Identification of novel AP-1 target genes in fibroblasts ... - Nature

Oncogene (2004) 23, 7005–7017

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Identification of novel AP-1 target genes in fibroblasts regulated during cutaneous wound healing Lore Florin1, Lars Hummerich2, Bernd Thilo Dittrich1, Felix Kokocinski2, Gunnar Wrobel2, Sabine Gack1, Marina Schorpp-Kistner1, Sabine Werner3, Meinhard Hahn2, Peter Lichter2, Axel Szabowski1 and Peter Angel*,1 1

Division of Signal Transduction and Growth Control, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 280, Heidelberg D-69120, Germany; 2Division of Molecular Genetics, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 280, Heidelberg D-69120, Germany; 3Institute of Cell Biology, ETH Zuerich, Hoenggerberg, Zuerich CH-8093, Switzerland

Mesenchymal–epithelial interactions are increasingly considered to be of vital importance for epithelial homeostasis and regeneration. In skin, the transcription factor AP-1 was shown to be critically involved in the communication between keratinocytes and dermal fibroblasts. After skin injury, the release of IL-1 from keratinocytes induces the activity of the AP-1 subunits c-Jun and JunB in fibroblasts leading to a global change in gene expression. To identify AP-1 target genes in fibroblasts, which are involved in the process of cutaneous repair, we performed gene expression profiling of wild-type, c-junand junB-deficient fibroblasts in response to IL-1, mimicking the initial phase of wound healing. Using a 15K cDNA collection, over 1000 genes were found to be Jun-dependent and additional 300 clones showed IL-1 responsiveness. Combinatorial evaluation allowed for the dissection of the specific contribution of either AP-1 subunit to gene regulation. Besides previously identified genes that are involved in cutaneous repair, we have identified novel genes regulated during wound healing in vivo and showed their expression by fibroblasts on wound sections. The identification of novel Jun target genes should provide a basis for understanding the molecular mechanisms underlying mesenchymal–epithelial interactions and the critical contribution of AP-1 to tissue homeostasis and repair. Oncogene (2004) 23, 7005–7017. doi:10.1038/sj.onc.1207938 Published online 26 July 2004 Keywords: expression profiling; Jun; mesenchyme; microarray; skin

Introduction The skin is the outermost barrier against environmental challenges and protects the body against radiation, dehydration and invading microorganisms. In order to *Correspondence: P Angel; E-mail: [email protected] Received 31 March 2004; accepted 21 May 2004; published online 26 July 2004

maintain its functionality, it is of vital importance that injuries are rapidly detected and efficiently repaired. Keratinocytes at the wound edges respond immediately with the release of preformed interleukin-1 (IL-1), which acts in concert with other soluble factors such as transforming growth factor b (TGF-b) and various chemokines to initiate the repair process (Martin, 1997; Singer and Clark, 1999; Werner and Grose, 2003). Rapid wound closure is subsequently achieved through the coordinated action of a variety of different cell types, including keratinocytes, dermal fibroblasts and cells of the immune system, which are activated and invade the injured area in response to cytokines in the wound fluid (Gillitzer and Goebeler, 2001; Werner and Grose, 2003). Different lines of evidence have suggested a critical role of the transcription factor AP-1 in skin homeostasis (Angel et al., 2001; Li et al., 2003; Zenz et al., 2003). AP1 represents a set of dimeric complexes composed of members of the Jun, Fos and ATF families, which assemble into homo- (Jun/Jun) or heterodimers (Jun/ Fos, Jun/Fra, Jun/ATF; van Dam and Castellazzi, 2001), and synergize or interfere with other transcription factors (Chinenov and Kerppola, 2001; Herrlich, 2001). AP-1 mediates gene expression in response to a variety of stimuli, including growth factors, cytokines, oncoproteins and chemical carcinogens (Angel et al., 2001). In skin, cytoskeletal proteins of the keratin family, markers of keratinocyte differentiation such as loricrin, dermal extracellular matrix components like collagen (I)-a2 as well as matrix metalloproteinases and cytokines are under AP-1 control (Angel et al., 2001). Although many of these genes are known to be induced upon tissue damage and have been reported to play an important role during wound healing, c-Jun expression in keratinocytes seems not to be required for epithelial repair. Specific deletion of c-Jun in keratinocytes of mice does not result in an overt phenotype, and wound closure seems to be only mildly affected (Li et al., 2003; Zenz et al., 2003). Notably, proliferation as well as migration of keratinocytes isolated from these mice are significantly impaired, but can be rescued by the addition of recombinant growth factors, such as

Jun-regulated genes during wound healing L Florin et al

7006

keratinocyte growth factor (KGF; fibroblast growth factor 7 (FGF-7)) and granulocyte–macrophage colony-stimulating factor (GM-CSF) (Zenz et al., 2003). It is tempting to speculate that in vivo these type of proteins, which are produced and secreted by mesenchymal cells, will compensate in a trans-regulatory manner for cell-autonomous defects in keratinocyte proliferation. This concept of trans-regulatory interaction between keratinocytes and cells of the mesenchyme, predominantly fibroblasts, is well supported by in vitro organ culture studies of Szabowski et al. (2000), who described a double paracrine mechanism involving AP-1 activity in fibroblasts, modulating keratinocyte proliferation and differentiation. In an in vitro reconstituted skin equivalent reflecting early molecular events of skin regeneration, the influence of genetically modified murine fibroblasts on proliferation and differentiation of cocultured primary human keratinocytes was investigated. Here, keratinocyte-derived IL-1 triggers the expression of AP-1 target genes in fibroblasts, including KGF and GM-CSF, which act back on keratinocytes to modulate proliferation and differentiation. In the presence of c-Jun-deficient fibroblasts, human keratinocytes proliferated only poorly and differentiation was also reduced, whereas hyperproliferation as well as increased differentiation was observed in cocultures with fibroblasts deficient in JunB. These phenotypes could be attributed, at least in part, to the antagonistic regulation of kgf and gm-csf by c-Jun and JunB in fibroblasts (Szabowski et al., 2000). Although the pathways in fibroblasts that are triggered by keratinocyte-derived IL-1 have been extensively studied, besides KGF and GM-CSF, little is known about AP-1 target genes mediating the cellular response to wounding. In the present study, we have employed large-scale gene expression profiling using a 15 000 cDNA collection to identify the genetic program of 1500 c-Jun/JunB-dependent and -independent genes in fibroblasts whose expression is regulated by IL-1 and which are likely to be involved in the process of cutaneous wound healing.

