Gene Expression Patterns as Potential Molecular Biomarkers for Malignant Transformation in Human Keratinocytes Treated with MNNG, Arsenic, or a Metal Mixture

Dong-Soon Bae*, Robert J. Handa{dagger}, Raymond S. H. Yang{ddagger} and Julie A. Campain{ddagger},1

* Laboratory of Cellular Carcinogenesis and Tumor Promotion, National Institutes of Health, Bethesda, Maryland 20892; {dagger} Department of Biomedical Sciences and {ddagger} Quantitative and Computational Toxicology Group, Center for Environmental Toxicology and Technology, Department of Environmental and Radiological Sciences, Colorado State University, Fort Collins, Colorado 80523

Received January 27, 2003; accepted April 17, 2003


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
In previous studies, treatment with 1-methyl-3-nitro-1-nitrosoguanidine (MNNG) enhanced malignant transformation of immortal human epidermal (RHEK-1) keratinocytes. In contrast, arsenic (As) alone or in a mixture of As, cadmium (Cd), chromium (Cr), and lead (Pb) inhibited this process. Microarray analysis showed unique gene expression patterns in RHEK-1 exposed to MNNG, As, or the metal mixture. From this analysis, we have selected 16 genes potentially involved in the enhancement or inhibition of transformation. These 16 genes, nine (IFN inducible protein 9-27, MAA A32, CCLB protein, integrin ß4, XRCC1, K8, K18, MT3, MAPKK6) of which were altered in a chemical-specific manner and seven (MIC1, bikunin, MTS1, BMP4, RAD23A, DOC2, vimentin) of which were commonly affected by the MNNG and As or mixture treatments, were examined for expression in detail by real-time RT-PCR. Qualitatively, both microarray and real-time RT-PCR analyses gave comparable results for 15 of 16 genes, i.e., genes were consistently induced or suppressed under the different treatment regimens when measured by either technique. Of the seven genes altered in their expression by multiple chemical treatments, five showed patterns consistent with a role in the transformation process, i.e., they were oppositely regulated in MNNG-transformed RHEK-1 cells (designated as OM3) as compared to the nonmalignant As- and mixture-exposed cells. Through time-course studies, we also identified markers whose expression correlates with acquisition of transformation-associated characteristics in OM3. Identification of a battery of genes altered during progressive transformation of RHEK-1 should aid in developing a mechanistic understanding of this process, as well as strengthening the utility of these genes as biomarkers.

Key Words: arsenic; metal mixture; gene expression; cDNA microarray; real-time RT-PCR; human keratinocytes; RHEK-1 cells.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
In previous work in our laboratory using the immortalized human epidermal keratinocyte cell line, RHEK-1, we analyzed the transforming potential of arsenic (As) alone and in the presence of cadmium (Cd), chromium (Cr), and lead (Pb) (Bae et al., 2002Go). These metals frequently occur together as environmental contaminants. These studies demonstrated that As was not capable of enhancing malignant transformation in RHEK-1, either by itself or in the metal mixture. In fact, chronic low-level exposure to As or the mixture inhibited spontaneous progression of these cells in a dose-dependent manner. This was in direct contrast to 1-methyl-3-nitro-1-nitrosoguanidine (MNNG), which greatly increased the rate and extent of conversion of RHEK-1 to a highly tumorigenic phenotype, likely through its activity as a DNA-damaging agent and mutagen. Microarray analysis of treated cultures demonstrated alterations in gene expression that were unique to the different chemicals (Bae et al., 2002Go). At least some of the genes identified in this manner are likely involved in the process of transformation in this cell type.

Although, based upon epidemiological evidence, As, Cd, and Cr are human carcinogens, the exact mechanisms involved in transformation by these agents are unclear. The primary target for As in exposed individuals is the skin; in fact, squamous and basal cell carcinomas have reached epidemic proportions in countries with As-contaminated water supplies (Chen et al., 1985Go; Tseng et al., 1968Go). Studies designed to characterize the transforming potential of As have historically proven difficult, as the metal has generally failed to act in as a carcinogen in skin model systems. In recent years, several investigators were able to demonstrate that As can act as a carcinogen or copromoter in defined skin carcinogenesis protocols in mice (Germolec et al., 1998Go; Ng et al., 1999Go). In vitro, although both As and Cr, a well-known skin sensitizer, have substantial effects on growth and differentiation of epidermal keratinocytes (Germolec et al., 1996Go, 1997Go; Ye et al., 1995Go), they have not acted to directly transform this cell type under culture conditions (Bae et al., 2002Go). That transformation by arsenic is cell-type specific is evident in the fact that the metal acts as an anticancer agent for some hematopoietic malignancies (Achanzar et al., 2002Go; Zhang et al., 1998Go). Clearly, the intracellular milieu, dictated by gene expression patterns, is extremely important in the ability of As to participate in the transformation process. Characterization of the relationship among As exposure, gene expression alterations, and transforming potential of the metal under different conditions should help in delineating important molecular events that are mechanistically linked to the carcinogenic process.

New technologies in expression analysis at the RNA and protein levels have led to the development of the field of toxicogenomics, i.e., the use of genetic information to address issues that are crucial in toxicology. Microarray analysis provides the advantage of being able to investigate expression of thousands of genes simultaneously in chemically treated versus control cells; thus enabling one to gain a more complete view of batteries of genes affected by exposure. When combined with cluster analysis, the qualitative effect of a chemical on multiple molecular pathways can be determined. One limitation to the microarray technology, however, is that it is less sensitive and quantitative than the traditional technologies that it has replaced. Both RT-PCR and Northern analysis allow for the relative quantification of expression of individual genes. Therefore, studies of global gene expression using microarrays are best if used in conjunction with these other more highly quantitative technologies.

