VHL mutation-mediated SALL4 overexpression promotes tumorigenesis and vascularization of clear cell renal cell carcinoma via Akt/GSK-3β signaling

The human ccRCC cell lines (ACHN, 786-O) and human umbilical vein endothelial cell (HUVEC) were cultured in RPMI 1640 (786-O) or DMEM/F-12 (ACHN, HUVEC) medium (Gibco) containing 10% fetal bovine serum (FBS, Bioind) and 1% penicillin/streptomycin (Gibco). All the cell lines were maintained in an incubator (37 °C) containing 5% (v/v) humidified CO₂.

Stable SALL4 knockdown cells were established by lentiviral shRNA infection. The lentiviral particles carrying shRNA against SALL4 (shSALL4, sh#1 and sh#2) or negative control (shNC) were generated and provided by Obio Technology (Shanghai, China). The cells (ACHN, 786-O) were transduced with lentivirus following the manufacturer’s instructions. To establish stable cell lines, the infected cells were treated with puromycin (2 μg/mL) for 7 days. The shRNA sequences for SALL4: GCCTTGAAACAAGCCAAGCTA (sh#1) and GAGGATGAAGCCACAGTAA (sh#2); the shRNA sequence for negative control: TTCTCCGAACGTGTCACGT (shNC).

Cells (1 × 10³ cells/well) were seeded in 96-well plates and maintained for indicated time. Cell growth was monitored by incubation with CCK-8 solution (Sangon Biotech) following the manufacturer’s protocols. Then a microplate reader (Bio-Rad) was used to detect absorbance at 450 nm. Three repetitions were conducted in triplicate.

Stable transfected cells (200 cells/well) were seeded into six-well plates and grown for 10 days. After fixed by 4% paraformaldehyde, the colonies were incubation with 0.1% crystal violet for 20 min. Then images were acquired and the number of colonies was counted.

For cell cycle analysis, cells were harvested and fixed with 70% ethanol, followed by sequential treatment with RNase A (100 μg/mL) and propidium iodide (PI) staining buffer. To analyze cell apoptosis, HUVECs treated as indicated were harvested and labeled with Annexin V-FITC and PI. Cell cycle distribution and apoptosis index were assessed by flow cytometry.

A SA-β-gal staining kit (Beyotime) was used to evaluate cellular senescence via detection of β-galactosidase activity following the manufacturer’s protocols. Cells were plated in six-well plates and cultivated for 48 h. Then cells were washed with PBS and treated with fixative buffer for 15 min. After washed with PBS for three times, cells were incubated with premixed staining solution in a CO₂-free atmosphere overnight. The images were acquired and SA-β-gal positive cells were counted under a microscope.

Cells seeded on six-well plates were cultivated to nearly full confluence. A yellow pipette tip was used to scratch a wound on the monolayer cells. Then photographs of scratched cells were taken under a microscope to monitor wound closure at indicated time.

Transwell assays were conducted to evaluate the capacities of migration and invasion. Cells were resuspended in medium without FBS and counted. The 8-μm-pore chambers (Corning) were coated with (for invasion assay) or without (for migration assay) Matrigel (BD Biosciences) and inserted into 24-well plates. Resuspended cells were placed into upper chamber for transwell migration (2 × 10⁴ cells/well) and invasion (1 × 10⁵ cells/well) assay. Chambers were then incubated with medium containing 10% FBS in 24-well plates for 24 h. Penetrated cells were stained with crystal violet and images were acquired for cell count under a microscope.

Stable transfected cells (ACHN-shNC, ACHN-sh#1) were re-suspended with DMEM/F-12 containing 10% FBS and plated in 60 mm dishes (5 × 10⁵ cells/dish). After incubation for 12 h, cells were washed and normal growth medium was replaced with serum-free medium (SFM) (5 ml/dish). After incubation for 24 h, cell culture supernatants from ACHN-shNC and ACHN-sh#1 cells were harvested. Cell debris in the supernatants was removed with a centrifuge.

The levels of VEGFA in conditioned medium from ACHN-shNC and ACHN-sh#1 cells were detected by using an ELISA kit (Sino Biological) according to the manufacturer’s protocols. Then a microplate reader (Bio-Rad) was used to measure absorbance at 450 nm.

