Critical role of SOX2–IGF2 signaling in aggressiveness of bladder cancer

Signaling elicited by the stem cell factors SOX2, OCT4, KLF4, and MYC not only mediates reprogramming of differentiated cells to pluripotency but has also been correlated with tumor malignancy. In this study, we found SOX2 expression signifies poor recurrence-free survival and correlates with advanced pathological grade in bladder cancer. SOX2 silencing attenuated bladder cancer cell growth, while its expression promoted cancer cell survival and proliferation. Under low-serum stress, SOX2 expression promoted AKT phosphorylation and bladder cancer cells’ spheroid-forming capability. Furthermore, pharmacological inhibition of AKT phosphorylation, using MK2206, inhibited the SOX2-mediated spheroid formation of bladder cancer cells. Gene expression profiling showed that SOX2 expression, in turn, induced IGF2 expression, while SOX2 silencing inhibited IGF2 expression. Moreover, knocking down IGF2 and IGF1R diminished bladder cancer cell growth. Lastly, pharmacological inhibition of IGF1R, using linsitinib, also inhibited the SOX2-mediated spheroid formation of bladder cancer cells under low-serum stress. Our findings indicate the SOX2–IGF2 signaling affects the aggressiveness of bladder cancer cell growth. This signaling could be a promising biomarker and therapeutic target for bladder cancer intervention.

In this study, we found SOX2 is a prognostic marker in bladder cancer patients, signifying poor survival. We also found SOX2 promotes AKT phosphorylation in bladder cancer cells by inducing IGF2 and IGF1R expression

Results
SOX2 expression is correlated with tumor malignancy in bladder cancer. Because factors in ESC signaling and iPSC reprogramming have been linked to tumor malignancy, we used the Cox's proportional hazards model to analyze the link between SOX2, KLF4, MYC and OCT4 expression and recurrence-free survival outcome for bladder cancer patients (Fig. 1a). Both univariate and multivariate regression analyses revealed that only SOX2 expression correlated with poor recurrence-free survival (Fig. 1a, and Supplementary Table 1). Boxand-whisker plots showed that SOX2 expression was also associated with advanced tumor grade of bladder cancer (Fig. 1b). Immunohistochemistry was used to verify SOX2 expression in primary bladder tumors, which showed SOX2 expression was high in tumors with "poorly differentiated" malignant grade (Fig. 1c). These data highlight SOX2's potential involvement in bladder cancer tumor malignancy.
SOX2 regulates the growth of bladder cancer cells. Assessing SOX2 expression in bladder cancer cell lines showed its expression was considerably lower in T24 cells than in 5637 cells (Supplementary Figure S1). To investigate its role in bladder cancer oncogenesis, SOX2 was ectopically expressed in T24 cells using the lentiviral transduction system, and its expression was confirmed with immunoblotting and qPCR (Fig. 2a left). Trypan blue cell exclusion and alamarBlue proliferation analysis showed that SOX2 expression promoted cell proliferation (  Figure S2a). Because 5637 represents a bladder cancer cell line with high SOX2 expression, we adopted the lentiviral shRNA system to knock down SOX2 in 5637 cells to further investigate the effect of eliminating SOX2 function. qPCR and immunoblotting assays indicated that endogenous SOX2 mRNA expression was suppressed by shSOX2 (Fig. 2b left). The trypan blue cell exclusion test, alamarBlue proliferation assay, and cell cycle analysis revealed that silencing SOX2 in 5637 cells inhibited cell proliferation due to S-phase arrest during cell cycle progression (Fig. 2b right and Supplementary Fig. S2b,c). In addition, clonogenic assays showed SOX2's ectopic expression increased T24 cells' colony-forming capability, whereas knockdown of SOX2 in 5637 cells weakened colony formation. (Fig. 2c). This suggests SOX2 expression promotes bladder cancer cell growth.

