Expression of E1AF/PEA3, an Ets-related transcription factor in human non-small-cell lung cancers: Its relevance in cell motility and invasion

Authors


Abstract

Cell invasion and metastasis characterize the malignant potential of non-small-cell lung cancers (NSCLCs). We have previously reported that E1AF, a member of the Ets-related transcription factor family, confers invasive phenotype on breast cancer and oral squamous-cell carcinoma cell lines. In our study, we analyzed the E1AF expression in cell lines and resected tumors of NSCLCs by Northern blot and in situ hybridization analyses and found that 15 of 17 cell lines and 12 of 19 tumors expressed E1AF mRNA while normal lung tissue and concomitant normal cells within tumors did not. To examine the biologic importance of E1AF in NSCLCs, we introduced the E1AF gene into VMRC-LCD and NCI-H226, NSCLC cell lines lacking E1AF expression, and examined cell motility and invasion activities. E1AF-transfected VMRC-LCD cells showed increased cell motility that was 2-fold that of parental and vector-transfected control cells (p < 0.01), and both cell motility and invasion were increased 1.6-fold in NCI-H226 (p < 0.01). Furthermore, hepatocyte growth factor (HGF), which is one of the most effective cell-scattering factors, stimulated the motile and invasive activities in E1AF-transfected VMRC-LCD and NCI-H226 cells but not in their parental or vector-transfected control cells. Ets-1 mRNA expression was found in E1AF-transfected VMRC-LCD cells but not in parental or vector-transfected cells. HGF further induced expression of the Ets-1 and urokinase-type plasminogen activator (uPA) genes specifically in E1AF-transfected cells. These findings suggest that E1AF plays a substantial role in the cell motility and invasion of NSCLCs. © 2001 Wiley-Liss, Inc.

Lung cancer is one of the leading causes of cancer death throughout the world. Non-small-cell lung cancers (NSCLCs) are characterized by local tumor aggressiveness and a marked propensity for dissemination to regional lymph nodes as well as distant metastasis.1 The clinical observations that patients having NSCLCs in comparable stages may run different clinical courses and may respond differently to similar treatments have yet to be fully understood. Clearly, a more sophisticated understanding of the pathogenesis and biology of these tumors could provide useful information.

The Ets-related oncoprotein family has a highly conserved ETS-domain and works as a transcription factor. V-ets of avian leukosis virus E26 was the first one described,2, 3 and more than 30 Ets-related oncogenes have been isolated. They play important roles in the regulation of gene expression during a variety of biological processes, including cell growth, oncogenesis and developmental programs in many organisms.4

E1AF, which binds to the enhancer elements of the adenovirus type 5 E1A gene,5 is a human member of the Ets-related oncoprotein family.6, 7 The E1AF Ets-domain is identical to that of mouse PEA3 (polyoma-virus enhancer activator 3)8 and there was about 94% identity in amino acids between them,6 indicating that E1AF is a human homologue of mouse PEA3. We have previously reported that E1AF can upregulate transcription from different subclasses of matrix metalloproteinase (MMP) genes in transient expression assays,9 that E1AF confers the invasive phenotype in MCF7, a human breast cancer cell line,10 and that E1AF expression is correlated with the transcription of MMPs and the invasive phenotype in oral squamous-cell carcinoma cell lines.11 However, little is known about the importance of Ets-related oncoproteins, including E1AF, in the invasion and metastasis of NSCLCs.12

Previous studies have indicated that transactivation activity of E1AF is regulated by the mitogen-activated protein (MAP)-kinase signaling pathway,13 which is activated by the hepatocyte growth factor (HGF)-Met signals.14, 15 Therefore, the function of E1AF may be regulated by the HGF-Met signals. It has been established that aberrant HGF-Met signaling plays a critical role in the migration and invasion of cancer cells.16, 17 Moreover, increased expression of Met and/or HGF has been demonstrated in a variety of human cancers, including NSCLCs, and is often associated with increased tumor grade and poor prognosis.18–20 However, the mechanisms by which the HGF-Met signals promote invasive and metastatic activities as well as the molecular changes that activate invasion-associated matrix-degrading proteases in cancers are still largely unknown.

