Microarray Analysis and RNA Silencing Link fra-1 to cd44 and c-met Expression in Mesothelioma¹

Asbestos is a naturally occurring group of mineral fibers that gives rise to pulmonary and pleural fibrosis, lung cancers, and mesotheliomas (1). The development of malignant mesotheliomas after exposures to crocidolite asbestos in man is well documented, although the molecular mechanisms of fiber-induced carcinogenesis are poorly understood.

Asbestos fibers may act at several stages of the multistage process of tumorigenesis. For example, fibers initiate DNA damage in vitro by causing oxidative lesions such as 8-oxo-deoxyguanosine in mesothelial cells and large base deletions associated with mutagenesis (2, 3). In addition, asbestos fibers stimulate DNA synthesis and hyperplasia in mesothelial and pulmonary epithelial cells after inhalation and are inflammatory agents (4, 5), thus suggesting their role as tumor promoters during the long latency period of tumor development. The chemical composition, length, and durability of different asbestos fiber types are critical in governing their deposition, reactivity, and biopersistence in lung and pleura and may be important in tumorigenicity (reviewed in Ref. 6).

We have demonstrated that crocidolite asbestos, the most pathogenic type of asbestos in the induction of human mesothelioma (1, 7), causes activation of ERK1³ and ERK2 after phosphorylation of the epidermal growth factor receptor in mesothelial cells (8, 9). These events are linked to the increased expression of AP-1 family members (fos/jun) and transactivation of AP-1-dependent gene expression (10, 11). Increases in fra-1 (fos-related antigen-1) expression are particularly striking and protracted in both pulmonary epithelial and mesothelial cells after exposure to crocidolite asbestos fibers as opposed to nonpathogenic particles (10, 11, 12, 13). Most recently, we have shown that fra-1 expression is causally related to phenotypic changes and anchorage-independent growth during the development of rat mesotheliomas (13).

The objective of studies here was to characterize, using oligonucleotide microarray analysis, subsets of up- or down-regulated gene expression in mesothelial cells after acute exposures to asbestos and in mesotheliomas. After confirming that fra-1 mRNA was dramatically increased after exposures of RPM cells to asbestos and in both human and rat mesotheliomas, we selected four candidate genes following patterns of fra-1 expression. After confirmation of changes in mRNA levels using real-time Q-PCR, we then used RNA interference technology to address the hypothesis that expression of some genes would be fra-1 dependent. Our results reveal that expression of c-met and cd44, genes encoding receptors linked to migration and invasiveness of tumors, is linked to fra-1 expression in asbestos-treated mesothelial cells and mesotheliomas.

RPM cells were isolated from the parietal pleura of Fischer 344 rats by methods described previously (10) and propagated in DMEM:Ham’s F-12 medium (Life Technologies, Inc.) containing 10% fetal bovine serum and hydrocortisone (100 ng/ml), insulin (2.5 μg/ml), transferrin (2.5 μg/ml), and selenium [2.5 ng/ml (Sigma)]. Cells were used within 10 passages. For all experiments, cells were grown to near confluence before exposure to agents.

Mesothelioma cell lines 11, 23, and 52 were developed in rats after i.p. injection of crocidolite asbestos and have been characterized previously (14). Cells were propagated as described above for RPM cells. Human normal mesothelial and human mesothelioma cells were kindly provided by Drs. Michele Carbone (Loyola University, Chicago, IL), Mauro Tognon (University of Ferrara, Ferrara, Italy), and Luciano Mutti (Laboratory of Clinical Oncology and Toxicology, “S. Maugeri” Foundation for Research and Care, Pavia, Italy).

[(Na₂(Fe³⁺)₂(Fe²⁺)₃Si₈O₂₂(OH)₂], a high iron-containing amphibole fiber associated with the causation of human mesothelioma (Ref. 1; National Institute of Environmental Health Sciences reference sample), was suspended in HBSS (Life Technologies, Inc.) at a concentration of 1 mg/ml, triturated eight times through a 22-gauge needle to obtain a homogeneous suspension, and added directly to the medium of confluent RPM cells at a final concentration of 5 μg/cm² dish for 24 h. Sham control dishes received medium without agents.

