signature=8be03c7ce1ab250184922bb484bc49c2,Gene expression profiling of initiated epidermal cells wi...-CSDN博客

Abstract

High-density oligonucleotide array technology was used to search for gene expression changes that may predict or cause sporadic tumorigenesis. To reduce variation among samples, a non-transformed mouse epidermal keratinocyte clone was compared with its carcinoma-producing or papilloma-producing 7,12-dimethylbenz[ a ]anthracene-initiated cell derivative. The majority of the ~12 325 genes or expressed sequence tags (ESTs) analyzed remained unaltered reflective of the clonal nature of the model. Consistent gene expression changes among biological replicates included 69 in malignancy-prone initiated cells, 46 in papilloma-precursor initiated cells and an additional 45 with changes in both initiated cell lineages. The changes covered a broad spectrum of cellular activities, implying that multiple pathways cooperate at the initiation of carcinogenesis. There was a tendency in the malignancy-prone cells for differences in proliferation and apoptosis-related genes. The benign lineage tended toward aberrant expression of genes involved in differentiation and epidermal barrier function. Several independently confirmed gene expression changes revealed plausible effectors to bypass an activated ras protein in proliferation pathways and to compromise the wild-type p53 and other apoptosis pathways at initiation of carcinogenesis. The differential expression patterns of the initiated cells are consistent with the hypothesis that changes in gene expression at initiation may cooperate with subsequent oncogenic changes to determine malignant fate. The cloned epidermal cell lineages allowed detection of putative early cancer genes in a model that is conducive to testing their direct role in epithelial multistep carcinogenesis in vitro and in vivo .

Introduction

Cancers arise as a result of accumulation of genetic or epigenetic alterations in genes that control cellular proliferation, differentiation, death and migration, with most cancer-causing proto-oncogenes or tumor suppressor genes falling into those key regulatory pathways. The clinical course of familial cancers such as familial retinoblastoma and hereditary colorectal cancer are predictable primarily because of their inherited genetic defects. However, the onset and order of genetic alterations that lead to development of most sporadic cancers remain undefined. Although tumorigenesis has been defined as a multistage process including initiation, pre-malignant lesions and malignant progression, molecular events that determine tumor fate, are largely unknown. Using skin lesions as an example, seborrheic keratosis (SK) and actinic keratosis (AK) are common skin lesions that along with the rarer keratoacanthoma (KA) are related to exposure to sunlight. SK almost never invades or metastasizes and thus is classified as a benign skin lesion ( 1,2 ). KA is an epidermal in situ malignancy that can metastasize at low frequency, yet has the potential to resolve spontaneously ( 3 ). AK is incipient intra-epidermal squamous cell carcinoma (SCC) with frequency associated with risk of metastasis ( 4–6 ). Head and neck cancers, in contrast, frequently progress in spite of surgical and radiation or chemotherapy treatment ( 7 ). The predicted benign versus malignant fates of these various lesions suggest that initial causes (sets of initiated genes) may predispose them to their distinctive malignant outcomes. In the animal two-stage carcinogenesis model, only 5% of dimethylbenz[ a ]anthracene (DMBA)-induced papillomas become malignant SCC whereas the majority remains as benign papillomas ( 8 ). However, UV induces SCC without apparent papilloma intermediates ( 9,10 ). This is not only due to immune suppression, but also probably due to different genetic alterations in the keratinocyte target cells ( 11 ). Taken together, the data indicate that particular sets of initiation genes may predict the aggressiveness of the disease.

Models for testing candidate genes for their role in sporadic cancers include skin grafting of cells initiated in vitro ( 12 ) and emerging transgenic `hit and run' approaches that may mimic focal disease ( 13 ). One of the major challenges to uncovering cancer-causing initiation genes is the inability to obtain pre-cancerous lesions with a known tumorigenic fate. In addition, the heterogeneity of most solid pre-cancerous or malignant lesions is an obstacle to defining the key molecular events. Cellular models in which normal, initiated and tumor cells are derived from a common genetic background offer advantages for selecting genetic events relevant to cancer development, as well as for characterizing the candidate genes in host cells using functional assays of initiation and malignant conversion. For the current study, we used a clonal mouse cancer model, represented in Figure 1 , in which non-transformed skin keratinocytes (cell strain 291) were treated with the carcinogen DMBA, and independently initiated cell clones (03C and 09C) were selected based on their ability to resist extracellular Ca 2+ -induced terminal differentiation ( 14 ). The tumor fate of each initiated cell type is known. The 03C generated poorly differentiated SCC cell line 03R, and 09C benign papilloma line 09R, as determined by skin grafting of initiated cells to sites in athymic nu/nu mice. In this model, p53 protein is wild-type but defective ( 15 ) and, as in most human epithelial cancers, the Ha-ras gene is not activated ( 16 ), thus providing a focus on alternative pathways of epithelial cancer development.

Previously, we have applied RNA differential display and found only one differentially expressed gene at the initiation stage compared with 15 changes at the malignant conversion stage using primer sets capable of detecting 5–10% of the total mRNA populations ( 17 ). The development of high-throughput technology like serial analysis of gene expression ( 18 ) and DNA microarray ( 19,20 ) makes it possible to globally examine complex patterns of gene expression changes. In the present study, we present the results of a broad-based gene expression screen using GeneChip microarray analysis to compare the non-transformed cell clone 291 with two of its initiated cell derivative clones, benign tumor-precursor 09C and malignant tumor-precursor 03C. In addition to certain shared gene expression changes, distinctive expression patterns were observed for 03C and 09C, respectively, supporting the concept that the fate of cancer development might be predisposed at an early stage by particular genetic changes. Knowledge of such changes should help to prioritize potential molecular therapeutic targets for prevention, diagnosis and treatment of epithelial cancers.

Materials and methods

Cell culture

Non-transformed mouse keratinocytes 291 were cultured in EMEM medium with a low concentration of Ca 2+ (0.04 mM), and initiated cells 03C and 09C were maintained in the high Ca 2+ -containing medium (1.4 mM) as described ( 21 ). To minimize expression variations due to cell culture environment, 291, 03C and 09C were amplified in their respective optimum growth conditions then incubated in a low Ca 2+ environment for 24 h prior to harvesting at 70% confluence. Total RNAs were isolated from each cell type performed in biological triplicate using TRIZOL according to the instructions of the manufacturer (Gibco BRL Life Technologies, Gaithersburg, MD).

Oligonucleotide array

Three independent sets of 291, 09C and 03C cells ( a , b and c ) were cultured and a total of nine RNA samples were harvested and subjected to analysis using the murine chip MG-U74Av2, that included 12 488 probe sets representing about 12 325 genes and expressed sequence tags (ESTs, Affymetrix GeneChip ® , Santa Clara, CA). The labeling of the targets and the chip hybridization were performed as described ( 22 ) by the OHSU Cancer Institute Gene Microarray Core. Briefly, 10 μg of total RNAs were initially converted into cDNA by reverse transcription and then labeled as biotinylated cRNA by in vitro transcription. The `target' cRNA population was fragmented before hybridization to the chip. The overall fluorescence intensity was globally scaled to a targeting intensity of 150. Genes that were differentially expressed in 03C and/or 09C compared with 291 were identified by pair-wise comparison of nine pairs, e.g. 291 a versus 03C a , 291 a versus 09C a , 09C a versus 03C a using the Affymetrix program MAS4.0.

Data analysis

The microarray assay noise and the background for each sample were determined using image processing software (Microarray Suite 4.0) provided by the array vendor (Affymetrix). The expression intensity information for all genes was exported into Microsoft Excel. A minimum gene expression intensity value of 50 was selected as a cut off value, being >4-fold above the measured noise and background ( Q ) of the hybridization assay ( Q is in a range of 2.44–7.01 in our study) and being judged as useful for screening out false positives in the data set (C.Harrington, OHSU Affymetrix Microarray Core Manager, personal communication). Housekeeping genes actin and GAPDH (which were synthesized on the chip as internal standards to monitor target labeling and hybridization) were shown to be present and unaltered by pair-wise comparison analysis among samples, consistent with northern and reverse transcription polymerase chain reaction (RT–PCR) analysis of those genes in our cell model system. Furthermore, an indication of excellent overall integrity of the targets interrogated was reflected by the signal ratio being close to 1 of 5′ to 3′ probe sets for both actin and GAPDH. The criteria for selecting differentially expressed genes included the following: (i) fold changes (differences in the expression level in initiated cells compared with 291) must be 2.5 or greater (which exceeds the 2-fold cut-off standard recommended by Affymetrix); (ii) difference calls (a variable generated by pair-wise comparison to show the increase or decrease of transcripts between the two samples) must be consistent (at least two out of three) in biological replicates; (iii) for genes that were down-regulated in the initiated cells compared with 291, the expression intensity of that gene transcript in 291 must be at least 50 and must be called `present'; for those that were up-regulated in the initiated cells compared with 291, the expression intensity in initiated cells must be at least 50 and be called `present' in the initiated cells.

