Identifying the role of apolipoprotein A-I in prostate cancer.

Although localized prostate cancer (PCa) can be cured by prostatectomy and radiotherapy, the development of effective therapeutic approaches for advanced prostate cancer, including castration-resistant PCa (CRPC) and neuroendocrine PCa (NEPC), is lagging far behind. Identifying a novel prognostic and diagnostic biomarker for early diagnosis and intervention is an urgent clinical need. Here, we report that apolipoprotein A-I (ApoA-I), the major component of high-density lipoprotein (HDL), is upregulated in PCa based on both bioinformatics and experimental evidence. The fact that advanced PCa shows strong ApoA-I expression reflects its potential role in driving therapeutic resistance and disease progression by reprogramming the lipid metabolic network of tumor cells. Molecularly, ApoA-I is regulated by MYC, a frequently amplified oncogene in late-stage PCa. Altogether, our findings have revealed a novel indicator to predict prognosis and recurrence, which would benefit patients who are prone to progress to metastasis or even NEPC, which is the lethal subtype of PCa.

INTRODUCTION

Prostate cancer (PCa) is the most commonly diagnosed noncutaneous cancer in men, especially in Western countries, which leads to approximately 30 000 deaths yearly in the USA.1 Although the androgen receptor (AR) signaling axis plays an extremely critical role in diagnosis and treatment, an increasing number of limitations have been uncovered. For instance, screening of the AR-regulated effector prostate-specific antigen (PSA), which is widely used to determine the likelihood of PCa in the clinic, has been shown to cause aggressive overtreatment in patients who have already been treated sufficiently with active surveillance.2 Moreover, PSA is not an ideal tool for some extremely lethal subtypes of PCa, such as neuroendocrine PCa (NEPC), which loses its luminal identity (AR and PSA), but displays distinct NE markers (enolase 2 [ENO2] and chromogranin A [CHGA]).3 Additionally, despite the initial effectiveness, the efficacy of AR-directed therapies is rather limited even with the second generation of anti-androgen drugs, such as abiraterone and enzalutamide.4 Therefore, identifying a biomarker whose level may be strongly associated with disease aggressiveness and development as well as exploiting a novel therapeutic target to compensate for the insufficiency of AR-targeted therapies is warranted.

The rewiring of metabolism has been regarded as one of the significant hallmarks of cancer for many decades since the Warburg effect was recognized, which describes the distinct preference of cancer cells for generating energy predominantly through glycolysis rather than the much more efficient oxidative phosphorylation (OXPHOS) pathway, regardless of the presence of oxygen.5 In contrast to a large number of solid cancers that exhibit this reprogrammed glucose metabolic trait, PCa is generally a nonglycolytic cancer type,6 which is supported by the clinical fact that PCa is not an ideal candidate for F-deoxyglucose (FDG)-positron emission tomography (PET) detection due to the low absorption rate of infused radiolabeled glucose by PCa tumor cells.7 Instead of glucose, lipids are one of the critical nutrients and energy suppliers in PCa, as shown by the hyperactivity of lipid synthesis.8 In addition to the utilization of de novo lipogenesis, PCa can also consume exogenous lipids such as androgens to support tumor cell growth.9 The dependency on lipids is even more notable in advanced PCa, manifesting as elevated levels of a panel of rate-limiting enzymes involved in both lipid synthesis and uptake.9 Therefore, the hypermetabolic rates of lipids represent a promising diagnostic and prognostic indicator of PCa initiation and progression.

In addition to the enzymes that are directly responsible for lipid synthesis and utilization, there is another specialized category of proteins that mediate lipid transport to improve their distribution and usage, that are known as lipoproteins. Apolipoprotein A-I (ApoA-I) is the major component of high-density lipoprotein (HDL).10 It has been implicated as a pleiotropic factor, as it not only regulates cholesterol trafficking to reduce the risks of cardiovascular disease but also functions in cancer contexts. Given the canonical role of HDL as “good cholesterol ” that ameliorates atherogenic processes, it is conceivable that ApoA-I may be protective against cancer.10 Indeed, reduced protein levels of ApoA-I have been observed in both cancerous tissue and the serum of patients, particularly within those carrying gastrointestinal tract carcinoma.10 However, other published data demonstrate that ApoA-I is a potential oncogenic player in recurrent head-and-neck squamous cell carcinoma, retinoblastoma, and bladder cancer and a disease progression driver in breast cancer.11121314 This background suggests that ApoA-I can be either pro-oncogenic or tumor suppressive, depending on the cancer context. However, its role in PCa remains largely unexplored.