Advances in polymerase chain reaction (PCR) instrumentation and use of fluorescent labels for detection of amplification products have facilitated several approaches to real-time RT-PCR quantification of input target copies. In particular, hot-start PCR methodology with the glass capillary format of the LightCycler has been used for amplification and quantitative detection of a single copy of bacterial DNA of Chlamydia spp. (Huang et al., 2001aGo). Rajeevan et al. demonstrated that real-time RT-PCR based on LightCycler technology is well suited to validate DNA array results because it is quantitative, rapid, and requires 1000-fold less RNA than conventional assays (Rajeevan et al., 2001Go). Real-time RT-PCR has been used to confirm the results of microarray analysis on global gene expression in a variety of systems (Huang et al., 2001bGo; Shultz et al., 2001Go; Waters et al., 2001Go).

We describe here an extension of a previous work comparing microarray analysis with real-time RT-PCR on a defined subset of genes altered in their expression by transformation enhancing and inhibiting chemical treatments in immortalized human keratinocytes.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Chemicals.
Sodium metaarsenite (NaAsO2), cadmium chloride (CdCl2), chromium oxide (CrO3), chromium chloride (CrCl3), lead acetate ((C2H3O2)2Pb•3H2O), and dimethyl sulphoxide (DMSO) were purchased from Sigma Chemical Co. (St. Louis, MO). N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) was obtained from Aldrich (Milwaukee, WI).

Cell culture.
The AD12/SV40 immortalized keratinocyte cell line (RHEK-1) was originally obtained from Dr. J. Rhim (Center for Prostate Disease Research, Rockville, MD; Rhim et al., 1985Go). RHEK-1 cells and derived malignant populations were cultured in Dulbecco’s Modified Eagle’s medium supplemented with 100 U/ml penicillin, 0.1 mg/ml streptomycin, 10 mM L-glutamine, and 10% fetal bovine serum (Summit Biotechnology, Ft. Collins, CO).

Cell line establishment following exposure to MNNG, As, or As-containing mixture.
The detailed protocol used to obtain the RHEK-1 cell lines utilized in these studies is described in Bae et al.(2002)Go. The RHEK-1 cell lines were designated as follows: (1) OM1: control cell population treated with 0.5% DMSO for 24 h and then cultured for 6 months; (2) OM3: cells treated with 0.1 µg/ml MNNG for 24 h and then cultured for 6 months; (3) As-Con: control population treated with sterile distilled water for 6 months; (4) As-High: cells treated chronically with 14 ppb As3+ for 6 months; (5) Mix-Con: control population treated with sterile distilled water for 6 months; and (6) Mix-High: cells treated with 14 ppb As3+, 104 ppb Cr3+ and 6+ (1:1), 618 ppb Cd2+, and 332 ppb Pb2+ for 6 months. These metal concentrations correspond to the LC10 for each metal as obtained by the MTT assay (Bae et al., 2001Go). All control and test cultures were passaged at simultaneous time points and, other than for chemical exposure, were treated in an identical manner.

Real-time RT-PCR.
The real-time RT-PCR protocol was as adaptation of a previously published protocol (Solum and Handa, 2002Go). RNA from OM1 and OM3 cells at p2, p7, p13, p16 was analyzed. In order to eliminate one potential source of variability between the microarray and real-time RT-PCR, the same RNA samples from the different RHEK-1 populations established in Bae et al.(2002)Go were utilized for both types of analysis. Thus, the gene expression data generated from these two methodologies can be directly compared. To ensure consistency across RNA samples from the same populations and to eliminate the complicating factors of cell density and other culture conditions, multiple flasks of cells from each control and treatment group were grown to approximately 70% confluency and were pooled prior to RNA preparation. For real-time RT-PCR analysis, 2–3 separate reverse transcriptase reactions were carried out on each pooled RNA sample to produce 2–3 individual cDNAs for further amplification. The genes analyzed by real-time RT-PCR and the primers utilized are shown in Table 1Go. The sequences for gene-specific primers corresponding to the PCR targets on the Atlas Human Cancer 1.2 Array were obtained from Clontech. Primers themselves were synthesized and HPLC-purified by Genset (La Jolla, CA). External DNA standard was synthesized for each gene of interest by conventional PCR and the appropriate primer set. PCR products were run on an electrophoresis gel and purified using the Ambion spin columns (Austin, TX). The concentration of purified PCR product was determined by absorbance at 260 nm and was used to make a set of standard dilutions. The quantitative measurement of target transcript in chemically-treated and control cells was measured in a LightCycler (Roche, Indianapolis, IN) with FastStart DNA Master SYBR Green I and a hot-start PCR method as follows and as shown schematically for a representative DOC2 gene in Figure 1AGo. Briefly: (1) A standard curve for each primer pair was created using the five to seven serially diluted known amounts of cDNA product; (2) from this standard curve, the absolute concentration of the unknown gene was determined; and (3) a subsequent melting curve analysis and length verification by gel electrophoresis were carried out to confirm the specificity of PCR products. As a negative control, template cDNA was replaced with PCR-grade water. As demonstrated in Figure 1AGo, from our studies on the DOC2 gene, the estimated absolute amount of the DOC2 gene in the p25 OM1 sample was 36.5 fg which corresponds to a crossing point (Cp) of 20.9, that is the cycle number at which the amplified products of all the standards and unknowns emit the same fluorescence above the threshold/noise band (i.e., located at 0.5 in Fig. 1AGo). A representative melting curve for the DOC2 gene is shown in Figure 1BGo. The specific DOC2 product melts at 86°C, while the primer dimer melts at 80.5°C.