Semi-quantitative RT–PCR

For semi-quantitative RT–PCR, 10 μg of total RNAs were treated with 2.5 U of RNase-free DNase I (Invitrogen, Carlsbad, CA) in the presence of 1 U of RNA nuclease inhibitors in a procedure described by the vendor. cDNA was synthesized from 2 μg of total RNA by RT with 0.3 μM poly dT primer (20mer) and 200 U of Superscript II RNaseH − reverse transcriptase in a 20 μl reaction mixture as described (Invitrogen). Genomic contamination in the cDNA was ruled out using actin primers (Primer 1 AGG CTG GCC TTT GAC ACC TAC CAG G; Primer 2 TCT GTT GTG TTT CCT CCC TGT TGG) that span the intron/exon junction, giving distinct bands for cDNA (300 bp) and genomic actin (481 bp). PCR was carried out in a 50 μl reaction containing 2.5 mM MgCl 2 , 0.1 μM of each primer set, 0.2 μM dNTP mix, 1 μl of cDNA template out of a total of 20 μl produced and 2.5 U of Taq polymerase (MBI Fermentas, Hanover, MD). After hot start at 94°C for 5 min, PCR was programmed as follows: denaturation at 94°C for 30 s, annealing at 58°C for 45 s and extension at 72°C for 30 s. The linear amplification range was determined by ethidium bromide staining of PCR products. The sizes of the PCR product and the primer pairs used for the amplification are as follows: tissue inhibitor of metalloproteinases 3 TIMP-3 (242 bp): 5′-CAGCCCTGTGATACTTGGGT-3′ (forward) and 5′-CAAGCTTCCAGCCAAACTTC-3′ (reverse); stem cell factor (SCF) (266 bp): 5′-AATGCACAACTGCCATCTCC-3′ (forward) and 5′-TCTGTGAAGCCATATAAATGTCAGA-3′ (reverse); MAP kinase phosphatase-1 (MKP-1) (300 bp): 5′-CCATCTGCCTTGCTTACCTC-3′ (forward) and 5′-AGCTTGGAGAGGTGGTGATG-3′ (reverse); TIP-30 (267 bp): 5′-AAATAAGGCCATCCTCCACC-3′ (forward) and 5′-CTGTTCTCCAAAAGCTGCCT-3′ (reverse); MEK5 (332 bp): 5′-GGATTTTACGGGGCATTTTT-3′ (forward) and 5′-TGGATCCCATACTGCTCTCC-3′ (reverse); and cyclophilin C (281 bp): 5′-TCCCAGTATTTGGGAAGTGC-3′ (forward) and 5′-TCACCCAAATGGAGATCACA-3′ (reverse). A set of primers specific to hypoxanthine-guanine phosphoribosyltransferase (HPRT): 5′-CCTGCTGGATTACATTAAAGCACTG-3′ (forward) and 5′-GTCAAGGGCATATCCAACAACAAAC-3′ (reverse), was used in each RNA sample to amplify HPRT cDNA as a reference under the same PCR conditions.

Results and discussion

Analysis of differential gene expression

Each hybridization to compare gene expression changes in initiated derivatives 03C and 09C with non-transformed progenitor keratinocyte 291 resulted in ~50% of the probe sets showing detectable expression of the represented genes and ESTs. Using the criteria described in the Materials and methods, the percentage of differentially expressed genes for one pair-wise comparison between each initiated clone and 291 was ~1.53% (172–199 out of a total of 12 488 probe sets, including internal controls). Because of the possibility that one of the triplicates might be an outlier, we focused on gene changes that occurred in at least two out of three sets. This resulted in a total of 160 differentially expressed genes in the initiated cell clones compared with their non-transformed progenitor. The chance of any given gene being selected in two or three of the triplicate data sets by random events is estimated to be about nine genes {12 488[(1.53%) 3 + 3(1.53%) 2 (1 – 1.53%)] = 8.7}, far fewer than the number of differentially expressed genes observed. The fact that the majority of the genes remained unaltered among 291, 03C and 09C is consistent with the clonal nature of the cell model. Among the total of 160 differentially expressed genes, 69 of the changes were specific for 03C and categorized as malignancy-related changes, 46 were specific for 09C and classified as papilloma-related, and 45 were shared by both 09C and 03C and categorized as initiation-related, with four of the latter displaying opposite expression changes in 03C and 09C. The known genes are listed in Table I . They have been placed in groups according to their known or probable functions as will be discussed below. A complete list of genes and their gene expression intensity values can be found at www.ohsu.edu/som-Dermatology/research/initiation_profile.html .

Independent validation of microarray results

Although the Affymetrix GeneChip ® provided a screen for differentially expressed genes based on consistent performance in the biological triplicates, the inherent variability due to limitations of the technologies ( 23 ) may obscure the true biological variations. Therefore, independent verification of microarray results was carried out by RT–PCR. Several low to moderately abundant genes were selected, as determined by quantitatively spiked standards—bacterial genes BioB, BioC and BioD and a phage gene Cre ( 28 ). We included genes with documented activities that made them theoretically likely to be causal agents in tumorigenesis in the absence of ras or p53 gene mutations. They included changes specific for 03C, such as TIMP-3 that was up-regulated 28-fold, SCF up-regulated 4.7-fold, MKP-1 up-regulated 3.8-fold and tat-interacting protein 30 (TIP-30) that was down-regulated 4.7-fold, and alterations shared by both 03C and 09C, including mitogen-activated protein kinase MEK5, that was elevated 3.5-fold in 03C and 4.1-fold in 09C cells, and cyclophilin C that was down-regulated 29-fold in 03C and 34-fold in 09C. Using primers designed in proximity to the 3′ end of the cDNA, each gene was amplified by PCR following RT, and amplicons were removed and quantified within the exponential phase of amplification as exemplified by HPRT. As shown in Figure 2 , the trend of the amplification intensity measured by the RT–PCR approach correlated with the findings by microarray analysis in that TIMP-3, SCF and MKP-1 were up-regulated in 03C, and MEK5 was up-regulated in 03C and 09C, while TIP-30 and cyclophilin C were down-regulated in 03C only or in both 03C and 09C, respectively. In the future, these and additional gene expression changes will be examined in independently derived mouse and human tumors to test their broader relevance. However, differentially expressed genes in our study have already been documented to be altered in expression in a variety of human cancers. These include MKP-1 ( 24 ), IGFBP2 ( 25 ), metallothionein 1 and 2 ( 26,27 ), Pcdh7 ( 28 ) and TIP30 ( 29 ). The current study supports these previous findings and further indicates that these changes may be early events in human cancer development.

Expression patterns in initiated cell precursors of benign or malignant tumors

The differentially expressed genes identified in both 03C and 09C cover a broad spectrum of cellular processes including cell proliferation, differentiation, cell–cell, cell–matrix interaction and apoptosis. The genes that are differentially expressed in 03C and 09C were categorized in an attempt to suggest genetic expression profiles altered in initiated cells that, in combination with genetic alterations at the malignant conversion stage, might lead to their distinctive malignant phenotypes.

Resistance to terminal differentiation

An important regulator of keratinocyte maturation in vivo is the intracellular and extracellular Ca 2+ concentration, which forms an increasing gradient from the proliferative basal cell compartment to the more superficial granular and cornified cell layers. An opposing gradient of vitamin A may also contribute to this process ( 30 ). Both human and murine keratinocytes are subject to regulation by calcium ion and vitamin A ( 31,32 ). In contrast to non-transformed keratinocytes, the initiated keratinocytes grow and can be selected on the basis of their ability to evade Ca 2+ -induced terminal differentiation. Accordingly, a number of terminal differential markers including small proline-rich proteins SPRR-1b, -2a and -2b were down-regulated in 09C. Other epidermal barrier function proteins were down-regulated in both 03C and 09C (SPRR2H) or over expressed in 03C only (SPRR1a). Members of the SPRR protein family show distinctive expression patterns ( 33 ) and are differentially regulated by a variety of factors in the process of keratinocyte terminal differentiation ( 34 ). The differentiation expression patterns of those terminal differentiation markers in the initiated cells may be a direct reflection of distinctive inactivation and/or activation of transcription factors/pathways in 03C versus 09C in the process of terminal differentiation. It is noted, however, that other markers of keratinocyte differentiation such as filaggrin remain elevated in 03C compared with its parental cell 291, indicating that the carcinoma precursor has undergone differentiation to some extent. However, the expression pattern of keratin 16 (up or sustained in 03C but down-regulated in 09C) revealed by microarray analysis indicated that 03C might undergo alternative differentiation, featuring hyperproliferative potential. In fact, over expression of filaggrin was reported in v-Ras transformed keratinocytes ( 35 ), and precocious appearance of filaggrin was noted in keratinocytes undergoing active proliferation ( 36 ). Therefore, under certain circumstances, expression of differentiation markers may not correlate to their differentiation status and proliferative potential due to multiple and complex regulation of terminal differentiation. Furthermore, some of the differentiation traits were subjected to further selection during tumor promotion, resulting in phenotypes preferential for aberrant growth and expansion. Indeed filaggrin is lost in 03R, the poorly differentiated SCC derived from 03C, and certain differentiation-associated keratins in 03R are absent or no longer Ca 2+ -dependent (unpublished data). Nonetheless, the expression patterns of differentiation markers at the initiation of carcinogenesis may be employed as indicators for their malignant potential.

