Oncogene (2006) 25, 2601–2614 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE The UVB-induced expression profile of human epidermis in vivo is different from that of cultured keratinocytes

CD Enk1, J Jacob-Hirsch2, H Gal3, I Verbovetski4, N Amariglio2, D Mevorach4, A Ingber1, D Givol3, G Rechavi2 and M Hochberg1

1Department of Dermatology, The Hadassah-Hebrew University Medical Center, Jerusalem, Israel; 2Department of Pediatric Hemato-Oncology and Functional Genomics, Safra Children’s Hospital, Sheba Medical Center and Sackler School of Medicine, Tel-Aviv University,Tel Aviv, Israel; 3Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot, Israel and 4The Laboratory for Cellular and Molecular Immunology, Department of Medicine, The Hadassah-Hebrew University Medical Center, Jerusalem, Israel

In order to obtain a comprehensive picture of the radiation. UVB, with a wavelength range between 290 molecular events regulating cutaneous photodamage of and 320 nm, represents one of the most important intact human epidermis, suction blister roofs obtained environmental hazards affectinghuman skin (Hahn after a single dose of in vivo ultraviolet (UV)B exposure and Weinberg, 2002). To protect itself against the were used for microarray profiling. We found a changed DNA-damaging effects of sunlight, the skin disposes expression of 619 . Half of the UVB-regulated genes over highly complicated cellular programs, including had returned to pre-exposure baseline levels at 72 h, cell-cycle arrest, DNA repair and (Brash et al., underscoring the transient character of the molecular 1996). Failure in selected elements of these defensive cutaneous UVB response. Of special interest was our strategies may result in propagation of cutaneous finding that several of the central target genes . Handlingof photodamageby the skin constitu- remained unaffected following UVB exposure in spite of tes a crucial survival mechanism, and the regulation and p53 accumulation. We next compared the in vivo activity of cutaneous UVB-regulated genes represents an expression profiles of epidermal sheets to that of cultured attractive model for the functioningof central protective human epidermal keratinocytes exposed to UVB in vitro. cellular responses under physiological conditions. We found 1931 genes that differed in their expression The advent of microarray technology allows the profiles between the two groups. The expression profile in generation of global expression profiles and differential intact epidemis was geared mainly towards DNA repair, expression of thousands of genes (Park et al., 2002; Joos whereas cultured keratinocytes responded predominantly et al., 2003; Sellheyer and Belbin, 2004). Several recent by activating genes associated with cell-cycle arrest and studies have employed microarray profilingto study apoptosis. These differences in expression profiles might UVB-regulated in cultured human reflect differences between mature differentiating kerati- keratinocytes (Becker et al., 2001; Li et al., 2001; nocytes in the suprabasal epidermal layers versus Murakami et al., 2001; Sesto et al., 2002; Takao et al., exponentially proliferating keratinocytes in cell culture. 2002; Dazard et al., 2003; Pisarchik et al., 2004; Lee Our findings show that extreme care should be taken when et al., 2005), revealing that a wide range of genes extrapolating from findings based on keratinocyte cultures regulating transcription factors, cytoskeletal , to changes in intact epidermis. secreted signaling molecules and controlling cellular Oncogene (2006) 25, 2601–2614. doi:10.1038/sj.onc.1209292; functions such as signal transduction, terminal differ- published online 23 January 2006 entiation and apoptosis, are induced by UVB. However, since these studies differ considerably in terms of time Keywords: UVB; epidermis; p53; in vivo; in vitro; kinetics, UVB dose, light sources and hybridization microarrays techniques, it is difficult to generate a comprehensive picture of the complex changes in gene expression taking place in UVB-irradiated keratinocytes. An additional drawback common to all these studies Introduction is the fact that they were performed on cultured keratinocytes under in vitro conditions. Skin tissue The skin constitutes the outer barrier of the human cultures differ substantially from the intact tissues they organism towards external stimuli and ultraviolet (UV) are supposed to imitate: Keratinocytes are usually cultured in the presence of hormones and growth factors Correspondence: Dr CD Enk, Department of Dermatology, Hadassah that might affect cell-cycle regulation, stress response Medical Center, P.O. Box 12000, IL 91010, Israel. and apoptosis (Gibbs et al., 1998), thereby modifying E-mail: [email protected] Received 11 July 2005; revised 6 October 2005; accepted 28 October the response to UVB. Furthermore, confluence 2005; published online 23 January 2006 in keratinocyte cultures affects the regulation of UVB-induced gene expression profile of human epidermis CD Enk et al 2602 programmed cell death, and a significant difference in sensitivity to UVB-induced apoptosis between subcon- fluent cultures and exponentially growing cells have been reported (Gniadecki et al., 1997). Finally, keratino- cyte cultures constitute a homogeneous population of cells lackingthe additional cellular constituents of intact epidermis such as Langerhans cells and intraepidermal inflammatory cells that together with the keratinocytes play important roles in the response of human epidermis to UV exposure (Baadsgaard, 1991). Usinga unique model system based on UV-exposed human epidermis followed by isolation of epidermal cells from suction blister roofs, we have recently studied the global expression profile in intact human epidermis usingoligonucleotide microarrays (Enk et al., 2004). With this in vivo setup that overcomes several of the limitations outlined above, we identified more than 800 UVB-regulated genes, some of which not previously known to be UVB sensitive. In the present study, we expand this investigation by monitoring the differen- Figure 1 Number up- and downregulated genes at different time- tially expressed genes over a 72 h period following UVB points followingUVB irradiation of intact human epidermis. exposure. We found that the majority of the in vivo UVB-regulated genes changed their expression at the 24 h time-point, but had returned to background levels within 72 h. We next compared the expression profile of sion profiles (Kannan et al., 2001). Applyingthis the in vivo exposed epidermal cells to that of cultured method to genes that were at least twofold changed in UVB-exposed normal human epidermal keratinocytes all volunteers at least at one time-point, triggered (NHEK), thus providingthe first direct comparison distinct, well-ordered gene expression and partitioned between the UV response of intact epidermis and the samples accordingto interval after UV exposure keratinocyte cultures. We here report that the expression rather than accordingto individuals (Figure 2). This profile in intact epidermis was geared mainly towards profilingpattern testifies to the robustness of the data set DNA repair, whereas cultured keratinocytes responded and indicates that our results reflect an overall UVB predominantly by activatinggenesassociated with cell- response. The algorithm showed that the most closely cycle arrest and apoptosis. related expression profiles were those of the 0, 2 and 72 time-points, whereas the 24 h expression profile dis- played greatest dissimilarity from the other time-points. Results and Discussion This suggests that the majority of UVB-induced changes in epidermal gene expression is transient and has UVB-induced gene expression reversed to background expression within 72 h. The To study the global gene expression profile in intact findingthat the early UVB response involves a relatively human epidermis following a single physiological UVB small number of genes, whereas the 24 h response is exposure, we exposed the inner forearms of three more complex and expresses a larger number of genes volunteers to 4 MED of UVB in vivo. Suction blisters corresponds well with the recent findings of Lee et al. were raised at 2, 24 and 72 h followingexposure, and (2005) who examined the differentially expressed genes mRNA extracted from the blister roofs underwent gene of UVB-irradiated immortalized HaCat keratinocytes at profilingusingAffymetrix Human Focus oligonucleotide 0.5, 6 and 24 h. arrays as described in Methods. Applyingan arbitrary The dendogram revealed eight major, stable clusters filter level of twofold changes in the ratios of gene containingdifferent expression kinetics of up- and expression in all volunteers at least at one time-point, we downregulated genes at 2, 24 and 72 h after UV found 619 genes whose expression was modified by UVB exposure (Figure 2). Interestingly, in five of the eight (Figure 1), of which 246 genes were upregulated and 373 clusters (cluster 1, 2, 5, 7 and 8), representing316 of the were downregulated, representing approximately 12% of 619 UVB-regulated genes (51%), the expression pattern the ‘valid’ genes (1.5–2% of the ). In our had returned to pre-exposure baseline levels a 72 h, list, the majority, 53% of the differentially expressed again underscoring the transient character of the genes, were found at the 24 h time-point, whereas 21 and genomic changes taking place in human skin following 26% of the UVB-regulated genes were found 2 and 72 h, a single, low-dose, UVB exposure. respectively, followingexposure.