Results Induction of known Jun target genes by IL-1 in fibroblasts To examine the gene expression profile in untreated versus IL-1 stimulated wt, c-jun- or junB-deficient fibroblasts, we first determined the appropriate time point for RNA isolation by analysing the kinetics of the induction of kgf and gm-csf (Figure 1a), two direct AP-1 target genes known to be upregulated in response to IL-1 in fibroblasts (Szabowski et al., 2000). Maximal induction of kgf and gm-csf was observed between 6 and 8 h of IL-1 treatment in wt and junBdeficient fibroblasts. In agreement with the strict c-Jun dependency of expression in untreated wt fibroblasts (Szabowski et al., 2000), neither kgf nor gm-csf transcripts were detected in cells lacking c-Jun, even after IL-1 treatment. Based on these results, the Oncogene

Figure 1 (a) Induction kinetics of two known AP-1 target genes upon IL-1 treatment. Wild-type, c-jun- or junB-deficient MEFs were treated with 5 ng/ml IL-1a for the indicated time periods. Semi-quantitative RT–PCR for kgf and gm-csf was performed to determine the time point of maximal induction of Jun target genes by IL-1a. b-Tubulin was used for standardization. (b) Schematic illustration of the experimental design. The first set of hybridizations (solid arrows) separately addressed the gene expression profiles of untreated (open boxes) versus IL-1-stimulated (shadowed boxes) wild-type, c-jun and junB-deficient fibroblasts. In a second set of experiments (dashed arrows), the IL-1-stimulated cells of all three genotypes were compared with each other. All hybridizations were performed in duplicate including a color switch

intermediate time point of 7 h was selected for the identification and global expression analysis of IL-1responsive genes. RNA integrity and effective IL-1 stimulation were confirmed by RT–PCR for kgf and gmcsf (data not shown).


7007

Experimental design To perform large-scale gene expression profiling of untreated and IL-1-treated wild-type and mutant cells, the 15K NIA cDNA clone collection was used, which comprises of 15 248 cDNA sequences from genes expressed during embryonic development (Ko et al., 1998; Tanaka et al., 2000). Of these, around 75% are ESTs without functional annotation, thereby allowing the identification of yet unknown genes. In total, we performed 12 independent microarray hybridizations (Figure 1b): in one experimental setup, gene expression in untreated versus IL-1-stimulated wild-type, c-jun- and junB-deficient fibroblasts was analysed separately for all of the three genotypes. In a second set of hybridizations, the expression profiles of stimulated c-jun/ and junB/ cells were compared to each other and to wt cells. All experiments were performed in duplicate with reversed assignment of fluorophors (‘color switch’). In light of the replicative design of the chip with each clone being represented by two spots, this experimental setting ensured a high degree of internal controls. Identification of novel Jun-dependent and IL-1-responsive genes Of the overall 15 248 sequences printed on the 15K NIA microarray chip, 10 180 clones (67%) were obtained as evaluable data sets. The remaining 5067 clones showed no detectable expression across the series of hybridizations, suggesting very low levels of expression in fibroblasts. Out of the 10 180 expressed cDNAs, 1526 (15%) were differentially expressed by more than 2.46fold between unstimulated or IL-1-treated wt and mutant fibroblasts. We identified 1198 Jun-regulated clones, most of which (1097) were IL-1 independent (Set I), whereas 101 genes showed not only Jun dependency but also significantly altered expression upon IL-1 treatment (Set II). In addition, we found 329 cDNA sequences to be differentially expressed after IL-1 stimulation, which were not affected by the deletion of either c-Jun or JunB (Set III; gene names and microarray data are listed in Tables 1 and 2). Using an advanced combinatorial evaluation method, we were able to further determine the specific contribution of c-Jun and JunB to target gene expression. In this extended query, fluorescence ratios representing differential expression were categorized as up-, down- or not regulated, rather than considering absolute values. Thus, genes were included whose expression levels could be below the level of detection in one of the three genotypes analysed (e.g. kgf and gm-csf in c-jun/ fibroblasts; Figure 1a). Responsiveness to IL-1 treatment was determined with respect to gene expression values in the wt cells or, if no expression was detectable in these cells, by the average ratios measured in c-junand junB-deficient fibroblasts. Using this approach, we defined distinct subsets of genes (Figure 2b) that were activated or repressed by

single deletion of c-Jun or JunB, while targeted deletion of the respective other Jun member had no influence on the expression level. Another subgroup of genes was properly expressed only in wt cells harbouring both AP-1 subunits. Interestingly, we also identified an array of 32 genes that were antagonistically controlled by cJun and JunB. Notably, nearly equal numbers of genes were found to be activated and repressed by either Jun protein, the major group even being negatively regulated by both AP-1 components as indicated by elevated mRNA levels in either one of the mutant cell lines, when compared to wt fibroblasts. Validation of microarray results In order to confirm the microarray-derived data by an independent assay, we performed semi-quantitative RT– PCR for an initial set of 34 genes using the same RNA that was used for hybridization (Figure 3). For 28 out of these 34 genes (83%), the pattern of differential gene expression obtained by chip analysis could be confirmed. The collection of 28 validated genes comprised representative examples of all of the subgroups depicted in Figure 2b, including seven IL-1-responsive clones. For instance, mRNA levels of Jun-activated genes, such as matrix GLA protein (mglap) and cd24a as examples for c-Jun and JunB targets, respectively, were shown to be reduced or not detectable in the respective jundeficient cells when compared to wt cells. Lipocalin-2 (lcn2), as a representative of the class of newly identified IL-1-inducible genes, was verified to be repressed by JunB as deduced from elevated basal expression in the absence of JunB and further induction upon stimulation of the cells with IL-1. Thus, the expression profile extracted from the microarray data could be largely confirmed. Identification of novel wound regulated genes Since the release of IL-1 from keratinocytes represents an early event in response to skin injury in vivo, we asked whether the Jun target genes identified in the microarray screening were regulated during wound healing. Indeed, screening of the literature revealed at least 19 of the differentially expressed genes, for which regulation in skin wounds has been described (Table 3). For some of them, for example, b1-integrin, osteopontin, lumican and decorin, a critical role in skin homeostasis and repair has been demonstrated in functional studies using knockout and transgenic animals (Danielson et al., 1997; Chakravarti et al., 1998; Liaw et al., 1998; Grose et al., 2002, respectively). In order to further identify novel genes involved in cutaneous repair, we analysed the expression of the 28 validated clones (Figure 3) during the wound-healing process. RNA was isolated from full-thickness excisional back skin wounds of mice at various time points after injury and gene expression levels were monitored by semi-quantitative RT–PCR (Figure 4). Out of 28 genes analysed, seven (25%) were transiently upregulated during wound repair (Figure 4a). Interestingly, Oncogene


7008

only two of them (lcn2 and mail pending) were also upregulated in response to IL-1 in fibroblasts. Conversely, the yet uncharacterized clone rikenA030007 was confirmed to be induced by IL-1, but its expression levels remained unchanged during wound healing. Kgf

and gm-csf, which are known wound-healing-associated genes (Werner et al., 1992; Mann et al., 2001), served as internal controls (Figure 4b). Five Jun-regulated genes (disabled-2 (dab2), rras2, TGF-b1-induced transcript 4 (tgfb1i4), mglap and hca112) showed increased expres-