In this study, we identified a new role of ApoA-I as a metabolic biomarker strongly associated with carcinogenesis and tumor evolution from multiple perspectives. We also propose a putative therapeutic regimen that targets ApoA-I to benefit advanced PCa patients who have few treatment choices.

MATERIALS AND METHODS

Cell culture

LNCaP, C4-2, PC3, and DU145 cells were cultured in Roswell Park Memorial Institute (RPMI)-1640 medium (Gibco, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin (Gibco). LAPC4 cells were maintained in Iscove's modified Dulbecco medium (IMDM; Gibco). RWPE-1 cells were maintained in keratinocyte serum-free medium (Gibco). Androgen deprivation therapy was modeled by charcoal-stripped FBS (CS-FBS; Gibco) application. Cells were preserved at 37°C in 5% CO2.

Gene silencing and overexpression

APOA1 knockdown was achieved by stable transduction of two sets of short hairpin RNA (shRNA) purchased from Sigma (TRCN0000371319 and TRCN0000371327; St. Louis, MO, USA). APOA1 complementary DNA (cDNA) was purchased from Origene (Rockville, MD, USA). All the shRNA and cDNA vectors were further packaged into lentiviruses. Briefly, each lentiviral vector was co-transfected with packaging vectors (pMDL, pVSVg, and pRev) into 293T cells by using Lipofectamine 3000 reagent (Invitrogen, Waltham, MA, USA). After 48 h, the viral particles were collected and purified with a 0.22-μm filter. Cells were infected by viruses followed by puromycin selection. Knockdown of MYC was performed by using two sets of small interfering RNAs (siRNA; Santa Cruz, Dallas, TX, USA). The efficiency of overexpression and knockdown was validated by western blot.

Cell proliferation, colony formation, and transwell assays

Cells were seeded in 96-well plates at an initial density of 1000 cells per well. At each indicated time point, cell viability was determined with the MTS reagent (Promega, Madison, WI, USA) according to the manufacturer's protocol. The colony formation assay was performed as described previously.15 A total of 1 × 104 cells were plated on a 60-mm cell culture dish and allowed to grow for up to 14 days. Colonies were fixed with cold methanol and then stained with 1% crystal violet. To determine the cell migration capability, 1 × 104 cells were plated in an 8-μm pore Boyden chamber (Sigma) with medium supplemented with 10% FBS surrounding the chamber in a 24-well plate. Chambers were coated with basement membrane extract (BME) solution and coating buffer. After 48 h of incubation, crystal violet was used to stain the migrated cells.

β-gal staining senescence measurement

LNCaP cells were treated with CS-FBS. At each indicated time point, cells were fixed and stained to determine the senescence-associated β-gal activity (Cell Signaling, Danvers, MA, USA). Positive cells were counted in several randomly selected fields to obtain the standard deviation.

Metabolite profiling

Cells were seeded in 6-well plates and grown until 80% confluence was reached. Then, the cells were extracted with prechilled 80% methanol. Macromolecules and debris were removed by centrifugation at 20 000g (Microfuge 20R, Beckman Coulter, Indianapolis, IN, USA) for 20 min at 4°C. The supernatant was dried by a speed vacuum at room temperature for 3 h. Dry pellets were further analyzed by liquid chromatography-mass spectrometry (LC/MS). The amount of HDL in the cell culture medium was determined by an HDL enzyme-linked immunosorbent assay (ELISA) kit (MYBioSource, San Diego, CA, USA) according to the manufacturer's instructions.

Gene expression and correlation analysis

Datasets were downloaded from the Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo) or cBioPortal (https://www.cbioportal.org). Relative APOA1 expression was compared between different groups (e.g., normal and primary tumor, primary tumor and metastasis, and primary-adeno and NEPC). Correlation matrices were computed by the Pearson's method.

