Methylation Induced Gene Silencing of HtrA3 in Smoking-Related Lung Cancer

Lung cancer is the leading cause of cancer-related deaths in the United States and worldwide (1, 2). Standard first line treatment for small-cell lung cancer (SCLC) includes platinum-based chemotherapy, generally with the DNA and protein cross-linking agent cis-platinum-(II)-diamine dichloride (cisplatin) or carboplatin plus etoposide, a topoisomerase II inhibitor, with or without radiation (3–6). Treatment of non–small cell lung carcinoma may include surgical resection, radiotherapy, and/or platinum-based chemotherapy (7–10). Despite a promising initial response to therapy, however, some 86% of lung cancer patients still die within 5 years of diagnosis, many from chemoresistant disease (2, 11–14). This grim prognosis underscores the need to characterize not only those factors involved in lung cancer etiology and progression but also those that compromise treatment by promoting chemoresistance.

Smoking increases the risk of all histologic types of lung cancer but is most strongly linked with small-cell lung cancer and squamous cell carcinoma (15, 16). Among carcinogens that have been isolated from cigarette smoke, the tobacco-specific nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) is widely accepted as a major causative agent for lung cancer in smokers (17–19). NNK metabolites can interact directly with DNA, forming methyl adducts primarily in target tissues most affected by NNK carcinogenicity (17, 18). Further, of adenocarcinomas induced by NNK in rats, 94% showed gene silencing through promoter hypermethylation within p16 (20, 21). In humans, p16 silencing is associated with promoter hypermethylation in non–small cell lung cancers (22).

HtrA3 was first characterized as an upregulated factor during mouse uterus placentation (23). More recently, HtrA3 has been identified as a possible target for cigarette smoke–induced changes in normal human bronchial epithelial cells, showing dramatic transcriptional downregulation within 2 hours of condensed cigarette smoke exposure in contrast to the steady HtrA1 and HtrA2 expression observed over the same time period (24). Long and short HtrA3 splice variants that differ in the COOH-terminal PDZ domain, which is lacking in the short isoform, and in the last seven COOH-terminal residues of the short form have been cloned (23). Of the four known human High-temperature requirement A (HtrA) family members, HtrA3 shares the greatest sequence homology and amino acid identity with HtrA1 (25). We previously reported that HtrA1 is downregulated, in part, through promoter hypermethylation in ovarian cancer (26).

In the present study, we have shown that HtrA3 is epigenetically silenced through cigarette smoke–induced methylation. Additionally, we provide preliminary evidence that clearly implicates HtrA3 loss as a factor in promoting resistance to etoposide and cisplatin cytotoxicity. More importantly, using patient-derived lung tumor samples, we show a statistically significant correlation between heavy smoking history and low HtrA3 expression. Collectively, these results suggest that cigarette smoke–induced methylation of HtrA3 could contribute to the etiology of chemoresistant disease in smoking-related lung cancer.

Cell lines were cultured according to the American Type Culture Collection guidelines.

RNA extraction, cDNA synthesis, and reverse transcription-PCRs (RT-PCR) were done as previously described (19). HtrA3 RT-PCR primers and control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) primers are listed in Supplementary Table S1. Real-time duplex PCR was carried out as previously described (16) with HtrA1, HtrA2, and HtrA3 primers obtained from SABiosciences.

The anti-HtrA3 monoclonal antibody was developed at the Mayo Antibody Core Facility Rochester. It is characterized in Supplementary Fig. S1. The antibody was raised against a polypeptide corresponding to amino acids 144 to 453 of HtrA3 in BALB/c mice. Supernatants from mouse hybridoma cultures were tested over several weeks until optimal HtrA3 recognition with minimal cross-reactivity to purified HtrA1 was obtained by ELISA and Western with culture medium as a negative control. Anti-HtrA3 monoclonal antibodies were purified from supernatants of these hybridomas using ImmunoPure Immobilized Protein A Plus IgG Purification kit (Pierce). Purified antibodies were tested for HtrA3 specificity by Western analysis using purified protein and lysates from confirmed HtrA3-expressing cell lines and from cell lines lacking the HtrA3 expression.

