|Home | About | Journals | Submit | Contact Us | Français|
Caveolin-1 protein has been called a ‘conditional tumor suppressor’ because it can either suppress or enhance tumor progression depending on cellular context. Caveolin-1 levels are dynamic in non-small-cell lung cancer, with increased levels in metastatic tumor cells. We have shown previously that transactivation of an erythroblastosis virus-transforming sequence (ETS) cis-element enhances caveolin-1 expression in a murine lung epithelial cell line. Based on high sequence homology between the murine and human caveolin-1 promoters, we proposed that ETS proteins might regulate caveolin-1 expression in human lung tumorigenesis. We confirm that caveolin-1 is not detected in well-differentiated primary lung tumors. Polyoma virus enhancer activator 3 (PEA3), a pro-metastatic ETS protein in breast cancer, is expressed at low levels in well-differentiated tumors and high levels in poorly differentiated tumors. Conversely, Net, a known ETS repressor, is expressed at high levels in the nucleus of well-differentiated primary tumor cells. In tumor cells in metastatic lymph node sites, caveolin-1 and PEA3 are highly expressed, whereas Net is now expressed in the cytoplasm. We studied transcriptional regulation of caveolin-1 in two human lung cancer cell lines, Calu-1 (high caveolin-1 expressing) and NCI-H23 (low caveolin-1 expressing). Chromatin immunoprecipitation-binding assays and small interfering RNA experiments show that PEA3 is a transcriptional activator in Calu-1 cells and that Net is a transcriptional repressor in NCI-H23 cells. These results suggest that Net may suppress caveolin-1 transcription in primary lung tumors and that PEA3 may activate caveolin-1 transcription in metastatic lymph nodes.
Caveolin-1 protein is essential for the formation of caveolae, which are plasma membrane invaginations that sequester proteins, enzymes and signaling molecules such as epidermal growth factor receptor and Src family kinases, and generally maintain them in an inactive form (1). Caveolin-1 levels are dynamic in tumorigenesis, and it is probably that changes in caveolin-1 protein levels affect tumor progression by influencing cell signaling. In lung adenocarcinomas, virtually all metastatic lymph nodes have increased caveolin-1 expression in tumor cells by immunohistochemistry. Conversely, in most primary lung tumors, caveolin-1 expression is exceedingly low (2–5). Little is known about the regulation of caveolin-1 transcription in lung tumorigenesis, whether regulation is dependent on cellular context and which environmental factors influence expression.
Evidence suggests that caveolin-1 has a causal role in tumorigenesis and acts as a ‘conditional tumor suppressor’ that can prevent tumor development and promote metastasis depending on cellular context (6). The principle of a protein being either tumor suppressing or promoting depending on context has been demonstrated for other proteins including transforming growth factor beta (1). In support of the tumor-suppressing function of caveolin-1, overexpressing caveolin-1 in murine fibroblast cell lines blocks cell cycle progression in G0/G1 (7). Caveolin-1−/− mice are more susceptible to skin tumors upon carcinogenic exposure and caveolin-1−/− mice crossed with inducible breast cancer mice exhibit early-onset tumor formation and increased metastases (8,9). Caveolin-1 has been mapped to a region of chromosome 7q31.1 often deleted in cancers (10).
Known mechanisms for loss of tumor suppressor function of caveolin-1 are tyrosine phosphorylation at the N-terminus, which causes recruitment of proteins involved in anchorage-independent growth; serine phosphorylation at Ser80, which converts caveolin-1 to a secreted protein and a dominant-negative P132L mutation in breast cancer, which leads to a misfolded caveolin-1 protein (reviewed in ref. 1). In addition, E-cadherin expression is lost as tumors progress, and caveolin-1 requires E-cadherin to function as a tumor suppressor (11).
In several malignancies, caveolin-1 promotes the malignant phenotype. Increased caveolin-1 expression has been correlated with metastasis in breast, prostate and colon carcinomas (1). In lung, there is no caveolin-1 expression in low-invasive lung adenocarcinoma cell lines, but there is abundant expression in highly invasive cell lines. Induced caveolin-1 gene expression in a low-invasive lung adenocarcinoma cell line increases cell migration and invasiveness (4). In prostate cancer, crossing caveolin-1−/− mice with TRAMP mice that develop spontaneous prostate tumors has shown that loss of caveolin-1 impedes progression to highly invasive and metastatic disease (12).