In our study, we examined E1AF expression in resected NSCLCs as well as NSCLC cell lines. Furthermore, we investigated the biologic importance of E1AF and its association with the HGF-Met signals in NSCLCs using E1AF-transfected NSCLC cell lines.

MATERIAL AND METHODS

NSCLC cell lines and tumor specimens

Seventeen human NSCLC cell lines were cultured in RPMI 1640 supplemented with 10% FCS and 0.03% glutamine at 37°C in an atmosphere 5% CO2. They included 6 adenocarcinoma (A549, ABC-1, VMRC-LCD, RERF-LC-OK, RERF-LC-MS and PC-3) and 11 squamous-cell carcinoma cell lines (EBC-1, LC-1 sq, VMRC-LCP, LK-2, PC-10, NCI-H157, NCI-H226, NCI-H520, NCI-H1264, NCI-H1373 and NCI-H1703). All of the adenocarcinoma cell lines and 5 squamous-cell carcinoma cell lines (EBC-1, LC-1 sq, VMRC-LCP, LK-2 and PC-10) were obtained from the Health Science Research Resources Bank of Japan (Osaka, Japan), and 6 squamous-cell carcinoma cell lines (NCI-H157, NCI-H226, NCI-H520, NCI-H1264, NCI-H1373 and NCI-H1703) were kindly provided by Dr. H. Oie of the NCI-Navy Medical Oncology Branch, National Cancer Institute (Bethesda, MD).

Tumor specimens from 19 patients with NSCLC were obtained by surgery at the Minami-ichijo Hospital during 1997 and 1998. None of the patients were subjected to either chemotherapy or irradiation before surgery. According to the 1981 WHO classification,21 tumor specimens were histopathologically diagnosed as adenocarcinoma (n = 16), squamous-cell carcinoma (n = 2) and large-cell carcinoma (n = 1). The postsurgical pathologic TNM stage (pTNM) was determined according to the guidelines of the American Joint Committee on Cancer.22 The tumor specimens represented 9 stage I, 3 stage II and 6 stage III tumors, as well as 1 stage IV tumor.

Northern blot analysis

Total cellular RNA was prepared from cells or tissues by guanidine isothiocyanate lysis and subsequent CsCl gradient ultracentrifugation (Beckman, Fullerton, CA).23 For Northern blot analysis, total cellular RNA (10 μg) was electrophoresed through denaturing formaldehyde-containing gels24 and transferred to nylon membranes (Pall BioSupport, Glen Cove, NY). Hybridization of the nylon membranes with [α-32P]dCTP (ICN Biomedicals, Irvine, CA) random-primed E1AF, Ets-1, MMP-1, MMP-3, MMP-9, urokinase-type plasminogen kinase (uPA), uPAR, c-Met or cardiac actin cDNA fragments was performed by standard methods.23 Human cDNA probes were obtained from the following sources: E1AF (0.6 kb XbaI/BamHI fragment of pCMVETS), Ets-1 (1.45 kb HindIII fragment of Ets-1 cDNA; a kind gift from Dr. H. Suzuki, Hokkaido University, Sapporo, Japan), MMP-1 (1.7 kb EcoRI fragment of pUC19MMP-1), MMP-3 (a kind gift from Dr. M. Seiki, Kanazawa University, Kanazawa, Japan), MMP-9 (1.2 kb PstI/EcoRI fragment of pUC19MMP-9), uPA (1.5 kb PstI fragment of pEMBL8uPA; ATCC, Rockville, MD), uPAR (1.0 kb EcoRI fragment of pSG5uPAR; ATCC), cardiac actin (a kind gift from Dr. F. Gunning, Stanford University and Veterans Affairs Medical Center, Palo Alto, CA). Autoradiography was done on Kodak XAR-5 film at −70°C with an intensifying screen.

