Defective Expression of Transforming Growth Factor β Receptor Type II Is Associated with CpG Methylated Promoter in Primary Non-Small Cell Lung Cancer

Transforming growth factor (TGF)-β is one of the most well known and potent inhibitors of epithelial cell growth (1). Resistance to growth inhibition by TGF-β has been considered an important step in tumorigenesis and contributes to development of many types of tumors (2, 3). TGF-β binds directly to TGF-β receptor type II (TGFβRII), a constitutively active transmembrane serine/threonine kinase, and is then recognized by TGF-β receptor type I, which is phosphorylated and activated by TGFβRII (4). Therefore, inactivation of TGFβRII can result in TGF-β resistance (5), and TGFβRII has been shown to function as a tumor suppressor in many solid tumors. In contrast, variants of the TGFβRII gene are rarely seen in primary non-small cell lung cancer [NSCLC (6, 7, 8, 9, 10, 11, 12, 13, 14, 15)], even though reduction of TGFβRII expression has been found frequently in NSCLC cells and tissues (16, 17). These findings stimulated our attempt to investigate the molecular mechanisms underlying defective TGFβRII expression in NSCLC.

Gene expression can be frequently inactivated by reversible epigenetic events rather than by genetic events (18). DNA methylation alterations are now widely recognized as a contributing factor in human tumorigenesis. In neoplasia, aberrant methylation serves as an alternative to point mutation and chromosome deletion in inactivating tumor suppressor genes or the genes that preserve normal cell function such as VHL and the DNA mismatch repair gene MLH1 (19). Methylation of the TGFβRII promoter sequence has been identified in primary human esophageal cancer (20). However, there is no experimental evidence reported thus far on CpG methylation within the promoter sequence of the TGFβRII gene or on association between the methylation event and TGFβRII expression in NSCLC.

To address these key questions, the present study carried out PCR-denaturing gradient gel electrophoresis (DGGE) to scan for mutations in the entire coding regions of the TGFβRII gene in 43 tumor and paired normal tissue specimens from patients with primary NSCLC. Then, we used immunohistochemistry and methylation-specific PCR (MSP) assays to determine TGFβRII expression and the nested MSP analysis to explore CpG methylation within the TGFβRII gene promoter. Our experimental data support that defective expression of TGFβRII plays a significant role in the development of NSCLC, whereas mutation of the gene is rare in the NSCLC samples under study. In addition, the present study provides clear evidence that the reduced expression of TGFβRII in NSCLC patients would be explained by down-regulation of gene expression at transcriptional level due to aberrant 5′ CpG methylation.

Fresh tumor and the corresponding paracarcinoma tissue specimens were collected after having obtained formal, written consent from 43 patients who attended the research and underwent pulmonary resection for primary NSCLC at Shanghai Hospital for Pulmonary Diseases. These patients had received neither radiotherapy nor chemotherapy before surgery. The pathological stages were determined according to the Revised International System for Staging Lung Cancer (21). The demographic and clinical features of these patients are described in Table 1.

Genomic DNA was extracted from the tumor and paired normal tissue samples by standard proteinase K digestion and phenol-chloroform extraction (22).

All seven exons of the TGFβRII gene were assayed by use of PCR-DGGE. The PCR primers used to amplify each of these exons are listed in Table 2. The PCR reaction conditions were described in our previous report (23), except for the annealing temperatures that are listed in Table 2. The DGGE experiment was performed with Dcode Universal Mutation Detection System (Bio-Rad). The PCR products of 20 μl were loaded onto an 8.0% (w/v) polyacrylamide gel (acrylamide/bis solution, 37.5:1) containing a linear chemical gradient of denaturants (Table 2) and subjected to electrophoresis at 150 V for 5–10 h in 1× TAE electrophoresis buffer. After electrophoresis, the gels were stained with ethidium bromide solution for 10 min and visualized under UV illumination.

The PCR products were separated on agarose gels. Bands of the expected size were excised and purified using the QIAquick Gel Extraction Kit (Qiagen) and then sequenced directly with the BigDye Terminator Cycle Sequencing Reaction Kit (PE Applied Biosystems) on an ABI 377 automatic DNA sequencer (PE Applied Biosystems).

