Enz treatment promoted the invasion of prostate cancer and the expression of SALL4

Early studies indicated that treatment with the recently approved antiandrogen Enz could increase PCa patients’ survival by an extra 4.8 months, yet might also have some adverse effects of increasing PCa cell invasion. To dissect the mechanism underlying this phenomenon, we first confirmed that treating with Enz could increase cell invasion significantly in both PCa 22Rv1 and C4-2 cells (Fig. 1A). We then applied the RNA-seq analysis to identify genes that might be modulated under Enz treatment in C4-2 cells, and found several oncogenes whose expressions were increased significantly (Fig. 1B).

Fig. 1

Enz promotes prostate cell invasion through SALL4. (A) Chamber-transwell invasion assay in both C4-2 and 22Rv1 cells treated with DMSO or 10 µM Enz (left panel), quantitation of the invasion assay (right panel). (B) The RNAseq list of most significant genes upregulated after Enz treatment. (C) TCGA data from 400 PCa patients divided into two groups according to SALL4 expression. Survival curve showed that patients who have a lower SALL4 expression have a higher overall survival. (D) The TCGA data of SALL4 expression in PCa patients with different stages (T1-T4) and different gleason scores (G6-G9). (E) Chamber-transwell invasion assay in C4-2 cells transduced with/without knocked down SALL4 gene (sh-SALL4) cells treated with DMSO or 10 µM Enz (left panels), quantitation of the invasion assay (right panel). (F) Chamber-transwell invasion assay of 22Rv1 (left panels) and C4-2 (right panels) cells transduced with pLKO or knocked down SALL4 (sh-SALL4) or pWPI or overexpressed SALL4 (oe-SALL4). In Fig. A, and E–F, the invaded cells were counted in 10 randomly chosen microscopic field (100 ×) of each experiment and pooled and quntitations are on the right. Scale bar: 20 µm. In A and E–F, each sample was run in triplicate and in multiple experiments. Data are presented as mean ± SD, P values listed in Figure are compared to the controls. Scale bar: 20 µm. All result were run in triplicate and in multiple experiments

We analyzed a human clinical sample survey via the TCGA database (http://cancergenome.nih.gov) using UCSC Cancer Genome Browser (https://genome-cancer.soe.ucsc.edu/) to link the oncogene expression to the PCa progression, which revealed that PCa patients with a higher expression of the SALL4 oncogene have a worse survival (Fig. 1C). Furthermore, we verified the expression of SALL4 in tissue samples from patients before and after Enz treatment using Western Blot (WB) analysis. The results indicated that Enz treatment significantly increased SALL4 expression (Supplementary Fig. 1A). Bioinformatics analysis also showed that the SALL4 oncogene expression was positively correlated with Gleason scores and T stage, the key indicators for the degree of PCa malignancy (Fig. 1D). Interestingly, we observed high SALL4 expression may be associated with tumor metastasis (Supplementary Fig. 1B). This suggests that SALL4 may play a crucial role in drug resistance, tumor metastasis, and poor prognosis in prostate cancer patients.

The key role of SALL4 of Enz tratement

We first constructed two sh-SALL4 constructs and one SALL4 overexpression construct, and chose the most effective sh-SALL4 and oe-SALL4 plasmids for later use (Supplementary Fig. 2A). To minimize the potential impact of different vectors on gene expression, we analyzed the changes in protein expression levels of C4-2 cells separately transfected with pLKO, pLVTHM, and pWPI vectors. The results demonstrated that these three vectors did not induce any significant differences in SALL4 expression (Supplementary Fig. 2B). We discovered that Enz significantly increased the expression of SALL4 and showd a similar trend in its action as the overexpression of SALL4 (Supplementary Fig. 2C and D).

In addition, we verified that knocking down SALL4 expression reversed/blocked the Enz-induced increase in C4-2 cell invasion (Fig. 1E). On the contrary, Enz did not enhance invasiveness in the C4-2 and 22RV1cells with oe-SALL4 (Supplementary Fig. 2E). The MTT assay also demonstrated that oe-SALL4 could enhance the drug resistance ability to Enz (Supplementary Fig. 2F). The above conclusions show that Enz treatment promotes prostate cancer invasion through SALL4.

Therefore, we further validated the effects of sh-SALL4 and oe-SALL4 on tumor invasion. The results revealed that sh-SALL4 inhibited tumor invasion, whereas oe-SALL4 had the opposite effect (Fig. 1F).

