EXO1/P53/SREBP1 axis-regulated lipid metabolism promotes prostate cancer progression

Data related to the TCGA-PRAD project were obtained through the UCSC Xena Browser (https://​xenabrowser.​net/​) [15]. GSE6919 from the GEO database (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​) [16]was used in this study [17, 18].

Detailed methods and principles have been reported in previous studies [19‐21]. Briefly, expression and clinical trait data for the TCGA-PRAD project were downloaded from its official website. We used the “limma” package to obtain differential genes (adj. P < 0.05, |logFC|> 1), and the differential genes were subsequently used in weighted co-expression network analysis with a soft threshold set to 4. The differential genes were categorized into 9 modules based on their expression patterns. Receiver operating characteristic (ROC) curves were generated to assess the diagnostic potential of EXO1 expression in PCa tissue. According to EXO1 expression, prostate cancer patients were categorized into two groups (high-expression group and low-expression group). The disease-free survival (DFS) of the two groups was expressed by Kaplan–Meier curves. Additionally, gene set enrichment analysis (GSEA) [22, 23] was conducted using the GSEA software (version 4.1.0, USA) to explore the enrichment of gene sets based on the expression data from the two groups mentioned.

20 pairs of PCa tissues and normal tissues were acquired from patients at the Department of Urology, Renmin Hospital of Wuhan University. None of the patients received adjuvant therapy before surgery. By Wuhan University's Human Research Ethics Committee regulations, we conducted this study. According to the Helsinki Declaration, all procedures were followed in our research. To extract RNA, we utilized the Trizol reagent extraction kit (Cat: CW0580S, CWBIO, China) and used 1000 ng of RNA for cDNA synthesis (Cat: RK20429, Abclonal, China). The resulting cDNA was diluted 1:10 for quantitative PCR (qPCR). For qPCR, the cDNAs were combined with SYBR green PCR Master Mix (Cat: RK21203, Abclonal, China) and gene-specific primers listed in Additional file 6: Table S1. The qRT-PCR was performed on a LightCycler­^® 480 Instrument II from Roche. The obtained values were then normalized to the expression of the housekeeping gene β-actin. To extract protein from the samples, triplicate wells were lysed in chilled RIPA buffer (Cat: P0013C, Beyotime, China) containing complete protease inhibitors (Cat: HY-K0010, MedChemExpress, America) and Phostop phosphatase inhibitors (Cat: HY-K0021, MedChemExpression, America). The protein concentration of the lysates was quantified using the BCA protein quantification kit (Cat: P0010S, Beyotime, China). For SDS-PAGE (Cat: PG112, Epizyme, China) and western blotting analyses, the protein samples were suspended in Laemmli buffer and sonicated 15 times for 30 s with intermittent breaks before being loaded onto a gel. The samples were boiled and used for western blot analysis after the addition of β-mercaptoethanol and bromophenol blue. The membranes were blocked in EpiZyme fast-blocking buffer and incubated overnight at 4 °C with the primary antibodies in the blocking buffer containing 0.2% Tween-20. The primary antibodies used were EXO1 (Cat: 16253-1-AP, Proteintech, China) and ACTB (Cat: AC004, Abclonal, China). For secondary antibody treatment, the membranes were either blocked in fast blocking buffer (Cat: PS108P, EpiZyme, China) at room temperature for 10–15 min or incubated directly in the diluent-blocking buffer containing 0.2% Tween-20 and 0.01% SDS for 1 h at room temperature. The membranes were imaged using fluorescence on a Biorad Imager and processed using Adobe Photoshop CC 2018. To analyze gene expression, RNAs and protein samples were extracted from 10 paired tissue samples. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to analyze RNA samples, while immunoblotting tests were conducted for protein analysis. Meanwhile, four pairs of samples were analyzed via immunohistochemistry. For cell-based experiments, we used the normal prostate cell line RWPE1 and PCa cell lines PC3, DU145, and LNCaP, which were purchased from the American Type Culture Collection (ATCC). The medium was prepared by adding 10% fetal bovine serum (Cat: C2910-0500, BioMed, China) to 1640 medium (Cat: PM150110, Pricella, China) with 1% 100 × penicillin (Cat: PB180120, Pricella, China). Cells were incubated at 37 ℃ with 5% CO₂ under controlled conditions.