View this table:
[in this window]
[in a new window]
 
TABLE 1 Sequences of Primers Used in the Real-Time RT-PCR
 


View larger version (35K):
[in this window]
[in a new window]
 
FIG. 1. Schematic representation of DOC2 mRNA quantification in an unknown sample using LightCycler technology. (A) A representative real-time RT-PCR LightCycler reaction for the DOC2 gene. RT-PCR was carried out as described in Materials and Methods using the DOC2 primer set and p25 OM1 RNA. The standard curve used to calculate the concentration of mRNA in the unknown test sample was generated using amounts of DOC2 DNA standard between 0.069 and 69,000 fg. For quantitative analysis, the Fit Points Method with two fit points (designated with two upper +) was utilized to determine robust crossing points. Cp, crossing point. (B) A melting curve analysis for the DOC2 gene. From this analysis, the specificity of PCR products for all 16 genes tested was verified by observing a single peak with higher melting temperature than a primer dimer. T, temperature.

 
Statistical analysis.
One-way ANOVA followed by Dunnett’s test was used to analyze differences between control and chemical-treated samples in real-time RT-PCR studies. p values < 0.05 were considered statistically significant.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Analysis by Real-Time RT-PCR of Expression Levels of Selected Genes Altered in a Chemical-Specific Manner in RHEK-1 Cells
To confirm our findings with microarray analysis (Bae et al., 2002Go) and to further explore these alterations in gene expression, we have performed real-time RT-PCR on a selected subset of genes which were altered in their expression in RHEK-1 in a chemical-specific manner. There are 1185 human cDNA clones potentially associated with carcinogenesis on the Clontech Atlas Human Cancer 1.2 array. The total number of genes altered greater than twofold in expression using this array were 63, 70, and 76, for OM3, As-High, and Mix-High, respectively, as compared to control, OM1. Two groups of genes were selected for real-time RT-PCR analysis: (1) those that exhibited the highest magnitude of induction or suppression in individual chemically treated populations, i.e., chemical-specific expression pattern, or (2) those that were altered under multiple chemical treatments.

In MNNG-treated OM3, the genes selected were interferon (IFN) inducible protein 9–27, melanoma associated antigen A32 (MAA A32), and the colon carcinoma laminin binding (CCLB) protein genes (which were all induced ≥ fourfold) and integrin ß4 (which was downregulated). No detectable changes in these four genes were observed in As- or mixture-treated RHEK-1 cells (data not shown). The genes chosen for further analysis from As-High, were DNA repair protein XRCC 1 (which was substantially upregulated in treated versus control cells) and those encoding the simple epithelial keratins 8 and 18 (K8 and K18); these latter two genes were representative of the many cytokeratins suppressed in As-High populations. There were no measurable changes in expression of these three genes in OM3 and mixture-exposed cells. As would be expected, RHEK-1 cells exposed to the four metal mixture (which included Cd) demonstrated increased mRNA levels for metallothionein 3 (MT 3). We chose this gene for further analysis and the gene encoding the important mitogenic signal transduction protein, mitogen activated protein kinase kinase 6 (MAPKK 6) as a representative of repressed genes in this cell population.

Pooled results from multiple real-time RT-PCR analyses on the nine genes selected from the chemical-specific gene expression profiles are shown in Table 2Go. Careful scrutiny of this data revealed that in approximately half of the cases, gene-specific comparative analysis between the three chemically treated samples and their respective controls gave very consistent absolute RT-PCR values in all samples analyzed, with SE ≤ 20%. In the other instances, however, while a consistent induction or suppression was observed in all three replicates from a treatment condition, the absolute fold alteration in expression varied by >20%. Table 2Go also demonstrates for comparison, the alterations in expression of these genes in the three chemically treated RHEK-1 populations as measured initially by microarray analysis. From a qualitative standpoint, both types of analysis gave comparable results for these specific genes, that is, genes were consistently induced or suppressed under the different treatment regimens when measured by either technique. The absolute level of change in expression measured by microarray versus real-time RT-PCR varied in our hands by a maximum of approximately fourfold, with, in most cases, the values of induction or suppression being larger by microarray analysis.


View this table:
[in this window]
[in a new window]
 
TABLE 2 Comparison of Gene Expression Changes in Chemically Treated RHEK-1 as Measured by Microarray and Real-Time RT-PCR: Selected Genes Altered in a Chemical-Specific Manner
 
Real-Time RT-PCR on Selected Genes Altered by Multiple Chemical Treatments in RHEK-1 Cells
We were interested in identifying genes that were altered in response to more than one of our tested chemical treatments. As MNNG acted to enhance, and both metal treatments inhibited, spontaneous transformation in RHEK-1, correlations made between these different outcomes and specific gene expression changes may provide mechanistic clues to the cancer process in this cell type. Seven genes in this category were more closely analyzed by real-time RT-PCR. Table 3Go demonstrates the mean induction/suppression (± SE) of these genes as measured by real-time RT-PCR in multiple samples from each of the individual chemical treatment groups as compared to its respective control. As discussed above for the genes induced in a chemical-specific fashion, the variability in triplicates for these multiple samples ranged from 6% to as high as 63%; approximately half of the samples varied by less than 20%. All seven genes were, however, consistent in their pattern of induction or suppression among the triplicates; this data also agreed well with the microarray analysis for six of the seven genes. Our original microarray data on bikunin indicated repression in both OM3 and the As-High samples. However, the signal intensity from this data was very low, indicating potential problems in analysis. Once a more sensitive analysis of expression of this gene was carried out by real-time RT-PCR, we were able to demonstrate that the bikunin gene is repressed by approximately 40% in OM3 and, in contrast, is induced in both the As-High and Mix-High cells. Of the seven genes analyzed in this manner, five showed expression patterns consistent with a role in the transformation process, i.e., they were oppositely regulated in OM3 as compared to the nonmalignant As-High and Mix-High cells. These genes were macrophage inhibitory cytokine 1 (MIC1), bikunin, multiple tumor suppressor 1 (MTS1), bone morphogenetic protein 4 (BMP4), and UV excision repair protein RAD23A. MIC1 was induced by MNNG and suppressed by both As and the metal mixture. Two tumor suppressor genes, bikunin and MTS1, and another TGFß-like protein, BMP4, showed opposite regulation, being suppressed in OM3 and overexpressed in the lines derived from As- and metal-mixture treatment. RAD23A also showed this latter pattern. The remaining two genes, vimentin and DOC2, were substantially upregulated in RHEK-1 exposed to both MNNG and As, and showed decreased expression in the mixture treated cells. We have recently confirmed by proteomic analysis that the vimentin protein is indeed overexpressed in OM3 (data not shown).