AR and SALL4 effected CSC-like

To further study how Enz-enhanced SALL4 expression can increase PCa cell invasion. We focused on the CSC-like, since recent studies indicated that SALL4 might play important roles to alter the CSC-like development (Tatetsu et al. 2016), and altered CSC-like population might influence the tumor cell invasion (Boccaccio and Comoglio 2006).

Using the sphere formation assay in C4-2 cells (Yuan et al. 2004), we found knocking down SALL4 significantly decreased the sphere numbers, and in contrast, oe-SALL4 significantly increased the sphere numbers (Fig.2A). Similar results were also obtained when we used 22Rv1 cells (supplementary Fig. 3A).

Fig. 2

Enz may influence PCa stem cell phenotype via altering SALL4 gene expression. (A) Sphere formation assays were performed to evaluate the cancer stem cells-like (CSC-like) numbers in C4-2 cells with sh-SALL4 and oe-SALL4 (B) C4-2 cells transfected with sh-AR and oe-AR. After 14 days of incubation, colonies in five random fields per each well were counted under a microscope. (C) Sphere formation assays were performed to evaluate the CSC-like numbers in C4-2 cells treated with Enz or transduced with sh-SALL4. After 14 days of incubation, colonies in five random fields per each well were counted under a microscope. In Quantitations for each set are at the right (A-C) images. Data are presented as mean ± SD, P values listed in Figure are compared to the controls. Scale bar: 20 µm. All result were run in triplicate and in multiple experiments

AR is a clear target of Enz, and a large number of studies have shown that Enz affects tumor growth and progression through AR. We first knockout and overexpression of AR, and found that sh-AR increased the sphere numbers in C4-2 cells, while oe-AR decreased the number (Fig. 2B). Besides, we initially found that Enz promoted the number of tumor spheres, while sh-SALL4 significantly reversed/blocked this effect (Fig. 2C). In invasion assay, sh-AR also increased the cell invasion significantly (supplementary Fig. 3B). Therefore, we speculate that Enz affect the expression of downstream SALL4 through AR.

AR increased CSC-like population via altering the SALL4 expression

To further substantiate the above conclusion, we also detected the CSC-like cell via qPCR measuring the expression of CSC-like markers, and results revealed that adding Enz could increase the expression of CD44, CD133, SOX2, and OCT4 in C4-2 cells. In contrast, adding the androgen R1881 could decrease the expression of those CSC-like markers by WB (Supplementary Fig. 3C). As anticipated, our findings revealed that sh-SALL4 could notably downregulate the expressions of CD44, CD133, SOX2, and OCT4 in both C4-2 and 22RV1 cell lines; conversely, Enz exerted an opposing effect (Fig. 3A). Besides, the expression of EMT pathway-related genes was also examined, and it was found that sh-SALL4 significantly inhibited the transcription of N-cadherin and Vimentin while promoting the transcription of E-cadherin and β-catenin (Supplementary Fig. 3D). Importantly, to demonstrate that Enz can regulate CSC-like marker expression via targeting the AR, we then used sh-AR and oe-AR. The results showed that oe-AR could decrease the expression of those CSC-like markers, and sh-AR could increase those CSC-like markers in both C4-2 and 22Rv1 cells (Fig. 3B).

Fig. 3

Enz/AR/SALL4 modulated CSC-like formation altered the cell invasion and stem cell phenotype in PCa cells. (A) The C4-2and 22RV1 cells were treated with 10 µM Enz for 48 h and transduced with sh-SALL4. Total RNAs were analyzed by qPCR to examine prostate CSC-like markers CD44, CD133, SOX2 and OCT4. (B) qPCR analysis of CSC-like markers CD44, CD133, SOX2, and OCT4 in C4-2 (left) and 22Rv1 (right) cells transduced with sh-AR or oe-AR. (C) The 22Rv1 cells transduced with sh-AR and sh-SALL4. (D) C4-2 cells transduced with oe-AR and oe-SALL4. (E) Chamber-transwell invasion analyzed C4-2 that transduced with sh-AR and sh-SALL4. (F) 22RV1 cells transduced oe-AR and oe-SALL4. (G-H) Sphere formation assay was performed to evaluate the CSC-like numbers in C4-2 cells (G) transduced with oe-AR and oe-SALL4 or in 22Rv cells (H) transduced with sh-AR and sh-SALL4. Each sample was run in triplicate and in multiple experiments. Data are presented as mean ± SD, P values listed in Figure are compared to the controls. Scale bar: 20 µm

As expected, sh-AR could also increase SALL4 and OCT4 protein expression in both 22Rv1 and C4-2 cells, while simultaneously sh-AR and sh-SALL4 could suppress the OCT4 expression (Fig. 3C), suggesting that AR regulated OCT4 expression via altering SALL4 expression. Consistent with this conclusion, oe-AR decreased the expression of SALL4 and OCT4, while this effect can be reversed via oe-SALL4 (Fig. 3D). In addition, the results of cell lines and clinical samples also indicated that Enz promoted the expression of SALL4, SOX2, and OCT4, while inhibiting the expression of AR (Supplementary Fig. 3E and F).