Cells or ground tissue were lysed using RIPA lysis buffer (Cat: P0013C, Beyotime, China) supplemented with proteasome inhibitors (Cat: ST507, Beyotime, China) for 30 min, during which time they were sonicated three times 10 s each. bicinchoninic acid (Cat: P0010S, Beyotime, Cina) was added to the lysate for 30 min at 37 °C. Then, the protein concentration of the lysate was measured and mixed with 5X loading buffer (Cat: BL502B, biosharp, China). After boiling at 100 °C for 10 min, the mixture was used for electrophoresis for 70 min, followed by transfer onto 0.45um PVDF membrane (Cat: IPVH00010, Merck, German). After that, 5% skimmed milk (Cat: GC310001, Servicebio, China) was soaked close to the PVDF membrane for 90 min. Next, the PVDF membrane was incubated with primary and secondary antibodies. After being washed with TBST (Cat: G0001, Servicebio, China) three times, the PVDF membrane was subjected to Electrochemiluminescence (ECL). The detailed information on primary and secondary antibodies employed in this study is presented in Additional file 1: Table S1.

The detailed experimental procedures were carried out as described in a study [24]. Trizol reagent (Cat: CW0580S, CWBIO, China) and RNApure Tissue & Cell Kit (Cat: CW0560, CWBIO, China) were used to extract RNA following the manufacturer’s protocol. 1 ng RNA was used for reverse reaction and then used for qPCR reaction with 2 × Universal SYBR Green Fast qPCR Mix (Cat: RK21203, Abclonal, China). ACTB served as an internal control. The primer sequences [25, 26] employed in this study are presented in Additional file 6: Table S1.

Fresh tissues were immersed in 4% paraformaldehyde (Cat: G1101, Servicebio, China) and subsequently embedded into paraffin. Tissue sections were deparaffinized in water and subsequently washed well using PBS (Cat: G420202, Servicebio, China). Subsequently underwent antigen repair with hydrogen peroxide (Cat: 10011208, Sinopharm, China) and BSA (Cat: 4240GR005, BioFroxx, German) closure, placed in primary antibody working solution, and incubated overnight at 4 °C, followed by incubation of secondary antibody (Cat: ab205804, Abcam, Britain) as well as color development. Finally, structurally intact areas were selected for microscopic photographs. The detailed information of the primary employed in this study is presented in Additional file 6: Table S1.

Plasmids overexpressing or knocking down EXO1 (Cat: GS1-23040034, Genecreate, China), siRNA targeting P53, and their corresponding negative controls were constructed in Genecreate (Wuhan, China). PCa cells were collected after being transfected for 48 h using lipofectamine 3000 (Invitrogen, Carlsbad, CA) reagent, following the manufacturer's instructions. Lentivirus for the knockdown of EXO1 was constructed in Genechem (Shanghai, China) with a lentivirus backbone.. plasmids for the expression of SREBP1 were constructed in Genechem (Cat: GV272, China). With a lentivirus backbone, the sequence information of siRNA and shRNA is as follows.

PCa cells were cultured in 6-well plates. Oleic acid (OA) was formulated in two concentrations of 100uM, and 200uM according to the dissolution protocol of the manual (Cat: HY-N1446, MedChemExpress, America). PCa cells were divided into three groups, which were pulled into corresponding concentrations of OA, the control was BSA. After culturing cells for 48 h, transwell nesting (Cat: 3422, Corning, America) was used to evaluate the abilities of migration and invasion. Cell numbers of the image results were counted by Image J (National Institutes of Health, America), The statistical graphs were made up of GraphPad Prism 9 (America).

