| Background and objectives:Lung cancer is the leading cause of cancer death worldwide. Despite advances in early detection and treatment, NSCLC patients still have a poor five year overall survival. The World Health Organization (WHO) has divided it into two major groups based on its biology and treatment:small cell lung cancer and non-small cell lung cancer (NSCLC). NSCLC accounts for more than 85% of all lung cancer cases. For many cancers, treatments for early stage tumors have proven effective while metastatic disease often carries a very poor prognosis Although progress in personalized medicine, metastatic lung cancer still has a poor prognosis, with an overall survival rate of <5% at 5 years. On one hand, it is due to lack of effective screening method; on the other hand, it is due to the fact that the key molecular events that fuel initiation and spread of this disease remains unclear. Therefore, identification and characterization of those key molecular events attracts more and more scientists. During the last 25 years, investigators have identified many of the genetic changes underlying the appearance of NSCLC including mutations of BRAF, KRAS, EGFR, FHIT, HER2/NEU, RB, p16INK4A, and p53. In addition, epigenetic silencing of the p16INK4A and CDH1 also play a role. A study demonstrating the poor survival of patients with 4 epigenetically silenced genes further emphasizes the importance of understanding the contribution of epigenetic mechanisms to NSCLC development.This complex, first discovered in S. cerevisiae, shows strong conservation from yeast to Drosophila to mammals and contains approximately 10-12 components. The human SWI/SNF complex, employing either BRG1 or BRM as a catalytic subunit, alters nucleosome arrangement along DNA in an ATP-dependent manner to modulate transcription. The SWI/SNF complex subunits includes BRG1/SMARCA4, ARID 1 A, PBRM and SNF5/INI1, some of which are tumor suppressor genes and some of which can work in concert with other tumor suppressor genes. Previous studies have indicated that the SWI/SNF chromatin remodeling complex participates in various biological processes, including gene transcription, cell cycle regulation and cell differentiation. The mechanism of cancer development involving the SWI/SNF complex remains unknown and is becoming a hot research area. SWI/SNF loss induced cancer development may be related to DNA repair, cell cycle regulation, transcription, as well as nucleosome positioning. Recent NGS studies have shown consistent mutations in key SWI/SNF complex members including BRG1/SMARCA4, ARID1A, PBM1 and SNF5/INI1 across abroad range of human tumors. This observation has proved especially true for human NSCLCs where mutation frequencies for BRG1 and ARID1A have ranged from 10-20%. Therefore, the significant number of these mutations emphasizes the need to understand the effects of loss of SWI/SNF complex activity during progression of NSCLC.NRF2 (NF-E2-related factor) belongs to the Cap’N’Collar (CNC) family members, contains an evolutionarily conserved basic Leucine zipper (bZIP) region. NRF2 widely found in most organs, its main function is to serve as transcription factors by activating about 200 genes to protect from oxidative stress damage. NRF2 is mainly regulated by KEAP1.Under physical condition, NRF2 mainly exists in the cytoplasm, where KEAP1 Kelch domain binds to its ETGE and DLG motif.KEAP1 has a stronger affinity with ETGE motif than DLG motif. Based on the above observations, McMahon and Tong proposed the hinge-and-latch model:KEAP1 first binds to NRF2 ETGE motif (hinge); once the interaction is established, DLG motif (latch) enters into a nearby uncoupled KEAP1 Kelch domain, thus promoting NRF2 degradation. Therefore, the KEAP1-CUL3-E3 ubiquitin ligase complex precisely regulates NRF2 levels in cells, keeping it at low levels under physiological conditions. Under oxidative stress conditions, several KEAP1’s cysteine residues are covalently modified, causing KEAP1’s conformational changes. NRF2 is released from its low affinity DLG motif, inhibiting NRF2’s degradation. Therefore, NRF2 binding domain on KEAP1 is saturated by NRF2 and newly synthesized NRF2 enter the nucleus.NRF2 form heterodimer with Maf protein family members to promote the expression of downstream genes. By activating those genes, NRF2 can initiate the protective mechanism in cells to promote cell survival. NRF2 could act as a tumor suppressor to prevent cancer. However, recent studies showed NRF2 could protect both normal cells and cancer cells.We have previously shown that re-expression of BRG1 in human cell lines lacking expression of both mutually exclusive ATPases, BRG1/SMARCA4 