We previously identified a cis-element, containing the erythroblastosis virus-transforming sequence (ETS) DNA-binding sequence 5′-GGAA/T3-′, in the proximal mouse caveolin-1 promoter. Activation of this element by the ETS protein ERM (ETS-related molecule PEA3) strongly enhances caveolin-1 gene expression in a mouse lung epithelial cell line (13,14). In Ewing's sarcoma, caveolin-1 is a major target of the ETS fusion protein EWS-FLI-1 (Ewing sarcoma gene-friend leukemia 1). Knockdown of caveolin-1 expression downregulates the malignant phenotype of Ewing's sarcoma cell lines. Loss of caveolin-1 expression reduces the growth of Ewing sarcoma gene cell-derived tumors in nude mice xenografts (15).
We hypothesized that the caveolin-1 promoter is differentially regulated, probably by ETS proteins, in primary lung tumors and in metastatic lymph node sites, given that caveolin-1 expression levels differ markedly in these cell populations. We identified Net as a putative ETS transcriptional repressor responsible for suppressing caveolin-1 in primary lung tumors and Polyoma virus enhancer activator 3 (PEA3) as a candidate transcriptional activator. Net expression in normal adult human lung is not well characterized, but immunohistochemistry in normal rat lung shows Net to have an inverse correlation with caveolin-1, being highly expressed in type II cells and minimally expressed in type I cells (data not shown). In cervical cancer, loss of Net as a transcriptional repressor has been implicated in cellular transformation (16).
PEA3 has little to no expression in normal lung but is expressed in distal lung epithelium during lung development and in human lung tumor cells (17–19). PEA3 is a member of the ETS subfamily composed of PEA3, ER81, and ERM. These members can transactivate target genes interchangeably via the PEA3 consensus element (20). PEA3 factors mediate pro-metastatic signatures and have been implicated in proliferation, differentiation and tumorigenesis (21,22). Known target genes of PEA3 include osteopontin, matrilysin, urokinase plasminogen activator (uPA), matrix metalloproteinase-1 and -9 and cyclooxygenase-2 (21,23). It has recently been reported that PEA3 regulates a cigarette smoke-responsive region in the matrix metalloproteinase-1 promoter (24).
Herein, we correlate expression patterns of caveolin-1 and the ETS transcription factors Net and PEA3 in human lung cancer specimens by immunohistochemistry. Our findings comparing primary human lung cancer specimens with metastatic lymph node specimens show increased PEA3 and caveolin-1 expression in metastatic cells. The expression pattern of Net, a known ETS repressor, on the other hand, changes from nuclear staining in primary tumors to cytoplasmic staining in advanced and metastatic sites. These findings suggest that in metastatic cells, Net resides in the cytoplasm in an inactive form. Chromatin immunoprecipitation (ChIP) assays, mutation analysis and small interfering RNA (siRNA) experiments in NCI-H23 (low-caveolin-1 expressing) and Calu-1 (high caveolin-1 expressing) lung cancer cell lines show that Net is a transcriptional repressor of caveolin-1 in NCI-H23 cells and PEA3 is a transcriptional activator in Calu-1 cells. Subsequent ChIP assays and siRNA studies in A549 and NCI-H1299 cells confirm the role of ETS proteins in the transcriptional regulation of caveolin-1. Since caveolin-1 is thought to promote cell migration and invasiveness, understanding the molecular mechanisms regulating its dynamic expression levels has important implications for preventing and treating metastatic non-small-cell lung cancer (NSCLC).
Archived paraformaldehyde-fixed, paraffin-embedded human lung adenocarcinomas were studied. A pathologist reviewed hematoxylin and eosin-stained sections and assigned a tumor grade of well, moderate or poorly differentiated (25). Thyroid transcription factor immunohistochemistry confirmed lung origin of specimens (26). Antigen retrieval for caveolin-1 is 0.05% citraconic anhydride (pH 7.4, 98°C, 45 min) and for PEA3 and Net is Vector unmasking buffer (microwave low power, 15 min) (27,28). Sections were incubated with primary antibody [mouse anti-caveolin-1 antibody (610406, BD Biosciences, San Jose, CA), mouse monoclonal anti-PEA3 antibody (sc-113, Santa Cruz Biotechnology, Santa Cruz, CA) or goat polyclonal anti-Net antibody (sc-17860, Santa Cruz Biotechnology)] in phosphate-buffered saline (4°C, 16 h). Antibody binding was detected using Vectastain Elite ABC kit with diaminobenzidine as chromagenic substrate. Control slides lacking primary antibody were included in all procedures. Sections were counterstained with methyl green, hematoxylin or left unstained and photographed in a Leitz Aristopan microscope using ImagePro software. Photographs shown are representative from n=5 samples.