Immunoblotting was done as previously described (16). HtrA2 (V-17) antibody was obtained from Santa Cruz Biotechnology (dilution of 1:200 in TBST+5% bovine serum albumin).

Immunoblots were analyzed by the Bio-Rad Quantity One 4.40 software. Arbitrary intensity values were normalized to β-actin loading controls and expressed as fold change over controls.

Treatments were done as previously described (16). LBH589 was purchased from Novartis Pharmaceuticals and was dissolved in DMSO before use.

These procedures were done as previously described (19). Primers used for methylation analysis are given in Supplementary Table S1.

NNK (ABCR GmbH & Co.) was prepared in DMSO to a final concentration of 25 mmol/L. BEAS-2B cells were seeded at 1 × 10⁵ cells in 35-mm tissue culture dishes. Either NNK or DMSO was added to the culture medium after 24 h in culture. Culture medium was changed every 48 h.

Lentiviral supernatants were generated by transfecting short hairpin RNA (shRNA) transfer vectors for nontarget control, or HtrA3 shRNA targeting either the 3′ untranslated region (UTR) or the open reading frame into 293T in combination with both VSV-G and gag/pol vectors (Sigma). Cells were seeded at 5 × 10⁶ in 10-cc tissue culture dishes. A total of 12 μg of DNA was included in the transfection mixture with 6 μg transfer vector and 3 μg each of VSV-G and gag/pol vectors. Transfections were carried out 24 h after seeding using Lipofectamine LTX (Invitrogen). Transfection medium was replaced with 5 mL culture medium after overnight incubation. Supernatants containing lentiviral particles for transduction were collected 48 h after transfection. Transduction was carried out using 2 mL supernatant on BEAS-2B and EKVX 24 h after seeding at 1.5 × 10⁵ in six-well plates. Efficiency of knockdown was determined by immunoblot after 48 h of transduction. We selected stable batch clones in BEAS-2B using 0.5 μg/mL puromycin after 2 d of transduction.

Rescue experiments were done as previously described (19). The full-length HtrA3 open reading frame was obtained from the Invitrogen Full-Length Mammalian Gene Collection and subcloned into pcDNA3.1(+)myc-His A (Invitrogen). Inactive mutants containing an S305A point mutation within the conserved trypsin H-S-D catalytic triad were generated by site-directed mutagenesis using the QuickChange Site Directed Mutagenesis kit (Stratagene). Primers are given in Supplementary Table S1. A polyclonal antibody raised against a peptide corresponding to amino acids 250 to 300 within the trypsin domain was obtained from Abcam (dilution of 1:200 in TBST+5% bovine serum albumin) and was used to detect PDZ-deleted variants.

Treatment was carried out as previously described (21). Etoposide and cisplatin were purchased from Sigma-Aldrich and were dissolved in DMSO before use.

MTT assays were carried out as previously described (19).

Clonogenic assays were carried out as previously described (22).

Tissue microarray (TMA) construction and digital imaging system were done as previously described (18). Levels of HtrA3 expression were calculated and compared by two separate observers in a double blinded fashion and categorized as low (weak staining, 0), moderate (1+), or high (intense staining, 2+).

Immunohistochemistry for HtrA3 in normal lung and lung tumors was done using the HtrA3 monoclonal antibody as previously described (12).

Two-tailed Student's t test, ANOVA, and two-sided χ² analyses were done using Prism 3.0 (GraphPad Software). P < 0.05 and α = 0.05 were considered statistically significant.