Total RNA was prepared using TRIzol reagent (Invitrogen Life Technologies, Inc., Carlsbad, CA) according to the manufacturer’s protocol. After the ethanol precipitation step in the TRIzol extraction procedure, a cleanup was done using Rneasy Total RNA Isolation kit (Qiagen, Valencia, CA).

Microarrays were performed using normal RPM, asbestos-exposed RPM, and three rat mesothelioma cell lines. The RNA target (biotin-labeled RNA fragments) was produced from 5 mg of total RNA collected from the pooling of five different experiments by first synthesizing double-stranded cDNA followed by an in vitro transcription reaction and a fragmentation reaction. A hybridization mixture containing the fragment cRNA, probe array control (Affymetrix, Santa Clara, CA), BSA, and herring sperm DNA was prepared and hybridized to the probe array at 45°C for 16 h. The hybridized probe array was then washed, and bound biotin-labeled cRNA was detected with a streptavidin phycoerythrin conjugate. Subsequent signal amplification was performed with a biotinylated anti-streptavidin antibody. The washing and staining procedures were automated using the Affymetrix fluidics station. Each probe array, Rat Genome U34A (Affymetrix), was scanned twice (Hewlett-Packard GeneArray Scanner), the images were overlaid, and the average intensities of each probe cell were compiled.

Total RNA (1 μg) was reverse transcribed with random primers using the Promega Avian Myeloblastosis Virus Reverse Transcriptase kit (Promega, Madison, WI) according to the recommendations of the manufacturer. To quantify gene expression, we amplified the cDNA by TaqMan real-time Q-PCR using the 7700 Sequence Detector (Perkin-Elmer Applied Biosystems, Foster, CA). Reactions contained 1× TaqMan Universal PCR Master Mix, 900 nm forward and reverse primers, and 200 nm TaqMan probes. Thermal cycling was performed using 40 cycles of 95°C for 15 s and 60°C for 1 min. Original input RNA amounts were calculated with relative standard curves for both the mRNAs of interest and the HPRT control. Duplicate assays were performed with RNA samples isolated from at least 2 independent experiments. The values obtained from cDNAs and HPRT controls provided relative gene expression levels for the gene locus investigated. The primers and probe sequences used are presented in Table 1.

Sequence information on fra-1 mature mRNA was extracted from the expressed sequence tag database.⁴ The open frame region from the cDNA sequence around 100 nucleotides downstream of the start codon of the gene was selected to develop the 21-base duplex siRNA molecule (AAGCGCAGACACAGACAGUCC). The sequence was BLAST-searched (National Center for Biotechnology Information database) against expressed sequence tag libraries to ensure the specificity of the siRNA molecule. siRNA duplexes (Dharmacon Research, Lafayette, CO) were transfected into RPM cells using OligofectAMINE reagent (Invitrogen Life Technologies, Inc.) as recommended by the manufacturer. Transient transfections were carried out on subconfluent RPM cells using the siRNA fra-1 duplex and a siRNA scramble duplex (control; Dharmacon Research). All experiments were performed in duplicate.

EMSAs were used to assess the binding of AP-1 to DNA and the composition of AP-1 complexes. Nuclear extracts were prepared and analyzed as described by Janssen et al. (15). The amount of protein in each sample was determined using the Bio-Rad protein assay (Bio-Rad, Hercules, CA). For supershift assays, nuclear extracts were incubated with an antibody to Fra-1 (Santa Cruz Biotechnology, Santa Cruz, CA) for 15 min at room temperature before the addition of labeled oligonucleotide. Gels were quantitated using a Bio-Rad phosphorimager.

The microarray data were processed using GeneSpring software (Silicon Genetics, Redwood, CA). For all other experiments, we used duplicate or triplicate determinations (N = 2–3) per group, and experiments were performed in duplicate. Results were evaluated by one-way ANOVA using the Student-Newman-Keuls procedure for adjustment of multiple pairwise comparisons between treatment groups. Differences with Ps ≤ 0.05 were considered statistically significant.