View this table:
[in this window]
[in a new window]
 
TABLE 3 Comparison of Gene Expression Changes in Chemically Treated RHEK-1 as Measured by Microarray and Real-Time RT-PCR Analyses: Genes Altered by Multiple Chemical Treatments
 
Time-Course Studies on MNNG-Associated Genes by Real-Time RT-PCR
We are interested in identifying molecular markers that correlate with acquisition of measurable transformation-associated phenotypic characteristics in our MNNG-transformed keratinocytes. To address this issue, we generated time-course expression data on the IFN inducible protein, integrin ß4, MAA A32, CCLB, and DOC2 genes in OM3, along with its control OM1, with continued passage in culture from the initial chemical treatment to acquisition of a fully malignant phenotype. The schematic of chemical transformation studies in RHEK-1 human keratinocytes is described in Figure 2Go. Gene expression patterns were measured at p2, p7, p13, p16, and p25 by real-time RT-PCR in OM3 as compared to OM1 and are graphically illustrated in Figure 3Go. These passages in OM3 correspond to (1) cells immediately following chemical treatment and with no discernible phenotypic changes, (2) the first passage of cells demonstrating subtle morphological alterations, (3) morphologically distinct cells that had acquired the ability to grow in an anchorage-independent manner (AIG+), (4) a tumorigenic population in immunocompromised mice, and (5) an extremely malignant population, generating tumors greater than 1 cm in three weeks or less and exhibiting a colony forming efficiency in methylcellulose of approximately 20%. It should be noted that during continual passage, OM1 also underwent transformation-associated changes, albeit at a much slower rate than that observed in OM3. This phenomenon is likely due to the Ad12/SV40 viral infection used to immortalize the cells. In contrast to OM3, the p11, p13, and p16 OM1 cells were anchorage-dependent. By p25, however, OM1 began to demonstrate phenotypic changes that were consistent with partial transformation; among these changes was acquisition of anchorage-independent growth. Approximately 30% of mice injected with OM1 cells at p25 developed small, slowly growing squamous cell carcinomas. The phenotypic changes are described in detail in a previous publication (Bae et al., 2002Go). In summary, p11 to p16 in OM3 and p25 in OM1 appeared to be important time points for analysis of gene expression alterations that correlate with malignant conversion in these cell lines.



View larger version (15K):
[in this window]
[in a new window]
 
FIG. 2. Schematic of chemical transformation studies in RHEK-1 human keratinocytes. RHEK-1 cells were treated with MNNG, As, or the metal mixture as described previously (Bae et al., 2002Go) and as summarized in Materials and Methods. Functional endpoints that have previously been associated with malignant transformation in this cell type (Rhim et al., 1985Go) were assayed in treated and control cultures approximately every other passage (p). One passage represents approximately seven days in culture. Passages at which cells underwent detectable alterations in morphology and/or acquired a more transformed phenotype are indicated. AIG, anchorage-independent growth; D, day.

 


View larger version (23K):
[in this window]
[in a new window]
 
FIG. 3. Time-dependent changes in expression of the IFN inducible protein 9–27, Integrin ß4, MAA A32, CCLB protein, and DOC 2 genes. Real-time RT-PCR was carried out as described in Materials and Methods using the appropriate primer sets as shown in Table 1Go. Cell populations analyzed were from various subcultures (up to passage [p] 25) after parental RHEK-1 cells were treated with 0.1 µg/ml MNNG or DMSO control. Both populations were analyzed at identical passage numbers, and had been cultured for the same cumulative number of days. The absolute amount of (A) IFN inducible protein 9–27, (B) Integrin ß4, (C) MAA A32, (D) CCLB protein, and (E) DOC2 mRNA in OM3 as compared to OM1 at various time points is plotted. Mean value is shown with SE from duplicate experiments. *Statistically different from p2 cells at p < 0.05. The tumorigenic potential of OM3 cells at p25, OM3 cells at p16, and OM1 cells at p25 (Bae et al., 2002Go) was very high, high, and moderate, respectively.