In addition to differential expression of markers of keratinocyte differentiation in the initiated cells, there are distinctive expression patterns of genes involved in pathways that control keratinocyte differentiation. For example, down-regulation of cellular retinoic acid binding protein II (CRABPII) in both 03C and 09C indicated alterations in retinoic acid (RA)-mediated gene expression pathway at initiation of tumorigenesis. RA, the biologically active derivative of vitamin A, functions through binding to its nuclear receptors RAR and RXR and is subject to regulation by CRABPII ( 37 ). Up-regulation in 09C of 24p3, a member of the lipocalin family of small secreted polypeptides, also has the potential to regulate accessibility of RA as it was shown to be involved in RA transport ( 38 ). In addition, the visinin-like proteins may play a role in Ca 2+ /calmodulin-dependent signaling ( 39 ). The loss of such a protein in the papilloma lineage is a potential lead for explaining the defect in response to Ca 2+ -mediated terminal differentiation in initiated cells.

Acquisition of proliferative potential and by-pass of ras activation

The reduced dependency on exogenous growth factors in cancer cells indicates that tumor cells have acquired self-sufficiency for mitogenic stimuli. Expression patterns of malignant-prone versus papilloma-prone initiated cells in this study indicate that the growth signaling autonomy in tumorigenic cells may be acquired at the initiation of carcinogenesis. The pathways that may be involved in this process, based on differentially expressed genes identified by microarray, are represented schematically in Figure 3 . A number of growth factors or mitogenic pathway components were up-regulated in both 03C and 09C such as the proto-oncogenes c-Jun, c-fos and MEK-5, a MAPK kinase that selectively activates ERK5 ( 40 ). While the proliferation-enhancing activities of c-Jun and c-fos have been well documented, MEK5 poses an attractive alternative to the ras gene mutation observed in most models of mouse two-stage carcinogenesis. Increased activity of a downstream effector of the ras/raf pathway could bypass the requirement for ras protein activation in the mouse cell model or in human surface epithelial cancers, such as epidermal and head and neck cancers, most of which also lack ras mutation. MEK5/ERK5 can be activated by epidermal growth factor (EGF) and nerve growth factor in a ras-dependent manner ( 41 ). Activation of ERK5 by Ha-ras protein suggests that the MEK5/ERK5 pathway is an extension of the ras signaling pathway ( 42 ). Moreover, dominant negative mutants of MEK5 and its downstream effector ERK5 inhibit Raf-dependent cellular transformation. Since MEK5 does not activate ERK1/2 ( 43 ) MEK5/ERK5 may represent a novel pathway to mediate ras transformation via a mechanism independent of the classical ERK1/2 effector pathway (Figure 3 ). MEK5/ERK5 was shown to activate its downstream effector genes, such as Sap1a or myocyte enhancer factor 2C (MEF2C), a transcriptional factor known to activate c-Jun ( 40 ). c-Jun was shown to be up-regulated in the initiated cells. Thus, over expression of MEK5 in 03C and 09C could be an oncogenic mechanism to bypass the requirement for sustained ras activation and activate a spectrum of key transcriptional factors important for transformation. However, it is important to realize that the activities of MEK5 and many other proteins are regulated primarily through post-translational modifications. While mRNA expression changes reflect only one aspect of the dynamic alterations of these proteins in the initiated cells, they serve as potential leads for identifying causal events at the initiation of carcinogenesis.

In addition to the changes in mitogenic pathways shared by both 03C and 09C, there were distinctive patterns of expressed genes important for promoting cell proliferation. While growth arrest-specific 6 gene (GAS6), SCF and stromal cell-derived factor 1 (SDF1) were up-regulated only in 03C, epidermal growth factor receptor (EGFR) pathway substrate 8 (Eps8) and amphiregulin, a ligand for EGFR, were over expressed in 09C. These suggest that distinctive pathways that regulate growth potential of 03C and 09C have been activated at the initiation of tumorigenesis.

Survival under stress

Evasion of cell attrition is just as critical as acquisition of proliferative potential for tumor cell expansion. Mutations that allow initiated cell populations to expand in the face of anti-growth or apoptosis signals would confer a selective advantage over normal cells. Aberrant induction of survival factors that override the anti-growth and/or apoptotic signals may contribute to the ability of 03C and 09C to proliferate under conditions that induce terminal differentiation of normal keratinocytes. Bone morphogenetic factor 7 (BMP-7), which belongs to the TGF-β superfamily, may be one of the anti-growth factors ( 44 ) up-regulated in both initiated cell types in response to increased Ca 2+ . TIMP-3, one of the anti-apoptotic factors ( 45 ) found in extracellular matrix was up-regulated in 03C. TIMP-3 may inhibit or regulate important mediators of apoptosis such as enzymes involved in converting tethered TNF-α to its soluble form ( 46 ). TNF-α has been shown to mediate apoptosis via a number of pathways ( 47 ). In breast cancer cells, TNF-α-induced apoptosis was shown to be dependent upon caspase 6 ( 48 ) one of the downstream effectors of apoptosis. Consistent with this, caspase 6 was down-regulated in both initiated cell types in the current study. This indicates that certain apoptotic pathways, in particular those induced by TNF-α, might have been targeted in initiated cells at multiple levels (Figure 3 ). Besides these shared changes, however, there were additional changes in genes involved in survival/apoptosis in malignancy-associated 03C cells, including SCF and TIP30. SCF is a growth factor specifically required for the growth of mast cells, melanocytes and hematopoietic stem cells. SCF binds to c-Kit and activates Akt through phosphatidylinositol 3′ kinase (PI3-K). Activated Akt modulates and inhibits either pro-apoptotic proteins such as Bad ( 49 ) and caspase 9 ( 50 ) or transcriptional factors important for expression of pro-apoptotic genes such as Forkhead-like (FKHR)/Afx ( 51 ) (Figure 3 ). The loss of PTEN, a negative regulator of Akt, has been identified in a number of human cancers, indicating a critical role of Akt activation in carcinogenesis ( 52 ). The down-regulation of TIP30, a serine/threonine kinase shown to promote expression of pro-apoptotic genes such as Bad ( 29 ), also supports the concept that cell survival and homeostasis have been altered in the carcinoma-prone initiated cells. Taking these findings together, it is likely that additional disruption of apoptotic pathways occurred in 03C beyond those altered in common between 03C and 09C.

Cell–cell, cell–matrix interactions

The changes in extracellular composition and cell adhesion molecules affect a wide spectrum of cellular behaviors such as proliferation, differentiation, apoptosis, motility and invasive capabilities. T-cadherin is a cell surface protein found over expressed in tumor-derived endothelial cells from a variety of tumors ( 53 ). It was implicated in calcium-mediated cell–cell interaction and might also be involved in cell signaling since it was found to co-localize with Src family kinase. Meltrin alpha, also known as ADAM12 (a disintegrin and a metalloprotease domain), binds to cell surface syndecans and mediates beta1 integrin-dependent cell spreading ( 54 ). The over expression of T-cadherin and down-regulation of ADAM12 in both 03C and 09C underline the concept that there are some common routes controlling cell–cell, cell–matrix interactions in the initiated cells that contribute to tumorigenesis. However, the genes that are different between the initiated cell precursors of benign or malignant fates may be those that contribute to differences in malignant potential. For example, up-regulation of TIMP-3 in 03C may regulate a variety of matrix metalloproteinases. MMPs were shown to be important in cell invasion and metastasis. Transmembrane-4 superfamily protein (TM4SF; tetraspanin) was shown to be a linker to recruit protein kinase C to the plasma membrane to interact with a specific integrin ( 55 ). The over expression of TM4SF in 03C implicates the PKC-mediated signal pathway in the carcinoma-prone initiated cells. For 09C, the up-regulation of secretory leukoprotease inhibitor is consistent with the finding that expression level of skin-derived anti-leukoproteinase was correlated with the degree of differentiation: high in well-differentiated papilloma and low in poorly differentiated SSC ( 56 ).