Western blot

Cell lysates were prepared with RIPA buffer supplemented with protease and phosphatase inhibitors. The protein concentration was determined with the Bradford assay. Equal amounts of protein (20 μg) were loaded onto 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels. The protein was then transferred to polyvinylidene difluoride (PVDF) membranes followed by blocking with 5% nonfat milk. The following primary antibodies were applied with overnight incubation: anti-ApoA-I (1:200 dilution; Santa Cruz), anti-N-Myc (1:200 dilution; Santa Cruz), anti-c-Myc (1:1000 dilution; Abcam, Cambridge, MA, USA), and anti-actin (1:200 dilution; Santa Cruz).

RNA extraction and quantitative real-time PCR

Total RNA was extracted by using TRIzol reagent (Invitrogen) and then reverse-transcribed into cDNA by applying a PrimeScript RT reagent kit (Takara Bio, Shiga, Japan) according to the manufacturer's instructions. Quantitative polymerase chain reaction (qPCR) was performed by using SYBR Taq Premix (Quanta, Beverly, MA, USA). The primer sequences were as follows: APOA1, forward 5'- AAAGCTGCGGTGCTGACCTT-3', reverse 5'-ATCTTGCTGCCAGAAATGCCG-3'; actin, forward 5'-AGAGCTACGAGCTGCCTGAC-3', reverse 5'-AGCACTGTGTTGGCGTACAG-3'.

Immunohistochemistry (IHC)

Formalin-fixed and paraffin-embedded samples from PCa patients were obtained from prostate biopsy, prostatectomy, and transurethral resection of prostate (TURP) at the Department of Urological Oncology, The First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, University of Science and Technology of China (Hefei, China). As described previously,16 5-μm-thick paraffin sections were further prepared. The tissue slides were sequentially deparaffinized and rehydrated with xylene with graded ethanol. Endogenous peroxidase activity was blocked in 3% H2O2 for 10 min, and the tissue was boiled in citrate buffer for heat-induced antigen retrieval. The tissue was then probed with ApoA-I antibody (1:100 dilution), and the nuclei were stained with hematoxylin. The intensity was assessed by the Q-score.

Statistical analyses

Statistically significant differences between two groups were determined by using a two-tailed paired t-test. Comparisons between more than two groups were carried out by analysis of variance (ANOVA) with Tukey's multiple-comparison test. Differences were considered statistically significant at P < 0.05. The enriched gene signature was evaluated by gene set enrichment analysis (GSEA; www.gsea-msigdb.org). Metabolic pathway enrichment and impact analyses were performed with MetaboAnalyst online software (www.metaboanalyst.ca). All statistical analyses were performed with GraphPad Prism software (Insightful Science, San Diego, CA, USA), Microsoft Excel (Microsoft, Redmond, WA, USA), and R (Revolution Analytics, Mountain View, CA, USA).

RESULTS

ApoA-I expression is upregulated in PCa

To determine the potential role of ApoA-I in PCa, we compared APOA1 gene expression in normal prostatic tissues and primary tumors via bioinformatics analyses.171819 A panel of datasets from various studies171819 showed a consistent expression pattern, in which APOA1 was upregulated in the primary PCa samples (). IHC staining of paraffin tissues using ApoA-I antibody further validated the elevated protein level of ApoA-I in human PCa samples (Figure and ). Consistently, in vitro, ApoA-I expression was barely observed in the normal epithelial prostatic cell line RWPE-1, while human PCa cells displayed substantial protein levels of ApoA-I, as detected by western blot (). Loss of tumor suppressor genes (TSGs) is known to facilitate tumorigenesis.20 Phosphatase and tensin homolog (PTEN) is among the most frequently lost or mutated TSGs in PCa, which is observed in up to 70% of localized PCa.21 To confirm that the increase in APOA1 in PCa is universal and can be caused by a genetically altered background, we interrogated a genetically engineered mouse model (GEMM) sequencing dataset20 and found that primary PCa tumors resulting from PTEN deficiency displayed significantly higher levels of APOA1, whereas APOA1 was undetectable in tissues with wild-type PTEN (). These findings suggest that PCa harbors a high level of ApoA-I, and, thus, APOA1 might be a potential oncogene involved in PCa initiation.

