Research ArticleProstate Cancer

Suppressing fatty acid uptake has therapeutic effects in preclinical models of prostate cancer

See allHide authors and affiliations

Science Translational Medicine  06 Feb 2019:
Vol. 11, Issue 478, eaau5758
DOI: 10.1126/scitranslmed.aau5758

Leave the fat out

Prostate cancer is one of the most common tumors in men. Although characterized by slow growth rate, preventing prostate cancer progression to an aggressive stage is a major challenge. Watt et al. focused on cancer metabolism and showed increased fatty acid uptake in human malignant prostate cancer tissue. The increased uptake was mediated by up-regulation of the fatty acid translocase CD36. Silencing CD36 in human prostate cancer cells reduced fatty acid uptake and cell proliferation. In prostate cancer mouse and human preclinical models, Cd36 ablation or inhibition reduced prostate cancer severity. The data suggest that CD36 might be targeted for treating prostate cancer.


Metabolism alterations are hallmarks of cancer, but the involvement of lipid metabolism in disease progression is unclear. We investigated the role of lipid metabolism in prostate cancer using tissue from patients with prostate cancer and patient-derived xenograft mouse models. We showed that fatty acid uptake was increased in human prostate cancer and that these fatty acids were directed toward biomass production. These changes were mediated, at least partly, by the fatty acid transporter CD36, which was associated with aggressive disease. Deleting Cd36 in the prostate of cancer-susceptible Pten−/− mice reduced fatty acid uptake and the abundance of oncogenic signaling lipids and slowed cancer progression. Moreover, CD36 antibody therapy reduced cancer severity in patient-derived xenografts. We further demonstrated cross-talk between fatty acid uptake and de novo lipogenesis and found that dual targeting of these pathways more potently inhibited proliferation of human cancer-derived organoids compared to the single treatments. These findings identify a critical role for CD36-mediated fatty acid uptake in prostate cancer and suggest that targeting fatty acid uptake might be an effective strategy for treating prostate cancer.


Alterations of metabolic activities have been shown to support the malignant properties of cancer cells (1, 2). Recent studies highlighted the relevance of glucose, glutamate, and fatty acids derived from de novo lipogenesis in modulating the bioenergetic processes and macromolecule synthesis required to sustain growth and proliferation (1, 3). Fatty acids are also derived from adipose tissue lipolysis or the breakdown of triglycerides contained in circulating chylomicrons and lipoproteins. In most nontumorigenic cells, these exogenous fatty acids are a preferred source for adenosine 5´-triphosphate production, membrane biosynthesis, energy storage, and the generation of a wide array of signaling molecules (4). However, the role of exogenous fatty acids and their transporters in tumor biology has received relatively little attention.

Prostate cancer is the second most commonly diagnosed cancer in men, representing 15% of male cancer diagnoses and 8% of all cancer cases (5). Although it is a common cancer, it is a relatively slow-growing malignancy, and many men have indolent or low-risk disease that takes decades to elicit clinical symptoms and is readily treated by active surveillance or curative intent therapy (6). The major clinical challenge is to prevent progression to aggressive disease in men with moderate- or high-risk prostate cancer.

The slow disease progression and ineffective identification of prostate cancer by 18F-deoxyglucose positron emission tomography (7) indicate that prostate cancer may not be subjected to the typical metabolic reprogramming observed in rapidly proliferating tumors, where glucose is considered the dominant metabolic substrate. Heightened fatty acid production from de novo lipogenesis is reported in many cancers (3), including prostate cancer (8), and pharmacological blockade of this process limits tumor growth (9, 10). Mounting evidence also implicates adipose-derived fatty acids in tumor malignancy, as exemplified in ovarian and breast cancer, where adipocytes reside in close proximity to the tumor foci and their secreted products affect disease progression (1113). In this regard, the prostate is covered by the prominent periprostatic adipose tissue, which has the capacity to supply substantial quantities of fatty acids to alter the prostate tumor microenvironment (14, 15). Moreover, fatty acids are the dominant metabolic substrate in immortalized prostate cancer cells (16, 17), and prostate cancer cell viability is reduced by blocking fatty acid oxidation via pharmacological and genetic inhibition of CTP1, the rate-limiting enzyme for mitochondrial fatty acid transfer (18). Preclinical and epidemiological studies also indicate that higher levels of dietary saturated fat increase the risk of all-cause mortality in men with localized prostate cancer (1921). In linking these observations, the increased expression of fatty acid translocase (FAT)/CD36, a major transporter for exogenous fatty acids into cells (22), correlates with a poor prognosis in lung squamous cell, glioblastoma, bladder, and luminal A breast carcinomas, whereas inhibition of CD36 impairs epithelial-to-mesenchymal transition and metastases in human melanoma and breast cancer–derived tumors (2325).

In this study, we established CD36-mediated fatty acid uptake as a critical process for the production of lipid biomass and the generation of oncogenic signaling lipids in prostate cancer. Furthermore, we showed that CD36 monoclonal antibody (mAb) therapy reduced prostate cancer growth in patient-derived xenografts (PDXs) of high-risk localized disease and that the efficacy of CD36 blockade could be enhanced by combined inhibition of de novo lipogenesis. These data suggest that blocking fatty acid uptake might be a promising therapeutic approach for treating prostate cancer.


Fatty acid uptake and biomass storage are increased in malignant human prostate tissue

We initially sought to identify the key metabolic changes in prostate cancer in humans. Paired benign and malignant prostate tissue was obtained from patients with prostate cancer with low to intermediate risk of progression (table S1) and was used for the assessment of metabolism using radiolabeled glucose and fatty acids ex vivo. Glucose uptake was increased in malignant compared with benign prostate tissue (Fig. 1A); however, glucose oxidation was not increased (Fig. 1B). De novo lipogenesis, which is elevated in most cancers to provide fatty acids for membrane production and is important for prostate cancer progression (26), was increased in malignant tissue (Fig. 1C). Fatty acid uptake was increased in tumors compared with benign tissue in 10 of the 11 patients (Fig. 1, D and E) and was not accompanied by an increase in fatty acid oxidation (Fig. 1F). Rather, the fatty acids were stored within complex lipids including ceramides, phospholipids, diacylglycerol, and triacylglycerol, as shown by their increase in malignant compared to benign tissue (Fig. 1G). There was no increase in cholesterol ester production, indicating specificity in lipid storage rather than generalized bulk storage as a result of increased fatty acid flux. Thus, marked remodeling of substrate metabolism occurs in malignant human prostate tissue, highlighted by increased fatty acid uptake and storage.