As shown in Fig. 1, after microarray analysis using the RG-U34A chip from Affymetrix, subsets of genes were identified that increased or decreased during the development of mesotheliomas or in response to asbestos. The gene expression data (average difference as calculated by Affymetrix algorithms) were normalized against the control sample (RPM cells), and the fold changes were determined using GeneSpring software (Silicon Genetics). Genes exhibiting >1.5-fold changes and present in comparison with control cells were considered altered using the Affymetrix absolute call algorithm. Of the 8799 genes present in the chip, 576 (6.5%) were up-regulated in RPM cells exposed to asbestos. Seventy-eight (0.9%) of those genes were up-regulated after exposure to asbestos but down-regulated in all mesotheliomas. In contrast, 353 (4%) genes were up-regulated in mesotheliomas. Of these, 73 (0.8%) were up-regulated in mesotheliomas and down-regulated in asbestos-exposed cells. Overall, 77 genes (0.9%) were found to be up-regulated in both RPM cells exposed to asbestos and in mesotheliomas, and 242 genes (2.75%) were found to be down-regulated in both RPM cells exposed to asbestos and in mesotheliomas.

Although only 42% of the up-regulated genes in both acute exposure to asbestos and in mesotheliomas could be classified, 53% of those classified are enzymes, and 32% are involved in transport. Of the common down-regulated genes that could be classified (51%), 63% belong to the enzyme ontology (including phosphatases), 32% are signal transducer ligands, and 21% are cellular components. In accordance with our previous studies (10, 16), c-fos expression was increased by asbestos but was absent in mesotheliomas, whereas fra-1 message levels were elevated in both asbestos-exposed RPM cells and mesotheliomas. The absence of c-fos in mesotheliomas suggests that substitution of fra-1 for c-fos in AP-1 complexes occurs during tumorigenesis (13).

Table 2 presents the fold increases in expression of nine selected genes after comparative analyses using real-time Q-PCR and microarray expression analysis. In addition to fra-1, which was dramatically increased, we focused on eight other genes at different levels of expression. The real-time Q-PCR probes developed and presented in Table 1 were designed to capture all isoforms of these genes. Although the algorithms used to obtain the fold changes in both cases are not the same, and the probes developed for real-time Q-PCR are detecting different isoforms of the same gene, the patterns of expression, as shown using this technique and microarrays (Table 2), are comparable. The fact that the genes selected were at different levels of expression (low, medium, and high according to arbitrary range) validates the results found in the microarrays for all genes in the analysis.

Because increases in fra-1 expression were most dramatic by both microarray and TaqMan analyses, and in line with our hypothesis that this early response proto-oncogene may be critical to the expression of intermediate response genes involved in the development of mesotheliomas, we also examined fra-1 expression by real-time Q-PCR in normal human mesothelial cells exposed to asbestos and in human mesotheliomas (Fig. 2,A). As can be seen, fra-1 mRNA levels increased strikingly in response to asbestos and in human mesotheliomas, validating our results in the rat model. Also, an EMSA supershift confirmed the increased level of Fra-1 protein in the AP-1 complex (Fig. 2 B).

We then used the RNA interference technique to silence the fra-1 gene in RPM cells exposed to asbestos and in a mesothelioma cell line to determine whether its expression was critical to induction of selected genes involved in cell signaling, invasiveness, and DNA damage. To verify the specificity of the sifra-1 RNA, we first performed real-time Q-PCR and EMSAs on RPM cells transfected with the sd and sifra-1 before a 24-h exposure to asbestos (Fig. 3). As shown in Fig. 3,A, RPM cells transfected with the sd showed increased expression of fra-1 after exposure to asbestos, whereas fra-1 expression was inhibited in sifra-1-transfected cells. Supershift analyses also showed decreased Fra-1 in AP-1 complexes of sifra-1-infected cells (Fig. 3 B).

In subsequent analyses, we transfected cells with sd or sifra-1 and examined expression of four selected genes in RPM cells alone and after exposure to asbestos (Fig. 4,A) and in mesothelioma 23 (Fig. 4 B). These studies revealed that from the subset of genes mimicking fra-1 patterns of expression, cd44 and c-met expression was strikingly diminished by siRNA gene silencing of fra-1, whereas mRNA levels of hmgI(Y) and gadd45 were unaffected after fra-1 silencing.