Profiling of expression changes revealed dramatic differences between carcinoma-precursor 03C and papilloma-precursor 09C with regard to expression of genes involved in the acquisition of proliferative potential, evasion of anti-growth/apoptosis signals and cell–cell, cell–matrix interactions. We hypothesized that these changes at initiation may act in conjunction with changes at the malignant conversion to determine the tumorigenic fate of the initiated cells. In fact, the alterations at the initiation stage may predispose or contribute to the changes seen at the malignant progression. For example, while the p53 gene remains wild-type in the tumorigenic cells of our model, functional activities of wild-type p53 proteins are inactivated in 03R, the malignant cell derivative of 03C (unpublished data). In its precursor 03C, p53 proteins retain their sequence-specific DNA binding activity and ability to transactivate p53 downstream genes. However, the higher basal p53 steady-state protein levels and less dramatic induction upon DNA damage in initiated cells 03C than in progenitor 291 cells suggest that loss of wild-type p53 protein functions may be progressive and begin at the initiation stage. Up-regulated MKP-1 in 03C, shown by microarray analysis, provides a plausible lead for the aberrant behavior of p53 in 03C. MKP-1 was shown to preferentially inhibit JNK activity ( 57,58 ). While JNK mediates p53 protein degradation under non-stressful growth conditions ( 59 ), it stabilizes and activates p53 in response to stress ( 60 ). Over expression of MKP-1 in 03C may compromise the wild-type p53 activities by deregulating its turnover rate under both stress and non-stressful conditions (Figure 3 ), thereby conferring on initiated cells a susceptibility to genomic instability.

The microarray approach using clonal lineages of known tumor fates is cogent in its demonstration of altered expression of many known cancer-associated genes. However, its usefulness is primarily speculative in that it provides a starting point for testing new hypotheses about (i) patterns of coordinated and perhaps co-regulated gene expression changes, (ii) the early timing of changes in expression of cancer-associated genes, (iii) new pathways for initiation of cancer and (iv) changes that may selectively predict or instigate a benign or malignant process.

In summary, a mouse clonal cancer model combined with microarray technology permitted a global view of gene expression changes that reveal potentially novel pathways of initiation relevant to tumor fate. Distinctive patterns of differentially expressed genes in 03C and 09C indicate that there are likely distinctive, multiple pathways at the initiation stage, which in cooperation with oncogenic events at the malignant conversion stage such as additional p53 functional defects, may determine the fate of malignancy. The expression profiling of 03C and 09C provides an overall view of genes whose expression has changed and been selected due to DMBA treatment and resistance to Ca 2+ -induced terminal differentiation. Identification of regulatory genes/pathways that are responsible for the observed changes are likely to provide leads for causal events in tumorigenesis. One of the advantages of the clonal cellular model is that it provides host cells for introduction of candidate genes and testing of their causal role in initiation in vitro ( 21 ) and, by skin-grafting, tumorigenesis in vivo ( 14 ). These functional assays can provide a basis for prioritizing genes for targeting to epithelial cells in vivo in transgenic and knock-out mice, with the potential to provide new models for development of molecular targeted chemoprevention and therapy.

Table I.

Differentially expressed genes at the initiation of carcinogenesis

Cellular activity.Genes involved a.Accession no..O3C b.O9C b.a Only known genes are listed (ESTs are excluded). Genes are organized according to expression: up-regulation from greatest to least followed by changes shared between 03C and 09C initiated cells.

b Fold changes are the average of the two or three biological replicates selected according to the criteria described in the Materials and methods.

c NC (no change) represents unaltered expression level of the gene between the samples compared.

d Four genes showed opposite expression changes in 03C and 09C.

e Genes that have been verified by RT–PCR are represented by bold letters.Differentiation/barrier

SPRR1aAF05715610.80NC c

FilaggrinJ034589.55NC

Caspase-14AF0929973.50NC

Suppressor of cytokine signaling-2 (SOCS-2)U883272.73NC

Tumor necrosis factor, alpha-induced protein 2L24118−11.70NC

24p3X81627NC19.40

SPRR2bAJ005560NC−3.30

Brain-derived neurotrophic factorX55573NC−3.95

GATA-binding protein 3X55123NC−5.60

Keratin 16AF053235NC−12.85

SPRR2aAJ005559NC−20.93

SPRR1bX91825NC−24.07

Neural visinin-like Ca 2+ -binding protein type 1 (NVP-1) dD211652.95−3.93

Cellular retinoic acid binding protein IIM35523−2.43−2.57

SPRR2HAJ005566−3.03−8.45

SRY-box containing gene 2X94127−5.57−7.93

Osteoblast specific factor 2D13664−11.33−10.23

Proliferation

GAS6X5984612.55NC

Neuroblastoma myc-related oncogene 1M1273112.15NC

Sphingosine kinase (SPHK1a)AF0687488.40NC

Inhibitor of DNA binding 4 (Id4)AJ0019728.00NC

Stromal-derived factor 1, alphaL120297.80NC

Metallothionein 2K022366.40NC

Metallothionein 1V008355.13NC

Insulin-like growth factor binding protein 2X815805.05NC

Stem cell factor eM576474.77NC

Coagulation factor II (thrombin) receptor-like 1Z480433.10NC

Stromal-derived factor 1, betaL120302.65NC

Phenylalkylamine Ca 2+ antagonist (emopamil) binding proteinX97755−2.47NC

Wee1 kinaseD30743−3.25NC

Interferon regulatory factor 1M21065−4.17NC

TRA1D78354−4.95NC

Eps8L21671NC7.20

cdc2/CDC28-like protein kinase 3 (Clk3)AF033565NC3.15

Dlxin-1AB029448NC−2.97

Max interacting protein 1L38822NC−3.63

Inhibitor of DNA binding 2 (Id2)AF077861NC−10.33

Bone morphogenetic protein 7X5690617.9530.25

Protein tyrosine phosphatase, non-receptor type substrate 1(BIT)AB01819410.8511.30

Jun oncogeneX1276110.835.35

c-fosV007273.853.13

MEK5 eAB0193743.504.15

Parathyroid hormone-like peptideM60057−3.05−2.53

Amphiregulin dL41352−4.252.75

Apoptosis

TIMP-3 eU2643728.07NC

Interleukin 1 receptor, type IIX597693.30NC

B-cell leukemia/lymphoma 6U414653.23NC

TIP30 eAF061972−4.77NC

MAST205U02313NC2.75

Serine protease inhibitor 6U96700−4.70−7.90

Caspase 6Y13087−7.73−9.33

Cell–cell, cell–matrix interaction

Tetraspan TM4SFAF0534549.57NC

Cysteine-rich glycoprotein SPARCX040177.43NC

Ankyrin 3, epithelialL406315.55NC

Procollagen, type IV, alpha 5Z351684.00NC

Pcdh7AB006758−2.77NC

Ena-vasodilator stimulated phosphoproteinU72519−4.77NC

Secretory leukoprotease inhibitorAF002719NC6.50

Integral membrane protein 2 BU76253NC−3.00

Neural cell adhesion moleculeX15052NC−4.30

Fibronectin (FN)M18194NC−5.30

Desmocollin 3Y11169NC−9.10

T-cadherinAB02210014.8511.50

Glypican 4X835773.402.55

Fibulin 2X75285−2.53−8.67

Minopontin dX13986−2.803.60

Transforming growth factor, beta induced, 68 kDaL19932−2.80−3.05

ADAM12D50411−7.15−4.95

Miscellaneous

IL-18D4994912.55NC

Homeobox D8X565614.75NC

Mouse A-X actinJ041814.70NC

MKP-1 eX619403.87NC

Hexosaminidase AU058373.00NC

Downstream of tyrosine kinase 1U788182.55NC

Mouse complement component C3K02782−2.43NC

FollistatinZ29532−2.47NC

Heterogeneous nuclear ribonucleoprotein G, splice variant 1AJ237847−2.57NC

Interleukin 1 alphaM14639−3.55NC

5–3 exonucleaseX91617−5.60NC

FXYD domain-containing ion transport regulator 5U72680−5.87NC

Mouse protein tyrosine phosphatase (70zpep)M90388NC5.50

Putative glycogen storage disease type 1bAF080469NC4.40

Glutamate-cysteine ligase, modifier subunit(Glclr)U95053NC3.15

Gycerol kinaseU48403NC2.90

Semaphorin EX85994NC−4.00

Mouse glutathione S -transferase class mu (GST5-5)J04696NC−5.00

Mouse calcium-dependent phospholipid binding proteinM72394NC−5.10

Adenylate cyclase 7U12919NC−5.10

Lysosomal acid lipaseZ31689NC−5.70

DNA ligase I, ATP-dependentU19604NC−7.20

SOX11AF009414NC−7.47

Schwannoma-associated protein (SAM9)AF026124NC−7.60

Cathepsin CU74683NC−8.75

Stearoyl-coenzyme A desaturase 1M21285NC−12.35

Glucocortoid-regulated inflammatory prostaglandin G/H synthase (griPGHS)M88242NC−14.17

Aldehyde dehydrogenase 3 (aldh3)AF0330347.2521.85

Cathepsin HU061193.154.20

Alpha-2,3-sialyltransferaseD289412.652.60

17-beta-hydroxysteroid dehydrogenase type 7Y15733−2.23−4.23

LatexinD88769−2.95−3.20

Mouse germline interleukin 1 receptor antagonist (1L-1rn) gene dL32838−4.555.65

Chemokine orphan receptor 1AF000236−5.60−8.57

Protease (prosome, macropain) 28 subunitAB007136−6.33−6.53

Cyclophilin C eM74227−29.03−33.97

b Fold changes are the average of the two or three biological replicates selected according to the criteria described in the Materials and methods.