Immunohistochemical staining for TGFβRII in NSCLC tissues was performed on 5-μm paraffin sections. Before incubation with primary antibody (i.e., rabbit polyclonal antibody against human TGFβRII gene product sc-400; Santa Cruz Biotechnology, Inc.), the sections were blocked with 10% normal goat serum for 15 min at room temperature. After the blocking serum was removed, the sections were incubated overnight at 4°C with the primary antibody at a 1:300 dilution. The sections were rinsed three times with PBS (3 min each time), incubated with biotinylated goat antirabbit secondary antibody (Zymed Laboratories Inc.) for 15 min at 37°C, rinsed three times with PBS (3 min each time), sequentially incubated with streptavidin-horseradish peroxidase (Zymed Laboratories Inc.) for 15 min at 37°C, and, finally, rinsed three times with PBS (3 min each time).

Immunoperoxidase reactions were visualized by making use of a mixture of 3,3′-diaminobenzidine tetrahydrochloride (Sigma Chemical Co.) and hydrogen peroxide. To generate a negative control, the sections were also treated with PBS instead of the primary antibody.

Human normal bronchial epithelium was used as control because NSCLC develops from bronchial epithelium precursors.

To take into account variation in both the proportion and intensity of the epithelial cells stained, the samples were first grouped into four grades: 0 (<10% of cells stained); 1 (10–33% of cells stained); 2 (34–66% of cells stained); and 3 (>66% of cells stained). They were then divided, according to the staining intensity, into four additional classes: 0 (negative staining); 1 (weak staining); 2 (moderate staining); and 3 (strong staining). The final score was determined as the product of the proportion and intensity scores and varied from 0 to 9. A sample was classified as a preserved (Pr) type of TGFβRII expression if its final score was within the 6–9 range; otherwise, it was classified as a reduced (Re) type of TGFβRII expression.

Genomic DNA extracted from the tumor and corresponding normal tissues was treated with sodium bisulfite as described previously (Ref. 24; the protocol in detail is available from the corresponding author on request). Methylation patterns in the TGFβRII promoter sequence [from the site −127 to –5 bp to the transcription start (25), GenBank accession number U37070] were determined using a nested PCR approach. In the first round of PCR, the primers were 5′-atttaatagattggagtat-3′ (forward) and 5′-ccaacaactaaacaaaac-3′ (reverse). The PCR reaction was performed in a 25-μl volume composed of 200 ng of bisulfite-modified DNA, 12.5 pmol of forward and reverse primers, 10 mm Tris-HCl (pH 9.0), 50 mm KCl, 0.1% Triton X-100, 6.0 mm MgCl₂, 1.2 mm deoxynucleotide triphosphates, 5.0% DMSO, and 1.5 units of Taq polymerase. Amplification conditions were as follows: initial denaturation at 94°C for 5 min; 40 cycles of 94°C for 45 s, 44°C for 30 s, and 72°C for 45 s; and a final 10-min extension at 72°C. The size of the product after this initial PCR reaction was 378 bp. For the second round of PCR, this product was diluted 1:10 in water, and the diluted solution of 1 μl was taken to carry out the MSP as suggested in Ref. 26. The nested primer sequences of TGFβRII for the unmethylation reaction were 5′-ttgaaagttggttaaagtttttgga-3′ (forward) and 5′-aaacaaaacctctctccaccca-3′ (reverse), and primer sequences for the methylated reaction were 5′-gaaagtcggttaaagttttcgga-3′ (forward) and 5′-acaaaacctctctccgcccg-3′ (reverse). The PCR condition was the same as that described above, except that the annealing temperatures for the unmethylated and methylated reaction were 45°C and 67°C, respectively. The products of the unmethylated and methylated reaction were accordingly 123 and 119 bp in size. Genomic DNA treated with SssI methylase was used as a positive control for the methylation experiment, but DNA extracted from normal blood without bisulfite treatment served as a negative control. The methylation status of the PCR products was determined by electrophoresis.

Total RNA from the tissue specimens was isolated by use of Tripure RNA Purification Kit (Roche Applied Science). RNA quality and concentration were estimated spectrophotometrically at 260 nm. RT-PCR was performed using SUPERSCRIPT II Kit (Life Technologies, Inc.). The cDNA was synthesized from total RNA of 1 μg of oligo(dT)₁₅ primer and RNase H⁻ reverse transcriptase (from Moloney murine leukemia virus).