In situ hybridization

Tissue samples for in situ hybridization were obtained from 19 resected NSCLCs and fixed in 4% (w/v) paraformaldehyde at 4°C for 48 hr after surgical resection, followed by embedding in paraffin. In situ hybridization was performed as described previously.11 In brief, 5 μm-thick paraffin-embedded specimens were cut, mounted on 3-aminopropyl triethoxysilane-coated glass microscope slides (APC-coated glass slides; Matsunami, Kishiwada, Japan) and deparaffinized with xylene. They were then digested in a solution containing 0.1% pepsin in 0.1 N HCl at room temperature for 10 min. After acetylation (0.1 mol/L triethanolamine, 0.25% acetic anhydride), tissue slides were covered with an in situ hybridization mixture and a biotin-labeled RNA probe. The biotin-labeled RNA probe was synthesized by in vitro transcription of the linearized double-stranded DNA template of a 472 bp E1AF fragment (924–1396),11 using biotin-UTP for the sense and antisense strands with T3 and T7 RNA polymerases, respectively. Hybridization was carried out at 50°C for 36 hr. After hybridization, RNAse digestion and high-stringency washing were undertaken. Tissue slides were then stained with a GenPoint system (DAKO JAPAN, Kyoto, Japan), followed by counterstaining with methyl green.

Transfection

VMRC-LCD and NCI-H226 cells were cotransfected with the E1AF-expression vector pCMVE1AF6 (2.5 μg) plus pRSVneo (0.125 μg) or the empty vector pEV3S25 (2.5 μg) plus pRSVneo (0.125 μg) by the standard DNA-calcium phosphate coprecipitation method. Transfected cells were selected for resistance to active Geneticin® (G418; Boehringer Mannheim, Indianapolis, IN) at 0.2 mg/ml. E1AF gene expression was confirmed by Northern blot analysis as described above.

In vitro motility and invasion assays

Transwell® cell culture chambers (pore size 8 μm; Costar, Cambridge, MA) were used for the motility and invasion assays. For the motility assay, cells (1 × 105 for VMRC-LCD and 2 × 104 for NCI-H226) were suspended in serum-free RPMI 1640 with 0.1% BSA (Sigma Chemical, St. Louis, MO) and added to the upper chamber. The reverse side of the upper chamber's filter was coated with human cellular fibronectin (10 μg)(Biomedical Technologies, Stoughton, MA) as a chemoattractant. Serum-free RPMI 1640 with 0.1% BSA with or without 20 ng/ml HGF (Toyobo, Osaka, Japan) (600 μl) was added to the lower chamber. For the invasion assay, cells (2 × 105 for VMRC-LCD and 1 × 105 for NCI-H226) were suspended in serum-free RPMI 1640 with 0.1% BSA and added to the upper chamber. The upper chamber's filter was coated with mouse Matrigel® (10 mg) (Becton Dickinson Labware, Franklin Lakes, NJ), and the reverse side of the filter with human cellular fibronectin (10 μg) as a chemoattractant. Serum-free RPMI 1640 with 0.1% BSA with or without 20 ng/ml HGF (600 μl) was added to the lower chamber. Cells were incubated for 24 hr at 37°C in a CO2 incubator. At the end of the incubation, cells on the upper surface of the filter were completely removed by wiping with a cotton swab. Cells were fixed in phosphate-buffered saline containing 3.7% formaldehyde and stained with Giemsa solution. Cells that moved into or invaded the lower surface of the filter were counted under a light microscope at a magnification of 200×. Each assay was performed in triplicate and repeated 3 times.

RESULTS

Expression of the E1AF gene in NSCLC cell lines and resected NSCLCs

We examined expression of the E1AF gene in 6 lung adenocarcinoma cell lines, 11 squamous-cell carcinoma cell lines and 19 resected NSCLCs. Northern blot analysis indicated that 2.5 kb E1AF mRNA was present in 5 of the 6 adenocarcinoma cell lines (the exception was VMRC-LCD) and in 10 of the 11 squamous-cell carcinoma cell lines (the exception was NCI-H226) but not in normal lung tissue (Fig. 1). Twelve of the 19 NSCLCs, including 11 of 16 adenocarcinomas, 1 of 2 squamous-cell carcinomas and none of 1 large-cell carcinoma, and 8 of 9 stage I, all 3 stage II, 2 of 6 stage III and none of 1 stage IV tumors expressed E1AF mRNA, although normal lung tissue again did not express E1AF mRNA (Fig. 2).