c NC (no change) represents unaltered expression level of the gene between the samples compared.

d Four genes showed opposite expression changes in 03C and 09C.

e Genes that have been verified by RT–PCR are represented by bold letters.Differentiation/barrier

SPRR1aAF05715610.80NC c

FilaggrinJ034589.55NC

Caspase-14AF0929973.50NC

Suppressor of cytokine signaling-2 (SOCS-2)U883272.73NC

Tumor necrosis factor, alpha-induced protein 2L24118−11.70NC

24p3X81627NC19.40

SPRR2bAJ005560NC−3.30

Brain-derived neurotrophic factorX55573NC−3.95

GATA-binding protein 3X55123NC−5.60

Keratin 16AF053235NC−12.85

SPRR2aAJ005559NC−20.93

SPRR1bX91825NC−24.07

Neural visinin-like Ca 2+ -binding protein type 1 (NVP-1) dD211652.95−3.93

Cellular retinoic acid binding protein IIM35523−2.43−2.57

SPRR2HAJ005566−3.03−8.45

SRY-box containing gene 2X94127−5.57−7.93

Osteoblast specific factor 2D13664−11.33−10.23

Proliferation

GAS6X5984612.55NC

Neuroblastoma myc-related oncogene 1M1273112.15NC

Sphingosine kinase (SPHK1a)AF0687488.40NC

Inhibitor of DNA binding 4 (Id4)AJ0019728.00NC

Stromal-derived factor 1, alphaL120297.80NC

Metallothionein 2K022366.40NC

Metallothionein 1V008355.13NC

Insulin-like growth factor binding protein 2X815805.05NC

Stem cell factor eM576474.77NC

Coagulation factor II (thrombin) receptor-like 1Z480433.10NC

Stromal-derived factor 1, betaL120302.65NC

Phenylalkylamine Ca 2+ antagonist (emopamil) binding proteinX97755−2.47NC

Wee1 kinaseD30743−3.25NC

Interferon regulatory factor 1M21065−4.17NC

TRA1D78354−4.95NC

Eps8L21671NC7.20

cdc2/CDC28-like protein kinase 3 (Clk3)AF033565NC3.15

Dlxin-1AB029448NC−2.97

Max interacting protein 1L38822NC−3.63

Inhibitor of DNA binding 2 (Id2)AF077861NC−10.33

Bone morphogenetic protein 7X5690617.9530.25

Protein tyrosine phosphatase, non-receptor type substrate 1(BIT)AB01819410.8511.30

Jun oncogeneX1276110.835.35

c-fosV007273.853.13

MEK5 eAB0193743.504.15

Parathyroid hormone-like peptideM60057−3.05−2.53

Amphiregulin dL41352−4.252.75

Apoptosis

TIMP-3 eU2643728.07NC

Interleukin 1 receptor, type IIX597693.30NC

B-cell leukemia/lymphoma 6U414653.23NC

TIP30 eAF061972−4.77NC

MAST205U02313NC2.75

Serine protease inhibitor 6U96700−4.70−7.90

Caspase 6Y13087−7.73−9.33

Cell–cell, cell–matrix interaction

Tetraspan TM4SFAF0534549.57NC

Cysteine-rich glycoprotein SPARCX040177.43NC

Ankyrin 3, epithelialL406315.55NC

Procollagen, type IV, alpha 5Z351684.00NC

Pcdh7AB006758−2.77NC

Ena-vasodilator stimulated phosphoproteinU72519−4.77NC

Secretory leukoprotease inhibitorAF002719NC6.50

Integral membrane protein 2 BU76253NC−3.00

Neural cell adhesion moleculeX15052NC−4.30

Fibronectin (FN)M18194NC−5.30

Desmocollin 3Y11169NC−9.10

T-cadherinAB02210014.8511.50

Glypican 4X835773.402.55

Fibulin 2X75285−2.53−8.67

Minopontin dX13986−2.803.60

Transforming growth factor, beta induced, 68 kDaL19932−2.80−3.05

ADAM12D50411−7.15−4.95

Miscellaneous

IL-18D4994912.55NC

Homeobox D8X565614.75NC

Mouse A-X actinJ041814.70NC

MKP-1 eX619403.87NC

Hexosaminidase AU058373.00NC

Downstream of tyrosine kinase 1U788182.55NC

Mouse complement component C3K02782−2.43NC

FollistatinZ29532−2.47NC

Heterogeneous nuclear ribonucleoprotein G, splice variant 1AJ237847−2.57NC

Interleukin 1 alphaM14639−3.55NC

5–3 exonucleaseX91617−5.60NC

FXYD domain-containing ion transport regulator 5U72680−5.87NC

Mouse protein tyrosine phosphatase (70zpep)M90388NC5.50

Putative glycogen storage disease type 1bAF080469NC4.40

Glutamate-cysteine ligase, modifier subunit(Glclr)U95053NC3.15

Gycerol kinaseU48403NC2.90

Semaphorin EX85994NC−4.00

Mouse glutathione S -transferase class mu (GST5-5)J04696NC−5.00

Mouse calcium-dependent phospholipid binding proteinM72394NC−5.10

Adenylate cyclase 7U12919NC−5.10

Lysosomal acid lipaseZ31689NC−5.70

DNA ligase I, ATP-dependentU19604NC−7.20

SOX11AF009414NC−7.47

Schwannoma-associated protein (SAM9)AF026124NC−7.60

Cathepsin CU74683NC−8.75

Stearoyl-coenzyme A desaturase 1M21285NC−12.35

Glucocortoid-regulated inflammatory prostaglandin G/H synthase (griPGHS)M88242NC−14.17

Aldehyde dehydrogenase 3 (aldh3)AF0330347.2521.85

Cathepsin HU061193.154.20

Alpha-2,3-sialyltransferaseD289412.652.60

17-beta-hydroxysteroid dehydrogenase type 7Y15733−2.23−4.23

LatexinD88769−2.95−3.20

Mouse germline interleukin 1 receptor antagonist (1L-1rn) gene dL32838−4.555.65

Chemokine orphan receptor 1AF000236−5.60−8.57

Protease (prosome, macropain) 28 subunitAB007136−6.33−6.53

Cyclophilin C eM74227−29.03−33.97

Table I.

Differentially expressed genes at the initiation of carcinogenesis

b Fold changes are the average of the two or three biological replicates selected according to the criteria described in the Materials and methods.

c NC (no change) represents unaltered expression level of the gene between the samples compared.

d Four genes showed opposite expression changes in 03C and 09C.

e Genes that have been verified by RT–PCR are represented by bold letters.Differentiation/barrier

SPRR1aAF05715610.80NC c

FilaggrinJ034589.55NC

Caspase-14AF0929973.50NC

Suppressor of cytokine signaling-2 (SOCS-2)U883272.73NC

Tumor necrosis factor, alpha-induced protein 2L24118−11.70NC

24p3X81627NC19.40

SPRR2bAJ005560NC−3.30

Brain-derived neurotrophic factorX55573NC−3.95

GATA-binding protein 3X55123NC−5.60

Keratin 16AF053235NC−12.85

SPRR2aAJ005559NC−20.93

SPRR1bX91825NC−24.07

Neural visinin-like Ca 2+ -binding protein type 1 (NVP-1) dD211652.95−3.93

Cellular retinoic acid binding protein IIM35523−2.43−2.57

SPRR2HAJ005566−3.03−8.45

SRY-box containing gene 2X94127−5.57−7.93

Osteoblast specific factor 2D13664−11.33−10.23

Proliferation

GAS6X5984612.55NC

Neuroblastoma myc-related oncogene 1M1273112.15NC

Sphingosine kinase (SPHK1a)AF0687488.40NC

Inhibitor of DNA binding 4 (Id4)AJ0019728.00NC

Stromal-derived factor 1, alphaL120297.80NC

Metallothionein 2K022366.40NC

Metallothionein 1V008355.13NC

Insulin-like growth factor binding protein 2X815805.05NC

Stem cell factor eM576474.77NC

Coagulation factor II (thrombin) receptor-like 1Z480433.10NC

Stromal-derived factor 1, betaL120302.65NC

Phenylalkylamine Ca 2+ antagonist (emopamil) binding proteinX97755−2.47NC

Wee1 kinaseD30743−3.25NC

Interferon regulatory factor 1M21065−4.17NC

TRA1D78354−4.95NC

Eps8L21671NC7.20

cdc2/CDC28-like protein kinase 3 (Clk3)AF033565NC3.15

Dlxin-1AB029448NC−2.97

Max interacting protein 1L38822NC−3.63

Inhibitor of DNA binding 2 (Id2)AF077861NC−10.33

Bone morphogenetic protein 7X5690617.9530.25

Protein tyrosine phosphatase, non-receptor type substrate 1(BIT)AB01819410.8511.30