Gene clustering Functional groups Clusteringthe DNA chip data by an unsupervised UVB-responsive genes were categorized into clusters approach organizes genes by similarities in their expres- based on known or presumed functions. Table 1 depicts

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2603

Figure 2 Cluster algorithm showing clustering of the 619 genes with at least twofold expression level changes in all volunteers in at least one time-point following in vivo UVB exposure of intact human epidermis. The normalized expression level for each gene (rows) in each sample (columns) is indicated by a color code (green downregulated and red upregulated). When applying an unsupervised algorithm, the samples were partitioned according to interval after UV exposure and showed greatest similarities among the 0, 2 and 72 h expression patterns. Z-score analysis grouped the genes into eight kinetically distinguishable clusters. The number of genes in each cluster is indicated. In five of the clusters, representinghalf of the UVB-modulated genes,the expression pattern had returned to baseline patterns at 72 h. selected genes of interest (http://www.hadassah.org.il/ (Maelandsmo et al., 1997). In normal human keratino- Departments/photobiology; for the full list). The 619 cytes, S100A2 nuclear stainingdisappears following differentially expressed genes were found in a wide range treatment with H2O2, and S100A2 protein might be of functional categories at one or more time-points after involved in keratinocyte responses to ROS (Zhang et al., UVB irradiation. 2002). The downregulation of S100A2 expression follow- ingUVB radiation reported in our study could thus result from the activity of free radicals on the epidermal cells. S100. Several skin disorders have been associated with the S100 family of calcium-bindingproteins, including disorders of keratinization, proliferation and differentia- Serine protease inhibitors. Our data show that a tion. The S100A7, S100A8 and S100A9 protein genes are number of serine proteinase inhibitors were induced absent or minimally expressed in normal epidermis, but followingUVB irradiation. The squamous cell carci- are markedly overexpressed in psoriatic lesions, in noma antigen (SCCA1 or SERPINB3) plays a role in keratinocytes with abnormal differentiation and in cuta- differentiation of squamous epithelium (Crombach neous tumors (Boni et al., 1997; Alowami et al., 2003; et al., 1989; Kato, 1996) and significantly inhibits certain Broome et al., 2003). S100A8 and S100A9 gene expression forms of apoptosis (Suminami et al., 2000). In human is upregulated after injury of murine and human epidermis skin, SCCA is expressed in epidermis overlyinginflam- (Thorey et al., 2001), and Grimbaldeston et al. (2003) matory lesions such as psoriasis and dermatitis (Hor- demonstrated that S100A8, but not S100A9,wasover- iuchi et al., 1994; Takeda et al., 2002). There is no expressed in the epidermis of UVA-irradiated BALB/c consensus regarding the expression in normal epidermis mice in response to reactive oxygen species (ROS). We (Duk et al., 1989; Cataltepe et al., 2000; Takeda et al., here report upregulation of S100A6, S100A7, S100A8, 2002). Our data show that SCCA1 is dramatically S100A9 and S100P protein genes in human epidermis increased 24 and 72 h after UVB exposure. Of specific followingUVB radiation. All genes showed interest is our findingof moderate but significant the same pattern of expression with upregulation at 24 h upregulation of serpinB13 (hurpin/headpin/PI13). This and sustained activity at 72 h. S100A2 is expressed in new member of the human ov-serpin family has benign nevi but not in metastatic , and loss of previously been demonstrated in the human immor- S100A2 activity might play a role in tumorigenesis and talized HaCaT keratinocyte cell line, in cultured normal constitute an early event in development human keratinocytes, in lesional skin from psoriatic

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2604 Table 1 Gene expression changes in intact human epidermis following in vivo UVB irradiation GenBank Symbol Gene/protein Fold change from nonirradiated 2h 24h 72h

Epidermal differentiation complex, calcium-binding proteins NM_005978.2 S100A2 S100 calcium-bindingprotein A2 0.34 1.75 1.76 NM_014624.2 S100A6 S100 calcium-bindingprotein A6 (calcyclin) 0.79 1.64 4.24 NM_002963.2 S100A7 S100 calcium-bindingprotein A7 (psoriasin 1) 0.69 22.43 12.02 NM_002964.2 S100A8 S100 calcium-bindingprotein A8 (calgranulin A) 1.32 14.57 12.52 NM_002965.2 S100A9 S100 calcium-bindingprotein A9 (calgranulin B) 0.73 22.04 15.19 NM_005980.1 S100P S100 calcium-bindingprotein P 1.30 3.61 3.53

Serpins (serine proteases inhibitors)

AF119873.1 SERPINA1 Serine (or cysteine) proteinase inhibitor, clade A (alpha-1 antiproteinase, 3.21 7.22 2.19 antitrypsin), member 1 AI554300 SERPINB1 Serine (or cysteine) proteinase inhibitor, clade B (ovalbumin), member 1 0.45 3.95 0.99 AF169949.1 SERPINB13 Serine (or cysteine) proteinase inhibitor, clade B (ovalbumin), member 13 1.03 1.97 2.69 BC005224.1 SERPINB3 Serine (or cysteine) proteinase inhibitor, clade B (ovalbumin), member 3 0.89 13.11 9.91

Extracellular matrix affecting factors NM_006665.1 HPSE Heparanase 0.79 5.12 3.95 NM_002421.2 MMP1 Matrix metalloproteinase 1 (interstitial collagenase) 0.18 15.56 2.57 NM_002422.2 MMP3 Matrix metalloproteinase 3 (stromelysin 1, progelatinase) 0.10 4.61 1.03 NM_003254.1 TIMP1 Tissue inhibitor of metalloproteinase 1 (erythroid potentiatingactivity, 1.22 5.43 3.41 collagenase inhibitor) L10343 SKALP/PI3 Elafin: neutrophil and pancreatic elastase-specific inhibitor of skin 0.52 21.55 14.19

Cell cycle NM_001211.2 BUB1B BUB1 buddinguninhibited by benzimidazoles 1 homologbeta (yeast) 1.01 0.63 2.24 NM_001237.1 CCNA2 Cyclin A2 2.31 1.67 8.50 NM_004701.2 CCNB2 Cyclin B2 1.46 0.92 5.90 NM_001759.1 CCND2 Cyclin D2 0.93 1.49 4.00 AW134535 CCNG2 Cyclin G2 0.92 2.70 1.05 AL524035 CDC2 Cell division cycle 2, G1 to S and G2 to M 0.92 0.46 4.19 NM_001255.1 CDC20 CDC20 cell division cycle 20 homolog( ) 2.29 1.17 6.12 NM_021873.1 CDC25B Cell division cycle 25B 1.71 1.36 3.61 AF213033.1 CDKN3 Cyclin-dependent kinase inhibitor 3 (CDK2-associated dual specificity 1.70 0.94 5.40 phosphatase) NM_001826.1 CKS1B CDC28 protein kinase regulatory subunit 1B 1.42 0.89 2.02 NM_001953.2 ECGF1 Endothelial factor 1 (platelet-derived) 1.27 4.40 4.85 NM_022346.1 HCAP-G condensation protein G 1.05 0.96 2.52 NM_006101.1 HEC Highly expressed in , rich in leucine heptad repeats 4.15 3.77 21.36 NM_002358.2 MAD2L1 MAD2 mitotic arrest deficient-like 1 (yeast) 1.66 0.97 3.86 NM_004526.1 MCM2 MCM2 minichromosome maintenance deficient 2, mitotin (S. cerevisiae) 1.38 1.07 3.13 NM_002388.2 MCM3 MCM3 minichromosome maintenance deficient 3 (S. cerevisiae) 1.34 0.85 2.26 AA807529 MCM5 MCM5 minichromosome maintenance deficient 5, cell division cycle 46 1.72 0.72 3.22 (S. cerevisiae) NM_005915.2 MCM6 MCM6 minichromosome maintenance deficient 6 (MIS5 homolog, S. pombe) 1.52 1.56 3.13 (S. cerevisiae) NM_002592.1 PCNA Proliferatingcell nuclear antigen 1.77 1.78 3.38 BC001422.1 PGF Placental growth factor, vascular endothelial growth factor-related protein 0.23 13.02 9.11 NM_005030.1 PLK Polo-like kinase (Drosophila) 1.44 1.52 3.89 NM_000946.1 PRIM1 Primase, polypeptide 1, 49 kDa 2.08 0.91 2.43 NM_021000.1 PTTG1 Pituitary tumor-transforming1 1.29 0.79 3.76 BC003186.1 Pfs2 DNA replication complex GINS protein PSF2 1.93 0.82 4.69 BC001866.1 RFC5 Replication factor C (activator 1) 5, 36.5 kDa 1.78 1.20 2.29 NM_006397.1 RNASEH2A Ribonuclease H2, large subunit 2.11 0.87 2.83 BC005264.1 RPA3 Replication protein A3, 14 kDa 1.71 1.15 3.05 BC001886.1 RRM2 Ribonucleotide reductase M2 polypeptide 1.81 0.77 8.55 NM_014264.1 STK18 Serine/threonine kinase 18 0.80 0.67 2.49 NM_003236.1 TGFA Transforminggrowthfactor, alpha 0.54 2.80 1.54 AF098158.1 TPX2 TPX2, microtubule-associated protein homolog( Xenopus laevis) 1.26 0.41 2.57 NM_001952.2 E2F6 E2F transcription factor 6 1.01 0.63 2.24 NM_003620.1 PPM1D Protein phosphatase 1D magnesium-dependent, delta isoform 2.31 1.67 8.50 AF022375.1 VEGF Vascular endothelial growth factor 1.46 0.92 5.90 NM_002094.1 GSPT1 G1 to S phase transition 1 0.93 1.49 4.00 NM_014574.1 STRN3 Striatin, calmodulin bindingprotein 3 0.92 2.70 1.05 NM_002748.1 MAPK6 Mitogen-activated protein kinase 6 0.92 0.46 4.19 NM_002467.1 MYC v-myc myelocytomatosis viral oncogene homolog (avian) 2.29 1.17 6.12 NM_002266.1 KPNA2 alpha 2 (RAG cohort 1, alpha 1) 1.71 1.36 3.61 NM_006732.1 FOSB FBJ murine osteosarcoma viral oncogene homolog B 1.70 0.94 5.40

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2605 Table 1 (continued ) GenBank Symbol Gene/protein Fold change from nonirradiated 2h 24h 72h

NM_005197.1 CHES1 Checkpoint suppressor 1 1.42 0.89 2.02 N33167 CDKN1C Cyclin-dependent kinase inhibitor 1C (p57, Kip2) 1.27 4.40 4.85 M73554.1 CCND1 Cyclin D1 (PRAD1: parathyroid adenomatosis 1) 1.05 0.96 2.52 Z24459 MTCP1 Mature T-cell proliferation 1 4.15 3.77 21.36 NM_001253.1 CDC5L CDC5 cell division cycle 5-like (S. pombe) 1.66 0.97 3.86 L20320.1 CDK7 Cyclin-dependent kinase 7 (MO15 homolog, Xenopus laevis, cdk-activatingkinase) 1.38 1.07 3.13