Figure 1.

Northern blot analysis of E1AF gene expression in cultured human NSCLC cell lines. Upper lanes: a 2.5 kb single band of the E1AF mRNA was detected in 10 of the 11 squamous cell carcinoma cell lines (lanes 1–11) and in 5 of the 6 adenocarcinoma cell lines (lanes 12–17). E1AF mRNA was not detected in normal lung tissue. Middle lanes: cardiac actin (2.2 kb) was used as an internal control. Lower lanes: ethidium bromide staining of the agarose gel. LC-1, LC-1 sq; LCP, VMRC-LCP; H157, NCI-H157; H226, NCI-H226; H520, NCI-H520; H1264, NCI-H1264; H1373, NCI-H1373; H1703, NCI-H1703; LCD, VMRC-LCD; OK, RERF-LC-OK; MS, RERF-LC-MS; 28S, 28S ribosomal RNA; 18S, 18S ribosomal RNA.

Figure 2.

Northern blot analysis of E1AF gene expression in 19 resected NSCLCs. Upper lanes: 12 of the 19 NSCLCs expressed E1AF mRNA, but normal lung tissue did not express E1AF mRNA. Middle lanes: cardiac actin (2.2 kb) was used as an internal control. Lower lanes: ethidium bromide staining of the agarose gel. See Figure 1 for abbreviations.

In situ hybridization in resected NSCLC tissues

We performed in situ hybridization analysis for E1AF expression in resected NSCLCs to examine its expression in tumor cells and concomitant normal cells separately, using E1AF antisense RNA as a probe. Cytoplasmic hybridized signals of E1AF were seen in cancer cells but not in concomitant interstitial cells in all 12 NSCLCs examined (Fig. 3a), which showed E1AF expression in Northern blot analysis, while no such signals were observed in either cancer or concomitant interstitial cells in NSCLCs that did not show E1AF expression in Northern blot analysis (data not shown). In addition, in situ hybridization analysis, using E1AF sense RNA as a probe, did not show any signals for the same specimens.

Figure 3.

In situ hybridisation analysis of E1AF transcripts in resected NSCLCs (a,b). Cytoplasmic hybridized signals were observed in cancer cells. However, no such signals were seen in concomitant interstitial cells. Scale bars = 20 μm.

In vitro motility and invasion activities of E1AF-transfectants

We next studied the biologic importance of E1AF expression in NSCLCs by analyzing motile and invasive activities of E1AF transfectants in vitro, using Transwell® cell culture chambers. VMRC-LCD is an adenocarcinoma cell line of the lung not having E1AF expression (Figs. 1 and 4, lane 1). VMRC-LCD cells transfected with E1AF showed a high level of E1AF expression (Fig. 4, lane 3). E1AF-transfected VMRC-LCD cells exhibited increased cell motility compared to both parental and vector-transfected cells (p < 0.01) (Fig. 5a), while cell invasion activity did not differ much between before and after E1AF transfection in this cell line (Fig. 5b). Then we studied motile and invasive activities with 20 ng/ml of HGF, since this VRMC-LCD showed a high level of met mRNA by Northern blot analysis (data not shown), and the HGF-Met signals have been shown to be involved in the cell motility and invasion.16, 17 HGF stimulated neither motile nor invasive activity in either parental or vector-transfected VMRC-LCD cells. In contrast, both motile and invasive activities were increased with HGF treatment in E1AF-transfected VMRC-LCD cells compared to those without HGF treatment (p < 0.01) (Fig. 5a,b). In addition, we analyzed motile and invasive activities of the E1AF transfectant of a squamous-cell carcinoma cell line of the lung, NCI-H226, which lacked E1AF expression (Figs. 1 and 4, lane 4) and had met expression (data not shown). E1AF-transfected NCI-H226 cells showed a high level of E1AF expression (Fig. 4, lane 6) and exhibited increased cell motile and invasive activities with or without HGF treatment compared to both parental and vector-transfected cells (p < 0.01) (Fig. 6a,b). These results indicated that E1AF activated cell motility and invasion in NSCLCs and that E1AF had a substantial role for the activation of cell motility and invasion with or without the HGF-Met signals. In addition, met expression was analyzed by Northern blot analysis in 12 NSCLC cell lines, and no correlation was found between met and E1AF expression; 7 of 10 E1AF-positive cell lines and both E1AF-negative cell lines showed met expression (data not shown).