Jun oncogeneX1276110.835.35

c-fosV007273.853.13

MEK5 eAB0193743.504.15

Parathyroid hormone-like peptideM60057−3.05−2.53

Amphiregulin dL41352−4.252.75

Apoptosis

TIMP-3 eU2643728.07NC

Interleukin 1 receptor, type IIX597693.30NC

B-cell leukemia/lymphoma 6U414653.23NC

TIP30 eAF061972−4.77NC

MAST205U02313NC2.75

Serine protease inhibitor 6U96700−4.70−7.90

Caspase 6Y13087−7.73−9.33

Cell–cell, cell–matrix interaction

Tetraspan TM4SFAF0534549.57NC

Cysteine-rich glycoprotein SPARCX040177.43NC

Ankyrin 3, epithelialL406315.55NC

Procollagen, type IV, alpha 5Z351684.00NC

Pcdh7AB006758−2.77NC

Ena-vasodilator stimulated phosphoproteinU72519−4.77NC

Secretory leukoprotease inhibitorAF002719NC6.50

Integral membrane protein 2 BU76253NC−3.00

Neural cell adhesion moleculeX15052NC−4.30

Fibronectin (FN)M18194NC−5.30

Desmocollin 3Y11169NC−9.10

T-cadherinAB02210014.8511.50

Glypican 4X835773.402.55

Fibulin 2X75285−2.53−8.67

Minopontin dX13986−2.803.60

Transforming growth factor, beta induced, 68 kDaL19932−2.80−3.05

ADAM12D50411−7.15−4.95

Miscellaneous

IL-18D4994912.55NC

Homeobox D8X565614.75NC

Mouse A-X actinJ041814.70NC

MKP-1 eX619403.87NC

Hexosaminidase AU058373.00NC

Downstream of tyrosine kinase 1U788182.55NC

Mouse complement component C3K02782−2.43NC

FollistatinZ29532−2.47NC

Heterogeneous nuclear ribonucleoprotein G, splice variant 1AJ237847−2.57NC

Interleukin 1 alphaM14639−3.55NC

5–3 exonucleaseX91617−5.60NC

FXYD domain-containing ion transport regulator 5U72680−5.87NC

Mouse protein tyrosine phosphatase (70zpep)M90388NC5.50

Putative glycogen storage disease type 1bAF080469NC4.40

Glutamate-cysteine ligase, modifier subunit(Glclr)U95053NC3.15

Gycerol kinaseU48403NC2.90

Semaphorin EX85994NC−4.00

Mouse glutathione S -transferase class mu (GST5-5)J04696NC−5.00

Mouse calcium-dependent phospholipid binding proteinM72394NC−5.10

Adenylate cyclase 7U12919NC−5.10

Lysosomal acid lipaseZ31689NC−5.70

DNA ligase I, ATP-dependentU19604NC−7.20

SOX11AF009414NC−7.47

Schwannoma-associated protein (SAM9)AF026124NC−7.60

Cathepsin CU74683NC−8.75

Stearoyl-coenzyme A desaturase 1M21285NC−12.35

Glucocortoid-regulated inflammatory prostaglandin G/H synthase (griPGHS)M88242NC−14.17

Aldehyde dehydrogenase 3 (aldh3)AF0330347.2521.85

Cathepsin HU061193.154.20

Alpha-2,3-sialyltransferaseD289412.652.60

17-beta-hydroxysteroid dehydrogenase type 7Y15733−2.23−4.23

LatexinD88769−2.95−3.20

Mouse germline interleukin 1 receptor antagonist (1L-1rn) gene dL32838−4.555.65

Chemokine orphan receptor 1AF000236−5.60−8.57

Protease (prosome, macropain) 28 subunitAB007136−6.33−6.53

Cyclophilin C eM74227−29.03−33.97

b Fold changes are the average of the two or three biological replicates selected according to the criteria described in the Materials and methods.

c NC (no change) represents unaltered expression level of the gene between the samples compared.

d Four genes showed opposite expression changes in 03C and 09C.

e Genes that have been verified by RT–PCR are represented by bold letters.Differentiation/barrier

SPRR1aAF05715610.80NC c

FilaggrinJ034589.55NC

Caspase-14AF0929973.50NC

Suppressor of cytokine signaling-2 (SOCS-2)U883272.73NC

Tumor necrosis factor, alpha-induced protein 2L24118−11.70NC

24p3X81627NC19.40

SPRR2bAJ005560NC−3.30

Brain-derived neurotrophic factorX55573NC−3.95

GATA-binding protein 3X55123NC−5.60

Keratin 16AF053235NC−12.85

SPRR2aAJ005559NC−20.93

SPRR1bX91825NC−24.07

Neural visinin-like Ca 2+ -binding protein type 1 (NVP-1) dD211652.95−3.93

Cellular retinoic acid binding protein IIM35523−2.43−2.57

SPRR2HAJ005566−3.03−8.45

SRY-box containing gene 2X94127−5.57−7.93

Osteoblast specific factor 2D13664−11.33−10.23

Proliferation

GAS6X5984612.55NC

Neuroblastoma myc-related oncogene 1M1273112.15NC

Sphingosine kinase (SPHK1a)AF0687488.40NC

Inhibitor of DNA binding 4 (Id4)AJ0019728.00NC

Stromal-derived factor 1, alphaL120297.80NC

Metallothionein 2K022366.40NC

Metallothionein 1V008355.13NC

Insulin-like growth factor binding protein 2X815805.05NC

Stem cell factor eM576474.77NC

Coagulation factor II (thrombin) receptor-like 1Z480433.10NC

Stromal-derived factor 1, betaL120302.65NC

Phenylalkylamine Ca 2+ antagonist (emopamil) binding proteinX97755−2.47NC

Wee1 kinaseD30743−3.25NC

Interferon regulatory factor 1M21065−4.17NC

TRA1D78354−4.95NC

Eps8L21671NC7.20

cdc2/CDC28-like protein kinase 3 (Clk3)AF033565NC3.15

Dlxin-1AB029448NC−2.97

Max interacting protein 1L38822NC−3.63

Inhibitor of DNA binding 2 (Id2)AF077861NC−10.33

Bone morphogenetic protein 7X5690617.9530.25

Protein tyrosine phosphatase, non-receptor type substrate 1(BIT)AB01819410.8511.30

Jun oncogeneX1276110.835.35

c-fosV007273.853.13

MEK5 eAB0193743.504.15

Parathyroid hormone-like peptideM60057−3.05−2.53

Amphiregulin dL41352−4.252.75

Apoptosis

TIMP-3 eU2643728.07NC

Interleukin 1 receptor, type IIX597693.30NC

B-cell leukemia/lymphoma 6U414653.23NC

TIP30 eAF061972−4.77NC

MAST205U02313NC2.75

Serine protease inhibitor 6U96700−4.70−7.90

Caspase 6Y13087−7.73−9.33

Cell–cell, cell–matrix interaction

Tetraspan TM4SFAF0534549.57NC

Cysteine-rich glycoprotein SPARCX040177.43NC

Ankyrin 3, epithelialL406315.55NC

Procollagen, type IV, alpha 5Z351684.00NC

Pcdh7AB006758−2.77NC

Ena-vasodilator stimulated phosphoproteinU72519−4.77NC

Secretory leukoprotease inhibitorAF002719NC6.50

Integral membrane protein 2 BU76253NC−3.00

Neural cell adhesion moleculeX15052NC−4.30

Fibronectin (FN)M18194NC−5.30

Desmocollin 3Y11169NC−9.10

T-cadherinAB02210014.8511.50

Glypican 4X835773.402.55

Fibulin 2X75285−2.53−8.67

Minopontin dX13986−2.803.60

Transforming growth factor, beta induced, 68 kDaL19932−2.80−3.05

ADAM12D50411−7.15−4.95

Miscellaneous

IL-18D4994912.55NC

Homeobox D8X565614.75NC

Mouse A-X actinJ041814.70NC

MKP-1 eX619403.87NC

Hexosaminidase AU058373.00NC

Downstream of tyrosine kinase 1U788182.55NC

Mouse complement component C3K02782−2.43NC

FollistatinZ29532−2.47NC

Heterogeneous nuclear ribonucleoprotein G, splice variant 1AJ237847−2.57NC

Interleukin 1 alphaM14639−3.55NC

5–3 exonucleaseX91617−5.60NC

FXYD domain-containing ion transport regulator 5U72680−5.87NC

Mouse protein tyrosine phosphatase (70zpep)M90388NC5.50

Putative glycogen storage disease type 1bAF080469NC4.40

Glutamate-cysteine ligase, modifier subunit(Glclr)U95053NC3.15

Gycerol kinaseU48403NC2.90

Semaphorin EX85994NC−4.00

Mouse glutathione S -transferase class mu (GST5-5)J04696NC−5.00

Mouse calcium-dependent phospholipid binding proteinM72394NC−5.10

Adenylate cyclase 7U12919NC−5.10

Lysosomal acid lipaseZ31689NC−5.70

DNA ligase I, ATP-dependentU19604NC−7.20

SOX11AF009414NC−7.47

Schwannoma-associated protein (SAM9)AF026124NC−7.60

Cathepsin CU74683NC−8.75

Stearoyl-coenzyme A desaturase 1M21285NC−12.35

Glucocortoid-regulated inflammatory prostaglandin G/H synthase (griPGHS)M88242NC−14.17

Aldehyde dehydrogenase 3 (aldh3)AF0330347.2521.85

Cathepsin HU061193.154.20

Alpha-2,3-sialyltransferaseD289412.652.60

17-beta-hydroxysteroid dehydrogenase type 7Y15733−2.23−4.23

LatexinD88769−2.95−3.20

Mouse germline interleukin 1 receptor antagonist (1L-1rn) gene dL32838−4.555.65

Chemokine orphan receptor 1AF000236−5.60−8.57

Protease (prosome, macropain) 28 subunitAB007136−6.33−6.53

Cyclophilin C eM74227−29.03−33.97

Fig. 1.