 
As demonstrated in Figure 3Go, our time-course analysis revealed the following: (1) mRNA levels for the IFN-inducible protein gene changed with time in both OM1 and OM3; expression of the gene decreased sequentially in OM1, showing a maximum inhibition at passage 16 of 203-fold as compared to p2. In OM3, however, expression of IFN-inducible protein initially decreased at p7, but then proceeded to rise again until reaching a level slightly higher than basal at p25; none of the changes in expression of the IFN-inducible gene observed in OM3 were, however, statistically significant as compared to the p2 control (Fig. 3AGo). (2) The gene for integrin ß4 also showed a unique pattern of expression throughout the time course. In OM1, mRNA for integrin ß4 increased in a time-dependent fashion, reaching maximum levels of expression at p25, which was a 19-fold increase over basal levels at p2. In OM3, however, expression of the gene decreased in a linear fashion up until p16, after which it rose to levels slightly higher than basal at p25 (Fig. 3BGo). (3) mRNA levels for the MAA A32, CCLB protein, and DOC2 genes increased with increasing time in culture in both OM1 and OM3. Expression levels of all three genes peaked at p25 in both cell lines, the last time point measured (Figs. 3C–3EGo). (4) Maximum expression of the MAA A32 and DOC2 genes was approximately three- to fourfold higher in OM3 as compared to OM1. (5) The CCLB protein gene was expressed at very low levels in both OM1 and OM3 through p16; when measured in p25 cells, however, expression was substantially induced, 75- and 36-fold as compared to basal p2, respectively for OM1 and OM3. (6) The MAA A32 and DOC2 protein genes exhibited very interesting kinetics in both the control and MNNG-treated cells, increasing expression of these genes occurred earlier (at or before p16) in OM3 and continued to rise through p25. In contrast, in OM1, MAA A32, and DOC2 levels were very low at p16 and did not begin to rise substantially until p25 (Figs. 3CGo and 3EGo).


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Previous microarray studies in our laboratory demonstrated unique expression profiles of cancer-associated genes following exposure of RHEK-1 keratinocytes to As, an As-containing chemical mixture, or MNNG (Bae et al., 2002Go). The studies described herein utilized real-time RT-PCR to accurately quantify expression in these cell lines of subsets of genes representing those showing a chemical-specific expression pattern and those altered in more than one treatment group.

In OM3 versus OM1, the genes that were analyzed by real-time RT-PCR were the IFN inducible protein, integrin ß4, MAA A32, CCLB, and the DOC2 genes in both p25, when the OM3 cells were fully transformed, and in earlier passages in our time-course study. Of these, the MAA A32, CCLB protein, and DOC2 genes were regulated in a time-dependent manner consistent with a role in transformation. The more rapid induction of these genes in OM3 as compared to OM1 and/or the higher levels of expression in the more malignant cells at p25 suggest that these genes may be markers of the malignant process. Our findings are consistent with work from other laboratories on the MAA A32 and CCLB protein genes. MAA A32 is a marker of tumor progression in human melanoma (Lehmann et al., 1989Go). The CCLB protein is a mitotic phosphoprotein that is upregulated in cancer cells in correlation with their metastatic phenotype (Jackers et al., 1996Go). Time-dependent changes in expression of the DOC2 gene in OM1 and OM3 suggest that overexpression might be a late event involved in acquisition of characteristics that support tumor formation in mice. Previous studies have linked the DOC2 gene to activation of the cell cycle (Xu et al., 1995Go). Overexpression of DOC2 may reflect the aggressive growth characteristics of late passage, and tumorigenic, OM3 cells.

Although the genes for integrin ß4 and IFN-inducible protein were also altered in both OM1 and OM3 in a time-dependent fashion, these expression patterns indicated that neither gene is likely involved in malignant progression in RHEK-1. Decreased expression of integrin ß4 has previously been implicated in tumor cell growth and metastasis (Gille and Swerlick, 1996Go). It is not clear why expression of this gene increases in OM1 with time in culture, while the cells are spontaneously gaining more characteristics of the malignant phenotype. Likewise, in OM3, substantial suppression of the integrin ß4 gene occurs between p2 and p16; at p25, however, when the cells are their most tumorigenic, expression of the gene returns to basal. The IFN-inducible protein gene, while not significantly altered over the time course in OM3, decreased in OM1 with continued time in culture. The significance of our findings concerning this gene is unclear, as previous studies suggested that its protein product, the Leu-13 antigen, is often overexpressed in malignant conditions (Evans et al., 1990Go).

Genes further analyzed in As and the metal-mixture treated RHEK-1 populations were XRCC1, K8, and K18 for As-High and MT3 and MAPKK6 for Mix-high cells. The human XRCC1 gene product is involved in DNA repair (Thompson et al., 1990Go); induction of this gene in cells treated with As may well be the normal response to the clastogenic activities of the metal. The two simple epithelial cytokeratins, K8 and K18, are highly expressed in several epithelial cancers and, are only expressed in keratinocytes under abnormal conditions (Trask et al., 1990Go; Welsh et al., 2001Go). Their suppression in As-treated RHEK-1 cells is consistent with the observed inhibition of spontaneous cell transformation. Induction of MT3 expression in Mix-High populations was not surprising due to the presence of Cd in the mixture (Hellemans et al., 1999Go). MAPKK6 is involved in activation of the p38 MAP kinase signal transduction pathway (Raingeaud et al., 1996Go) and has been linked to cellular transformation (Chiariello et al., 2000Go). Repression of MAPKK6 in Mix-High cells correlates with the poor tumorigenic potential of this cell population.

We were highly interested in genes that were altered in RHEK-1 by more than one chemical treatment. These genes, if expressed in a manner consistent with a role in transformation, i.e., oppositely regulated by MNNG, which enhances transformation, and the As/mixture treatments, which are inhibitory to the process, may potentially provide us with candidate molecular markers for malignant conversion. In our microarray studies, only the MIC1 gene was found to be expressed at high levels in OM3 and repressed in As-High and Mix-High. The MIC1 gene encodes a macrophage inhibitory cytokine; higher expression of this gene has recently been demonstrated in advanced and more aggressive prostatic tumors than in noncancerous tissues (Nakamura et al., 2003Go). Although it is unclear exactly how MIC1 may be involved in transformation, the protein is a divergent member of the TGFß superfamily. Many members of this family are highly active in keratinocytes (Bootcov et al., 1997Go). Transformed cells often express TGFß; production of the cytokine by resistant tumor cells could potentially give them a substantial growth advantage over surrounding normal cells (Fynan and Reiss, 1993Go).