Fig. 1 Fatty acid uptake and storage are increased in malignant compared with benign human prostate tissue.

(A) Glucose uptake (n = 8), (B) glucose oxidation (n = 8), (C) de novo lipogenesis (n = 12), (D) fatty acid uptake (n = 11), (E) the percent difference in fatty acid uptake between patient-matched malignant (mal) and benign tissue, (F) fatty acid oxidation (n = 14), and (G) fatty acid incorporation into various lipids (n = 6 to 8 per lipid type). Cer, ceramide; PL, phospholipid; CE, cholesterol ester; DAG, diacylglycerol; TAG, triacylglycerol. *P < 0.05 malignant versus benign by two-tailed paired t test. For (A) to (G), data are presented as means ± SEM. (H) Kaplan-Meier curve showing estimated biochemical recurrence (BCR)–free probability after radical prostatectomy in patients with high (upper quartile) or low levels of tumor CD36 mRNA in the TCGA (provisional) cohort (27). This cohort contains genomic and transcriptomic data from 491 tumor and 40 normal human prostate samples. P = 0.022 and hazard ratio (HR) of 1.8 were determined using a log-rank test. (I) Histogram showing the alteration frequency of CD36 in seven distinct prostate cancer cohorts.

Fatty acid uptake is regulated by protein-mediated transport (22), and we reasoned that if free fatty acids (FFAs) are important in prostate cancer aggressiveness, then fatty acid transporter expression might be related to clinicopathological parameters of prostate cancer. We used The Cancer Genome Atlas (TCGA) prostate cancer cohort (27) to interrogate known fatty acid transporters including CD36, which encodes a FAT, and SLC27A genes, which encode the fatty acid transport proteins 1 to 6. Kaplan-Meier survival analysis showed that high expression of CD36 was associated with reduced relapse-free survival (Fig. 1H and fig. S1, A and B). In contrast, none of the SLC27A genes were associated with relapse-free survival (fig. S1, C to H). Further analysis of additional prostate cancer genomic datasets revealed that the CD36 gene was frequently gained or amplified but rarely lost in prostate cancer (Fig. 1I) and that such copy number changes occurred at an increased incidence in metastases compared to primary tumors (fig. S1I). Supporting this finding, copy number gain was associated with increased CD36 mRNA expression (fig. S1J), and CD36 was expressed at a higher level in metastatic disease (fig. S1K). These observations aligned with a recent pan-cancer analysis showing that CD36 is amplified at a higher frequency in metastatic versus primary tumors (28) and another demonstrating that CD36 is a metastasis-promoting factor in oral carcinoma (23). Last, we showed that CD36 protein was expressed in human prostate tissues in benign and tumor regions, as well as in both stromal and epithelial cells (fig. S1L), although there was no evidence for increased expression in tumor regions.

Silencing of CD36 reduces fatty acid uptake and attenuates cancer aggressiveness in prostate cancer cells

The increase in fatty acid uptake in malignant tissue prompted us to assess the effects of fatty acids on prostate cancer progression. Fatty acid transport and storage into PC3 and LNCaP prostate cancer cells have been shown to be increased after exposure to a physiological mixture of FFAs (29). Here, we showed that fatty acid exposure was associated with increased cell proliferation in PC3 and LNCaP prostate cancer cell lines but not in benign prostate BPH-1 cells (fig. S2A). Furthermore, fatty acid uptake was increased in malignant (PC3 and LNCaP) compared with nonmalignant (BPH-1) cells (fig. S2B).

To test whether CD36 knockdown would affect fatty acid metabolism and reduce tumorigenicity, we transfected PC3 cells with control short hairpin RNA (shRNA) designed to silence CD36 gene expression. The control shRNA did not affect CD36 gene expression or lipid metabolism compared with the parental PC3 cells (Fig. 2, A to D). In contrast, CD36 shRNA resulted in almost complete ablation of CD36 gene expression (Fig. 2A), a 35% reduction in FFA uptake (Fig. 2B), reduced FFA oxidation (Fig. 2C), and reduced incorporation of fatty acids into complex lipids (Fig. 2D) compared to parental PC3 cells. Exposing parental PC3 cells to fatty acids increased cell proliferation by 44%, and silencing of CD36 completely attenuated these proliferative effects (Fig. 2E). In addition, CD36 shRNA reduced migration efficiency (Fig. 2, F and G) and decreased the number of PC3 migrating cells (Fig. 2H) in the presence of FFAs. Reexpression of CD36 into CD36 shRNA cells restored fatty acid uptake and proliferation rates to levels measured in parental PC3 cells (fig. S3, A to C). Similar to androgen receptor (AR)–negative PC3 cells, CD36 silencing reduced fatty acid uptake and proliferation in the AR-positive LNCaP prostate cancer cells (fig. S3, D to H).

Fig. 2 shRNA-mediated silencing of CD36 in prostate cancer cells reduces fatty acid uptake and impairs cancer aggressiveness.