Importantly, we found that knocking down AR increased PCa cell invasion dramatically in C4-2 cells (Fig. 3E), and this sh-AR-increased PCa cell invasion could be reversed/blocked via suppressing the SALL4 expression (Fig. 3E). In contrast, increasing AR led to decrease the cell invasion in 22Rv1 cells (Fig. 3F) and this AR-decreased cell invasion was reversed/blocked via increasing the SALL4 expression (Fig. 3F).

Sphere formation assays also showed that oe-AR reduced the sphere number dramatically, and sh-SALL4 reversed the decrease of sphere number (Fig. 3G). In contrast, suppressing AR led to increase sphere number, and this increase could be reversed via the sh-SALL4 (Fig. 3H).

Together, results from Figs. 2–3 demonstrate that Enz treatment can increase the PCa cell invasion via altering the CSC-like population that involved the modulating the AR/SALL4/SOX2-OCT4 signaling.

Transcriptional activation of SALL4 by the Enz/AR axis

To dissect the molecular mechanisms of how Enz/AR axis can alter the SALL4 expression, we first proved that targeting the AR led to alter the SALL4 expression in both C4-2 and 22Rv1 cells at the protein levels (Fig. 4A and B).

Fig. 4

Mechanism dissection of how Enz treatment increases SALL4 expression. (A-B) C4-2 and CWR22Rv1 cells transduced with sh-AR and oe-AR. Total protein was analyzed by western-blot for SALL4. (C) Bioinformatic analysis of potential AR binding sites in SALL4 promoter. (D) Lysates of C42 cells were subjected to ChIP assay. ChIP products were amplified by PCR reaction. (E) Lysates of C4-2 cells were subjected to ChIP assay. ChIP products were determined by semi-quantitative qPCR assay. (F-G) Co-transfection of SALL4 promoter constructs containing wild type or mutant seed regions into C42 cells (F) and luciferase assay (G) was applied to detect the luciferase activity. For A, B, D, E, and G, each sample was run in triplicate and in multiple experiments. The data were presented as mean ± SD, ns P > 0.05, compared to the controls.

We then studied the AR effects at the transcriptional regulation, and found four putative androgen-response-elements (AREs) located on the SALL4 5’ promoter region (−9 ~ −17; −598 ~ −606; −1515 ~ −1523; and −1617 ~ −1625) (Fig. 4C). We applied ChIP assay in vetro to verify the AR transcriptional regulation, and found AR could bind to the ARE-2 located at 598–606 bp away from the transcription start site of SALL4 in C4-2 cells (Fig. 4D and E).

Importantly, using luciferase reporter containing 1.5 kb upstream sequence with normal or mutated ARE-2 (Fig. 4F), we found oe-AR significantly decreased the reporter activity in C4-2 cells (Fig. 4G).

Together, results from Fig. 4A-F suggest that Enz/AR axis can transcriptionally regulate the SALL4 expression via binding to the ARE located on its 5'promoter region.

SALL4 modulates CSC-like via SOX2/OCT4 activation

To examine the mechanism of how Enz/AR/SALL4 axis can alter the CSC-like population, we focus on those CSC-like related biomarkers, and found sh-SALL4 led to drastically decrease the mRNA expression of CSC-like markers including CD44, CD133, SOX2 and OCT4, while oe-SALL4 led to increase CD44, CD133, SOX2 and OCT4 expression in C4-2 and 22Rv1 cells (Fig. 5A).

Fig. 5

Mechanism dissection how SALL4 regulates cancer stem cells. (A) C4-2 cells (left) and CW22Rv1 cells (right) transduced with sh-SALL4 and oe-SALL4 for 48 h, total RNA was analyzed by qPCR for expression of CSC-like markers CD44, CD133, SOX2, and OCT4. (B-C) C4-2 cells and 22Rv1 cells transduced with sh-SALL4 and oe-SALL4, then SOX2 and OCT4 protein expression was determined by Western blot. (D) Possible binding motifs of SALL4. (E) Co-transfection of OCT4 and SOX2 promoter constructs containing wild type regions into C4-2 cells (left) and luciferase assay (right) was applied to demonstrate transcriptional activation by SALL4. (F) The correlation of SALL4 with SOX2 and OCT4 by GEPIA website. For A, B, C and E, each sample was run in triplicate and in multiple experiments. The data were presented as mean ± SD

Results from WB assay furthers confirmed that sh-SALL4 led to SOX2 and OCT4 downregulation in both C4-2 and 22Rv1 cells, and oe-SALL4 led to an opposite effect (Fig. 5B and C).