and BRM/SMARCA2, induces expression of genes often associated with epigenetic silencing. We also observed some overlap between genes activated by BRG1 expression and those activated by treatment with the DNA methyltransferase (DNMT) inhibitor 5dAzaC. However, we did not assess the effects of histone acetylation in this study, another mechanism for gene silencing. Because we only examined a limited number of genes, we could not determine how commonly genes activated by BRG1 expression overlapped with those induced by DNMT inhibition or by HDAC inhibition. To address the question of how BRG1 inactivation contributes to NSCLC development, we carried out a gene expression array analysis on a BRG1/BRM-deficient cell line treated with a DNMT inhibitor, a HDAC inhibitor or infected with an adenovirus expressing BRG1. One caveat from studies using adenovirus infection to express proteins is the significant production of protein by the adenovirus infection. To address this issue, we re-expressed BRG1 in the H522 cell line by gene transfection to minimize overexpression. We used restriction landmark genomic scanning (RLGS) and a published DNA methylation array data to further investigate the relationship between BRG1 loss and DNA methylation.To further investigate the mechanism on BRG1 loss in the development of NSCLC, we used shRNA technology to generate BRG1 and/or BRM-depleted clonal cell lines using the human NSCLC H358 cell line that expresses wild-type BRG1 and BRM genes. Cells with reduced BRG1 expression showed a change in morphology from a tightly packed and cuboidal appearance to an elongated morphology with distinct cellular borders. Parental, Control, and BRG1 knockdown cell lines were inoculated into the lungs of nude mice. Mice with reduced BRG1 expression showed a shortened survival, and tend to for larger tumors, indicating an increased tumorigenic potential. Due to recent next generation sequencing (NGS) studies in lung cancer have shown a significant number of activating mutations in the KEAP1/NRF2/ARE signaling pathway, is there any relation between inactivation of BRG1/BRM and KEAP1-NRF2-ARE signaling pathways? Can BRG1/BRM loss activate KEAP1-NRF2-ARE pathway to induce tumor progression?Method1.In order to evaluate if re-expression of BRG1 and BRM in BRG1/BRM deficient NSCLC cell lines leads to re-expression of epigenetically silenced genes,SW13 or H522 cells were either treated with 5-dAzaC and/or TSA or transfected with the empty expression vector pcDNA3 or expression vectors for BRG1, DNBRG1 or BRM. After 48 hours, protein was extracted, separated and immunoblotted for BRG1, BRM, CDH1 and CK-18.2.To further address the relationships among BRG1 re-expression, DNA methylation and histone acetylation, we carried out a gene expression array analysis on the BRG1/BRM-deficient human H522 NSCLC cell line after infection with Ad-BRG1-GFP or Ad-GFP or treatment with vehicle (DMSO),5μM 5-dAzaC or 100nM TSA.3. One caveat from studies using adenovirus infection to express proteins is the significant production of protein by the adenovirus infection. To address this issue, we re-expressed BRG1 in the H522 cell line by gene transfection to minimize overexpression. We also included a second BRG1/BRM-deficient cell line, A427. H522 and A427 cells were transfected with BRG1 (pBJ5-BRG1) or empty vector (pcDNA3). After 48 hours, cells were harvested for either total RNA or protein. Gene expression (CDH1, CD44, CDH3, EHF and RRAD) was then evaluated by QPCR or by western blotting.4. To further investigate the relationship between BRG1 loss and DNA methylation, we asked whether BRG1 re-expression altered DNA methylation in NSCLC cell lines. H522 cells were transfected with BRG1 (pBJ5-BRG1) or empty vector (pcDNA3). After 48 hours, cells were harvested for DNA and used for RLGS to detect if DNA methylation changes in the 2 BRG1/BRM-deficient NSCLC cell lines after BRG1 re-expression.5. We examined whether the DNA methylation status of these genes would predict their response to BRG1 re-expression i.e. do high levels of DNA methylation inhibit the effects of BRG1 re-expression? To address this issue, we took advantage of a recent study that used DNA methylation arrays to compare the patterns among 69 human NSCLC cell lines including H522 and A4276. We generated BRG1 and/or BRM-depleted clonal cell lines using the human NSCLC H358 cell line that expresses wild-type BRG1 and BRM genes, and shRNA RNAi technology.7. NRF2 target gene protein expression was measured by western blot analysis using whole cell lysates. NRF2 target gene mRNA expression was determined by qPCR using total RNA.8. We generated BRG1 knockout cell lines