Repair D21089.1 XPC Xeroderma pigmentosum, complementation group C 2.52 2.39 1.81 U64315.1 ERCC4 Excision repair cross-complementingrodent repair deficiency, complementation 0.71 0.80 1.42 (XPF) group F

Histones BC002649.1 HIST1H1C 1, H1c 2.31 1.22 2.57 NM_021052.1 HIST1H2AE Histone 1, H2ae 6.86 2.41 1.76 NM_003523.1 HIST1H2BE Histone 1, H2be 4.09 3.67 2.48 NM_003522.1 HIST1H2BF Histone 1, H2bf 4.67 2.98 1.42 NM_003525.1 HIST1H2BI Histone 1, H2bi 3.43 2.76 2.14 BC000893.1 HIST1H2BK Histone 1, H2bk 1.32 2.62 1.34 AA451996 HIST2H2AA Histone 2, H2aa 3.83 2.00 0.94 NM_003528.1 HIST2H2BE Histone 2, H2be 6.16 1.90 0.71

Apoptosis NM_001168.1 BIRC5 Baculoviral IAP repeat-containing5 (survivin) 2.22 0.47 7.54 NM_000633 BCL2 B-cell CLL/lymphoma 2 1.24 0.49 0.66 U66879 BAD BCL2-antagonist of cell death 1.39 0.71 1.37 NM_004281 BAG3 BCL2-associated athanogene 3 0.49 0.66 0.75 NM_001188 BAK1 BCL2-antagonist/killer 1 0.88 1.87 1.37 AA457021 BAG5 BCL2-associated athanogene 5 0.76 0.63 0.45 AF060922 BNIP3L BCL2/adenovirus E1B 19 kDa interactingprotein 3-like 0.85 0.81 2.06 NM_004346 CASP3 Caspase 3, apoptosis-related cysteine protease 0.59 0.88 0.64 U25804 CASP4 Caspase 4, apoptosis-related cysteine protease 1.31 1.64 1.48 BF439983 CASP8 caspase 8, apoptosis-related cysteine protease 1.33 1.97 1.28 NM_003655.1 CBX4 Chromobox homolog4 (Pc class homolog, Drosophila) 1.85 0.40 0.66 NM_015322.1 FEM1B fem-1 homologb ( Caenorhabditis elegans) 0.24 0.99 0.81 NM_002015.2 FOXO1A Forkhead box O1A (rhabdomyosarcoma) 0.40 0.53 0.70 NM_001562.1 IL18 Interleukin 18 (interferon-gamma-inducing factor) 1.16 0.35 1.55 NM_007350.1 PHLDA1 Pleckstrin homology-like domain, family A, member 1 0.20 2.60 2.55 NM_014456.1 PDCD4 Programmed cell death 4 (neoplastic transformation inhibitor) 1.15 0.21 0.78 NM_006378.1 SEMA4D Sema domain, immunoglobulin domain (Ig), transmembrane domain (TM) and 0.93 0.36 0.83 short cytoplasmic domain, (semaphorin) 4D NM_004760.1 STK17A Serine/threonine kinase 17a (apoptosis-inducing) 0.18 1.26 0.53 NM_004226.1 STK17B Serine/threonine kinase 17b (apoptosis-inducing) 0.27 0.42 0.50 BF224259 SPF30 Splicingfactor 30, survival of motor neuron-related 0.35 0.90 0.54 NM_006290.1 TNFAIP3 Tumor necrosis factor, alpha-induced protein 3 0.41 0.62 0.49 NM_006048.1 UBE4B ubiquitination factor E4B (UFD2 homolog, yeast) 1.32 0.45 0.66 NM_002467.1 MYC v-myc myelocytomatosis viral oncogene homolog (avian) 0.43 0.16 0.31 NM_004052.2 BNIP3 BCL2/adenovirus E1B 19 kDa interactingprotein 3 1.31 1.86 3.27 NM_001279.1 CIDEA Cell death-inducingDFFA-like effector a 0.89 1.36 2.99 AF053641.1 CSE1L CSE1 chromosome segregation 1-like (yeast) 1.43 2.23 2.54 NM_004938.1 DAPK1 Death-associated protein kinase 1 1.14 5.41 2.45 NM_004944.1 DNASE1L3 Deoxyribonuclease I-like 3 0.92 7.81 3.13 NM_002462.1 MX1 Myxovirus (influenza virus) resistance 1, interferon-inducible protein p78 (mouse) 1.61 4.07 5.14 BC001808.1 NME6 Nonmetastatic cells 6, protein expressed in (nucleoside-diphosphate kinase) 1.17 2.99 2.22 NM_007350.1 PHLDA1 Pleckstrin homology-like domain, family A, member 1 0.20 2.60 2.55 AF016266.1 TNFRSF10B TRAIL, tumor necrosis factor receptor superfamily, member 10b 0.34 2.33 0.78

Cytokines and chemokines NM_004887.1 CXCL14 Chemokine (C-X-C motif) ligand 14 1.26 0.21 0.87 NM_006729.1 DIAPH2 Diaphanous homolog2 ( Drosophila) 0.98 0.16 0.58 AF196186.1 PARD3 Par-3 partitioningdefective 3 homolog( C. elegans) 1.14 0.34 1.01 NM_020299.1 AKR1B10 Aldo-keto reductase family 1, member B10 (aldose reductase) 1.90 10.23 11.95 NM_001175.1 ARHGDIB Rho GDP dissociation inhibitor (GDI) beta 1.03 2.94 1.83 BG327863 CD24 CD24 antigen (small-cell lung cluster 4 antigen) 0.68 2.82 1.84 M92934.1 CTGF Connective tissue growth factor 1.60 0.99 4.65 NM_004942.2 DEFB4 Defensin, beta 4 1.18 59.75 30.81 NM_001935.1 DPP4 Dipeptidylpeptidase 4 (CD26, adenosine deaminase complexingprotein 2) 1.60 1.68 3.58 NM_000505.2 F12 Coagulation factor XII (Hageman factor) 2.01 3.68 3.80

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2606 Table 1 (continued ) GenBank Symbol Gene/protein Fold change from nonirradiated 2h 24h 72h

NM_005532.1 IFI27 Interferon, alpha-inducible protein 27 2.87 3.63 12.10 NM_006435.1 IFITM2 Interferon-induced transmembrane protein 2 (1-8D) 1.77 2.62 2.40 BF338947 IFITM3 Interferon-induced transmembrane protein 3 (1-8U) 1.92 4.60 5.28 NM_001560.1 IL13RA1 Interleukin 13 receptor, alpha 1 1.12 3.58 1.07 NM_003856.1 IL1RL1 interleukin 1 receptor-like 1 0.32 8.19 6.01 NM_001562.1 IL18 Interleukin 18 (interferon-gamma-inducing factor) 1.16 0.35 1.55 NM_004633.1 IL1R2 Interleukin 1 receptor, type II 0.24 1.05 0.44 NM_003856.1 IL1RL1 Interleukin 1 receptor-like 1 0.32 8.19 6.01 NM_014432.1 IL20RA Interleukin 20 receptor, alpha 1.37 0.36 0.67 NM_006850.1 IL24 Interleukin 24 0.61 4.15 1.34 NM_000565.1 IL6R Interleukin 6 receptor 0.15 0.50 0.41 AF043337.1 IL8 Interleukin 8 2.15 1.12 0.27 NM_001557.1 IL8RB Interleukin 8 receptor, beta 0.96 2.83 1.78 NM_004030.1 IRF7 Interferon regulatory factor 7 2.66 1.78 1.09 NM_006674.1 MICA MHC class I polypeptide-related sequence A 3.78 1.79 0.97 NM_006187.1 OAS3 20-50-oligoadenylate synthetase 3, 100 kDa 1.34 2.00 2.84 NM_003999.1 OSMR Oncostatin M receptor 0.84 2.73 0.91 AI346350 PMSCL1 Polymyositis/scleroderma autoantigen 1, 75 kDa 1.44 0.97 5.78 NM_006404.1 PROCR Protein C receptor, endothelial (EPCR) 1.01 3.86 3.62 AB005043.1 SOCS1 Suppressor of cytokine signaling 1 3.02 3.04 1.89

Oxidatave stress related NM_001752.1 CAT 1.15 0.24 0.89 NM_002085.1 GPX4 Glutathione 4 (phospholipid hydroperoxidase) 1.76 0.51 1.43 NM_005109.1 OSR1 Oxidative-stress responsive 1 0.29 1.41 0.74 L19185 PRDX2 2 1.34 0.39 1.29 NM_000963.1 PTGS2 Prostaglandin-endoperoxide synthase 2 (prostaglandin G/H synthase and 0.76 0.18 0.71 cyclooxygenase) NM_005410.1 SEPP1 Selenoprotein P, plasma, 1 1.29 0.26 0.79

Genes with at least twofold changes in expression in all volunteers at least at one time-point are listed.