Human NSCLC cell lines were studied based on messenger RNA (mRNA) and protein analyses showing that Calu-1 and NCI-H1299 cells express high caveolin-1 levels, A549 cells express moderate caveolin-1 levels and NCI-H23 cells express low caveolin-1 levels (29,30). Calu-1 cells are derived from a pleural lung squamous carcinoma metastatic site. NCI-H1299 cells are derived from a metastatic lymph node. NCI-H23 and A549 cells are derived from primary lung carcinomas.
Calu-1 cells were maintained in McCoy's 5A (modified) medium (Invitrogen, Carlsbad, CA); NCI-H1299 and NCI-H23 cells were maintained in RPMI 1640 medium (Invitrogen) with 10 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid and A549 cells were maintained in Dulbecco’s modified Eagle’s medium (Invitrogen). All cells were maintained in media containing 10% fetal bovine serum, 2 mM glutamine, 100 U/ml penicillin G and 100 μg/ml streptomycin sulfate, were incubated at 37°C in 5% CO2 and harvested at 80% confluence for experiments.
We amplified the proximal 1.0 kb of the published human caveolin-1 promoter sequence (GenBank accession number AJ133269) using primers designed with adapters for KpnI and SacI. We characterized two ETS sites, the −190 ETS site (described previously) (15) and the −966 ETS site (identified by Transcription Element Search System). Two-step polymerase chain reaction (PCR) was performed with Advantage-HF 2 PCR kit (Clontech, Mountain View, CA) [94°C, 30 s (94°C, 30 s; 68°C, 4 min; five cycles) and (94°C, 15 s; 65°C, 4 min; 30 cycles) 68°C, 10 min] using the CTB-11K1 BAC clone (chromosome 7) as template (Invitrogen). DNA was purified using QIAquick PCR purification kit (Qiagen) and digested with KpnI and SacI. Fragments were run on a 1% agarose gel, purified with QIAquick Gel Extraction kit (Qiagen), ligated with pGL3-basic vector (Promega, Madison, WI) and sequenced. In the −477 to −557 region, which is highly guanine rich, all clones contained at least one base pair difference from the published sequence. All constructs contain an A substituted for a G at position −490.
Similarly, constructs containing wild-type (−190WT and −966WT) and mutated ETS sites (−190MUT and −966MUT) were created by PCR. Each mutated construct contained a one base pair substitution in the core ETS site. The same mutation at the −190 ETS site decreased caveolin-1 transcription in Ewing's sarcoma cell lines (15). After sequence verification, MatInspector confirmed that no other known cis-elements were created in the mutated constructs.
Oligonucleotides are as follows: −190WT forward—5′-CAGGGTACCGCGCAGCACACGTCCGGGCCAA-3′, −190MUT forward—5′-CAGGGTACCGCGCAGCACACGTACGGGCCAA-3′, −966WT forward—5′-CAGGGTACCGTCAAATCTTTCCTCACAGCC-3′, −966MUT forward—5′-CAGGGTACCGTCAAATCTTCCCTCACAGCC-3′ and reverse—5′-GACTGAGCTCGGGCTGTGCTTTAAGGGAAC-3′ (nucleotide differences in bold underlined alphabets).
Total RNA was isolated from cell lines with TRIzol (Invitrogen) and treated for DNA contamination using DNA free (Ambion, Austin, TX). A total of 500 ng RNA was reverse transcribed using TaqMan reagents (Applied Biosystems, Foster City, CA) (25°C, 10 min; 37°C, 60 min and 95°C, 5 min).
Caveolin-1, PEA3, Net and Ets1 mRNA expression were analyzed in cell lines by quantitative real-time (QRT) reverse transcription (RT)–PCR using the ABI Prism 7000 sequence detector (Applied Biosystems). RTs were diluted 1:25. Primers and probe sequences are as follows: Caveolin-1: forward—5′-CTAATCCAAGCATCCCTTTGCC-3′, reverse—5′-TTTATTACTGCCTCCTCCCCCA-3′; PEA3: forward—5′-CCCTACCAACACCAGCTGTC-3′, reverse—5′-GAGAAGCCCTCTGTGTGGAG-3′ (18); Net: forward—5′-TCCACTGCTCTCCAGCATAC-3′, reverse—5′-AATTGTGGCCAGACGTCATC-3′(18) and β-actin: forward—5′-CCCTGAAGTACCCCATCGAG-3′, reverse—5′-CAGATTTTCTCCATGTCGTCCC-3′, probe 5′-ACGGCATCGTCACCA-3′. For Ets1, TaqMan gene expression assay Hs00901425_m1 (Applied Biosystems) was used. Reactions were performed in 50 μl and amplified (95°C, 10 min; 40 cycles: 95°C × 15 s and 60°C × 1 min) using SYBR Green or TaqMan PCR Master Mix (Applied Biosystems). The relative amounts of mRNA for caveolin-1, PEA3, Net and Ets1 were determined using calibration curves from human adult total lung (caveolin-1, Net and Ets1) and Calu-1 cells (PEA3) and normalized to β-actin. All experiments were three different samples performed in duplicate. Data were analyzed by Student's t-test with P < 0.05 considered significant.