The TGFβRII (581-bp) gene fragment was coamplified with β-actin (933 bp) as an internal control. For the two genes, amplification was carried out in the same tube containing 1× PCR buffer, 1.5 mm MgCl₂, 2 units of Taq DNA polymerase, and 10 pm of each of the flanking primers. After initial denaturation at 94°C for 5 min, amplification conditions were as follows: 30 cycles of 94°C for 45 s, 51°C for 30 s, and 72°C for 1 min; followed by a final 7-min extension at 72°C. The RT-PCR products were separated on a 1.5% agarose gel. The semiquantitative of TGFβRII mRNA was evaluated according to the absorbance ratio of mRNA/β-actin.

The primer sequences were 5′-cggaagctcatggagttcagcgagc-3′ (forward) and 5′-agcccaaagtca cacaggcagcagg-3′ (reverse) for TGFβRII and 5′-accctgaagtaccccatc-3′ (forward) and 5′-ctagaagcatttgcggtg-3′ (reverse) for β-actin.

All seven exons of the TGFβRII gene were screened through PCR-DGGE analysis in all 43 NSCLC tissues and their corresponding paracarcinoma tissues. Fig. 1,A demonstrated a DGGE variant detected in the boundary area of exon 4, which was also seen in the corresponding paracarcinoma tissues. Sequence analysis of the variant showed that it was an A/T substitution at the fourth-to-last base in intron 3 (Fig. 1 B). The DGGE analysis excluded any other mutation in the whole coding region of TGFβRII in all of the tumor and normal samples.

Fig. 2 illustrates immunostaining views of four representative samples. The normal bronchial epithelium was strongly stained (Fig. 2,A), indicating normal expression of TGFβRII. Among the 43 tumor samples, immunostaining was distributed mainly on membrane and in the cytoplasm of epithelial cells and was heterogeneous in some of the tumor sections. The phenomenon of immunostaining in the cytoplasm could be explained by the fact that the epithelia were largely broken after being treated with immunohistochemical agents. According to the criteria described above to classify the tumor samples, we found that 55.8% (24 of 43) of samples showed preserved TGFβRII expression (Fig. 2,D), and 44.2% (19 of 43) of samples had TGFβRII reduced expression (Fig. 2, B and C).

We first examined four tissue samples of normal bronchial epithelium collected from individuals who had no record of NSCLC. Their methylation analyses are shown in Fig. 3,A. The figure showed that the TGFβRII promoters of these samples were free of methylation, suggesting an effective control of false positive in the nested MSP experiment. In contrast, Fig. 3,B demonstrated that the MSP approach revealed clear evidence of the methylated TGFβRII promoter in 42% of the tumor samples. Table 3 summarized analysis of association between TGFβRII methylation and its expression from the immunohistochemistry experiment. It revealed that the reduced TGFβRII expression in the NSCLC samples was highly significantly associated with the methylated promoter of the gene (χ²_{df = 1} = 19.25; P < 10⁻⁴).

RT-PCR analysis was carried out on the 19 samples that showed a reduced TGFβRII expression and 5 other randomly selected samples that showed normal expression of TGFβRII. Fig. 4,A demonstrated the mRNA change in both tumor and paired normal tissues from 8 representative NSCLC samples of the 19 cases. It showed that the mRNA level was comparable between the tumor and normal tissues in cases 1 and 2, but for the cases 3 to 6 and 8, the mRNA in the tumor tissues was remarkably reduced when compared with the corresponding normal tissues. The most distinct example was case 7, for which the mRNA in the tumor tissue was undetectable. However, the mRNA levels were not changed between tumor and normal tissues for the five samples in which the protein expression was preserved (Fig. 4 B).

Of the 19 samples under investigation, 16 (84.2%) tumor tissues showed clear reduction or loss in quantity of mRNA in comparison with their normal pairs, whereas the levels of mRNA remained unchanged between the tumor and paracarcinoma tissues for those 5 samples (Table 4). Table 4 illustrated the relationship between TGFβRII expression assay and the methylation analysis. It demonstrated that the reduced gene expression revealed by the immunohistochemistry analysis was highly correlated to the methylation event, which was further confirmed by the mRNA analysis.

Resistance to TGF-β is common in NSCLC cells (16). As a receptor of TGF-β, TGFβRII plays a key role in this step because impaired expression of TGFβRII may cause refractoriness to TGF-β. This has been described in several types of cancers including NSCLC. It has been suggested that TGF-β resistance stemming from inactivation of TGFβRII could be a multiple process involving both genetic and epigenetic events (20).

