PTEN-knockout regulates metabolic rewiring and epigenetic reprogramming in prostate cancer and chemoprevention by triterpenoid ursolic acid

Abstract

PTEN (phosphatase and tensin homologue deleted on chromosome 10) is one of the most frequently mutated/deleted tumor suppressor genes in many human cancers. Ursolic acid (UA) is a natural triterpenoid possessing antioxidant, anti-inflammatory and anticancer effects. However, how PTEN impacts metabolic rewiring and how UA modifies PTEN-driven metabolic and epigenetic reprogramming in prostate cancer (PCa) remains unknown. In the current study, we found that UA protects against PTEN knockout (KO)-induced tumorigenesis at different stages of PCa. Epigenomic CpG methyl-seq revealed UA attenuated PTEN KO-induced differentially methylated regions (DMRs) profiles. Transcriptomic RNA-seq showed UA abrogated PTEN KO-induced differentially expressed genes (DEGs) of PCa related oncogenes’ Has3, Cfh and Msx1 overexpression, indicating UA plays a crucial role in PTEN KO-mediated gene regulation and its potential consequences on cancer interception. Association analysis of DEGs and DMRs identified that the mRNA expression of tumor suppressor gene BDH2, and oncogenes Ephas, Isg15 and Nos2 were correlated with the promoter CpG methylation status in the early-stage comparison groups indicating UA could regulate the oncogenes or tumor suppressor genes through modulating their promoter methylation at early stage of prostate tumorigenesis. Metabolomic study showed UA attenuated PTEN KO-regulated cancer-associated metabolism like purine metabolism/metabolites correlating with RNAseq findings, glycolysis/gluconeogenesis metabolism, as well as epigenetic-related metabolites pyruvate and lactate indicating UA plays a critical role in PTEN KO-mediated metabolic and epigenetic reprogramming and its consequences on cancer development. In this context, UA impacts metabolic rewiring causing epigenetic and transcriptomic reprogramming potentially contributing to the overall protection against prostate-specific PTEN KO-mediated PCa.

1. INTRODUCTION

Prostate cancer (PCa) is one of the most commonly diagnosed cancer and the third leading cause of cancer-related death in men in the United States (1). The phosphatase and tensin homologue (PTEN) tumor suppressor on chromosome 10 and the phosphoinositide 3-kinase (PI3K) signaling axis restrained by PTEN are among the most commonly altered pathways in primary human PCa (2, 3). Inactivation of PTEN is identified in ~20% of primary prostate tumor tissue samples at radical prostatectomy and in 50% of castration-resistant tumors. Loss of PTEN function results in activation of the PI3K-AKT (RAC-alpha serine/threonine-protein kinase) pathway, which is highly linked with adverse oncological outcomes, making PTEN a potentially useful genomic marker to distinguish indolent from aggressive disease in patients with clinically localized tumors. In addition, deletion of Pten has been highly linked to inflammation, in prostate-specific Pten KO mice, the expression of CXCL8/IL-8, a pro-inflammatory chemokine promoting tumorigenesis was significantly increased (4). So, the cellular regulatory role of PTEN points to the fact that PTEN signaling pathway as well as PTEN mutation-mediated biological alternations could be used as a primary target in cancer chemoprevention and therapy.

Cellular metabolism is the set of chemical reactions that occur in living organisms in order to allow organisms to grow and reproduce, maintain their structures, and respond to environmental changes. Cancer metabolism is an essential aspect of tumorigenesis, as cancer cells have increased energy requirements in comparison to normal cells. Thus, an enhanced metabolism is needed in order to accommodate tumor cells’ accelerated biological functions, including increased proliferation, vigorous migration during metastasis, and adaptation to different tissues from the primary invasion site (5). The characteristic metabolic hallmark of tumor metabolism is aerobic glycolysis or the Warburg effect. Unlike normal cells that produce energy mostly through the oxidation of pyruvate in the mitochondria, cancer cells predominantly produce energy via enhanced glycolysis in the cytosol, even under aerobic conditions. Aerobic glycolysis in cancers is the combined result of oncogenes, tumor suppressors, a hypoxic microenvironment and others (6). Understanding the complex cancer energy metabolism will help to develop new approaches for cancer interception. This study will cover the tumor suppressor gene PTEN-mediated metabolism, including the analysis of altered enzymatic gene expressions that are involved in cancer-related purine metabolism, and discuss current strategies of targeting metabolic pathways such as pyruvate metabolism for cancer interception at early-stage.

Epigenetics is the study of genomic alterations, such as DNA methylation and histone modifications, which alter DNA accessibility and chromatin structure, thus influencing patterns of gene expression, but without altering the genetic code itself (7). Epigenetic dysregulation occurs at every phase of PCa and plays critical role in PCa initiation, progression, and treatment resistance. These aberrations are responsible for silencing tumor-suppressor genes, activating oncogenic drivers, and driving therapy resistance (8). Specifically, CpG hyper-/hypo-methylation in DNA promoter regions is believed to play a key role in regulating gene expression, perhaps by interfering binding of transcription factors (9). Previous studies reported the epigenetic silencing of PTEN gene in melanoma with high percentages of PTEN methylation were associated with low PTEN transcription levels in melanoma, therefore, resulting in increased tumorigenesis (10). Interestingly, increasing evidence suggests that cellular metabolism plays an important role in cancers and that the cellular metabolism/metabolites are tightly linked to the basic epigenetic machinery (11). Epigenetic modifications including DNA methylation and histone acetylation are sensitive to cellular metabolic status (12). Strong molecular link between metabolic rewiring and epigenetic modifications through key metabolic intermediates, such as α-ketoglutarate (aKG), acetyl-CoA (AcCoA) of the tricarboxylic acid (TCA) cycle and S-adenosyl methionine (SAM) of methionine cycle which are co-factors for the epigenetic enzymes and work as hubs between epigenetic processes and therapeutic modalities (13–15). So, modulation of epigenetic such as DNA de-/methylation, histone de-/acetylation as well as transcriptomics through regulating metabolomics may prevent diseases and protective against cancer development and metastasis (16).