Cell monolayers were trypsinized, washed in phosphate-buffered saline, centrifuged, resuspended in lysis buffer with protease inhibitors and incubated with rotation (4°C, 60 min) as described (14). Lysate was centrifuged (10 min, 13000 r.p.m., 4°C). Supernatant (20–50 μg protein) was electrophoresed on a 12% polyacrylamide gel and transferred to polyvinylidene difluoride membranes. For caveolin-1, polyvinylidene difluoride membranes were blocked in 1X Tris-buffered saline Tween-20 containing 5% dry milk (1 h, RT), exposed overnight at 4°C to mouse anti-human caveolin-1 antibody (1:1000) followed by anti-mouse secondary antibody (1:10000, 1 h, RT). For PEA3, mouse anti-PEA3 antibody (1:1000) and anti-mouse secondary antibody (1:20000) were used. For Ets1, rabbit polyclonal anti-Ets1 antibody (1:1000) (sc-350, Santa Cruz Biotechnology) and anti-rabbit secondary antibody (1:20000) were used. Immunoblots were probed for β-actin to control for equal loading. Binding of labeled horseradish peroxidase-secondary antibodies was detected with Super-Signal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL). All experiments were performed in triplicate. Densitometry was performed using the Luminescent Image Analyzer (LAS-4000, Fujifilm, Valhalla, NY). Data were analyzed using Student's t-test with P values <0.05 considered significant.
Cells were fixed with 1% formaldehyde, incubated (37°C, 15 min), washed with phosphate-buffered saline, resuspended in lysis buffer [1% sodium dodecyl sulfate (SDS), 10 mM ethylenediaminetetraacetic acid (EDTA), 50 mM Tris pH 8, 1 mM phenylmethylsulfonyl fluoride, 1 mM pepstatin A and 1 mM aprotinin] and sonicated on ice to 500–1000 base pair fragments with a Fisher Scientific sonicator (Power 5, five cycles of 5 min, 25 s on, 5 s off). Lysate was centrifuged (RT, 4000 r.p.m., 5 min). Supernatant was divided into aliquots. One aliquot was stored as input DNA. Three micrograms of anti-PEA3, anti-Net, anti-Ets1 or non-specific IgG (mouse IgG for PEA3, goat IgG for Net and rabbit IgG for Ets1) antibodies were added to each of the other aliquots. Dilution buffer (0.01% SDS, 1% Triton X-100, 2 mM EDTA pH 8, 20 mM Tris–HCl pH 8 and 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride and 1 mM Aprotinin) was added and samples were incubated (4°C, overnight) with rotation.
AG beads (Santa Cruz Biotechnology) pretreated with 9:1 dilution: lysis buffer, bovine serum albumin 100 μg/ml and salmon sperm 500 μg/ml were added to samples. Samples were rotated (4°C, 2 h) and centrifuged (4000 r.p.m., 2 min). Pellets were washed in wash buffer (1% Triton X-100, 0.1% SDS, 150 mM NaCl, 2 mM EDTA pH 8, 20 mM Tris–HCl pH 8, 1 mM phenylmethylsulfonyl fluoride, 1 mM aprotinin) and then with final wash buffer (1% Triton X-100, 0.1% SDS, 500 mM NaCl, 2 mM EDTA pH 8, 20 mM Tris–HCl pH 8). Immune complexes were eluted with elution buffer (1% SDS, 100 mM NaHCO3), incubated with rotation (RT, 15 min) and centrifuged (4000 r.p.m., 2 min). Proteinase K (500 μg/ml) and RNase A (500 μg/ml) were added to supernatants and input DNA and then incubated (37°C, 30 min). Cross-links were reversed (65°C, overnight) and DNA was purified.
DNA fragments were analyzed by PCR for caveolin-1 fragments spanning the −190 ETS site (234 bp amplicon), −966 ETS site (175 bp amplicon) and for β-actin (171 bp amplicon, control for non-specific binding). Primers for caveolin-1 −190 ETS site were previously published (15). Primers for caveolin-1 −966 ETS site: 5′-CAGGAACAGACAAAATACTTTAATCG-3′ and 5′-CCATATTTGCAAAATACACAAAATGT-3′; primers for β-actin: 5′-CCAAAACTCTCCCTCCTCCT-3′ and 5′-CTCGAGCCATAAAAGGCAAC-3′.