Figure 4.

Northern blot analysis of the E1AF gene expression in E1AF-transfected NSCLC cell lines. Upper lanes: a 2.5 kb transcript was clearly detected in E1AF-transfected VMRC-LCD (lane 3) and in E1AF-transfected NCI-H226 (lane 6), but no signals were observed in parental VMRC-LCD (lane 1), vector-transfected VMRC-LCD (lane 2), parental NCI-H226 (lane 4), or vector-transfected NCI-H226 (lane 5). Lower lanes: ethidium bromide staining of the agarose gel. LCD-vector, vector-transfected VMRC-LCD; LCD-E1AF/PEA3, E1AF-transfected VMRC-LCD; H226-vector, vector-transfected NCI-H226; H226-E1AF/PEA3, E1AF-transfected NCI-H226. See Figure 1 for other abbreviations.

Figure 5.

In vitro motility and invasion assays with and without HGF treatment in VMRC-LCDs. (a) In vitro motility assay of VMRC-LCDs. (b) In vitro invasion assay of VMRC-LCDs. Cells were incubated without HGF (open bar) or with HGF (closed bar) for 24 hr, and then migrated cell numbers were counted at a magnification of 200×. See Figure 4 for abbreviations.

Figure 6.

In vitro motility and invasion assays with and without HGF treatment in NCI-H226s. (a) In vitro motility assay of NCI-H226s. (b) In vitro invasion assay of NCI-H226s. Cells were incubated without HGF (open bar) or with HGF (closed bar) for 24 hr, and then migrated cell numbers were counted at a magnification of 200×. See Figure 4 for abbreviations.

HGF induces expression of Ets-1 and uPA mRNA in E1AF-transfected VMRC-LCD

To elucidate how E1AF enhances invasiveness, we examined expression of several molecules potentially involved in the invasion and metastasis in E1AF-transfected VMRC-LCD cells. Expression of Ets-1 (6.8 and 2.7 kb mRNAs) was detected even before HGF treatment, and both Ets-1 mRNAs proportionally increased 3 hr after the addition of HGF in E1AF-transfected cells (Fig. 7). But Ets-1 expression was not detectable either before or after HGF treatment in parental or vector-transfected cells. Furthermore, the expression of uPA mRNA was increased 3 hr after the addition of HGF in E1AF-transfected VMRC-LCD cells. In contrast, no such increase of uPA expression was seen in parental or vector-transfected cells. The mRNA encoding for uPAR was expressed at a low level and no increase was seen after the addition of HGF in parental, vector-transfected or E1AF-transfected VMRC-LCDs (data not shown). The mRNAs for MMP-1, MMP-3 and MMP-9 were not detectable either before or after HGF treatment in any of these 3 VMRC-LCDs (data not shown).

Figure 7.

Northern blot analysis of E1AF, Ets-1 and uPA mRNA in VMRC-LCD cells before and after the HGF treatment. Upper lane: total cellular RNA was extracted from VMRC-LCD cells before and 3, 6, 9, 12 or 24 hr after the addition of HGF (20 ng/ml), and expression of E1AF, Ets-1 and uPA was determined by Northern blot analysis. Cardiac actin was used as an internal control. Middle lane: ethidium bromide staining of the agarose gel. Lower lane: densitometric determination of the uPA mRNA level compared to the cardiac actin mRNA level in each sample. See Figures 1 and 4 for abbreviations.