Cells were collected by centrifugation and then resuspended using the medium at a density maintained at 10⁴ per ml of medium. Subsequently, 200 μl of culture medium was pulled into the wells of the 96-well culture plate. Cell proliferation assays were performed using CCK-8 (MedChemExpress, USA) and measured at 0, 24, 48 h, and 72 h according to the method provided by the manufacturer. Subsequently, the optical density at 450 nm (OD450) of the cells was measured and recorded using GraphPad Prism software over four days to reflect the proliferative ability of the cells. EdU-555 assay kit (C0075S, Beyotime) was applied for EdU assays. Following the protocol, when the cells reached 50% confluency, they were fixed and stained with the appropriate reagents. The recommended reaction time of 2 h was used. A fluorescence microscope (Olympus, Japan) at 200 × magnification was used to capture images. Cell proliferative ability was assessed, and three times were repeated for each group.

Briefly, PCa cells were starved using a serum-free medium (Cat: 8122122, Gibco, America) for 8 h, then digested and centrifuged to collect the cells, which were subsequently resuspended using a serum-free medium. We used transwell nesting (Cat: 3422, Corning, America) to assay the invasive and migratory capacity of cells. For migration experiments, 40,000 cells were added to each chamber. For invasion experiments, 80,000 cells were added to each well. The time was set to 24–36 h. A microscope (Olympus, Japan) at 200 × magnification was used to capture images. Each group was repeated three times.

Cells from each group were cultured in 6-well plates. When the fusion of the cells was close to 100%, the same size wound was created using a sterile lance tip. The wounds were observed and photographed at 0 h and 48 h, respectively.

Cells (1000 PCa cells) were grown in wells in 6-well plates and cultured under normal conditions for 12 days. To fix the colonies, 4% paraformaldehyde was added to plates in each group for 10 min. The crystal violet (G1014, Servicebio) was used for staining. Finally, the colonies in each group were photographed.

When PCa cells fusion reached 30%-50%, they were fixed by 4% paraformaldehyde and stained using Oil Red O staining solution at room temperature for 1 h. After being washed three times using PBS and air-drying, the cells were photographed using a microscope at 400 × magnification.

Directed by the manufacturer's instruction, cells were cultured using 10 cm dishes and collected for lysis using 100 μl 2% TritonX-100 (Cat: BS084, Biosharp, China). Separate groups were established for blanks, calibration, and sample, each with four replicate wells. Then, the groups were subjected to a total cholesterol assay kit (Cat: A111-1-1, Njjcbio, China) or triglyceride assay kit (Cat: A110-1-1, Njjcbio, China) incubated in a 37℃ cell culture incubator for 10 min. Then, the signal values of each group were detected at 500 nm wavelength. Finally, the total cholesterol and cholesterol content of each group were calculated according to the following formula. T-CHO/TG (mmol/gprot) = [(A sample-A blank)/(A calibration—A blank)] * C calibration /Cpr (C calibration: the concentration of calibration, Cpr: the concentration of sample).

Animal experiments were approved by the Laboratory Animal Ethics Committee of Wuhan University. PC3 cell lines with stable knockdown of EXO1 were constructed using lentivirus. The above cells were digested centrifuged resuspended using PBS and injected subcutaneously into the axilla of Balb/c nude mice (aged 4 weeks, procured from Cyagen Corporation and housed under pathogen-free conditions at the Animal Biosafety Level 3 Laboratory of Wuhan University) using a syringe cell injection of 5 × 10⁶ cells per mouse. The nude mice were kept in an SPF environment for approximately 30 days. Tumor growth was measured every 5 days starting from day 10. The mice were euthanized after the experiment, and the tumors were surgically removed and photographed. They were then used for IHC and tissue oil red staining.