Expression analysis of the human bronchial cell line BEAS-2B and 23 lung cancer cell lines by RT-PCR revealed that HtrA3 was expressed in BEAS-2B but was transcriptionally downregulated in 10 of 22 lung cancer cell lines (Fig. 1A, top). Immunoblot analysis using an HtrA3 monoclonal antibody (Supplementary Fig. S1) revealed that HtrA3 was expressed in BEAS-2B but was lost in over half (13 of 23) of lung cancer cell lines analyzed including adenocarcinoma lines H23, H1648, and H2122, which each showed mRNA expression (Fig. 1A, bottom). Collectively, these data suggest that HtrA3 is expressed in the normal lung but may be specifically lost in lung cancer.

As epigenetic modification has been implicated in regulating HtrA1 expression in ovarian cancer (26, 27), we assayed lung cancer cell lines lacking the HtrA3 expression for aberrant methylation and histone deacetylation. Treatment of H157, A549, CL1.0, and H460 with increasing concentrations of the methyltransferase inhibitor 5-aza-2′-deoxycytidine (5-aza-dC) resulted in a dose-dependent increase in HtrA3 transcription (Fig. 1B). Treatment with the histone deacetylase inhibitor trichostatin-A did not result in HtrA3 reexpression and no synergistic effects on HtrA3 expression were observed in combination with 5-aza-dC, indicating that HtrA3 expression is primarily under methylation control (Fig. 1C). However, although real-time PCR analysis showed that the novel histone deacetylase inhibitor LBH589 did not induce HtrA3 re-expression in A549 or H157, a synergistic effect on HtrA3 induction was observed in A549 when 5-aza-dC and LBH589 were combined (Supplementary Fig. S2A-B). Further, HtrA3 expression was induced by LBH589 in H358 and a synergistic effect on HtrA3 expression was again observed in combination with 5-aza-dC, indicating that HtrA3 expression in lung cancer cell lines may also be regulated by histone deacetylation in a manner that is cell type specific (Supplementary Fig. S2C).

To identify the target regions of HtrA3 methylation, we analyzed the methylation status of the HtrA3 promoter and of the first exon in BEAS-2B and in nine lung cancer cell lines, and observed differential methylation within the first exon, but not within the promoter, between cell lines with and without HtrA3 expression (Supplementary Fig. S3A; Table 1A). Sequence analysis revealed that 78% of lung cancer cell lines analyzed (7 of 9) showed an increase in methylation frequency in the first exon with corresponding a loss of HtrA3 expression (Table 1A). In contrast, cell lines with HtrA3 expression were consistently unmethylated at 19 of 23 CpG sites. These sites were consistently methylated in cell lines with low HtrA3 expression.

We next induced HtrA3 expression in H157 and A549 using 5.0 μmol/L 5-aza-dC and determined the methylation status of CpG sites within this region by direct sequencing following sodium bisulfite treatment (Fig. 1D). Sites 4, 7, 8, 11, 17, 20, and 21 within the first exon became unmethylated following 5-aza-dC treatment in H157 and A549 with corresponding increase in HtrA3 mRNA and protein expression (Supplementary Fig. S3A; Table 1A). These sites were unmethylated in cell lines with HtrA3 expression but were methylated in cell lines without HtrA3 expression (Table 1A).

To determine if HtrA3 was methylated in primary tumors from lung cancer patients with a history of smoking, we next assessed HtrA3 methylation and expression in primary lung tumors from heavy smokers (>40 pack-years) and from light smokers (<10 pack-years). Our analysis revealed an increase in the frequency of HtrA3 methylation in 69% (11 of 16) of tumors from heavy smokers with corresponding loss of HtrA3 expression when compared with tumors from light smokers (Supplementary Fig. S3B; Table 1B). Collectively, these data support CpG island methylation within the first exon of HtrA3 as a major mechanism of regulating HtrA3 expression in lung cancer, particularly related to smoking.