SOX2 promotes the survival of bladder cancer cells by activating AKT signaling. To test whether
SOX2 plays a role in cell survival, we assessed SOX2 expression in T24 cells under a low-serum stress. Clonogenic analysis showed that SOX2 expression promoted T24 cell growth under a low-serum (1% FBS) condition (Fig. 3a). We further validated the effect of SOX2 expression on T24 cell-spheroid formation under low-serum stress. The T24 cells formed spheroids in a 3D culture system under the normal-serum (10% FBS) condition, wherein SOX2 expression did not affect spheroid formation (Fig. 3b). By contrast, long-term culturing of T24 spheroids under low-serum condition (1% FBS) attenuated the size of the spheroids; however, SOX2 expression sustained the T24 spheroid-forming capability under the low-serum condition, indicating SOX2 is involved in bladder cancer cell survival (Fig. 3b). In addition, the cell cycle analysis revealed that SOX2 expression sustained the S-phase in T24 cells under the low-serum condition ( Fig. 3c and Supplementary Figure S2c bottom left). These findings suggest that SOX2 expression helps bladder cancer cells overcome low-serum stress.
Because AKT signaling contributes to cell survival, we examined SOX2's effect on AKT (Ser473) phosphorylation. Immunoblotting analysis revealed that SOX2 expression in T24 cells sustained AKT phosphorylation under serum-free and low-serum (1% FBS) conditions (Fig. 3d). Moreover, knockdown of SOX2 decreased AKT phosphorylation in 5637 cells (Supplementary Figure S3a). To further verify whether AKT activation mediates SOX2-induced bladder cancer survival, we pharmacologically inhibited AKT activation with MK2206. MK2206 treatment reduced the AKT phosphorylation in 5637 cells as well as in SOX2-expressing T24 cells (Fig. 3e, Supplementary Figure S3b). Clonogenic and 3D colony-forming assays revealed MK2206 also reduced the colony numbers and sizes with respect to the SOX2-expressing T24 cells (Fig. 3f). These results highlight SOX2's activation of AKT signaling to boost bladder cancer cell survival.

IGF2 expression is regulated by SOX2 under an epigenetic control.
To identify the genes responsible for SOX2-mediated bladder cancer cell survival, gene expression profiling analysis was performed using the SOX2-expressing T24 cells and control T24 cells, and the result was uploaded as GSE145826. qPCR analysis confirmed IGF2 levels were considerably higher in SOX2-expressing T24 than in control T24 cells (Fig. 4a). Similarly, SOX2-high 5637 cells exhibited higher IGF2 level than SOX2-low T24 cells, while SOX2 silencing in 5637 cells inhibited IGF2 expression (Fig. 4b). Because SOX2 and IGF2 expression has been linked to embryonic development, we examined SOX2 and IGF2 expression in ESC and fibroblasts. RNA-seq and qPCR assays showed that both SOX2 and IGF2 levels were downregulated in the differentiated fibroblasts compared to ESC ( Fig. 4c and Supplementary Figure S4). Since SOX2 mediates stem cell differentiation and iPSC reprogramming mainly via epigenetic regulation, whereby H3K4me3 marks sites with active gene expression 35,36 , we examined whether H3K4me3 signal on the IGF2 locus is affected by SOX2 expression in bladder cancer cells. ChIP-seq analysis showed a strong H3K4me3 signal on the IGF2 locus in ESC but not in fibroblasts (Supplementary Figure S5).
(Ctrl). Clonogenic analysis (right) to assess the SOX2 knockdown effect on the colony-forming ability in 5637 cells transduced with the lentiviral vector encoding shSOX2 or scrambled control vector (SC). Colonies were subjected to crystal violet staining and quantified by ImageJ analysis. Results are the average of three replicates and expressed as the mean ± S.D. *P < 0.05, **P < 0.01. (2020) 10:8261 | https://doi.org/10.1038/s41598-020-65006-z www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/ ChIP-qPCR analysis revealed the H3K4me3 signal on the IGF2 locus increased because of SOX2 expression in T24 cells and decreased on account of SOX2 knockdown in 5637 cells (Fig. 4d). We also found IGFBP1 expression decreased because of SOX2 expression in T24 cells and increased on account of SOX2 knockdown in 5637 cells (Fig. 4e). These results indicate SOX2 expression regulates IGF2 signaling.