Clonal cell lineage model of keratinocyte carcinogenesis. Non-transformed mouse epidermal cell clone 291 was treated with DMBA. Initiated cell clones 03C and 09C were selected based on altered terminal differentiation response to extracellular Ca 2+ . The initiated cells gave rise to reproducible lineage-dependent tumor phenotypes upon transplantation to athymic nu/nu mice [14].

Fig. 1.

Fig. 2.

Confirmation by semi-quantitative RT–PCR ( A ) of microarray results ( B ). TIMP-3 ( 27 ), SCF ( 30 ), MKP-1 ( 27 ), TIP30 ( 30 ), MEK5 ( 33 ), cyclophilin C ( 27 ) and HPRT (number of cycles in parentheses) were amplified and products were separated in a 1.5% agarose gel and stained with ethidium bromide. HPRT was used as the control for equivalent RNA template among the three cell samples in the PCR reactions. The fold change of microarray analysis was calculated from the mean of biological replicates of 03C and/or 09C.

Fig. 2.

Fig. 3.

Model of potential pathway alterations at the initiation of carcinogenesis. Differentially expressed genes in malignant precursor 03C are shown in bold with the changes shared by both 03C and 09C underlined.

Fig. 3.

To whom correspondence should be addressed Email: kuleszma@ohsu.edu

We are grateful to Dr Melissa Wong of the OHSU Department of Dermatology and Dr Chris Harrington, Yi-Ching Hsieh and Edwin Quick of the OHSU Cancer Institute Gene Microarray/Affymetrix Core facility for comments on data analysis and the manuscript. We also thank Michelle Bryant, Jodi Johnson and Salwa El-Hizawi for assistance with preparation of the manuscript. This research was supported by the Department of Dermatology at OHSU, NIH R01 CA31101 and the OHSU Cancer Institute CCSG 5P30 CA69533.

References

Sloan,J.B. and Jaworsky,C. (

1993

) Clinical misdiagnosis of squamous cell carcinoma

in situ

as seborrheic keratosis. A prospective study.

J. Dermatol. Surg. Oncol.

413

–416.

Gordon,M.L. and Hecker,M.S. (

1967

) Care of the skin at midlife: diagnosis of pigmented lesions.

Geriatrics

–58.

Connors,R.C. and Ackerman,A.B. (

1976

) Histologic pseudomalignancies of the skin.

Arch. Dermatol.

112

1767

–1780.

Salasche,S.J. (

2000

) Epidemiology of actinic keratoses and squamous cell carcinoma.

J. Am. Acad. Dermatol.

–7.

Epstein,E., Epstein,N.N., Bragg,K. and Linden,G. (

1968

) Metastases from squamous cell carcinomas of the skin.

Arch. Dermatol.

245

–251.

Cockerell,C.J. (

2000

) Histopathology of incipient intraepidermal squamous cell carcinoma (`actinic keratosis').

J. Am. Acad. Dermatol.

–17.

Lingen,M.W., Emami,B. and Clark,J.I. (

2000

) New therapeutic strategies for the treatment and prevention of head and neck cancer.

Exp. Opin. Invest. Drugs

2855

–2872.

Brown,K., Quintanilla,M., Ramsden,M., Kerr,I.B., Young,S. and Balmain,A. (

1986

) v-ras genes from Harvey and BALB murine sarcoma viruses can act as initiators of two-stage mouse skin carcinogenesis.

Cell

447

–456.

Nishigori,C., Tanaka,M., Moriwaki,S., Imamura,S. and Takebe,H. (

1992

) Accelerated appearance of skin tumors in hairless mice by repeated UV irradiation with initial intense exposure and characterization of the tumors.

Jpn. J. Cancer Res.

1172

–1178.

van Kranen,H.J. and de Gruijl,F.R. (

1999

) Mutations in cancer genes of UV-induced skin tumors of hairless mice.

J. Epidemiol.

S58

–S65.

Nishigori,C., Yarosh,D.B., Ullrich,S.E., Vink,A.A., Bucana,C.D., Roza,L. and Kripke,M.L. (

1996

) Evidence that DNA damage triggers interleukin 10 cytokine production in UV-irradiated murine keratinocytes.

Proc. Natl Acad. Sci. USA

10354

–10359.

Kulesz-Martin,M.F., Blumenson,L.E., Manly,K.F., Siracky,J. and East,C.J. (

1991

) Tumor progression of murine epidermal cells after treatment

in vitro

with 12-

-tetradecanoylphorbol-13-acetate or retinoic acid.

Cancer Res.

4701

–4706.

Berns,A. (

2001

) Cancer: improved mouse models.

Nature

410

1043

–1044.

Kulesz-Martin,M.F., Penetrante,R. and East,C.J. (

1988

) Benign and malignant tumor stages in a mouse keratinocyte line treated with 7,12-dimethylbenz[

]anthracene

in vitro

Carcinogenesis

171

–174.

Han,K.A., Rothberg,P. and Kulesz-Martin,M. (

1990

) Altered levels of endogenous retrovirus-like sequence (VL30) RNA during mouse epidermal cell carcinogenesis.

Mol. Carcinogen.

–82.

Schneider,B.L., Bowden,G.T., Sutter,C., Schweizer,J., Han,K.A. and Kulesz-Martin,M.F. (

1993

) 7,12-Dimethylbenz[

]anthracene-induced mouse keratinocyte malignant transformation independent of Harvey ras activation.

J. Invest. Dermatol.

101

595

–599.

Liu,Y. and Kulesz-Martin,M.F. (

1998

) Altered gene expression in a clonal epidermal cell model of carcinogenesis identified by RNA differential display.

Carcinogenesis

683

–686.

Velculescu,V.E., Zhang,L., Vogelstein,B. and Kinzler,K.W. (

1995

) Serial analysis of gene expression.

Science

270

484

–487.

Lipshutz,R.J., Fodor,S.P., Gingeras,T.R. and Lockhart,D.J. (

1999

) High density synthetic oligonucleotide arrays.

Nature Genet.

–24.

Brown,P.O. and Botstein,D. (

1999

) Exploring the new world of the genome with DNA microarrays.

Nature Genet.

–37.

Kulesz-Martin,M., Kilkenny,A.E., Holbrook,K.A., Digernes,V. and Yuspa,S.H. (

1983

) Properties of carcinogen altered mouse epidermal cells resistant to calcium-induced terminal differentiation.

Carcinogenesis

1367

–1377.

Lockhart,D.J., Dong,H., Byrne,M.C.

et al

. (

1996

) Expression monitoring by hybridization to high-density oligonucleotide arrays.

Nature Biotechnol.

1675

–1680.

Schadt,E.E., Li,C., Su,C. and Wong,W.H. (

2000

) Analyzing high-density oligonucleotide gene expression array data.

J. Cell. Biochem.

192

–202.

Loda,M., Capodieci,P., Mishra,R., Yao,H., Corless,C., Grigioni,W., Wang,Y., Magi-Galluzzi,C. and Stork,P.J. (

1996

) Expression of mitogen-activated protein kinase phosphatase-1 in the early phases of human epithelial carcinogenesis.

Am. J. Pathol.

149

1553

–1564.

Sallinen,S.L., Sallinen,P.K., Haapasalo,H.K., Helin,H.J., Helen,P.T., Schraml,P., Kallioniemi,O.P. and Kononen,J. (

2000

) Identification of differentially expressed genes in human gliomas by DNA microarray and tissue chip techniques.

Cancer Res.

6617

–6622.

Tsangaris,G.T., Vamvoukakis,J., Politis,I., Kattamis,A.C. and Tzortzatou-Stathopoulou,F. (

2000

) Metallothionein expression prevents apoptosis. II: Evaluation of the role of metallothionein expression on the chemotherapy-induced apoptosis during the treatment of acute leukemia.

Anticancer Res.

4407

–4411.