Ursolic acid (UA) is a natural pentacyclic triterpenoid carboxylic acid derived from medicinal herbs, fruits and vegetables (e.g. cranberry, blueberry, apple peels and mushrooms) and exerts a wide range of biological activities, including antioxidant, anti-inflammatory, chemopreventive as well as antitumor effects (17–19). UA has been reported to suppress the proliferation and induce apoptosis in both in vitro tumor cells and in vivo animal models, therefore, to inhibit tumor promotion, metastasis and angiogenesis (20). Our recent studies reported that UA decreased the CpG methylation of Nuclear factor E2-related factor 2 (Nrf2) promoter region, which was associated with the Nrf2 activation in mouse epidermal JB6 cells (21). In addition, UA has been reported to regulate several cellular metabolites and metabolism-related signaling pathways including SAM, methionine, glycolysis, and nucleotide sugars metabolism which are tightly coupled to epigenetic machinery and cancers (22). Lodi et al. (23) also reported that combinatorial treatment with natural compounds including UA in prostate cancer inhibits prostate tumor growth and leads to key modulations of cancer cell metabolism.

Our previous study (24) involved crossing the probacin (Pb)-Cre promoter transgenic mice and Pten-flox allelic mice to obtain prostate-specific Pten−/− (KO) mouse prostatic adenocarcinoma model, and found that Pten deletion drives global changes in DNA CpG methylation and transcriptomic gene expression which are highly associated with several inflammatory and immune molecular pathways, such as NF-kB signaling, IL-6 signaling and PI3K in B lymphocytes signaling, during PCa development. However, the role of the UA in regulating metabolic rewiring, CpG methylomic reprogramming, and transcriptomic network in blocking PTEN deletion-mediated biological alterations and eliciting cancer interception effect in PCa remains unknown. So, as a continuation of our Pten deletion study, previous Pten wild-type (WT) vs KO (24) incorporated with current UA treated Pten KO datasets were reanalyzed to examine 1) the role of Pten deletion in the cancer interception effect of UA in prostate-specific Pten KO prostatic adenocarcinoma mouse model; 2) the underlying intricate biological connectivity between metabolomic, epigenomic and transcriptomic regulation by UA.

2. MATERIALS AND METHODS

2.1. Chemicals and animal diets

UA was purchased from Toronto Research Chemicals (North York, ON, Canada) and blended into AIN-93M rodent diet (Research Diet, Inc. New Brunswick, NJ, USA) at a final concentration of 0.1 % (w/w). The diet was stored at 4 °C throughout the animal experimentation period. All mice were housed in our animal facility in accordance with the protocol approved (1–16) by the Rutgers University Institutional Animal Care and Use Committee (IACUC). All mice were maintained under standard 12-h light/12-h dark cycles with water and diet provided ad libitum unless otherwise specified.

2.2. Animal studies

A mouse model that uses the Pb-Cre promoter and Pten-flox has been commonly used to examine the role of Pten deletion in prostate tumorigenesis (25, 26). Previous studies reported that in prostate-specific Pten KO mice, at week 10, the low-grade (LG) prostatic intraepithelial neoplasia (PIN) was observed; at week 12, the high-grade (HG) PIN would appear, and adenocarcinoma developed starting from week 12 to week 20 (at week 20, 100 % of the mice developed adenocarcinoma) (27, 28). Pb-Cre4 mice (strain: B6.Cg-Tg(Pbsn-cre)4Prb/Nci) and Pten(flox/flox) mice (C;129S4-Ptentm1Hwu/J) obtained from National Cancer Institute and Jackson Laboratories, respectively, were used to crossbreed and generate the prostate-specific Pten KO mouse (29, 30). F2 generation prostate-specific Pten KO male offspring (Pb-cre/Pten(flox/flox)) were generated by crossing male Pb-Cre4 mice with female Pten(flox/flox) mice. The detailed prostate Pten KO breeding scheme was presented in our previous publication (24). For simplicity, Pb-cre/Pten(flox/flox) mice are referred to as Pten KO, and Pten(flox/flox) mice are considered as Pten WT. The mice were genotyped using PCR, and only the male mice that were Cre carriers and homozygous Pten(flox/flox) were used for the treatments. The detailed primers used for the Pb-Cre4 and Pten genotyping were also reported in our previous study (24).

2.3. Animal grouping

Pten KO mice were randomly assigned to four experimental groups (n=5). Briefly, 6-week (wk)-old Pten KO mice were fed with either AIN-93M control diet or 0.1 % UA diet for 6 wk (age: 12 wk old) and 14 wk (age: 20 wk old) and then sacrificed by CO₂ asphyxiation. Mouse prostates were collected immediately and either dissected into ventral & lateral prostate (VLP) and dorsal & lateral prostate (DLP) and snap-frozen and then stored at −80 °C for epigenomics and metabolomics analyses.

2.4. Isolation of nucleic acid and next-generation sequencing (NGS)

Total RNA and DNA were extracted from prostate tissues using an AllPrep DNA/RNA Mini Kit (Qiagen, Valencia, CA, USA). The quality and concentration of the extracted RNA and DNA were determined using an Agilent 2100 Bioanalyzer and a NanoDrop spectrophotometer, respectively. 8 RNA samples and 8 DNA samples [2 groups (Pten KO, Pten KO+UA) * 2 time points (6 and 14 wk) * 2 (n=2)] were subjected to RNA-Seq and SureSelect Methyl-Seq, respectively. Library preparation and sequencing were performed by RUCDR Infinite Biologics. The DNA samples were further processed using an Agilent Mouse SureSelect Methyl-Seq Target Enrichment System (Agilent Technologies Inc., Santa Clara, CA, USA) and sequenced on an Illumina NextSeq 500 instrument with 76-bp single-end reads, resulting in the generation of 30–45 million reads per sample. For RNA-Seq, 75-bp paired-end reads produced approximately 30 million reads per sample. The further details of the DNA-seq and RNA-seq procedures were described previously (31).