(A) CD36 mRNA expression in PC3 parental cells, cells transfected with a shRNA with a scrambled sequence (shRNA neg), and two independent PC3 cell lines stably transfected with shCD36. n = 4 per group. *P < 0.05 versus PC3 by one-way analysis of variance (ANOVA) and Bonferroni multiple comparisons test. (B to D) Fatty acid metabolism was assessed in cells treated with 500 μM oleate (1 mCi/ml of 1-14C-oleate) for 2 hours for (B) fatty acid uptake, (C) fatty acid oxidation, and (D) fatty acid incorporation into cellular lipids. Data are presented as means ± SEM, n = 10 to 12 per group from four independent experiments. *P <0.05 versus PC3 by one-way ANOVA and Bonferroni multiple comparisons test. (E) In vitro proliferation of PC3 cells without or with CD36 knockdown using shRNA. Data from PC3 parental and shRNA neg are reported together as PC3. Cells were treated without (−FA) or with (+FA) fatty acids (oleate:palmitate, 1:1 conjugated to 1% bovine serum albumin) for 24 hours. Data represent means ± SEM, n = 3 per condition from one of three independent experiments. *P < 0.05 by two-way ANOVA and Bonferroni multiple comparisons test. (F and G) Migration efficiency calculated by recovery of the cell surface after manually scratching the cell surface. Data are presented as means ± SEM, n = 4 to 6 per group from three independent experiments. *P < 0.05 by two-way ANOVA and Bonferroni multiple comparisons test. (F) Representative images of cells at start and end point. Scale bars, 50 μm. (H) Cell motility assessed by Transwell cell migration assays calculated as the number of cells migrated through the Transwell filter relative to control after 18 hours. Data are presented as means ± SEM, n = 3 to 7 per group from three independent experiments. *P < 0.05 by two-way ANOVA and Bonferroni multiple comparisons test. (I) In vivo growth of subcutaneous tumors after injection of shCD36 PC3 cells compared and shRNA neg PC3 control cells in SCID mice (subcutaneous implantation of 1.8 × 106 cells in collagen). Tumor volume was measured at the indicated time points. Data are presented as means ± SEM; n = 6 for shRNA neg, n = 10 for shCD36 from independent clones #1 and #3. Data assessed by two-way ANOVA. Ptime×genotype = 0.07, Ptime < 0.0001, Pgenotype = 0.005.

To determine whether these findings were relevant in vivo, we injected control and CD36 shRNA PC3 cells subcutaneously in immune-deficient severe combined immunodeficient (SCID) mice. Tumor volume increased sixfold between days 5 and 30 in the control shRNA grafts and was reduced (32%) in the CD36 shRNA grafts (P < 0.005; Fig. 2I). Together, these proof-of-concept experiments show that reducing fatty acid uptake by silencing CD36 slows proliferation of prostate cancer cell lines.

Genetic ablation of Cd36 in prostate epithelial cells alters fatty acid metabolism in mice prone to prostate cancer

We developed a novel genetic approach to determining the role of Cd36 deletion in mice with susceptibility to prostate cancer. We first established a mouse model of prostate cancer by deleting the tumor suppressor Pten in prostatic epithelial cells using the Probasin (PBi)-Cre promoter (Ptenfl/fl.PB-Cre) (30). These mice were crossed with Cd36 fl/fl mice to generate mice with (Pten−/−) or without (Pten−/−.Cd36−/−) Cd36 in prostatic epithelial cells. Cd36 protein was expressed in prostate epithelial cells of wild-type (WT) and Pten−/− mice and reduced in Pten−/−.Cd36−/− mice (Fig. 3A).

Fig. 3 Genetic ablation of Cd36 alters fatty acid metabolism in the prostate tissue of Pten-deficient mice.

(A) Immunohistochemical staining of Cd36 in lateral prostate lobes of WT, Pten−/−, and Pten−/−.Cd36−/− mice (isotype-matched negative control in bottom right on Pten−/−.Cd36−/− mice). Images are representative of n = 4 animals per genotype. Scale bars, 50 μm. (B to E) Fatty acid metabolism was assessed in the anterior prostate of mice using 500 μM oleate (1 mCi/ml of 1-14C-oleate) for 2 hours. (B) Fatty acid uptake, (C) fatty acid oxidation, (D) fatty acid storage into various lipids, and (E) the fatty acid storage to oxidation ratio. Data are presented as means ± SEM; n = 11 for WT, n = 4 for Pten−/−, n = 15 for Pten−/−.Cd36−/− mice. (F) De novo lipogenesis was assessed in the anterior prostate lobe using 5 mM glucose (3 μCi/ml of [14C(U)]-glucose). Data are presented as means ± SEM; n = 9 for WT, n = 7 for Pten−/, n = 5 for Pten−/−.Cd36−/− mice. For all figures, horizontal lines denote P < 0.05 between adjoining bars. Data are assessed by one-way ANOVA and Bonferroni multiple comparisons tests.

Metabolic assessment in prostate tissue demonstrated marked increases (55%) in fatty acid uptake in Pten−/− compared to WT mice (Fig. 3B), which was not accompanied by changes in fatty acid oxidation (Fig. 3C) but rather by marked increases in fatty acid storage into a variety of lipid pools (Fig. 3D). This distinct pattern of lipid storage in Pten−/− mice mimicked that observed in malignant compared with benign human prostatic tissues. Fatty acid metabolism was modulated in Pten−/−.Cd36 −/− mice highlighted by complete attenuation of fatty acid uptake induced by Pten deletion (Fig. 3B) and restored lipid storage rates to those observed in WT mice (Fig. 3D). This metabolic remodeling was illustrated by restoration in the fatty acid storage-to-oxidation ratio in Pten−/−.Cd36 −/− mice (Fig. 3E). In light of the reduction in fatty acid uptake in Pten−/−.Cd36−/− mice, we hypothesized that de novo lipogenesis might be increased to produce sufficient fatty acid substrate for membrane production. De novo lipogenesis was increased in the prostates of Pten−/−.Cd36−/− mice compared with both WT and Pten−/− mice (Fig. 3F).

Cd36 inactivation reduces the primary tumor burden in prostate-specific Pten−/− mice

Individual prostate lobes were dissected from WT, Pten−/−, and Pten−/−.Cd36−/− mice at 8, 12, and 24 weeks of age. The mass of the prostatic lobes was not generally different between genotypes at 8 weeks (Fig. 4A). By 12 weeks, the mass of all prostatic lobes was increased in Pten−/− compared to WT mice (Fig. 4A), and by 24 weeks, the prostatic lobes weighed on average 400% more in Pten−/− compared to WT mice. The Pten−/−-mediated increase in tissue mass was abrogated in Pten−/−.Cd36 −/− mice, reducing the dorsal, lateral, and ventral prostate masses by 24, 33, and 34% by 12 weeks of age (Fig. 4A). By 24 weeks of age, there were notable reductions in tissue mass in all prostate lobes of Pten−/−.Cd36 −/− compared with Pten−/− mice (Fig. 4A).