Caveolin-1 promoter-luciferase [wild-type (−190WT; −966WT) or mutated ETS sites (−190MUT; −966MUT)] and Renilla luciferase control plasmids were transiently cotransfected into cell lines using GeneJammer transfection reagent (Stratagene, La Jolla, CA) as described (13). Optimal conditions for NCI-H23 cells were 2 × 105 cells/35 mm dish, 1 μg DNA constructs and 6 μl transfection reagent. Optimal conditions for Calu-1 cells were 3 × 105 cells/35 mm dish, 2 μg DNA constructs and 12 μl transfection reagent.
Cells incubated in standard growth conditions for 48 h were harvested and analyzed for firefly and Renilla luciferase activity with the Dual-Luciferase Reporter Assay System (Promega). Luminescence was detected in a Berthold Lumat LB 9501 luminometer (Berthold, Nashua, NH). Four experiments were performed, each in duplicate. Firefly luciferase activity was normalized to Renilla luciferase activity. For each experiment, fold difference in normalized luciferase activity between mutated and wild-type constructs (with wild-type designated as 1) was calculated. Data are expressed as the mean of four experiments ±SE and analyzed by Student's t-test with P <0.05 considered significant.
ON-TARGETplus SMARTpool siRNA mixtures targeting PEA3 (L-004207-00), Net (L-010320-00), ETS1 (L-003887-00) and non-targeting control siRNA (D-001810-10) were obtained (Dharmacon, Lafayette, CO). Cells (2–3 × 105/35 mm dish) were treated with 50 nM siRNA mixture targeting PEA3 (Calu-1), Net (NCI-H23) or Ets-1 (A549, NCI-H1299) using DharmaFECT General Transcription Protocol. Controls were cells transfected with non-targeting control siRNA. siRNA-treated cells were cultured in standard conditions for 48–72 h. RNA was isolated, DNase treated, reverse transcribed and analyzed by QRT RT–PCR. Protein was isolated, electrophoresed and analyzed by western blot. Data (n=3) are expressed as the average fold difference between targeting siRNA compared with non-targeting control siRNA (with non-targeting control siRNA designated as 1). Herein, Calu-1 is referred as Calu-1 (cav-high) and NCI-H23 as NCI-H23 (cav-low).
Immunohistochemistry for caveolin-1 was performed on primary tumor specimens (n=15; 5 well-, 5 moderate- and 5 poorly differentiated adenocarcinomas). Consistent with the known expression pattern in normal adult lung, areas of adjacent normal lung express caveolin-1 in blood vessels and alveolar type I (but not type II) cells, serving as technical controls within each section. Well- and moderately differentiated primary lung adenocarcinomas have little or no caveolin-1 expression in tumor cells (Figure 1a). Poorly differentiated primary tumors have little or no caveolin-1 expression in 4/5 specimens (Figure 1a) and 1/5 specimen shows high expression in tumor cells. All metastatic lymph nodes (n=5) express caveolin-1 in epithelial tumor cells (Figure 1a–b). Adjacent lymphocytes do not express detectable levels of caveolin-1. As expected, blood vessels within lymph nodes express high levels of caveolin-1 (Figure 1b).
Our findings are consistent with previous immunohistochemical studies of primary and metastatic lung adenocarcinomas (2,4,5). In an immunohistochemical study on 95 human lung adenocarcinomas, in just 4% of cases did the percentage of caveolin-1-positive tumor cells in the primary tumor exceed 30%, and all four of these patients had tumors that were metastatic to lymph nodes. A total of 34/35 metastatic lymph nodes had high caveolin-1 expression in this study (4).
We characterized the expression pattern of PEA3 in primary tumors and metastatic lymph nodes. We found that caveolin-1 and PEA3 protein expression positively correlate. PEA3 expression is minimal in well-differentiated tumors and is increasingly positive in the nuclei of tumor cells in moderate- and poorly differentiated tumors (Figure 1a–b). PEA3 is expressed in the nucleus of metastatic lymph node tumor cells.
Net, on the other hand, is highly expressed in the nucleus of primary tumors (Figure 1a). While Net staining is mostly nuclear in well-differentiated and moderately differentiated primary tumors, it is mostly cytoplasmic in poorly differentiated tumors. High-power view of metastatic lymph nodes shows positive Net staining in cytoplasm and not in nuclei of tumor cells (Figure 1b).