ApoA-I expression is upregulated in PCa. (a) APOA1 mRNA levels in PCa datasets (Yu et al.17 2004; Taylor et al.18 2010; and Grasso et al.19 2012) separated by normal versus primary tumor. (b) Representative ApoA-I IHC images of benign prostate, primary PCa, CRPC, and NEPC tissue (scale bars = 100 μm). (c) Q-scoring analysis of tissue staining intensity using ApoA-I antibody. (d) Western blot showing the protein level of ApoA-I in various human prostate cell lines. Actin was loaded as the internal control. (e) Expression levels of APOA1 mRNA in GEMM samples from WT, PT (PTEN KO), PT (PTEN, RB1 KO), and Met (PTEN, RB1 KO) mice. Two-tailed t-test. *P < 0.05, **P < 0.01, ***P < 0.001. ApoA-I: apolipoprotein A-I; PCa: prostate cancer; IHC: immunohistochemistry; PT: primary tumor; Met: metastasis; CRPC: castration-resistant PCa; NEPC: neuroendocrine PCa; GEMM: genetically engineered mouse model; PTEN: phosphatase and tensin homolog; KO: knock out; WT: wild-type; RB1: RB transcriptional corepressor 1.

ApoA-I is associated with disease progression

To test whether ApoA-I has the potential to drive the disease development of PCa, we next focused on the comparisons of ApoA-I expression between primary and advanced tumors. Surprisingly, although APOA1 was significantly upregulated in primary tumors, its level was even higher in tissues from metastatic lesions obtained by biopsies according to bioinformatics analyses1718 (). An extremely high level of APOA1 was also observed in several datasets of NEPC,22232425 the terminal stage of PCa, which frequently emerges after long-term hormonal therapy treatment23 (). Concordantly, both castration-resistant PCa (CRPC) and NEPC human tissues exhibited an increased ApoA-I IHC staining intensity (Figure and ). Additionally, we observed a similar trend in ApoA-I protein levels within PCa cell lines representing different stages of PCa. ApoA-I experienced a step-wise elevation across the disease spectrum (from androgen dependent to the CRPC and NEPC stages), as shown by the fact that PC3 and DU145, therapy-resistant cells with NE features, were found to display the highest level of ApoA-I in comparison to other hormone-sensitive PCa cell lines (LNCaP and LAPC4) and CRPC cells (C4-2), as shown in . In GEMMs of PCa, although PTEN deficiency is able to promote tumor formation, its inactivation alone only leads to minimally invasive features after a long latency.2627 The additional loss of the TSG retinoblastoma tumor suppressor gene (RB1) has been reported to drive metastasis.2028 Intriguingly, prostate-specific simultaneous deletion of PTEN and RB1 resulted in an increased level of APOA1, and this level was further elevated in double-knockout metastatic mouse tumors (). These results suggest the pro-oncogenic role of ApoA-I associated with PCa development. Indeed, pair-wise correlation analyses of two well-established datasets revealed that APOA1 negatively correlated with TSGs (PTEN, RB1, and tumor protein P53 [TP53])2028 and the NEPC repressor gene (RE1 silencing transcription factor [REST])29 but positively associated with NE differentiation genes (enhancer of zeste 2 polycomb repressive complex 2 subunit [EZH2], aurora kinase A [AURKA], MYCN proto-oncogene [MYCN], forkhead box A2 [FOXA2], SRY-box transcription factor 2 [SOX2], serine/arginine repetitive matrix 4 [SRRM4], and one cut homeobox 2 [ONECUT2]),2328,[30–33 NEPC hallmark genes (CHGA and synaptophysin [SYP]),22 the CRPC-related gene (ubiquitin conjugating enzyme e2 C [UBE2C]),34 and the epithelial–mesenchymal transition (EMT) marker gene (snail family transcriptional repressor 1 [SNAI1])34 as shown in .

Figure 2

ApoA-I is associated with disease progression. (a) APOA1 mRNA levels in primary versus metastatic PCa from public datasets. (b) Expression levels of APOA1 mRNA in primary-adeno versus NEPC from public datasets. (c) Correlogram heatmap displaying correlations between mRNA expression levels of APOA1 and other canonical genes involved in tumor suppression, NE differentiation, and CRPC-driven and EMT pathways. Two-tailed t-test. *P < 0.05, **P < 0.01, ***P < 0.001. ApoA-I: apolipoprotein A-I; PCa: prostate cancer; NEPC: neuroendocrine prostate cancer; CRPC: castration-resistant prostate cancer; EMT: epithelial-mesenchymal transition; adeno: adenocarcinoma; NE: neuroendocrine; mRNA: messenger RNA. All the full names of the genes in are listed in .

