SOX2 suppresses CDKN1A to sustain growth of lung squamous cell carcinoma

Abstract

Since the SOX2 amplification was identified in lung squamous cell carcinoma (lung SCC), SOX2 transcriptional downstream targets have been actively investigated; however, such targets are often cell line specific. Here, in order to identify highly consensus SOX2 downstream genes in lung SCC cells, we used RNA-seq data from 178 lung SCC specimens (containing tumor and tumor-associated cells) and analyzed the correlation between SOX2 and previously-reported SOX2-controlled genes in lung SCC. In addition, we used another RNA-seq dataset from 105 non-small cell lung cancer cell lines (NSCLC; including 4 lung SCC cell lines) and again analyzed the correlation between SOX2 and the reported SOX2-controlled genes in the NSCLC cell lines (no tumor-associated cells). We combined the two analyses and identified genes commonly correlated with SOX2 in both datasets. Among the 99 genes reported as SOX2 downstream and/or correlated genes, we found 4 negatively-correlated (e.g., CDKN1A) and 11 positively-correlated genes with SOX2. We used biological studies to demonstrate that CDKN1A was suppressed by SOX2 in lung SCC cells. G1 cell cycle arrest induced by SOX2 siRNA was rescued by CDKN1A siRNA. These results indicate that the tumorigenic effect of SOX2 in lung SCC cells is mediated in part by suppression of CDKN1A.

Figure 1. SOX2 positively or negatively correlated genes in the human lung squamous cell carcinoma (SCC) samples from the TCGA.

The Cancer Genome Atlas (TCGA) human lung squamous cell carcinoma (SCC) samples (n = 178) were divided into two groups that have high RNA expression of SOX2 (HIGH, n = 89) or low expression of SOX2 (LOW, n = 89). Two-tailed Welch’s t-test was used to determine the significance of gene expression between the two groups. Genes with significantly higher expression in the HIGH group (p < 0.05 and average expression > 1) were considered as SOX2 positively correlated genes, while genes with significantly higher expression in the LOW group (p < 0.05 and average expression > 1) were considered as SOX2 negatively correlated genes. The expression was measured in RPKM and quantile normalized. PLA2G4B was not present in the TCGA SCC dataset. Results represent the mean ± S.E.M. *indicates p < 0.05, **indicates p < 0.01, and ***indicates p < 0.001.

Figure 2. SOX2 positively or negatively correlated genes in the human non-small cell lung cancer (NSCLC) cell lines.

Human non-small cell lung cancer (NSCLC) cell lines (n = 105) were separated into two groups that have high RNA expression of SOX2 (HIGH, n = 20) or low expression of SOX2 (LOW, n = 85). The expression of previously reported SOX2 downstream or correlated genes was assessed as to whether the expression of those genes was correlated with that of SOX2. Two-tailed Welch’s t-test was used to determine the significance of gene expression between the two groups. Genes with significantly higher expression in the HIGH group (p < 0.05 and average expression >1) were considered as SOX2 positively correlated genes, while genes with significantly higher expression in the LOW group (p < 0.05 and average expression >1) were considered as SOX2 negatively correlated genes. The expression was measured in RPKM and quantile normalized. Results represent the mean ± S.E.M. *indicates p < 0.05, **indicates p < 0.01, and ***indicates p < 0.001.

Figure 3. SOX2 suppresses CDKN1A in human lung SCC cells.

(A) Immunoblot analysis of SOX2, CDKN1A, CDKN1B and TP53 in the indicated lung SCC cell lines. The expression level of β-actin is shown as a control. (B) qPCR demonstrated that CDKN1A mRNA was significantly increased after SOX2 silencing by SOX2 siRNAs (siSOX2 #1 and siSOX2 #2) in EBC2, LK2 and H520 lung SCC cells. Non-targeting siRNA was used as a control (siCtrl). Detection of GAPDH was used for normalization. Results represent the mean ± SD (n = 3). Statistical significance was defined as p < 0.01 (*). (C) Immunoblot analysis shows increased CDKN1A expression 48 hours after SOX2 siRNAs (siSOX2 #1 and siSOX2 #2) transfection in EBC2, LK2 and H520 lung SCC cells. TP53 expression was not changed after SOX2 siRNAs transfection. Non-targeting siRNA was used as control (siCtrl). (D) The qPCR analysis showed that CDKN1A mRNA expression was decreased 48 hours after Ad-CMV/SOX2 (Ad-SOX2) infection at a MOI of 100 in EBC1 and at a MOI of 10 in H226 lung SCC cells. Ad-CMV/Luc (Ad-Luc) did not change CDKN1A mRNA expression in both lung SCC cell lines. Non-treated cells were used as a control (Ctrl). Immunoblot analysis demonstrated that CDKN1A protein was suppressed 48 hours after Ad-SOX2 infection in EBC1 and H226 lung SCC cells. Ad-Luc did not change CDKN1A protein expression in the cells. TP53 expression was not changed in the cells. Statistical significance was defined as p < 0.01 (*).

Figure 4. CDKN1A siRNAs released G1 cell cycle arrest induced by SOX2 siRNA.

Flow cytometry cell cycle analysis of EBC2 and LK2 lung SCC cells transiently transfected with non-targeting siRNA (siCtrl), SOX2 siRNA (siSOX2), CDKN1A siRNA (siCDKN1A) or the mixture of SOX2 and CDKN1A siRNAs (siSOX2 + siCDKN1A). Cells were harvested at 72 hours after siRNA transfection. Indicated in the inset of each panel are the percentages of each phase of cell cycle, where values represent the mean ± SD of triplicate measurements.