Data from each group were presented as mean ± standard deviation (SD). Student's t-test or paired Student's t-test was used to test differences between groups. The relationship between EXO1 expression and clinical traits of PCa patients was tested by Pearson's χ² test. GraphPad Prism (California, USA) was used for all the aforementioned analyses above. Univariate and multivariate Cox regression analyses were performed with SPSS 25.0. The significance level was set at P < 0.05. Three repetitions of each experiment were conducted. The hypothetical mechanistic draw was made by Figdraw. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Error bars indicate mean ± SEM.

We obtained a total of 4175 differentially expressed genes (DEGs) from The Cancer Genome Atlas Prostate Adenocarcinoma (TCGA-PRAD) project meeting the criteria of |logFC|> 1, and adj P < 0.05. Among DEGs, 3222 were found to be lowly expressed genes, while 953 were highly expressed genes (Fig. 1A). We performed WGCNA to explore the expression patterns of these 4175 DEGs. Gene dendrogram and module colors were displayed (Fig. 1B). In this study, a scale-free network was determined with the soft threshold (β = 4) (Fig. 1C). Modules with similarity ≥ 0.75 were merged (Fig. 1D, E). Ultimately, nine modules were identified, and their relationships with clinical traits were demonstrated (Fig. 1F). Notably, the pink module exhibited the strongest association with clinical traits (T-stage: r = 0.38, P = 2e − 18; N stage: r = 0.29, P = 2e − 11; Gleason score: r = 0.49, P = 2e − 31). Therefore, the pink module, which contained 110 genes, was considered the hub module. Their protein interactions were revealed using the STRING database [27] (Fig. 1G). Gene Ontology (GO), and Kyoto Encyclopedia of Gene, and Genomes (KEGG) analyses [28, 29]showed that pink module genes have various functions in biological processes including cell cycle, apoptosis, and P53 signaling pathway (Fig. 1H).

Thirty pink module genes that were closest to the T stage and Gleason scores were selected to form two datasets. The gene EXO1 was obtained after intersecting these two gene sets with the P53-signaling-related gene sets, and cell-cycle-related gene sets (Fig. 2A). Analysis of the TCGA-PRAD project and Gene Expression Omnibus (GEO) database revealed that EXO1 expression was higher in PCa tissue compared to normal tissue, and increased with elevated T-stage (Fig. 2B–F). ROC curves indicated that EXO1 had a good diagnostic value (Figs. 2G, H). The Kaplan–Meier curves indicated that high EXO1 expression predicted poor prognosis (Fig. 2I). By compiling clinical data from patients with high, and low EXO1 expression, the results of chi-square analysis demonstrated that EXO1 expression was related to PCa T-stage, Gleason score, and prognosis (Fig. 2J–L). Univariate and multivariate regression analyses demonstrated that EXO1 expression was a prognostic factor independent of TNM staging, and Gleason score (Table 1). Immunohistochemistry (IHC), RT qPCR, and western blotting (WB) were used to examine EXO1 expression in PCa and paracancerous tissues. The results showed that EXO1 RNA levels were higher in cancerous tissue than in paracancerous tissues in 16/20 patients. Meanwhile, EXO1 protein expression levels were higher in cancerous tissue than in paracancerous tissues in all patients. (Fig. 2M–N). The IHC results showed that EXO1 was significantly higher in cancerous tissues than in paracancerous tissues (Fig. 2O). To investigate the reasons for the high expression of EXO1 in PCa, we explored the copy number variation (CNV) of EXO1 and its promoter methylation changes in the TCGA-PRAD project through the GSCA database [30]. The results showed that the CNV of EXO1 is amplified in PCa, whereas its promoter methylation level was hypomethylated (Additional file 1: Fig. S1A, C). Further exploration revealed that the expression level of EXO1 was positively correlated with its CNV and negatively correlated with its promoter methylation level (Additional file 1: Fig. S1B, D). These findings that the high expression of EXO1 in PCa may be caused by the amplification in CNV and the low level of promoter methylation.