Jayasurya,A., Bay,B.H., Yap,W.M. and Tan,N.G. (

2000

) Correlation of metallothionein expression with apoptosis in nasopharyngeal carcinoma.

Br. J. Cancer

1198

–1203.

Yoshida,K., Yoshitomo-Nakagawa,K., Seki,N., Sasaki,M. and Sugano,S. (

1998

) Cloning, expression analysis, and chromosomal localization of BH-protocadherin (PCDH7), a novel member of the cadherin superfamily.

Genomics

458

–461.

Xiao,H., Palhan,V., Yang,Y. and Roeder,R.G. (

2000

) TIP30 has an intrinsic kinase activity required for up-regulation of a subset of apoptotic genes.

EMBO J.

956

–963.

Yuspa,S.H. (

1998

) The pathogenesis of squamous cell cancer: lessons learned from studies of skin carcinogenesis.

J. Dermatol. Sci.

–7.

Tu,C.L., Oda,Y. and Bikle,D.D. (

1999

) Effects of a calcium receptor activator on the cellular response to calcium in human keratinocytes.

J. Invest. Dermatol.

113

340

–345.

Fisher,G.J. and Voorhees,J.J. (

1996

) Molecular mechanisms of retinoid actions in skin.

FASEB J.

1002

–1013.

Hohl,D., de Viragh,P.A., Amiguet-Barras,F., Gibbs,S., Backendorf,C. and Huber,M. (

1995

) The small proline-rich proteins constitute a multigene family of differentially regulated cornified cell envelope precursor proteins.

J. Invest. Dermatol.

104

902

–909.

Cabral,A., Voskamp,P., Cleton-Jansen,A.M., South,A., Nizetic,D. and Backendorf,C. (

2001

) Structural organization and regulation of the small proline-rich family of cornified envelope precursors suggest a role in adaptive barrier function.

J. Biol. Chem.

276

19231

–19237.

Dlugosz,A.A., Cheng,C., Williams,E.K., Dharia,A.G., Denning,M.F. and Yuspa,S.H. (

1994

) Alterations in murine keratinocyte differentiation induced by activated rasHa genes are mediated by protein kinase C-alpha.

Cancer Res.

6413

–6420.

Lutzow-Holm,C., Heyden,A., Huitfeldt,H.S., Brandtzaeg,P. and Clausen,O.P. (

1995

) Topical application of calcitriol alters expression of filaggrin but not keratin K1 in mouse epidermis.

Arch. Dermatol. Res.

287

480

–487.

Delva,L., Bastie,J.N., Rochette-Egly,C., Kraiba,R., Balitrand,N., Despouy,G., Chambon,P. and Chomienne,C. (

1999

) Physical and functional interactions between cellular retinoic acid binding protein II and the retinoic acid-dependent nuclear complex.

Mol. Cell. Biol.

7158

–7167.

Chu,S.T., Lin,H.J., Huang,H.L. and Chen,Y.H. (

1998

) The hydrophobic pocket of 24p3 protein from mouse uterine luminal fluid: fatty acid and retinol binding activity and predicted structural similarity to lipocalins.

J. Pept. Res.

390

–397.

Sathyanarayanan,P.V., Cremo,C.R. and Poovaiah,B.W. (

2000

) Plant chimeric Ca

/Calmodulin-dependent protein kinase. Role of the neural visinin-like domain in regulating autophosphorylation and calmodulin affinity.

J. Biol. Chem.

275

30417

–30422.

Kato,Y., Kravchenko,V.V., Tapping,R.I., Han,J., Ulevitch,R.J. and Lee,J.D. (

1997

) BMK1/ERK5 regulates serum-induced early gene expression through transcription factor MEF2C.

EMBO J.

7054

–7066.

Kamakura,S., Moriguchi,T. and Nishida,E. (

1999

) Activation of the protein kinase ERK5/BMK1 by receptor tyrosine kinases. Identification and characterization of a signaling pathway to the nucleus.

J. Biol. Chem.

274

26563

–26571.

English,J.M., Pearson,G., Baer,R. and Cobb,M.H. (

1998

) Identification of substrates and regulators of the mitogen-activated protein kinase ERK5 using chimeric protein kinases.

J. Biol. Chem.

273

3854

–3860.

Zhou,G., Bao,Z.Q. and Dixon,J.E. (

1995

) Components of a new human protein kinase signal transduction pathway.

J. Biol. Chem.

270

12665

–12669.

Andrews,P.W., Damjanov,I., Berends,J., Kumpf,S., Zappavigna,V., Mavilio,F. and Sampath,K. (

1994

) Inhibition of proliferation and induction of differentiation of pluripotent human embryonal carcinoma cells by osteogenic protein-1 (or bone morphogenetic protein-7).

Lab. Invest.

243

–251.

Fata,J.E., Leco,K.J., Voura,E.B., Yu,H.Y., Waterhouse,P., Murphy,G., Moorehead,R.A. and Khokha,R. (

2001

) Accelerated apoptosis in the Timp-3-deficient mammary gland.

J. Clin. Invest.

108

831

–841.

Amour,A., Slocombe,P.M., Webster,A.

et al

. (

1998

) TNF-alpha converting enzyme (TACE) is inhibited by TIMP-3.

FEBS Lett.

435

–44.

Sundararajan,R. and White,E. (

2001

) E1B 19K blocks Bax oligomerization and tumor necrosis factor alpha-mediated apoptosis.

J. Virol.

7506

–7516.

Nyormoi,O., Wang,Z., Doan,D., Ruiz,M., McConkey,D. and Bar-Eli,M. (

2001

) Transcription factor AP-2a is preferentially cleaved by caspase 6 and degraded by proteasome during tumor necrosis factor alpha-induced apoptosis in breast cancer cells.

Mol. Cell. Biol.

4856

–4867.

Datta,S.R., Dudek,H., Tao,X., Masters,S., Fu,H., Gotoh,Y. and Greenberg,M.E. (

1997

) Akt phosphorylation of BAD couples survival signals to the cell-intrinsic death machinery.

Cell

231

–241.

Cardone,M.H., Roy,N., Stennicke,H.R., Salvesen,G.S., Franke,T.F., Stanbridge,E., Frisch,S. and Reed,J.C. (

1998

) Regulation of cell death protease caspase-9 by phosphorylation.

Science

282

1318

–1321.

Biggs,W.H., III, Meisenhelder,J., Hunter,T., Cavenee,W.K. and Arden,K.C. (

1999

) Protein kinase B/Akt-mediated phosphorylation promotes nuclear exclusion of the winged helix transcription factor FKHR1.

Proc. Natl Acad. Sci. USA

7421

–7426.

Simpson,L. and Parsons,R. (

2001

) PTEN: life as a tumor suppressor.

Exp. Cell Res.

264

–41.

Wyder,L., Vitaliti,A., Schneider,H., Hebbard,L.W., Moritz,D.R., Wittmer,M., Ajmo,M. and Klemenz,R. (

2000

) Increased expression of H/T-cadherin in tumor-penetrating blood vessels.

Cancer Res.

4682

–4688.

Iba,K., Albrechtsen,R., Gilpin,B.

et al

. (

2000

) The cysteine-rich domain of human ADAM 12 supports cell adhesion through syndecans and triggers signaling events that lead to beta1 integrin-dependent cell spreading.

J. Cell Biol.

149

1143

–1156.

Zhang,X.A., Bontrager,A.L. and Hemler,M.E. (

2001

) Transmembrane-4 superfamily proteins associate with activated protein kinase C (PKC) and link PKC to specific beta(1) integrins.

J. Biol. Chem.

276

25005

–25013.

Alkemade,H.A., Vlijmen-Willems,I.M., van Haelst,U.J., van de Kerkhof,P.C. and Schalkwijk,J. (

1994

) Demonstration of skin-derived antileukoproteinase (SKALP) and its target enzyme human leukocyte elastase in squamous cell carcinoma.

J. Pathol.

174

121

–129.

Bokemeyer,D., Sorokin,A., Yan,M., Ahn,N., Templeton,D. and Dunn,M. (

1996

) Induction of mitogen-activated protein kinase phosphatase 1 by the stress-activated protein kinase signaling pathway but not by extracellular signal-regulated kinase in fibroblasts.

J. Biol. Chem.

271

639

–642.

Franklin,C.C. and Kraft,A.S. (

1997

) Conditional expression of the mitogen-activated protein kinase (MAPK) phosphatase MKP-1 preferentially inhibits p38 MAPK and stress-activated protein kinase in U937 cells.

J. Biol. Chem.

272

16917

–16923.

Fuchs,S., Adler,V., Buschmann,T., Yin,Z., Wu,X., Jones,S.N. and Ronai,Z. (

1998

) JNK targets p53 ubiquitination and degradation in nonstressed cells.

Genes Dev.

2658

–2663.

Fuchs,S.Y., Adler,V., Pincus,M.R. and Ronai,Z. (

1998

) MEKK1/JNK signaling stabilizes and activates p53.

Proc. Natl Acad. Sci. USA

10541

–10546.