ApoA-I increases the survival, proliferation, invasion, and therapy resistance of PCa cells

Because ApoA-I is highly expressed in the late stage of PCa, we next investigated whether ApoA-I is able to enhance the biological aggressiveness of the cellular behaviors of PCa cells. We employed APOA1 cDNA to ectopically overexpress ApoA-I in hormone-dependent primary LNCaP cells (). We observed a significant increase in cell proliferation and survival, as determined by the MTS and colony formation assays, respectively (Figure ). Conversely, we used a lentivirus-based vector to express either a control shRNA sequence that does not target any known sequence in the mammalian genome (shCont) or independent, nonoverlapping shRNAs targeting ApoA-I (shApoA-I-1 and shApoA-I-2) to inhibit the expression of ApoA-I in CRPC (C4-2) and NEPC cells (PC3), in which ApoA-I is endogenously abundant ( and ). As expected, the knockdown of APOA1 attenuated cell viability as well as colony formation ability in advanced tumor cells (Figure and ). In contrast, knocking down APOA1 in hormone-sensitive LNCaP cells had minimal effect on cell viability, likely due to its low baseline level (). To determine the enhanced invasiveness induced by ApoA-I, we performed Boyden chamber assays and found an increase in the invading cell number after ApoA-I overexpression in LNCaP cells (Figure and ). In contrast, PC3 cells with APOA1 knockdown showed diminished ability to transmigrate through the chamber pores in response to FBS-coated stimulation (Figure and ).

ApoA-I promotes cellular proliferation, clonogenicity, invasiveness, and therapeutic resistance in PCa. (a) Western blot showing ApoA-I expression in LNCaP cells by ectopically overexpressing APOA1 cDNA (left panel). Cell viability was monitored at the indicated time points (right panel). (b) Representative colony formation images of LNCaP cells with ApoA-I overexpression and PC3 cells with APOA1 knockdown. (c) Quantification of colony number from b. (d) Western blot showing the efficiency of shRNA-mediated knockdown of APOA1 in PC3 cells (left panel). Cell viability was compared between PC3 cells in the presence or absence of ApoA-I (right panel). (e) Representative images of the invasion capacity of LNCaP cells with ApoA-I overexpression and PC3 cells with APOA1 knockdown passing through a transwell barrier. Scale bars = 200 μm. (f) Quantification of the relative migrated cell number. (g) Representative images of senescent cells after ADT treatment stained at the indicated time points. Scale bars = 20 μm (left panel). Bar plot showing the percentage of β-gal positively stained cells (right panel). Two-tailed t-test. ***P < 0.001. ApoA-I: apolipoprotein A-I; PCa: prostate cancer; cDNA: complementary DNA; shRNA: short hairpin RNA; ADT: androgen deprivation therapy; LNCaP: lymph node carcinoma of the prostate; Cont: control.

Hormonal therapy-induced senescence is an early response of primary PCa to androgen-deprivation therapy (ADT), which can be measured by SA-β-gal staining.15 Evasion of this treatment-induced senescence has been demonstrated to promote the outgrowth of CRPC.35 We noticed that ApoA-I overexpression reduced the percentage of SA-β-gal-positive senescent cells (), suggesting that ApoA-I leads to resistance to hormonal therapy in tumor cells. Taken together, these results indicate that ApoA-I is a critical factor in accelerating PCa progression that leads to a more proliferative, invasive, and therapy-resistant phenotype.

ApoA-I displays a global effect on lipid metabolism in PCa

Because lipoprotein is a crucial player in coordinating lipid metabolism within organisms, we then wanted to know whether the altered concentrations of ApoA-I, which is an indispensable component of lipoprotein (specifically HDL), might affect reprogramming of lipid metabolism in PCa. Concomitant with the elevation in expression, the ApoA-I-involved lipoprotein metabolism pathway was significantly activated in metastatic PCa and NEPC in comparison to primary adenocarcinoma, as revealed by GSEA (Figure and ). Because HDL mainly comprises ApoA-I, next, we measured the concentration of HDL in cell culture medium and found that medium derived from PC3 and DU145 cells had higher levels of HDL than the medium used to culture LNCaP and LAPC4 cells (), indicating that the increase in ApoA-I in advanced PCa contributes to increased synthesis of additional HDL products.