Drug sensitivity analysis [31] revealed that EXO1 was associated with sensitivity to a variety of antitumor drugs including vorinostat (Additional file 2: Fig. S2A, E). Immune infiltration analysis [32] showed that EXO1 promoted infiltration of activated CD4⁺positive T cells in multiple tumor types including prostate cancer (Additional file 2: Fig. S2F–I). Furthermore, prostate cancer patients were classified into four different immune subtypes (C1, C2, C3, and C4), and EXO1 expression was found to be the lowest in the C3 subtype (Additional file 2: Fig. S2J). As EXO1 is a promising prognostic biomarker for PCa, we investigated its role in PCa progression. EXO1 was stably knocked down in PCa cell lines (Fig. 3A–C). The results of Cell Counting Kit-8 (CCK-8), 5-ethynyl-2′-deoxyuridine (EdU) assay, and clone formation assays showed that the proliferative ability of PCa cells was effectively weakened by knocking down EXO1 (Fig. 3D–G). Transwell and wound healing assays showed that the metastatic ability of PCa cells was decreased after knocking down EXO1 (Fig. 3H–J). Conversely, over-expression of EXO1 significantly enhanced the cell proliferation, and migration abilities of PCa cells (Fig. 4A–J).

The results of Gene Set Enrichment Analysis (GSEA) indicated that EXO1 potentially regulated lipid metabolism (Fig. 5A–D). To find the relationship between PCa and lipid metabolism, different concentrations of OA were added to the culture medium. The results of CCK-8, transwell, and EdU assay suggested that high OA can promote the abilities of proliferation, migration, and invasion (Additional file 3: Fig. S3A–D). Among the key regulators of lipid resynthesis, the SREBP family, including SREBP1, and SREBP2, plays a significant role. In contrast to SREBP2, which regulated cholesterol resynthesis and uptake, SREBP1 was mainly responsible for fatty acid resynthesis, and regulated enzyme expression such as fatty acid synthase (FASN) and stearoyl-CoA desaturase 1 (SCD1). The results based on the Gene Expression Profiling Interactive Analysis (GEPIA) database [33] using the TCGA-PRAD project suggested that EXO1 expression in PCa was closely related to SREBP1, FASN, and SCD (Fig. 5E–G). SREBP1 was positively correlated with FASN expression in PCa (Fig. 5H). We found that SERBP1 expression was decreased, whereas SREBP2 expression remained relatively unchanged after EXO1 knockdown (Fig. 5I). Interestingly, FASN and SCD1 showed a consistent changing trend with SREBP1 (Fig. 5J–K). To validate the effects of EXO1 on lipid metabolism, we labeled intracellular lipid droplets using oil red O staining and examined the triglyceride (TG) and cholesterol levels in each group. Subsequently, the results of oil red O staining, TG, and cholesterol were consistent with the results of SREBP1 (Fig. 5L–N). By contrast, the expression level of SREBP1 increased after EXO1 overexpression (Fig. 5O). FASN and SCD1 followed the SREBP1 change (Fig. 5P–Q). We found that EXO1 promoted the TG, and cholesterol accumulation in PCa (Fig. 5R–T). Then we performed a rescue experiment to look for the relationship between EXO1 and SREBP1. In the plasmids-lentivirus-mediated EXO1 knockdown PCa cell lines, we conducted SREBP1 overexpression using plasmids and confirmed the upregulated expression of SREBP1 through WB analysis (Additional file 4: Fig. S4A). PCa cells proliferation, migration, and invasion become stronger than another group of knocking down EXO1 (Additional file 4: Fig. S4B, C, and H). The results of TG and cholesterol were consistent (Additional file 4: Fig. S4D–G). These results collectively suggested that EXO1 could upregulate SREBP1 to promote lipid synthesis.