Fig. 4 Cd36 inactivation slows prostate cancer progression and reduces the primary prostate tumor burden in prostate-specific Pten−/−mice.

(A) Prostate tissue mass of lateral, anterior, ventral, and dorsal lobes from WT, Pten−/−, and Pten−/−.Cd36−/− mice fed a standard chow diet for 8, 12, and 24 weeks. Data are presented as means ± SEM and were analyzed using two-way ANOVA and Bonferroni post hoc test. For each genotype, the n is reported for 8, 12, and 24 weeks, respectively, as follows: n = 8, 20, and 9 for WT, n = 10, 10, and 9 for Pten−/−, n = 8, 17, and 9 for Pten−/−.Cd36−/−. P < 0.05 denoted by “8” for WT versus Pten−/−, “#” for WT versus Pten−/−.Cd36−/−, and “†” for Pten−/− versus Pten−/−.Cd36−/−. Flow cytometric quantitation of (B) total prostate cellularity, (C) prostate epithelial cells, (D) epithelial cell size [mean forward scatter (FSC)], and (E) percentage of epithelial cell proliferation (ki67+) as quantitated by flow cytometric analysis of whole prostate glands from WT, Pten−/−, and Pten−/−.Cd36−/− mice aged 12 to 17 weeks. Results are presented as means ± SEM for all datasets and assessed by one-way ANOVA and Bonferroni multiple comparisons tests. Horizontal lines denote P < 0.05 between adjoining bars; n = 5 for WT, n = 6 to 8 for Pten−/−, n = 4 for Pten−/−.Cd36−/−. (F) Hematoxylin and eosin staining of lateral prostate lobes from mice aged 12 weeks. Pten−/− prostates contain CIS, which is evident by glandular ducts filled with noninvasive malignant cells (black arrow) contained within the basement membrane (white arrow), while Pten−/−.Cd36−/− prostates predominantly show HG-PIN (less CIS), containing cytological atypical cells, but retain the ductal lumen (“*”). Scale bars, 100 μm. (G) Pathology was scored as normal, HG-PIN, or CIS. The proportion of each pathology is shown for 8 weeks (n = 8 for WT, n = 10 for Pten−/−, n = 8 for Pten−/−.Cd36−/− mice), 12 weeks (n = 20 for WT, n = 10 for Pten−/−, n = 17 for Pten−/−.Cd36−/− mice), or 24 weeks (n = 9 for WT, n = 9 for Pten−/−, n = 9 for Pten−/−.Cd36−/− mice). Results are presented as means ± SEM for all datasets. Data assessed by one-way ANOVA and Bonferroni multiple comparisons tests. “*” denotes P < 0.05 between Pten−/− and Pten−/−.Cd36−/− mice.

Flow cytometric analysis was used to characterize the cellular composition of WT, Pten−/−, and Pten−/−.Cd36−/− prostates, including epithelial (basal and luminal) and stromal (fibroblasts, endothelial, and white blood cells) subsets (fig. S4A). Total cellularity was increased in Pten−/− compared to WT mice (Fig. 4B) and was evident in all cell subsets examined including epithelial (Fig. 4C), stromal, basal, luminal, endothelial, and white blood cells (fig. S4, B to F). In all cases, there were significant reductions in cell numbers in the prostates of Pten−/−.Cd36−/− compared with Pten−/− mice (P < 0.05; Fig. 4, B and C, and fig. S4, B to F). There was no change in epithelial cell size in Pten−/− or Pten−/−.Cd36−/− compared to WT mice, as determined by forward scatter properties (Fig. 4D). Epithelial cell proliferation was increased in Pten−/− compared to WT mice, and this proliferative effect of Pten−/− ablation was completely abrogated in Pten−/−.Cd36−/− mice (Fig. 4E). Together, these data suggest that the reduction in Pten−/−.Cd36−/− mouse prostate size is likely to be due to decreased subset cellularity and reduced epithelial cell proliferation.

Consistent with the reported phenotype in Pten−/− mice (3033), disease progression to prostate cancer involved high-grade prostatic intraepithelial neoplasia (HG-PIN) characterized by multifocal luminal hyperplasia and increased numbers of atypical cells, leading to a stratified epithelial layer within preexisting prostatic ducts. Incidence of prostate cancer was dominated by carcinoma in situ (CIS) lesions, characterized by prostatic ducts filled with hyperplastic luminal cells and evidence of cytoplasmic vacuolization (Fig. 4F) (34). At 8 weeks, Pten−/− mice showed predominately HG-PIN, with some evidence (8%) of CIS (Fig. 4G). At 12 and 24 weeks, the percentage of CIS in Pten−/− mice was increased to 24 and 43%, respectively (Fig. 4G). At the oldest age examined (24 weeks), Pten−/− mice showed no evidence of malignant cells invading the basement membrane in regions of CIS (Fig. 4F), and consistent with updated observations in Pten-null prostate tumors (35, 36), there was no evidence of metastasis at 24 weeks. WT mice (Fig. 4G) and WT mice with Cd36 deletion in the prostate epithelial cells (WT.Cd36−/− mice) had normal tissue mass, benign prostate pathology, and prostate lipid metabolism (fig. S5).

Pten−/−.Cd36−/− mice showed a similar pathology as Pten−/− mice at 8 weeks (Fig. 4G). At 12 and 24 weeks, the percentage of CIS in Pten−/−.Cd36−/− mice was reduced to 15 and 27%, respectively, representing a halving of the amount of carcinoma compared to Pten−/− mice (Fig. 4G). Together, the large differences in tissue weight and reduction in the CIS prevalence showed that disease progression was slower and the overall tumor burden in Pten−/−.Cd36−/− mice was reduced.