ApoA-I affects lipid metabolism in PCa. (a) GSEA of lipoprotein metabolism signatures in metastatic versus primary PCa from Yu et al.17 2004. (b) GSEA of lipoprotein metabolism signatures in NEPC versus primary-adeno from Beltran et al.23 2016. (c) Relative HDL concentration in LNCaP, LAPC4, DU145, and PC3 cell culture medium. (d) Pathway enrichment analysis showing the enriched metabolic pathways in LNCaP and PC3 cells. (e) Pathway impact analysis showing the effect of overexpression of ApoA-I in LNCaP cells on global metabolic pathways. (f) Pathway impact analysis showing the effect of loss of ApoA-I in PC3 cells on global metabolic pathways. Two-tailed t-test. ***P < 0.001. ApoA-I: apolipoprotein A-I; PCa: prostate cancer; GSEA: gene set enrichment analysis; NEPC: neuroendocrine prostate cancer; HDL: high-density lipoprotein; adeno: adenocarcinoma; ES: enrichment score; NES: normalized enrichment score; LNCaP: lymph node carcinoma of the prostate; NS: not significant; Cont: control.

Lipid metabolism involves a complex network that includes de novo biosynthesis and oxidative catabolism. Products and intermediates within numerous metabolic pathways contribute to lipogenesis. For instance, glucose-derived pyruvate fluxes into the tricarboxylic acid (TCA) cycle to generate citrate, which is a critical intermediate linking fatty acid synthesis and glycolysis and an important precursor for lipogenesis.8 Alternatively, in cancer cells with malfunctioning mitochondria or that are undergoing hypoxia, citrate is produced from the reductive carboxylation of glutamine-derived a-ketoglutarate.36 Thus, both glucose and glutamine are crucial carbon providers for the anabolism of fatty acids. To determine the potential rewiring of metabolic features during disease progression, we performed metabolite profiling experiments to obtain an overview of the altered metabolic phenotypes. Compared to the primary LNCaP cells, the advanced PC3 cells exhibited the enrichment of pathways more closely associated with lipid-related metabolism processes, such as the Warburg effect (aerobic glycolysis), TCA cycle, fatty acid metabolism, fatty acid biosynthesis, and oxidation of fatty acids (). To further investigate the impact of ApoA-I on metabolic reprogramming, we conducted two more metabolite profiling assays by employing ApoA-I-overexpressing LNCaP and ApoA-I-suppressed PC3 cell models. Both genetic modifications had significant effects on the global metabolic system (Figure and ). Specifically, the most remarkably affected pathways were involved in fatty acid metabolism, such as fatty acid elongation and fatty acid degradation (Figure and ). These metabolic investigations corroborate that ApoA-I has a profound effect on lipid metabolism and results in a shift in PCa to a lipid-producing phenotype, which is a key turning point in disease progression.

ApoA-I is upregulated by MYC in PCa

MYC is a well-established oncogenic protein family that includes c-Myc (encoded by MYC), N-Myc, and L-Myc.37 Among them, c-Myc and N-Myc have been implicated to be pro-tumorigenic and metastasis driven in PCa, and their amplification is seen in up to 40% of tumor foci.23 Interestingly, they are believed to govern different directions of histological differentiation, as c-Myc amplification has been highly detected in CRPC tumors that maintain adenocarcinoma characteristics, while N-Myc appears to be exclusively overexpressed in NEPC.37 Because ApoA-I is correlated with increased disease severity, we hypothesized that MYC might account for the overexpression of ApoA-I in advanced PCa. Indeed, interrogation of various advanced PCa publicly available datasets revealed that APOA1 showed a positive correlation with either MYCN or MYC in a histology-variant setting17192225 (Figure and ). GSEA analysis showed that genes involved in the lipoprotein metabolism pathway were significantly enriched in N-Myc-overexpressing LNCaP cells ().38 Moreover, in in vitro studies, both c-Myc and N-Myc overexpression in LNCaP cells caused upregulation of APOA1 transcript levels (). Conversely, knocking down MYC by two sets of siRNAs significantly reduced APOA1 mRNA levels in PC3 cells (). These results suggest that ApoA-I is regulated by the MYC family (both c-Myc and N-Myc) in PCa.