Table 1 (a)–(d) c-Jun and/or JunB target genes in fibroblasts Gene

(a) Selected c-Jun-dependent genes Activated by c-Jun Matrix gamma-carboxyglutamate protein (mglap) EST, weakly similar to PRP2 mouse prolin Procollagen, type I, a2 ESTs Biglycan Gelsolin Cytochrome P450, 1b1, benz[a]anthracene inducible 6030455P07Rik Solute carrier family 1, member 7 Stromal cell-derived factor 1 Asparagine synthetase Mus musculus, similar to gene overexpress Kruppel-like factor 4 (klf4) Pleiotrophin Lymphotoxin B receptor (ltbr) 1110048B16Rik TGF-b 1-induced transcript 4 (tgfb1i4) [y] Repressed by c-Jun Creatine kinase, brain Telomeric repeat binding factor 1 DNA segment, Chr 12, ERATO Doi 553, expressed Sdccag28 Integrin a3 Low-density lipoprotein receptor-related protein 2 Mm.137948 Nestin 2610206D03Rik Mesothelin Disabled homolog 2 (dab2; Drosophila) [y] (b) Selected JunB-dependent genes Activated by JunB Capping protein (actin filament), gelsolin-like Transmembrane 4 superfamily member 6 Myeloid ecotropic viral integration site-related 1 f-Box-only protein 3 ESTs, weakly similar to KDGA mouse diacyl CD9 antigen Mm.44032 1200003E16Rik 9130023P14Rik Expressed sequence AI504062 CD24a antigen Nuclear protein 1 (nupr) [y] Repressed by JunB H19 fetal liver mRNA Clusterin Lipoprotein lipase Ubiquitin-like 1 (sentrin) activating enzyme E1B Mus musculus, similar to hypothetical protein Inhibitor of DNA binding 2 Erythrocyte protein band 4.1-like 4b Decorin 9330166G04Rik Cytochrome b-245, a polypeptide Glutathione S-transferase, a 4 [y] Oncogene

UniGene/GB

Fold difference compared to wt

Tested by RT–PCR

Mm.193459 Mm.103269 Mm.4482 Mm.182669 Mm.2608 Mm.21109 Mm.4443 Mm.29997 Mm.1056 Mm.465 Mm.2942 Mm.5510 Mm.4325 Mm.3063 Mm.3122 Mm.4948 Mm.20927

133.0 103.9 78.2 45.7 42.5 22.3 20.0 19.0 12.9 11.6 10.7 10.2 9.5 6.9 4.5 4.0 3.7

X

Mm.16831 Mm.4306 Mm.27563 Mm.28896 Mm.57035 Mm.23847 Mm.137948 Mm.23742 Mm.28878 Mm.17510 Mm.34248

54.3 43.1 33.7 29.6 24.9 22.2 17.5 17.3 15.5 14.8 7.1

X

Mm.18626 Mm.46701 Mm.31436 Mm.143768 Mm.9747 Mm.2956 Mm.44032 Mm.27917 Mm.100720 Mm.31672 Mm.6417 Mm.18742

13.8 12.9 12.2 12.0 11.2 11.1 10.3 10.1 8.4 8.3 6.9 5.9

X X


23.9 17.1 16.6 14.1 14.0 13.9 13.5 11.4 10.0 9.8 8.6

X X X

X


7009 Table 1 (continued) Fold difference in wt versus Gene (c) Selected genes, dependent on c-Jun and JunB Genes, activated by c-Jun and JunB PDZ and LIM domain 1 (elfin) Expressed sequence AA673245 Inactive X-specific transcripts Fibulin 2 Fragile X mental retardation syndrome 1 homolog Ser/(cys) proteinase inhibitor, clade B1a (serpinb1a) Tumor protein D52-like 1 p53 apoptosis effector related to Pmp22 ESTs 1500005K14Rik Serum/glucocorticoid-regulated kinase (sgk) 2810012H18Rik Four and a half LIM domains 1 (fhl1) EBNA1 binding protein 2 Fyn proto-oncogene Mm.44220 [y] Repressed by both subunits 1810009M01Rik 2310035M22Rik Hepatocellular-associated antigen 112 (hca112) Vhlh-interacting deubiquitinating enzyme 1 EST Aldehyde dehydrogenase family 1, subfamily A1 0610010I23Rik Glutathione S-transferase, mu 2 Complement component factor h 2010008E23Rik Protein tyr. phosphatase, F-interacting, BP 2 (ppfibp2) Paternally expressed 3 (peg3) Fibroblast growth factor receptor-like 1 (fgfrl1) Expressed sequence AI115348 ESTs ESTs, weakly similar to rat F-spondin (Mm.10295) [y] (d) Selection of antagonistically regulated genes Antagonistically controlled, c-Jun activated Procollagen, type II, a1 Transglutaminase 2, C polypeptide ESTs Necdin Similar to ASF-1 Growth factor receptor-bound protein 10 Inhibitor of DNA binding 3 0610038H10Rik IGF binding protein 4 (igfbp4) FK506 binding protein 5 (51 kDa) Expressed sequence C85539 [y] Antagonistically regulated, c-Jun repressed genes Expressed sequence C80998 Adamts4 Pellino 1 H2A histone family, member Z 2310008N12Rik Thymopoietin Expressed sequence AU022220 ESTs, similar to zinc-finger protein Expressed sequence AU042434 Immediate early response 5 1200008D14Rik Related RAS viral oncogene homolog 2 (rras2) [y]

UniGene/GB

Mm.5567 Mm.6149 Mm.4095 Mm.6120 Mm.3451 Mm.46316 Mm.7821 Mm.28209 Mm.29116 Mm.34131 Mm.28405 Mm.3886 Mm.3126 Mm.29906 Mm.4848 Mm.42200

c-jun/

junB/

33.8 26.5 13.1 11.5 11.3 10.6 9.5 9.5 8.3 6.9 6.1 6.0 5.2 4.5 4.0 4.2

6.8 5.3 12.9 5.6 3.3 17.6 5.2 8.5 5.6 6.1 3.9 2.7 6.6 12.8 5.4 3.5

Mm.28385 Mm.41535 Mm.27061 Mm.24383 BG071273 Mm.4514 Mm.133825 Mm.14601 Mm.8655 Mm.101274 Mm.2817 Mm.7952 Mm.35691 Mm.21597 Mm.28110 Mm.102935

39.1 29.0 19.6 15.4 15.1 11.1 11.0 8.3 6.6 6.4 6.3 6.2 4.3 4.2 4.0 2.9

39.2 5.9 28.5 5.1 9.2 10.7 8.7 7.1 4.5 6.9 4.3 6.6 5.0 8.9 58.7 5.3


13.2 10.0 9.7 8.2 7.2 6.0 4.9 4.9 4.3 4.1 3.1

4.9 3.1 4.5 5.9 3.2 2.7 4.8 3.5 18.7 5.7 6.0

Mm.172875 Mm.87750 Mm.28957 Mm.916 Mm.30060 Mm.124 Mm.26304 Mm.26852 Mm.24807 Mm.12246 Mm.2252 Mm.23219