2.5. Bioinformatics analyses

2.5.1. RNA-seq

Cutadapt (32) was used to remove the Illumina Universal Adapter sequence. Hierarchical indexing for spliced alignment of transcripts (HISAT2) was adapted for aligning the reads to the mouse genome (mm10) and remove PCR duplicates (33). Genomic features with overlapping reads were counted by featureCounts (version 1.5.1) (34) and then data were further analyzed for differential expression genes (DEGs) with DEGSeq (version 1.36.0) in R (35).

2.5.2. DNA SureSelect methyl-seq

The DNA reads were aligned to the in silico bisulfite-converted mouse genome (mm10) using the Bismark (version 0.15.0) alignment algorithm (36). And the DMRfinder (version 0.1) was used to extract methylation counts and to cluster CpG sites into differentially methylated regions (DMRs) (37). Each DMR contained at least three CpG sites. Methylation differences greater than 0.1 (10%) with P-values < 0.05 were considered statistical significance. Genomic annotation was performed with ChIPseeker (version 1.10.3) in R (38).

2.5.3. Ingenuity pathway analysis (IPA)

DEGs with false discovery rates (FDR) adjusted P value (q value) < 0.05 coupled to log2-fold change >1.0 or <−1.0 were subjected to IPA (IPA 4.0, Ingenuity Systems). The input DEGs were mapped to the IPA knowledge base, and the biological functions, networks, and pathways related to Pten WT vs. KO (reanalyze from our published datasets) (24) and Pten KO vs KO+UA treatment in different time points (6 and 14 wk) were identified.

2.6. LC-MS metabolomic analysis

LC-MS metabolomic analysis was performed in Metabolomics Shared Resources, Rutgers Cancer Institute of New Jersey (CINJ) according to our previously reported protocol (39–41). Prostate VLP tissues from three mouse groups (Pten WT, KO and KO+UA) were subjected to organic extraction of cellular metabolites for metabolomic analysis. Briefly, 15–30 mg tissue were weighed and pulverized with Yttria Grinding ball using CryoMill at 20 Hz for 2 min to ensure completely homogenized tissue. Then, metabolites were extracted with (tissue weight (mg) * 40)/2 μl volume of cold 40:40:20 methanol:acetonitrile:water solution with 0.5% formic acid, and followed by 5 cycles of ice cold sonication with 30s ON and 30s OFF using Bioruptor UCD-200 sonication machine and another 10 min incubation on ice, and then sequentially neutralized with 50 μL/1mL extraction buffer of 15% NH₄HCO₃. The cleared supernatant was used for LC-MS metabolomic analysis. The tissue pellets were further lysate with RIPA buffer for the protein extraction for the normalization of LC-MS metabolomic data.

LC separation was performed on a XBridge BEH Amide column (2.1 mm × 150 mm, 2.5 μm particle size, 130 Å pore size; Waters) coupled with a Waters XBridge BEH XP VanGuard cartridge (2.1 mm × 5 mm, 2.5 μm particle size, 130 Å pore size) guard column. The solvent A prepared by water/acetonitrile (95:5, v/v) with 20 mM NH3AC and 20 mM NH₃OH at pH 9; and solvent B prepared by acetonitrile/water (80:20, v/v) with 20 mM NH₃AC and 20 mM NH₃OH at pH 9 in the following solvent B percentages over time: 0 min, 100%; 3 min, 100%; 3.2 min, 90%; 6.2 min, 90%; 6.5 min, 80%; 10.5 min, 80%; 10.7 min, 70%; 13.5 min, 70%; 13.7 min, 45%; 16 min, 45%; 16.5 min, 100%. The flow rate was set to 300 μL/min with an injection volume 5 μL. The column temperature was set at 25 °C. MS scans were obtained in both positive and negative ion modes with a resolution of 70,000 at m/z 200, in addition to an automatic gain control target of 3 × 10⁶ and m/z scan range of 72 to 1000.

2.7. Dosage information/Dosage regimen

Animal diets containing UA (≥95% purity, Sigma-Aldrich U6753) 0.1% w/w kg diet were prepared in AIN-93M control diets and the specially prepared diet excluded the use of antioxidants to avoid interference. The oxidation was prevented by frequent change of diet and storing it in a nitrogen atmosphere in the cold. The average daily consumption of feed for an adult 25 g mouse is 3–5 g, corresponds to 3–5 μg/day. Previous studies used 0.1%–2% UA-supplemented diets for the long-term treatment (≥ 6 weeks) in both healthy and disease animal models (42) and the toxicity studies indicated that the animals received daily doses of 1000 mg/kg/day via oral gavage for 90 days does not result in toxic effects and concluded that the no-observed-adverse-effect-level (NOAEL) for UA is likely to be higher than 1000 mg/kg/day. A clinical pharmacokinetic (PK) and safety study in healthy adult volunteers of UA at single oral doses up to 1000 mg also found no serious adverse event (43). In terms of the pharmacological effects, previous animal studies reported that 0.1%–2% UA-supplemented diets could initiate the therapeutic potential of UA (42). Based on these reports that the dose used in the current animal study is non-toxic and would potentially show pharmacological effects of UA.

2.8. Statistical analysis

Statistical significance was tested with one-way ANOVA followed by Dunnett’s post hoc test for differences among multiple experimental groups and with Student’s t-test for differences between two experimental groups. The values are presented as the mean ± standard deviation (SD). P-values ≤ 0.05 were considered statistically significant.