Cd36 deletion reduces prostate cancer severity in the setting of lipid oversupply

Obesity (37) and high dietary saturated fat intake (19, 20) are associated with poor prostate cancer prognosis. Accordingly, we assessed the impact of Cd36 deletion on prostate cancer in mice fed a high-fat diet. Fatty acid uptake into the prostate was increased by 62% in Pten−/− versus WT mice, and this increase was completely attenuated in Pten−/−.Cd36−/− mice (fig. S6A). There was no change in de novo lipogenesis between Pten−/− and Pten−/−.Cd36−/− compared to WT mice (fig. S6B). High-fat feeding increased the lateral and dorsal prostate weights by 22%, independent of genotype, whereas there was no effect of diet on ventral and anterior prostate weight (fig. S6C). Although there were no significant changes in tissue mass between Pten−/−.Cd36−/− and Pten−/− mice (fig. S6C), the prevalence of CIS was reduced by 62% in Pten−/−.Cd36−/− mice (fig. S6D).

Pten deficiency induces marked alterations in the prostate lipidome of mice, and this is partially restored with Cd36 deletion

Targeted lipidomic analysis was used to better understand how lipid metabolism is remodeled in Pten−/−-mediated prostate cancer and how Cd36 deletion could affect this lipidomic profile (Fig. 5A). Pten deletion resulted in remodeling of the lipidome, including broad reductions in membrane phospholipids, ether lipids, and diacylglycerol neutral lipid species and increases in the fatty acid oxidation metabolites acyl carnitines (ACs), monoacylglycerols (MAGs), and phospholipid hydrolysis products acyl and ether lysophospholipids (Fig. 5B). Together, these data suggest that Pten deletion may lead to the release of fatty acids and lysophospholipids from membrane phospholipids and ether lipids to be used for fatty acid oxidation and signaling, respectively.

Fig. 5 Cd36 inactivation suppresses oncogenic lipid production of murine prostate cancer.

Lipidomic analysis of prostate tissue in WT, Pten−/−, and Pten−/−.Cd36−/− mice aged 12 weeks. (A) Heat map showing the changes of the prostate lipidome in Pten−/− and in Pten−/−.Cd36−/− mice (bottom). Lipid class and individual lipid species are denoted at the top of the figure. (B) Representative lipids altered in Pten−/− compared with WT mice. *P < 0.05 by unpaired t test. (C) Lipids detected in Pten−/− and Pten−/−.Cd36−/− mice. Data are assessed by one-way ANOVA and Bonferroni multiple comparisons tests, and horizontal lines denote P < 0.05 between adjoining bars. n = 7 for WT, n = 6 for Pten−/−, n = 5 for Pten−/−.Cd36−/−.

To better understand the mechanism through which Cd36 deletion impaired tumorigenesis in the Pten deletion background, we next analyzed the lipidomic data for lipid species that were dysregulated in Pten-deleted tumors but rescued back to WT levels in the Pten−/−.Cd36−/− tumors (Fig. 5C). Most lipid species were not altered by Cd36 deletion, suggesting that CD36-mediated fatty acid transport controls a specific subset of lipids in tumors driven by the loss of Pten. ACs, MAGs, and several acyl and ether lysophospholipids that were elevated by Pten deletion were reversed upon Cd36 deletion (Fig. 5C). These lipidomic data suggest that Cd36 deletion may reduce cancer cell proliferation by impairing fatty acid oxidation and lysophospholipid-mediated tumor-promoting signaling pathways (fig. S7).

CD36 has been linked with phospholipid remodeling in other cell types including cardiomyocytes (38) and macrophages (39). Examination of the TCGA prostate cancer dataset showed that CD36 expression is correlated with PLA2G4A (fig. S8A), which encodes cytosolic phospholipase A2α (cPLA2α) but not with other phospholipases. Similar to CD36, PLA2G4A was frequently gained or amplified in prostate cancer (fig. S8B). On the basis of these observations in which the lipidomic data show that many phospholipid species are decreased and lysophospholipid species are increased in Pten−/− mice and that this is reversed with Cd36 deletion, we speculated that cPLA2α could be a mediator of Cd36’s effect on lysophospholipid metabolism. In support of this hypothesis, activation of cPLA2α, as measured by its localization to membranes, was increased in the prostates of Pten−/− mice when compared with WT mice, and this increase was abolished in Pten−/−.Cd36−/− prostate tissue (fig. S8, C and D). Moreover, shRNA-mediated knockdown of PLAG2G4A in PC3 cells reduced proliferation by 27% (fig. S8, E and F). Although they do not demonstrate causality, these data suggest that cPLA2α inhibition is linked to the effects of Cd36 deletion on prostate cancer cells.

CD36 mAb therapy reduces cancer severity in human prostate tissue

To assess the therapeutic potential of blocking CD36 in human prostate in vivo, we administered mAbs (20 μg) against CD36 intraperitoneally to non-obese diabetic scid gamma (NSG) mice harboring PDXs of localized high-risk prostate cancer. Two independent PDX lines were selected on the basis of their similarities including Gleason 4+5 and pT3b pathology, hormone dependency and intact AR signaling [AR-, prostate-specific antigen (PSA)–, and prostate-specific membrane antigen (PSMA)–positive], and CD36 expression (Fig. 6A), but with differing capacities for fatty acid uptake (Fig. 6B; low versus high). The PDX line characterized by high fatty acid uptake rates (PDXhi) stored more fatty acid in cellular lipids and exhibited higher rates of de novo lipogenesis and similar rates of fatty acid oxidation compared to the other PDX line with low fatty acid uptake rates (PDXlo; Fig. 6, C to E). Prostate tumor xenografts were larger in PDXhi compared with PDXlo mice [PDXhi (309 ± 42 mg) versus PDXlo (46 ± 9 mg); Fig. 6F], indicating more aggressive cancer. CD36 mAb treatment did not affect prostate cancer growth in PDXlo tumors, whereas there was a reduction in tumor volume in PDXhi tumors (Fig. 6F). This was associated with a 22% reduction in fatty acid uptake in the PDX graft (Fig. 6G) but occurred independently of changes in body mass or blood lipids (fig. S9, A to C), indicating the absence of pronounced systemic effects of the mAb therapy. Last, we compared the antitumorigenic effects of CD36 mAb therapy with docetaxel, a frontline prostate cancer therapy. Tumor volume was reduced equally with both treatments in the PDXhi grafts (Fig. 6H), providing evidence that systemic blocking of CD36 can be exploited for therapeutic gain in human prostate cancer.