The Boyden chamber (Millipore) assay was performed to determine the effect of ccRCC cells on HUVECs recruitment. Briefly, HUVECs were resuspended with SFM and placed into upper chamber (2 × 10⁴ cells/well). Chambers were then incubated with indicated CMs (SFM, shNC-CM, sh#1-CM) or co-cultured with ccRCC cells (ACHN-shNC, ACHN-sh#1) in lower chamber for 24 h. Penetrated cells were stained with crystal violet and images were acquired for cell count under a microscope.

Matrigel (BD Biosciences) (50 μl/well) was used to coat the wells of 96-well plates. After solidification for 1 h at 37 °C, HUVECs (4 × 10⁴ cells/well) were re-suspended with indicated CMs (SFM, shNC-CM, sh#1-CM) and plated onto the matrigel. After 6-h incubation at 37 °C for tube formation, tubules were observed and pictures were acquired by microscopy. ImageJ software was used to analyze the formed tubes.

Immunoblotting assay was carried out as we previously described [16]. Cells were lysed using RIPA buffer containing protease and phosphatase inhibitors. The protein sample was resolved by SDS-PAGE, followed by transference of proteins onto PVDF membranes. The membrane was then treated with 5% bovine serum albumin, followed by overnight incubation with specific antibodies against Akt (1:1000, Cell Signaling Technology, CST), p-Akt (1:1000, CST), GSK-3β (1:1000, CST), p-GSK-3β (1:1000, CST), VEGFA (1:500, Sangon Biotech), SALL4 (1:1000, CST) and GAPDH (1:2000, Sangon Biotech). The corresponding HRP-conjugated secondary antibody and ECL substrate were used to visualize the protein bands.

Cells were plated and grown on coverslips overnight. After fixed in 4% paraformaldehyde, cells were treated with 0.5% triton X-100 for cell permeabilization. The coverslips were then immersed in blocking solution (Beyotime) for 1 h, followed by incubation with anti-p-GSK-3β (1:300, CST) and anti-p-Akt (1:200, CST) overnight. The Alexa Fluor 594-conjugated secondary antibody (1:200, Zsbio) was used for fluorescence labeling. Coverslips were incubated with DAPI for nuclei staining. Cells were observed and pictures were acquired by fluorescence microscope.

Immunodeficient BALB/c mice (six-week-old) were provided by Experimental Animal Center of Fourth Military Medical University (FMMU) and kept under SPF conditions. The procedures involving animals received approval of the Animal Research Ethics Committee of FMMU and the research was conducted in accordance with institutional guidelines. For tumor growth assay, 786-O sublines (1 × 10⁷ cells) were implanted subcutaneously into the flanks of mice (six per group). A vernier caliper was used to measure tumor size every 5 days. The formula (volume = length × width²/2) was used to calculate tumor volume. Mice were sacrificed 5 weeks after inoculation. Tumor nodules were collected for further examination.

Immunohistochemistry staining was carried out as we previously described [16]. Morphology analysis of formalin-fixed and paraffin-embedded (FFPE) sections were performed with hematoxylin and eosin staining. For IHC staining, heat-based antigen unmasking was conducted using microwave in citrate buffer after dewax and rehydration of sections. Tissue sections were exposed to 3% H₂O₂ to block endogenous peroxidase, followed by treatment with blocking solution for nonspecific binding. Subsequently the sections were stained with specific antibodies against SALL4 (1:100, Abcam) overnight. After incubation with corresponding secondary antibodies, a DAB substrate kit was used to visualize positive antigen binding. Haematoxylin counterstaining was performed and pictures were captured by microscopy.

To identify the expression pattern and clinical characteristics of SALL4, public datasets (TCGA Renal 2 and TCGA-KIRC cohorts) were analyzed in Oncomine (http://​www.​oncomine.​org/​) and UALCAN [17] (http://​ualcan.​path.​uab.​edu/​) databases. The correlation between SALL4 expression and survival in ccRCC patients was assessed by GEPIA (http://​gepia.​cancer-pku.​cn/​) and Kaplan-Meier Plotter (http://​kmplot.​com/​analysis/​) databases. Genome-wide association analysis of SALL4 was performed with Cancer Regulome Tools (http://​explorer.​cancerregulome.​org/​). Gene correlation analysis, mutation analysis and methylation analysis were carried out using LinkedOmics [18] (http://​www.​linkedomics.​org/​) and TCGAportal (http://​tumorsurvival.​org/​) databases.