3. RESULTS

3.1. Overview of DMRs regulated by UA in the prostate of Pten KO mice

Our previous histology analysis showed that almost all glands of the 12 and 20 wk-old Pten prostate-specific KO mice developed low-grade (LG) or high-grade (HG) PIN. The severity of PIN and inflammation development gradually increased as Pten KO mice aged (24). HG PIN resembling chronic inflammation or prostatitis is considered as a precursor to prostate adenocarcinoma since it may convert into prostate cancer over time but may be rescued by chemopreventive agent to reduce the risk of prostate cancer. Hence, we treated the Pten KO mice with UA at different time points (12 and 20 wk age or 6 and 14 wk treatment). The size of prostates in Pten KO and UA treated Pten KO mice were measured as shown in Figure S1. Briefly, deletion of Pten resulted in progressively enlarged prostate lobes in an age-dependent manner and the sizes of prostates significantly different between Pten KO and Pten KO+UA mice.

To identify the overall regulatory effect of UA on Pten KO-induced DNA methylation alterations, we performed single base-pair resolution CpG Methyl-seq with DNA prostatic samples from Pten KO and KO+UA mice. Sequencing reads were aligned to an in-silico C-T converted mouse genome (mm10) and deduplicated. Individual CpG sites were clustered into DMRs, and the average methylation ratio for each DMR was calculated. We then collected DNA methylation data for a total of 137,897 DMRs. These DMRs were further annotated with gene features using ChIPseeker. As shown in Figure S2A, majority of the DMRs were detected in the promoters and the distal intergenic regions (> 3 kb upstream of the transcription start site (TSS) or downstream of the 3’ untranslated region (UTR)). Similarly, analysis of the distribution of DMRs based on number of CpGs and region showed that the number of CpG sites in the promoter, gene body, and downstream regions were much greater than that in other regions (Figure S2B). We next compared the DNA methylation levels in Pten KO and KO+UA groups. As shown in Figure S2C, no significant methylation difference was observed among each individual treatment group. However, CpG methylation in the promoters was much lower than that in other regions for these groups. Principal Component Analysis (PCA) revealed a clear separation for each treatment groups, showing that there were clear differences between Pten KO mice and Pten KO+UA groups (Figure S2D). The overall DMRs in Pten WT mice were reported in our previous study (24) which will be reanalyzed with Pten KO and KO+UA datasets to uncover the role of Pten deletion in the epigenetic regulation function of UA in prostate-specific Pten KO prostatic adenocarcinoma mouse model, as discussed below.

3.2. UA rescues epigenetic CpG methylation changes in the prostate of Pten KO mice

To examine the epigenetic regulatory effects of UA on Pten KO mice after 6 and 14 wk treatments, we compared the CpG Methylseq profiles between the Pten WT, KO and KO+UA groups. Firstly, the DMR regulations of Pten KO and UA protective effect on Pten KO-mediated CpG methylation were identified by comparing Pten WT vs. KO and KO vs. KO+UA counterparts with the threshold methylation difference ≥ 10%, P-value ≤ 0.05 (Figure 1A). Briefly, in 6 wk counterpart, Pten KO significantly increased 3986 and decreased 5330 DMRs while UA reversed 1514 (1514/3968) and 2121 (2121/5330) DMRs, respectively, indicating the epigenetic regulatory effect of UA on Pten KO-mediated CpG methylation. In 14 wk counterpart, Pten KO exacerbate the epigenetic modulation with extra hypomethylated DMRs (14109) compare to 6 wk counterparts, while UA reversed 4262 (4262/14109) Pten KO-regulated DMRs. These results indicate the later stage of PCa drives severe epigenetic CpG methylation which may be responsible for silencing tumor-suppressor genes and activating oncogenic drivers, etc. (8), while UA partially reversed the Pten KO-induced epigenetic alterations in both early and late stage (6 and 14 wk treatment or 12 and 20 wk age). Based on these findings, to further explore the effect of UA on Pten KO-mediated methylation modulations at different PCa stages (6 wk vs. 14 wk treatment or 12 wk vs. 20 wk age in current study), a cutoff with higher thresholds of methylation difference ≥ 20 % coupled to P-value ≤ 0.05 were used to identify and plot the UA-regulated DMRs by comparing Pten KO mice with Pten KO+UA mice (KO vs. KO+UA) at different stages (Figure 1B–C). Specifically, 307 DMRs were hypermethylated and 924 DMRs were hypomethylated after UA treatment for 6 wk while 274 and 957 DMRs were hyper- and hypomethylated, respectively, in response to the 14 wk UA treatment. Of the DMRs regulated by UA treatment, a total of 1038 DMRs were found to affect methylation in same directions after both 6 and 14 wk treatment (Figure 1D). Specifically, 194 DMRs that were hypermethylated in both 6 wk (194/307) and 14 wk (194/274) UA treatment. And almost all of the hypomethylated DMRs in 6 wk (844/924) and 14 wk (844/957) UA treatment groups were completely overlapped indicates that UA possesses the stable CpG methylation in the prostate of Pten KO mice.

Figure 1. DNA Methyl-seq profiles in Pten WT, Pten KO and UA treated Pten KO mice at 6- and 14-weeks.

(A) Venn diagram showing the DMRs in response to Pten WT vs KO and Pten KO vs KO+UA after 6- and 14-week. The cutoff threshold of methylation difference ≥ 10% or ≤ −10%, P-value ≤ 0.05; (B) MA plots showing DMRs in response to Pten KO mice treated with UA at 6- and 14-week with cutoffs of methylation difference ≥ 20% or ≤ −20% coupled to p<0.05; (C) Heatmap comparing the DMRs in response to Pten KO mice treated with UA for 6- and 14-week; (D) Venn diagrams showing the overlapped hyper- and hypo-methylated genes after 6 and 14 week of UA treatment in Pten KO mice.