using the human NSCLC H358 cell line and CRISPR technology. Protein expression of BRG1, BRM, KEAP1, NRF2, HMOX1 and NQO1 was measured by western blot analysis using whole cell lysates.9. We designed primers for functional NRF2 binding sites in the HMOX1 promoter. We then performed ChIP analyses on the BRG1 and/or BRM deficient cell lines, treated with either ethanol (vehicle control) or 75μM tBHQ for six hours. Nucleoprotein complexes were immunoprecipitated with antibodies against NRF2, RNAP Ⅱ, BRG1 or a rabbit IgG control and precipitated DNA was determined by qPCR with oligonucleotide primers complimentary to the ARE sequences of HMOX1 to quantify NRF2 promoter occupancy.10. We next examined whether changes in NRF2 expression in the H358 BRG1 knockdown cells might be related to altered intracellular ROS levels. After 48 hours exposure to 10mM NAC, cells were harvested and protein extracted. Total cellular proteins were separated on a 4-12% SDS-polyacrylamide gel and probed with KEAP1, NRF2, HMOX1 and NQO1 antibodies. ChIP assays were carried out using cross-linked chromatin from H358 control and Brgli.2 cells treated with either vehicle or lOmM NAC for 48 hours. Nucleoprotein complexes were immunoprecipitated with antibodies against NRF2, RNAP Ⅱ, BRG1 or a rabbit IgG control and precipitated DNA was determined by QPCR with oligonucleotide primers complimentary to the ARE sequences of HMOX1 to quantify NRF2 promoter occupancy.11. In order to validate our results obtained with human NSCLC cell lines, we assessed KEAP1-NRF2 signaling in TCGA primary human NSCLC samples using the 15 gene NRF2 target gene signature to assess if low BRG1 expression levels in primary human NSCLC are associated with increased KEAP1-NRF2 signaling. Samples with nonsense or frameshift mutations were used to identify a threshold for low SMARCA4/BRG1 expression.12. Statistical analysisStatistical analysis was performed on SPSS 13.0. Results are shown as mean±standard deviation (x±s). Difference between two groups are compared using student t test. Difference between multiple groups are compared using one way ANOVA. For multiple comparsison test, if the variances are homogeneous, LSD method was used; if the variances are not homogeneous, Dunnett method was used. P<0.05 was considered statistically significant. For RNA-seq data, kruskal-wallis H test was used to compare differences between multiple groups. P<0.05 was considered as statistically significant. For multiple comparsison test, a two-sided Wilcoxon rank sum test was applied. a=0.05/C24=0.0083, P<a was considered statistically significant. R 2.15.1 was used to perform statistical analysis and generate figures.Results1. Re-expression of BRG1 and BRM in BRG1/BRM-deficient NSCLC cell lines leads to re-expression of epigenetically silenced genes.Either treatment with TSA or 5-dAzaC or re-expression with BRG1 or BRM induced expression of CDH1 protein in 2 BRG1/BRM-negative cell lines, SW13, derived from an adrenal carcinoma and H522, derived from a NSCLC. In contrast, treatment with a dominant-negative form of BRG1 (DNBRG1) that lacks ATPase or with vehicle (DMSO) had no effect.2. Analysis of BRG1 re-expression on gene expression in the H522 NSCLC cell line.We first analyzed these data by hierarchically clustering both the genes and the arrays in the expression data and then creating a heat map to look for common patterns of gene expression. The 4 replicates from each treatment group clustered together, which shows consistency of gene expression patterns among the replicates. The groups infected with adenovirus showed greater similarity to each other than to either TSA or 5-dAzaC treatments. The DMSO treatment control showed the least similarity to any of the other treatment groups while TSA and 5-dAzaC treatments showed the most similarity.We first looked for genes that showed changed expression after Ad-BRGl infection. Our results found expression levels for 5527 genes increased and 6510 decreased after BRG1-GFP re-expression normalized to GFP expression alone. However, this number represents an over estimation because some genes showed decreased expression under all treatment conditions. In a similar vein, expression of 2436 genes increased and 2763 genes decreased after 5-dAzaC treatment. In contrast, we observed fewer changes in gene expression after TSA treatment, where 560 genes went up and 995 genes went down.We identified genes whose expression increased after 2 different treatments. We observed 429 genes that showed increased expression after either 5-dAzaC treatment or BRG1 re-expression. However, to be conservative in these analyses, we first focused only on those genes whose