ApoA-I is regulated by the MYC family. Correlation analyses of APOA1 with either (a) MYCN or (b) MYC from public datasets. (c) GSEA of lipoprotein metabolism signatures in the LNCaP parental cell line and LNCaP with N-Myc overexpression cell line. (d) Overexpression of N-Myc or c-Myc upregulating APOA1 mRNA levels. Left panel showing the efficiency of N-Myc or c-Myc in LNCaP cells determined by western blot. Right panel showing APOA1 transcript levels measured by qPCR. (e) Repression of c-Myc downregulating APOA1 mRNA expression. Left panel showing immunoblot validation of MYC knockdown by two sets of siRNAs. Right panel showing APOA1 transcript levels measured by qPCR. Two-tailed t-test. ***P < 0.001. ApoA-I: apolipoprotein A-I; GSEA: gene set enrichment analysis; qPCR: quantitative polymerase chain reaction; siRNA: small interfering RNA; Cont: control; NES: normalized enrichment score.

DISCUSSION

Proliferating cancer cells comprehensively rewire their metabolism to sustain growth and survival, particularly in harsh conditions, such as hypoxia, nutrition scarcity, and drug treatment.39 Unlike many other solid tumors, PCa is less glycolytic but highly lipogenic.7 In highlighting the importance of lipids to PCa, one established piece of evidence is that obesity significantly increases the mortality of PCa.40 Furthermore, nonmetastatic PCa mouse models elucidate that a high-fat diet is sufficient to trigger metastasis by inducing lipid accumulation.41 In a metabolic molecular setting, it is mainly because PCa is more reliant on OXPHOS, driven by either glucose or glutamine, to generate citrate, which is the precursor for de novo fatty acid biosynthesis.9 Consistently, our findings identify a highly enriched lipid metabolism-related signature in the advanced PCa cell line, which strengthens the accumulating evidence that PCa can be facilitated by reprogrammed lipid homeostasis.

Hypermetabolic rates are usually reflected by increased expression of enzymes. For instance, the most extensively studied lipogenic enzyme, fatty acid synthase (FASN), has been reported to be overexpressed in a variety of cancers, including PCa. It serves as a biomarker of PCa progression, with increased association with a more aggressive phenotype.7 Additionally, carnitine palmitoyltransferase I (CPT1) and alpha-methylacyl-CoA racemase (AMACR), the two enzymes that are responsible for fatty acid β-oxidation, have also been implicated as being elevated in PCa.9 In contrast to the direct functions in lipid synthesis and utilization, ApoA-I and its assembled lipoprotein are more responsible for modulating lipid movement for reasonable distribution. This process is essential because in cancer cells, lipid metabolism is often dysregulated.9 Appropriate rearrangement is required to satisfy increased demand of tumor cells for these building blocks and energy sources, which are critical for the proliferation and evasion of apoptosis. Our analyses from both public datasets and our own experimental results uncover that ApoA-I-mediated lipoprotein metabolism is highly activated in advanced PCa, including the lethal form of PCa, NEPC, and its alteration is able to cause global metabolic reprogramming, particularly lipid metabolism. Our observations fill a conceptual gap in the understanding of metabolic underpinnings of PCa, suggesting that in addition to the well-studied lipogenesis and β-oxidation, ApoA-I regulation of lipid transporters enables metabolic flux to be better arranged to desired sites, which further gives rise to the increased lipogenic characteristic of PCa. Future efforts will be aimed at more specific details about how ApoA-I coordinates lipids in cancer.

Our current work reveals the unexpected step-wise elevation of ApoA-I expression across the spectrum of PCa progression from multiple perspectives. This finding is somewhat contradictory with the canonical role of ApoA-I as a protector in cancer prevention, which indicates that ApoA-I might be a versatile player in a tissue-dependent manner. In the cardiovascular and gastrointestinal systems, ApoA-I and HDL decrease lipid accumulation to reduce the risks of diseases such as atherosclerosis and even prevent normal tissue from malignancy.10 However, in other organs, especially those who are hormone-dependent, such as the breast, ApoA-I-related lipids become crucial energy suppliers for tumor-promoting processes.42 The high expression of ApoA-I in PCa and the positive association between ApoA-I and disease progression support ApoA-I as an ideal diagnostic and prognostic indicator for early diagnosis (Figures and ). Consistently, its resultant product, HDL, has also been epidemiologically and experimentally reported to be correlated with increased risks of both overall and high-grade PCa by many other publications.434445464748 This positive linkage is found more dramatic in the disease-recurred population.45 This is likely because ADT leads to elevated HDL levels to beneficially feed androgen-independent PCa cells, such as DU145 and PC3.4748 Interestingly, HDL has few impacts on androgen-sensitive LNCaP cells,46 which together with our current findings, further highlights that HDL and ApoA-I are much more important in advanced PCa than primary tumors.

