The mutation information including mutations of the 26 BAF genes and the top-10 frequently mutated genes in human lung carcinoma was obtained from the COSMIC (http://www.sanger.ac.uk/cosmic) database (15,16). Each type of simple mutation (point mutation, small insertion and deletion) was represented by a specific mutation value, ‘0’, without somatic mutation; ‘1’, synonymous mutation; ‘2’, missense mutation; ‘3’, nonsense or frameshift mutation. The accumulated mutations of the 26 BAF genes in several lung carcinoma samples were then indicated by a defined BAF mutation index (BMI), equal to the sum of the mutation values from the 26 BAF genes.

For the construction of the slippage-luciferase (slippage-Luc) vector, the luciferase-coding sequence with a mutated initiation codon (ATG to CTG) was first PCR-amplified from the pGL3 vector (Promega), and cloned into the BamHI and HindIII site of the pcDNA3.1/Hygro vector (Invitrogen). An ATG initiation codon followed with the oligonucleotides containing CA repeats leads to out of frame luciferase expression and was cloned into the XhoI and BamHI site of the recombinant pcDNA3.1/Hygro-Luc construct. The oligonucleotides containing CA repeats were synthesized as follows: XhoI-ATG-(CA)₁₇ (5′-TCGAGCGCCATGGAA(CA)₁₇G-3′) and BamHI-(CA)₁₇ (5′-GATCC(TG)₁₇TTCCATGGCGC-3′). The annealed oligonucleotides were then ligated into the XhoI and BamHI site of the modified pcDNA3.1/Hygro-Luc construct to acquire the slippage-Luc vector.

The slippage-Luc vector was then linearized by BglII digestion and transfected into Calu-3 cells using the Lipofectamine transfection reagent (Invitrogen) according to the manufacturer’s instructions. After 48 h, the transfected cells were subsequently exposed to selection in medium containing 500 μg/ml Hygromycin (Clontech). The cells were fed with selective medium every 3–4 days until hygromycin-resistant colonies could be visible. Expanded colonies stably expressing the slip-luciferase transcripts were assessed by PCR.

For SMARCA4 knockdown, 4 fragments of sequences targeting against SMARCA4 were as follows: GCACCAGGAATACCTCAAT, GGTCAATGGTGTCCTCAAA, GGAGCACAAACGCATCAAT and GGCTCTGAGTACTTCATCT, and their corresponding oligonucleotides for transcribing short hairpin RNA (shRNA) were synthesized. Then these annealed phosphorylated oligos were inserted into pSUPER carrying the neomycin-resistance gene, respectively; pSUPER carrying an irrelevant nucleotide not matching any human gene sequences was used as a negative control (pSUPER-shNC). RNAi knockdown efficiency of these pSUPER constructs against SMARCA4 were evaluated 3 days after transfection in the Calu-3 cell sublines. Stable SMARCA4 knockdown was identified from double stable colonies with hygromycin- and G418-resistance of Calu-3 cells.

Genomic DNA was extracted from the stable cell lines using the DNeasy Tissue kit (Qiagen), and RNA was extracted using TRIzol^® reagent (Invitrogen) according to the protocols recommended by the manufacturer, respectively. The total RNA was treated by RNase-free DNase I (Fermentas) to remove DNA contamination.

The first-stand cDNA was transcribed by PrimerScript™ RT reagent kit with gDNA Eraser (Takara). Amplified PCR products were observed by electrophoresis on a 2% agarose gel and visualized after staining with ethidium bromide. Real-time RT-PCR was performed using a CFX96 system (BioRad), and relative expression levels were calculated as 2^−ΔΔCt according to the manufacturer’s protocol. All experiments were performed in triplicate.

Total protein extracts of stable cell lines were subjected to protein gel electrophoresis using 8% SDS-PAGE and were then transferred onto a nitrocellulose membrane (Amersham). After blocking with phosphate-buffered saline (PBS) containing 5% nonfat milk, the membrane was incubated for immunoblotting assay with the SMARCA4 antibody (1:200 dilution, Santa Cruz) at 4°C overnight. The immunostaining signal was visualized with the Odyssey infrared imaging system (Li-Cor Biosciences).

Luciferase activity was measured using the Luciferase reporter assay system (Promega) according to the instructions of the manufacturer. The experiments were performed at least 6 times, and the data were subjected to Luria-Delbrück fluctuation analyses (17,18).

The two-tailed t-test was used to compare mutated genes or simple mutations, and the statistical difference between mutation rates was determined by the Chi-square test. Pearson correlation analysis was used for investigating the relationship between BAF mutations and overall mutations in lung cancer samples. The statistical analyses were performed using SPSS software. A P-value of <0.05 was considered to indicate a statistically significant difference.

The 26 genes encoding mammalian SWI/SNF subunits are listed in Table I. Based on the data from Sanger COSMIC, we first investigated the mutation type and prevalence of these BAF genes in human lung carcinoma samples. All BAF genes except SMARCE1 were mutated to varying degrees (ranging 0.23–7.94%) in the lung carcinoma samples analyzed (Table II). Of 803 lung carcinoma samples covering all the 26 BAF genes, 282 (35.12%) exhibited at least 1 mutated BAF gene, and the BAF-mutated frequency indicated no significant difference between various histological types of lung cancer (P=0.093; Table III). These data demonstrated that BAF complexes were frequently mutated, ranking second to TP53, and played tumor-suppressive roles in lung cancer. To integrate more related information of the lung cancer samples, a mutation profile of BAF complexes was described in these BAF-mutated lung carcinoma samples, where BAF mutation index (BMI), gender, histological types and genotyping of driver genes of lung cancer are shown (Fig. 1). In addition, we also observed mutation ratios of the acknowledged driver genes with frequent mutations in lung cancer, including TP53, EGFR, KRAS, MLL3, MLL2, ZNF521, NF1, PDE4DIP, CDKN2A and STK11 (Table III). Similar to BAF, the mutation frequencies of MLL3, ZNF521, NF1 and PDE4DIP had no significant differences between the 3 histological types of lung carcinoma, implying that these tumor suppressors play a triggering role in the origin of tumor cells, regardless of histological types of lung cancer.