Malignant mesotheliomas are of contemporary importance because their incidence is increasing in several countries, and they have most recently been associated with the presence of SV40 T antigen as well as asbestos exposures (17). The diagnostic tools and treatment regimens for these tumors are disappointing, and patients generally survive less than 18 months after initial diagnosis (18). In an effort to dissect critical genes and signaling pathways involved in the development of mesotheliomas, we performed microarray analysis of mRNA expression patterns in normal RPM cells exposed to asbestos and in rat mesotheliomas. Because expression of the early response proto-oncogene, fra-1, was strikingly increased in both rat and human mesothelial cells exposed to asbestos and in mesotheliomas, and because fra-1 expression is critical to morphological transformation of mesothelial cells (13, 19), we hypothesized that its expression would govern the regulation of intermediate response genes linked to phenotypic changes in mesothelial cells. We show here, using a novel RNAi approach, that silencing of fra-1, a prominent component of AP-1 complexes in mesotheliomas (13, 19), results in abrogation of cd44 and c-met expression, genes that are critical to growth and invasiveness of tumors.

Many oncogenic and mitogenic pathways converge at the AP-1 transcription factor, suggesting an important role of this complex in cell transformation or responses to carcinogens and tumor promoters (reviewed in Ref. 20). The dimeric AP-1 complex is comprised of members of the Fos, Jun, and ATF family, and the constituents of this complex, particularly the interaction of binding partners, may govern cell responses. In contrast to other members of the Fos/Jun family, Fra-1 has only recently received attention. Some reports indicate that it lacks a functional transactivation domain (21, 22) and is merely a suppressor of gene transcription. However, recent reports indicate that ERK-dependent transactivation of Fra-1 is critical to mitogenesis and tumor promotion in skin (23) and that both c-Fos and Fra-1 are capable of modulating transcription of target genes (24). These studies support our results here and recently published work showing a functional role of ERK-dependent Fra-1 expression in transformation of mesothelial cells (13).

The fra-1-dependent expression of cd44 is intriguing and has multiple ramifications in the process of mesothelioma development. CD44 is the principal cell surface receptor for the extracellular matrix glycosaminoglycan hyaluronan, which is increased in mesotheliomas (25). The cell surface hyaluronan receptor CD44 is expressed in a variety of normal tissues and tumors. Binding of CD44 to hyaluronan mediates cell attachment and migration (26). An association between overexpression of CD44 or its alternative spliced variants and aggressiveness and metastasis of a variety of human tumors shows the importance of this protein in tumor invasiveness in vitro (27, 28) and in vivo (29, 30). CD44 also has been linked causally to the development of chronic inflammatory responses in lung (31, 32).

In support of an association between fra-1 and cd44, p21ras promotes transcription of the cd44 gene via an AP-1 binding site (33). Moreover, a link between AP-1, CD44 induction, and cell invasion into extracellular matrix has been demonstrated (34). Epidermal growth factor, a cytokine causing increased ERK activity in RPM cells in a manner similar to asbestos (8, 35), up-regulates CD44-dependent astrocytoma invasion in vitro (36). Moreover, the ERK pathway regulates alternative pre-mRNA splicing variants of CD44 (37).

Our observation that fra-1 expression up-regulates c-met also has mechanistic implications in the development of mesotheliomas by asbestos and SV40 T antigens, further supporting the concept that these diverse agents may act through similar cell signaling pathways (38). The c-met proto-oncogene encodes a transmembrane tyrosine kinase receptor (Met) for HGF (or scatter factor), a multifunctional protein involved in tissue repair as well as in cancer and metastasis. Others have found a predominant role of HGF in mesothelioma cell invasion and regulation of matrix metalloproteinases and their inhibitors (39). Met is also overexpressed in mesotheliomas and linked to growth, migration, and invasion of tumors (40, 41). Recent studies have confirmed that HGF is increased in bronchoalveolar and pleural lavage fluids after administration of crocidolite asbestos to rats and causes mitogenesis in mesothelial cells (42). Moreover, SV40 viral replication in human mesothelial cells induces HGF/Met receptor activation via an autocrine loop that causes cell cycle progression into S phase (43). RNA silencing using sifra-1 did not inhibit hmgI(Y) or gadd45 expression induced by asbestos. Both gadd45 and gadd153 were highly up-regulated in both asbestos-exposed RPM cells and mesotheliomas, observations that may relate to the fact that asbestos induces DNA damage and growth arrest in mesothelial cells (44).

HMGI(Y)s are chromatin-associated proteins that are overexpressed in many types of human malignancies (45). HMGI(Y) expression is also induced by tumor promoters in transformation-sensitive cells (46). Furthermore, HMGI(Y)-positive carcinomas have invasive growth patterns, whereas the HMGI(Y)-negative carcinomas are either carcinomas in situ or tumors with minimal invasion (45). This evidence suggests that HMGI(Y) may play a vital role in the oncogenic transformation of cells. Transcription of cd44 by AP-1 proteins is enhanced by HMGI(Y) (47).