expression increased>2 fold compared to the parental cell line. Of these 429 genes,145 genes increased more than 2 fold after BRG1 re-expression including genes. In a similar analysis for genes whose expression increased after TSA treatment or BRG1 re-expression, we found 186 genes.140 genes from this group that went up by more than 2-fold after BRG1 re-expression. We identified genes that showed increased expression after TSA or 5-dAzaC treatment but not BRG1. We found 170 genes, of which 148 increased by at least 2-fold after TSA treatment. Finally, we looked for genes whose expression went up under all three conditions. This group contained the fewest genes, of which 55 genes increased by more than 2 fold after BRG1 re-expression.3. Validation of gene expression array dataTwo independent sample t test was applied. Comparing to control group (H522 pCDNA3), expression of BRG1, CDH1, CDH3, EHF and RRAD was upregulated in BRG1 re-expression group (H522 BRG1) (t=-4.992,P=0.015;t=-7.181 P=0.005;t=-9.311, P=0.003;t=-7.686, P<0.001;t=-12.737, P<0.001) Comparing to control group (A427 pCDNA3), expression of BRG1, CD44, EHF and RRAD was upregulated in BRG1 re-expression group (A427 BRG1) (t=-12.005, P=0.001;t=-5.376, P=0.005; t=-2.521, P=0.045;t=-4.958, P=0.015) Differences are statistically significant. The Western blot results appeared consistent with the QPCR findings.4. Gene expression changes after BRG1 reexpression in BRGl/BRM-deficient NSCLC cells does not correlate with DNA methylation levelsRLGS analysis of H522 cells transfected with vector or BRG1 revealed only three prominent landmarks that appeared after BRG1 re-expression. However, sequencing of these DNAs revealed that they originated from the BRG1 transgene and not from changes in methylation of the H522 cellular DNA. Similar results were observed for the A427 cell line.DNA methylation correlates poorly with altered gene expression after BRG1 re-expression. We took advantage of a recent study that used DNA methylation arrays to compare the patterns among 69 human NSCLC cell lines including H522 and A427. A427 cells displayed significantly more DNA methylation along the length of the CD44 gene than H522 cells. However, BRG1 re-expression induced CD44 expression only in the A427 cell line. In contrast, H522 cells showed significantly less methylation in the promoter region of CDH3 than A427 cells, consistent with BRG1 re-expression inducing its expression in H522 alone. Furthermore, RRAD basal expression was higher in A427 cells despite the presence of significantly more DNA methylation along the entire promoter and coding region. Similarly, we did not observe an association between CDH1 methylation and expression.5. BRG1 and BRM shRNAs inhibits mRNA and protein expression of BRG1 and BRM.One way ANOVA was used. BRG1 and BRM mRNA expression between H358 BRG1/BRM knockdown and control cell lines are different (F=134.901, P<0.001; F=154.963, P<0.001). For multiple comparison, compared to control group, mRNA expression of BRG1 in H358 BRG1 knockdown and BRG1/BRM double knockdown groups are greatly reduced, P<0,05, differences are statistically significant. mRNA expression of BRM in H358 BRM knockdown and BRG1/BRM double knockdown groups are greatly reduced, P<0,05, differences are statistically significant. The Western blot results appeared consistent with the QPCR findings.6. Knockdown of BRG1 and/or BRM modulates expression of NRF2 target genesCompared to control group, BRG1 reduction results in the inactivation of KEAP1 and an increase in NRF2 protein levels in the H358 BRG1 knockdown and double knockdown cell lines. NRF2 mRNA expression remained the same across all the cell lines (F=0.971, P=0.462), indicating the inactivation of KEAP1. In contrast, reduction in BRM expression in parental cells did not appreciably alter their NRF2 or KEAP1 dimer protein levels.One way ANOVA was applied. mRNA expression of HMO1, GSTM4, GCLM and NQO1 between H358 BRG1/BRM knockdown and control cell lines are different (F=36.083,P<0.001; F=50.924,P<0.001; F=30.126,P<0.001; F=57.529,P<0.001).For multiple comparison, compared to control group, mRNA expression of HMOX1 and GSTM4 in H358 BRG1 knockdown and BRG1/BRM double knockdown groups are greatly increased, P<0.05, differences are statistically significant; mRNA expression of HMOX1 and GSTM4 in H358 BRM knockdown group maintained the same, P=0.154 and 0.259, differences are not statistically significant. Compared to control group, mRNA expression of GCLM and NQO1 in H358 BRM knockdown group was increased, P<0.05, differences are statistically significant; mRNA expression of GCLM in H358 BRG1 knockdown (Brgli.2) and BRG1/BRM double