3.3. UA mediates differentially expressed genes (DEGs) in the prostate of Pten KO mice

To identify gene expression changes in Pten KO mice as well as Pten KO mice treated with UA for 6 and 14 wk, RNA-seq was performed with RNA samples extracted from Pten WT, KO and KO+UA mice prostate tissues. Figure S3A shows that the counts and the raw gene expression overviews across the Pten KO and KO+UA groups were measured within the similar magnitude. PCA (Figure S3B) and Euclidean distance clustering (Figure S3C) showed the Pten KO groups to be clearly separated from 12 and 20 wk UA treatment groups. The DEGs overview of Pten WT mice were reported in our previous publication (24). To examine the effect of UA treatment on Pten KO-mediated gene regulation, the DEG profiles between Pten WT vs. KO and KO vs. KO+UA were further analyzed and plotted with the cutoff p-value ≤ 0.01 coupled to log2 fold change ≥ 1.0 or ≤ −1.0 (Figure 2A). When comparing the 6 wk groups, the MA plot showing that 2,398 and 803 genes were significantly up- and down-regulated after Pten KO respectively (Pten KO vs. WT); while 1,157 and 1,658 genes were significantly up- and down-regulated respectively, when Pten KO mice treated with UA diet (Pten KO vs. KO+UA). Within these 2 comparisons in the 6 wk counterparts, UA offsets 1,741 Pten KO-modulated (1,576 genes were up-regulated, while 165 were down-regulated after Pten KO) genes (Figure 2B). Similarly, the gene expression profiles in the 14 wk counterparts were shown in Figure 2C–D. In addition, 14 wk Pten KO exacerbate the gene regulation with an increased DEGs compare to that in 6 wk counterpart. However, 6 wk UA treatment elicited stronger gene regulatory effect compared to the 14 wk counterparts manifested as more filtered DEGs (Figure 2A&C). These results indicate that UA possesses more potent gene regulatory effects on the early stages of PCa which may elicit stronger cancer protective effects. Notably, UA significantly reversed or suppressed the Pten deletion-induced PCa related oncogenes such as Has3, Cfh and Msx1 overexpression (Figure 2E) indicating UA plays a crucial role in Pten KO-mediated gene regulation and its potential consequences on cancer prevention.

Figure 2. mRNA transcriptomic profiles regulated by Pten KO and the protective effects of UA on Pten KO-mediated gene regulation.

(A,C) MA plots showing DEGs in response to Pten WT vs KO and Pten KO vs KO+UA at 6-week (A) and 14-week (C) with cutoffs of p<0.01 and log2(fold change) ≥ 1.0 or ≤ −1.0; (B,D) Venn diagrams comparing the upregulated and downregulated genes in Pten WT vs KO and Pten KO vs KO+UA comparison groups at 6-week (B) and 14-week (D); (E) Prostate cancer-relevant oncogenes reported from RNA-seq analysis. The mRNA levels were shown as RNAseq reads per million in different groups (Pten KO, WT and KO+UA) at 6- and 14-week time points. All the data are presented as the means ±SEM of three independent experiments. *, P < 0.05 and **, P < 0.01 indicate significant differences between two groups. Student’s t test was used to calculate the significance of the differences between groups.

3.4. UA reverses Pten KO-induced signaling pathway

The regulated genes shown in Figure 3A&C which were filtered with the thresholds p-value < 0.01 coupled to log2 fold change ≥ 1.0 or ≤ −1.0 from the RNAseq datasets were further used to conduct canonical pathway analysis using IPA to uncover potential biological functions of Pten deletion the treatment effect of UA. Briefly, Pten KO significantly (-logP<1.3 or P<0.05) regulated 163 and 193 signaling pathways in 6 and 14 wk groups, respectively, compared to their corresponding Pten WT mice, while UA modulated 167 and 39 signaling pathways after 6 and 14 wk treatment respectively. These results indicate that the signaling pathways in later stage (14 wk treatment or 20 wk age) of PCa are highly modulated compared to early stage (6 wk treatment or 12 wk age), while UA exerts stronger signaling protective effects in early stage rather than late stage (Supplemental File 1). To further explore the effects of UA on Pten KO-mediated signaling pathways, the shared pathways between Pten WT vs. KO and KO vs. KO+UA comparison groups in 6 & 14 wk were filtered. Top 25 shared signaling pathways were further plotted with their ‘activation z scores’ (Figure 3A and Supplemental File 2). Among the 6 wk counterpart, 126 and 11 signaling pathways were significantly activated and inhibited by Pten KO while UA completely offsets these Pten KO-mediated signaling pathways (Supplemental File 2). For the 12 wk counterparts, UA only reversed 34 Pten KO-mediated signaling pathways (30 upregulated and 4 downregulated) (Supplemental File 2). These results indicate that the effects exerted by UA treatment on the pathways were stronger at 6 wk than that at 14 wk, consistent with earlier findings. Interestingly, blockade of the Pten KO-induced AMP-activated protein kinase (AMPK: one of the central regulators of cellular and organismal metabolism) signaling pathway by UA treatment indicating that UA could play critical roles in regulating growth and reprogramming metabolism as well as cellular processes such as autophagy and cellular polarity in Pten KO mice prostate (44). UA-mediated AMPK signaling inhibition may further block glycolysis metabolism which has been considered as an attractive anticancer strategy (45, 46). Other inflammatory response and immune response pathways, such as NF-kB signaling, iNOS signaling, PI3K in B lymphocytes signaling, LXR/RXR activation, IL-6 signaling, IL-8 signaling, CD28 signaling in T helper cells and iCOS-iCOSL signaling in T helper cells, among others were also highly regulated in a reverse manner by UA in Pten KO mice. It was not surprising that these findings suggested the inhibitory effects of UA on the level of inflammation induced by Pten deletion at 6 wk are stronger than those at 14 wk by affecting inflammatory response and immune response pathways. In this context, the totality of these modulated signaling pathways would contribute to the biological effects of UA and provide new windows of therapeutic opportunity against PCa in early stages of tumor development.

Figure 3. Ingenuity pathway analysis (IPA) and correlation analysis of DEGs with DMRs.