Fig. 6 CD36 mAb therapy reduces tumor growth in prostate cancer PDXs.

PDXs were established in NSG male mice from localized hormone-sensitive prostate cancer tissues. Two independent PDX lines were used (PDXlo and PDXhi), which indicate their propensity for fatty acid uptake. (A) Pathology and histological assessment including hematoxylin and eosin (H&E), and expression of AR, PSA, and PSMA. Scale bars, 100 μm. Metabolic assessment of PDXlo and PDXhi showing (B) fatty acid uptake, (C) fatty acid oxidation, (D) fatty acid storage into lipids, and (E) de novo lipogenesis, which was assessed in the PDX grafts using 500 μM oleate (1 mCi/ml of [1-14C]-oleate) for 2 hours. Data are presented as means ± SEM, n = 8 for PDXlo and n = 3 for PDXhi. For therapeutic studies, pair-matched precision slices of PDX were established in IgA control or CD36 mAb–treated mice. (F) PDX graft volume from individual grafts at harvest in PDXlo and PDXhi lines. Paired samples are joined by dashed lines. n = 15 control and treatment for PDXlo and n = 7 for control and treatment for PDXhi. Data assessed by paired Student’s t test, *P < 0.05, main effect of treatment by two-way ANOVA. n = 8 to 12 per group. (G) Metabolic assessment of PDXhi grafts from vehicle and CD36 mAb–treated mice showing fatty acid uptake. Data are presented as means ± SEM, n = 3 PDX grafts per group. Data assessed by paired Student’s t test, *P < 0.05. (H) PDX graft volume from individual grafts at harvest in PDXhi lines after CD36 mAB treatment compared to docetaxel treatment for 3 weeks. Paired samples are joined by lines. Data assessed by paired Student’s t test, *P < 0.05, main effect of treatment by two-way ANOVA. n = 8 to 12 per group.

Inhibition of de novo lipogenesis increases the efficacy of CD36 blockade

Prostate cancer is characterized by increased de novo lipogenesis, and Cd36 deletion exacerbated this effect in cancer-prone Pten−/− mice, raising the intriguing possibility that dual targeting of fatty acid uptake and de novo lipogenesis would increase the efficacy of CD36 blockade in prostate cancer. We tested this hypothesis in organoid cultures established from human PDX tumors. C75, a validated synthetic fatty acid synthase inhibitor (20 μg/ml), and CD36 mAb (5 μg/ml) treatment independently reduced proliferation compared with untreated organoids, and dual treatment ablated growth by 95% (Fig. 7, A and B).

Fig. 7 Inhibition of de novo lipogenesis increases the efficacy of CD36 blockade.

(A and B) Human PDX-derived organoids treated with C75 fatty acid synthase inhibitor, CD36 mAb, or both in combination for 11 days. (A) Images of day 11 control/treated organoid. Scale bars, 200 μm. (B) Organoid viability after treatment was assessed using the PrestoBlue assay. Data are presented as means ± SEM; n = 6 per group; relative to vehicle control (VC). P < 0.05 is shown by adjoining lines between groups by one-way ANOVA and Bonferroni multiple comparisons test. (C to E) Mouse organoids established from Pten−/− and Pten−/−.Cd36−/− mice treated with and without C75 fatty acid synthase inhibitor for 7 days. (C) Fatty acid uptake was assessed in cells treated with 500 μM oleate (1 mCi/ml of 1-14C-oleate) for 2 hours. (D) De novo lipogenesis was assessed using 5 mM glucose (3 μCi/ml of [14C(U)]-glucose). Data are presented as means ± SEM; n = 6 for all groups. For (C) and (D), horizontal lines between adjoining bars denote main effect for genotype, P < 0.05; *P < 0.05 versus −C75 within the same genotype. Data assessed by one-way ANOVA and Bonferroni multiple comparisons tests. (E) Organoid viability after treatment was assessed using the PrestoBlue assay. Data are presented as means ± SEM; n = 6 per group from two independent experiments. *P < 0.05 versus vehicle control; adjoining line denotes main effect for genotype by one-way ANOVA and Bonferroni multiple comparisons test.

To confirm these results, we developed organoids from Pten−/−- and Pten−/−.Cd36−/−-purified prostate luminal epithelial cells and treated these with C75. Fatty acid uptake was decreased by 50% in Pten−/−.Cd36−/− compared with Pten−/− organoids, and C75 had no effect on fatty acid uptake (Fig. 7C). De novo lipogenesis was increased in Pten−/−.Cd36−/− compared with Pten−/−-derived organoids, and C75 reduced lipogenesis in organoids from both genotypes (Fig. 7D). In accordance with our in vivo data, Cd36 deletion reduced growth by ~50% in the Pten−/− prostate organoids, and this was further reduced to ~80% by inhibiting lipogenesis (Fig. 7E), thereby confirming enhanced antitumorigenic efficacy of the dual therapy.


Although the production of fatty acids via heightened lipogenesis is a well-known metabolic adaptation in prostate cancer (40), several lines of evidence indicated a prominent role for exogenous fatty acids to support prostate cancer pathogenesis (14, 17, 41). In this study, we have demonstrated increased fatty acid uptake and significant lipidomic remodeling in human prostate cancer that is, at least partly, mediated by CD36. Supporting this concept, CD36 is frequently gained or amplified in prostate cancer and is associated with poor patient prognosis (27). The present work further extends on other recent efforts examining CD36 in tumors (23). We show that CD36 deletion restricts fatty acid uptake from the tumor microenvironment, reduces cancer-mediated lipid biosynthesis from fatty acid precursors and the generation of oncogenic lipid signaling pathways, and attenuates cancer growth. These data provide the impetus to target CD36 as a therapy for prostate cancer.