IGF2/IGF1R signaling is essential for SOX2-mediated bladder cancer cell survival. To understand whether IGF2/IGF1R signaling is responsible for high-SOX2 expressing bladder cancer cell growth, we knocked down IGF2 and IGF1R (Fig. 5a-d). Clonogenic assays showed that IGF2 and IGF1R knockdown weakened the colony-forming capability of the high-SOX2 expressing 5637 cells (Fig. 5b,d). Immunoblotting assays revealed that AKT phosphorylation decreased on account of knockdown of IGF2 or IGF1R in 5637 cells (Fig. 5a,c). Linsitinib, an IGF1R inhibitor, was used to block IGF2/IGF1R signaling-mediated AKT phosphorylation in SOX2-expressing bladder cancer cells. Immunoblotting assays revealed that linsitinib inhibited AKT (a) and SOX4 (b) expression with recurrence-free survival in patients from TCGA_BLCA cohort. The significance was examined by log-rank test. *P < 0.05. (c) Kaplan-Meier analysis to assess the correlation of SOX2-high/SOX4-low and SOX2-low/SOX4-high signatures with recurrence-free survival in patients from TCGA_BLCA cohort. The significance was examined by log-rank test. *P < 0.05. (d) Kaplan-Meier analysis of IGF1R expression with recurrence-free survival in patients from TCGA_BLCA cohort. The significance was examined by log-rank test. **P < 0.01. (e) Kaplan-Meier analysis to assess the correlation of SOX2-high/IGF1Rhigh and SOX2-low/IGF1R-low signatures with recurrence-free survival in patients from TCGA_BLCA cohort. The significance was examined by log-rank test. ***P < 0.001. (f) The schematic model of SOX2-mediated oncogenesis of bladder cancer. SOX2 induces the expression of IGF2 and IGF1R, but suppresses that of IGFBP1, thus promoting AKT phosphorylation, with increased proliferation and survival of bladder cancer cells.

IGF1R signaling serves as a prognostic biomarker for bladder cancer.
Our present results confirmed that SOX2 regulates IGF2/IGF1R signaling in bladder cancer cells. SOX4, another SOX family member, is reported to be amplified in bladder cancer 14,15 . We confirmed that SOX4, but not SOX2, was mainly amplified in primary bladder tumors (Supplementary Figure S6). To further examine the potential of SOX2, SOX4, and IGF1R signaling in bladder cancer prognosis, we correlated the expression of these molecules with recurrence-free survival in primary bladder tumors. Kaplan-Meier analysis showed that SOX2, but not SOX4, expression correlated with a poor recurrence-free survival in bladder cancer patients, and the patients harboring a SOX2-high/SOX4-low signature had a worse recurrence-free survival outcome than those with SOX2-low/SOX4-high signature (Fig. 6a-c). Moreover, SOX4, but not SOX2, correlates with a good overall survival, and patients harboring the SOX4-low/SOX2-high signature show a worse overall survival than those with the SOX4-high/SOX2-low signature (Supplementary Figure S7). Furthermore, IGF1R expression was also associated with a poor recurrence-free survival outcome (Fig. 6d). We observed that the patients harboring a SOX2-high/IGF1R-high signature had a worse recurrence-free survival outcome than those with SOX2-low/IGF1R-low signature (Fig. 6e). Both univariate and multivariate regression assays of IGF signaling molecules showed that IGF1R is the only independent predictor of poor recurrence-free survival (Table 1). These findings suggest that IGF1R can be a potential biomarker for predicting poor survival outcomes.

Discussion
Although signaling elicited by the stem cell factors SOX2, OCT4, KLF4, and MYC has been associated with cancer progression in several tumors, it has remained unclear how these signaling molecules mediate bladder cancer progression. We found that SOX2, but not OCT4, KLF4, or MYC expression, correlates with poor prognosis and histologic differentiation in bladder cancer. Moreover, we found that SOX2 promotes bladder cancer cell survival by inducing the IGF2/IGF1R pathway, thereby activating AKT survival signaling. Pharmacological inhibition of IGF1R or AKT inhibits bladder cancer cell survival. Our findings provide insights on SOX2-mediated oncogenesis in bladder cancer and highlight putative therapeutic targets.