Group differences were determined using IBM SPSS 18.0 software. Experimental values are represented as means ± SD. Two-tailed Student’s t-test (two groups) and one-way ANOVA test (three or more groups) were conducted as appropriate for differences comparison. Kaplan-Meier method was employed to analyze patient survival and Pearson correlation coefficient was applied for gene expression correlation analysis. Difference was considered significant as indicated (* P < 0.05, ** P < 0.01, *** P < 0.001 and **** P < 0.0001).

To elucidate SALL4 expression signatures in human malignancies, we conducted pan-cancer analysis of public TCGA datasets in UALCAN database. Our results showed that SALL4 was significantly upregulated in most of human tumors (Additional file 1: Figure S1a), including ccRCC, indicating an oncogenic role for SALL4 in tumorigenesis. Further analysis of publicly available datasets was performed to evaluate the expression levels of SALL4 in ccRCC and normal kidney tissues. We found that SALL4 was dramatically overexpressed in ccRCC tissues compared with normal kidney tissues (Fig. 1a, b and Additional file 1: Figure S1b, c). To validate the upregulated SALL4 expression in ccRCC patients, we performed IHC staining in ccRCC and adjacent noncancerous samples from the same patients and observed higher protein levels of SALL4 at both plasmolemma and nucleus in ccRCC specimens (Fig. 1c). Moreover, SALL4 expression varied depending on tumor invasion depth, lymph node status, distant metastasis, histological grade and AJCC stage (Fig. 1d-h and Additional file 1: Figure S1d-g).

In addition, the correlation between SALL4 expression and ccRCC patients’ clinicopathological characteristics was analyzed to investigate the clinical significance of SALL4 in ccRCC. Our results revealed that SALL4 expression positively correlated with T stage, tumor grade and AJCC stage (Table 1, P < 0.05). Kaplan–Meier survival analysis revealed the close relationship between SALL4 expression and the prognosis of ccRCC patients. The overall survival time was significantly shorter in ccRCC patients with high SALL4 expression than in those with low SALL4 expression (Fig. 1i-k and Additional file 2: Figure S2). Univariate and multivariate analyses were performed to identify risk factors of prognosis in ccRCC patients. SALL4 was found to be an independent prognostic factor of overall survival for ccRCC patients (Table 2, P < 0.05). To further explore the diagnostic potential of SALL4, the receiver-operating characteristic (ROC) curve analysis was performed by measuring the area under the curve (AUROC). Our data indicated that SALL4 level could distinguish ccRCC patients from healthy controls (Fig. 1l) and diagnose patients with deeper invasion (T3–4) (Fig. 1m), distant metastasis (Fig. 1n) and advanced stage (Stage III-IV) (Fig. 1o). Taken together, these data suggest that upregulation of SALL4 contributes to ccRCC progression and is associated with poor prognosis of ccRCC patients.

