SOX2 and cancer: current research and its implications in the clinic

Tumor development occurs when a cell accumulates genetic alterations that modify normal cell cycle progression. There are several means of genetic aberrations including gene deletions or gene amplification. Gene amplification is defined as a copy number increase of a particular chromosomal region. SOX2 amplification is due to multiplication of the 3q26.3 gene locus [12, 13]. SOX2 amplification has previously been found in several cancer types including glioblastoma, small-cell lung cancer (SCLC) and many forms of squamous cell carcinoma (SCC) [12, 14‐20]. A summary of SOX2 amplification in various cancer types with corresponding references can be found in Table 1.

The latest SOX2 studies have focused on the co-amplification of SOX2 with other critical genes. Justilien and colleagues revealed the co-amplification of two oncogenes, PRKCI and SOX2, is responsible for the cancer stem cell phenotype seen in lung squamous cell carcinoma (LSCC) [61]. Moreover, another study performed FISH analysis in 447 resected non-small cell lung cancer (NSCLC) tissue samples and SOX2 amplification was associated with increased gene copy number of FGFR1 and PI3KCA genes [62].

As described by Hanahan and Weinberg, cancer is a disease characterized by determined hallmarks some of which are: sustained proliferative signaling, activation of invasion and metastasis, and evasion of cell death [73]. Studies have strongly associated SOX2 to these respective cancer hallmarks and thus far SOX2 has been shown to promote cellular proliferation (breast, prostate, pancreatic and cervical cancers) [21, 28, 57, 58], evade apoptotic signals (prostate, gastric cancer and NSCLC) [37, 58, 63] and promote invasion, migration and metastasis (melanoma, colorectal, glioma, gastric, ovarian cancer and hepatocellular carcinoma) [15, 29, 47, 49, 55]. We summarized SOX2 amplification and resulting alterations in cellular functions in all cancer types in Table 1 and showed examples of SOX2’s role in oncogenic signaling in Figure 2. Below we highlight a few functional examples of SOX2 in cancer before we review the latest SOX2 research in different aspects of cancer physiology including: cellular proliferation, apoptosis and invasion/migration/metastasis. For a complementary and closer examination into SOX2 and oncogenic signaling, protein-protein interactions and miRNAs see review by Liu et al.[74].

Cellular proliferation is tightly regulated by SOX2 in many cancer types. SOX2 knockdown in pancreatic cancer cells resulted in cell growth inhibition through cell cycle arrest, not apoptosis, via the transcriptional induction of p21^Cip1 and p27^Kip1[57]. When SOX2 was overexpressed, cell proliferation was promoted through cyclinD3 (CCND3) transcriptional induction and allowed further S-phase entry (Figure 2B) [57]. Recently, Hütz and colleagues confirmed this in gastric cancer [37]. SOX2 was functionally inhibited via cellular transfection with a tetracycline-inducible C-terminally truncated version of SOX2, termed dnSOX2. Despite lacking the transactivation domain, dnSOX2 could bind SOX2 recognition sites on DNA and compete with wild-type SOX2. This inhibition led to the decrease in cellular proliferation in AZ-521 cells and further analysis using a RNA gene expression microarray, revealed an upregulation of p21 and downregulation of Δp63 (splice variant of p63) [37]. Additionally, Fang and colleagues found in LSCC, SOX2-silencing inhibited cellular proliferation via the upregulation of BMP4 [64]. After performing chromatin immunoprecipitation and luciferase experiments, SOX2 was found to transcriptionally repress the BMP4 promoter. The authors therefore suggest that BMP4 is playing a tumor suppressor role in LSCC, while SOX2 repression of BMP4 transcription causes cell growth [64]. It’s important to note that the involvement of SOX2 in cell proliferation has been controversially discussed in colorectal and gastric cancer [30, 31, 37, 38]. The contradictory effect of SOX2 in cell proliferation suggests that SOX2 plays a differential role depending on the type of cancer (Figure 2B).

SOX2 also plays an important role in evading apoptotic signals. In prostate cancer, in vitro and in vivo xenograft experiments using DU145 SOX2-overexpressing cells in NOD/SCID mice revealed that SOX2 caused an increase in apoptotic resistance by decreasing store-operated Ca²⁺ entry via repressing ORAI1 expression (Figure 2) [58]. Equivalently, upon silencing of SOX2 in NSCLC cell lines, apoptosis was induced [63]. Hütz and colleagues, in addition to proliferation analysis, investigated apoptosis using, caspase 3/7 assays performed in AZ-521 gastric cancer cells [37]. After 48 hours of SOX2 inhibition, nearly 60% of cells were apoptotic compared to only 20 – 40% of the control cells [37]. These studies prove SOX2’s involvement in apoptosis inhibition and therefore in direct promotion of uncontrolled cell growth.

Finally, research has indicated that SOX2 is a novel regulator of cell invasion, migration and metastasis. For example in melanoma, SOX2 knockdown in A2058 cells resulted in a 4.5-fold decrease in invasion in vitro and adopted this phenotype via the upregulation of matrix metalloproteinase (MMP)-3 [49]. Likewise in colorectal cells, SOX2 was involved in cellular migration and invasion in vitro, but mediated these effects through MMP-2 [29]. This invasive phenotype was also confirmed in malignant glioma, since siRNA-mediated downregulation of SOX2 resulted in a significant decrease in migration and invasion capabilities [15]. Moreover SOX2 overexpression in the SOX2-negative glioma cell line U-87 resulted in a significant increase in the number of migratory and invasive cells [15]. Numerous gain and loss of function studies in several cancer types (gastric cancer, ovarian cancer and hepatocellular carcinoma) reinforced the link between SOX2 and cellular invasion and migration [15, 37, 47, 55]. Recently, Yang and colleagues showed SOX2’s involvement in promoting invasion and migration in laryngeal cancer cells through the induction of MMP-2 and the PI3K/AKT/mTOR pathway (Figure 2A) [48].