patients and only weakly expressed in normal skin (Abts expression at 24 h and sustained upregulation at 48 h et al., 1999). Hurpin overexpression in HaCaT cells after 2 MED of solar simulator radiation. In our study, confers resistance to UV-induced apoptosis (Welss et al., TIMP-1 was upregulated already at 2 h and remained 2003). Our findingof overexpression of hurpin in UVB- high for 72 h with a distinct peak at 24 h. It is intriguing irradiated normal human epidermis suggests a role for that a single low dose of UVB induces long-lasting hurpin in the hyperproliferative state conferred to UV- activity of MMP-1 with deleterious effects on dermal irradiated epidermis (Lee et al., 2002). extracellular matrix. However, the ratio of MMP-1 and TIMP-1 is considered to be relevant for determiningthe amount of active MMP-1 (Sudel et al., 2003). Calculat- Extracellular matrix. The matrix metalloproteinases ingthese ratios with our data (0.15, 3.38 and 0.75 for (MMP) form a family of structurally and functionally the 2, 24 and 72 h time-points) indicates that MMP-1 related zinc endopeptidases capable of degrading activity followinga single UVB dose is short lived and is structural proteins in connective tissue with implications abolished at 72 h. SKALP/PI3, a recently described on tissue plasticity and migration of tumor cells proteinase inhibitor absent from normal epidermis but (Birkedal-Hansen et al., 1993; Kahari and Saarialho- found in hyperproliferative skin such as psoriasis and Kere, 1997). MMPs are tightly regulated by a special healingwounds, and regulated by UVB and proinflam- class of tissue inhibitors of matrix protease, TIMP-1. matory cytokines (Pfundt et al., 2000; Tanaka et al., MMP induction by UV has been implicated in the 2000b), was strongly upregulated at 24 h and remained qualitative and quantitative alterations in the composi- high at 72 h in our list. tion of the dermal extracellular matrix, a hallmark of photoaged skin (Scharffetter et al., 1991; Fisher et al., 1996). In our list, MMP-1 and MMP-3 expression was . Cell-cycle progression and proliferation is a dramatically induced after 24 h and remained upregu- tightly controlled process induced by a wide range of lated though to a lower degree at 72 h. These findings exogenous events including UVB radiation that is are in accord with those of Fisher et al. (1996) regulated by a complex series of endogenous processes who demonstrated induction of MMP-1 and MMP-3 (Celis et al., 1987). We found that most of the cell-cycle at 16–24 h in human skin biopsies following2 MED regulator genes were strongly upregulated at 72 h of UVB radiation, and those of Lahmann et al. (2001) but only partially at earlier time-points, indicatinga and Sudel et al. (2003) who found a peak of MMP-1 late burst of cell-cycle progression and proliferation.

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2607 Interestingly, cyclins A2, B2, D2 and G2 as well as exposure. The interaction of photoproducts with his- PCNA were upregulated, consistent with the lack of tones results in modulation of the structure GADD45 and WAF1/CIP1/ upregulation (Table 2). causingaltered cromatin structure with increased accessibility to repair (Brown et al., 1993; Liu Repair. UVB causes DNA lesions in epidermal cells, et al., 2000). Interestingly, an early and strong upregu- predominantly cyclobutane pyrimidine dimers (CPD) lation of eight different histone genes was included in and pyrimidine–pyrimidone [6–4] photoproducts (Sage, our list. 1993). The biological significance of these DNA lesions depends upon the capacity of the cell to repair the Apoptosis. UV-induced apoptosis is a highly complex damage before it can be incorporated permanently into process in which different molecular pathways such as the genome. Typically, DNA damage is rapidly repaired DNA damage, activation of the at a relative high rate by inducing upregulation of p53, triggering of cell death receptors, and ROS are excision repair (Young et al., 1996). In our involved (Aragane et al., 1998; Wehrli et al., 2000). In list, XPC, which is involved in DNA excision repair with our list, genes belonging to the death receptor pathway special affinity to DNA UV photoproducts (Kosmoski such as TRAIL-R2 and DAPK-1 were upregulated at et al., 2001), was clearly upregulated already 2 h after the 24 h time-point, and were shut down at 72 h,

Table 2 p53 target gene expression changes in intact human epidermis following in vivo UVB irradiation Symbol Gene product 2 h 24 h 72 h

Patient Patient Patient

I II III I II III I II III

APAF1 Apoptotic protease-activatingfactor 0.9 0.7 0.7 0.3 0.9 1.6 0.5 1.1 0.3 BAI1 Brain-specific angiogenesis inhibitor 1 1.2 0.5 0.6 0.5 0.6 0.6 0.3 0.2 0.4 BAX BCL2-associated X protein 1.3 1.0 1.5 1.9 2.0 2.0 1.9 0.2 0.4 BID BH3 interactingdomain death agonist 2.5 1.1 0.6 6.5 1.9 0.7 4.0 1.1 0.8 BTG2 BTG family, member 2 1.4 1.0 1.4 1.0 0.5 0.9 0.8 0.6 0.8 CASP6 Caspase 6, apoptosis-related cysteine protease 1.7 3.2 1.6 0.5 1.1 0.6 1.0 1.9 0.9 CCNG1 Cyclin G1 0.6 1.0 0.8 1.6 2.6 1.5 1.4 2.0 1.6 CDKN1A Cyclin-dependent kinase inhibitor 1A (p21, Cip1) 0.5 0.6 0.8 1.0 0.9 1.1 0.7 0.7 0.7 GADD45A Growth arrest and DNA-damage-inducible, alpha 1.1 0.9 0.9 1.1 0.6 0.8 1.0 0.5 0.5 GADD45B growth arrest and DNA-damage-inducible, beta 3.2 1.4 1.1 0.6 0.6 0.7 0.8 0.7 0.9 GADD45G Growth arrest and DNA-damage-inducible, gamma 1.1 1.9 3.0 0.6 2.6 2.3 1.4 0.4 2.1 GGA1 Golgi associated, gamma adaptin ear-containing, 1.3 1.1 1.1 1.1 1.1 1.1 1.1 1.3 1.1 ARF-bindingprotein 1 GML GPI anchored molecule-like protein 10.6 6.5 0.4 4.9 12.1 2.0 8.0 3.0 0.9 HERC1 Hect (homologous to the E6-AP (UBE3A) carboxyl 1.2 0.9 1.1 0.6 0.4 0.4 0.6 0.7 0.9 terminus) domain and RCC1 (CHC1)-like domain (RLD) 1 IGFBP3 Insulin-like growth factor-binding protein 3 0.5 0.3 0.6 1.0 0.6 0.5 1.5 0.6 0.5 MDM2 Mdm2, transformed 3T3 cell double minute 2, p53-binding 0.5 1.4 0.8 1.1 1.4 1.9 1.4 2.3 0.7 protein (mouse) MDM4 Mdm4, transformed 3T3 cell double minute 4, p53-binding 1.7 3.7 0.3 1.2 0.5 0.3 16.0 0.7 0.8 protein (mouse) NOX1 NADPH oxidase 1 8.0 2.3 0.9 1.0 1.4 2.3 2.3 0.8 0.4 NOX3 NADPH oxidase 3 1.3 0.7 1.7 0.8 0.5 1.2 1.3 1.1 1.2 NOX4 NADPH oxidase 4 1.0 1.5 2.0 0.4 1.5 4.6 1.3 1.1 2.3 NOX5 NADPH oxidase, EF hand calcium-bindingdomain 5 1.9 5.7 0.5 0.9 9.2 1.1 1.2 6.5 0.8 P53AIP1 p53-regulated apoptosis-inducing protein 1 1.5 2.3 1.9 0.4 0.4 0.3 1.1 2.0 3.0 PERP p53-induced protein PIGPC1 1.1 1.2 1.1 0.8 0.9 1.1 0.9 1.1 1.0 PLAGL1 Pleiomorphic adenoma gene-like 1 1.1 1.2 1.0 0.7 0.4 0.8 0.6 1.1 1.1 REPRIMO Candidate mediator of the p53-dependent G2 arrest 0.9 0.9 1.1 0.8 0.4 0.8 0.1 0.3 0.7 TNFRSF6 Tumor necrosis factor receptor superfamily, member 6 1.1 0.7 0.7 1.4 1.2 1.5 1.1 0.9 1.0 TP53 Tumor protein p53 (Li–Fraumeni syndrome) 1.6 1.5 1.3 0.7 0.6 0.5 0.9 0.9 0.8 TP53BP1 Tumor protein p53 bindingprotein, 1 2.0 1.7 1.3 1.2 0.8 0.1 1.3 1.1 1.4 TP53BP2 Tumor protein p53-bindingprotein, 2 0.4 0.3 0.6 1.1 1.2 1.1 0.7 0.8 0.7 WIG1 p53 target zinc finger protein 1.1 0.8 0.8 1.1 1.5 1.0 0.9 0.7 0.9 ERCC3 Excision repair cross-complementing(xeroderma 1.5 1.6 0.9 1.1 1.1 0.9 0.9 1.1 0.9 (XPB) pigmentosum group B complementing) XPA Excision repair cross-complementing(xeroderma 1.9 1.5 1.5 0.9 0.8 0.9 0.6 0.9 1.2 pigmentosum group A complementing) XPC Excision repair cross-complementing(xeroderma 2.5 1.6 1.7 1.9 1.6 2.5 1.0 0.9 2.1 pigmentosum group C complementing) ERCC5 Excision repair cross-complementing(xeroderma 1.2 2.1 1.7 0.5 0.7 0.5 0.7 1.4 1.4 (XPG) pigmentosum group G complementing)

Numbers show fold change of expression.