Other genes up-regulated in mesothelioma were related to changes in metabolic rates and are commonly expressed during tumorigenesis. For example, tumor cells are known to be highly glycolytic, and increased expression of glycolytic enzymes including hexokinase II, which catalyzes the phosphorylation of glucose, has been reported (48). Glucose metabolism is partially regulated by phosphofructokinase C and is also increased during cell transformation (49). Also, lysophosphatidic acid up-regulates exokinases, requiring both Ca²⁺-independent protein kinase Cs and mitogen-activated protein kinase pathways (50).

The transport of pyruvate and lactate across cellular membranes is an essential process in mammalian cells and is mediated by monocarboxylate transporters that are highly expressed in cancer cells but low in normal cells (51).

Acyl-CoA thioesterase, which cleaves acyl-CoA thioesters to free fatty acids, and CoA may serve to modulate cellular levels of acyl-CoAs to affect various cellular functions, including lipid metabolism (52). LAT1, which transports neutral amino acids, is also highly expressed in tumors, presumably to support the increased protein synthesis needed for cell growth (53). In addition, isovaleryl CoA dehydrogenase, an enzyme involved in the turnover of toxic metabolites and a product of endogenous catabolism, is also expressed in malignancies (54).

Asbestos fibers generate reactive oxygen and nitrogen species and cause inflammation associated with the development of both cancers and fibrosis (55). The up-regulated genes in mesotheliomas that are related to the development of inflammation include those belonging to the kallikrein-kinin system, including the bradykinin B2 receptor, which is activated by bradykinin, enhancing vascular permeability and producing tissue edema (56). Bradykinin has also been linked to allergic airway inflammation and regulation of AP-1-related gene expression (57).

In pulmonary inflammation, increases in nitric oxide production via inducible nitric oxide synthase and availability of the nitric oxide synthase substrate, l-arginine, are required. The increased nitrotyrosine levels in asbestos-exposed lungs (58) may be associated with increases in l-arginine resulting from an increase in cationic amino acid transporter 1 (59). The ability of the known fibrotic agent, transforming growth factor β, to up-regulate l-arginine transport and cause its metabolism to polyamines and l-proline, processes associated with transforming growth factor β-stimulated collagen production, may contribute to remodeling at sites of damage (60). Increased expression of interleukin 6, a cytokine increased by asbestos in macrophages and epithelial cells in vitro (61) and an autocrine growth factor for human cells (62), was also observed in mesotheliomas. Increased levels of expression of this cytokine and its receptor have been observed during chronic inflammatory disease as well as tumorigenesis (63). Other genes increased in our studies were glutathionate-dependent dehydroascorbate reductase, which reduces dehydroascorbic acid to ascorbic acid after it is oxidized during normal aerobic metabolism, and SOD-2 (MnSOD), a mitochondrial protein up-regulated in rat lungs after inhalation of asbestos (64).

A number of protein kinases and proto-oncogenes including c-myc were increased in all samples. These include the serum and glucocorticoid-induced protein kinase, sgk, a multifunctional kinase that can be phosphorylated through a phosphatidylinositol 3′-kinase-dependent signaling pathway (65), and the ras subfamily of GTPases, ralB. Ras proteins can activate at least three downstream signaling cascades mediated by the raf-mitogen-activated kinase (MEK)-ERK kinase family, phosphatidylinositol 3′-kinase, and Ral-specific guanine nucleotide exchange factors. The importance of the interaction between ERK pathways in the production of a malignant phenotypes has been demonstrated previously (66).

In conclusion, our studies are the first to use a robust microarray analysis to demonstrate increases and decreases in gene expression in asbestos-exposed mesothelial cells and mesotheliomas. In support of our findings, a previous study using cDNA microarray methodology has shown increases in fra-1, egfr, and c-myc in asbestos-exposed RPM cells and in rat mesotheliomas (67). The ability to modify gene expression of critical genes involved in metastasis of tumors by silencing of fra-1, an ERK-dependent gene, suggests that the development of novel therapeutic approaches to modify AP-1 and associated signaling pathways may be promising in the treatment of mesothelioma.