32.0 9.9 7.8 4.1 4.0 3.9 3.6 3.5 3.2 3.1 2.9 2.5

10.0 5.5 8.4 5.7 3.1 8.9 3.3 7.3 3.4 4.1 2.8 2.5

Tested by RT–PCR

X

X X X X

X

X X X X

X

X

X

In all, 10 genes of each subgroup exhibiting highest differential expression compared to wild type are listed. In addition, genes showing modulated expression after IL-1 stimulation were analysed by RT–PCR. Gene denomination and UniGene/GenBank entries were provided by NIA. The complete list of identified genes is available on demand Oncogene


7010 Table 2 Selected IL-1-responsive genes Gene

UniGene/GB

Induced by IL-1 stimulation ESTs ESTs, highly similar to mouse LMA3 Metallothionein 2 Expressed sequence AI385631 ESTs ESTs Septin 9 1810027P18Rik ESTs ESTs, similar to HPRP3 Expressed sequence C81521 Riken 2810434B10 9130023F12Rik Macrophage gal-N-acetyl-galactosamine-specific lectin Lipocalin 2 (lcn2) ESTs Expressed sequence AL022637 Osmotic stress protein 94 kDa ELKL motif kinase Mail-pending House-keeping protein 1 ESTs Mm.21179 4930406E12Rik Solute carrier family 9, isoform 3 regulator 1 2310058G22Rik Dystroglycan 1 Frizzy-related protein 1 (Drosophila) Cyclin B1, related sequence 1 Mus musculus, similar to metallothionein A030007L17Rik Radical fringe gene homolog (rfng; Drosophila) Kruppel-like factor 4 (klf4) Serum/glucocorticoid-regulated kinase (sgk) [y]

Mm.172997 Mm.174029 Mm.142740 Mm.9845 Mm.24268 BG071856 Mm.38450 Mm.29180 Mm.29490 Mm.8581 Mm.13352 Mm.24337 Mm.190602 Mm.7574 Mm.9537 Mm.173998 Mm.200806 Mm.4150 Mm.4082 Mm.3732 Mm.103 Mm.197686 Mm.21179 Mm.34525 Mm.27842 Mm.25702 Mm.7524 Mm.24202 Mm.22569 Mm.192991 Mm.27337 Mm.871 Mm.4325 Mm.28405

39.1 23.6 22.9 21.4 20.5 20.1 19.5 19.0 16.6 16.5 14.0 13.9 12.5 12.1 11.6 9.1 8.3 8.2 6.8 6.2 5.4 5.1 4.8 4.6 4.5 4.5 4.3 4.3 4.2 4.1 3.2 2.8 0.4 0.8

Mm.27856 Mm.5831 Mm.18062 Mm.30820 Mm.38367 Mm.18203 Mm.4173 Mm.14897 AU041552 Mm.18574 Mm.41552 Mm.22480 Mm.173941 Mm.148717 Mm.38211 BG085183 Mm.43612 Mm.35059 Mm.25150 Mm.42200 Mm.21146 Mm.71886 Mm.79916 Mm.43444 Mm.29674 Mm.153446 Mm.104900 Mm.143753 Mm.26949 Mm.86651 Mm.63468 Mm.1408

43.5 18.9 12.3 11.7 9.2 9.1 9.1 8.9 8.8 8.7 8.5 8.5 8.0 7.5 7.5 6.0 5.6 5.1 4.7 3.4 3.4 3.3 3.2 3.2 3.2 3.1 3.1 3.1 3.0 2.9 2.9 2.8

Downregulated after IL-1 treatment Protein tyrosine phosphatase, receptor type, K ATP-binding cassette, sub-family E (OABP) 1 Expressed sequence AU044632 Mm., clone MGC:27904 IMAGE:3500 Mm.38367 Embryonic ectoderm development DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide 26 Cadherin 2 ESTs 2700087I09Rik Mm., clone MGC:28734 IMAGE:4460 Cyclin D binding myb-like transcription factor 1 ESTs Mm.148717 ESTs ESTs 8-oxoguanine DNA-glycosylase 1 rho GTPase activating protein 5 Clone IMAGE:4948318; Mus musculus Mm.44220 GATA binding protein 6 Expressed sequence AA536666 Mm.79916 MAD2 (mitotic arrest deficient, homolog)-like 1 PTK2 protein tyrosine kinase 2 Phospholipase A2, group V N-acylsphingosine amidohydrolase 2 Exportin 4 Mus musculus, similar to electron transfer ESTs, moderately similar to protocadherin ESTs, weakly similar to C16C10.6.p [c.e.] Adrenomedullin (adm) Oncogene

Fold induction

Regulation by c-Jun/JunB

Tested by RT–PCR

ant., c-Jun act. c-Jun act.

c-Jun rep. JunB rep. Both rep. Ant., c-Jun act. c-Jun act.

X

JunB rep.

X

JunB act. X JunB act. Ant., c-Jun act. JunB rep. Both rep. JunB act. Both rep. c-Jun rep. c-Jun rep. Both rep. c-Jun act. JunB rep. Both rep. c-Jun act. Ant., c-Jun act.

X X X X

Ant., c-Jun rep. c-Jun act.

JunB act.

Both rep. c-Jun rep. Both rep. JunB rep. Both rep. Both rep. Both act. c-Jun act. c-Jun act. c-Jun act. JunB rep. c-Jun rep. JunB rep. c-Jun act. c-Jun rep. c-Jun rep. Both act. JunB act. c-Jun act.

X X

X


7011 Table 2 Gene Bloom syndrome homolog (human) Four and a half LIM domains 1 (fhl1) [y]

(continued )

UniGene/GB Mm.12932 Mm.3126

Fold induction 2.8 1.3

Regulation by c-Jun/JunB Both act. c-Jun act.

Tested by RT–PCR X

The top 30 genes induced or repressed by IL-1 are presented. Jun dependency in fibroblasts, if determined, is indicated on the right. Gene names, UniGene and Gene Bank numbers were obtained from NIA. The complete list is available on demand. Ant: antagonistically regulated

sion between 0.5 h up to 7 days after injury (Figure 4a), which is likely to be independent of IL-1 (Figure 3). To our knowledge, none of these seven woundregulated genes has so far been implicated in skin regeneration, underscoring the value of our approach to identify novel genes involved in this process. Given the reliability of the microarray data of at least 83%, one might expect a total of over 200 Jun and IL-1-regulated genes, exhibiting alterations in expression during the process of wound healing. Cell type-specific gene expression in wounded skin In order to define exactly the cell type responsible for the expression of the newly identified wounding-associated genes in vivo, in situ hybridization analysis was performed on skin biopsies, which were isolated 1 day after wounding. The JunB target gene lcn2 and the cJun-regulated tgfb1i4, both of which exhibit maximal induction at this time point, showed strong expression in fibroblasts (Figure 5), confirming expression analysis in fibroblasts in tissue culture. In addition, expression of both genes is also observed in keratinocytes (Figure 5), whereas transcripts of the c-Jun target dab2 were detected mainly in fibroblasts.