QRT–PCR and western blots confirm that Calu-1 (cav-high) and NCI-H23 (cav-low) cells differ markedly in caveolin-1 mRNA and protein levels (Figure 2). In Calu-1 (cav-high) cells, PEA3, but not Net, binds at the both −190 and −966 ETS sites (Figure 3).
Transient transfections of wild-type and mutated constructs into Calu-1 (cav-high) cells show no difference in luciferase activity between wild-type and mutated −966 ETS site constructs (Figure 3). Mutation of the −190 ETS site, however, results in a 50% decrease in luciferase activity, indicating that this ETS site transactivates the caveolin-1 promoter.
We transfected either an siRNA mixture targeting PEA3 or a non-targeting siRNA control into Calu-1 (cav-high) cells. We confirm effective knockdown of PEA3 mRNA and protein by QRT–PCR and western blot (Figure 4). Silencing of PEA3 results in a statistically significant decrease in caveolin-1 mRNA and protein levels, further supporting the role of PEA3 as a direct transcriptional activator of caveolin-1 in lung cancer cells.
In NCI-H23 (cav-low) cells, neither PEA3 nor Net bind at the −190 ETS site, whereas both Net and PEA3 bind at the −966 ETS site (Figure 5). Net does not bind the caveolin-1 promoter at these sites in Calu-1 (cav-high) cells. This preferential binding of Net is unlikely due to differences in transcription factor abundance since both cell lines express Net and PEA3 mRNA at similar levels by QRT–PCR (data not shown). The cellular distribution (cytoplasmic versus nuclear) of Net and PEA3 in NCI-H23 (cav-low) and Calu-1 (cav-high) cells is not known and may be contributing to the differential binding patterns to the caveolin-1 promoter in these cell lines.
Transient transfections of wild-type and mutated constructs into NCI-H23 (cav-low) cells show no difference in luciferase activity between wild-type and mutated −190 and −966 ETS sites constructs (Figure 5). Since neither Net nor PEA3 bind the −190 ETS site, mutation of this site did not result in a difference in luciferase activity, as expected. Both Net and PEA3 bind the −966 ETS site, and it is possible that mutation of this site allowed neither the repressor nor the activator to bind, thus resulting in no difference in luciferase activity.
We transfected either an siRNA mixture targeting Net or a non-targeting siRNA control into NCI-H23 (cav-low) cells. We confirm effective knockdown of Net mRNA by QRT–PCR (Figure 6). Silencing of Net results in a statistically significant increase in caveolin-1 mRNA and protein levels. Taken together, the ChIP assays and siRNA data confirm that Net is a transcriptional repressor of caveolin-1 in NCI-H23 (cav-low) cells.
ChIP assays were performed in A549 and NCI-H1299 cells to confirm ETS transcription factor binding to the caveolin-1 promoter. In both cell lines, neither PEA3 nor Net binds at the −190 ETS site or the −966 ETS site. Since Ets1 has been shown to be a pro-metastatic ETS transcription factor, we performed ChIP assays to determine whether Ets1 binds to the caveolin-1 promoter. Ets1 binds at the −190 ETS site, but not the −966 ETS site, in both A549 and NCI-H1299 cells, thus confirming the −190 site as an important ETS-transactivating site. Supplementary Figure 1 (available at Carcinogenesis Online) is a representative gel from NCI-H1299 cells.
We transfected either an siRNA mixture targeting Ets1 or a non-targeting siRNA control into A549 and NCI-H1299 cells. We confirm effective knockdown of Ets1 mRNA by QRT–PCR. Silencing Ets1 results in a minimal decrease in caveolin-1 mRNA and protein levels in both A549 and NCI-H1299 cells. Representative graphs and immunoblots are from NCI-H1299 cells (Supplementary Figure 1 is available at Carcinogenesis Online).
In lung tumorigenesis, there is clear evidence that caveolin-1 expression is altered. In normal adult lung, caveolin-1 is expressed in alveolar type I epithelial cells, endothelial cells and fibroblasts. Consistent with the published literature demonstrating dynamic changes in caveolin-1 expression, we confirm by immunohistochemistry that most primary lung adenocarcinomas do not express detectable caveolin-1 protein but that virtually all metastatic lymph nodes have high caveolin-1 expression in tumor cells.
Based on our previous findings that the ETS protein ERM strongly enhances caveolin-1 gene expression in a mouse lung alveolar epithelial cell line, we hypothesized that caveolin-1 is an ETS target gene in lung tumorigenesis. Identifying the ETS transcription factors regulating caveolin-1 expression is complex. In humans, there are 27 ETS family members with multiple members expressed in any individual cell (18). Many different ETS proteins can bind the same ETS cis-element, and ETS factors are able to repress or activate transcription depending on promoter context and cell type (20). Genes such as uPA, upregulation of which confers a pro-metastatic phenotype in breast cancer, can be regulated by more than one specific ETS factor. Binding of ETS proteins is facilitated by binding of other transactivating factors to cis-elements in proximity to ETS sites (22).