Tumors display altered metabolism relative to benign tissues (1). An important aspect of this work relates to the marked alterations in fatty acid metabolism in human prostate cancer. Recent studies have reported dependency on CD36 for oral cancer metastasis (23) and fatty acid oxidation for triple-negative breast cancer tumor growth (42); however, the bioenergetic requirement for fatty acid uptake and metabolism had not been formally tested in primary human cancer tissues. Here, we report substantial alterations in lipid metabolism in malignant prostate tissue from men with high-risk localized disease, highlighted by increased fatty acid uptake and preferential channeling of these exogenously derived fatty acids into lipid pools that support the requirement for cell division, bulk energy storage (triglycerides and sterol lipids), and membrane production (phospholipids and ceramides)—processes required for tumorigenicity. Some aggressive cancer cells co-opt lipolysis of intracellular triglycerides to enhance FFA availability, which induces high levels of malignancy (43). Although this hypothesis cannot be tested in human tissue, there are more triglyceride-rich lipid droplets in malignant compared with benign human prostatic tissue. Together, our results demonstrate that exogenous fatty acids are of equal importance as glucose for energy provision and contribute to the production of complex lipids in human prostate cancer.

Extending on the human studies, we obtained several lines of evidence that blocking CD36 reduces fatty acid uptake and slows cancer progression. The present data show that reducing fatty acid uptake via loss of Cd36 modulates specific lipid metabolism nodes in tumors driven by loss of Pten, rather than a generalized remodeling of the entire prostate lipidome. In this regard, Pten deletion leads to the release of fatty acids and lysophospholipids from membrane phospholipids and ether lipids. Acyl and ether lysophospholipids, such as lysophosphatidylcholine, lysophosphatidylethanolamine, and lysophosphatidic acid, have all been shown to fuel cancer cell proliferation, motility, and invasiveness by acting as signaling lipids or providing additional sources of fatty acids (4348). Cd36 deletion reduces fatty acid uptake and prevents the accumulation of these lipids and their metabolites, strongly indicating blockade of these tumor-promoting pathways. Aside from fatty acid uptake, we also show that Cd36 deletion causes cPLA2α activation and production of arachidonic acid from phospholipids, which is linked to the severity of prostate cancer in men (49) and is associated with the Cd36 cancer phenotype shown here. Although these changes in lipid metabolism provide a likely molecular explanation for the antitumorigenic effects of CD36 deletion, CD36 also functions as a signaling receptor capable of activating SRC family kinases, mitogen-activated protein kinases, and reactive oxygen species pathways through its recognition of extracellular “danger” ligands, which include oxidized low-density lipoproteins, β-amyloid peptide, Staphylococcus aureus–derived microbial diacylglycerides and lipoteichoic acid, and Mycoplasma macrophage-activating lipopeptide-2 (50, 51). Each of these has oncogenic potential, so it is possible that changes aside from lipid metabolism may contribute to the antitumorigenic effects of CD36 ablation reported here. However, our findings showing reduced CD36 and cancer progression in culture cells (PC3 and LNCaP) and organoids, where danger ligands are absent or in very low abundance, provide strong support for fatty acid uptake and metabolism as an important component of CD36’s tumorigenic effects.

We note the apparent discrepancy with respect to fatty acid oxidation in Pten−/− mice. The acute experiments in isolated prostate tissue using 14C-18:1 FFA show no effect of cancer on fatty acid oxidation, whereas the lipidomic analysis in the same mice demonstrates marked acetylcarnitine accumulation, which is indicative of increased fatty acid oxidation in vivo. In reconciling these data, we speculate that C18:1 FFA transported into prostate tumors is first distributed into phospholipid storage, where they can then be cleaved by cPLA2α to produce fatty acids for oxidation to support energetic requirements. Note that this reaction would produce lysophospholipids containing unsaturated fatty acids (18:1) and saturated fatty acid for oxidation. This biology would not be elucidated using the 14C-18:1 FFA tracer methodology.

An in-depth understanding of metabolism has provided insights into the clinical utility of cancer therapeutics (52, 53). The translational relevance for using CD36 blockade as a therapeutic for patients with prostate cancer is highlighted by our studies using human PDXs of high risk, locally invasive disease, which shows that systemic CD36 mAb administration slows tumorigenicity in prostate cancer that is characterized by high rates of fatty acid uptake. Although CD36 was previously implicated in metastases in several other cancer types (23, 25, 28), the present studies define early localized disease with high-risk features as an appropriate therapeutic window for treatment.

Cancer progression was delayed in Pten−/− Cd36 knockout mice despite a compensatory increase in fatty acid production by de novo lipogenesis. This raises the intriguing possibilities of cross-talk between these pathways to maintain fatty acid flux in transformed cells and/or the existence of an intracellular fatty acid–sensing mechanism/rheostat that could induce multiple pathways to ensure availability of fatty acids to maintain fitness of tumor cells. Such communication within a fatty acid regulatory network was previously described in immortalized prostate cancer cells (54), although it is not yet clear how this metabolic cross-talk is regulated. Our parallel finding that de novo lipogenesis was increased in response to reduced fatty acid uptake prompted our examination of dual targeting of fatty acid transport and de novo lipogenesis as a more effective therapeutic approach in prostate cancer. We confirmed that the combination therapy of CD36 and fatty acid synthase (FASN) inhibition reduces tumorigenesis in prostate cancer organoids more effectively than either treatment alone.