Discussion In an attempt to identify novel AP-1-dependent target genes involved in the process of wound healing, we analysed the gene expression profile of untreated and IL-1-stimulated wild-type and either c-jun- or junBdeficient fibroblasts. For this purpose, we generated a 15K microarray, which contains the NIA clone collection consisting mainly of unknown genes (Ko et al., 1998; Tanaka et al., 2000). The fibroblasts were cultured under low serum conditions to mimic the quiescent state of their dermal counterparts in vivo. Subsequently, cells were treated with IL-1, mimicking its immediate release by keratinocytes and its early expression by polymorphonuclear leukocytes in the wound (Hu¨bner et al., 1996) and thereby stimulating AP-1 activity in fibroblasts (Muegge et al., 1989; Williams et al., 1992). The time point for the subsequent chip analysis was chosen as a result of the induction kinetics of kgf and gm-csf, two known AP-1-dependent target genes (see Figure 1a) in response to IL-1. Taken into account that IL-1 regulation is not necessarily coupled to Jun-dependent gene expression and vice versa, a combinatorial analysis

Figure 2 Identification of Jun-dependent and IL-1-responsive genes by microarray analysis. (a) Evaluation of microarray data resulted in 41500 clones that are differentially expressed by more than 2.5-fold. These genes can be classified to be either Jun regulated, but IL-1 independent (Set I), Jun dependent and IL-1 responsive (Set II) or IL-1 dependent irrespective of c-Jun and JunB (Set III). (b) Partition of the three sets of genes according to the regulation by Jun proteins and their responsiveness to IL-1 on gene transcription. Jun-dependent genes are divided into IL-1independent (hatched boxes) and IL-1-responsive genes (black sections above). The regulation by the respective Jun protein(s) is indicated below (act. ¼ activated), rep. ¼ repressed); for antagonistically controlled genes, the specification refers to the activity of cJun. Y-axis indicates gene numbers

was applied (see Figure 1b), which should lead to the identification of genes, which are Jun dependent, IL-1 regulated or fulfil both criteria. More than 1000 cDNA clones could be identified, which show a deregulated expression in c-Jun- or JunBdeficient fibroblasts compared to wild-type cells (see Figure 2a and Table 1). The considerably higher number Oncogene


7012

Figure 3 Verification of the microarray results by semi-quantitative RT–PCR. Analysis of the RNA used in gene expression profiling confirmed differential expression for an initial set of 28 randomly selected Jun target genes. The genes are grouped by their dependency on Jun proteins and sorted for being positively or negatively regulated by c-Jun and JunB (as indicated on the right of each panel). Genes for which the prediction could only be partially substantiated are indicated by an asterisk (*). Primer pairs used for PCR analysis are listed in the Supplementary Table 1

of c-Jun- and/or JunB-dependent genes (1097) compared to the amount of IL-1-regulated genes (328) (see also Table 2) may be due to the fact that the applied approach does not distinguish between direct Jun target genes and indirectly Jun-regulated genes. However, the stimulation for 7 h favors the identification of direct IL1-responsive Jun-regulated genes. As expected, the number of genes in the intersection of IL-1- and Junregulated genes is reduced to 101 (see Figure 2a). In Oncogene

addition to induction of the JNK pathway, which modulates the activity of c-Jun as well as other transcription factors like ATF-2 and Elk-1 (Weston and Davis, 2002), the inductive stimulus IL-1 simultaneously triggers the IKK–NF-kB signal transduction pathway (Baud et al., 1999). There is increasing evidence for a crosstalk between the IKK/NFkB and JNK pathway and a convergence at the level of target gene promoters, leading to an additive and/or synergistic


7013 Table 3 Known wound regulated genes found in the screen Gene name

Symbol

UniGene

Dependency on AP-1

Decorin Fibronectin 1 Fibulin 2 Follistatin-like 3 Granulocyte–macrophage colony-stimulating factor Hypoxia-inducible factor 1, a-subunit Keratinocyte growth factor (Fgf7) Insulin-like growth factor 1 Insulin-like growth factor 2 Insulin-like growth factor binding protein 3 Lumican Metallothionein 1, 2 Nidogen 1 Osteopontin (secreted phosphoprotein 1) Platelet-derived growth factor receptor, a Platelet-derived growth factor receptor, b Tenascin C Transforming growth factor, b2 Transglutaminase 2, C polypeptide

Dcn Fn1 Fbln2 Fstl3 Gm-csf Hif1a Kgf/Fgf7 Igf1 Igf2 Igfbp3 Lum Mt2 Nid1 Spp1 Pdgfra Pdgfrb Tnc Tgfb2 Tgm2

Mm.1987 Mm.193099 Mm.6120 Mm.28863 Mm.4922 Mm.3879 Mm.57177 Mm.2770 Mm.3862 Mm.29254 Mm.18888 Mm.142740 Mm.4691 Mm.321 Mm.2924 Mm.4146 Mm.980 Mm.18213 Mm.18843

JunB rep c-Jun act Both act c-Jun act Ant., c-Jun act c-Jun act Ant., c-Jun act c-Jun act Both rep c-Jun rep JunB rep c-Jun act c-Jun act — c-Jun act c-Jun act c-Jun act c-Jun act Ant, c-Jun act

Reference

IL-1

ind ind

ind ind

Sayani et al. (2000); Soo et al. (2000) Muro et al. (2003) Fassler et al. (1996) Wankell et al. (2001) Mann et al. (2001) Elson et al. (2000), Scheid et al. (2000) Werner et al. (1992) Brown et al. (1997) Brown et al. (1997) Thorey et al. (2004) Saika et al. (2000) Iwata et al. (1999), Lansdown (2002) Fassler et al. (1996) Liaw et al. (1998) Beer et al. (1997) Beer et al. (1997) Fassler et al. (1996) Sullivan et al. (1995), Frank et al. (1996) Haroon et al. (1999)

Known wound-regulated genes are listed alphabetically with representative reference. Jun dependency and IL-1 responsiveness are indicated (ind ¼ IL-1 induced)

activation (Kralova et al., 1996; Li et al., 2000; Tang et al., 2001). Therefore, it is reasonable to assume that genes that show an altered Jun-dependent basal expression are still IL-1 responsive via the regulation of these transcription factors. Accordingly, in some of the identified IL-1-regulated Jun target genes (e.g. Mailpending, an Bcl-3-related IkB family member, and the proinflammatory cytokine Lipocalin-2), NF-kB binding sites have been mapped within the promoter (Cowland and Borregaard, 1997; Shiina et al., 2001). Based on the initial observation that the protooncogene c-Jun exhibited a transforming activity when overexpressed in fibroblasts, and that cell-cycle regulators such as Cyclin D1 as well as growth factors (such as KGF, GM-CSF and HB-EGF) were found to be activated by c-Jun (Shaulian and Karin, 2002), c-Jun was considered for a long time as the general activator of AP-1-dependent transcription, whereas JunB was proposed to exert opposing functions by mediating repression of AP-1 target genes and acting as a general repressor (Chiu et al., 1989). However, the subgroup of antagonistically regulated genes, which fit into this category, represents the smallest subset of genes identified in this screen (see Figures 2b and 3). Moreover, it was recently demonstrated that the role of c-Jun in stimulating proliferation is not only exclusively based on the activation of target genes but also on the repression of cell-cycle inhibitors like p16INK4 and p53 (Mechta-Grigoriou et al., 2001; Shaulian and Karin, 2002). Moreover, JunB is able to regulate positively gene expression, for example, by influencing cell-cycle control via activating p16 and Cyclin A (Andrecht et al., 2002; Passegue´ et al., 2002). The ability of JunB to compensate functionally the loss of c-Jun during embryonic development (Passegue´ et al., 2002) supports the idea that JunB is able to act as an activator, at least at some