The TGFβRII gene has been mapped to chromosome 3p22, and loss of heterozygosity on chromosome 3p has been detected in 63% of NSCLC (15). Moreover, extensive analyses found that mutations within the coding sequence of the TGFβRII gene were rare in NSCLC. These findings have made interpretation of the role of molecular mechanisms of TGFβRII expression in affecting NSCLC carcinogenesis unclear.

To address these ambiguities, the present study first screened the whole coding sequences of TGFβRII in 43 pairs of NSCLC tumor and normal tissue specimens and detected no pathogenic mutation, but a polymorphic site in the boundary area between exons 3 and 4 was detected. This polymorphism has been reported elsewhere (27). Another mutational analysis on the TGF-β receptor type I gene also failed to detect any pathogenic mutation (23).

Furthermore, we carried out immunohistochemistry analysis on the same set of samples and observed clear differentiation in expression of TGFβRII between the tumor and paired paracarcinoma samples. Of the 43 pairs, 19 tumor samples showed a remarkable reduction in TGFβRII expression, and the expression was even undetectable in 1 of them. This high frequency (44.2%) of reduced or lost expression of TGFβRII suggested that the gene would be an essential molecular factor in influencing NSCLC development. By taking the same experimental approach, Kim et al. (16) examined 21 surgically resected NSCLC tissues and observed reduced TGFβRII expression in 6 of them. Reduced TGFβRII expression was also observed in 5 of 13 human NSCLC cell lines (17). Furthermore, we carried out the significance test for association between reduced TGFβRII expression and prognostic factors such as gender of patients, tumor size, nodal status, and histological types and found that none of these tests was statistically significant (P > 0.43; Table 5). This agrees with the observation in small cell lung cancer by de Jonge et al. (28).

Although methylation in promoter regions of many tumor suppressor genes has been widely accepted as one of major molecular events in silencing the suppressors (29, 30, 31, 32, 33), data regarding the relationship between methylation and expression of TGFβRII in human esophageal cancer were very limited and only recently available (20). The question arises as to whether or not the reduced TGFβRII expression in the NSCLC samples can be explained by methylation in the promoter of the gene. Interestingly, our study reveals that highly significant association does exist between the reduced or lost expression and CpG methylation in the promoter of the gene. The results did not exclude the occurrence of the methylation in five paracarcinoma tissue samples, even though they had been histologically resected as distant from the tumor as possible. However, the possibility that these paracarcinoma tissues are contaminated by tumor cells cannot be discarded. This could explain the occurrence of methylation in the paracarcinoma samples because the nested MSP experiment has such high sensitivity that methylated alleles can be detected in as little as 5% of the total DNA sample (19). In addition, methylation could also be possible in both the somatic and germ-line stages (34). Given the fact that methylation was never found in any paracarcinoma samples whose tumor pairs were free of methylation, we would anticipate that the chance of the germ-line methylation would be negligible.

Gene silencing associated with methylation is mediated by methyl-binding proteins that bind to methylated cytosines and recruit a complex of proteins that repress transcription (35). DNA methyltransferase inhibitors such as 5-aza-2′-deoxycytidine were a recommended medical treatment for human cancers (36). The efficacy of the treatment depends largely on the ability of the drugs to remedy the defective expression of tumor suppressor genes. Our findings suggest that such a strategy might be a potential therapy for NSCLC patients.

Our fourth experiment has been focused on the level of mRNA from those examples that showed different patterns of TGFβRII expression (reduced, lost, or normal). It demonstrated a clear agreement between the decrease in the mRNA level and the reduced expression of TGFβRII in the tumor samples and suggested that the defective TGFβRII expression in NSCLC tumors very likely stemmed from the altered transcription of the gene. The decrease or loss of TGFβRII expression at transcriptional and posttranscriptional level has also been observed in many hormone-dependent cancers such as prostate (37) and breast (38) carcinomas.

In conclusion, the present study suggests that TGFβRII plays an important role as a tumor suppressor in primary NSCLC carcinogenesis and that the reduced expression of TGFβRII could be one of major molecular indicators for the disease. The aberrant 5′ CpG methylation of the gene has explained the down-regulation of the gene at a transcriptional level.

We are grateful for the participation and cooperation of the patients. The criticisms and comments by two anonymous reviewers and communicating editor have been very helpful in improving presentation of the paper.