(A) Top 25 shared signaling pathways which modulated by Pten KO and reversed by UA treatment at 6- and 14-week time points; (B) Correlation analysis between gene expression (DEGs) and DNA methylation (DMRs) in Pten KO vs KO+UA groups at different time points. The DMRs which were negatively correlated with their corresponding gene expression with the cutoff of methylation difference ≥ 10% or ≤ −10% coupled to gene expression log2(fold change) ≥ 2.0 or ≤ −2.0 were selected as show in the blue boxes (locate in the upper left and bottom right corners). The DMR locations (gene features: Region, Body, Downstream and Promoter, etc.) are indicated by the colors. (C) mRNA expression levels of tumor suppressor genes/oncogenes and their corresponding promoter CpG methylation ratio (red font numbers) in Pten KO and KO+UA mice. All the data are presented as the means ±SEM of three independent experiments. *, P < 0.05 and **, P < 0.01 indicate significant differences between two groups. Student’s t test was used to calculate the significance of the differences between groups.

3.5. Integrated methyl-seq and RNA-seq analysis identifies DNA methylation and gene expression patterns

The current dogma of DNA methylation implicates regulation of transcription of many downstream target genes in mammalian cells (47) and DMRs have been shown to play an important role in the transcriptional control of many crucial genes. It is generally accepted that the promoter CpG methylation ratio of DMRs is negatively associated with their transcription of downstream genes. The correlation between DEGs and DMRs induced by UA treatments were further analyzed to examine the underlying linkage between DNA methylation and transcriptomic gene expression. A cutoff threshold of ≥ 0.1 (10%) or ≤ −0.1 (10%) for DNA methylation changes combined with log2-fold change ≥ 2.0 or ≤ −2.0 for gene expression were applied for the correlation analysis (Figure 3B and Supplemental File 3). Briefly, with the 6 wk groups, a total of 55 DMRs in the promoter regions were negatively correlated (32 genes’ promoter methylation increased with downregulated mRNA expression; 23 gene promoter methylation decreased with upregulated mRNA expression) with their corresponding gene expression after UA treatment comparing Pten KO control diet versus UA treated (Supplemental File 3). These negatively correlated genes were further used for the starburst plot and the negatively correlated genes were highlighted in the blue boxes (Figure 3B). Interestingly, this associative analysis revealed that UA significantly decreased the tumor suppressor gene 3-hydroxybutyrate dehydrogenase 2 (BDH2)’s (48) promoter methylation by 11.2% (from 17.4% in Pten KO to 6.2% in Pten KO+UA) and increased the gene expression by 2.7-fold comparing the Pten KO at 6 wk. In addition, the promoter methylation of PCa relevant oncogenes ephrin type-A receptor 2 (Ephas) (49), interferon-stimulated gene 15 (Isg15) (50) and nitric oxide synthase (Nos2) (51) were increased by 15.7%, 26.3% and 25.0% respectively, while the gene expression were decreased by 86.4%, 83.0% and 98.0% respectively, compared to Pten KO at 6 wk (Figure 3C and Supplemental File 3). These results indicated that UA could regulate the tumor promoter/suppressor genes through modulating their promoter methylation at the early stage of PCa. Collectively, these results suggest an important subset of genes associated with PCa development and the chemopreventive activity of UA was identified through investigation of the correlation between mRNA expression and DNA CpG methylation.

3.6. UA attenuates Pten KO-driven metabolomic reprogramming

To unravel the potential underlying molecular links between metabolic reprogramming and epigenetic modifications as well as the impact of UA on Pten KO-modulated epigenetic effects on carcinogenesis through regulating mitochondrial metabolic pathways and metabolites, the prostate tissues collected from Pten WT, KO and KO+UA mice were used to perform the LC-MS metabolomic analysis. A total of 205 metabolites were identified (Supplemental File 4) of which top 30 regulated metabolites (one-way ANOVA computed P value) across 3 groups (n=5 in each group) were further plotted (Figure 4A). Pathway analysis (Figure 4B) combined with pathway enrichment analysis between Pten WT vs. KO and KO vs. KO+UA comparison groups revealed that both Pten deletion and UA treatment significantly reprogram the cancer associated metabolisms including purine metabolism, glycolysis/gluconeogenesis as well as the epigenetic related metabolism such as pyruvate metabolism (Figure 4B and Table S1 and S2). These results suggested that Pten KO and UA alter the mitochondrial metabolism/metabolites, and also linking the down-stream impact on epigenetic reprogramming and PCa, as discussed below.

Figure 4. Metabolomic profiles of prostate samples from Pten WT, KO and UA treated Pten KO mice.

(A) Top 30 regulated metabolites after Pten KO and UA treatment (n=5); (B) Metabolism pathway analysis of Pten WT vs KO and Pten KO vs KO+UA comparison groups; (C) The major metabolites quantification of purine metabolism in Pten WT, KO and UA treated Pten KO mice. All the data are presented as the means ±SEM of three independent experiments. *, P < 0.05 and **, P < 0.01 indicate significant differences between two groups. Student’s t test was used to calculate the significance of the differences between groups.

3.7. Pten KO and UA regulates epigenetics and PCa via cellular metabolism

Purine or purine metabolism is the most abundant and most critical metabolic substrate for all living organisms by providing essential components for DNA and RNA which is tightly linked to cancers (52). UA treatment completely reversed the Pten KO-upregulated cancer-relevant metabolisms including purine metabolism/metabolites (Figure 4C). In the context of purine metabolism, RNAseq revealed that the transcription levels of purine biosynthesis pathway catalyzing enzymes including Adenylosuccinate Synthase (Adss), Adss1, Phosphoribosylaminoimidazole Carboxylase And Phospho-ribosyl-amino-imidazole-succino-carboxamide Synthase (Paics) and Guanine Monophosphate Synthase (Gmps) were also upregulated by Pten KO and attenuated by UA (Figure 5A), which are highly in agreement with the metabolomic profiles. The metabolomic along with the transcriptomic results from RNAseq indicated that UA would potentially exert its cancer preventive or therapeutic effects via the attenuation/reversal of Pten KO-induced metabolic rewiring. Therefore, these results provide the evolving landscape of purine synthesis inhibitors/purine metabolism enzyme inhibitors can be potentially exploited for the inception of PCa.