c-Jun-regulated genes. In line with these observations, our data do not indicate functional preferences between c-Jun and JunB to either activate or repress gene transcription. Instead, distinct sets of c-Jun and JunB target genes were identified, of which nearly equal numbers were positively and negatively modulated (see Figures 2b and 3). Interestingly, the largest subgroup of genes represents those, which turned out to be synergistically activated or repressed by both, c-Jun and JunB (see Figures 2b and 3). The underlying molecular mechanism of this regulation may involve combinations of direct as well as indirect effects via the control of upstream components. Since c-Jun/JunB heterodimers are not able to bind to DNA (Deng and Karin, 1993), it is more likely that both subunits dimerize with specific partners and bind to different AP-1 sites to act synergistically on the same promoter. To confirm the results from the combinatorial microarray analysis by an independent method, semiquantitative RT–PCR analysis was performed for 34 genes representing all groups concerning the regulation mentioned above (see Figure 3, Tables 1 and 2 and data not shown). The high degree of overlap (83%) of both methods mirrors the typical rate of 15% false-positives previously reported for microarray studies (for review, see Lee et al., 2000). Since the experimental design of the screen was concentrated on the identification of novel Jun targets involved in wound healing, altered gene expression during the course of this process in vivo was the most stringent selection criteria. As listed in Table 3, a considerable number of newly identified Jun-regulated genes have already been associated with wound healing. Moreover, seven novel wound-regulated AP-1 target genes could be identified, which have so far not been Oncogene


7014

Figure 4 Expression of Jun target genes during early-stage skin regeneration. (a) Validated Jun-dependent genes showing regulation during skin repair. RNA was isolated from full-thickness excisional wounds of murine back skin taken at the indicated time points after wounding and subjected to semi-quantitative RT– PCR. (b) gm-csf and kgf are presented as internal controls for wound regulation. In addition, two genes are displayed whose expression remained unchanged during skin repair (cd24a and the riken clone A030007). Dependency on Jun proteins in fibroblasts is annotated on the right of each panel. All samples were standardized for b-tubulin levels

implicated in this process (see Figure 4a). Increased mRNA levels of these genes in wounded skin after 6 h is likely to be mostly due to enhanced transcription, rather than a contribution of some infiltrating immune cells, which are responsible for the subsequent boost of an inflammatory response. It is reasonable to assume that the initial IL-1 release by keratinocytes of the wound edge represents the triggering signal for lcn2, mail-pending, kgf and gm-csf, because IL-1 is also a strong stimulus of transcription in vitro. In contrast, the upregulation of the other genes in the wounded skin is not mimicked by IL-1 treatment in vitro, arguing that other cytokines are responsible for the observed effect. Various cytokines and growth factors are secreted Oncogene

Figure 5 Expression of selected wound-regulated Jun target genes in vivo. In situ hybridization was performed on cryosections from murine excisional back skin wounds at day 1 using DIG-labeled riboprobes. Riboprobes were detected enzymatically with DAB (reddish-brown) and sections were counterstained with hematoxylin. (a) Lcn2 expression in wounded skin at day 1 (bar ¼ 0.5 mm). (b) Higher magnification of the wound proximity, hybridized with riboprobes specific for lcn2, dab2 and tgfb1i4 (bar ¼ 0.1 mm)

in the early phases after wounding (Singer and Clark, 1999; Gillitzer and Goebeler, 2001; Werner and Grose, 2003) and have an impact on gene expression in fibroblasts. One likely candidate is TGF-b1, which is immediately released from platelets and described to serve as potent mitogen and chemoattractant for neutrophils, macrophages and fibroblasts (O’Kane and Ferguson, 1997). This idea is supported by the observation that tgfb1i4 turned out to be upregulated in wounds. In addition, TGF-b is able to antagonize the induction of genes by other stimuli, for example, as described for the IL-1 receptor antagonist (Bodo et al., 1998). Therefore, it is also not surprising that some in vitro IL-1-responsive genes are not wound


7015

regulated, as shown for the rikenA030007 sequence (see Figures 3 and 4). Since the screen is performed in fibroblasts in vitro and the elevated mRNA levels at 6 h postwounding in vivo most likely reflect transcriptional upregulation in cells residing in close proximity of the wound, particularly in fibroblasts, we still aimed to confirm the assumption of the in vivo source of the transcripts by in situ hybridization (see Figure 5). Besides keratinocytes, fibroblasts are indeed the major cell type, which expressed the identified genes, as shown for lcn2, dab2 and tgb1i4. In summary, we identified an intriguing number of novel c-Jun- and JunB-dependent genes expressed in fibroblasts, as well as more than 300 IL-1-responsive genes. Furthermore, for seven of these genes, a wound regulation in vivo could be shown and for three of them, fibroblasts were confirmed to be a source of expression in the healing skin. The provided data might serve as a basis for better understanding of the role of AP-1 in skin biology.

Materials and methods Cell lines and RNA preparation Immortalized wild-type, c-jun- and junB-deficient mouse embryonic fibroblasts (MEFs) have been described previously (Schreiber et al., 1999; Andrecht et al., 2002). Cells were seeded and grown to subconfluency. Subsequently, cells were starved by culturing in medium containing 0.5% FBS for 48 h and afterwards either left untreated or stimulated with IL-1a (R&D Biosystems GmbH, Wiesbaden, Germany) for 7 h. Total RNA was isolated using TRIzols reagent (Invitrogen GmbH, Karlsruhe, Germany) and subsequent column purification (RNeasy Kit, Quiagen, Hilden, Germany). Generation of the 15K NIA microarray The NIA 15K clone set containing 15 247 sequences collected from embryonic cDNA libraries (including preimplantation stage embryos from unfertilized egg to blastocyst, embryos at E7.5 and E12.5 female mesonephros/gonad) and newborn ovary tissues was utilized to generate the 15K NIA cDNA microarray (Ko et al., 1998; Ko et al., 2000; Tanaka et al., 2000; Kargul et al., 2001; sequence information is available at the web site of the National Institute on Aging (http:// lgsun.grc.nia.nih.gov/cDNA/15k.html). All clones are available as inserts of a modified pSPORT1 vector (Life Technologies, Rockville, MD, USA), propagated in clones of the Escherichia coli strain DH10B. To amplify cDNA inserts, PCR primers 50 -CCAGTCACGACGTTGTAAAACGAC-30 and 50 -GTGTGGAATTGTGAGCGGATAACAA-30 (Biospring, Frankfurt, Germany) were designed from sequences flanking the multiple cloning site of the vector. For each cDNA clone, amplification was performed as described in Tanaka et al. (2000). After precipitation, the PCR fragments were dissolved in 30 ml spotting buffer containing 3 SSC and 1.5 M betaine (Diehl et al., NAR 2002) and were spotted on QF epoxy substrates (Quantifoil Microtools, Jena, Germany) using an OmniGrid Microarrayer (GeneMachines, San Carlos, CA, USA) equipped with Stealth SMP3 Micro Spotting Pins (Telechem).