To further examine the SALL4 impacts on the transcriptional regulation of SOX2, we first searched for potential SALL4 response elements on the presumptive SOX2 promoter region using JASPAR database and PROMO database, but failed to identify such motifs in the 2.5 kb sequences upstream of SOX2 and OCT-4 transcription start site (Fig. 5D). However, using luciferase reporter with the 2.5 kb upstream sequence, we found increasing the SALL4 significantly increased the reporter activity in C4-2 cells control cells transfected with the wild type construct (Fig. 5E), suggesting SALL4 might be able to transactivate the CSC-like markers SOX2 and OCT-4 transcription likely through a non-canonical binding element.

Together, results from Fig. 5A-E suggest that Enz/AR/SALL4 axis may alter the CSC-like population via transcriptional regulation of SOX2 and OCT-4 likely through a non-canonical binding element.

Using GEPIA database (http://gepia.cancer-pku.cn/) we found that the positive correlation between SALL4 and SOX2 (Fig. 5E).

Together, results from human clinical sample survey also support all above in vitro cell lines studies showing Enz/AR/SALL4/SOX2 axis may play key roles during PCa progression,

In vivo validation of the AR/SALL4/SOX2 axis in metastasis

To confirm all above in vitro cell lines data in the in vivo animal, we then orthotopically xenografted the the 22Rv1 cells expressing firefly luciferase to the anterior prostate. Once the PCa tumors were detected, we then treated mice with Enz to compare the 22Rv1 cells infected with vector control vs sh-SALL4 groups, and monitered the progression with the In Vivo Imaging Systems (IVIS). After 8 weeks, we sacrificed the mice and examined the tumor metastasis in lung, liver, spleen, and kidney with the IVIS analysis and IHC staining.

The IVIS results revealed that mice xenografted with PCa 22Rv1cells with a sh-SALL4 expression had significantly small tumor growth than the control group (Fig. 6A-B). Importantly, further adding Enz (30 mg/kg) for three weeks resulted in much smaller tumors in both pLVTHM group and sh-SALL4 group (Fig. 6A). Results from sacrificed 8-weeks mice via assaying the tumor volumes and weight also confirmed IVIS data showing that sh-SALL4 suppresses tumorigenicity in mice with/without Enz treatment (Fig. 6B-C).

Fig. 6

The AR/SALL4/SOX2/OCT4 signaling axis in PCa xenograft model. (A) IVIS imaging was used to determine the tumor size and metastasis in mice xenografts from 22Rv1-Luc cells transduced with vector control or sh-SALL4. Mice groups were treated with/without Enz. Due to mouse deaths during the experiment the Groups 1 (vector control + DMSO), Group 2 (vector control + Enz), group 3 sh-SALL4 + DMSO) and Group 4(sh-SALL4 + Enz) had remaining mouse numbers of 4, 4, 6, 7, respectively. (B) After sacrificing the mice the tumor tissues were obtained for macroscopic analysis. (C) Weights of the xenografts were shown in the four groups. *p < 0.05. (D) Representative bioluminescent images of metastases to different organs. (E)The percentage of metastasis (meta) vs no-metastases (no-meta) in the four groups. (F) Representative images of IHC staining for AR, SOX2, and OCT4 of xenograft specimens for the four groups. Scale bar: 20 µm

The tumor metastasis was also evaluated via the IVIS analysis, which revealed that more metastases occurred in the Enz treatment group, while in the sh-SALL4 group there were no obvious metastases in either Enz treatment group or control group mice (Fig. 6D-E). The IHC staining also indicated that the expression of SOX2 and OCT4 decreased clearly in the sh-SALL4 group mice, while in the Enz-treatment group SOX2 and OCT4 expression increased dramatically (Fig. 6F). It is possible that as the sh-SALL4 group grew much slower the tumor that eventually grew may have lost the repression of SALL4, thus these tumor cells could still respond to Enz similar to the control group.

Together, these in vivo data suggest that the Enz may function via altering the AR/SALL4/SOX2-OCT4 signaling to alter the CSC-like population to increase the cell invasion, and targeting this newly identified signaling with small molecule of sh-SALL4 may decrease the Enz adverse effect of increasing cell invasion to better suppress the PCa progression.