Cigarette smoke is a known etiologic factor for lung cancer that exerts its effects, at least in part, through promoter hypermethylation in crucial tumor suppressor genes (22). We therefore investigated whether the cigarette smoke–derived carcinogen NNK regulates HtrA3 expression through methylation in vitro. BEAS-2B human bronchial cells were treated with increasing concentrations of NNK and were assayed for mRNA and protein expression. We observed a 2-fold loss of HtrA3 protein expression within 24 hours of 500 μmol/L NNK treatment compared with HtrA3 expression at treatment initiation and a more rapid 2-fold loss within 2 hours of treatment when the dose was doubled (Fig. 2A). In contrast, HtrA3 expression was unsuppressed in the untreated cell population. Real-time PCR further revealed significant downregulation of HtrA3 transcription within 2 hours of 1.0 mmol/L NNK treatment (Fig. 2B). In contrast, HtrA1 and HtrA2 mRNA and protein expression were induced by 1.0 mmol/L NNK treatment (Fig. 2A-B) and HtrA2 protein expression doubled within 24 hours of treatment at 500 μmol/L (Fig. 2A).

As chronic smokers are subject to long-term, low-dose cigarette smoke exposure, we assessed the effect of prolonged low-dose NNK treatment on HtrA3 expression in BEAS-2B. We observed a steady decline in HtrA3 protein and mRNA expression, with an 8-fold decrease in protein expression over 15 days (Fig. 2C-D). In contrast, both HtrA1 and HtrA2 were induced over the same time period. Interestingly, we also observed HtrA3 protein re-expression 72 hours after an initial 24-hour treatment with 500 μmol/L NNK, suggesting that NNK-mediated downregulation of HtrA3 expression was reversible (Supplementary Fig. S4A). Further, we found that NNK treatment slowed BEAS-2B cell proliferation when measured by mitochondrial reduction of the tetrazolium salt MTT (Supplementary Fig. S4B).

In addressing whether NNK-mediated HtrA3 suppression occurred through methylation, bisulfite-modified DNA from NNK-treated samples was analyzed by direct sequencing. We observed an increase in methylation frequency in the first exon of HtrA3 with corresponding loss of HtrA3 expression with both 500 μmol/L NNK treatment over 24 hours and 10 μmol/L treatment over several days when compared with untreated controls (Table 2A-B). Collectively, these data suggest that HtrA3 may be a specific target for cigarette smoke carcinogen-mediated regulation through methylation in smoking-related lung cancer.

Downregulation of human HtrA family members has been implicated in resistance to cisplatin and paclitaxel in ovarian and gastric cancers and in renal cells (28, 29). We therefore investigated whether the loss of HtrA3 expression could result in chemoresistance in vitro. We observed a 2- to 3-fold decrease in HtrA3 protein expression in pooled BEAS-2B clones stably transduced with shRNAs directed at either the open reading frame or the 3′UTR compared with endogenous HtrA3 expression in a clonal line stably transduced with a nontargeted shRNA control (Fig. 3A). MTT assay (Fig. 3B, top) and clonogenic survival assay (Fig. 3B, bottom) showed that endogenous HtrA3 expression sensitized cells to both etoposide and cisplatin treatment compared with clonal lines in which HtrA3 was stably downregulated (P < 0.05).

We next subcloned the long HtrA3 isoform (full length) and a shortened variant lacking the PDZ domain (PDZ deleted) as well as catalytically inactive forms of each. We transiently re-expressed each form in a clonal cell line in which HtrA3 was stably suppressed by a 3′UTR-targeted shRNA (Fig. 3C). Re-expression of the functional full-length and PDZ-deleted HtrA3 variants in BEAS-2B significantly attenuated proliferation as determined by MTT assays with either etoposide or cisplatin treatment compared with the re-expression of protease inactive variants or expression of vector control (P < 0.05; Fig. 3D). Moreover, re-expression of functional variants significantly decreased clonogenic survival with etoposide treatment compared with the re-expression of protease-inactive variants or vector control (P < 0.05). Taken together, these data implicate HtrA3 as a key factor in modulating sensitivity to etoposide and cisplatin in smoking-related lung cancer that mediates this function, at least in part, through its catalytic activity.