Labeling and hybridization of samples Fluorescence-labeled cDNA was generated by reverse transcription of 40 mg total RNA according to standard protocols (DeRisi et al., 1997) using an Omniscript RT Kit (Qiagen) and Cy3/Cy5-dUTPs (NEN Life Science Products, Ko¨ln, Germany). Subsequently, Cy3- and Cy5-labeled cDNA samples were pooled, purified and concentrated using Microcon YM30 PCR filter units (Millipore, Eschborn, Germany) as described previously (DeRisi et al., 1996; Schena, 1996; DeRisi et al., 1997). Samples were diluted in DigEasyHyb (Roche Diagnostics, Mannheim, Germany) and hybridized to microarray slides for 16 h at 371C under gentle agitation. Thereafter, the slides were washed at room temperature with: (i) 1 SSC/ 0.1% (w/v) SDS; (ii) 0.1 SSC/0.1% SDS for 5 min; (iii) 70% ethanol for 30 s; (iv) 100% ethanol for 30 s; and (v) finally airdried (Schlingemann et al., 2003). To account for the systemic error caused by different properties of the fluorescent dyes concerning incorporation, mean brightness and background noise, the hybridizations were always performed in duplicate experiments with interchanged assignment of fluorescent dyes (‘color switch’). Data acquisition and analysis Microarrays were scanned and processed using a GenePix 4000A microarray scanner in combination with the software GenePix Pro 3.0. (Axon Instruments Inc., Union City, NJ, USA). Subsequent data analysis and selection were performed using Microsoft Excel. After local background subtraction for each individual spot, the mean of signal intensities and logarithmic fluorescence ratios (Cy5/Cy3) of the replicate spots representing the same gene was calculated. For each hybridization, mean ratios were normalized by the median logarithmic ratio (ln ratio) of all genes found in that experiment. Accurate differential expression values for each gene were obtained by calculating the average of normalized ratios of replicate experiments after reversing the values obtained in the respective color switch experiment. Data sets representing differentially expressed genes were selected applying the following criteria: (i) evaluable data obtained in at least two independent experiments, (ii) mean signal intensities beyond the threshold set at 300 U, and (iii) mean of normalized logarithmic ratio higher than 1.3 (meaning differential expression of more than 2.46-fold). RT–PCR analysis The same RNA samples used for hybridization were applied to semi-quantitative RT–PCR for validation of the microarray data. cDNA synthesis was performed according to standard protocols using oligo(dT) and random hexamer oligonucleotides. Gene-specific fragments were obtained by linear phase PCR amplification with b-tubulin serving for standardization. Primer sequences are listed in the Supplementary Table 1. Wound healing experiments Four full-thickness excisional wounds were generated on the back of female BALB/c mice (10–12 weeks old) by dissecting skin and panniculus carnosus as described (Werner et al., 1992). The wounds were left untreated. For RNA expression studies, mice were killed at different time points after injury. Complete wounds including 2 mm of the epidermal margins were excised, immediately frozen in liquid nitrogen and stored at 801C until used for RNA isolation. Wound-healing experiments were performed with permission from the local authorities in Zurich, Switzerland. Oncogene


7016 In situ hybridization In situ hybridization was performed on 7 mm cryosections of full-thickness excisional mouse back skin wounds at day 1 after injury. Sense and antisense probes were generated by run-off in vitro transcription of the corresponding PCR fragments cloned into the vector pGEM-T-Easy (Promega GmbH, Mannheim, Germany) in the presence of digoxygenin (DIG)-labeled dUTPs. The hybridization was performed at 551C for 4 h. Hybridized probe was detected by peroxidaseconjugated anti-DIG antibodies (Roche Diagnostics) using

diaminobenzidine (DAB; Vector Laboratories Ltd, Peterborough, UK) as substrate. Sections were counterstained with hematoxylin. Acknowledgements We thank Dr Jochen Hess for careful proofreading of the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft (An 182/8-2) and by the Research Training Network (RTN) Program of the European Community.

References Andrecht S, Kolbus A, Hartenstein B, Angel P and SchorppKistner M. (2002). J. Biol. Chem., 277, 35961–35968. Angel P, Szabowski A and Schorpp-Kistner M. (2001). Oncogene, 20, 2413–2423. Baud V, Liu ZG, Bennett B, Suzuki N, Xia Y and Karin M. (1999). Genes Dev., 13, 1297–1308. Beer HD, Longaker MT and Werner S. 1997. J. Invest. Dermatol., 109, 132–138. Bodo M, Carinci P, Baroni T, Bellucci C, Giammarioli M, Pezzetti F and Becchetti E. (1998). Cytokine, 10, 353–360. Brown DL, Kane CD, Chernausek SD and Greenhalgh DH. 1997. Am. J. Pathol., 151, 715–724. Chakravarti S, Magnuson T, Lass JH, Jepsen KJ, LaMantia C and Carroll H. (1998). J. Cell Biol., 141, 1277–1286. Chinenov Y and Kerppola TK. (2001). Oncogene, 20, 2438–2452. Chiu R, Angel P and Karin M. (1989). Cell, 59, 979–986. Cowland JB and Borregaard N. (1997). Genomics, 45, 17–23. Danielson KG, Baribault H, Holmes DF, Graham H, Kadler KE and Iozzo RV. (1997). J. Cell Biol., 136, 729–743. Deng T and Karin M. (1993). Genes Dev., 7, 479–490. DeRisi J, Penland L, Brown PO, Bittner ML, Meltzer PS, Ray M, Chen Y, Su YA and Trent JM. (1996). Nat. Genet., 14, 457–460. DeRisi JL, Iyer VR and Brown PO. (1997). Science, 278, 680–686. Diehl F, Beckmann B, Kellner N, Hauser NC, Diehl S and Hoheisel JD. (2002). Nucleic Acids Res., 30, e79. Elson DA, Ryan HE, Snow JW, Johnson R and Arbeit JM. (2000). Cancer Res., 60, 6189–6195. Fassler R, Sasaki T, Timpl R, Chu ML and Werner S. (1996). Exp. Cell Res., 222, 111–116. Frank S, Madlener M and Werner S. (1996). J. Biol. Chem., 271, 10188–10193. Gillitzer R and Goebeler M. (2001). J. Leukocyte Biol., 69, 513–521. Grose R, Hutter C, Bloch W, Thorey I, Watt FM, Fassler R, Brakebusch C and Werner S. (2002). Development, 129, 2303–2315. Haroon ZA, Hettasch JM, Lai TS, Dewhirst MW and Greenberg CS. (1999). FASEB J., 13, 1787–1795. Herrlich P. (2001). Oncogene, 20, 2465–2475. Hu¨bner G, Brauchle M, Smola H, Madlener M, Fassler R and Werner S. (1996). Cytokine, 8, 548–556. Iwata M, Takebayashi T, Ohta H, Alcalde RE, Itano Y and Matsumura T. (1999). Histochem. Cell Biol., 112, 283–290. Kargul GJ, Dudekula DB, Qian Y, Lim MK, Jaradat SA, Tanaka TS, Carter MG and Ko MS. (2001). Nat. Genet., 28, 17–18. Ko MS, Kitchen JR, Wang X, Threat TA, Wang X, Hasegawa A, Sun T, Grahovac MJ, Kargul GJ, Lim MK, Cui Y, Sano Y, Tanaka T, Liang Y, Mason S, Paonessa PD, Sauls AD,