Because the introduction of SOX2, OCT4, MYC, and KLF4 is sufficient to reprogram differentiated cells into iPSC with ESC properties, the combined expression of these factors has been hypothesized to initiate tumors and promote cancer progression [37][38][39][40] . On the one hand, tumors harboring ESC-like gene expression signature with active targets of SOX2, OCT4, and MYC were associated with poor pathological differentiation and poor prognosis in brain, breast, and bladder cancer patients 6 . On the other hand, tumor malignancy has been attributed only to MYC and not to other ESC factors 41 . Here, we observed that only SOX2 expression correlates with poor recurrence-free survival of bladder cancer patients. Moreover, we found that SOX2 expression is associated with poor pathological differentiation, emphasizing its involvement in bladder cancer malignancy. This is in accordance with the results of Ruan et al., who reported that SOX2 and Ki67 expression in T1 early stage, non-muscle, invasive bladder tumors correlate with poor recurrence-free survival 18 . Currently, SOX2 is thought to be a cancer stem cell marker in relation to bladder tumors and Sox2 knockout within primary invasive bladder cancer caused enhanced tumor regression 42 . SOX4, another SOX family member, is overexpressed in bladder tumors harboring the genomic 6p22 amplification 15,43,44 . Here, we observed that SOX4 is mainly amplified in primary bladder tumors, whereas only SOX2 expression is associated with poor recurrence-free survival in patients with bladder cancer. SOX2 expression did not affect SOX4 levels in T24 cells (Supplementary Figure S8). Together, these data suggest that SOX4 and SOX2 play distinct roles in tumor initiation and progression. www.nature.com/scientificreports www.nature.com/scientificreports/ The molecular mechanism linking SOX2 expression to poor prognosis in bladder cancer has not been well understood. We found that SOX2 expression not only promotes cell proliferation but also enhances cell survival under low-serum stress. We discovered that, under low-serum stress, SOX2 expression induces AKT phosphorylation and sustains bladder cancer cells' spheroid-forming capability. In many cancers, AKT phosphorylation promotes cell survival by inducing drug resistance and desensitizing radiation therapy. We observed that pharmacological inhibition of AKT phosphorylation attenuates bladder cancer cells' SOX2-mediated survival and spheroid-forming capability. Furthermore, we identified IGF2 and IGF1R as AKT upstream molecules which induce AKT phosphorylation in SOX2-positive bladder cancer cells. IGF2 expression boosts cancer cell survival and tumor progression in colon cancer 45,46 . To search for SOX2-binding targets in colon cancer cells, Fang et al. performed ChIP-seq analysis and identified IGF1R as one of direct binding targets of SOX2 47 . We observed that SOX2 expression in T24 cells increased IGF1R expression, whereas SOX2 knockdown in 5637 cells did not affect IGF1R expression (Supplementary Figure S9). In contrast, SOX2 expression increased IGF2 expression, and SOX2 knockdown decreased IGF2 level, supporting IGF2 as a potential downstream target of SOX2. Our ChIP-qPCR analysis did not show SOX2 binding to the IGF2 locus (data not shown). In addition to directly binding to target genes, SOX2 regulates gene expression via histone modifications during stem cell differentiation and lung cancer cell plasticity 16,48 . ChIP-qPCR analysis showed that the signal of H3K4me3, a predominant mark of active promoters, increased upon SOX2 expression in bladder cancer cells. Knockdown of IGF2 and IGF1R attenuated the growth of bladder cancer cells. All these data verify SOX2 promotes bladder cancer cell growth and survival by inducing IGF2/IGF1R signaling.