For our studies, we selected PEA3 as a candidate activator of caveolin-1 transcription in metastastic lung tumors because it has been linked to oncogenesis in many carcinomas, including lung (31). Introducing the PEA3 gene into NSCLC cell lines that lack exogenous PEA3 expression causes increased motility and invasiveness by activating the Rho/ROCK pathway (18,19,32). Induced expression of PEA3 induces epithelial/mesenchymal transition of lung epithelial cells (33).
In primary lung tumors where caveolin-1 expression is low, we hypothesized that the potent transcriptional inhibitor Net represses caveolin-1 transcription by binding at an ETS cis-element. Although promoter methylation is known to play a role in the silencing of caveolin-1 in breast, ovarian, prostate and colorectal cancer (34–37) and a dominant-negative P132L mutation is known to occur in breast cancer (38), these gene silencing mechanisms are not known to play a role in lung cancer.
Little is known about Net in normal lung or in lung tumorigenesis. During development, Net is expressed at sites of vasculogenesis and angiogenesis. It is expressed in pulmonary vasculature and to a lower level throughout the lung parenchyma during mouse development (39). Net is expressed in adult human lung by QRT–PCR (18), but the specific expressing cells are not known. Interestingly, Net can switch to a transcriptional activator upon RAS (rat sarcoma)/ERK (extracellular signal-regulated kinase) activation and phosphorylation by mitogen-activated protein kinases (40,41). Phosphorylated Net has been shown to positively regulate vascular endothelial growth factor expression and promoter activity in NIH-3T3 cells (42). Studies of human tumors (Kaposi's sarcoma, prostate cancer and head and neck cancer) show phosphorylated Net to be highly expressed in tumor cells but not in normal surrounding tissue (42).
We correlated the expression pattern of caveolin-1 with PEA3 and Net by immunohistochemistry in human lung specimens. We found that caveolin-1 protein expression positively correlates with PEA3 expression. Similar to caveolin-1, PEA3 staining is negative in well-differentiated tumors and highly expressed in metastatic tumor cells. Net, a known transcriptional repressor, is highly expressed in primary tumor cells, where caveolin-1 is not expressed. While Net staining is mostly nuclear in well-differentiated primary tumors, it is mostly cytoplasmic, and therefore likely inactive, in poorly differentiated tumors and metastatic tumor cells.
We chose two representative human lung cancer [Calu-1 (cav-high) and NCI-H23 (cav-low)] cell lines as models to investigate the transcriptional regulation of caveolin-1 by PEA3 and Net. ChIP assays were performed to characterize Net and PEA3 binding to two ETS cis-elements in the human proximal caveolin-1 promoter in both cell lines. We show that PEA3 binds the caveolin-1 promoter at both ETS sites (−190 and −966) in Calu-1 (cav-high) cells. Transient transfections show a 50% decrease in luciferase activity in the mutated −190 ETS site construct compared with the wild-type construct in Calu-1 (cav-high) cells. There is no difference in luciferase activity between wild-type and mutated −966 ETS site constructs. Interestingly, the same mutation of the −190 ETS site decreases caveolin-1 transcription in Ewing's sarcoma cells. Cotransfection of the Ewing sarcoma gene-friend leukemia 1 fusion protein with either wild-type or mutated −190 ETS site constructs shows that the fusion protein only transactivates the wild-type construct (15). Supporting data from our siRNA experiments show that silencing PEA3 in Calu-1 (cav-high) cells causes a decrease in caveolin-1 mRNA and protein expression. Thus, PEA3 binds the caveolin-1 promoter at an important ETS-transactivating site and activates caveolin-1 transcription in Calu-1 (cav-high) cells.
In NCI-H23 (cav-low) cells, both Net and PEA3 bind at the −966 ETS site. The mechanism by which ETS family members compete for occupancy of the same promoter is not known (22). Although ETS proteins generally bind as monomers, co-operative interactions including homo-dimerization occur (20,43). Silencing Net in NCI-H23 (cav-low) cells increases caveolin-1 mRNA and protein expression. We show by ChIP assays and siRNA experiments that Net binds to the promoter and represses caveolin-1 in NCI-H23 (cav-low) cells.