knockdown (Brgli.2brm) was reduced, P<0.05. The Western blot results appeared consistent with the QPCR findings.Consistent with shRNA study, NRF2, KEAP1 dimer and HMOX1 protein expression was increased, but NQO1 protein expression remained unchanged after BRG1 was knocked-out, using CRISPR technology.7. Reduced BRG1 expression impacts NRF2 binding to HMOX1 ARE sequences.In response to tBHQ treatment, recruitment of NRF2 to the regulatory regions of its targets genes was greatly increased, however, NRF2 does not bind to transcription starting site (TSS) and 500bp downstream from TSS(+500bp). We observed a significant increase in NRF2 binding to HMOX1 enhancers (EN1 and EN2) in H358 BRG1 knockdown cells and double knockdown cells either treated or untreated with tBHQ. When untreated with tBHQ, compared to H358 parental and control cell lines, recruitment of RNA polymerase Ⅱ to its regulatory regions (EN2, TSS and +500bp) was greatly increased (F=27.947, P<0.001; F=19.001, P<0.001; F=16.996, P<0.001). When treated with tBHQ, compared to H358 parental and control cell lines, recruitment of BRG1 to its regulatory regions (EN2, TSS and +500bp) was also increased (F=30.788, P<0.001; F=52.721, P<0.001; F=22.443, P<0.001) When untreated with tBHQ, compared to H358 parental and control cell lines, recruitment of BRG1 to its regulatory regions (EN2, EN1, TSS and +500bp) was greatly reduced (F=8.875, P=0.002; F=19.817, P<0.001; F=7.123, P=0.005; F=4.195, P=0.030); When treated with tBHQ, compared to H358 parental and control cell lines, recruitment of BRG1 to its regulatory regions (EN2, EN1, TSS and +500bp) was also redcued (F=40.734, P<0.001; F=25.318, P<0.001; F=11.231, P=0.001; F=6.057, P=0.009)8. Loss of BRG1 causes increased ROS productionAfter NAC treatment, western blot and ChIP were performed. Western blot revealed that expression of HMOX1 and KEAP1 dimer were reduced in BRG1 knockdwon group, whereas control group remained the same. When untreated with NAC, compared to control cell line, recruitment of NRF2 to its regulatory regions was greatly increased (t=-6.061, P=0.001;t=-3.128, P=0.020). When treated with NAC, compared to control cell lines, recruitment of NRF2 to its regulatory regions was almost the same (t=-0.059, P=0.955;t=-0.638, P=0.558). When untreated with NAC, compared to control cell line, recruitment of RNA polymerase Ⅱ to its regulatory regions (EN2,TSS,+500bp) was greatly increased (/=-7.048, P=0.004;t=-3.139, P=0.020;t=-4.736, P=0.003). When treated with NAC, compared to control cell lines, recruitment of RNA polymerase Ⅱ to its regulatory regions was almost the same (t=-0.627, P=0.553;t=0.301, P=0.774;t=1.581, P=0.165) When untreated with NAC, compared to control cell line, recruitment of BRG1 to its regulatory regions (EN2, EN1, TSS,+500bp) was greatly reduced (t=5.193, P=0.002;t=7.668, P<0.001;t=5.436, P=0.002; t=4.101, P=0.006).When treated with NAC, compared to control cell lines, recruitment of BRG1 to its regulatory regions was also reduced (t=5.371, P=0.002;t=5.348, P=0.002;t=3.832, P=0.009;t=6.299, P=0.001)9. Low BRG1 expression levels in primary human NSCLC are associated with increased KEAP1-NRF2 signalingKruskal-wallis H test was applied to test if KEAP1-NRF2-ARE signaling activity was different between groups. (H=95.881, P<0.001, differences are statistically significant). Wilcoxon rank rum test was used to do multiple comparison. Samples with KEAP1/NRF2 mutations (group 2 vs group 1) and with low BRG1 expression (group 3 vs group 1) display increased NRF2 signaling activity compared to the tumors with wild-type genes (group 1). P is less than modified a, differences are statistically significant. Compared to samples with KEAP1/NRF2 mutations (group 2), NRF2 signaling activity in samples with low BRG1 expression and KEAP1/NRF2 mutations (group 4) display an increased tendency, but not statistically significant due to its small sample size (n=4). Two way Wilcoxon rank sum test’s P value is 0.013, greater than modified a, but one way Wilcoxon rank sum test’s P value is 0.006, less than modified a.Conclusion1. Epigenetic silencing induced by BRG1 loss during NSCLC development does not correlate with DNA methylation and is an independent mechanism for gene silencing.2. Reexpression of BRG1 in BRG1/BRM deficient NSCLC cell lines leads to reexpression of epigenetically silenced genes, i.e.CDH1, CDH3, CD44, EHF and RRAD.3. In human NSCLC cell line, loss of BRG1 activates the KEAP1-NRF2-ARE pathway.4. In primary human NSCLC, loss of BRG1 activates the KEAP1-NRF2-ARE pathway. Therapeutic approaches that inhibit NRF2 or NRF2 target genes might benefit BRG1 deficient NSCLC patients. |
PDF Full Text Request