Currently, limited targeted therapies are available for treating this aggressive disease. Since our evidence suggests that SOX2 activates AKT survival signaling and promotes spheroid-forming ability by inducing IGF2/IGF1R in bladder cancer cells, IGF2 and IGF1R may be potential therapeutic targets for treating bladder cancer. IGF2 expression promotes aggressiveness in lethal prostate cancer and has been proposed to be a druggable target 31 . Indeed, we found that pharmacological inhibition of IGF2/IGF1R signaling with linsitinib decreased AKT phosphorylation and attenuated bladder cancer cells' SOX2-mediated colony-forming ability. These findings highlight IGF2/IGF1R's potential as therapeutic targets to treat bladder cancer. Moreover, we observed that IGF1R expression correlates with poor recurrence-free survival in bladder cancer patients. The tumors harboring high-SOX2/ high-IGF1R signature are associated with the worst survival outcome in bladder cancer patients. These data suggest IGF1R may serve as a novel biomarker for monitoring bladder cancer progression.
The limitation of this study is that we mainly used T24 and 5637 cells, which represent muscle invasive bladder cancer cell lines maintained in different media, to dissect the potential oncogenic role of SOX2 in bladder cancer. Thus, these data may not be sufficient to represent all bladder cancer. Through gene expression analysis, 5637 cells have been classified into basal subtype bladder cancer, which displays cancer stem cell markers and is associated with poor prognosis [49][50][51] . SOX2 has been characterized as s a marker for stem-like tumor cells in bladder cancer 42 . Hence, our findings that SOX2 activates IGF2/IGF1R signaling and predicts poor prognosis further link IGF2/ IGF1R signaling to SOX2 mediated aggressiveness in bladder cancer.
In conclusion, we demonstrated that SOX2 stimulated IGF2 expression to activate AKT signaling, enhancing the survival and spheroid-forming capability of bladder cancer cells. Moreover, we observed SOX2 and IGF1R levels are significantly correlated with poor prognosis in bladder cancer patients. The fact that SOX2-IGF2/IGF1R signaling axis confers aggressiveness in bladder cancer suggests that IGF2/IGF1R may serve as therapeutic targets in treating bladder cancer.
Trypan blue cell exclusion assay. Cells (10 5 cells/well) were seeded in 24-well plates and incubated in triplicate tests. In indicated times, cells were trypsinized and then counted using Trypan Blue staining in low magnification under the microscope.
Cell-cycle analysis by flow cytometry. Cells were plated at a density such that they would be 50% confluent on the day of analysis. Cells were trypsinized and fixed in cold 70% ethanol for 10 min and then stained with 3D Colony-forming assay. 60-mm plastic tissue-culture petri plates were coated with 0.7% agarose, and 200 cells/plate were seeded. The seeded cells were then placed into the incubator and incubated for 7 days to form unattached floating spheroid colonies. The floating spheroid colonies were then subjected to the indicated treatment for another 7 days. The volumes of spheroid colonies were calculated by using the formula V = (4/3)*pi*r 3 , where the radius r was measured by the microscope.
ChIP-qPCR. ChIP-qPCR was performed using the protocol as previously described 16 , with the antibodies against H3K4me3 (GTX128954, GeneTex) or control IgG (GTX35035, GeneTex). For the detailed RT-primer and probe sequences used in ChIP-qPCR, see Supplementary Table S2. Public domain data analysis. Public gene expression profiling datasets of bladder cancer patients were accessed from TCGA_BLCA database via UCSC cancer browser, and GSE32894 and GSE73211 via GEO browser. The means of SOX2, SOX4, KLF4, MYC, OCT4, and IGF1R expression were used as cut-off points for statistics analysis. The ChIP-seq results of H3K4me3 epigenetic regulation in embryonic stem cells and fibroblasts were collected from ENCODE (https://www.encodeproject.org/). The genetic alteration rates of SOX2 and SOX4 in bladder cancer were analyzed from cBioPortal (http://www.cbioportal.org/). For the used public domain databases, see Supplementary Table S4.
Statistical analysis. The attribution of different gene expression and clinical associated variables in death risks of bladder cancer were calculated using the Cox's proportional hazards regression analysis and demonstrated in forest plot or tables. Recurrence-free survival and overall survival curves of bladder cancer patients were analyzed by the Kaplan-Meier plots and compared the difference between groups by the log-rank test. All statistics analysis was presented using SAS software, version 9.4. Significance difference was set in which P value was less than 0.05.