It is probably that other ETS proteins, cis-elements and transcription factors are involved in caveolin-1 regulation in lung tumorigenesis. The caveolin-1 gene is regulated by Sp1-, E2F/DP-1-, p53- and steroid response element-specific enhancers in skin fibroblasts (44,45). The tumor suppressor gene adenomatous polyposis coli, which is lost in 85% of sporadic colon cancers, induces caveolin-1 transcription via FOXO1a (46).
Observations suggest that multiple ETS factors act as an ‘ETS regulatory network’ to regulate the pathways involved in tumorigenesis. In breast cancer, ETS proteins such as ETS-1, ETS-2 and PEA3 are pro-metastatic, whereas the ETS proteins PDEF (prostate-derived ETS factor), Ese(epithelial specific ETS3)-2, and Ese-3 are antimetastatic (22). Ets-1 regulates hypoxia-inducible genes that facilitate tumor cell survival (47). Ets-1 appears to be a predictor of poor prognosis after surgical resection in lung adenocarcinoma patients (48).
While we show that silencing PEA3 and Net in human lung cancer cell lines statistically alter caveolin-1 mRNA and protein levels, these effects are not complete. We believe that similar to breast cancer, there are ETS regulatory networks, that transcriptionally regulate caveolin-1 in lung tumorigenesis. To explore whether redundant ETS proteins regulate caveolin-1 expression, we performed ChIP assays and siRNA experiments in A549 and NCI-H1299 NSCLC cells. We show that Ets1, but not PEA3 or Net, bind to the caveolin-1 promoter in these cells. Silencing Ets1 in these cells has minimal effect on endogenous caveolin-1 expression. Differences in cofactor abundance probably account for these differential binding patterns.
In addition, we raise the possibility that an ETS fusion protein regulates caveolin-1 expression, with the first fusion protein in NSCLC, a rare fusion tyrosine kinase EML4-ALK (echinoderm microtubule associate protein like 4-anaplastic lymphoma kinase), recently reported (49). Gene fusions involving four different ETS factors, including all three PEA3 family members, are present in >50% of all human prostate cancers, and ETS fusion proteins have been described in breast cancer and acute myelogenous leukemia (50–52).
It is hypothesized that loss of caveolin-1 expression is important in transformation of normal epithelial cells to cancer cells, with caveolin-1 functioning as a tumor suppressor gene in this context. The inverse correlation of Net and caveolin-1 by immunohistochemistry in primary lung tumors and our results from ChIP and siRNA experiments in NCI-H23 (cav-low) cells suggest that Net represses caveolin-1 transcription in primary lung tumors. When tumors become metastatic, caveolin-1 levels are increased, and it has been suggested that the function of caveolin-1 changes to that of a metastasis promoter. We believe that, based on our results, there are several potential mechanisms for increased caveolin-1 in metastasis. One possibility is that, since there is increased PEA3 expression as tumors progress, PEA3 is more readily able to bind, perhaps by competing with Net binding, to the caveolin-1 promoter.
Alternatively, since Net immunohistochemical staining is nuclear in the primary tumor cells and cytoplasmic in the metastatic tumor cells, perhaps a change in cellular location leads to inability of Net to bind and repress the caveolin-1 promoter in metastasis. Nuclear location has been shown to be important for the transcription factor nuclear factor-κB to be active and tumor promoting (53). Another possibility is that Net is non-phosphorylated in primary tumors but is phosphorylated in metastatic tumors, causing Net to change from a repressor to an activator. Future studies will include ChIP assays in fresh human lung tumor samples to determine if the promoter-binding patterns seen in our cell line model are representative of primary tumors and metastatic sites. We will also characterize the phosphorylation status of Net in primary tumors versus metastatic sites.
We have shown by ChIP assays, promoter analysis and siRNA experiments that PEA3 is a transcriptional activator and that Net is a transcriptional repressor of caveolin-1 in lung cancer cell lines. Based on these results and the patterns of expression of PEA3 and Net we observed in human lung cancer specimens, it is probably that Net is suppressing caveolin-1 transcription in primary lung tumors and that PEA3 is activating caveolin-1 transcription in metastatic sites. Understanding the transcriptional regulation of caveolin-1 in lung tumorigenesis may provide insight into how to prevent caveolin-1 upregulation, probably an important step in metastasis. If overexpression of caveolin-1 promotes cancer cell migration and seeding, blocking caveolin-1 expression in metastatic tumor cells is a potential therapeutic target in advanced NSCLC.
American Cancer Society (IRG-72-001-34-IRG) to H.K.; American Lung Association Lungevity to H.K.; National Institute of Health/National Heart, Lung, and Blood Institute Training Grant to K.A.S.
Conflict of Interest Statement: None declared.