To further validate its oncogenic activities in ccRCC, we set out to check the putative function of SALL4 on ccRCC cell growth. Lentiviral shRNA-mediated knockdown of SALL4 was conducted and downregulated SALL4 protein levels in ccRCC cells (ACHN, 786-O) were detected (Fig. 2a). The cells were then subjected to CCK-8 and colony formation assays to determine the influence of SALL4 downregulation on ccRCC cell proliferation. We found that knockdown of SALL4 in ACHN and 786-O cells resulted in slower growth rate compared with control cells (Fig. 2b). Similarly, the number of colonies formed by cells with downregulated SALL4 was significantly reduced (Fig. 2c). To test whether SALL4 also drives cell cycle progression, flow cytometry analysis was performed. The results showed that remarkable changes of cell cycle distribution were induced by SALL4 silencing in ccRCC cells. As indicated by increased G1-phase cells and decreased S/G2-phase cells, downregulation of SALL4 in ccRCC cells arrested cell cycle by restraining G1-S transition (Fig. 2d). Resistance to senescence or apoptosis has been identified as a hallmark of cancer cells and plays a crucial role in cell survival and tumorigenesis [19]. In particular, it has been demonstrated that some cells are more prone to senescence rather than apoptosis even following intensive exogenous stress [20]. SA-β-gal is the most frequently used marker for senescence and senescent cell exhibits high SA-β-gal activity. To further elucidate the functional role of SALL4 in cell senescence, ccRCC cells with stable SALL4-targeted or control shRNA were assayed using SA-β-gal staining kit. We observed that depletion of SALL4 in ACHN and 786-O cells upregulated SA-β-gal synthesis (Fig. 2e) indicating that SALL4 depletion triggered cells senescence. By analyzing a public dataset of 533 ccRCC patients from TCGA, we found that SALL4 mRNA level was significantly correlated with the transcripts of genes related to proliferation, senescence and cell cycle, including CCNE1 (r = 0.4145, P < 0.0001), CDK3 (r = 0.3811, P < 0.0001), E2F1 (r = 0.3302, P < 0.0001) and RB1 (r = − 0.3032, P < 0.0001) (Fig. 2f-i and Additional file 3: Figure S3, Additional file 4: Table S1). Next, to investigate the oncogenic activity of SALL4 in ccRCC tumorigenesis in vivo, tumor formation was evaluated by subcutaneous inoculation of 786-O sublines in nude mice. we found that downregulation of SALL4 in ccRCC cells resulted in a dramatic decrease in tumorigenic potential, as evidenced by decreased tumor size, repressed tumor growth and reduced tumor weight (Fig. 2j-l). Together, these findings validate that SALL4 drives ccRCC cell growth by promoting cell cycle progression and restraining cell senescence.

Next, to explore whether SALL4 also function as a prometastatic factor in ccRCC, we performed a series of loss-of-function studies in ACHN and 786-O cells stably transfected with SALL4-targeted or control shRNA. The wound healing assays demonstrated that SALL4 downregulation markedly suppressed cell migration to delay healing of the scratched cell monolayer in ccRCC cells (Fig. 3a, c). Similar results were observed in transwell migration assays. We found that SALL4 silencing in ccRCC cells significantly impaired the migratory ability as measured by cells attached to the lower membrane surfaces. Consistently, in matrigel invasion assays of ACHN and 786-O cells, less cells were observed to penetrate through the matrigel barrier upon SALL4 knockdown, indicating a decrease in invasion potential (Fig. 3b, d). These results were consistent with our finding that SALL4 was upregulated in metastatic ccRCC tumors (Fig. 1f). The epithelial-mesenchymal transition has been reported to be involved in SALL4-mediated tumor metastasis [21]. In agreement with previous findings, we found that compared with the control cells, SALL4-deficient ACHN cells seemed to exhibit a tighter organization of cells in colonies (Additional file 5: Figure S4a). In addition, we analyzed the RNA-seq data of 533 ccRCC patients from TCGA and found that SALL4 mRNA level was significantly correlated with the transcripts of matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) related to tissue remodeling and tumor invasiveness [22, 23], including MMP9 (r = 0.1227, P = 0.0005), MMP25 (r = 0.2601, P < 0.0001), TIMP3 (r = − 0.2221, P < 0.0001) and TIMP4 (r = − 0.2761, P < 0.0001) (Fig. 3e-h and Additional file 6: Figure S5, Additional file 7: Table S2). Collectively, these results support that SALL4 functions as a prometastasis oncogene in ccRCC progression.

