, Lund LR, Eriksen J, Pyke C, Kristensen P and Res. Commun., 301, 386–391. Dano K. (1994). J. Invest. Dermatol., 102, 519–522.

Oncogene Stat3, wound healing and cancer DJ Dauer et al 3408 Sano S, Itami S, Takeda K, Tarutani M, Yamaguchi Y, Miura Tusher VG, Tibshirani R and Chu G. (2001). Proc. Natl. Acad. H, Yoshikawa K, Akira S and Takeda J. (1999). EMBO J., Sci. USA, 98, 5116–5121. 18, 4657–4668. Van Gelder RN, von Zastrow ME, Yool A, Dement WC, Saunders NA, SmithRJ and Jetten AM. (1994). Am. J. Resp. Barchas JD and Eberwine JH. (1990). Proc. Natl. Acad. Sci. Cell Mol. Biol., 11, 147–152. USA, 87, 1663–1667. Schumann RR, Kirschning CJ, Unbehaun A, Aberle HP, Vultur A, Cao J, Arulanandam R, Turkson J, Jove R, Greer Knope HP, Lamping N, UlevitchRJ and Herrmann F. P, Craig A, Elliott B and Raptis L. (2004). Oncogene, 23, (1996). Mol. Cell. Biol., 16, 3490–3503. 2600–2616. Severgnini M, Takahashi S, Rozo LM, Homer RJ, Kuhn C, Wakabayashi H, Cavanaugh PG and Nicolson GL. (1995). Jhung JW, Perides G, Steer M, Hassoun PM, Fanburg BL, Cancer Res., 55, 4458–4464. Cochran BH and Simon AR. (2004). Am. J. Physiol. Lung. Wang T, Niu G, Kortylewski M, Burdelya L, Shain K, Zhang Cell Mol. Physiol., 286, L1282–L1292. S, Bhattacharya R, Gabrilovich D, Heller R, Coppola D, Shou J, Soriano R, Hayward SW, Cunha GR, Williams PM and Dalton W, Jove R, Pardoll D and Yu H. (2004). Nat. Med., Gao WQ. (2002). Proc. Natl. Acad. Sci. USA, 99, 2830–2835. 10, 48–54. Song L, Turkson J, Karras JG, Jove R and Haura EB. (2003). Warrington JA, Nair A, Mahadevappa M and Tsyganskaya Oncogene, 22, 4150–4165. M. (2000). Physiol. Genomics, 2, 143–147. Steinman RA, Wentzel A, Lu Y, Stehle C and Grandis JR. Yan H, Dobbie Z, Gruber SB, Markowitz S, Romans K, (2003). Oncogene, 22, 3608–3615. Giardiello FM, Kinzler KW and Vogelstein B. (2002). Nat. Sue RD, Belperio JA, Burdick MD, Murray LA, Xue YY, Dy Genet., 30, 25–26. MC, Kwon JJ, Keane MP and Strieter RM. (2004). Yu H and Jove R. (2004). Nat. Rev. Cancer, 4, 97–105. J. Immunol., 172, 3860–3868.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc)

Oncogene