As HtrA3 expression is regulated by cigarette smoke in vitro, we next addressed whether HtrA3 expression in primary lung tumors is strongly correlated with smoking history in vivo. HtrA3 expression was determined by immunohistochemistry in normal lung tissue and on a TMA containing 70 primary lung tumors from patients identified as light (<10 pack-years, 0), moderate (between 10 and 40 pack-years, 1), and heavy (>40 pack-years, 2) smokers. Levels of HtrA3 expression were calculated and compared by two separate observers in a double-blinded fashion and were categorized as low (weak staining, 0), moderate (1+), or high (intense staining, 2+; Fig. 4A-B). Supplementary Table S2 shows the relevant patient and tumor characteristics. Statistical analyses were done by V. S. and D. B. Light and moderate smoking groups were combined for statistical analysis. As given in Fig. 4C, 100% (35 of 35) of tumors from heavy smokers showed either low or moderate HtrA3 expression with 66% (23 of 35) of tumors showing markedly low HtrA3 expression. Likewise, 91% (32 of 35) of tumors from light and moderate smokers combined showed moderate to high levels of HtrA3 expression with 49% (17 of 35) showing markedly high expression. Results given in Fig. 4C show a statistically significant difference between the low- and high-staining groups (P < 0.0001). The odds ratio for light smoking history in the cohort with low levels of HtrA3 expression was 0.05 (95% confidence interval, 0.0221-0.2313). No significant correlations were noted between HtrA3 expression and patient gender or tumor grade, stage, or histology. These results clearly support a strong correlation between heavy smoking history and low HtrA3 expression in primary lung tumors.

Gene silencing through aberrant promoter and first exon hypermethylation has been widely implicated in a variety of pathologic processes including cancer induction, progression, and resistance to chemotherapy. Here, we have characterized a novel mechanism for HtrA3 regulation through cigarette smoke–mediated first-exon methylation in smoking-related lung cancer. We have shown increased frequency of methylation with corresponding HtrA3 downregulation both in vitro and in vivo, uniquely correlated to smoking, and have identified chemoresistance as one functional consequence of HtrA3 loss.

Our finding that HtrA3 expression is regulated through its first exon, although unexpected, is not unique. Although rare, hypermethylation of the first exon in the absence of significant promoter methylation has been implicated in gene silencing and tumorigenesis for the BRCA2 gene in sporadic breast cancer and for the p73 gene in some leukemias, lymphomas, and brain tumors as well as in ovarian cancer cell lines (30, 31). In smoking-related lung cancer, cigarette smoke carcinogens, particularly NNK, promote tumorigenesis by forming methyl adducts in hypomethylated areas, which commonly include the promoter and first exon (32, 33). Studies in aging and cancer have also suggested that DNA methyltransferases bind preferentially to methylated and hemimethylated nucleation sites in mediating de novo CpG methylation through spreading (34). Thus, in smoking-related lung cancer, prolonged cigarette smoke exposure may result in methyl adducts within the HtrA3 promoter region that serve as nucleation sites for DNA methyltransferase–mediated spreading to unmethylated CpG sites within the first exon. However, methylation analysis of primary lung tumors revealed a small minority of samples for which HtrA3 expression was downregulated in the absence of methylation (Table 1B). Thus, there clearly are mechanisms other than methylation that may regulate HtrA3 expression that may include histone deacetylation as well as microRNA, loss of heterozygosity, and genomic mutation, some of which have been implicated in HtrA1 silencing in ovarian and gastric cancers (26, 27).