There were several limitations to this study. Given that the focus of this study was on localized prostate cancer rather than metastatic or therapy-resistant disease, we did not assess the effect of Cd36 deletion or blockade on survival in mice. Typically, Pten−/− mice start to die between the ages of ~8 and 20 months depending on the laboratory and the background strain (55, 56). Although it was not tested in this study, given that we show reduced tumor burden and progression in Pten−/−.Cd36−/−, a longer latency of disease progression and prolonged survival in these mice is expected. In further support of CD36’s role in prostate cancer progression, it is worth noting that copy number gain and increased expression of the CD36 gene are evident in metastatic disease compared with localized disease. Another limitation was that we were unable to correlate fatty acid uptake rates in human prostate in vivo with clinical outcomes in patients; hence, it is currently unclear whether fatty acid uptake predicts a worse prognosis at diagnosis, a concept that will be tested in future studies. Assessing CD36 protein expression in primary tissue is unlikely to provide strong diagnostic value because CD36 protein is expressed in human prostate tissues in benign and tumor regions and there was no evidence for increased expression in tumor regions. Although this contrasts the transcript data from other datasets, showing elevated expression of CD36 in primary tumors from men who experienced more rapid disease recurrence, CD36 activity (CD36-mediated fatty acid uptake) is difficult to evaluate by immunohistochemistry because it is heavily dependent on translocation from the cytoplasm to the cell membrane and posttranslational modifications (57). In summary, we provided preclinical evidence that targeting the metabolic differences in fatty acid metabolism between tumor and normal cells might be an effective anticancer strategy and that CD36 might be a pharmacological target for early treatment in high-risk localized prostate cancer.


Study design

The research objective of our study was to determine the effect of fatty acid metabolism in prostate tumorigenesis using analysis of human tissues, cell lines, genetically modified mice, and human PDXs and patient-derived organoids. Patients with prostate cancer undergoing radical prostatectomy were included in the human study (details provided in table S1). No randomization was performed for the human studies; however, investigators were blinded to the allocation of benign or malignant tissue during metabolic analyses. For mouse studies in genetically modified mice, 8- to 24-week-old male mice were used, and for PDX in vivo studies, 8- to 12-week-old male mice were used. One investigator was responsible for group allocation of mice, and subsequent investigators were blind to the genotypes or treatment at tissue collection, metabolic assays, and histological assessment. Experimental replicates were variable for each experiment and are detailed in the figure legend. Raw data for all the experiments are reported in table S2.

Statistical analysis

Statistical analysis was performed using unpaired one- or two-way Student’s t tests, two-way ANOVA, or repeated-measures ANOVA where appropriate. Multiple comparisons were performed using a Bonferroni post hoc analysis when required. Statistical significance was set a priori at P < 0.05. Data are reported as means ± SEM. In vitro experiments were performed in triplicate and expressed as a summary of the mean of replicates with SEM, unless specified otherwise. Histological quantitation of mouse prostate tissue was performed using an Aperio ScanScope digital scanner and viewed using ImageScope software (Aperio).


Materials and Methods

Fig. S1. Survival curves for CD36 and other genes encoding proteins regulating fatty acid uptake.

Fig. S2. Fatty acid uptake in prostate cancer cell lines.

Fig. S3. shRNA knockdown of CD36 in PC3 and LNCaP cells.

Fig. S4. Mouse phenotyping by flow cytometric analysis.

Fig. S5. Phenotype of WT.Cd36−/− mice.

Fig. S6. Assessment of metabolism and cancer pathology in Pten−/−.Cd36−/− mice fed a high-fat diet for 6 weeks.

Fig. S7. Schematic depicting the changes in Pten−/−-induced lipid metabolism that link the production of oncogenic lipids to prostate cancer progression.

Fig. S8. Linking cPLA2α inhibition to the antitumorigenic effects of Cd36 deletion.

Fig. S9. Systemic effects of CD36 mAb treatment in mice.

Table S1. Patient characteristics of specimens used for metabolism studies.

Table S2. Raw data (Excel file).

References (5878)


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 58.
  58. 59.
  59. 60.
  60. 61.
  61. 62.
  62. 63.
  63. 64.
  64. 65.
  65. 66.
  66. 67.
  67. 68.
  68. 69.
  69. 70.
  70. 71.
  71. 72.
  72. 73.
  73. 74.
  74. 75.
  75. 76.
  76. 77.
  77. 78.
Acknowledgments: We thank R. Legaie, H. Nim, X. Chen, and M. Richards (Monash University) for technical and bioinformatics assistance, the Australian Prostate Cancer Bioresource, and the Melbourne Urological Research Alliance (MURAL). The results published here are, in part, based on data generated by TCGA, established by the National Cancer Institute and the National Human Genome Research Institute, and we are grateful to the specimen donors and relevant research groups associated with this project. Funding: This work was supported by the Prostate Cancer Foundation of Australia (ID: PCFA–NCG 3313, awarded to M.J.W. and R.A.T.) and the Diabetes Australia Research Trust (awarded to M.J.W.). M.J.W., M.K.M., L.A.S., and G.P.R. are supported by the National Health and Medical Research Council of Australia (APP1077703, APP1143224, APP1102752, and APP1121057), and R.A.T. is supported by the Victorian Cancer Agency (MCRF15023). L.F. is supported by the Department of Health and Human Services acting through the Victorian Cancer Agency (MCRF16007). Author contributions: M.J.W. and R.A.T. conceived and designed experiments. M.J.W., S.T.W., V.R.H., P.R.W., M.M., J.L., C.H., and B.N. performed metabolism experiments. K.E.A. and D.K.N. performed lipidomic experiments. L.A.S., C.H., and R.B.S. performed the omic analysis. A.K.C., S.T.W., P.R.W., M.M., N.L., L.H.P., B.N., R.R., L.F., and R.A.T. performed prostate cancer experiments including cell culture, PDX, and organoid experiments. M.K.M., G.P.R., M. Febbraio, M.P., S.N., and M. Frydenberg provided reagents, clinical materials, and mouse models. M.J.W. and R.A.T. wrote the paper. All authors edited the manuscript. Competing interests: D.K.N. is a cofounder, shareholder, and adviser of Frontier Medicines and Artris Therapeutics and the director of the Novartis-Berkeley Center for Proteomics and Chemistry Technologies. K.E.A. is a scientist from Frontier Medicines. Data and materials availability: All the data are included in the main text or in the Supplementary Materials.
View Abstract

Stay Connected to Science Translational Medicine

Editor's Blog

Navigate This Article