To further characterize SALL4’s biologic functions and discern its oncogenic mechanisms, we first characterized the differential association signature between SALL4 mRNA level and individual gene expression by analyzing the RNA-seq data of 533 ccRCC patients from TCGA. As shown in Additional file 8: Figure S6a, we identified 6315 (dark red dots) and 2862 (dark green dots) genes that were positively and negatively correlated with SALL4 (FDR < 0.01), indicating that SALL4 may exert a widespread influence on the transcription profiles. We then performed cluster analysis of the correlation profiles and constructed a heatmap to visualize the top 50 most significant genes that were positively and negatively correlated with SALL4 transcripts (Additional file 8: Figure S6b, c). Statistical scatter plots for top-ranked genes were created and revealed a strong positive correlation between SALL4 mRNA level and NOD-like receptor family CARD domain containing 5 (NLRC5) gene expression (positive rank #2, r = 0.62, P = 6.66e-57) (Additional file 8: Figure S6d). NLRC5 was previously reported to be upregulated in ccRCC and positively associated with tumor progression [24]. Recent evidence indicates that NLRC5 is implicated in vascular remodeling and required for vascular intimal hyperplasia [25]. we next conducted gene ontology term enrichment and KEGG pathway analyses of SALL4-associated genes and found that SALL4 expression significantly correlated with genes controlling vasculogenesis, endothelium development and cellular response to vascular endothelial growth factor stimulus (Fig. 4a), confirming the potential role of SALL4 in modulating vascular remodeling. Vasculogenesis refers to the formation of new blood vessels. Endothelium development is a complex process in which endothelial cells are oriented towards their specific fate and specialized endothelial cells form the endothelium of an organ such as vasculature, lymph vessel and heart. Furthermore, gene set enrichment analysis (GSEA) were performed in human ccRCC samples and revealed a large number of significantly enriched gene sets, including vasculogenesis (Fig. 4b) and endothelium development (Fig. 4c). Genome-wide association of SALL4 mRNA expression with multifarious molecular signatures, including gene expression, protein level (reverse phase protein arrays, RPPA), microRNA expression, DNA methylation, copy number variation and somatic mutation, was analyzed and visualized using a circos plot (Additional file 8: Figure S6e) and a network (Additional file 8: Figure S6f). In agreement with our observations in Additional file 8: Figure S6a-d, we found that the transcription levels of NLRC5 and SALL4 mRNA expression were strongly associated in ccRCC patients (Additional file 8: Figure S6f). In summary, these data suggest that SALL4 may be involved in vasculogenesis and vascular remodeling.

Given the potential role of SALL4 in vasculogenesis and vascular remodeling, we sought to investigate whether SALL4 modulates angiogenesis in ccRCC. In response to treatment with CM collected from ACHN sublines, we observed that HUVECs cultured in CM from ACHN-sh#1 cells showed significant reduction in viable cell number as measured by CCK-8 assay (Fig. 4d). Flow cytometry analysis of HUVECs treated with CM from ACHN-sh#1 cells revealed an arrested cell cycle compared with that of cells treated with CM from ACHN-shNC cells (Fig. 4e), but no significant alteration in cell apoptosis was observed (Additional file 5: Figure S4b, c). Scratch assay was performed on HUVECs treated with CMs. We found that migration of HUVECs was markedly alleviated after treatment with CM from ACHN-sh#1 cells (Fig. 4f). In HUVECs recruitment assay, we observed that the chemotactic response of HUVECs migrating toward CM from ACHN-sh#1 cells was drastically suppressed in comparison with that of cells stimulated with CM from ACHN-shNC cells (Fig. 4g). To explore the direct effect of SALL4 deficiency on the function of endothelial cells, we established a coculture system of HUVECs with ACHN sublines. Consistent with the above findings, we noticed that much fewer HUVECs were recruited toward ACHN cells lacking SALL4 (Fig. 4g). To substantiate the relevance of SALL4 to angiogenesis in ccRCC, HUVECs were treated with CMs from ACHN sublines and assayed for tube formation. We observed that as expected, HUVECs exhibited diminished tube formation activity following stimulation with CM from SALL4-deficient ACHN cells (Fig. 4h, i). Furthermore, we analyzed the levels of VEGFA in CMs from ACHN sublines and found that the concentration of VEGFA in CM from ACHN-sh#1 cells decreased (Fig. 4j). Taken together, these in vitro observations validate the potent role of SALL4 in ccRCC angiogenesis.