We showed that 1.0 mmol/L NNK resulted in a 2-fold decrease in HtrA3 transcription in human bronchial cell line BEAS-2B within 2 hours of treatment (Fig. 2B). As de novo methylation requires at least one round of DNA replication and cell division, however, it is unlikely that this rapid transcriptional downregulation resulted from aberrant methylation (35). Analysis of the HtrA3 first exon identified the core xenobiotic response element (XRE) consensus sequence, 5′-TNGCGTG-3,′ as a target of differential methylation at CpG position five in lung cancer cell lines and in primary tumor samples (Supplementary Fig. S3; Table 1). XRE are enhancer DNA elements located proximal to the promoter region of genes implicated in metabolizing cigarette smoke carcinogens. They bind aryl-hydrocarbon receptors, which are induced in response to xenobiotic factors, such as cigarette smoke (36). It has been proposed that sudden activation of the XRE pathway results in the rapid recruitment of transcription effectors away from other less crucial genes, downregulating their expression while the products of XRE responsive genes are upregulated (37, 38). HtrA1 and HtrA2 transcriptional and protein upregulation may implicate their involvement in XRE-mediated response to xenobiotic stress. Along these lines, a consensus XRE sequence was identified proximal to the HtrA1 promoter region. Initial induction of HtrA3 protein expression observed within 4 hours of 500 μmol/L NNK treatment followed by HtrA3 suppression within 24 hours (Fig. 2A) could suggest that NNK-induced methylation of the HtrA3 XRE both attenuates the HtrA3 stress response and contributes to HtrA3 silencing.

Comparison of Tables 1 and 2 reveals that in some instances, overall levels of methylation in A549 and H157 after treatment with 5-aza-dC, which induced HtrA3 expression, were similar to levels of methylation following NNK treatment, which resulted in HtrA3 suppression. A more detailed analysis identified CpG sites 4, 6, 7, 8, 11, 17, 20, and 21 as sites demethylated in both A540 and H157 within 48 hours of 5-aza-dC treatment. Sites that become methylated with NNK treatment seem to differ with dose and the duration of treatment. With 500 μmol/L NNK treatment, CpG sites 4, 6, 7, 8, 11, and 19 were fully methylated, whereas site 20 was hemimethylated within 24 to 48 hours of treatment. In contrast, with 10 μmol/L NNK treatment, CpG sites 4, 6, 8, 11, 20, and 21 were fully methylated, whereas sites 7 and 17 were hemimethylated within 15 days of treatment. Based on these data, minimal CpG sites within the first exon of HtrA3 involved in the regulation of HtrA3 expression seem to be sites 4, 6, 8, 11, and 20.

The identification of HtrA3 as a novel target for suppression through cigarette smoke–mediated methylation and the evidence of its role in modulating sensitivity to etoposide and cisplatin in vitro may contribute to studies of both biomarkers of chemoresistance, particularly related to smoking, and targeted epigenetic therapies in lung cancer. CpG sites 4, 6, 8, 11, and 20 could eventually serve as screening sites with strong predictive value with regard to treatment outcome that could alter the target approach to lung cancer treatment from a general to an individual treatment strategy. Further, as HtrA3 is downregulated in lung tumors by epigenetic mechanisms, it represents a strong potential target for epigenetic modulation (39). Chemotherapeutic strategies for cancer treatment have included 5-aza-dC in the treatment of myelodysplastic syndrome, chronic myelogenous leukemia, and acute myelogenous leukemia (40). Future clinical trials involving the evaluation of epigenetic therapies should eventually include HtrA3 as a therapeutic marker to assess the biological response of epigenetic therapies and should evaluate its prognostic potential to predict tumor response to combined epigenetic and conventional chemotherapy. Until then, the analysis of HtrA3 expression and function still warrants further investigation in confirmatory studies and perhaps even in different tumor types.

No potential conflicts of interest exist.

We thank Giovanni Aletti for the assistance in statistical analyses, Keishi Narita and Thomas Beito for the assistance with HtrA3 protein purification and monoclonal antibody generation, Melissa Scheerer the for assistance with 5-aza-dC treatments, the members of the Shridhar laboratory, as well as S.H. Kaufmann, G. Isaya, and D. Mukhopadhyay for their discussions and critical review of the data.

Grant Support: D. Beleford is a Bonner Scholar in Biochemistry and Molecular Biology and a recipient of Mayo Clinic College of Medicine M.D./ PhD Training Program funding. This work was also supported by funds from Edith and Bernard Waterman Foundation and the Mayo Clinic (V. Shridhar).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.