Oncogene

DePalma GE, Sharara R, Rowe LB, Eppig J, Morrell C and Doi H. (2000). Development, 127, 1737–1749. Ko MS, Threat TA, Wang X, Horton JH, Cui Y, Wang X, Pryor E, Paris J, Wells-Smith J, Kitchen JR, Rowe LB, Eppig J, Satoh T, Brant L, Fujiwara H, Yotsumoto S and Nakashima H. (1998). Hum. Mol. Genet., 7, 1967–1978. Kralova J, Schatzle JD, Liss AS, Bargmann W and Bose Jr HR. (1996). Oncogene, 12, 2595–2604. Lansdown AB. (2002). Wound Repair Regen., 10, 130–132. Lee ML, Kuo FC, Whitmore GA and Sklar J. (2000). Proc. Natl. Acad. USA, 97, 9834–9839. Li G, Gustafson-Brown C, Hanks SK, Nason K, Arbeit JM, Pogliano K, Wisdom RM and Johnson RS. (2003). Dev. Cell, 4, 865–877. Li JJ, Cao Y, Young MR and Colburn NH. (2000). Mol. Carcinogen., 29, 159–169. Liaw L, Birk DE, Ballas CB, Whitsitt JS, Davidson JM and Hogan BL. (1998). J. Clin. Invest, 101, 1468–1478. Mann A, Breuhahn K, Schirmacher P and Blessing M. (2001). J. Invest. Dermatol., 117, 1382–1390. Martin P. (1997). Science, 276, 75–81. Mechta-Grigoriou F, Gerald D and Yaniv M. (2001). Oncogene, 20, 2378–2389. Muegge K, Williams TM, Kant J, Karin M, Chiu R, Schmidt A, Siebenlist U, Young HA and Durum SK. (1989). Science, 246, 249–251. Muro AF, Chauhan AK, Gajovic S, Iaconcig A, Porro F, Stanta G and Baralle FE. (2003). J. Cell Biol., 162, 149–160. O’Kane S and Ferguson MW. (1997). Int. J. Biochem. Cell Biol., 29, 63–78. Passegue´ E, Jochum W, Behrens A, Ricci R and Wagner EF. (2002). Nat. Genet., 30, 158–166. Saika S, Shiraishi A, Saika S, Liu CY, Funderburgh JL, Kao CWC, Converse RL and Kao WWY. (2000). J. Biol. Chem., 275, 2607–2612. Sayani K, Dodd CM, Nedelec B, Shen YJ, Ghahary A, Tredget EE and Scott PG. (2000). Histopathology, 36, 262–272. Scheid A, Wenger RH, Christina H, Camenisch I, Ferenc A, Stauffer UG, Gassmann M and Meuli M. (2000). Pediatr. Surg. Int., 16, 232–236. Scheena M. (1996). BioEssays, 18, 427–431. Schlingemann J, Hess J, Wrobel G, Breitenbach U, Gebhardt C, Steinlein P, Kramer H, Furstenberger G, Hahn M, Angel P and Lichter P. (2003). Int. J. Cancer, 104, 699–708. Schreiber M, Kolbus A, Piu F, Szabowski A, Mohle-Steinlein U, Tian J, Karin M, Angel P and Wagner EF. (1999). Genes Dev., 13, 607–619. Shaulian E and Karin M. (2002). Nat. Cell Biol., 4, E131–E136. Shiina T, Morimatsu M, Kitamura H, Ito T, Kidou S, Matsubara K, Matsuda Y, Saito M and Syuto B. (2001). Immunogenetics, 53, 649–655.


7017 Singer AJ and Clark RA. (1999). N. Engl. J. Med., 341, 738–746. Soo C, Hu FY, Zhang X, Wang Y, Beanes SR, Lorenz HP, Hedrick MH, Mackool RJ, Plaas A, Kim SJ, Longaker MT, Freymiller E and Ting K. (2000). Am. J. Pathol., 157, 423–433. Sullivan KM, Lorenz HP, Meuli M, Lin RY and Adzick NS. (1995). J. Pediatr. Surg., 30, 198–202. Szabowski A, Maas-Szabowski N, Andrecht S, Kolbus A, Schorpp-Kistner M, Fusenig NE and Angel P. (2000). Cell, 103, 745–755. Tanaka TS, Jaradat SA, Lim MK, Kargul GJ, Wang X, Grahovac MJ, Pantano S, Sano Y, Piao Y, Nagaraja R, Doi H, Wood WH, III, Becker KG and Ko MS. (2000). Proc. Natl. Acad. Sci. USA, 97, 9127–9132. Tang G, Minemoto Y, Dibling B, Purcell NH, Li Z, Karin M and Lin A. (2001). Nature, 414, 313–317.

Thorey IS, Hinz B, Hoeflich A, Kaesler S, Bugnon P, Elmlinger M, Wanke R, Wolf E and Werner S. (2004). J. Biol. Chem., 279, 26674–26684. van Dam H and Castellazzi M. (2001). Oncogene, 20, 2453–2464. Wankell M, Kaesler S, Zhang YQ, Florence C, Werner S and Duan R. (2001). J. Endocrinol., 171, 385–395. Werner S and Grose R. (2003). Physiol. Rev., 83, 835–870. Werner S, Peters KG, Longaker MT, Fuller-Pace F, Banda MJ and Williams LT. (1992). Proc. Natl. Acad. Sci. USA, 89, 6896–6900. Weston CR and Davis RJ. (2002). Curr. Opin. Genet. Dev., 12, 14–21. Williams DH, Jeffery LJ and Murray EJ. (1992). Biochim. Biophys. Acta, 1180, 9–14. Zenz R, Scheuch H, Martin P, Frank C, Eferl R, Kenner L, Sibilia M and Wagner EF. (2003). Dev. Cell, 4, 879–889.

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

Oncogene