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2608 underscoringthe transient character of the apoptotic sigma (Hermeking et al., 1997). p53 target genes present response. Supportingthe picture of an early burst of on the Affymetrix Human Focus oligonucleotide arrays proapoptotic activity, the antiapoptotic gene, Bcl-2, was are listed in Table 2. Surprisingly, applying the filter downregulated at 24 h, whereas the apoptosis inhibitor, used throughout this paper, that is, twofold change in all BIRC5 (Survivin), was strongly upregulated at 72 h. three volunteers, only GML was upregulated and BAI1 Evidence of involvement of the ROS pathway is found downregulated at 24 and 72 h, respectively. Applying a in the downregulation of catalase, glutathione perox- less strict filter of 1.5-fold changes, additional p53 target idase, oxidative-stress responsive 1, peroxiredoxin 2 and genes were changed, including BAX (up at 24 h), CASP6 prostaglandin-endoperoxide synthase. (up at 2 h), CCNG1 (up at 24 h), HERC1 (down at 24 h), Recently, death-receptor-mediated, transcription-in- IGFBP3 (down at 2 h), TP53BP2 (down at 2 h), XPA dependent signaling modulation has emerged as an (up at 2 h) and XPC (up at 2 and 24 h). However, important determinant of cell survival duringdevelop- even with these less restrictive cutoff criteria, key p53 ment and cellular homeostasis. These mechanisms targets such as WAF1/CIP1/p21, GADD45 and Mdm2 involve key post-translational modifications that affect remained unaffected followingUVB exposure. the activities of proteins at different levels in the To verify the validity of these profilingdata, we first signaling pathways (Tran et al., 2004). Thus, UVB- ascertained the accumulation of p53 protein in human induced apoptosis, here evidenced by the presence of epidermis followingexposure to 4 MED of UVB in vivo sunburn cells (Figure 3a) and by a diffuse, epidermal (Figure 4). Our results confirmed that p53 has a very low annexin V staining (Figure 3b), might be mediated by basal expression in human skin and accumulates upon both gene activation and by transcription-independent UV irradiation (Hall et al., 1993; Liu et al., 1994; mechanisms. Courtois et al., 1997). We next confirmed the microarray data by semiquantitative RT–PCR usingprimers for key p53 target. As can be seen in Figure 5, neither WAF1/ p53. The ‘guardian of the genome’, p53, is a tumor CIP1/p21, GADD45, Bax or Mdm2 gene expression was suppressor protein that plays a key role in the protective upregulated in spite of p53 protein accumulation 24 h responses against exogenous injuries such as UV by followingUVB exposure, whereas SCCA1, included as a initiatinga cascade of fine-tuned events through positive control, was clearly increased. Thus, our data transactivation of its target genes. Target genes of p53 show that in spite of p53 protein accumulation in include genes involved in cell-cycle arrest such the epidermal cells followinga single in vivo UVB exposure, cyclin-dependent kinase inhibitor WAF1/CIP1/p21 (el p53 did not cause transactivation of its known target Deiry et al., 1993), the proto-oncogene Mdm2 (Barak genes. et al., 1993), a negative regulator of p53, the growth Our current understandingof the regulation of p53 in arrest and DNA damage-inducible GADD45 (Kastan human skin followingUV damageis mostly based on et al., 1992), Bax (Miyashita et al., 1994), Cyclin G keratinocyte cultures. To which degree such models (Okamoto and Beach, 1994), PCNA (Shivakumar et al., reflect genuine physiological processes is largely un- 1995), IGF-BP3 (Buckbinder et al., 1995), B99 (Utrera known. Thus, the p53 repertoire and signaling pathways et al., 1998), p53R2 (Tanaka et al., 2000a) and 14-3-3 of the cell is determined by several biological parameters such as cell type, oncogenic composition, the nature and the intensity of the extracellular stimulus, and the level of p53 (Lassus et al., 1996; Li and Ho, 1998; Sionov and Haupt, 1999; Zhao et al., 2000). Furthermore, the resistance of keratinocytes to apoptosis depends on the differentiation status and the location of the cells within the epidermal layers (Qin et al., 2002). Many of these parameters are not well controlled in the available cell culture models. Our unexpected findingthat only few of the p53 target genes were activated following a single in vivo UVB exposure might reflect crucial differences in

Figure 3 Apoptosis stainingof human epidermis 24 h after exposure to UVB. (a) Sunburn cells stained by H&E; (b) Frozen Figure 4 p53 stainingof human epidermis 24 h after exposure to sections stained for annexin V. UVB. (a) Unexposed; (b) UVB-exposed skin.

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2609

Figure 5 Reverse transcriptase–polymerase chain reaction (RT– PCR) verification of selected p53 target genes 24 h after in vivo UVB irradiation of intact human epidermis. Nonirradiated epidermis served as a control. RNA was obtained from four volunteers different from those tested by microarray analysis. G3PDH was used as housekeepinggene.SCCA1 served as a positive control. The PCR product was stained with ethydium bromide (a), and signal intensities were normalized for G3PDH by Figure 6 Venn diagram representing up- and downregulated genes densitometry (b). in in vivo UVB-irradiated intact human epidermis and in vitro exposed NHEK. The diagram shows the number of genes with the indicated expression patterns. terms of cellular composition, differentiation, intra- cellular p53 levels and UV sensitivity existingbetween Comparison of UVB-regulated genes in intact epidermis cell cultures and in vivo models like the one we have and cultured keratinocytes used. It is of interest to note that in a recent paper, Intact human epidermis is a well-ordered multilayer Marionnet et al. (2003) examined the expression profile organ in which the majority of the cells, that is, cells of of human epidermis following in vivo low-dose solar the suprabasal layers, are keratinocytes undergoing simulator UV exposure and found GADD45 and Bax to various stages of differentiation. In contrast, subcon- be moderately upregulated. Although a different light fluent cultured NHEK is a monolayer system consisting source was used in this study and verification of the of exponentially proliferatingkeratinocytes. In order to upregulation of the two genes was not performed, the investigate whether these differences in maturity influ- data appear to contradict our findings. However, ence the molecular repertoire of epidermal cells in whereas the epidermal sheets in our study originated response to UVB, we compared the in vivo expression from chronically sun-exposed forearm skin, the derma- profiles of epidermal sheets from the present study to tome specimens in the Marionnet et al. (2003) paper that of NHEK exposed to UVB in vitro. The profiling were obtained from buttock skin, not previously results of the NHEK were previously published (Dazard exposed to sunlight. This might partially explain the et al., 2003) and the full list of UVB-regulated genes can apparent discrepancy in GADD45 and Bax regulation, be found at http://www.weizmann.ac.il/home/ligivol/ since pre-exposure of human keratinocytes to a low dose UVB_project /table2.xls. In total, we identified 1931 of UVB significantly alters the p53 signaling pathways genes that differed in their expression profiles among in and apoptotic response compared to cells that were not vivo and in vitro irradiated epidermal cells, for which the pre-exposed (Decraene et al., 2004). overlappingand nonoverlappinggenesamongthe Possible mechanisms resultingin lack of activity of groups are illustrated in Figure 6. When directly p53 transcription targets despite increase in p53 protein comparingthe UVB response of individual genes levels include inhibitory modification of the p53 proteins (Table 3), we found multiple differences in genes such as dephosphorylation, mono-ubiquitination, ned- governing crucial cellular functions such as p53, cell- dylation (Shieh et al., 1997; Siliciano et al., 1997; Wang cycle arrest, DNA repair and apoptosis. The overall et al., 2004; Xirodimas et al., 2004) or the expression of picture exhibited by these differences in expression fits a inhibitory DNp63 (Yang et al., 1998). Elucidation of the pattern of exponentially proliferatingcells reactingto mechanisms involved in the silencingof the p53 UVB damage by initiating cellular processes associated signaling pathways following in vivo UVB irradiation with cell-cycle arrest and apoptosis, whereas differen- of human skin might provide clues for understanding tiatingcells respond by DNA repair (http://www. the processes allowingthe proliferation of transformed hadassah.org.il/Departments/photobiology; for the full cells in human skin. list) (Qin et al., 2002; Chaturvedi et al., 2004). Thus, key

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2610 Table 3 Comparison of UVB-induced expression changes of in vivo exposed intact epidermis and in vitro irradiated NHEK GenBank Symbol Gene/protein Epidermis NHEK

Cell cycle NM_003914.1 CCNA1 Cyclin A1 ¼ m NM_001759.1 CCND2 Cyclin D2 mk NM_004701.2 CCNB2 Cyclin B2 mk AW134535 CCNG2 Cyclin G2 mk NM_000946 PRIM1 DNA primase, G1 to S cell-cycle reactome m ¼ NM_001269 CHC1 Chromosome condensation 1, (G1/S transition of m ¼ mitotic cell cycle DNA packaging) N33167 CDKN1C Cyclin-dependent kinase inhibitor 1C (p57, Kip2) km NM_005197 CHES1 Checkpoint suppressor 1 k ¼ M68520 CDK2 Cyclin-dependent kinase 2 k ¼ U49844 ATR Ataxia telangiectasia and Rad3 related kk

Repair NM_000380.1 XPA Xeroderma pigmentosum, complementation group A ¼ k D21089.1 XPC Xeroderma pigmentosum, complementation group C m ¼ NM_000107 DDB2 Damage-specific DNA-binding protein 2, nucleotide-excision repair m ¼ NM_002691 POLD1 Polymerase delta 1, DNA replication DNA repair response to UV m ¼ NM_006297 XRCC1 X-ray repair complementingdefective repair m ¼ AL080203 POLE Polymerase (DNA directed), epsilon; DNA replication and DNA repair m ¼ AJ243797 TREX1 Three prime repair exonuclease 1 m NM_002690 POLB Polymerase (DNA directed), beta (DNA replication and DNA repair) ¼ m NM_001983 ERCC1 Excision repair cross-complementingrodent repair deficiency, ¼ m complementation group 1 NM_002412 MGMT O-6-methylguanine-DNA methyltransferase; DNA ligation, DNA repair ¼ m