Four genes, bikunin, MTS1, BMP4, and RAD23A were all significantly repressed in OM3 and induced in the populations derived from As or metal mixture treatment. Evidence would support a potential role for any of these genes in inhibition of the carcinogenic process in keratinocytes. Reduced expression of the bikunin gene has been implicated in progression of human glioblastoma (Hamasuna et al., 2001Go). MTS1 (p16-INK4) is an inhibitor of cyclin/cyclin-dependent kinase complexes (Luca et al., 1995Go). Under normal conditions, the MTS1 gene is involved in induction of the growth arrest that accompanies terminal differentiation in keratinocytes; squamous cell carcinoma lines are often defective in induction of MTS1 in response to differentiating agents (Harvat and Jetten, 2001Go). The gene is deleted or mutated with high frequency in human melanoma cell lines and familial melanoma patients (Luca et al.,1995Go). The role of TGFß-like BMP4 in carcinogenesis is not yet defined. One function, however, suggested by Glozak and Rogers, is that BMP4 may induce apoptosis in some cell types (Glozak and Rogers, 1996Go). Members of the BMP family, particularly BMP2 and BMP6, are important regulators of keratinocyte growth and differentiation. Altered expression of these proteins has been demonstrated in cancers of epidermal origin (Jin et al., 2001Go). The remaining gene, RAD23A or HHR23A, is a nucleotide excision repair protein whose normal function is important for prevention of skin cancer (Masutani et al., 1994Go).

The vimentin gene, along with the DOC2 gene as discussed above, was induced in OM3 and As-High populations, but repressed in cells treated with the mixture. Overexpression of vimentin, an intermediate filament protein, is frequently observed in cancer cells and correlates with malignancy (Dandachi et al., 2001Go). There are several possible explanations for the expression pattern we observe with the DOC2 and vimentin genes: (1) these two genes are unrelated to the process of transformation in RHEK-1 and their altered pattern of expression is the indirect result of other changes in chemically-treated cells; (2) expression of the gene(s) may be associated with, but not sufficient for malignant conversion, and other gene products present in OM3 are necessary for expression of the tumorigenic phenotype; or (3) other proteins present in As-H cells interfere in the process of transformation either through direct inhibition of DOC2 or vimentin activity or indirectly by blocking other steps in the pathway.

The overall goal of our studies is to identify potential target genes involved in carcinogenic activity of environmentally relevant chemicals in keratinocytes. To this end, we have utilized cDNA microarray analysis and real-time RT-PCR validation to correlate gene expression changes in chemically treated RHEK-1 cells with time-sensitive acquisition of transformation-associated characteristics. Presently, we are expanding our effort to study gene expression profiles of additional carcinogenic and noncarcinogenic chemicals in RHEK-1. Once candidate genes have been identified, mechanistic studies will be required to define the role of these genes in transformation. As RHEK-1 cells spontaneously transform in culture, one possibility is that the same "cancer-associated" gene expression alterations will eventually occur in chemically treated and control populations, albeit at very different rates. These alterations could read-out as perturbations in a common growth or differentiation pathway in keratinocytes such as those involving ras or protein kinase C. Alternatively, strong mutagens such as MNNG, by nature of their mode-of-action, could induce different gene expression changes each time. Control cells could either be genetically unstable and also undergo a variety of carcinogenic changes each time the cells are allowed to transform or could be altered along the same pathway each time, both possibilities would likely be due to viral infection and abrogated function of key tumor suppressor genes. These different possibilities will all be amenable to further study. This work should aid us in our pursuit of reliable toxicity markers for use in computational risk assessment approaches.


    ACKNOWLEDGMENTS
 
This study was supported by the Agency for Toxic Substances and Disease Registry (ATSDR) Cooperative Agreement U61/ATU881475, and by the National Institute for Environmental Health Sciences (NIEHS) Superfund Basic Research Program Project P42 ES05949.


    NOTES
 
1 To whom correspondence should be addressed. Fax: (970) 491-8304. E-mail: julie.campain{at}colostate.edu. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Achanzar, W. E., Brambila, E. M., Diwan, B. A., Webber, M. M., and Waalkes, M. P. (2002). Inorganic arsenite-induced malignant transformation of human prostate epithelial cells. J. Natl. Cancer Inst. 94, 1888–1891.[Abstract/Free Full Text]

Bae, D. S., Gennings, C., Carter, W. H., Jr., Yang, R. S. H., and Campain, J. A. (2001). Toxicological interactions among arsenic, cadmium, chromium, and lead in human keratinocytes. Toxicol. Sci. 63, 132–142.[Abstract/Free Full Text]

Bae, D. S., Hanneman, W. H., Yang, R. S. H., and Campain, J. A. (2002). Characterization of gene expression changes associated with MNNG, arsenic, or metal mixture treatment in human keratinocytes: Application of cDNA microarray technology. Environ. Health Perspect. 110(Suppl. 6), 931–941.[ISI][Medline]

Bootcov, M. R., Bauskin, A. R., Valenzuela, S. M., Moore, A. G., Bansal, M., He, X. Y., Zhang, H. P., Donnellan, M., Mahler, S., Pryor, K., et al. (1997). MIC-1, a novel macrophage inhibitory cytokine, is a divergent member of the TGF-beta superfamily. Proc. Natl. Acad. Sci. U.S.A. 94, 11514–11519.[Abstract/Free Full Text]