Multigene GO and KEGG enrichment analysis of the pink module, where EXO1 was located, revealed an enrichment of the P53 signaling pathway (Fig. 1H). GSEA results further indicated that EXO1 was a potential pathway to regulate P53 (Fig. 6A). GEPIA results showed that EXO1 expression was significantly correlated with PCNA, and CDC25C, which are important components of the P53 signaling pathway (Fig. 6B). Additionally, IHC of P53 in human tissues also demonstrated that P53 levels were distinctly lower in cancerous than in paracancerous tissues (Fig. 6C). To investigate the role of EXO1 in the regulation of P53 signaling, we examined the RNA, and protein levels of P53 after EXO1 knockdown. The results showed that EXO1 suppression led to a growth in P53 mRNA, and protein levels (Fig. 6D-E). Furthermore, we examined the expression of classical genes downstream of P53 including CDC25C and PCNA. CDC25C and PCNA expressions decreased visibly (Fig. 6F–I and Additional file 5: Fig. S5I). Conversely, when EXO1 was over-expressed, opposite changes were observed in these genes (Fig. 6J–O). Previous studies have shown that P53 signaling, when activated, promotes apoptosis by inhibiting BCL2 and promoting BAX; and inhibits the cell cycle by suppressing CDC25 and PCNA. To further confirm the effect of EXO1 on the P53 signaling pathway, we examined the expression of the P53 downstream signals BCL2, BAX, and CDC25C with PCNA after changing the expression level of EXO1. We found that EXO1 promoted BCL2, CDC25, and PCNA and inhibited BAX expression (Additional file 5: Fig. S5A–I), and further experiments showed that EXO1 inhibited apoptosis in prostate cancer cells. The above results suggest that EXO1 inhibits P53 signaling. In conclusion, these results supported that EXO1 inhibited P53 signaling.

To determine whether P53 mediated the EXO1 promotion of PCa progression and lipid synthesis, siRNA was used to inhibit the activity of P53 signaling after knocking down EXO1. The results of western blotting analysis suggested that P53 knockdown rescued the decreased expression of SREBP1 caused by EXO1 knockdown (Fig. 7A). Subsequently, many rescue experiments were conducted. The Results of CCK-8 and EdU assays showed that the proliferation ability of PCa cells was effectively rescued after P53 knockdown (Fig. 7B–D). Consistent with this, Transwell assays indicated that P53 knockdown rescued the metastasis ability of PCa cells (Fig. 7E). Additionally, oil Red O staining assays indicated that P53 knockdown effectively attenuated the lipid reduction induced by EXO1 knockdown (Fig. 7F). This finding was further validated by the results of TG, and cholesterol measurement experiments (Fig. 7G). Taken together, these results suggested that EXO1 promoted PCa progression by inhibiting P53 signaling. On this basis, we conclude that EXO1 promotes lipid synthesis and progression through PCa P53 signaling.

Based on the results of the above in vitro experiments, we further examined the tumor growth-promoting and lipid accumulation effects of EXO1 in vivo. First, we constructed a PC3 cell line with stable knockdown of EXO1 using lentivirus and injected it subcutaneously in nude mice. The results showed that the tumor mass and volume were significantly reduced in the EXO1 knockdown group compared to the control group (Fig. 8A–C). Subsequently, subcutaneous tumors were used for IHC and oil red O staining. The results showed that the Ki67 scores remarkably decreased in the EXO1 knockdown group, suggesting that EXO1 significantly reduces the malignancy of PCa. The IHC results of EXO1, P53, and SREBP1 indicated that EXO1 inhibited P53 signaling, and promoted SREBP1 expression in vivo (Fig. 8D). Furthermore, the oil red O staining results showed that EXO1 knockdown significantly suppressed lipid accumulation in PCa (Fig. 8E). Collectively, our findings suggested that the EXO1-P53-SREBP1 axis promoted PCa progression and lipid accumulation, as observed in both in vitro and in vivo experiments (Fig. 8F).