Apoptosis M14745 Bcl-2 ¼ m NM_004052 BNIP3 BCL2/adenovirus E1B-interactingprotein 3 (antiapoptosis) m ¼ AL117381 BCL2L1 BCL2-like 1 (antiapoptosis) m ¼ M59465 TNFAIP3 TNFa- induced protein 3 km NM_001066 TNFRSF1B Tumor necrosis factor receptor superfamily, member 1B ¼ m BF439983 CASP8 Caspase 8, apoptosis-related cysteine protease mk U13699 CASP1 Caspase 1, apoptosis-related cysteine protease ¼ m AK024029 MOAP1 Modulator of apoptosis 1 ¼ m U83981 PPP1R15A , apoptosis response to DNA damage stimulus ¼ m BF686824 DAPK3 Death-associated protein kinase 3 k ¼ NM_001166 BIRC2 Baculoviral IAP repeat-containing2 (antiapoptosis) ¼ k

p53 target genes NM_015675.1 GADD45B Growth arrest and DNA-damage-inducible, beta ¼ m NM_000389.1 CDKN1A Cyclin-dependent kinase inhibitor 1A (p21, Cip1) ¼ m NM_001924.2 GADD45A Growth arrest and DNA-damage-inducible, alpha ¼ m M31159.1 IGFBP3 Insulin-like growth factor-binding protein 3 k ¼ NM_005426.1 TP53BP2 Tumor protein p53-bindingprotein, 2 k ¼ NM_003922.1 HERC1 Hect (homologous to the E6-AP (UBE3A) carboxyl terminus) k ¼ domain and RCC1 (CHC1)-like domain (RLD) 1 NM_002656.1 PLAGL1 Pleiomorphic adenoma gene-like 1 k ¼ NM_005657.1 TP53BP1 Tumor protein p53-bindingprotein, 1 ¼ k

Growth factors and oncogenes NM_002467.1 MYC v-Myc myelocytomatosis viral oncogene homolog (avian) kk NM_001953.2 ECGF1 Endothelial cell growth factor 1 (platelet-derived) mm

Epidermal differentiation complex, calcium-binding proteins NM_002965.2 S100A9 S100 calcium-bindingprotein A9 (calgranulin B) mm

Histones NM_017445.1 H2BFT H2B histone family, member T mm NM_003528.1 H2BFQ H2B histone family, member Q mm AA451996 H2AFO H2A histone family, member O mm NM_003522.1 H2BFG H2B histone family, member G mm BC000893.1 H2BFT H2B histone family, member T mm NM_021052.1 H2AFA H2A histone family, member A mm BC002649.1 H1F2 H1 histone family, member 2 mm NM_006026.1 H1FX H1 histone family, member X mm

Serpins (Serine proteases inhibitors) AF119873.1 SERPINA1 Serine (or cysteine) proteinase inhibitor, clade A (alpha-1 antiproteinase, m ¼ antitrypsin), member 1

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2611 Table 3 (continued ) GenBank Symbol Gene/protein Epidermis NHEK

AI554300 SERPINB1 Serine (or cysteine) proteinase inhibitor, clade B (ovalbumin), member 1 m ¼ BC005224.1 SERPINB3 SCCA1 Serine (or cysteine) proteinase inhibitor, clade B (ovalbumin), member 3 m ¼

Cytokines and chemokines NM_005532.1 IFI27 Interferon, alpha-inducible protein 27 mm NM_003856.1 IL1RL1 Interleukin 1 receptor-like 1 mm NM_004591.1 CCL20 Chemokine (C-C motif) ligand 20 mm NM_001560.1 IL13RA1 Interleukin 13 receptor, alpha 1 mm AF043337.1 IL8 Interleukin 8 mm NM_001562.1 IL18 Interleukin 18 kk

Extracellular matrix affecting factors NM_002421.2 MMP1 Matrix metalloproteinase 1 (interstitial collagenase) mm NM_001814.1 CTSC Cathepsin C mm NM_003254.1 TIMP1 Tissue inhibitor of metalloproteinase 1 mm

Oxidative stress related NM_002084.2 GPX3 3 (plasma) mm NM_000581.1 GPX1 Glutathione peroxidase 1 mm NM_002083.1 GPX2 Glutathione peroxidase 2 (gastrointestinal) mm NM_001752.1 CAT Catalase kk

Heat shock X51757 HSPA6 Heat shock 70 kDa protein 6 (HSP70B0) kk m, Upregulated; k, downregulated; ¼ , no change. cell-cycle progression markers such as cyclins B2, D2 in the present study was performed with 150–250 mJ/cm2, and G2 were all downregulated and p57/KIP2 upregu- whereas the keratinocyte cultures were exposed to lated in NHEK but not in intact epidermis, where 40 mJ/cm2, usinglight sources with identical UVB spectra. an opposite picture indicatingcell-cycle progression Keepingin mind that approximately 100 mJ/cm 2 of was found. Also amongapoptosis-related genes, a UVB can be encountered upon exposure to 60–90 min functional dichotomy was evident, since the proapop- of midday Summer sun, it is worth noticingthat the titic genes, TNFAIP3, TNFRSF1B, CASP1, MOAP1 UVB exposure delivered to both intact epidermis and and PPP1R15A, were upregulated only in NHEK, to the keratinocyte cultures is environmentally relevant; whereas antiapoptotic genes such as BNIP3 and (3) The hypothesis-generating nature of microarray BCL2L1 were upregulated only in intact epidermis. profilingappears to justify our approach of comparing Likewise, in the DNA repair group, XPC, DDB2, two bioinformatic analyses despite the fact that they POLD1, XRCC1, POLE and TREX1 were only were not performed in parallel. However, future upregulated in intact epidermis. And in the p53 group parallel, hypothesis-driven studies are required to where key target genes such as GADD45A, GADD45B compare selected candidate genes for their function and WAF1/CIP1/p21 were unaffected in intact epider- and relevance under in vivo and in vitro conditions. mis, the same genes were upregulated in NHEK. The fact that the comparison between intact epidermis and cultured keratinocytes was not performed in parallel raises a number of issues: (1) Although both microarray Methods profilings were performed in the same laboratory, Array processing different Affymetrix microarrays were used in the two All experiments were performed usingAffymetrix Human experiments. To allow a direct comparison between the Focus oligonucleotide arrays, as described (http://www. two experiments, a series of compensatory bioinformatic affymetrix.com/support/technical/datasheets/human datasheet. procedures were performed as detailed in Methods to pdf). Total RNA from each sample was used to prepare compensate for potential pitfalls due to differences in biotinylated target RNA, with minor modifications from construction of the microarrays,; (2) UV light impinging the manufacturer’s recommendations (http://www.affymetrix. on intact human epidermis is absorbed or scattered by com/support/technical/manual/expression_manual.affx). pigment cells and by chromophores in the stratum Briefly, 5 mgof mRNA was used to generatefirst-strand corneum, and only 20–30% of the incominglight cDNA by usinga T7-linked oligo(dT)primer. After second- strand synthesis, in vitro transcription was performed with reaches the keratinocytes of the deeper layers of the biotinylated UTP and CTP (Enzo Diagnostics), resulting in epidermis. None of these physiological structures approximately 300-fold amplification of RNA. The target diminish the amount of UV light reaching the mono- cDNA generated from each sample was processed as per the layer keratinocytes in culture. To compensate for these manufacturer’s recommendation usingan Affymetrix Gene- differences in UV accessibility, the in vivo UVB exposure Chip Instrument System (http://www.affymetrix.com/support/

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2612 technical/manual/expression_manual.affx). Briefly, spike con- that were compared to the list of 642 downregulated genes trols were added to 15 mg fragmented cRNA before overnight from the in vivo study. hybridization. Arrays were then washed and stained with streptavidin–phycoerythrin, before beingscanned on an Volunteers Affymetrix GeneChip scanner. Additionally, quality and Three healthy volunteers were recruited after informed consent amount of startingRNA was confirmed usingan agarose and prior approval from the Ethical Committee on Experi- gel. After scanning, array images were assessed by eye to ments on Humans (Helsinki Committee). Inclusion criteria confirm scanner alignment and the absence of significant were skin type II–III, age 18-30 years, non-smokers, no regular bubbles or scratches on the chip surface. 30/50 ratios for or concomitant medication. GAPDH and beta-actin were confirmed to be within acceptable limits (1.01–1.59 and 1.04–1.82), and BioB spike UVB irradiation controls were found to be present on all chips, with BioC, Areas of 9 cm2 on previously unexposed skin of the inner BioD and CreX also present in increasingintensity. When forearms were irradiated with UVB from a bank of four FS40 scaled to a target intensity of 150 (using Affymetrix MAS 5.0 fluorescent lamps (Philips) that emit wavelengths between array analysis software), scalingfactors for all arrays were 280 and 320 nm with a peak at 313 nm. Light intensity, deter- within acceptable limits (1.07–2.08), as were background, mined usinga Waldmann (Waldmann GBH, Schwenningen, Q values and mean intensities. Details of quality control Germany) UV radiometer, was 0.75 mW/cm2 at a distance of measures can be found at http://www.ncbi.nlm.nih.gov/geo/ 20 cm from the light source. Single doses of 150–250 mJ/cm2, and at http://eng.sheba.co.il/genomics. equivalent to 4 MED, were used.