Chen, C.-J., Chuang, Y.-C., Lin, T.-M., and Wu, H.-Y. (1985). Malignant neoplasms among residents of a blackfoot disease-endemic area in Taiwan: High-arsenic artesian well water and cancer. Cancer Res. 45, 5895–5899.[Abstract]

Chiariello, M., Marinissen, M. J., and Gutkind, J. S. (2000). Multiple mitogen-activated protein kinase signaling pathways connect the cot oncoprotein to the c-jun promoter and to cellular transformation. Mol. Cell Biol. 20, 1747–1758.[Abstract/Free Full Text]

Dandachi, N., Hauser-Kronberger, C., More, E., Wiesener, B., Hacker, G. W., Dietze, O., and Wirl, G. (2001). Co-expression of tenascin-C and vimentin in human breast cancer cells indicates phenotypic transdifferentiation during tumour progression: Correlation with histopathological parameters, hormone receptors, and oncoproteins. J. Pathol. 193, 181–189.[CrossRef][ISI][Medline]

Evans, S. S., Lee, D. B., Han, T., Tomasi, T. B., and Evans, R. L. (1990). Monoclonal antibody to the interferon-inducible protein Leu-13 triggers aggregation and inhibits proliferation of leukemic B cells. Blood 76, 2583–2593.[Abstract]

Fynan, T. M., and Reiss, M. (1993). Resistance to inhibition of cell growth by transforming growth factor- beta and its role in oncogenesis. Crit. Rev. Oncog. 4, 493–540.[ISI][Medline]

Germolec, D. R., Spalding, J., Boorman, G. A., Wilmer, J. L., Yoshida, T., Simeonova, P. P., Bruccoleri, A., Kayama, F., Gaido, K., Tennant, R., et al. (1997). Arsenic can mediate skin neoplasia by chronic stimulation of keratinocyte-derived growth factors. Mut. Res. 386, 209–218.[CrossRef][ISI][Medline]

Germolec, D. R., Spalding, J., Yu, H.-S., Chen, G. S., Simeonova, P. P., Humble, M. C., Bruccoleri, A., Boorman, G. A., Foley, J. F., Yoshida, T., et al. (1998). Arsenic enhancement of skin neoplasia by chronic stimulation of growth factors. Am. J. Pathol. 153, 1775–1785.[Abstract/Free Full Text]

Germolec, D. R., Yoshida, T., Gaido, K., Wilmer, J. L., Simeonova, P. P., Kayama, F., Burleson, F., Dong, W., Lang, R. W., and Luster, M. I. (1996). Arsenic induces overexpression of growth factors in human keratinocytes. Toxicol. Appl. Pharmacol. 141, 308–318.[CrossRef][ISI][Medline]

Gille, J., and Swerlick, R. A. (1996). Integrins: Role in cell adhesion and communication. Ann. N.Y. Acad. Sci. 797, 93–106.[Abstract]

Glozak, M. A., and Rogers, M. B. (1996). Specific induction of apoptosis in P19 embryonal carcinoma cells by retinoic acid and BMP2 or BMP4. Dev. Biol. 179, 458–470.[CrossRef][ISI][Medline]

Hamasuna, R., Kataoka, H., Meng, J. Y., Itoh, H., Moriyama, T., Wakisaka, S., and Koono, M. (2001). Reduced expression of hepatocyte growth factor activator inhibitor type- 2/placental bikunin (HAI-2/PB) in human glioblastomas: Implication for anti-invasive role of HAI-2/PB in glioblastoma cells. Int. J. Cancer 93, 339–345.[CrossRef][ISI][Medline]

Harvat, B. L., and Jetten, A. M. (2001). Decreased growth inhibitory responses of squamous carcinoma cells to interferon-gamma involve failure to recruit cki proteins into cdk2 complexes. J. Invest. Dermatol. 117, 1274–1281.[Abstract/Free Full Text]

Hellemans, G., Soumillion, A., Proost, P., Van Damme, J., Van Poppel, H., Baert, L., and De Ley, M. (1999). Metallothioneins in human kidneys and associated tumors. Nephron 83, 331–340.[CrossRef][ISI][Medline]

Huang, J., DeGraves, F. J., Gao, D., Feng, P., Schlapp, T., and Kaltenboeck, B. (2001a). Quantitative detection of Chlamydia spp. by fluorescent PCR in the LightCycler. Biotechniques 30, 150–157.[ISI][Medline]

Huang, Q., Dunn, R. T., Jayadev, S., DiSorbo, O., Pack, F. D., Farr, S. B., Stoll, R. E., and Blanchard, K. T. (2001b). Assessment of cisplatin-induced nephrotoxicity by microarray technology. Toxicol. Sci. 63, 196–207.[Abstract/Free Full Text]

Jackers, P., Minoletti, F., Belotti, D., Clausse, N., Sozzi, G., Sobel, M. E., and Castronovo, V. (1996). Isolation from a multigene family of the active human gene of the metastasis-associated multifunctional protein 37LRP/p40 at chromosome 3p21.3. Oncogene 13, 495–503.[ISI][Medline]

Jin, Y., Tipoe, G. L., Liong, E. C., Lau, T. Y., Fung, P. C., and Leung, K. M. (2001). Overexpression of BMP-2/4, -5 and BMPR-IA associated with malignancy of oral epithelium. Oral Oncol. 37, 225–233.[CrossRef][ISI][Medline]