Figure 5.

(A) Quantified purine metabolism catalyze enzyme mRNA expression levels (the mRNA levels were shown as RNAseq reads per million) in Pten WT, KO and UA treated Pten KO mice at different time points; (B) Quantified epigenetic associated metabolites in Pten WT, KO and UA treated Pten KO mice groups. *, P < 0.05 and **, P < 0.01 indicate significant differences between two groups. Student’s t test was used to calculate the significance of the differences between groups..

The Warburg effect is an alteration in the metabolism of most cancer cells that enables them to convert glucose into lactate, even in the presence of abundant oxygen; a process known as aerobic glycolysis (53). Pten KO significantly regulated the Glycolysis/Gluconeogenesis (a reverse glycolysis pathway) (Figure 4B) indicating the effect the Pten KO in reprogramming cancer-associated cellular metabolism. Recent research reports that the aberrant metabolism in cancer is not only involved in maintaining a high proliferative rate or survival but also have consequences that impact epigenetic mechanisms such as DNA methylation, histone post-translational modifications, triggering oncogenes activation or loss of tumor suppressor genes expression resulting in tumor development (54). In current study, UA significantly blocked the Pten KO-regulated pyruvate metabolic metabolites including pyruvate and lactate (Figure 5B) which are highly linked with to the basic epigenetic machinery (55, 56). Advances in understanding of prostate cancer metabolism might help to explain many of the biological responses such as epigenetic modulations that are induced by treatment, which might, in turn, lead to the attainment of more effective therapeutic effects.

4. DISCUSSION

Carcinogenesis is caused by a cumulative and multistage process that primarily consists of initiation, promotion, and progression stages. PTEN alteration is strongly implicated in PCa development (57). To achieve Pten prostate-specific deletion, we crossed Ptenloxp/loxp mice (30) to the ARR2 Probasin (Pb)-Cre transgenic line, PB-Cre4, in which the Cre recombinase is under the control of a modified rat prostate-specific probasin (PB) promoter (29). Since Cre-mediated recombination event is a unidirectional process, cells with Cre-mediated gene deletion are likely to increase and accumulate over time (28). Previous studies also demonstrated that Pten KO lead to a significant shortened latency of PIN formation and results in prostate cancer progression to a metastatic stage which are also in agreement with our previous (24) and current findings. PTEN-controlled signaling pathways are frequently altered in human PCa and the major function of PTEN relies on its phosphatase activity and subsequent antagonism of the pro-growth phosphatidylinositol 3′-kinase (PI3K)/AKT pathway (58, 59). Alterations in Pten/PI3K/AKT and p53 can all impact cellular metabolism (60). The PI3K pathway has been shown to play a major role in tumor proliferation and survival for a wide variety of human cancers (61). Activation of PI3K results in the downstream activation of AKT and stabilization of hypoxia-inducible factor (HIF)-1. The PI3K enzyme itself antagonizes the tumor-suppressor PTEN, and the loss of PTEN increases glycolysis by activation of AKT and HIF-1 (62). Interestingly, we found UA significantly decreased the PI3K/AKT Signaling in Pten KO mice (P<0.05) (Supplemental File 1) indicating the potential PCa cancer prevention effects of UA.

Recent studies also suggest that PTEN may function through AKT-independent mechanisms (62, 63). Chronic inflammation is now known to contribute to several forms of human cancers, with an estimated 20% of adult cancers including PCa attributable to chronic inflammatory conditions. Men with chronic inflammation in non-cancerous prostate tissue may have nearly twice the risk of actually having prostate cancer (63). In current study, IPA pathway analysis revealed that Pten deletion significantly modulates the inflammation response pathways such as NF-κB signaling, iNOS signaling, PI3K in B lymphocytes signaling, IL-6 signaling, IL-8 signaling and CD28 signaling in T helper cells, among others. Most importantly, UA treatment offset these Pten KO-induced inflammatory signaling responses (Figure 3A and Supplemental file 1&2) indicating the potential anti-inflammatory effects of UA which would further contribute to its cancer interception effect. Specifically, NF-κB activation associated with inflammation are known to contribute to prostate cancer malignancy. Inflammatory signals have also been associated with the development of castration resistance and resistance against other androgen depletion strategies, which is a major therapeutic challenge (64). Chronic inflammation also has been reported highly associated with epigenetic alterations mediated by DNA and histone modifications, thus driving changes in the expression of many inflammation-related genes, such as IL1R1, IL-1β, cyclooxygenase-2 (COX2), CXCL14, CCL25, CXCL6, IL13, IL17C, and IL4R (65–67). Epidemiological data also indicate that patients diagnosed with chronic inflammatory prostatitis have an increased risk of developing PCa at the later age (24). Hence, our results underscore the importance of inflammation/anti-inflammation in PCa progression/prevention and provide a plausible explanation based on whole-genome methylation and transcription profiling.