Data analysis p53 immunohistochemistry The 8757 probe sets contained in the Affymetrix Human Focus Formalin-fixed, paraffin-embedded tissue sections of punch- oligonucleotide array were filtered using Mas 5 algorithm biopsies taken from nonirradiated and 24 h after 4 MED (http://www.hadassah.org.il/Departments/photobiology). A UVB-irradiated human skin, and a positive control section list of 5380 ‘valid’ probe sets, representingprobe sets with (carcinoma of the colon) were cut at 4–5 mm, deparaffinized in signals higher than 20 and detected as present (P) in at least a clearingsolution and rehydrated in alcohol. Endogenous one sample, was obtained. Treated and control samples were peroxidase was blocked with 3% hydrogen peroxidase in compared in each time-point. The comparison generated a list distilled water for 5 min. Antigen retrieval was performed by 1 of 619 ‘active genes’ representing probe sets changed by at least microwave heatingat 92 C for 20 min in 1 mmol EDTA, twofold (as calculated from the MAS 5 LogRatio values) pH 8.0. Slides were then immunostained in a Ventana Nexes automated stainingsystem with an AEC stainingkit. Primary (LRX1orLRpÀ1) and detected as ‘Increased’ or as as ‘Decrease’ (I or D, P-value 0.0025) or ‘Marginal Increased’ or antibodies were applied (monoclonal mouse antihuman p53 as or Marginal Decrease (MI or MD, P-value 0.003 ) in all protein, clone D0-7; DAKO, Glostrup, Denmark) in dilution 1 treated sample as compared to all the control samples in at 1/50. Incubation time was 32 min at 37 C. least one time-point. This list excluded upregulated genes in all treated samples with signals lower than 20 or detected as Staining for apoptosis absent, and downregulated gene with base line signals lower Punch biopsies obtained 24 h after UVB irradiation were than 20 and detected as absent in the control samples. stained for apoptosis by H&E (sunburn cells) and annexin V. Hierarchical clusteringwas performed usingSpotfire Deci- Samples were immediately frozen in OCT in liquid nitrogen sionSite for Functional Genomics (Somerville, MA, USA). and were stained with AnV-FITC (Roche Diagnostics, Genes were classified into functional groups using the GO Mannheim, Germany). annotation tool (Dennis Jr et al., 2003). Over-representation calculations were carried out usingEase (Hosack et al., 2003). Epidermal sheets and RNA extraction Functional classifications with an ‘Ease score’ lower than 0.05 Suction blisters (2.5 cm in diameter) were induced 2, 24, and were marked as over-represented. 72 h after irradiation, and RNA was extracted from the blister roofs (Enk and Katz, 1994). Blisters from adjacent nonirra- diated skin served as control. Total RNA was extracted from Data comparison of in vivo and in vitro UVB-regulated genes the blister roofs usingthe SV Total RNA Isolation System The list of UVB-regulated genes following in vivo exposure (Promega, Madison, WI, USA). of intact epidermis obtained from the present study was compared to a previously published list of differentially Semiquantitiative RT–PCR regulated genes in NHEK following in vitro UVB irradiation RT–PCR was performed with the Access RT–PCR system (Dazard et al., 2003). (Promega). Total RNA (1 mg) was used as templates in In order to compare the two experiments, initial compensa- reaction mixtures containing2 ml25mM MgSO4,1ml10mM tion had to be made for the use of different arrays (Affymetrix dNTP mix, 20 pmols of each of the primers for all genes, 5 U Human Focus for the in vivo and U95 Affymetrix arrays for AMV reverse transcriptase, 5 U Tfl DNA polymerase and the in vitro study). To select the modulated genes in the in vivo reaction buffer in a total volume of 50 ml. RT was performed at study, we applied a filter of at least twofold change at 2 or 24 h 481C for 45 min and 951C for 2 min, followed by PCR. Cycling or at both time-points in at least two out of the three conditions and mRNA concentrations for each gene were volunteers. In total, 381 and 642 genes passed this filter for the determined by prior titration experiments. The following up- and downregulated genes, respectively. In the in vitro cyclingtimes and temperatures were used: 31cycles at 94 1 for study, the original list of 273 genes was first intersected with 60 s, 601 for 60 s and 721 for 90 s (p21); 28 cycles at 941 for 60 s, the list of probe sets present on the Focus array, leaving247 601 for 60 s and 721 for 90 s (GAAD45, Bax and mdm2); genes that were then compared to the list of 381 upregulated 28 cycles at 941 for 60 s, 551 for 60 s and 721 for 90 s (G3PDH). genes from the in vivo study. Similarly, the list of 363 In total, 10 ml of the PCR products were run on 2% agarose downregulated genes in the in vitro study was first intersected ethidium bromide gel electrophoresis and visualized under with the list of genes present on Focus array, leaving 202 genes UV. Semiquantitative estimations were obtained by

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2613 correctingfor the G3PDH housekeepinggene.Band intensities profiles between the two groups: The expression profile of in were determined by densitometry measurements usingMulti- vivo exposed intact epidermis was geared towards DNA repair, Analystt/PC Version1.1 (Bio-Rad, CA, USA). Reactions were whereas in vitro irradiated NHEK responded by activating regularly performed in the absence of reverse transcriptase or genes associated with cell-cycle arrest and apoptosis. These template to control for contamination of genomic DNA or differences in expression profiles might reflect differences carryover of cDNA. between mature differentiatingkeratinocytes in the suprabasal Primers sequences were: epidermal layers versus exponentially proliferatingNHEK in p21 sense : 50-AGGATCCATGTCAGAACCGGCTGG-30; cell culture. Of special interest is the p53 profile: In contrast to antisense: 50-CAGGATCCTGTGGGCGGATTAGGGCT-30 the paradigm that key target genes such as GADD45, WAF1/ GAAD45 sense: 50-AGA ACG ACA TCA ACA TCC TGC-30; CIP1/p21, Bax and Mdm2 are upregulated in a properly antisense: 50-AAT GTG GAT TCG TCA CCA GA-30 functioningp53 tumor suppressor cascade, we found that in Bax sense: 50-GGGGACGAACTGGACAGTAA-30; spite of p53 protein accumulation in epidermal cells, several antisense: 50-CAGTTGAAGTTGCCGTCAGA-30 central downstream genes were unaffected by a single in vivo Mdm2 sense: 50-CAGCTTCGGAACAAGAGACC-30; UVB exposure. Assumingthat under physiologicalconditions, antisense: 50-GTCCGATGATTCCTGCTGAT-30; a single UVB exposure has little carcinogenic potential, the SCCA1 sense: 50-CCTGAAGGTAATATTGGCAGC-30; lack of activation of genes involved in cell-cycle arrest and antisense: 50-GGGATGAGAATCTGCCATAG-30; p53 target genes suggests that signaling pathways that can G3PDH sense: 50-TGAAGGTCGGAGTCAACGGATTT cause DNA repair without prior cell-cycle arrest, or that GGT-30; transcription-independent mechanisms are activated in intact antisense: 50-CATGTGGGCCATGAGGTCCACCAC-30 epidermis. It should be noted that we have limited our in depth comparison of UVB-induced expression profiles to the p53 tumor suppressor gene, since it plays a central role in the Conclusions regulation of the UV response. It is indeed possible that other Using global microarray profiling to investigate the molecular conclusions might have been reached had we broadened our response of intact human epidermis to a single, physiological investigation also to include additional molecular pathways dose of in vivo UVB irradiation, we found that genes such as the NF-kB family of transcription factors, known as controllingnumerous cellular functions were regulatedby ‘the central mediator of the human immune response’. Future UVB. Although the majority of the UVB-modulated genes studies should be carried out to elucidate such possibilities. were found at the 24 h time-point, differentially expressed Our findings show that extreme care should be taken when genes could be found already after 2 h. The transient character extrapolatingfrom findingsbased on keratinocyte cultures to of the UVB response was demonstrated by the findingthat half changes in intact epidermis. of the up- and downregulated genes had returned to back- ground levels within 72 h following exposure. Based on the assumption that keratinocyte cultures closely Acknowledgements mirror changes taking place in intact epidermis, our present understandingof the influence of UVB on human skin is This study was supported by grants from The Israel Cancer mainly based on in vitro models. Our study is the first to Association through the donation from The Brown Founda- directly compare the molecular changes in intact human tion, Florida (grant no. 20030004-B) and The Hadassah epidermal cells with those of in vitro exposed NHEK. We Women Zionist Organization of America (HWZOA) Com- found that close to 2000 genes differed in their expression pensatory Research Fund.