We next analyzed the numbers of mutated genes and simple mutations occurring in the lung carcinoma samples. Among the 803 lung carcinoma samples analyzed above, 282 BAF-mutated samples significantly exhibited more mutated genes and simple mutations than the BAF wild-type ones (Fig. 2A), which suggests that inactivated mutations of the BAF complexes were associated with genome instability of lung carcinoma. Moreover, we analyzed mutation frequencies of the acknowledged driver mutated genes in these lung carcinoma samples with a mutated-type and wild-type BAF genotype. We analyzed the available genotyping information of the top-10 driver mutated genes from the 803 lung carcinoma samples, 800 of which covering 9 driver mutated genes (no coverage with NF1) were available for further analysis. Significantly, TP53, MLL3, MLL2, ZNF521 and PDE4DIP were more often mutated in the BAF-mutated samples than the BAF wild-type ones (Table IV). According to BAF-mutated extents, these BAF-mutated lung carcinoma samples were categorized into two groups: one group with BMI ≥3, the other with BMI ≤2. The group with BMI ≥3 had more significant mutated genes and simple mutations than the other group with BMI ≤2 (Fig. 2B). To further establish the relationship between BAF complex mutations and genome instability, we correlated the BMI with the number of overall mutations. Pearson correlation analysis showed that a positive correlation existed between the BMI and the numbers of mutated genes (R²=0.718, P<0.001) or simple mutations (R²=0.682, P<0.001) (Fig. 2C). Collectively, these data suggest that the BAF complex mutations are linked with genomic instability of lung cancer types.

Next, we further investigated the overall mutation differences between the mutated-type and wild-type genotyping of each individual BAF gene in the lung carcinoma samples, respectively. More mutated genes or simple mutations significantly appeared in the lung carcinoma sample with the mutated-type genotyping of 6 genes, SMARCA4, ARID2, ARID1B, BCL11A, BCL11B and BRD9 (Fig. 2D). There were no significant differences between the mutated-type and wild-type genotyping of the other BAF genes in the lung carcinoma samples (data not shown). Accordingly, the 6 BAF genes primarily contribute to safeguard genome stability against mutation occurrence.

To evaluate the mutation ratio in vitro, we developed a mutation reporter of lung cancer cells. First, a slippage vector construct contains 17 CA-dinucleotide repeats cloned just downstream from the initiation codon of the luciferase gene, which leads to the luciferase-coding sequence out of frame. However, genome instability resulting in gains of one or loss of two CA repeats will restore the reading frame (Fig. 3A). Then the linearized slippage vector containing the hygromycin resistance gene for selection of stable cell lines was transfected into Calu-3 cells, a lung adenocarcinoma cell line. We cultured Calu-3 cells with selective medium every 3–4 days until hygromycin-resistant colonies were visible, and the 12 colonies were selected for further identification. The results showed that both target fragments of the CMV promoter and Luc element on the slippage vector could be amplified by designed primers from genomic DNA of the 12 colonies. To evaluate the transcript levels of the slippage-Luc, the same primers were used to amplify cDNA of the 12 cell sublines using RT-PCR. The slippage-Luc transcripts were detected by the primers for Luc element in 4 of the 12 cell colonies, where 2 colonies (H10, H12) exhibiting obvious slippage-Luc transcripts were selected for mutation reporters (Fig. 3B).

To confirm the role of BAF inactivation in genome stability in vitro, we selected the most frequently mutated BAF gene, SMARCA4, for the mutation ratio assay as loss of function in the established mutation reporters of Calu-3 cells without SMARCA4 mutations found (data not shown). pSUPER constructs containing shRNA (sh1, sh2, sh3 and sh4) against SMARCA4 were transfected into the two established Calu-3 cell sublines (H10 and H12), respectively. Three days after transfection, we evaluated the RNAi knockdown efficiency of these shRNAs, using quantitative RT-PCR assay. The result showed that the mRNA levels of SMARCA4 were efficiently and significantly decreased to <50% by sh-1 and sh-3, as compared to shNC (Fig. 4A). Subsequently, we screened G418-resistant colonies to establish double-stable sublines with SMARCA4 downregulation, based upon the two Calu-3 cell reporters H10 and H12. Of these double-stable colonies, 5 colonies (G5, G8, G19, G23 and G24) with <50% SMARCA4 expression levels compared with their progenitor cells and 4 colonies (G1, G3, G14 and G15) close to the original SMARCA4 levels in their progenitor cells were used for mutation ratio analysis (Fig. 4B). Furthermore, we performed western blot assays to examine SMARCA4 protein expression in these selected double-stable cells. The SMARCA4 protein level was markedly decreased in the 5 cell sublines with stable SMARCA4 knockdown, as compared with the controls with shNC in the two reporter cell lines (H10 and H12), respectively (Fig. 4C). These reporter cells were expanded for 3 weeks, where luciferase activity was measured and Luria-Delbrück fluctuation analyses were then used to calculate the mutation rate. Upon the pSUPER-mediated SMARCA4 knockdown, the mutation ratios were significantly increased in the mutation reporter cells (Fig. 4D). Collectively, these observations suggest that SMARCA4 knockdown may lead to the increased mutations in human lung cancer.