Lehmann, J. M., Riethmuller, G., and Johnson, J. P. (1989). MUC18, a marker of tumor progression in human melanoma, shows sequence similarity to the neural cell adhesion molecules of the immunoglobulin superfamily. Proc. Natl. Acad. Sci. U.S.A. 86, 9891–9895.[Abstract]

Luca, M., Xie, S., Gutman, M., Huang, S., and Bar-Eli, M. (1995). Abnormalities in the CDKN2 (p16INK4/MTS-1) gene in human melanoma cells: Relevance to tumor growth and metastasis. Oncogene 11, 1399–1402.[ISI][Medline]

Masutani, C., Sugasawa, K., Yanagisawa, J., Sonoyama, T., Ui, M., Enomoto, T., Takio, K., Tanaka, K., van der Spek, P. J., Bootsma, D., et al. (1994). Purification and cloning of a nucleotide excision repair complex involving the xeroderma pigmentosum group C protein and a human homologue of yeast RAD23. EMBO J. 13, 1831–1843.[Abstract]

Nakamura, T., Scorilas, A., Stephan, C., Yousef, G. M., Kristiansen, G., Jung, K., and Diamandis, E. P. (2003). Quantitative analysis of macrophage inhibitory cytokine-1 (MIC-1) gene expression in human prostatic tissues. Br. J. Cancer 88, 1101–1104.[CrossRef][ISI][Medline]

Ng, J. C., Seawright, A. A., Qi, L., Garnett, C. M., Chiswell, B., and Moore, M. R. (1999). Tumors in mice induced by exposure to sodium arsenate in drinking water. In Arsenic Exposure and Health Effects (W. R. Chappell, C. O. Abernathy, and R. L. Calderon, Eds.), pp. 217–223. Elsevier Science, New York, NY.

Raingeaud, J., Whitmarsh, A. J., Barrett, T., Derijard, B., and Davis, R. J. (1996). MKK3- and MKK6-regulated gene expression is mediated by the p38 mitogen-activated protein kinase signal transduction pathway. Mol. Cell Biol. 16, 1247–1255.[Abstract]

Rajeevan, M. S., Vernon, S. D., Taysavang, N., and Unger, E. R. (2001). Validation of array-based gene expression profiles by real-time (kinetic) RT-PCR. J. Mol. Diagn. 3, 26–31.[Abstract/Free Full Text]

Rhim, J. S., Jay, G., Arnstein, P., Price, F. M., Sanford, K. K., and Aaronson, S. A. (1985). Neoplastic transformation of human epidermal keratinocytes by AD12-SV40 and Kirsten sarcoma viruses. Science 227, 1250–1252.[ISI][Medline]

Shultz, V. D., Phillips, S., Sar, M., Foster, P. M., and Gaido, K. W. (2001). Altered gene profiles in fetal rat testes after in utero exposure to di(n-butyl) phthalate. Toxicol. Sci. 64, 233–242.[Abstract/Free Full Text]

Solum, D. T., and Handa, R. J. (2002). Estrogen regulates the development of brain-derived neurotrophic factor mRNA and protein in the rat hippocampus. J. Neurosci. 22, 2650–2659.[Abstract/Free Full Text]

Thompson, L. H., Brookman, K. W., Jones, N. J., Allen, S. A., and Carrano, A. V. (1990). Molecular cloning of the human XRCC1 gene, which corrects defective DNA strand break repair and sister chromatid exchange. Mol. Cell. Biol. 10, 6160–6171.[ISI][Medline]

Trask, D. K., Band, V., Zajchowski, D. A., Yaswen, P., Suh, T., and Sager, R. (1990). Keratins as markers that distinguish normal and tumor-derived mammary epithelial cells. Proc. Natl. Acad Sci. U.S.A. 87, 2319–2323.[Abstract]

Tseng, W. P., Chu, H. M., How, S. W., Fong, J. M., Lin, C. S., and Yeh, S. (1968). Prevalence of skin cancer in an endemic area of chronic arsenicism in Taiwan. J. Natl. Cancer Inst. 40, 453–463.[ISI][Medline]

Waters, K. M., Safe, S., and Gaido, K. W. (2001). Differential gene expression in response to methoxychlor and estradiol through ERalpha, ERbeta, and AR in reproductive tissues of female mice. Toxicol. Sci. 63, 47–56.[Abstract/Free Full Text]

Welsh, J. B., Zarrinkar, P. P., Sapinoso, L. M., Kern, S. G., Behling, C. A., Monk, B. J., Lockhart, D. J., Burger, R. A., and Hampton, G. M. (2001). Analysis of gene expression profiles in normal and neoplastic ovarian tissuesamples identifies candidate molecular markers of epithelial ovarian cancer. Proc. Natl. Acad. Sci. U.S.A. 98, 1176–1181.[Abstract/Free Full Text]

Xu, X. X., Yang, W., Jackowski, S., and Rock, C. O. (1995). Cloning of a novel phosphoprotein regulated by colony-stimulating factor 1 shares a domain with the Drosophila disabled gene product. J. Biol. Chem. 270, 14184–14191.[Abstract/Free Full Text]

Ye, J., Zhang, X., Young, H. A., Mao, Y., and Shi, X. (1995). Chromium(VI)-induced nuclear factor-kB activation in intact cells via free radical reactions. Carcinogenesis 16, 2401–2405.[Abstract]

Zhang, W., Ohnishi, K., Shigeno, K., Fujisawa, S., Naito, K., Nakamura, S., Takeshita, K., Takeshita, A., and Ohno, R. (1998). The induction of apoptosis and cell cycle arrest by arsenic trioxide in lymphoid neoplasms. Leukemia 12, 1383–1391.[CrossRef][ISI][Medline]