References

Abts HF, Welss T, Mirmohammadsadegh A, Kohrer K, Cataltepe S, Gornstein ER, Schick C, Kamachi Y, Chatson K, Michel G, Ruzicka T. (1999). J Mol Biol 293: 29–39. Fries J et al. (2000). J Histochem Cytochem 48: 113–122. Alowami S, QingG, Emberley E, Snell L, Watson PH. (2003). Celis JE, Madsen P, Celis A, Nielsen HV, Gesser B. (1987). BMC Dermatol 3:1. FEBS Lett 220: 1–7. Aragane Y, Kulms D, Metze D, Wilkes G, Poppelmann B, Chaturvedi V, Bacon P, Bodner B, Nickoloff BJ. (2004). Luger TA et al. (1998). J Cell Biol 140: 171–182. J Invest Dermatol 123: 1200–1203. Baadsgaard O. (1991). Arch Dermatol 127: 99–109. Courtois SJ, Woodworth CD, Degreef H, Garmyn M. (1997). Barak Y, Juven T, Haffner R, Oren M. (1993). EMBO J 12: Exp Cell Res 233: 135–144. 461–468. Crombach G, Scharl A, Vierbuchen M, Wurz H, Bolte A. Becker B, Vogt T, Landthaler M, Stolz W. (2001). J Invest (1989). Cancer 63: 1337–1342. Dermatol 116: 983–988. Dazard JE, Gal H, Amariglio N, Rechavi G, Domany E, Birkedal-Hansen H, Moore WG, Bodden MK, Windsor LJ, Givol D. (2003). Oncogene 22: 2993–3006. Birkedal-Hansen B, DeCarlo A et al. (1993). Crit Rev Oral Decraene D, Gan D, Mammone T, Maes D, Declereq L, Biol Med 4: 197–250. Germyn M. (2004). J Investig Dermatol 122: A140. Boni R, Heizmann CW, Doguoglu A, Ilg EC, Schafer BW, Dennis Jr G, Sherman BT, Hosack DA, YangJ, Gao W, Lane Dummer R et al. (1997). J Cutan Pathol 24: 76–80. HC et al. (2003). Genome Biol 4:3. Brash DE, Ziegler A, Jonason AS, Simon JA, Kunala S, Leffell Duk JM, Voorst Vader PC, ten Hoor KA, Hollema H, DJ. (1996). J Investig Dermatol Symp Proc 1: 136–142. Doeglas HM, de Bruijn HW. (1989). Cancer 64: 1652–1656. Broome AM, Ryan D, Eckert RL. (2003). J Histochem el Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, Cytochem 51: 675–685. Trent JM et al. (1993). Cell 75: 817–825. Brown DW, Libertini LJ, Suquet C, Small EW, Smerdon MJ. Enk CD, Katz SI. (1994). Arch Dermatol Res 287: 72–77. (1993). Biochemistry 32: 10527–10531. Enk CD, Shahar I, Amariglio N, Rechavi G, Kaminski N, Buckbinder L, Talbott R, Velasco-Miguel S, Takenaka I, Faha HochbergM. (2004). Photodermatol Photoimmunol Photo- B, Seizinger BR et al. (1995). Nature 377: 646–649. med 20: 129–137.

Oncogene UVB-induced gene expression profile of human epidermis CD Enk et al 2614 Fisher GJ, Datta SC, Talwar HS, WangZQ, Varani J, KangS Pfundt R, Wingens M, Bergers M, Zweers M, Frenken M, et al. (1996). Nature 379: 335–339. Schalkwijk J. (2000). Arch Dermatol Res 292: 180–187. Gibbs S, Boelsma E, Kempenaar J, Ponec M. (1998). Cell Pisarchik A, Wortsman J, Slominski A. (2004). Gene 341: Tissue Res 292: 107–114. 199–207. Gniadecki R, Hansen M, Wulf HC. (1997). J Invest Dermatol Qin JZ, Chaturvedi V, DenningMF, Bacon P, Panella J, 109: 163–169. Choubey D et al. (2002). Oncogene 21: 2991–3002. Grimbaldeston MA, Geczy CL, Tedla N, Finlay-Jones JJ, Sage E. (1993). Photochem Photobiol 57: 163–174. Hart PH. (2003). J Invest Dermatol 121: 1168–1174. Scharffetter K, Wlaschek M, Hogg A, Bolsen K, Schothorst A, Hahn WC, WeinbergRA. (2002). Nat Rev Cancer 2: 331–341. Goerz G et al. (1991). Arch Dermatol Res 283: 506–511. Hall PA, McKee PH, Menage HD, Dover R, Lane DP. (1993). Sellheyer K, Belbin TJ. (2004). J Am Acad Dermatol 51: Oncogene 8: 203–207. 681–692. HermekingH, LengauerC, Polyak K, He TC, ZhangL, Sesto A, Navarro M, Burslem F, Jorcano JL. (2002). Proc Natl Thiagalingam S et al. (1997). Mol Cell 1: 3–11. Acad Sci USA 99: 2965–2970. Horiuchi Y, Tsukahara T, Otoyama K. (1994). J Dermatol 21: Shieh SY, Ikeda M, Taya Y, Prives C. (1997). Cell 91: 325–334. 67–72. Shivakumar CV, Brown DR, Deb S, Deb SP. (1995). Mol Cell Hosack DA, Dennis Jr G, Sherman BT, Lane HC, Lempicki Biol 15: 6785–6793. RA. (2003). Genome Biol 4: R70. Siliciano JD, Canman CE, Taya Y, Sakaguchi K, Appella E, Joos L, Eryuksel E, Brutsche MH. (2003). Swiss Med Wkly Kastan MB. (1997). Genes Dev 11: 3471–3481. 133: 31–38. Sionov RV, Haupt Y. (1999). Oncogene 18: 6145–6157. Kahari VM, Saarialho-Kere U. (1997). Exp Dermatol 6: Sudel KM, Venzke K, Knussmann-HartigE, Moll I, 199–213. Stab F, Wenck H et al. (2003). Photochem Photobiol 78: Kannan K, Amariglio N, Rechavi G, Jakob-Hirsch J, Kela I, 355–360. Kaminski N et al. (2001). Oncogene 20: 2225–2234. Suminami Y, Nagashima S, Vujanovic NL, Hirabayashi K, Kastan MB, Zhan Q, el Deiry WS, Carrier F, Jacks T, Walsh Kato H, Whiteside TL. (2000). Br J Cancer 82: 981–989. WV et al. (1992). Cell 71: 587–597. Takao J, Ariizumi K, Dougherty II, Cruz Jr PD. (2002). Kato H. (1996). Anticancer Res 16: 2149–2153. Photodermatol Photoimmunol Photomed 18: 5–13. Kosmoski JV, Ackerman EJ, Smerdon MJ. (2001). Proc Natl Takeda A, Higuchi D, Takahashi T, Ogo M, Baciu P, Acad Sci USA 98: 10113–10118. Goetinck PF et al. (2002). J Invest Dermatol 118: 147–154. Lahmann C, YoungAR, Wittern KP, BergemannJ. (2001). Tanaka H, Arakawa H, Yamaguchi T, Shiraishi K, Fukuda S, Photochem Photobiol 73: 657–663. Matsui K et al. (2000a). Nature 404: 42–49. Lassus P, Ferlin M, Piette J, Hibner U. (1996). EMBO J 15: Tanaka N, Fujioka A, Tajima S, Ishibashi A, Hirose S. 4566–4573. (2000b). Br J Dermatol 143: 728–732. Lee JH, An HT, ChungJH, Kim KH, Eun HC, Cho KH. Thorey IS, Roth J, Regenbogen J, Halle JP, Bittner M, Vogl T (2002). Photodermatol Photoimmunol Photomed 18: 253–261. et al. (2001). J Biol Chem 276: 35818–35825. Lee KM, Lee JG, Seo EY, Lee WH, Nam YH, YangJM et al. Tran SE, Meinander A, Eriksson JE. (2004). Trends Biochem (2005). Br J Dermatol 152: 52–59. Sci 29: 601–608. Li D, Turi TG, Schuck A, FreedbergIM, Khitrov G, Utrera R, Collavin L, Lazarevic D, Delia D, Schneider C. BlumenbergM. (2001). FASEB J 15: 2533–2535. (1998). EMBO J 17: 5015–5025. Li G, Ho VC. (1998). Br J Dermatol 139: 3–10. WangX, Taplick J, Geva N, Oren M. (2004). FEBS Lett 561: Liu M, Dhanwada KR, Birt DF, Hecht S, PellingJC. (1994). 195–201. Carcinogenesis 15: 1089–1092. Wehrli P, Viard I, Bullani R, Tschopp J, French LE. (2000). Liu X, Mann DB, Suquet C, Springer DL, Smerdon MJ. J Invest Dermatol 115: 141–148. (2000). Biochemistry 39: 557–566. Welss T, Sun J, IrvingJA, Blum R, Smith AI, Whisstock JC Maelandsmo GM, Florenes VA, Mellingsaeter T, Hovig E, et al. (2003). Biochemistry 42: 7381–7389. Kerbel RS, Fodstad O. (1997). Int J Cancer 74: 464–469. Xirodimas DP, Saville MK, Bourdon JC, Hay RT, Lane DP. Marionnet C, Bernerd F, Dumas A, Verrecchia F, Mollier K, (2004). Cell 118: 83–97. Compan D et al. (2003). J Invest Dermatol 121: 1447–1458. YangA, KaghadM, WangY, Gillett E, FlemingMD, Dotsch Miyashita T, Krajewski S, Krajewska M, WangHG, Lin HK, V et al. (1998). Mol Cell 2: 305–316. Liebermann DA et al. (1994). Oncogene 9: 1799–1805. YoungAR, Chadwick CA, Harrison GI, Hawk JL, Nikaido Murakami T, Fujimoto M, Ohtsuki M, Nakagawa H. (2001). O, Potten CS. (1996). J Invest Dermatol 106: 1307–1313. J Dermatol Sci 27: 121–129. ZhangT, Woods TL, Elder JT. (2002). J Invest Dermatol 119: Okamoto K, Beach D. (1994). EMBO J 13: 4816–4822. 1196–1201. Park WY, HwangCI, Im CN, KangMJ, Woo JH, Kim JH Zhao R, Gish K, Murphy M, Yin Y, Notterman D, Hoffman et al. (2002). Oncogene 21: 8521–8528. WH et al. (2000). Genes Dev 14: 981–993.

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

Oncogene