Mutations that activate oncogenes or inactivate tumor suppressors can significantly affect activities of metabolic enzymes and have a key role in aerobic glycolysis and other metabolism of cancer (62, 68). AKT stimulates glycolysis by increasing the expression and membrane translocation of glucose transporters, and also by phosphorylation of glycolytic enzymes, such as hexokinase (HK) and phosphofructokinase (PFK) (62, 69). Moreover, AKT activates mammalian target of rapamycin (mTOR), which indirectly affects other metabolic pathways by activating HIF-1, even under normoxic conditions (69). p53 has been shown an inhibitory effect on glycolysis by upregulating the expression of TP53-induced glycolysis and apoptosis regulator (TIGAR), which decreases fructose-2,6-bisphosphatase (Fru-2,6-P2) by dephosphorylation (70). Increased expression of TIGAR results in a decreased level of Fru-2,6-P2 and a decreased glycolytic rate (71). Hence, loss of p53 induced by Pten deletion at 14 wk time point (Figure 3A) may shifts metabolism from mitochondrial respiration towards glycolysis. In current study, metabolomic analysis revealed UA blocks Pten KO-induced AMPK signaling pathway (Supplemental file 1&2) indicating UA plays important roles in regulating cell growth and reprogramming metabolism as well as cellular processes such as autophagy and cell polarity which are mediated by AMPK pathway (72). AMPK may act in cancer cells as a metabolic gatekeeper that functions to establish metabolic checkpoints that limit cell division, and its loss of function would enhance tumorigenesis and tumor progression (73). In addition, AMPK signaling also has been reported to regulates the Warburg effect, one of the characteristic features of cancer cell metabolism as aforementioned, in cancer cells and suppresses tumor growth in vivo (65). UA-mediated AMPK signaling inhibition may also further regulates glycolysis metabolism which has been considered an attractive anticancer strategy (45, 46). The defining distinction between neoplastic cells and their normal counterparts is the unregulated and increased rate of growth of the former. Purines are basic components of nucleotides in cell proliferation which provide essential components for DNA and RNA which is also tightly linked with cancers (52). Under the conditions with higher requirement for purine nucleotides, such as dividing cells and tumor cells, the de novo biosynthetic pathway is fundamental to replenish the purine pool. Nucleotide synthesis such as purine is a frequently limiting factor of proliferation, therefore, purine biosynthesis and its associated mitochondrial pathways have been targeted by chemotherapeutic agents for decades (74). The typical approach is by direct inhibition of this pathway using purine antimetabolites, analogs of nucleotides, or their precursors acting as competitive inhibitors. These have been proven to be effective treatments acting to stall DNA replication or cause apoptosis via DNA damage, and they reflect a significant percentage of currently available cancer treatment. Another class of purine antimetabolites, the purine deoxy nucleoside analogs fludarabine, cladribine, clofarabine, nelarabine, and pentostatin are US Food and Drug Administration (FDA)-approved drugs for the treatment of cancers (75). However, none of these currently approved drugs target purine biosynthesis directly; instead, they target up stream’s input availability and down stream’s utilization of synthesized purines. While several biosynthetic inhibitors are in development, a better understanding of the precise molecular mechanisms of these agents and identification of new enzyme and metabolite targets is crucial for improving options for cancer treatment (76). So, the regulation effects of UA on Pten KO-mediated purine metabolism (Figure 4B&C) along with its regulations on the biosynthesis catalyze enzymes including Adss, Adss1, Paics and Gmps may provide new therapeutic targets on the purine metabolism for cancer interception.

Recent studies show that the aberrant metabolism in cancer is not only involved in maintaining a high proliferative rate or survival but also have consequences in epigenetic reprogramming through key metabolic intermediates, such as pyruvate, lactate, AcCoA and aKG which work as hubs between epigenetic processes and oncogenes activation or loss of tumor suppressor genes expression (77). In many cancer cells, pyruvate is usually processed into lactate, which is actively transported to the extracellular matrix due to the upregulation of monocarboxylate 1 (MCT1) and giving the cells a high glycolytic rate. In current study, Pten KO significantly decreased the pyruvate and increased lactate production while UA treatment completely offsets Pten KO-modulated pyruvate and lactate (Figure 5B) indicating the blockade effect of UA on Pten KO-promoted pyruvate to lactate conversion. Lactate plays a key role in regulating gene transcription by inhibiting the histone deacetylase (HDAC) enzymes, promoting hyperacetylation in nucleosomes and active transcriptional state (78). Some studies also mentioned that the histone H4 acetylation levels increase when cells are treated with lactate, promoting changes in gene expression that favors the cancer establishment (79). In breast cancer, the overproduction of lactate induces tumor growth by demethylation of HIF-1α in patients’ tissue (80).

5. CONCLUSIONS

In summary, our current study integrates the latest Methyl-seq, RNA-seq and LC/MS/MS technologies in dissecting the potential underlying mechanism of epigenomic CpG methylation, mRNA transcriptomic gene expression and mitochondrial metabolic rewiring in Pten prostate-specific knockout mice after treatment with chemopreventive phytochemical UA. The results provide the experimental in vivo evidence for the interconnectivity of metabolomic, epigenomic, and transcriptomic effects of UA in Pten prostate-specific KO mouse model. UA rescues the metabolic profile, reprograms epigenetic CpG methylation, driving alterations of phenotypic gene expression, de-/activated signaling pathways in prostate-specific Pten KO mice. The correlative analyses of multi-omics data revealed the underlying intricated biological connectivity between metabolomic, epigenomic and transcriptomic regulation by UA. Thus, the overall cancer interception effect of UA may exhibit through regulation of metabolic rewiring and epigenetic reprogramming, making it a promising chemopreventive agent to reduce the risk of PCa. However, further research is needed for the translation of UA from preclinical model to clinic.

ABBREVIATIONS

AcCoA

acetyl-CoA

aKG

α-ketoglutarate

DEGs

differentially expressed genes

DMRs

differentially methylated regions

IPA

Ingenuity pathway analysis (IPA)

KO

knockout

NGS

next-generation sequencing

Nrf2

nuclear factor E2-related factor 2

PI3K

phosphoinositide 3-kinase

PCa

prostate cancer

PIN

prostatic intraepithelial neoplasia

PTEN

phosphatase and tensin homologue deleted on chromosome 10

SAM

S-adenosyl methionine

TCA

tricarboxylic acid

UA

Ursolic acid

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in the methods, results and/or supplementary material of this article.