Many tumors are derived from a single cell that has undergone malignant transformation through the acquisition of genomic aberrations, e.g. gene amplification, mutations or other mechanisms [75]. Clonal expansion of cells with tumorigenic properties is the next step towards tumor initiation. These selected cells have the ability to evade normal cell cycle checkpoints, rapidly proliferate and invade tissues [73, 75]. The orchestration of tumor initiation and maintenance has been shown in some cancers to be driven by cancer stem cells (CSCs), also termed tumor-initiating cells or cancer stem-like cells. These CSCs may acquire tumor-initiating and self-renewal properties through similar molecular mechanisms governing cellular reprogramming [76]. Evidence has linked induced cellular reprogramming to cancer and led to the assumption that CSCs may arise via a reprogramming-related mechanism [77‐79]. The reactivation of stem cell-associated markers or pluripotency factors may cause dedifferentiation and a more stem cell-like state [76]. Sussman and colleagues discovered that the ubiquitin-specific protease 22 (USP22) is responsible for controlling the cellular transition from stemness towards differentiation [80]. Moreover they found USP22 represses the SOX2 promoter in order to control the embryonic stem cell transition from self-renewal to differentiation [80] Therefore, not only is SOX2 an essential stem cell marker but its suppression is mandatory for cellular differentiation. For these reasons, SOX2 has been heavily investigated in CSCs in several cancer types.

SOX2 has shown to increase CSC markers in ovarian, pancreatic, lung cancer, but research also has proven its function in self-renewal [57, 81, 82]. Self-renewal capacity of a CSC is critical and enables the maintenance of the CSC subpopulation within a tumor and SOX2 has shown to mediate this in breast, gastric, ovarian, prostate cancer, glioma, osteosarcoma, lung adenocarcinoma and NSCLC [15, 42, 54, 59, 65, 66, 77, 82, 83]. Studies in gastric cancer using siRNA-mediated SOX2 knockdown, found reduced spheroid colony formation and increased apoptosis within sphere cells, highlighting the importance of SOX2 in self-renewal capacity [83]. In prostate cancer stem cells, the activation of EGFR signaling increased SOX2 expression and the self-renewal capacity [59]. Singh and colleagues showed similar results in NSCLC, when siRNA-mediated SOX2 knockdown led to a 2.5-fold reduction in sphere formation [65]. Moreover, EGFR/Src/Akt signaling influenced SOX2 protein expression since during EGFR or SRC inhibition using gefitinib or BIBW, respectively, levels of SOX2 were considerably decreased [65]. Taken together, SOX2 mediates self-renewal of CSCs through EGFR signaling in at least two cancer types and is a major mediator of self-renewal in several cancers through mechanisms that remain unclear.

Recent CSC studies have concentrated on SOX2 and its mechanisms in cancer stem cell maintenance and regulation. Santini and colleagues investigated SOX2 in melanoma initiating cells and Hedgehog-GLI (HH-GLI) signaling [50]. They functionally showed that the ectopic expression of SOX2 in vitro caused enhanced self-renewal capacity in melanoma cells. Moreover, GLI1 and GLI2 downstream transcription factors of HH-GLI Signaling were able to bind to the proximal promoter of SOX2 in primary melanoma cells in chromatin immunoprecipitation (ChIP) studies and therefore SOX2 is regulated by HH signaling [50]. To a similar extent, Justilien and colleagues heavily investigated Hedgehog (Hh) acyltransferase (HHAT) in LSCC and found that not only are PRKCI and SOX2 coamplified and cooperate in LSCC but also SOX2 becomes phosphorylated by Protein Kinase Cι (PKCι) [61]. Next, phosphorylated SOX2 is recruited and required for HHAT expression and therefore maintenance of the stem-cell phenotype. The authors provide evidence that the coamplifcation of both oncogenes is required to activate the PKCi-SOX2-HHAT signaling axis which impels the stem cell phenotype. Further work is being done on establishing PKCι inhibitors for LSCC treatment [61].

Lastly, a recent publication by Favaro and colleagues proved that SOX2 is required for in vitro CSC maintenance in a high-grade oligodendroglioma mouse model [43]. Oligodenrogliomas were generated in mice via the transduction of PDGF-B-IRES-GFP-encoding retrovirus within the brain at embryonic day 14.5. Additionally these embryos carried a homozygous SOX2^flox mutation, which allowed authors to excise SOX2 using lentiviral Cre recombinase virus. Wild-type oligodendroglioma cells from mentioned mouse model were transplanted into the brain of C57/Bl6 mice, which generated lethal tumors. However, when SOX2-deletion cells were transplanted into C57/Bl6 mice they remained tumor free [43]. This elegant study proved the obligatory function of SOX2 in oligodendroglioma tumor initiation. The requirement of SOX2 in oligodendroglioma suggested possible therapeutic intervention.