DISCUSSION

In our study, we showed that cell lines and resected tumors of NSCLCs expressed E1AF mRNA frequently, while normal lung tissue and concomitant normal cells in tumors did not express E1AF by Northern blot and in situ hybridization analyses. These data suggest that E1AF may play an important role in NSCLCs. To determine the role of E1AF expression in NSCLCs, we transfected an E1AF expression vector into NSCLC cell lines that did not express E1AF mRNA. In vitro motility assay revealed that motile activity increased significantly in E1AF-transfected NSCLC cells. In the presence of HGF, both motile and invasion activities increased significantly in E1AF-transfected NSCLC cells compared to in the absence of HGF, but no such increase was detected in either parental or vector-transfected cells. Northern blot analysis indicated that E1AF-transfected VMRC-LCD expressed Ets-1 mRNA, which was not expressed in either parental or vector-transfected VMRC-LCDs, and that 3 hr after the addition of HGF, Ets-1 and uPA mRNAs increased in E1AF-transfected VMRC-LCD but again no such increase was seen in either parental or vector-transfected VMRC-LCDs, suggesting that Ets-1 and uPA were involved in the motile and invasive activities of E1AF-expressing NSCLCs through the HGF-Met signaling pathway. We previously showed that Ets-1 but not E1AF activated the promoter of uPA in breast cancer.26 In addition, Fafeur et al.27 reported that HGF stimulated Ets-1 expression, which resulted in uPA expression and the cell dispersion in MDCK epithelial cells. Consistently, it has been shown that HGF stimulation of a variety of cells expressing Met induces uPA expression28, 29 through a pathway involving MAP kinase signaling.14, 15 Taken together, it is indicated that HGF-induced Ets-1 activates the transcription of the uPA gene in the E1AF-transfected VMRC-LCD, in which basal expression level of Ets-1 is upregulated by E1AF. This is the first study that has shown that E1AF can induce Ets-1 expression, although the effect of E1AF expression on the promoter activity of the Ets-1 and uPA genes was not examined in our study. Cross-talks among Ets-related transcription factors, such as between E1AF and Ets-1 as shown in our study, may be important for the motile and invasive activities of cancer cells.

Cell invasion is a major component of the complex multistep process of tumor metastasis. Invasion requires both cell motility and degradation of the surrounding extracellular matrix, the latter of which is mediated by a number of proteolytic enzymes.30 In our study, increased cell invasion by E1AF was thought to be a reflection of enhanced motile activity and of increased amounts and activities of uPA and its downstream proteolytic enzymes other than MMP-1, MMP-3 and MMP-9. It has been shown that a high level of uPA expression is associated with invasion and metastasis.31 However, the genetic changes that activate these invasion-associated matrix-degrading proteases in cancers remain to be determined. Many of the Ets-related oncoproteins have been shown to be transcription factors potentially involved in the activation of genes encoding for such enzymes.4, 32 It has also been shown that the transcriptional regulatory regions of the matrix-degrading protease genes often contain Ets-binding sequences,33, 34 and that the Ets-binding sequence is important for activating transcription of MMP genes.35, 36

E1AF is a transcription factor that binds to the Ets-binding sequences.6, 7 Studies using CAT assays have revealed that E1AF can upregulate promoter activities of the genes encoding for MMP-1, MMP-3 and MMP-9.9 It has been proposed that E1AF confers the invasive phenotype in a human breast cancer cell line,10 that E1AF expression is correlated with the transcription of the genes encoding for MMPs and invasive phenotype in oral squamous-cell carcinoma cell lines11 and that the transfection of antisense E1AF restrains invasion of oral cancer cells by downregulating the MMP genes.37 Moreover, the E1AF gene has been shown to form a fusion gene with the EWS gene by chromosome translocation in sarcoma, suggesting its causal involvement in the development of sarcoma.38 These studies, including our study, suggest that E1AF is one of the proto-oncogenes involved in the invasion and metastasis of several types of cancer, including NSCLCs.

In conclusion, E1AF is frequently expressed in resected tumors and cell lines of NSCLCs and is involved in their cell motility and invasion, suggesting that E1AF plays a substantial role in the invasion and metastasis of NSCLCs. These data provide a potential importance of E1AF in NSCLCs, including as a marker for the stratification of patients and a target of therapy.

Ancillary