The oncogenic role of ApoA-I in PCa, which regulates cellular behavior and mediates drug resistance, also suggests the potential of ApoA-I to be a putative therapeutic target, compensating for the limitations of the currently used anti-AR therapies. However, given the physiologically wide distribution of ApoA-I within organs, avoiding severe side effects is a major challenge. Drugs with super affinity and organ specificity are required. Future studies to employ GEMM with prostate-specific deletion of the APOA1 gene might address this concern.

Although metabolic rewiring has long been recognized in cancer, more recently, altered molecular drivers are also known to affect cancer metabolism, and the discovery of cross-talk between metabolism and cancer genetics as well as epigenetics broadens the contributions to carcinogenesis and tumor maintenance.49 We and others have found that MYC, which is frequently amplified in advanced PCa, displays a striking impact on cancer metabolism.1650 MYC has been implicated in promoting glycolysis by upregulating lactate dehydrogenase A activity.16 In glutaminolysis, MYC increases glutaminase to fuel the TCA cycle through a posttranscriptional repression mechanism.50 In the current study, our data reveal the additional regulatory effect of MYC on lipid metabolism by increasing ApoA-I expression. Future molecular studies will be aimed at obtaining insights into specific aspects of mechanistic regulation.

CONCLUSION

In summary, our current study uncovers the novel role of ApoA-I in PCa. Overexpression of ApoA-I allows PCa to initiate, survive, and proliferate after hormonal therapies. Incorporating ApoA-I into clinical practice might aid risk stratification of PCa cases that are likely to recur and progress.

AUTHOR CONTRIBUTIONS

JW, LFX, CZL, and YDF conceived the study and participated in its design. LFX, JW, and CL participated in data acquisition. LFX, JW, and TH participated in the data analysis and interpretation. JW drafted the original manuscript. LFX reviewed and helped to edit the manuscript. CZL and YDF revised and supervised the manuscript. All authors read and approved the final manuscript.

COMPETING INTERESTS

All authors declared no competing interests.

Loss of ApoA-I shows different effects between advanced and primary tumor cells. Western blot showing the efficiency of shRNA-mediated knockdown of ApoA-I in (a) C4-2 and (b) LNCaP cells. Cell viability was compared between the indicated cells with the presence or absence of ApoA-I in (a) C4-2 and (b) LNCaP cells. ApoA-I: apolipoprotein A-I; n.s: not significant; LNCaP: lymph node carcinoma of the prostate; shRNA: short hairpin RNA; Cont: control.

Supplementary Table 1

All the full names of the genes in Figure 2c

Gene symbol	Official full name
APOA1	Apolipoprotein A1
RB1	RB Transcriptional Corepressor 1
TP53	Tumor Protein P53
PTEN	Phosphatase and Tensin Homolog
FOXA1	Forkhead Box A1
REST	RE1 Silencing Transcription Factor
EZH2	Enhancer of Zeste 2 Polycomb Repressive Complex 2 Subunit
AURKA	Aurora Kinase A
MYCN	MYCN Proto-Oncogene
FOXA2	Forkhead Box A2
SOX2	SRY-Box Transcription Factor 2
SRRM4	Serine/Arginine Repetitive Matrix 4
CHGA	Chromogranin A
ENO2	Enolase 2
SYP	Synaptophysin
CXCR2	C-X-C Motif Chemokine Receptor 2
ONECUT2	One Cut Homeobox 2
CD44	CD44 Molecule
PROM1	Prominin 1
POU5F1	POU Class 5 Homeobox 1
SOX9	SRY-Box Transcription Factor 9
KLF4	Kruppel-Like Factor 4
IGF1R	Insulin-Like Growth Factor 1 Receptor
REG4	Regenerating Family Member 4
UBE2C	Ubiquitin Conjugating Enzyme E2 C
BCL2	BCL2 Apoptosis Regulator
E2F3	E2F Transcription Factor 3
SNAI1	Snail Family Transcriptional Repressor 1