SALL4 was previously implicated in chromatin remodeling via modulation recruitment of nucleosome remodeling deacetylase (NuRD) complex [26]. The tumor suppressor PTEN has been reported to be a target gene repressed by SALL4 and counteract the PI3K pathway [13]. To validate the involvement of PI3K/Akt signaling in SALL4-mediated oncogenic activities in ccRCC, we evaluated the effects of SALL4 downregulation on PI3K pathway markers. Immunoblotting analysis of lysates from ccRCC cells revealed that SALL4 knockdown decreased the levels of phosphoprotein markers of PI3K pathway activation (p-Akt, p-GSK3β) without altering the total protein levels in ACHN and 786-O cells (Fig. 5a, b). Consistently, immunofluorescent staining (Fig. 5c) further confirmed the observations in western blot assay. The extensive crosstalk and a positive feedback loop have been reported to exist between the PI3K/Akt and VHL/HIF pathways [7], contributing to ccRCC tumorigenesis and progression. Given the functional impact of SALL4 on angiogenesis and decreased VEGFA level in CM from SALL4-downregulated ACHN cells, we further investigated whether SALL4 modulates the synthesis of VEGFA, a downstream target of VHL/HIF pathway. We observed that depletion of SALL4 led to a dramatic reduction in VEGFA protein levels in ACHN and 786-O cells (Fig. 5a, b), thus suggesting a link between SALL4 expression and VHL/HIF pathway. Collectively, these results indicate that SALL4 expression positively correlates with the activation of Akt/GSK-3β signaling and VEGFA expression.

VHL protein has emerged as a potent tumor suppressor governing various signaling pathways that promote oncogenic phenotypes and fitness. Mutations in VHL are observed in roughly 70–80% of ccRCC and identified as causal event for ccRCC initiation and progression [27]. To investigate whether SALL4 overexpression is enriched in tumors with driver gene mutations, we analyzed the association between SALL4 expression and the mutation profile of most commonly altered genes (VHL, PBRM1, SETD2, KDM5C, BAP1, PTEN, MTOR, TP53 and PIK3CA) in ccRCC. Our results revealed a significant difference of SALL4 expression between tumors with VHL or PBRM1 mutations and non-mutated cases (Fig. 6a). Additionally, we explored the expression of VHL in TCGA Renal 2 dataset via Oncomine database and observed that VHL expression was significantly decreased in ccRCC samples when compared with that in normal kidney tissues (Fig. 6b, c). We then evaluated the relationship between SALL4 gene expression and the genetic status of VHL in ccRCC patients. Our data indicated that SALL4 gene copy number was higher in tumors with VHL deletion and methylation, especially VHL mutation (Fig. 6d-f), suggesting that SALL4 amplification was significantly co-occurred with VHL lesions. Nevertheless, no significant difference of SALL4 expression was observed among ccRCC patients with different VHL mutation subtypes (Additional file 9: Figure S7a). Next, to further acquire a better understanding of the functional role of VHL mutation in SALL4 expression, we analyzed SALL4 mRNA levels in ccRCC patients with or without different VHL point mutations (Fig. 6g-j and Additional file 9: Figure S7b-k). As shown in Fig. 6 g, a significant positive association was observed between VHL_p.L89H point mutation and SALL4 gene expression (P < 0.05).

Prior evidence demonstrates that VHL deficiency drives genome-wide changes in DNA methylation which has been implicated in the pathogenesis of ccRCC [28]. In addition, it is well-established that DNA methylation alterations are highly associated with aberrant expression of genes. We thus infer that DNA methylation may be also involved in VHL mutation-mediated upregulation of SALL4. By analyzing the correlation between SALL4 methylation status and its mRNA expression (Additional file 10: Figure S8), we found that six methylation regions of SALL4 were markedly hypomethylated in ccRCC tumor samples in comparison with paired normal kidney samples. Among the six hypomethylated SALL4 regions, a negative correlation between methylation status of chr20:51800437 and chr20:51800155 and SALL4 mRNA expression was observed in both ccRCC tumor samples and paired normal kidney samples (Additional file 10: Figure S8). Further exploration are required to determine the exact prevalence of promoter hypomethylation in association with SALL4 upregulation. Together, these data suggest that VHL mutation can function as a potential mechanism for SALL4 upregulation.